00001.tif

PDF

Download

Open document

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600 MATERNAL EDUCATION AND CHILDHOOD SURVIVAL IN URBAN AREAS OF ZAIRE: THE PATHWAYS OF INFLUENCE by N'zinga Luyinduladio A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Sociology) May 1995 Copyright 1995 N'zinga Luyinduladio UMI Number: 9617114 UMI Microform 9617114 Copyright 1996, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90007 This dissertation, written by N'zinga Luyinduladio under the direction of h..U... Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies D a te. DISSERTATION COMMITTEE Chairperson To my wife, my children, and my mother for all the love and patience. To the memory of my father. ACKNOWLEDGEMENTS I have immensely benefitted from the suggestions given at various occasions by Prof. David M. Heer, Prof. Maurice van Arsdol Jr., Prof. Eileen Crimmins, Prof. Timothy Biblarz. Professor David M. Heer provided the necessary direction from the beginning of the study. I have also benefitted from the comments of Prof. Samuel H. Preston of the University of Pennsylvania on my dissertation proposal. I also benefitted in Fall 1991 from the useful comments on my term paper (which became later my dissertation proposal) from Prof. Mark Hayward of the Pennsylvania State University in his class "Sociology 545: World Population Problems" while he was still at the University of Southern California. I shall also mention here the valuable comments made to me in 1987 by Dr. Eric Vilquin of the Institute of Demography, Universite Catholique de Louvain, on the French version of my dissertation proposal. Finally, I have benefitted from the computer expertise of Dr. Gerald P. Jones of University Computing Center of the University of Southern California. I thank all of them for their comments, advice, encouragements, and help. However, I remain the sole responsible of any mistake this study still contains. Throughout the last six years, I was supported by the African-American Institute: the African Graduate Fellowship Program and the Partners for International Education and Training, respectively. I am particularly indebted to them for iii their financial and administrative support. I am also particularly indebted to Professor Maurice van Arsdol Jr. and Professor David M. Heer who have done everything for me to succeed in my study at the University of Southern California. I am also grateful to Professor Samuel H. Preston and Professor Etienne van De Walle who had guided me during my Master Program at the University of Pennsylvania. They both were very helpful to me during this hard time where I had to learn English and, at the same time, pursue the academic program. In addition, I shall thank the William and Flora Hewlett Foundation (award # 92-5048 to the University of Southern California, Population Research Laboratory) for the help it has given me during the crucial moments in the course of my research, especially during the last semester (Spring 1995) I have spent at the University of Southern California. I will always remember that. Further, I am deeply indebted to all the Faculty members of the Department of Demography (University of Kinshasa) for the encouragements they gave me at various moments in this research. Last but not least, I am particularly thankful to Prof. Makwala ma Mavambu ye Beda of the University of Kinshasa, Department of Demography, who was my valuable mentor during my first steps as a Researcher in the Department of Demography, University of Kinshasa. My special thanks to Boga Fidzani for typing the tables of chapters four and five. I will always remember this hard work you did for me. I shall also thank Dr. Lukowa Kidima who iv has edited this dissertation. I thank Mr. Kayembe of the "Institut National de la Statistique" for providing me the access to the FONAMES/UNICEF data sets. I also thank the other members of the Technical Team of the FONAMES/UNICEF survey in 1986 and 1987 for their contributions in the study design, data collection, and data processing. The members of this team were Mr. Kinavwidi, Mr. Kayembe, Dr. Shodu, and myself. I learned a lot from all of you. My thanks to friends and members of my extended families in Zaire for the support they have given to my wife and children during my absence. Here, I shall thank my colleague, Prof. Lututala Mumpasi and Prof. Itimelongo Titi for their special regards to my family. v TABLE OF CONTENTS TABLE OF CONTENTS ........................................ vi LIST OF TABLES ....................................... ix LIST OF CHART & MAP ..................................... xvi ABSTRACT ............................................... xvii Introduction .............................................. 1 0.1. Taking Stock of the Situation .................. 2 0.1.1. Why the Maternal Education Debate Continues? ......................... 2 0.1.2. The Nature of this Statistical Association............ .................. 3 0.1.3. What are the Mechanisms Through which Education Affects Mortality? ....... 5 0.1.4. Two Assumptions Underlying the Maternal Education Effect ................. 14 0.2. Justification of the Proposed Research ............................. 14 Chapter 1 MATERNAL EDUCATION AND CHILDHOOD MORTALITY: A REVIEW OF THE LITERATURE ......................... 18 1.1. Aggregate Studies on Education- Childhood Mortality Association ......... 18 1.2. Micro-Level Studies on Education- Childhood Mortality Association ......... 29 1.3. A Theory of Health Transition ........... 57 Chapter 2 THE PROPOSED STUDY: MATERNAL EDUCATION AND CHILDHOOD MORTALITY IN URBAN AREAS OF ZAIRE. FACTORS AND DETERMINANTS, AND PATHWAYS OF EDUCATIONAL INFLUENCE .............................. 67 2.1. The Proposed Questions of Study......... 67 2.2. The Objectives of the Study ............. 69 2.3. The Hypothesis of the Research.......... 70 2.4. Theoretical Framework ................... 72 2.5. Methodological Approaches ............... 76 Chapter 3 THE EMPIRICAL DATA ............................... 78 3.1. Objectives of the Survey and Sampling Design...................... 78 3.1.1. Objectives of the Survey ......... 78 3.1.2. Sampling Design ................... 79 3.1.3. Stratification of the Universe at the First Stage........ 82 3.1.4. Determination of the Sample Size at the First Stage and Sampling Fractions at 1st. and 2nd Stages ...................... 83 3.1.5. Drawing of the sample at the First and Second Stages, Adjustment of Sampling Fractions, Drawing of Final Sample ............ 84 3.1.6. Techniques of Data Collection and Questionnaire ................... 87 3.1.7. Organization of the Survey ....... 97 3.2. Socio-Demographic Characteristics of the Population. Housing Conditions ......................... 100 3.2.1. Demographic Characteristics of the Population ...................... 101 3.2.2. Socio-economic Characteristics of the Population .................. 112 3.2.3. Housing Conditions ............... 115 Chapter 4 INFANT AND CHILD MORTALITY IN URBAN AREAS OF ZAIRE: LEVEL AND DETERMINANTS ................... 128 4.1. Level of Infant and Child Mortality in Urban Areas of Zaire in 1987 ........ 128 4.2. Maternal Education as a Determinant of Infant and Child Mortality in Urban Areas of Zaire in 1987 .................. 136 4.2.1. The Method ....................... 136 4.2.2. The Variables Included and Results of Multivariate Regressions ........................ 137 Chapter 5 FROM MATERNAL EDUCATION TO CHILD MORTALITY: THE PATHWAYS OF INFLUENCE ......................... 158 5.1. Education and Gestational Status of Women ......................... 159 5.1.1. Education of Mother and Birth Weight .................. 159 5.1.2. Education of Mother and Premature Birth ................... 164 5.2. Education and the Use of Maternity for Delivery ............................ 166 5.3. Education of Mother and Child Nutritional Status ...................... 168 5.4. Education and Knowledge of Sugar-Salt-Solution, Incidence of Diarrhoea and Fever, and Behavioral Responses to Diarrhoea or Fever ................................ 176 5.4.1. Incidence of Diarrhoea & Fever: Descriptive Statistics ......................... 180 vii 5.4.2. Incidence of Diarrhoea & Behavioral Responses of Mothers in Urban Zaire: Multivariate Results ............................ 184 5.4.3. Incidence of Fever and Behavioral Responses to Fever: A Multivariate Analysis .......... 200 5.5. Maternal Education and Immunization of Children 12-23 Months of A g e......... 214 5.5.1. Immunization of Children 12-23 Months: Cross-Tabulation ................... 215 5.5.2. Education and Child Immunization: A Multivariate Logistic Regression........ 220 5.6. Maternal Education and Occupation Status of Mothers ....................... 225 Chapter 6 CONCLUSION AND RECOMMENDATIONS .................... 238 6.1. The Findings of the Study ............... 240 6.2. Congruence of the Findings of this Study with Previous Studies ..... 251 6.3. Suggestions for Future Researches ....... 263 6.4. Policy Recommendations .................. 264 REFERENCE ............................................. 268 TABLE APPENDIXES ........................................ 281 viii LIST OF TABLES1 Table 0.1.1. Relative Risks of Death in Infancy and Childhood by Region and Education (an Average of WFS Survey) ..... 4 Table 0.1.2. Relative Risks of Death, Before and After Adjustment for Other Socio-economic Factors .................................... 7 Table 0.1.3. Relative Risk of Neonatal, Post-neonatal, and Childhood Deaths, Before and After Adjustment for Other Socio-economic Factors .............. 8 Table 1.2.1. Child Mortality Indices by Education of Mother, Respondents 15-49, Ibadan City, Nigeria 1973 ............... 30 Table 1.2.2. Probabilities of Dying by Age 1 through 5 by Maternal Years of Schooling, Sudan, 1973 ............................. 38 Table 3.1.1. Population in the 1984 Census, Number of Household Estimated, Estimated Size of the Sample, and Global Sampling Fraction per City ...................................... 81 Table 3.1.2. Distribution of Compounds, Households, and Population Surveyed Among the Six Strata. 13 Cities of Zaire, 1987 .......... 87 Table 3.1.3. Calendar of the Execution of the Survey .................................. 100 Table 3.2.1. Distribution of Population by Residence Status ......................... 101 Table 3.2.2. Composition of the Population by Age and by Gender ............................ 102 Table 3.2.3. Distribution of Population (in %) by Marital Status and Gender ................. 103 1 The logic of numbering is as follows: the first number indicates the chapter, the second indicates the section in the chapter, then the last number indicates the table number. Table 3.2.4. Distribution of Women by Age Group. Children Ever Born by Age, and Average Parity per Woman by Age .................. 104 Table 3.2.5. Distribution of Children Born in 1981-85, Children Born the last 12 months before the Survey. Period Fertility Rates (in thousand) in 1981-85 and in the last 12 months ................................. 105 Table 3.2.6. Level of Education of Male in July-August 1987 ......................... 104 Table 3.2.7. Level of Education of Female in July-August 1987 ...................... 109 Table 3.2.8. Educational Patterns by Strata and Gender in Urban Areas in Zaire, 13 Cities. Population 15 Years or more. (Percentages) ......... 110 Table 3.2.9. Average Number of years of Schooling of Population Over 15 Years .............. 110 Table 3.2.10. Distribution (in %) of Urban Population by Occupation Status and by Gender in August 1987 ............ Ill Table 3.2.11. Structure of Household Returns ("Recettes") July-August 1987 (in zaire currency) ...................... 113 Table 3.2.12. Structure of Household Consumption Expenditures. July-August 1987 (in zaire currency) ...................... 114 Table 3.2.13. Distribution of Income in July-August 1987 ......................... 114 Table 3.2.14. Distribution of Households (in %) by Number of Rooms ....................... 115 Table 3.2.15. Distribution of Households (in %) by Types of Materials Used to Build the Ground, Walls, and Roofs ............. 117 Table 3.2.16. Distribution Households by Ownership of the House ............................. 117 x Table 3.2.17. Distribution of households by Location, Mode of Use and Distance between the Latrine and the Housing Unit ....................................... 118 Table 3.2.18. Distribution of Households by Source of Supply of Drinking-Water and Distance of the Supply and the House ................................. 121 Table 3.2.19. Distribution of Households by Mode of Human and Household Wastes ............... 122 Table 3.2.20. Distance between the House and the Place Where Household Wastes are Thrown ................................ 123 Table Annex 1: Criteria of Stratification of Primary Sampling Units (quarters, blocs, cellules, etc.) ......................... 125 Table Annex 2: Distribution by Strata and by City of Primary Sampling Units (quarters, blocs, cellules) in 1987..... 126 Table Annex 3: Distribution by Sub-Strata and by City of Total and Sampled Primary Sampling Units, and Sampling Fractions at 1st, 2nd Stages, and Global .................. 127 Table 4.1.1. Distribution of Births and Deaths, and Mortality Birth Cohorts 1981-1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ............... 282 Table 4.1.2. Proportion of Children Dead Among the Ever Born by Mother's Age and Years of Schooling. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 .............................. 133 Table 4.1.3. Probabilities of Dying from Birth to Age X (in years). Cohort of Births 1981-1985................ 134 Table 4.1.4. Estimated Levels of Infant Mortality Using 5q , j as an Indicators of Entry in Coale and Demeny Life Table Models.......................... 135 xi 155 286 140 297 144 149 150 308 315 315 xii Variables Included in the Study ....... Odds Ratio and Standard Errors for Logistic Regression of Infant Mortality. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987.. Average Odds Ratios of Significant Factors of Infant Mortality in Urban Zaire, 1987....................... Odds Ratio and Standard Errors for Logistic Regression of Child Mortality. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987.. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Infant Mortality in Urban Areas of Zaire. The Last Model ... Mean Odds Ratio of Significant Factors of child Mortality in Urban Zaire, 1987....................... Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Child Mortality in Urban Areas of Zaire ................................... Parameter Estimates and Standard Error (in parenthesis) for Multinomial Logistic Regression of Birth Weight; Children Born in Cohort 1981-1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ........... Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Premature Birth. Cohort 1981-1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 .................................... Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Maternity for Delivery. Women who gave birth between 1981 and 1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987............. Table 5.3.1. Distribution (in %) of Children 1-4 Years by Age and Nutritional Status, Cohort 1983-1986 .......................... 172 Table 5.3.2. Parameter Estimates and Standard Error (in parenthesis) for Multinomial Logistic Regression of Nutritional Status of Children 12-59 Months. Urban Areas of Zaire, Cities. Fonames/Unicef Survey, 1987................ 323 Table 5.4.1. Incidence of Diarrhoea and Fever by Age. Children 0-59 months. Urban Areas of Zaire, 13 Cities. FONAMES/UNICEF Survey, 1987 ............... 181 Table 5.4.2. Incidence of Diarrhoea and Fever by Mother's Education. Children Aged 0-59 Months Reported Having Diarrhoea and Fever in the Two Weeks Reference by Education of Mother .................................. 183 Table 5.4.3. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Incidence of Diarrhoea among Children Aged 0-59 months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987................................ 332 Table 5.4.4. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Sugar-Salt- Solution by Mothers Whose Children are Aged 0-59 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987............................... 342 Table 5.4.5. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Modern Services in case of Diarrhoea by Mothers Whose Children are Aged 0-59 Months. Urban Areas of Zaire, 13 Cities. Fonames-Unicef Survey, 1987................ 354 xiii Table 5.4.6, Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Modern Medicine to Treat Diarrhoea. Urban Areas of Zaire, 13 Cities. Fonames/UNICEF Survey, 1987 ....................... 361 Table 5.4.7, Odds Ratio and Standard Error (in parentheses) for Logistic Regression of Promptness in Seeking Modern Health Care in case of Diarrhoea. Urban Areas of Zaire, 13 Cities. FONAMES/UNICEF Survey, 1987 ....................... 368 Table 5.4.8. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Prevalence of Fever among Chidren 0-59. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ........................... 375 Table 5.4.9. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of the Use of Modern Health Services for Child Fever. Urban Areas of Zaire, 13 Cities. FONAMES/UNICEF Survey, 1987........... 385 Table 5.4.10. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of the Use of Drugs in case of Fever Among Mothers whose Children are Aged 0-59 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ........... 396 Table 5.4.11. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Promptness in Seeking Health Care in Case of Fever. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ............................ 405 Table 5.5.1. Distribution of Children 12-23 Months by Types of Vaccine, Immunization Status and by Knowledge of Immunization Dates ..................... 216 xiv Table 5.5 Table 5.5 Table 5.6 .2. Percentage of Children 12-23 Months who Have Never Received any Vaccine; Percentage who Received all the Vaccine, Immunization Prevalence against DPT, Polio, Measles and BCG; Percentage of Children 12-23 Months whose Birth Took Place in Hospital Setting; and Percentage of Children with a Immunization Card by Socioeconomic & Biological Characteristics of Mother and Child, and by Types of Neighborhood ............................ .3. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Completeness of Immunization Among Children 12-2 3 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ............ .1. Parameter Estimates and Standard Error (in parenthesis) for Multinomial Logistic Regression of Occupational Status of Mothers with Preschool Children. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 ..................................... 231 417 423 XV LIST OF CHARTS & MAP Chart 2.4.1. The Routes of Maternal Education Effect ............................. 75 Chart 3.1.1. The Organization of Survey ................ 98 Chart 4.2.1. From the Proximate Determinants to Infant Death in Urban Zaire: The Significant Paths .......................................... 143 Chart 4.2.2. From the Proximate Determinants to Child Death in Urban Zaire: The Significant Paths ....................... 153 Chart 5.1.1. Pathways of the Effect of Maternal Education on Infant and Child Mortality ........................... 229 Map 3.1.1. The Official Map of Zaire in 1988 124 xv i ABSTRACT This study has two objectives. First it analyzes the effect of maternal education on childhood mortality. Second it investigates the pathways through which maternal education exerts its impact on child survival. The data for this study were collected in 13 Cities of Zaire (1987) from a representative sample of 6,500 households. A logistic regression was used for childhood mortality, use of maternity services, length of pregnancy, incidence of child morbidity, and behavioral responses to diarrhoea and fever. A multinomial logistic regression was used for determining the odds of being a working mother, the odds of normal birth weight or above normal weight, the odds of severe or moderate malnutrition versus "good" nutrition. The results indicate that infant mortality is significantly affected by: (1) gender (with female infants having lower mortality than males); (2) mother's age at birth (infants born to mothers of 25-34 years are less likely to die in infancy than infants of teenagers); (3) birth weight (infants born with normal weight and those born with above normal weight have lower mortality than those with low birth weight); (4) length of pregnancy (premature infants are more likely than full-term babies to die in infancy); (5) number of bedrooms (the more bedrooms the lower the mortality). Child mortality is significantly influenced by: (1) place of delivery (with children born in hospital having lower child mortality than those born at home); (2) type of neighborhood of residence (living in rich neighborhoods is life-protective for children); (3) occupation of mother (working mothers have higher childhood mortality than housewives); (4) the number of bedrooms (the more bedrooms the lower the childhood mortality). Concerning the pathways for educational effects, our study finds that education: (1) moderately increases the odds for a mother to deliver a baby of normal weight? (2) reduces the odds of giving birth to a premature baby; (3) increases the odds of mother to deliver in maternity hospital? (4) decreases the odds that a child will be moderately malnourished; (5) increases the odds that mothers use the sugar-salt-solution; (6) increases the odds that mothers use health services for malaria? (7) increases the odds that mothers work outside the home. xviii INTRODUCTION This dissertation has two objectives. First, it seeks to examine the impact of maternal schooling on infant and child mortality in urban areas of Zaire. Along with this objective, I seek to assess the relative contribution of economic, demographic, biological, and environmental factors in explaining the education-child survival relationship. The second objective of this dissertation is to investigate the pathways through which maternal education exerts an impact on child survival. Following the second objective, I will examine a certain number of issues discussed in the health transition theory in relation to child survival in the Third World. For example, can we find evidence that educated mothers compared to uneducated are most likely to be better nourished during pregnancy. If yes, is this because of education or is this a reflection of economic status? In addition, can we find evidence that education increases mothers' awareness about effective ways to prevent, recognize, and treat childhood ailments? Finally, does maternal schooling have an impact on child nutritional status? If the latter issue is true what is the role of economic and other factors on education-child nutrition relationship. 1 0.1. TAKING STOCK OF THE SITUATION 0.1.1. Whv The Maternal Education Debate Continues? One of the great findings in demographic research in Developing Countries is the robust and widespread effect of maternal schooling on demographic phenomenon, especially child mortality and fertility. However, this association is, according to LeVine et al. (1991, pp.459-460), "extremely familiar and persistently ambiguous." Therefore, the debate on the topic is still "a la une," i.e. is still going on. In studies on child mortality-maternal education association, two issues still channel the debate. First to what extent is this association a reflection of economic factors in the household. Secondly, to what extent is this association influenced by the health service use and domestic health practices, two factors believed to be affected by schooling. According to Caldwell (1986), this debate is not new. It represents a new twist to the old development versus medical technology debate. Other authors think that the debate is biased since the effect of education, particularly maternal education "may have been exaggerated." 2 0.1.2. The Nature of this Statistical Association Four features have been identified regarding the nature of the statistical association between maternal education and childhood mortality. The first feature is that this association is negative. On average, for example, each additional year of mother's education coincides with a 7-9 percent decrease in under five year-old mortality (Cochrane et al., 1980, 1982; United Nations, 1985).1 The second characteristic of maternal schooling-child mortality association is that there is no threshold in this inverse relationship. On this point Cleland (1989, p.402) mentions that from the very start of primary school each additional year's retention within the formal school system is associated with a fall of 3-5 per cent (net of economic factors) in childhood mortality. This feature suggests that neither the content of schooling nor the basic literacy nor the substance of the formal curriculum is of major importance, but rather the effect of education in changing mentality or outlook of people. This is what is called the ideational change due to education (LeVine et al., 1991). The third feature of this association is that maternal education-child mortality relationship persists in all major 1 This represents the gross effect. See below about the net effect. 3 regions of developing countries, suggesting that "highly culture-specific explanations can only be partial" (Cleland, 1989, p.402). Finally, the literature shows that this association is stronger in childhood (1-4 years) than in infancy (0-1 year). For example, Table 0.1.1. below indicates that the fall in mortality corresponding with 4-6 years of mother's years of schooling is about 20% in infancy, but it ranges from 30 to 58% in early childhood and from 43 to 72% in later childhood. This evidence strongly suggests that both infant and childhood mortality are sustained by different sets of causes and factors (Palloni, 1981; Kim, 1988). Table 0.1.1.: Relative Risks of Death in Infancy and Childhood by Region and Education (an average of WFS survey) Region 0 Years of 1-3 Schooling 4-6 7 + First Year Sub-Saharan Africa (n=8) 1.00 0.85 0.78 0.56 Latin America (n=9) 1.00 0.86 0.65 0.45 Arab countries (n=8) 1.00 1.05 0.76 0.66 Asia (n=9) 1.00 0.88 0.80 0.59 Second Year Sub-Saharan Africa 1.00 1.00 0.70 0.45 Latin America 1.00 0.73 0.42 0.18 Arab countries 1.00 0.69 0.60 0.20 Asia 1.00 0.95 0.63 0.15 Third-Fifth Year Sub-Saharan Africa 1.00 0.64 0.53 0.43 Latin America 1.00 0.75 0.41 0.17 Arab countries 1.00 0.52 0.28 0.13 Asia 1.00 0.75 0.57 0.27 Source; Cleland and Ginneken, 1988, op. cit., Table 1, p.1358 4 0.1.3. What are the Mechanisms Through which Education Affects Mortality? Basically there are four common channels through which education affects childhood survival: 1) reproductive health, 2) socioeconomic status of the family, 3) use of modern health services, and 4) the domestic health care of children (Cleland and Ginneken, 1988; Cleland, 1989). The first channel, the reproductive health, includes the family formation pattern (i.e., age of mother at birth of child, her parity, her pace of reproduction, and the rank of the child), the nutritional or gestational status during pregnancy, and the general health status of the mother. About this channel, the works by Hobcraft et al. (1984, 1985) and Cleland and Ginneken (1988) indicate that the benefit accorded by education has little to do with changes in procreative behavior. Also previous works show a lack of convergence in the findings concerning the benefit of education in changing procreative behavior. For example, Hobcraft et al. (1984, 1985) have shown that educated mothers have the tendency to reproduce at low risk ages; while a recent study by Ahmad et al. (1991) based on the 1986 DHS data from Liberia did not apparently confirm these findings. In contrast, this latter study establishes that younger mothers— in general more educated— have lower childhood survival rates than order mothers who, in large majority, are less educated. About nutritional status of mothers during pregnancy it is 5 thought that educated women may achieve higher average birth weights because they receive a larger share of food within the home; because they flout the taboos concerning the consumption of protein sources such as chicken and eggs during pregnancy; because they are more innovative in seeking antenatal care; because they engage in less heavy manual labor during the later months of pregnancy; or because they are a healthier group to start with (Ware, 1984, p.195). However, as discussed by Cleland and Ginneken (1988), mother's nutritional status during pregnancy cannot be the dominant explanation since it only helps to explain the effect of education on neonatal and, to a lesser extent, post-neonatal mortality. It does not provide answer to the question why the education impact is stronger in childhood than in infancy. The second channel through which education is thought to impact on mortality is the socioeconomic status. Here, the central question consists of determining if the strong education-mortality association simply reflects the impact of economic advantage on health and survivorship. On this issue, the review of literature made by Cleland and Ginneken (1988) shows that the effect of education continues to be statistically significant even after controlling for economic status. However, this broad finding merits some comments. First, as mentioned by Cleland and Ginneken (1988, p.1359-60), demographic surveys used in most of reviewed studies lack detailed data on household income (or expenditure), relying mostly on substitute measures of income which "do not adequately capture income or wealth." Second, though direct income measures are missing, the economic effect on childhood survival is not negligible. There are studies such as the comparative study (of 15 countries) by the United Nations (1985) which have shown that about half of the gross impact of mother's education could be attributed to economic advantage (Table 0.1.2.). Table 0.1.2.: Relative Risks of Death, before and after Adjustment for Other Socio-Economic Factors. Years of Schooling < 4 4-6 7 + Unadjusted 1.00 0.81 0.49 Adjusted 1.00 0.81 0.51 Source: Cleland and van Ginneken (1988, p.1359). Similarly, the analysis of the WFS data by Hobcraft et al. (1984) has revealed that economic advantage in the education-mortality relationship accounts for about half of the gross effect of mother's education, especially in childhood mortality, while for both neonatal and post-neonatal mortality the net education effect is reduced after controlling for economic status (Table 0.1.3.). 7 Table 0.1.3.: Relative Risks of Neonatal. Post-neonatal, and Childhood Deaths, before and after Adjustment for Other Socio-Economic Factors Years of Schooling 0 1-3 4-6 7 Neonatal Unadjusted 1.00 0.87 0.76 0.57 Adjusted 1.00 0.98 0.90 0.86 Post-neonata1 Unadjusted 1.00 0.82 0.73 0.45 Adjusted 1.00 0.98 0.90 0.78 Childhood Unadjusted 1.00 0.68 0.47 0.23 Adjusted 1.00 0.82 0.70 0.51 Source: Cleland and Ginneken (1988, p.1360). There are exceptions to the general finding that maternal education affects childhood survival. For example, in the study by Hobcraft et al. (1984) the following countries were found out of the general pattern. They include Lesotho, Sudan, Haiti, Jamaica, Trinidad and Tobago, Syria, and Fiji. In the study by the United Nations (1985) no significant net maternal effect was found for Nepal, Jamaica, and Sudan. Numerous explanations are given for these exceptions, namely: (1) guality of data or substantial omission of deaths for Sudan and the three English Caribbean countries, (2) limited cases of educated mothers for Nepal, (3) ethnicity for Fiji. It is interesting to note the fact that all the studies mentioned above did not include the interaction between 8 education, income and household facilities, admitting implicitly the assumption of additivity, i.e., the impact of education on mortality is the same at all economic levels. However, there are studies by Casterline et al. (1989) for Egypt, and Pebley and Stupp (1987) for Guatemala, which have shown the existence of interaction between education and income. The third mechanism through which education affects child mortality is the use of health services. About this pathway, the paper by Cleland and Ginneken (1988, p.1361) mentions the following role expectations related to education education is likely to impart a greater responsiveness to novel ideas and services, a greater identification with the outside world, more social confidence at handling officials and perhaps an enhanced ability and willingness to travel outside the home community in search of services. Cleland and Ginneken (1988) mention many empirical studies which have found a positive association between maternal education and use of modern preventive health services and, less often, of modern curative services. These conclusions prevail even after controlling for economic variables, demographic variables (maternal age, parity, etc.), and environmental variables (rural-urban residence, etc.).2 Other studies in Ibadan (Nigeria), and in Mexico3 support the 2 See Table no.3 shown in J.G. Cleland and J.K. van Ginneken (1988), op. cit., p.1361. 3 Cited by Cleland and van Ginneken, 1988, op. cit., p.1361, references nos.35-36. 9 assumption that educated mothers are more likely than the uneducated to seek medical attention with greater timeliness, they are more likely than the uneducated mothers to extract a higher quality of care and adhere to advice with greater persistence. However, the work by Caldwell et al. (1975, 1985), which assumed that the benefit accorded to maternal education may be greater in places where modern health facilities are available, found that, in contrast, differentials in childhood mortality by maternal education were in fact greater in a Nigerian village with a hospital than in a village without such a facility, indicating that the presence of health services improved child survivorship by 20 percent, while maternal education in the absence of services improved child survivorship by 33 percent. Note however that interaction of services and maternal education resulted in 87 percent improvement. Contrary to the argument of Caldwell et al. (1975, 1985), there are studies which argue that availability of health services has an equalizing influence. Of these studies three are worth mentioning. First the study by Palloni (1985), cited by Cleland and Ginneken (1988, p.1362), argues that "in societies where services are widely available, personal characteristics, including education, become less important determinants of health." In addition, the study by Rosenzweig and Schultz (1982), cited by Bicego and Boerma (1991, p.17) argues that "where severe obstacles of access exist, education 10 is key in placing a woman at a relative advantage in term of securing services." Finally the study by Cochrane et al. (1980), cited by Cleland and Ginneken (1988, p.1362) which has indicated that "educational differentials in mortality tend to decrease as per capita health expenditure rises." However, there are studies which have shown an independent effect of maternal education on health care provision. Evidence for this argument comes from developed as well as underdeveloped countries. In Africa, in particular, two studies in Kenya, one by Mosley (1985), another by Ewbank and Kekovole (1986),4 have shown that child mortality gradient by education is kept across provinces and districts independently of health services provision and/or utilization. The fourth channel through which education of mother is believed to affect childhood health is the domestic care of children. On this issue two questions are of importance. One is what are the attitudinal, social and cognitive consequences of education which lead to behavioral changes in mother's care within the domestic context. Evidence gathered to date, although still meager and inconclusive, indicates that educated mothers, compared to the uneducated may attach high value to the welfare and health of children; have greater decision-making power on health-related and other matters; be less fatalistic about disease and death; be more knowledgeable about disease prevention and cure; be more innovative in the use of remedies; be more 4 All of them are cited by Cleland and Ginneken, op. cit., p.1362. 11 likely to adopt new codes of behavior which improve the health of children though they are not perceived as having direct consequences for health (Cleland and van Ginneken, 1988, p.1363). The second question of importance is about identifying the behaviors, at the household level, through which maternal education could lead to improved health of children. Four behavioral factors have been found in the literature. First is the domestic health care. For example, evidence gathered in Kenya indicates that educated mothers are likely to treat measles and diarrhoea in an effective modern manner and less likely to use traditional remedies. In Nigeria, evidence gathered by Orubuloye and Caldwell (1975) shows that educated mothers are more likely to substitute modern medicines sold by pharmacies or hawkers for traditional herbal remedies. The second behavioral factor concerns nutrition. Here, Cleland and Ginneken (1988) indicates that few empirical data exist on the relationship between maternal education and quantity of food intake or nature of diet, except for the education effect on breastfeeding. About this latter, data reveal that because breastfeeding in developing countries acts as a partial safeguard against infection and nutritional deficiency in infancy, it is understandable that, first, maternal education is weaker at post-neonatal ages than in childhood; second, for the same reason, the stronger education-infant mortality association observed in Latin America than in other regions reflects the fact that breast 12 feeding is generally shorter in Latin America than elsewhere. This explanation may be also appropriate for the stronger education-infant mortality association observed in urban areas outside Latin America than in rural areas. In the present research, I will not only test if the effect of maternal education is weaker at post-neonatal ages than in childhood ages, but I will also investigate whether or not there is a stronger education-infant mortality association in rich quarters of urban areas than in poor areas. The third behavior which affects child health at household level concerns mother's employment away from home. Here, the rationale is that educated mothers are more likely than the uneducated to work outside the home. This is supposed to have an adverse effect on infant and childhood health, particularly if work reduces breastfeeding or curtails child care. However, the large majority of empirical studies have failed to confirm such an effect, except in few cases (Farah and Preston, 1982; Hobcraft et al., 1984; etc.). The last behavioral factor of concern is the preventive care in the home. Here again, data are both meager and inconclusive. There are studies which have shown that preventive health care is significantly associated with maternal education, after control of income. There are others which have indicated that preventive care is not associated with maternal education. 13 0.1.4. Two Assumptions Underlying the Maternal Education Effect The literature on maternal education suggests two basic assumptions. First, higher education is expected to be associated with higher survival because it is a proxy for increased command over resources, resulting in higher quality clothing, shelter, nutrition, medical care, sanitary facilities, and water supply. The second assumption is suggested by Caldwell (1979). It states that female education in particular helps break traditional norms and alter the sexual and generational balance of power in the household, which may result on a reallocation of resources from the old to the young. 0.2. JUSTIFICATION OF THE PROPOSED RESEARCH It clearly emerges from what is mentioned above that education is a powerful socio-cultural factor for childhood health and mortality. However, as Cleland (1989, p.415) has mentioned it "convincing explanations of the relationship, particularly that between maternal schooling and child survival, still elude us." There are practical and theoretical reasons to pursue the research on this topic. First, I agree with Cleland (1989, p.416) that the study of the education- mortality association is still justifiable especially on the ground of practical implications because this topic offers "an 14 excellent entry point for the study of general household determinants of health.” In that case, a deep understanding of mother's education and its role in child welfare will bear fresh understandings of the determinants of health care utilization, household decision-making and intra-household resource allocation, etc., (Cleland, 1989). On the ground of theoretical interests, the study on the relationship between maternal education and child survival is justifiable for several reasons. The first reason for pursuing this study is that the unavailability of data to capture variation in household wealth or disposable income noted in the previous studies make it difficult to respond adequately to the question "to what extent is the observed relationship merely a function of education's link to economic status?" (Bicego and Boerma, 1993, p.1207). Our study will answer that question by assessing the net effect of education on child mortality after controlling for household income. The second theoretical reason for pursuing this topic is the need to re-assess whether the education-mortality association is in fact a causal one and, if so, to examine new evidence to see if the theory of education-conditioned use of modern health services and change in pattern of family formation is conclusive. This particular reason is justifiable because the evidence relating education to the use of health services, on the one hand, and to change in pattern of family formation, on the other, is still inconclusive. About health 15 care utilization by mothers, there is considerable evidence from studies such as Tekge and Shorter (1984) for Jordan, Benyoussef and Wessen (1974) for Tunisia, Caldwell et al.(1983) for South India, Mbake and van de Walle (1987) for Sub-Saharan Africa, the Cebu Study Team (1991) in Filipino, and the study by Streatfield et al. (1990), etc. Finding from these studies support the thesis that educated women have greater use of health care and greater awareness of correct immunization schedules, etc. At the same time, there are studies such as Barbieri (1990) for Senegal, cited by Bicego and Boerma (1993, p.1208), that have not found conclusive causal relationship between maternal education and use of health services, that is, "the effect of maternal education on child survival does not operate through differential health service utilization or family pattern." About the impact of education on pattern of family formation, the above mentioned studies also convey the same mixed message. Thus, all these mixed conclusions make it very difficult to determine, with minimum doubt, the question what behaviors serve to mediate the education advantage. The last reason for pursuing this research is that no study on this topic has been done using data from Zaire. Therefore, my study, which intends to examine the household determinants of child health, will enlighten policy-makers in setting a new child health policy, in particular, and a responsible reproduction health policy. 16 This dissertation includes six chapters. In chapter 1 our goal is to review the literature related to the association maternal education-child survival. In chapter 2 I will present the questions, objectives, hypotheses, and methodology of the proposed study. Chapter 3 is devoted to presenting the data used in this study. Chapter 4 will analyze the effect of maternal education on infant and child mortality in urban areas of Zaire. Finally, chapter 5 will assess the pathways through which maternal education affects childhood survival. In chapter 6, I will draw the conclusions of the study. 17 Chapter 1 MATERNAL EDUCATION AND CHILDHOOD MORTALITY: A REVIEW OF THE LITERATURE This chapter will review the literature on the relationship between maternal education and child mortality in the developing countries with special attention on Africa.1 Because of the abundance of the material to be covered, the review is divided into three sections. Section 1 reviews aggregate level studies on the association between maternal education and childhood mortality. Section 2 presents micro level evidence for this association. Finally, section 3 discusses the theory recently built to account for the education-child mortality relationship in the Third World. 1.1. AGGREGATE STUDIES ON EDUCATION-CHILDHOOD MORTALITY ASSOCIATION At aggregate level of study, there exists a tremendous amount of evidence on the education-mortality association (e.g., Cleland, 1989; Alachkar and Serow, 1988; Preston, 1975, 1980; Cochrane et al., 1980, 1982; Flegg, 1982; etc.). In a review of the literature on maternal education and child survival, Cleland (1989) remarks that, although the majority 1 Here, the term child mortality is used in large sense. It includes mortality in the age period 0-4 years. 18 of studies reviewed share the data base gathered and published by the United Nations agencies, they display a substantial diversity in the following matter: selection and total number of countries, choice of the dependent variable, time period studied, and whether the studies are cross-sectional or trends. Beyond these remarks, Cleland (1989, p.400) indicates that the most remarkable standard result from all cross- sectional studies is the strong net association between adult literacy (or female adult literacy) and mortality, which holds despite the time periods under consideration and despite the problems of measurement associated with mortality as well as with socioeconomic variables. The author's review of literature suggests that this association is somewhat less strong for infant mortality than for life expectancy at birth. He also indicates that, in contrast to cross-sectional studies which display net association between maternal education and child survival, studies which attempt to account for mortality trends have not been effective in finding significant link maternal education and child survival. Another cross-sectional study by Alachkar and Serow (1988) measures the socioeconomic factors that influence mortality and isolates the determinants which have the strongest effect. (This study includes data from World Bank for 125 countries— developed and undeveloped countries— , six measures of mortality and six socioeconomic factors, all measured for 1983). What this study reveals (particularly the 19 multivariate analysis) is that there are three strong and consistent predictors of mortality, namely the reductions in fertility, the increases in the proportion of school-aged population enrolled in school, and the increases in the supply of health care providers (i.e., the reductions in the ratio of population to physicians). The relative weight of these factors varies slightly with the indicator of mortality under consideration. For infant mortality, for example, this study shows that a one-unit increase in fertility raises significantly this mortality by 0.4592 after all other socioeconomic variables have been controlled, while controlling for other socioeconomic variables, a one-unit increase in school enrollment significantly reduces infant mortality by 0.2522. The decrease in infant mortality induced by a one-unit rise in the supply of health care providers is 0.1801. The latter factor is also significant. For child mortality (1-4 years), the regression coefficients which correspond to these three factors are 0.3745, -0.2207, and -0.3922 for fertility, school enrollment, and supply of health care providers, respectively. What this cross-sectional study reveals precisely is that declines in fertility level (that correspond roughly to a decline of crude birth rate at level below 35 per 1000) are quite likely to induce declines in infant and child mortality, while the increase in educational attainment of population, especially at levels above the primary school, and the increase in supplies of health care, 20 will reduce the levels of mortality and increase the expectation of life. A recent aggregate-level study by Gbesemete and Jonsson (1993) focuses on the question whether there are factors outside the medical sphere that affect infant mortality rate in Africa. The study includes data from 28 low-and middle- income African countries for the year 1984.2 The study also reveals three factors which have a negative and significant association with infant mortality in this sample of countries. These factors are gross national product per capita (regression coefficient -0.11), school population as a percentage of the population 5-19 years of age (-0.26), percentage of the population with access to health care services (-0.13). Together these factors explain 80% of the variation in infant mortality in the sample. In addition, the authors indicate that population density is significantly and positively associated with this mortality with a regression coefficient of +0.06. However, female education, along with other factors such as water supply, food aid, calorie supply, health care expenditure, and degree of urbanization, is not found to be a significant factor in this study, though it is negatively associated with infant mortality. In light of these findings, Gbesemete and Jonsson (1993, p.155) insist that a 2 These countries are Ethiopia, Zaire, Burkina Faso, Malawi, Uganda, Niger, Tanzania, Somalia, Togo, Benin, Liberia, Ghana, Sierra Leone, Kenya, Sudan, Lesotho, Mozambique, Mali, Zambia, Nigeria, Cote D'Ivoire, Zimbabwe, Cameroon, Tunisia, Senegal, Congo, Morocco, and Botswana. reduction in infant mortality rate in Africa "is feasible only with changes on diverse fronts rather than by marginal improvements in a few determining factors." In their classical study, Cochrane et al. (1982) have examined a total of four aggregate-level studies which provide evidence on 14 data related to infant and/or child mortality. These studies illustrate that the educational factor is statistically significant, with the expected negative sign. For infant mortality, Cochrane et al. (1982) have estimated that one-percent increase in literacy reduces infant mortality by about one-half death per thousand, while "a change from zero to complete literacy would reduce infant mortality by 55 to 40 per thousand" (p.243). For child mortality, the authors estimate that a change from zero to total literacy would cut down this mortality by 25 to 18 per thousand. However, the authors warn that the use of aggregate data in order to generalize about the effect of education on child survival might be done with caution because "in aggregate data parental education is probably highly correlated with other factors which improve health and reduce mortality" (p.238). Education as a key factor for mortality is extensively discussed by Caldwell in two papers published in 1986 and 1990. In his 1986's paper, "Routes to Low Mortality in Poor Countries," Caldwell mentions particularly literacy among factors which can help reduce infant and child mortality in low income societies. In effect, he shows that infant 22 mortality and life expectancy at birth are highly correlated with the proportion of females in school a generation earlier. The correlation coefficients are -0.8563 and 0.8744, respectively. Other important indicators of low mortality mentioned by Caldwell (1986) are levels of family planning practice (correlation coefficient -0.8234), male school attendance (-0.6219), ratio of doctors to population (-0.6105), and nutrition levels (-0.6095). The study found, however, that levels of income have lower correlation with infant mortality (-0.3109) and expectation of life at birth (+0.3862). Further, Caldwell reports that, at community and national level, higher literacy levels are not only necessary to influence individual behaviors and beliefs, but they are also needed to induce popular pressures for fuller and more efficient health care systems. In another paper published in 1990, "Cultural and Social Factors Influencing Mortality Levels in Developing Countries", Caldwell attempts to build a more general theory of mortality transition by examining existing aggregate and individual- level evidence from developing countries. This evidence reveals that, in contrast to today's developed countries where the unprecedented decrease in mortality was due to the rise in real incomes of population induced by the Industrial Revolution and to the improvements in health services induced by the advances in science and technology, in today's developing countries levels of income and access to health 23 services are weak predictors of mortality levels and that social determinants (especially education) apparently play a key role in explaining the remarkable health achievements reached in these societies. Four aggregate-level studies cited by Caldwell (1990) are worth mentioning here. The first of these is the 1985 Rockefeller's study ("Good Health at Low Cost") , which examined four Third World populations which have reached, at low cost, remarkable low levels of mortality (life expectancies ranging from 66 to 70 years).3 Caldwell reports that the main forces for driving down mortality within an income constraint (per capita incomes in these populations are in the range of US$ 3 00-400, i.e. one-fortieth that of Western Europe and one-fiftieth of North America) are literacy, especially female literacy, fertility control, and female autonomy. The second aggregate study to be mentioned is the study by Rogers and Wofford (1989). This study, reports Caldwell (1990, p.47), identified two main factors for reducing mortality in Less Developed Countries, namely literacy and the proportion of the population working outside agriculture, and one factor the safety of the water supply. Rogers and Wofford (1989) have indicated that health inputs measured by the ratio of physicians to population have lower level of correlation, while nutrition was not found to be significant. The last two studies mentioned by Caldwell (1990) 3 These populations (countries) are Sri Lanka, Kerala State in South West India, China, and Costa Rica. 24 are that for Ghana in the 1960s by Gaisie (1969) and that for Latin American countries in the 1970s by Behm (1983) . Both studies have shown remarkable differentials in the survival of children by the level of maternal education. Another aggregate study which indicates the importance of literacy, especially female literacy, in reducing infant mortality was done by Flegg (1982). Here, two major findings deserve to be mentioned. The level of literacy is the key factor in lowering infant mortality, though the degree of equality in income and the level of medical care also play major roles. Specifically, Flegg's study suggests that reducing illiteracy from 39% to 2% reduces infant mortality rate from 94 deaths per 1,000 live births to 43 deaths per 1,000 live births, a 50 percent reduction in infant mortality rate, while reducing inequality from 1.2 to 0.7 implies a decline by 20 deaths per 1,000 live births. Another finding suggests that education has the greatest effect on infant mortality in societies with a relatively egalitarian distribution of income. The importance of (maternal) literacy in shaping the mortality patterns is also reported in Palloni's (1981) study which analyzes Latin America data.4 From this study, three observations deserve to be mentioned at large. Firstly, the 4 Other socioeconomic factors are mentioned by the Palloni. Among them, he indicates income per capita (see Palloni, 1981, Table 4, p.642), and contextual and community factors (p.644). 25 study reveals that maternal literacy has a much greater effect on mortality in the age interval 1-5 than on infant mortality because child mortality reflects the very high influence of education on variation in death rates attributed to water- food-airborne diseases, especially diarrhea and influenza- pneumonia-bronchitis. According to Palloni (1981, p.641), about 90 percent of variance in death rates attributed to these water-food-airborne diseases is explained by education alone. Secondly, the study reveals that educational effect is strongest among countries with high levels of excess mortality at 0-1 and 1-5. This is explained, according to the author, by the predominance of the water-food-airborne diseases in pattern of cause of death in those countries. Thirdly, Palloni's work has opened a new dimension in the debate of the relevance of education in the determination of mortality by distinguishing aggregate level from individual level. At the aggregate level, suggests Palloni (1981), the illiterate rate in a society is less an indication of the proportion of mothers with adequate knowledge to treat and nourish a sick child or to defy the authority of elders than a reflection of the degree of social and political maturity of the system above and beyond the amounts of wealth at its disposal and the degree of equality of its distribution (Palloni, 1981, p.643) . At the aggregate level, therefore, the illiteracy rate reflects not only the limitations of individuals but, more importantly, the capacity of a system to organize and mobilize to fulfill societal necessities (Palloni, 1981, p.642). 26 As will be seen in section 1.2 below, at the individual level, mother's education influences the amount of knowledge available to her to take care of the health of infants and children by promoting awareness and use of available health care services. For example, Caldwell (1990) argues that an educated mother exerts more authority within the extended family and is more likely to seek modern health care for infants and children than to follow traditional norms and practices. The key role of maternal education on child mortality is also indicated in the study by Mosley (1985), which analyzes child survival from birth to two years of age in Kenya. This study shows that changes in mortality over time from 1962 to 1979 and between regions could largely be explained by differences in maternal education and family income, while formal medical services have less impact on mortality. In effect, evidence reveals that the decline in child mortality attributed to a change in education level of childbearing women is estimated at about 100% in 1962-69, about 78% in the period 1969-79, and about 86% of the decline between 1962-79. During the last period (1969-79), for example, the remaining 14% of the decline in child mortality can be attributed to improvement in the household economic situation. In 1979, for example, the numbers of children dying by age two per 1,000 births by level of mother's education were as follows: 163 for uneducated mothers, 104 for mothers with 1-7 years of 27 education, and 61 for mothers with 8 or more years of education. In this study Mosley states that the observations that maternal literacy and level of income are associated with survival while modern health services generally are not, "are not novel in Africa" (Mosley, 1983, p.115). Three previous studies have reached similar conclusions. First, the 1979 macro level study on 41 African countries by the Economic Commission for Africa revealed that [child mortality] is strongly associated with literacy and per capita GNP, and no effect of person per doctor or per bed was established. Furthermore Mosley cites the study by Gaise (1979), which has revealed that the decline in infant mortality in Ghana from 205 to 132 per thousand births between 1950 and 1970 "could not be attributed to health services which had scarcely reached the majority of population, but must be related to general improvement in the standard of living" (Mosley, 1983, p.115). Finally, Mosley (1985) mentions the study by Caldwell in Nigeria in 1979 which showed that maternal education was an independent determinant of urban-rural differentials in child mortality in Nigeria. Mother's education, concluded the study, "was probably not operating as a proxy for other social variables but actually was indicative of patterns of maternal behavior which have a direct bearing on the survival of the child" (Mosley, 1985, p.115). Several micro-level studies, mostly from developing countries, have indicated the existence of differentials in 28 child survival rates associated with mother's education. I will review these studies in the next section. 1.2. MICRO-LEVEL STUDIES ON EDUCATION-CHILDHOOD MORTALITY ASSOCIATION Until the mid-1970s, the analysis of the influence of maternal education on children's mortality in developing countries was neglected despite the existence of many studies which found a strong inverse relationship between education, particularly maternal education, and child mortality. In addition, education was frequently and falsely considered as a reflection of the standard of living, and consequently this relationship was given a simple economic explanation. The turning point is the study by Caldwell (1979) which used data set from the 1973 Nigerian segment of the changing African Family Project. This is a probability sample of 6,606 Yoruba women aged 15-49 years carried out in May-June 1973 in Ibadan. In this study, Caldwell discusses the meaning of maternal education in its influence on child survival and also discusses two questions of interest. First, he wants to know if maternal education is a proxy for general economic development or a reflection of socio-economic status generated primarily by other conditions, such as husband's income (Caldwell, 1979, p.398). Second, is the proportion of children dead per woman "a proxy for infant and child mortality"? (Ibid., p.398). 29 The findings of the study are clear. Child mortality (CM) index for women with no education was 1.287, and 0.870 for women with only primary school education, and 0.507 for women with secondary school. This is clear evidence that mother's education deflates infant and child mortality. As shown in Table 1.2.1. below, Caldwell computed separately the child mortality index for younger women (15-29 years of age) and older women (3 0-49 years of age), abbreviated as CMY and CMO, respectively. Table 1.2.1.: Child Mortality Indices by Education of Mother, Respondents 15-49, Ibadan City, Nigeria 1973 Mother's Education Odds Ratio Odds Ratio5 All Index NS PS SS Mothers PS/NS SS/NS PS/NS SS/NS CM CMY CMO 1.287 0.870 0.507 1.097 0.68 0.39 0.66 0.43 0.321 0.218 0.133 0.257 0.68 0.41 0.67 0.57 0.966 0.652 0.374 0.840 0.67 0.39 0.66 0.39 NS=No Schooling PS=Primary Schooling SS=Secondary Schooling Source: J.C. Caldwell (1979), op. cit., Table 2, p.401 Basically, the same conclusion appears in this table: women's education decreases infant and child mortality. Later, Caldwell analyzes the maternal education-childhood mortality association after controlling for other socioeconomic factors, namely the occupation of both mother and father, the urban- rural area and residential location within urban areas, 5 Ratio of indices corrected to account for the different fertility schedules. 30 whether the marriage is monogamous or polygynous, and whether the parents practiced family planning or not. Here again, the study reveals that differentials in child mortality found by mother's education do not appear as a mere reflection of other socioeconomic factors. These differentials are clearly meaningful in their own right. Woman's education appears to be more important to cut down childhood mortality than even her most immediate environment. From this study seven points merit to be emphasized (Caldwell, 1979, pp.402-408). (i) There is a multicollinearity between occupation of mother and her education due to the fact that "merely half of the women with secondary education are working in white-collar occupations" (p.402) ; (ii) income or family income works as an indicator of capacity to purchase health or foods. This capacity "is better measured by family income than by education or occupation of either parent or by the areas of residence" (p.402); (iii) there are interactions between mother's education and some socioeconomic factors. Specifically, four factors interact with mother's education: place of residence, husband's occupation, husband's education, and type of marriage in which the mother finds herself; (iv) the practice of family planning appears to be a measure of significant changes in family relationships, especially parent-child relationships; (v) the impact of the mother's education on childhood mortality is greater than that of the father; (vi) father's occupational status, especially a white-collar occupation, plays a greater 31 influence when mother's education is low. In any case father's occupation plays "a distinct and substantial impact” (p.407); (vii) mother's social mobility (measured by her father's occupational status) appears to be of little importance. However, the latter conclusion is not confirmed in the study in Sudan by Farah and Preston (1982), which in contrast has found that mother's social mobility as measured by her father's occupational status has great impact in her child mortality. Another study indicating the role of maternal education was published by Caldwell and McDonald (1981). This study reexamines the conclusions reached by Caldwell in 1979. In fact, Caldwell and McDonald (1981) have confirmed that parental education is the key factor of child survival, with mother's schooling usually having the greatest impact. For example, the analysis indicates that children with mothers with primary schooling to have mortality 20-50 percent lower than those with uneducated mothers, and for those with mothers with secondary and tertiary education to be 30-60 percent lower and 60-90 percent lower respectively (Ibid., p.85). After using a multiple classification analysis to examine the reduction in the probability of dying in the first 2 years of life, q(2), the authors indicate that parent's education has over three times the impact on child mortality as does the father's occupation, mother's education is somewhat more important than father's education. The authors conclude that "schooling makes parents 32 increasingly part of a new culture." So "when parents have been to school, they, and even their illiterate relatives, assume that they are moving into a different society" (Caldwell and McDonald (1981, p.93). For example, they believe in modern health facilities and in the theory of health and cure. Brief, "the major impact of schooling is almost certainly the new family system it brings about" (Ibid., p.93), a system in which children (and women) are awarded higher priorities in care and consumption than in the traditional system. Although the authors' study is of great value, three objections can be pointed out. First, the analysis is based only on one index of child mortality q(2) and not on q(5). A recent study by Tawiah (1989) has shown that the effect of maternal education on child mortality is larger than the correspondent effect on the mortality during the first two years of life but lower than the effects of paternal education on q(5). Secondly, Preston's (1982) study has shown that knowledge and health care provision have a great impact on child health. Therefore, although parent's education is necessary, it is not sufficient in reducing or eliminating rural-urban differentials in child health. Thirdly, as mentioned by Cleland and van Ginneken (1988), there are studies in Bangladesh and India that show that sex differentials in child mortality are not eliminated even among 33 better educated mothers.6 This is contrary to Caldwell's hypothesis that in the new family system resulting from the schooling food is likely to be shared more evenly...therefore, reducing the sex differences in mortality. In an analysis of data on 1281 women of childbearing ages (15-45) from a stratified random survey of Nicaragua in 1977- 78, Wolfe and Behrman (1982) investigate the determinants of child health, mortality and nutrition in developing countries. To reach their goals, the authors build upon economic models of household behavior. This study reveals that income or generalized purchasing power is not a major factor of child mortality, health, and nutrition in developing countries. These findings contradict the World Bank (1980), Ward and Sanders (1980), and other studies which report that generalized increase in income will improve health and nutrition status of children in developing countries.7 Rather, Wolfe and Behrman's (1982) study indicates that parental schooling, especially mother's education, is the major determinant of child mortality, health, and nutrition, showing, for example, that a one-year increase in woman's 6 See, exactly references 60-63. 7 Both studies are: 1) World Bank, 1980, World Development Report. 1980. Washington, D.C.; 2) J.O. Ward, and J.H. Sanders, 1980, "Nutritional Determinants and Migration in the Brazilian Northeast: A Case Study of Rural and Urban Ceara," Economic Development and Cultural Change. Vol.29, pp.141-163. These two studies are cited by Wolfe and Behrman (1982). 34 education significantly reduces the child probability of dying by 9 per cent. In meantime, increases in weight, height, and biceps, for combined sample, resulting from a one-year increase in woman's schooling, are 4%, 5%, and 0.3%, respectively. Further, the study indicates that woman's participation in informal sector reduces child mortality but its impact is not statistically significant. But woman's participation in informal sector is positively and significantly associated with child weight for urban areas, child height in other urban areas, and child biceps circumference in central metropolis. Finally, the study shows that for child mortality (combined sample) an additional child under five years in the household increases child mortality by 58 percent, the association is statistically significant. In contrast, households which have a sewage disposal have significantly lower child mortality (by 55%) than those which do not have one. Children born with low birth weight have 61% higher mortality than those born with normal weight. Although Wolfe and Behrman (1982) built a complicated model and used so many control variables, they confirm, however, that parental education, particularly mother's schooling, may be particularly important in the production of commodities necessary for consumption. Here, the underlying assumption is that more educated parents tend to be more knowledgeable than those who are less educated about preventive and curative health and nutrition measures, more capable of following medical and nutritional advice 35 or instructions, less fatalistic about illness and therefore more prone to seek medical help for a sick child, and more child-oriented due to the larger role mothers play in intrafamilial decisions (Wolfe and Behrman, 1982, p.168). In a re-analysis of the same data, Wolfe and Behrman (1987) examine again the role of mother's schooling in determining children's health and nutrition. In this paper, however, the authors investigate whether controlling for unobserved mother's childhood-related-background influences the effect of mother's schooling on children's health. They report that women's schooling is the only individual estimate that is significantly related to child health (infant and child mortality, child height, child weight, and child arm circumference) . In both cases (individual and means estimates) other variables such as father's education, marital status of the mother, and nutritional inputs are not statistically significant. For example, individual estimates show that a one-year increase in mother's schooling has the following impact in child health: it decreases infant and child mortality rate by 1.1, while it increases child height, weight, and arm circumference by 10.4, 7.3 and 8.8, respectively. However, this study shows that deviation estimates are in clear contrast with individual and mean estimates in the sense that they indicate that once interfamilial differences in mother's childhood-related-background are controlled, mother's education is no longer significantly associated with 36 child health. In deviate estimates, only the length of breastfeeding has a significant negative impact on standardized weight controlling for intrafamilial variation (i.e, one-month increase in the average month of breastfeeding deflates the child weight of 230 per cent). But father's schooling, mother's marital status, calories, and length of breastfeeding (except in the case noted) continue to have insignificant coefficient estimates. According to Wolfe and Behrman (1987, p.247-252), the results exhibited by deviation estimates imply that the positive association between mother's schooling and child health often observed in standard estimates (i.e., individual and means estimates) is not primarily due to the effect of schooling per se; instead mother schooling is serving largely as a proxy for unobserved characteristics related to her childhood background such as number of siblings, whether her mother and father were present in her adolescence, whether her childhood was in urban area, whether she migrated, and her age at marriage. In other words, the authors' study implies that, in standard individual child health estimates as well as mean estimates, the apparent strong positive effect of mother's education reflects the failure to control for mother's childhood background related abilities, motivation, knowledge and tastes. Another individual-level study showing the importance of education in childhood survival was presented by Farah and Preston (1982). Two issues are addressed by the authors: one 37 is the role of parental, particularly maternal, education in reducing infant and child mortality; another issue is whether the regional differentials in child mortality in the Sudan are a result of differences in the distribution of household's socioeconomic attributes among the regions or these differentials persist independent of this distribution (Farah and Preston, 1982, p.368). The study is based on two bodies of data: One is the 1973 census; the second is an investigations of Khartoum conducted by the changing African Family Program. From the bivariate analysis, Farah and Preston (1982)'s study reveals that for all ages the ratio of aggregate mortality for mothers with no schooling to those with 7 years is greater than 2:1; and this mortality invariably declines as education is increased. Table 1.2.2. shows these relations: Table 1.2.2.: Probabilities of Dying by Age 1 through 5 by Maternal Years of Schooling, Sudan, 1973. q(a)a Maternal years of schooling None 1-3 4-6 7+ 1) 0.1911 0.1344 0.0970 0.0700 q(2) 0.2001 0.1500 0.1088 0.0811 q(3) 0.2090 0.1598 0.1194 0.0904 q(5) 0.2181 0.1698 0.1346 0.1079 Source: Farah and Preston, 1975, Table 2, p.369. aq(a) is estimated by applying Trussel's (1975) North model to proportion dead by age of mother and years of schooling. In addition, bivariate relations indicate that the effect of education is maintained even after controlling for urban- rural residence and paternal education. And the improvement in 38 child mortality associated with an additional year of parental education is much greater for mothers than for fathers. Finally, father schooling is more closely associated with household income than mother's education. Multivariate analysis also shows that mother's education has a large effect on mortality. In effect, each additional year of mother's schooling is associated with a 3.63 percent reduction in the proportion of children dead, meaning that five years of maternal schooling is expected to reduce child mortality by 18 percent. Father's education is also shown significant but its impact is about one-third as low as that of the mother. (In effect, each additional year of father's education reduces child mortality by 0.0115 so that five year of paternal schooling is expected to reduce child mortality by about 6 percent). Further, Farah and Preston (1982) report that children whose fathers are employers and own-account workers experience lower mortality,8 while children of working women had higher mortality than children of housewives. About this latter finding, Farah and Preston (1982, p.372) provide three explanations. Firstly, they mention that "children of working women are more deprived of adequate child care (direct effect)." Secondly, "women's work is an indicator of economic stress in the household, since full-time childbearing is a strongly sanctioned normative 8 The authors warn the readers to adopt some caution for this particular finding because of the quality of census data on occupation. 39 status of mothers in Sudan (proxy variable)." Thirdly, "women who have lost a child are freed from child care responsibility and therefore more likely to work (reverse causation)." Another interesting result obtained from this study is that there is large regional child mortality differentials in Sudan. It is shown, for example, that "being located in one of the three southern regions raises child mortality rates compared to those in Khartoum by 48-71 percent" (Farah and Preston, 1982, p.373). These differences remain even after controlling for father's and mother's education and employment status, type of marriage, housing structure, and so on... For Farah and Preston (1982), these regional differences in child mortality reflects differences in disease environment, particularly the higher endemicity of malaria in the South, and the general level of development, including the health system (Farah and Preston, 1982, p.373). According to both authors, this explanation has been also reported in two studies in Africa: one in Kenya by Anker and Knowlers (1971) and another in Tanzania by Hogen and Jiwani (n.d). Both studies found that levels of regional malaria endemicity had very strong effects on childhood mortality levels, quite apart from levels of household or community socioeconomic variables (Farah and Preston, 1982, p.374). In Khartoum, women's education again has a strong and persistent effect on child survival. The study shows that in this city the effect of mother's schooling maintains an independent role as well. For example, the impact of one 40 additional year of schooling is shown to reduce child mortality by 3.62 percent, meaning that controlling for all other variables achieving 5 years of schooling (by the mother) reduces child mortality by 18.1 percent. However, in contrast to the claim made by Caldwell (1979) that the impact of mother's schooling in Ibadan operates to deflate child mortality through the rise of the mother's status in the household, Farah and Preston (1982) show that in Khartoum the effect of mother's schooling seems to work through other paths to reduce child mortality, namely through husband's education, blood relation between spouses, and type of residence in the early years of marriage (nuclear family residence or extended family residence). In effect, the regression coefficients for maternal education in Khartoum passes from -0.0489 to -0.0329 when characteristics of the husband/marriage are introduced. This represents 32.7% reduction in the ability of maternal education to deflate childhood mortality. Another striking result for Khartoum is that there is an inter-generational effect on child mortality. That is, the social class origin of the mother influences the survival of her children. But the effects of these intergenerational factors diminish when other socioeconomic variables are controlled, suggesting that "class of origin is operating almost exclusively through other variables present in the model" (Ibid., p.3 75). 41 The results obtained by Farah and Preston relative to maternal and paternal's education are consistent with the average levels found in Cochrane et al. (1982)'s review of literature that examine the separate effects of father's and mothers's schooling. In another paper, Preston (1985) uses United States census data and today's data from underdeveloped countries and shows among other things that child mortality advantage for literate mothers relative to those of illiterate mothers in the U.S. was 30% in the period 1890-1900, while today this advantage is, on average, 43-47% in developing countries. This dissimilarity is strange when one considers two facts. First, the Unites States in the late 19th century was a relatively rich country. For example, between 1890 and 1900 the per capita GNP (in 1929 U.S. dollars) rose from $415 to $497. Converted into 1982 dollars, the range is from $2,148 to $2,572. But the life expectancy at birth was only of 49-50 years. The second fact is that the U.S. population in 1890- 1900 was relatively well educated. For example, 78-79% of youths aged 5-17 were enrolled in school, 87.7% of ever- married women were literate, as were 89.0% of their husbands. To the question "why did not these high levels of literacy and income translate into high levels of life expectancy?" (Preston, 1985, p.374), the author responds that "this failure reflects a substantial ignorance about health matters,..." (Preston, 1985, p.383). He adds that this ignorance "reflects 42 the widespread failure of urban residents to activate the political institutions that were capable, then as now, of sharply reducing mortality" (Ibid., p.383). And this ignorance includes both "ignorance about personal health and ignorance about what public institutions could accomplish in the area of health" (Ibid., p.383). According to the author, the fact that education has a much larger health payoff in the developing countries today than it had in the United States in 1900, helps shed light on the mechanisms through which this variable is operating today. It suggests that health knowledge itself (perhaps more of hygienic practices than of disease causality, as Lindenbaum (1983) suggests) is one of the most important routes through which education is operating. In a study using Sri Lankan and Korean data, Trussel and Preston (1982) indicate that several studies on the impacts of schooling on child morality have overestimated the effects of mother's (and possibly father's) education since most of these studies have used age of mother (instead of marital duration) as the basic indexing variable of children's exposure to the risk of mortality. They also indicate that mother's education and father's education are associated to about the same degree with child mortality in the two data considered in this study. For example, in South Korea, the completion of high school by the father reduces child mortality by 47.2 percent; for the mother the correspondent figure is 37.4 percent. In Sri Lanka, child mortality is reduced by 50-55 percent when the mother or 43 the father has completed 10 years of education. The effect of mother's education on infant and child mortality was also estimated in the analysis of World Fertility Survey (WFS) data made by Hobcraft et al. (1984, 1985). After these authors have examined data from 24 countries in detail, they concluded that "mother's education does have an important association with infant and child mortality, especially the latter, but not in all countries" (Hobcraft et al., 1984, p.221). In addition, they have shown that in 15 countries the best three-factor model (in term of mortality-deflating-effect) is that which contains mother's education, husband's occupation, and husband's education. Although Hobcraft et al. (1984, 1985) indicate clearly a strong association between maternal education and infant/child mortality, Cleland and van Ginneken (1988), based on the same WFS data, indicate that about half of the gross effect of mother's education can be attributed to economic advantage. In a remarkable review of the literature on the association between maternal education and child mortality, Cochrane et al. (1982) distinguished bivariate analysis, which measures the gross effect of mother's education, from multivariate analysis, which determines how much mother's schooling is capturing the effect of other highly correlated variables. The authors found that regardless of measurement technique chosen, the bivariate analyses indicate that the association between maternal education and child mortality is 44 uniform, indicating that an additional year of schooling cuts down infant and child mortality by nine per 1,000. Multivariate studies suggest that part of the effect of mother's education revealed in bivariate studies results from the fact that better-educated women tend to be married to better-educated husband. But husband's educational effect is about half the effect of the mother's, i.e., -3 and -6 per thousand, respectively. The impact of education on infant and child mortality was also found in Trussel and Hammerslough (1983). In that study, like in the work by Trussel and Preston (1982), mother's and father's education were found to have the biggest differential impact. But, unlike other studies such as Cochrane (1980), the impact of father's education was found substantially higher than that for mother's education. In effect, children of mothers with no education are 1.5 times as likely to die at every age when compared with children of mothers with 10 + years, while children of uneducated fathers have 2.2 times as likely to die at every age when compared with children of fathers with 10+ years. In addition, the study reveals that residents of urban areas and estates have death rates among children 14 percent and 45 percent higher, respectively. This means that living in a rural area seems to provide the healthiest environment, while living on an estate provides the least healthy. Further, the study indicates that children born to mothers less than 20 years of age and to women older than 45 35 years of age are 29 percent and 21 percent respectively higher than for children born to women 2 0-34 years of age. Recently, Bicego and Boerma (1993) have published a paper on maternal education and child survival in which a uniform analytical methodology was applied to Demographic and Health Survey (DHS) data from 17 developing countries. The aim of this study is to address a series of six issues regarding the positive association between maternal education and the health and survival of children under age two. Firstly, this study reveals that mother's education advantage in child survival is less pronounced in neonatal period where a substantial part of the education advantage is explained by education's link to economic status of the household. These results are congruent with those of Hobcraft et al. (1984). They are also congruent with the hypothesis that personal resources (including maternal education) condition key risk-modifying behaviors to a greater extent as the child ages and thus as household-based decisions regarding susceptibility and exposure to disease become more and more important in delineating mortality risk. Secondly, the study indicates that in post-neonatal period, mortality risk is roughly twice as sensitive to the effects of maternal education as neonatal risk, after controlling for economic conditions of the household. Here, education advantage appears more pronounced in urban settings as compared to rural settings. Bicego and Boerma (1993, p.1224), interpret this as an indication "... that, rather 46 than physical access to modern health services, it is access to broader social and economic support systems that is limiting in the urban context." Thirdly, the authors' study reveals that the association between stunting and maternal education tends to decrease when the economic status of the household is controlled, suggesting that the relationship between maternal education and stunting reflects more the differentials in economic status than the differentials in education. In the meantime, the association between maternal education and underweight status is more pronounced than for the education-stunting relationship, and is roughly of the same magnitude as was estimated in the post- neonatal analysis. The study indicates that undernutrition (due heavily not solely to economic conditions) is the presumable causal pathway through which mother's education exerts an impact on the survival prospects of her children during the post-neonatal period. The fourth issue which emerges from this study is that education advantage in the comparison of rural areas with urban areas is more pronounced in the rural setting. This is consistent with Rozenzweig and Schultz's (1982) hypothesis that "where severe obstacles of access exist, education is key in placing a woman at a relative advantage in terms of securing services."9 The fifth issues is that non-use of antenatal care is 9 Cited by Bicego and Boerma (1993, p.1225). 47 found higher among the uneducated women than women with secondary education, even after controlling for economic status of the household and the pattern of family formation. The sixth point emerging from the study is consistent with previous findings such as Hobcraft et al. (1984, 1985), Cleland and van Ginneken (1988), etc. That is, the pattern of family formation does little to explain the education advantage in child survival. The same result is found in the case of growth faltering. Another evidence of positive relationship between maternal education and child survival is reported by Tawiah (1989) , using WFS data from Ghana. According to the author, of the six socio-economic factors included in the analysis, mother's and father's education have the largest effect in lowering child mortality in Ghana, followed by husband's occupation and mother's occupation, in that order. In the first two crucial years of life, the effect of mother's education appears to be greater than that of husband's schooling. To that effect, this result is consistent with the findings of Caldwell and McDonald (1981) and Martin et al. (1983). The latter, for example, reported that mother's education had a greater effect in Pakistan and the Philippines while father's education was slightly more important in Indonesia. In addition, the study shows that mortality differentials by mother's education are larger before the second birthday than for the g(5) estimates. For father's 48 education, child mortality differences are lower for "infants" (less than 2 years) and larger for childhood (less than 5 years). In effect, the survival chances in the first 2 years of life are 80% higher for children whose mothers have 11- 22 years of education than for those with uneducated mothers; they are 34% higher for children born to mothers whose husbands have 1-6 years of education than for their counterparts born to mothers with uneducated husbands (Tawiah, 1989, p.354) . Unexpectedly, the study shows that children born to mothers in the 1-6 and 7-10 years of education groups experienced higher mortality rates before the second birthday than children born to uneducated mothers. To explain the effect of mother's education on child survival in the first two years of life, the author (p.354) reports that "an educated mother may be free from traditional constraints. She is more likely to recognize the symptoms and early signs of severe childhood diseases and take appropriate measures to combat them." For the author, beyond the first two years of life the greater effect of husband's than mother's education on child survival might be "a reflection of the effect of standard of living on child mortality" because, mentions Tawiah (1989, p.354), husbands with 11-22 years of education are more likely to be well-paid and have higher standards of living which can protect their children against the principal causes of death, e.g. measles and nutritional deficiency. Tawiah's study also indicates a large rural-urban differentials which "may partly be explained in terms of 49 access to and availability of health care facilities” or may reflect the "educational differences between the rural area and the city" since 75.6% of the uneducated women lived in the rural areas, while 13.5% and 10.9% lived in town and city respectively (Ibid., p.354). The association between education and infant and child mortality is also examined by Pant (1991), who used the data taken from the 1986 Nepal Fertility and Family Planning Survey. The author finds a significant lower mortality for children born to mothers or fathers with some years of schooling, in comparison to those born to mothers or father with no schooling. In addition, the study indicates that "the infant and child mortality rates in Nepal were significantly lower among the children who belonged to households which had access to toilets or electricity" (Ibid., p. 440). In this matter the study is consistent with other studies10 such as DaVanzo, Butz and Habicht (1983), Khan (1988), Jain (1985), and Meegama (1980), etc., where households characteristics (such as source of drinking water, access to toilet and electricity, use of tap or hand pump, etc.) were found important in lowering infant and child mortality. Neither mother's nor father's education is shown to be significant in households which did not have access to electricity or a toilet, suggesting that the impact of schooling on infant and child mortality emerges only if the 10 All these studies are cited by Pant (1991). 50 household has access to resources that can be converted into services. Thus, both education and resources are complementary to each other in the sense that access to both of these may improve child survival in Nepal. Recently O'Toole and Wright (1991) have examined the relationship of parental education on child mortality in Burundi, using the Cox's proportional hazard rate on the 1987 Demographic and Health Survey data set. The results of the study indicate that parental education is a key factor in lowering child mortality, with maternal education, particularly her secondary education, being stronger than paternal education, after all other factors have been controlled. No interaction between mother's and father's education is found statistically significant, suggesting that both maternal and paternal education operate independently of each other with respect to their combined effect on child mortality. With respect to other factors, this study found that father's occupation is not a major factor of child mortality, while mothers who live in urban areas are found experiencing a much lower child mortality than their counterparts who live in rural areas, after controlling for education. This particular result contradicts Caldwell's (1979) findings which have indicated that, controlling for parents' education, the rural-urban differential is minimized in Nigeria. Further, significantly clear regional differentials in child mortality emerge from these data 51 related to Burundi, denoting a regional variation in morbidity or disease prevalence. This latter result is consistent with Coosemans' (1985) and Farah and Preston's (1982) studies, respectively.11 Discussing their own results, O'Toole and Wright (1991) raise a number of important issues such as the issue of selectivity (i.e., are women who attended school selectively different from those who did not?) and the issue of clear mechanisms through which maternal education affect child survival, etc. Another recent evidence demonstrating the association between education and child mortality comes from Majumder and Islam (1993). This study used data from the 1989 Bangladesh Fertility Survey. The paper reveals that both wife's and husband's education result in improved child survival in Bangladesh. It also reveals that wife's education has a greater impact on child survival in Bangladesh than that of husband's education, meaning that "the influence of father's education is largely through his socio-economic status" (Majumder and Islam, 1993, p.317), while the influence of mother's education is through "knowledge about effective ways to prevent, recognise and treat childhood diseases" (Majumder and Islam, 1993, p.317). A number of other variables have been shown to be significantly associated with child survival in Bangladesh, particularly the respondent's working status 11 Coosemans (1985)'s study is cited by O'Toole and Wright (1991), p.260. 52 (negative impacts) and household electricity (positive impact). These findings, report the authors, correspond with earlier results by Frankenberg and Martin (1988) . According to Majumder and Islam (1993, p.317), the latter results can be interpreted to mean that "respondents having electricity disproportionately belong to better income or economic status families", while "respondents working in paid jobs belong disproportionately to the lower strata of the society, so the economic welfare that they bring to their families is at the cost of their children's health." However, the study reveals that a number of other variables have no impact on child survival, particularly household water supply and toilet facility, place of residence, religion, etc. According to the authors all these variables, except religion and pregnancy wastage, are captured by both the respondent's and her husband's educational status since they are significantly correlated with these latter. In a prospective and population-based study of 6011 children born in the three maternity hospitals of Pelotas (Brazil) during the 1982 study, Victora et al. (1992) examine the impact of maternal education in relation to a wide array of child health outcomes such as birthweight, perinatal and infant mortality, hospital admissions and nutritional status. Before adjusting for confounding effects, maternal education is found to be associated with perinatal and infant mortality, hospital admissions in the first 20 months of life and the 53 three nutritional indicators. After adjustment for confounding effects (age, income, race, and height), four patterns of the association appear. First, in perinatal period, birthweight and perinatal mortality are no longer associated with maternal education after adjusting for confounders. Second, concerning infant mortality, the odds ratios are reduced by approximately 50% after adjusting for confounders, though the education- mortality association is still significant. That is, "infants of uneducated mothers were almost three times more likely to die than those of mothers with 9 or more years of schooling" (Victora, et al., 1992, p.904). Third, the association between maternal education and hospital admissions, despite also being reduced by approximately 50% after adjustment for confounding variables... remained significant. Four, the association between maternal schooling and the three nutritional indices (length-for-age, weight-for-age, and weight-for-length) are reduced by 50% to 75% after controlling for other socioeconomic variables, but they remained significant. Victoria et al. (1992) also shows that controlling for family income only greatly decreases the effect of maternal education on birth weight, perinatal and infant mortality, but does not affect the strength and significance of the relationship between education and hospital admissions and nutritional status. Thus, the study supports the thesis that maternal education has an effect on child health which is partly independent from that of other socioeconomic factors. 54 It also suggests that maternal care, a social/behavioral factor, is more important than the biological factors of the mother since stronger impacts are observed for the late (postneonatal mortality, hospital admissions and nutritional status) than for the early (birthweight, perinatal mortality) outcomes. These findings confirm the results obtained by Puffer and Serrano (1973) and United Nations (1982) showing that post-neonatal mortality is influenced, in general, by exogenous factors while neonatal/perinatal mortality is more influenced by endogenous factors. The impact of maternal education on childhood survival is also mentioned in the study by Ahmad, Eberstein, and Sly (1991) based on the 1986 Demographic and Health Survey (DHS) of Liberia in which 5,239 women aged 15-49 were interviewed. Three findings about this impact are reported. First, higher maternal education is associated with a reduction in tendency to breastfeed (direct effect is -0.119). Second, education has both a positive direct effect on child survival, through prenatal care (0.261), and a negative indirect effect on child survival, through a reduced tendency to breastfeed. Third, the direct positive impact of education on child survival (0.167) is much smaller than the direct impact of breastfeeding (0.544) on child survival, indicating that breastfeeding is the single most important factor of child survival. Consistent with other studies discussed above (such as Palloni, 1981; Cochrane et al., 1982; Caldwell, 1979), maternal education is 55 found positively associated to child survival for a number of reasons. It is believed to provide the mother with the necessary skills for child care, to change the traditional balance of familial relationship with profound effects on a child, to determine changes in social attitudes as well as the readiness to accept new ideas, and to increase her awareness and attitudes to health, etc. Along these lines of reasoning, this study indicates that maternal education has a positive direct effect on prenatal care (0.261). Further, these findings are congruent with studies (such as Casterline et al., 1989; Goldberg et al., 1984; Kelly et al., 1982) which have indicated that maternal education has a poor or weak explanatory power of child survival once other factors such as income, housing and use of health services are controlled for. In addition, this study agrees with Farah and Preston (1982), etc., who have indicated that mother's education may have negative impact on child survival in developing countries because mothers in the middle and higher social classes (probably all educated mothers) are more likely to abandon traditional behavior and prolonged breast-feeding, to resort to artificial milk based hazardous products, and are more likely to resort to a baby-sitter for daytime child care, etc. All of these may increase the child's likelihood to catch infectious diseases or to experience injuries. However, the study by Ahmad et al. (1991) reveals that older mothers have lower risk of child mortality than younger ones for four 56 reasons. (1) As the age increases there is a cumulative increase in maternal experience; (2) older mothers have children who are old enough to assist her in the care of younger siblings; (3) older women have the tendency to observe traditional practices such as postpartum abstinence and prolonged breast-feeding; (4) older mothers in this study have been found more likely to breast-feed than the younger mothers. These latter findings are totally in contradiction with those of Hobcraft et al. (1983, 1984, 1985) which have reported that children born to very young or very old mothers experience excess mortality compared to those born to mothers in the intermediate age groups. This is so because the effects of younger age on child mortality is believed to pass through poor nutrition and immature maternal reproduction development factors, for very young mothers, while the effects of older age is believed to pass through maternal depletion syndrome and weakness of the maternal reproduction tissues. Having reviewed the literature on the maternal education- childhood mortality relationship, the next section of this chapter will discuss the health transition theory, which is designed to increase the understanding of this association. 1.3. A THEORY OF HEALTH TRANSITION This section aims at presenting the theoretical background which explains the impact of maternal education on 57 childhood mortality. Evidence has indicated that over the last 100 years, today's developed countries have witnessed an unprecedented increase in their life expectancy "by more than 50 percent from under 50 years to around 75" (Caldwell, 1990, p.45). This mortality decline is basically explained by the rise in real incomes of population induced by the Industrial Revolution. These material improvements and scientific advances had real effects on the life of people by allowing them to be better nourished and clothed; by allowing the construction of improved hospitals that the masses could afford to use; by providing the resources to treat drinkable water and to build sewage disposal system; by making possible safer medical procedures and by producing sulfa drugs, antibiotics, new vaccines and powerful insecticides (Caldwell, 1990). In today's developing countries, analyses of existing data have indicated that both at the level of national or other large aggregates and at individual level, "levels of income and health services are weak predictors of mortality levels and that social determinants apparently play a major role in determining mortality" (Caldwell, 1990, p.46). Here, it is noteworthy to emphasize that though individual-level studies in Third World have suggested the predominant role of social and cultural determinants in explaining the mortality declines, evidence has also suggested that alone, these determinants could not induce dramatic reductions in mortality 58 because these reductions are "the product of an interaction between certain cultural and social characteristics on the one hand and the easy accessibility of basic modern health services on the other" Caldwell (1990, p.51). Two examples can be mentioned here. In the case of Sri-Lanka, the provision of modern health services, measured more by its accessibility to a wide population, coupled with the social achievements, allowed the subsequent dramatic increase in life expectancy. Another example is given by Orubuloye and Caldwell (1975)'s study of two socio-economically similar populations in Nigeria, one with access to a hospital and doctors and the other isolated from such interventions. This study has showed that the gain in life expectancy was 20 percent when the sole intervention was easy access to adequate health facilities for illiterate mothers, 33 percent when it was education without health facilities and 87 percent with both (Caldwell, 1990, p.52). As indicated by Cleland and van Ginneken (1988) a wide range of studies in the Third World societies have suggested that the use of modern health services increases with education and that income also interacts with the provision in health service. All of this evidence makes it difficult to divide the "pure" effect of mother's schooling between modern medicine and behavioral and care factors that prevent children from catching sickness or having an accident in the first place. To interpret all of this evidence, Caldwell (1993, p.125) 59 has formulated a theory of health transition in which he discusses "the cultural, social and behavioural determinants of health (i.e. those determinants other than medical intervention and material standard of living)." Five propositions are included in this theory (Caldwell, 1990): Proposition I; There have always been socioeconomic differentials in mortality levels and that they predated the impact of modern medicine (Caldwell, 1990, p.53). Two major forces have always been at work to explain these differentials. The first force is income and the ability to eat better and enjoy other material comforts. Formulated by Malthus, the rationale of this explanation is that "ultimate constraints on mortality decline are those of material resources and hence of economic living standards" (Caldwell, 1986, p.171). The second force at work is social factors (ethnicity, etc.). Proposition II: A substantial part, probably the majority, of the explanation for social differentials in mortality in the contemporary Third World lies in the interaction with modern medicine. Here, Caldwell (1990, p.53) states that the breakthrough periods in reducing mortality levels in different Third World countries have been associated with the democratization of services not with an increase in the quality of medical technology. During these breakthrough periods, reports Caldwell (1990), the informal system of health services (made up of non- traditional untrained practitioners or charlatans, drugs 60 distributors, etc.) helps to change beliefs and practices with regard to illness and acts as a referral system to a more formal health sector. Let me agree with (Caldwell, 1990, p.53) that in most of the part of rural sub-Saharan Africa, even in urban areas inhabited by the poor, "informal sector is the only channel of modern medicine..." Proposition III: The various social mechanisms identified as playing a role in reducing mortality are really different facets of the same phenomenon, which might be called social modernization, or the rise of individualism or westernization (Caldwell, 1990, p.54) . According to Caldwell (1990), westernization is (1) the social counterpart of the transition from subsistence-production organization to the market economy; (2) the move toward a system where individuals have options and can make choices; (3) the dismantling of the subsistence-production organization and the dismantling of belief systems that were necessary to ensure its survival. In terms of demographic behavior, the dismantling of belief systems is very important because it helps people to realize that sickness and death are the result of non-divine and non-magical forces of the world. Indeed, it helps people to realize that something can be done for sickness, either in the form of careful behavior or seeking the best help. This process is called the secularization of health behavior (Caldwell, Reddy, and Caldwell, 1983). In the process of westernization, the impact of maternal 61 education on child survival is dominant. It occurs everywhere, in good as well as in poor schools. In addition, it helps create a sort of middle-class culture (or model) where educated mothers are expected to observe a certain level of domestic hygiene (Lindenbaum et al., 1985; and Lindenbaum, 1989). Further, as indicated by Preston (1985), maternal education produces much greater differentials in child mortality in today's Third World than it did one century ago in the West. Two reasons are suggested by Caldwell (1986, 1990) to explain Preston (1985)'s finding. First, modern medicine in 19th century had relatively little to offer. Second, Western education's message now is much closer to Western behavioral pattern of today's time while its message in 19th century was closer to Western behavioral pattern of that time. Thus, the impact of education on changing attitudes and behavior with health implications during the 19th century was much less than in today's Third World (Caldwell, 1990). Finally, the effects of mother's education (along with father's education impact) remains even after controlling for economic advantages (income, father's occupation, etc.). Proposition IV: Child mortality will fall more rapidly as the intergenerational wealth flow turns downward from parents to children (Caldwell, 1990, 56) . In terms of health transition, because of the strengthening of intrafamilial emotion and resource-allocation priorities, this 62 downflow trend in wealth leads parents to spend more effort and a greater proportion of income on child care and health, to place more emphasis on children relative to the elderly and to plan for their future, etc. However, two major obstacles may prevent social and family changes to induce effectively the health transition. The first is the lack of spousal emotional and economic bond. In Sub-Saharan Africa, for example, because of the lineage system, which gives predominance to extended-family residence rather than nuclear-family residence, and partly because of widespread polygamy, husbands and wives entertain a very limited emotional and economic bond. This worry can be attenuated because recent evidence by Omideyi (1993), using Yoruba data, reveals that changes are beginning to appear in household composition in Sub-Saharan Africa. There is an emergence of nuclear family which is having an effect on fertility and women's roles in the management of household resources. According to Omideyi (1993), nuclearization in Nigeria is the result of increasing modernization acting through the increasing value given to education of young children by parents. The second obstacle is the strong cultural tradition, still present in many Third World countries, which consists of limiting women's autonomy. This practice has a detrimental effect on child health "because of the limitation in mother's taking quick and effective action" (Caldwell, 1990, p.57). 63 Proposition V: Cultural, social, and behavioral factors have an impact both on an individual's mortality and on the mortality of an individual's dependents (Caldwell, 1990, p.57). Here, Caldwell (1990) discusses how behavior, especially mother's behavior can, in the Third World societies, be translated into lower child mortality. The basic argument implied in this proposition is that greater female autonomy or education increases her capacity in health management. (Health management is defined as "behavior that prevents sickness from occurring or limits the damage once it does occur" (Caldwell, 1990, p.58)). This proposition also implies that behavioral changes in the Third World affect not only mortality and fertility levels but the structure of society and all social relations. What the health transition theory described above is telling us is that increased maternal education reduces child mortality because education gives women the power and the confidence to take decision-making into their hands (Caldwell, 1979, 1981, 1986, 1990; Caldwell and McDonald, 1981; Farah and Preston, 1982; Preston, 1985; Wares, 1984; Schultz, 1984). On the question why maternal education has an influence on childhood health, this theory roughly suggests five reasons. First that schooling makes educated mothers break with tradition, or become less fatalistic about illness, and adopt many of the alternatives in child care and treatment of illness and treatment of illness becoming available in a rapidly changing society, thus profoundly influencing their 64 children's chances of survival (Caldwell, 1981, p.75). The second reason is that schooling may help women to be more capable of maneuvering the modern world, to have greater respect for modern bureaucratic authority, and to enhance self-esteem in increasing, for example, competence in dealing with bureaucratic organization (Caldwell, 1979, 1981; Cleland, 1989) . In health transition theory, this implies that mother's schooling increases "her knowledge about effective ways to prevent, recognize, treat childhood illnesses" (Cleland, 1989, p.411); it enhances her use of modern health services and modern remedies, and her skills in health care practices (Cleland and van Ginneken, 1988; Cleland, 1989; Mosley and Chen, 1984). This explanation also implies that an educated mother is more likely to be listened to by doctors and nurses...; she is more likely to know where the proper facilities are and to regard them as part of her world, and their use as a right, not a boon (Caldwell, 1981, p.75). In addition, an educated mother is "more exposed and sympathetic to, or more able to understand, health message disseminated by mass media or by less formal means" (Cleland, 1989, p.411). The third explanation is that education of women greatly alters the traditional nature of spousal relationships, with profound effects on child care, fertility (through change in family formation pattern), and on nutritional practices 65 (Caldwell, 1979, 1981; Cochrane et al., 1982; Ware, 1984). The fourth explanation, known as the Lindenbaum hypothesis, is that schooling has a positive impact on mother's ability for domestic hygiene, preventing her children to be exposed to repeated infections and eventual risk of death (Lindenbaum et al., 1985; Lindenbaum, 1989). Another explanation, called Levine's hypothesis states that maternal schooling changes positively the nature of mother-child relationship (LeVine, 1980; LeVine et al., 1991; Cleland, 1989). What Levine's thesis suggests is that not only are the emotional and verbal links between spouses fortified, but so do the emotional, verbal and visual interactions between the child and her mother, although the style of child- rearing in modern context entails less physical contact between the child and mother. In light of the review of literature and of the theory of health transition discussed above, Chart no. 2.4.1. (see Chapter 2) gives the pathways through which the effect of maternal education influences childhood health. 66 Chapter 2 THE PROPOSED STUDY: MATERNAL EDUCATION AND CHILDHOOD MORTALITY IN URBAN AREAS OF ZAIRE. FACTORS AND DETERMINANTS, AND PATHWAYS OF EDUCATIONAL INFLUENCE This chapter presents the proposed study and is divided into five sections. Section 2.1. discusses the research questions. Section 2.2. discusses the objectives of the research while Section 2.3. presents the hypothesis to be tested in this dissertation. The theoretical framework and the methodologies of the research will be discussed in Section 2.4. and 2.5., respectively. 2.1. THE PROPOSED QUESTIONS OF STUDY Along with the theory of health transition described in chapter 1, the central question of this dissertation is how attending school during childhood, in adolescence, or later, can affect a woman's child care behaviors so as to reduce infant and child mortality. As the behavior of a person is determined by his/her knowledge, beliefs, and resources, it goes without saying that education is an important force because it has the ability to increase family earnings through its positive effect on market productivity (Grosse and Auffrey, 1989); and because it can also increase knowledge and modify the belief system of people (Ware, 1984). In this research the issue is to what extent are knowledge and 67 resources gained by education converted into lower child morbidity and mortality. Three other questions will be of interest in this research. First, what are the socioeconomic, demographic, and cultural factors, and their mechanism of action, which influence infant and child mortality in urban areas in Zaire? Second, in order to set a social policy aiming at the reduction of infant and child mortality in the households and, given the reality that family incomes are low and shrinking, does the parents' education— especially the mother's education— affect really and negatively this mortality? If so (1) throughout what mechanism(s) does mother's education affect this mortality? (2) is (are) there other factor(s) which might play the same role as mother's education during this period? (3) what is the meaning of education in the household? Does it reflect the standard of living of the household? Or does it reflect the "degree" of opening of the household to ideas related to hygiene, to beliefs, attitudes and behaviors regarding child health and other issues of life? Schematically, the second question can be formulated as follows. Let us consider two families and F2 with corresponding levels of maternal education N and N+K respectively. Can it be established that family Fj^ will experience in its lifetime higher risk of child death than family F2? If this can be established, is it because that household F-^ has a lower level of maternal education than 68 family F2? The last question that this research will attempt to address can be formulated as follows. Should infant and child mortality be studied in the context of human nature with its biological or physiological composition which is insufficiently "adapted" to permit children to support the environment pressure1 or do we have to consider this mortality in terms of social environment insufficiently organized and, therefore, incapable of keeping alive all human beings born into its setting. 2.2. THE OBJECTIVES OF THE RESEARCH Two main objectives are pursued in this dissertation. First, I seek to examine the impact of maternal schooling on infant and child mortality in urban areas of Zaire. Along with this objective, I seek to assess the relative contribution of economic, demographic, biological, and environmental factors in explaining the education-child survival relationship. Further, along the first objective, I seek to identify the characteristics of families with high risk of child mortality. The second objective of this dissertation is to investigate the pathways through which maternal education exerts an impact on child survival. Following this second objective, I will examine a certain number of issues discussed 1 'Environment' is understood in a broad sense. 69 in the health transition theory in relation to child survival in the Third World. For example, can we find evidence that educated mothers compared to uneducated are most likely to be better nourished during pregnancy?. If yes, is this because of education or is this a reflection of the economic status? In addition, can we find evidence that education increases mothers' awareness about effective ways to prevent, recognize, and treat childhood ailments? Finally, does maternal schooling have an impact on child nutritional status? If the latter issue is true, what is the role of economic and other factors on education-child nutrition relationship?. In light of my assessment, I will attempt to suggest a social policy aimed at reducing child mortality, promoting health-seeking behaviors among parents/mothers in order to protect children, and reducing fertility. After having presented the problem of my research and defined the objectives that I want to pursue, the next section will present the hypotheses of my research. 2.3. THE HYPOTHESIS OF THE RESEARCH As has been suggested by many scholars, education is a kind of social investment which is the result of many years of efforts (Cochrane et al., 1980; Caldwell, 1979; Caldwell and McDonald, 1981; D'Souza and Bhuiya, 1982). Once an individual has got it, it is less probable that he/she may lose it. In 70 addition, education is not directly influenced by the fluctuations and changes of the economy. Along with the health transition theory, I formulate four hypotheses in this dissertation. Firstly, I expect that the higher the level of mother's education the lower she will experience infant and child mortality in her reproductive life. Specifically, this hypothesis implies that the statistical association between maternal education and childhood mortality is negative and persists even after controlling for economic resources. Secondly, I expect that the statistical association between maternal education and childhood mortality is stronger in childhood (1-4 years) than in infancy (0-1 year). Thirdly, I assume that when a family (household) experiences favorable socioeconomic and environmental conditions, the quality and the amount of social contacts facilitated by its environment and by its social status are enough to locate this family at the crossroads of non- traditional and modern ideas especially about health, hygiene, and nutrition, etc. Whatever the level of parents' education, this family will experience low risk of infant and child death. When these conditions are not favorable, the risk of dying for children will be inversely proportional to the level of parents' education, especially mother's education. This assumption implies that education is the key force in placing a woman at a relative advantage in terms of child health when severe obstacles of access exist (Rozenzweig and Schultz, 71 1982) . Fourthly, because schooling "enhances knowledge about effective ways to prevent, recognize, and treat childhood illnesses" (Cleland, 1989, p.411), I assume in this dissertation that maternal education results in improved infant/child survival because educated mothers are less likely than uneducated ones (1) to give birth to low birth weight babies, (2) give birth to premature babies. At the same time, I assume that educated mothers are more likely than the uneducated ones (1) to possess health knowledge, (2) to use health services that effectively prevent fatal childhood diseases, (3) to have better nourished children, (4) and to work outside the home. 2.4. THEORETICAL FRAMEWORK To assess the first objective, the study relies on the analytical framework developed by Mosley and Chen (1984), which posits that the background socioeconomic determinants (such as maternal education and household economic status) must operate through a set of proximate determinants to influence infant and child survival chances. For this study, the Mosley and Chen's framework has been adapted to meet the requirements of the study. (See Chart 2.4.1. the Framework adapted for the present study) . In this case, the study dependent variable is whether the child is dead or alive. 72 In order to assess the pathways through which maternal education affects child mortality, I will test whether or not children of educated mothers are better nourished, protected by immunization, suffer less from diarrhea and fever; and whether or not educated mothers display positive behaviors when their children have diarrhea or fever. Therefore, the following variables will be considered as dependent: (1) nutritional status of the child, as measured by arm circumference; (2) birth weight; (3) length of pregnancy; (4) mother's behavioral responses in case of diarrhea and fever; (5) mother's working status; (6) immunization status of children. Two independent variables are contrasted in the study: maternal education and the household economic status. Maternal education is measured as the number of years of formal schooling completed. The household economic status is measured by (1) the household expenditures for consumption; (2) the index of household wealth, captured by the materials used in the construction of the house where the household resides (floor, wall, roof). Five routes through which maternal education is operating to lower infant and child mortality are suggested. Among these routes four are considered as proximate determinants and one is another socioeconomic determinant. The first proximate determinant is called family formation pattern which includes age of mother at the birth of child, parity or gravidity, pace 73 of reproduction, and interactions of all these variables. The second proximate is the environment contamination which includes the housing conditions, overcrowding factors, and the conditions of immediate environment of habitat. The third proximate includes the nutrient deficiencies of the mother during pregnancy which is captured by birth weight, age of pregnancy, and previous birth loss. The fourth proximate includes measures of illness control, i.e., preventive measures (e.g. antenatal care, use of sugar and salt solution, use of oral rehydration therapy, knowledge of SSS) and curative measures including use of health services, use of modern medical technology (e.g., drugs), and promptness in taking sick child to the physician or dispensary. The fifth route through which education is believed to have an impact on child survival is the household economic status, i.e. household income and the household living conditions as captured by the materials of construction of the housing. By all accounts, this pathway also influences child survival through the four proximate determinants examined above. Figure 2.4.1. summarizes the conceptual framework of maternal education as it has been identified throughout the literature. This framework has its essential basis in Mosley and Chen's (1984) model, though other conceptual frameworks such as Chen (1983), Millard (1994) have also been carefully 74 consulted in this regard. Figure 2.4.1: The Route of Maternal Education Effect SOCIOECONOMIC FACTORS MOTHER'S EDUCATION HOUSEHOLD LEVEL: . Income . Housing constru ction materials — > PROXIMATE DETERMINANTS A.FAMILY FORMATION PATTERN . Age of mother at birth . Gravidity /parity . Birth interval . (interaction b/w these factors) B.ENVIRONMENTAL CONTAMI NATION . Housing conditions . Conditions of immediate environment of habitat . Crowding C. NUTRIENT DEFICIENCY OF MOTHER DURING PREGNANCY . Birth weight (Bweight) . Age of Pregnancy . Previous pregnancy loss D. ILLNESS CONTROL 1) Preventive Actions: . Antenatal care . Immunization . Knowledge of S.S.S. . Use of O.R.T./S.S.S. 2) Curative Actions: . Use of health services . Use of modern medical technology (e.g. drugs) . Promptness in taking sick child to physician or promptness in giving sick child appropriate medical treatment HEALTH STATUS CHILD Morbidity and/or Nutritiona status of the child — > — > J Infant & child survival 75 2.5. METHODOLOGICAL APPROACHES In accordance with the objectives of the research, two methodological approaches are defined. The first approach consists in relating maternal education to child survival, controlling for, on the one hand, the proximate determinants defined above, and on the other, the economic resource, maternal occupation and marital status, and demographic factors of the child. The appropriate techniques to be used in the relationship will be defined in chapter 4. The second methodological approach consists in establishing the relationship between maternal education and the factors considered as pathways to affect childhood survival. Here again, the appropriate techniques to be used in order to reach this objective will be defined in chapter 5. It will not be possible to apply a path analysis to investigate the pathways through which maternal education exerts its impact on child mortality. Three obstacles prevent us to use of this approach. First, the data set to be used in this study were obtained by a cross-sectional rather than a longitudinal approach. Second, some crucial data were obtained only for children who were still alive at the time of survey. This is the case of information on nutritional status of children, immunization status of children 12-23 months, and on diarrhoea and fever. Third, information on housing and 76 household income are related not to the time where the child was born rather to the time of the survey. 77 Chapter 3 THE EMPIRICAL DATA The goal in this chapter is to present the data set used in this dissertation. Two sections are developed in here. Section 3.1. shows how the data were gathered, i.e., the sampling procedures used. Section 3.2. presents the socio demographic characteristics of the population and the characteristics the housing or habitat.1 3.1. OBJECTIVES OF THE SURVEY AND SAMPLING DESIGN 3.1.1. Objectives of the Survey The data used in this dissertation come from a sample survey carried out in 13 Cities of Zaire. These cities are: The core of this chapter is written on the basis of the published and unpublished documents written for the survey. Among these documents I can mention: (1) the general report of the survey written in French by Ngondo a Pitshandenge, Dr. R. Gamboa, N. Luyinduladio, and N. Kinavwidi (1988); (2) the questionnaire and manual of the survey written in 1987 by N. Kinavwidi, Mr. Kayembe, and N. Luyinduladio; (3) the manual for the field workers which was written in 1987 by N. Kinavwidi and N. Luyinduladio; (4) the methodology for the survey and the methodology for the adjustment of sampling fractions which was written by Mr. Niwembo Kinavwidi and N'zinga Luyinduladio; (5) and many other code books released by the Institut National de la Statistique (INS) for use of data collection. The rest of the chapter is written on the basis of my personal experience in the field and on the data computed from four files of the raw data (these files are: Modules I, II, III, and VI) . I have used the SAS software under UNIX operating system to exploit these rich data. 78 Boma, Matadi, Bandundu, Kikwit, Mbandaka, Zongo, Kisangani, Bukavu, Lubumbashi, Likasi, Kolwezi, Kananga, and Mbuji-Mayi. Map 3.1.1. shows the location of these Cities. The data collection was done by the "Fonds National Medico-Sanitaire" (FONAMES), a public institution created by the government of Zaire in order to coordinate the health-related decisions. This survey was financed by UNICEF. (That is why its name is FONAMES-UNICEF Survey). Three objectives were followed in this survey. The first was to gather information on health status of mother and children as well as on social and economic conditions of the households. The second objective was to provide statistical indicators on the levels of infant, child and maternal morbidity and mortality. The third objective was to identify the causes and determinants associated to children's and mothers' health status. 3.1.2. Sampling Design A representative sample of households was drawn to meet the objectives set for the survey and to respond to financial and temporal constraints since the territory to cover was immense. Having considered all these constraints, a sample of 6,500 households was decided by the scientific team in charge of the data collection. To distribute the sample size in each city we proceeded as follows. First, we estimated the total number of households 79 living in each city on the basis the total population observed in the 1984 census. This estimation was obtained by dividing the population in each city by 5.5, the average number of persons per household observed in the 1984 census. Second, we estimated the ratio of the largest city to the lowest city (in terms of number of households). This ratio is 30 to 1. Third, we determined the minimum size of the sample in each city. This minimum size is 300 households. If this minimum size was drawn in each city, the total sample would be 3,900 households. A difference of 2,600 households would be left to reach the size set by scientific team of the survey operations (i.e. 6,500 households). Therefore, a fourth step was necessary. This step consisted of distributing the residue of 2,600 households proportionally in the 13 cities. Table 3.1.1. gives the estimated size of the sample in each city. 80 Table 3.1.1.: Population in the 1984 census. Number of Household estimated. Estimated Size of the Sample, and Global Sampling Fraction per Citv CITY POPULATION 1984 HOUSEHOLD NUMBER ESTIMATED % SAMPLE SIZE PER CITY GLOBAL SAMPLING FRACTION CITY( ) BOMA 88,556 16,101 3.29 388 1/41.50 MATADI 144,742 26,317 5.37 444 1/59.27 BANDUNDU 63,189 11,489 2.35 363 1/31.65 KIKWIT 146,784 26,688 5.45 445 1/59.97 MBANDAKA 125,263 22,775 4.65 355 1/28.28 ZONGO 18,072 3,286 0.67 318 1/10.33 KISANGANI 282,650 51,391 10.49 580 1/88.61 BUKAVU 171,064 31,103 6.35 470 1/66.18 LUBUMBASHI: 543,268 98,776 19.38 836 1/118.15 LIKASI 194,465 35,357 7.22 493 1/71.72 KOLWEZI 201,382 36,615 7.48 500 1/73.23 KANANGA 290,898 52,891 10.08 588 1/89.95 MBUJI-MAYI: 423,363 76,975 14.94 720 1/106.91 TOTAL 2,693,696 489,763 100.00 6,500 1/75.35 Source: Republique du Zaire, D^partement du Plan, Institut National de la Statistique, 1988, Le Zaire en chiffres. p.16 Notes: (*) Estimated by dividing the population of the city by the average number of persons per household observed in the 1984 survey. This number is 5.5 persons. (**) These fractions were adjusted after the census of compounds ("le releve parcellaire") was performed. (The city of Mbandaka had a total population of 125,263 inhabitants in 1984. During the sampling we mistakenly copied 55,225 instead of 125,263. By doing so, we have lost roughly 66 households in this city). The unit of analysis of the survey is the household. But since there is nowhere in Zaire a list of households available for sampling design, we used two sampling units in order to identify the households living in the 13 cities. One of the units is the quarter. In urban areas in Zaire, one can 81 establish the list of quarters existing in an administrative zone. (In some cities quarter was replaced by blocs, "cellules," etc.). Quarter or bloc is an administrative unit which is made up of a certain number of compounds. The second sampling unit used is the compound. A compound is made up of households (or families) which live within it. To minimize the cost of the survey as well as to maximize the precision of the sample, a two stage-sample-survey was carried out in which the sampling unit was not the individual compound but a cluster of inhabited compounds. At first stage of the sampling, the unit chosen was the quarter (or neighborhood) while at the second stage, we chose the inhabited compound as the sampling unit. At each stage, the statistical universe is the total sampling units identified. 3.1.3. Stratification of the Universe at the First Stage To increase the efficiency of the sampling, a double stratification of the sampling units (quarters, cellules, blocs, etc.) was done at the first stage. The first stratification, which is based on the socioeconomic and housing characteristics of the units constituting the first stage, was done according to the criteria previously defined by the National Institute of Statistic (I.N.S.). These criteria were defined in accordance with previous knowledge of the urban areas of Zaire, particularly the 1984 national census, and in accordance with the principle that Primary 82 Sampling Units (quarters, cellules, blocs, etc,) located within a stratum must be homogeneous among them, while the primary sampling units between strata must be heterogeneous among them. Table Annex 1 gives the criteria defined by I.N.S. used to create each stratum. Table Annex 2 shows the distribution of primary sampling units (quarters, blocs, etc.) by strata and by cities. The second stratification done in each city consisted of grouping primary sampling units into large subsets according to their population size in 1984 or 1986. These subsets are called sub-strata. Table Annex 3 gives the number of sub strata created and selected in each city. 3.1.4. Determination of the Sample Size at The First Stage and Sampling Fractions at 1st and 2nd Stages The size of the sample at the first stage was determined within the sub-strata by allowing a proportional distribution of primary-sampling units (quarters, cellules, blocs, etc.) in each sub-stratum. By doing so, we took into account not only the number of primary-sampling units in each sub-stratum but also the size of the population of each of them. To determine the sampling fraction at the first stage within each sub-stratum in each city, we divided the number of primary sampling units obtained in the sub-stratum by the total number of these units in the corresponding universe. The sampling fraction at the second stage was also computed within each sub-stratum. This fraction is the ratio of the global 83 sampling fraction defined for the city to the corresponding fraction obtained at the first stage. Table Annex 3 gives for in each city the sampling fractions defined in each sub stratum at 1st and 2nd stages. 3.1.5. Drawing of the Sample at the First and Second Stages. Adjustment of Sampling Fractions, and Drawing of Final Sample At the first stage of sampling, the sample of quarters (primary sampling-units) was drawn systematically from the sampling frame previously constituted. To do so, we used the sampling fractions defined at the first stage (Table Annex 3) . (The list of primary sampling units— quarters, cellules, blocs, etc.— selected in each city exists. This list gives not only the name of quarters selected at first sampling stage, but also the number of inhabited compounds and the sampling fractions at the first and second stages. Because this list is very long, it will not be published in this study). Before selecting the compounds to be surveyed, we first proceeded by a census of compounds in all the selected primary-sampling units (quarters, blocs, cellules, etc.). This census (called "releve parcellaire" in French) was justified by the fact that in the 13 cities surveyed there was not a list of inhabited compounds available. Therefore, the establishment of this list pursued a practical and a theoretical objectives. The practical objective is that this operation made it possible to obtain a sampling frame of all 84 inhabited compounds located in the selected quarters by identifying the compounds which were inhabited and those which were not inhabited in May-June 1987. The theoretical reason of this operation is to allow all the compounds (and consequently all the households) located in the selected quarters (or blocs, cellules, etc.) to have the same probability of being selected in the final sample of the study. This census allowed the estimation of the probability for a given compound (and all the households within it) to belong in the sample. After the identification of inhabited compounds within each sub-stratum, the global sampling fractions and the sampling fractions defined at the second stages were adjusted using the following relation: fg= fl * f2 Where fg is the global sampling fraction in a given sub stratum; f1 is the sampling fraction at the first stage; f2 is the sampling fraction at the second stage. After the adjustment of fractions fg and f2, the next sampling step was the selection of the final sample of the survey. (There is a methodological note which explains how the adjustment of the sampling fractions was done. It will not be published in this dissertation) . To draw the sample of compounds, a systematic sampling was performed on the sampling frame of all compounds identified during the census. The final sample included a total of 5,410 compounds in all of the 13 cities, an average of 1.22 households per 85 compound. Within these compounds, a total of 6,574 households were identified among which, 6,527 households accepted to be interviewed; i.e., the coverage rate is 99.3 percent. In addition, the survey identified 42,877 individuals, an average of 6.57 persons per household or 7.93 persons per compound in urban areas in Zaire. Table 3.1.2. presents the distribution of population, households, and compounds by strata. It also shows the following averages: households per compound, persons per household, and persons per compound. It appears in this Table that households which live in upper class neighborhoods (strata 0) and planned neighborhoods (strata 3) , two strata lived, in majority, by rich and upper middle class people, have higher average of persons per household, i.e. 7.01 and 7.26, respectively. But households which live in poor strata (formed by extended neighborhoods (strata 4) and eccentric neighborhoods (strata 6)), are less populated. 86 Table 3.1.2.: Distribution of Compounds. Households, and Population Surveyed Among the Six Strata. 13 Cities of Zaire. 1987. Average Number of: Strata Compounds Surveyed Households Surveyed Population Surveyed H/C P/H P/C 0 422 491 3,443 1.16 7.01 8.16 1 1,238 1, 544 9,837 1.25 6.37 7.95 2 712 912 5,937 1.28 6.51 8.34 3 982 1,071 7,778 1.09 7.26 7.92 4 1,773 2,226 13,810 1.26 6.20 7.79 6 283 330 2, 072 1.17 6.28 7.32 Total 5,410 6, 574 42,877 1.22 6.52 7.93 Source: Ngondo a Pitshandenge, Dr. R. Gamboa, N. Luyinduladio, and N. Kinavwidi, 1988, Table 2.01, p.38. Legend: H/C means number of Households per Compound; P/H means number of Persons per Household; and P/C means number of Persons per Compound. 3.1.6.: Techniques of Data Collection and Questionnaire The FONAMES-UNICEF survey is a single-round cross- sectional survey where all the households living in the sample of selected compounds were interviewed. The task of the field workers was to visit the selected compounds, identify all household living there, and to interview all individuals who lived in those households, be they residents or not. A written questionnaire was applied to all the households. This questionnaire was written in French. A manual was available to the interviewers for use in the field. It explains each single part of the questionnaire by indicating, for example, how the interviewers should be dressed, how they should approach the individuals, how they should introduce themselves and the survey, and how they should ask each 87 question. For most of the intimate questions, this manual explains what the filter questions are before asking the main question. Finally, the manual gives the codes of certain questions. In the field, the questionnaire was filled as follows, the field worker had to ask the questions in the way they were explained in the manual. Then he/she had to write the answers given by the individuals being interviewed. To increase the acceptability of the population surveyed, the head of the household was interviewed first. If he/she agreed, the interview started with him, then his wife/wives, his children and other members of the household. The questionnaire for this survey includes eight modules (parts). In order to allow the identification of households surveyed in each city, a key of the questionnaire (called "identification") was added in each "module." The following information is included in this identification: file code (column 1), strata (col 2)2 and sub-strata (col 3), province (col 4) , district (col 5-6) , city (col 7-8) , quarter or neighborhood (col 9-10), compound number (col 11-14), number of households in the compound (col 15-16) , rank of the household (col 17-18) , and date of the interview: day (col 19- 20) and month (col 21-22). The text below explains in detail each module of the questionnaire. (Again the questionnaire used in this survey is 2 COL means column 88 too long to be published in this dissertation). A) Module I: Composition of the Household This module intends to identify all the members living in each household. (Here, it should be underlined that in Zaire, a household may be composed of one or several nuclear families). Each member of the household was given a identification number (col 2 3-24). For each member living in the households surveyed twelve characteristics were identified: (1) Gender (col 25); (2) Relationship to the head of the nuclear family living in the household (col 26). When several families live in one household, the head of the individual family is linked to the head of the household; (3) Degree of "kinship" (col 27- 28), i.e., the person to whom the household member is identified (or attached). Here, the wife was identified along with her husband while children were identified along with the mother if the mother lives in the household. Otherwise children were attached to the parent living in the household; (4) Residence status (col 29), defines if the household member is resident or visitor; (5) Place of birth (col 30-32). For Zairians, the survey intended to note the administrative zone or the city where they were born. For foreigners, the country of birth was indicated; (6) Date of birth (col 33-34 and col 35-36): Information to be indicated is month and year, respectively; (7) Parents' survival status: Indicate whether 89 the father (col 37), and mother (col 38) of the person interviewed are still alive; (8) Ethnic group for Zairian or country of citizen for foreigners (col 39-41); (9) Matrimonial status (col 42); (10) Education (col 43-44): number of year completed; (11) Employment status (col 45); (12) Position in the job (col 46). B) Module II: Characteristics of the Housing Unit This module gives the characteristics of the housing unit of the households. The following characteristics are included:3 (1) Number of rooms (col 25-26); (2) Status of ownership of the residence (col 27) ; (3) Nature of the housing unit: Floor (col 28) , Wall (col 29) , and Roof (col 30) ; (4) Latrine: Location (col 31), mode of use (col 32), effective use (col 33), distance to the residence (col 34-37); (5) Water: source (col 38), distance to the residence (col 37-42) ; (6) Treatment of wastes: human wastes (col 43-44), household wastes (col 45) ; (7) Distance between the housing unit and garbage depository (col 46-47); (8) Care given to the trash- can of the household (col 48). C) Module III: Maternity History of Women Aged 13-49 Are included in this file all women aged 13 to 49 living in the households surveyed. To allow comparability with module 3 In each household, this information is attached to the head of the household (col 23-24), meaning that all members of the same household will have the same housing information. 90 I, the identification number of the woman in module III (col 23-24) is the same as the one used in module I (col 23-24). Each woman living in the households surveyed was required to provide information relative to each pregnancy she had in her lifetime. Each pregnancy was ranked between 0 to 36 for practical reasons (col 25-26). The following information was asked for each pregnancy: (1) Outcome of the pregnancy (col 31); (2) Date of delivery (or any other event related to the pregnancy): Day (col 32-33), Month (col 34-35), Year (col 36- 37); (3) Place of delivery (col 38); (4) Age of pregnancy at delivery (col 39); (5) Sex of the child born alive or still born (col 40); (6) Birth weight of child born alive (col 41); (7) Survival Status of child born alive (col 42); (8) Date of death: Day (col 43-44), Month (col 45-46), Year (col 47-48); (9) Employment status of the mother during the pregnancy (col 49); (10) Position in the job during pregnancy (col 50); (11) Matrimonial status of the mother during the pregnancy (col 51); (12) Education of the mother during the pregnancy (col 52-53) : This variable is given in number of years completed; (13) Contraceptive use by the woman in 1986 (col 54); (14) "Cause" of morbidity of children under 5 year-old (col 55-57): Here, the reported "cause" is based on mother's assessment; (15) Arm circumference for children of 12 to 59 months (col 58); (16) "Cause" of death for children dead under the age 5 (col 59-61). 91 D) Module IV: Maternal Morbidity This part of the questionnaire intends to collect information on the physiological status of women who were pregnant during the survey or those who gave birth 42 days before the interview (post-partum). For the time being, this part is not going to be exploited. Check the questionnaire to see the information gathered. E) Module V: Maternal Mortality This module concerns all women, formerly members of the households surveyed, who died 42 days before the interview. For all these women, information was ask to find out whether they were pregnant at death, or in labor at death, or in post partum. For this study, this part is not going to be exploited. See the questionnaire to know about the information gathered. F) Module VI: Household Budget In this module, the questionnaire has three parts. Part I concerns the revenues (the exact word from French is "recettes"), Part II concerns the expenditures for familial enterprises. These expenditures do not include expenditures for consumption in the household. Part III includes the expenditures for household consumption. In part I (Returns or "Recettes"), the following 92 information was gathered: (1) Returns from familial business (from French "recettes des entreprises familiales"). This information is located in col 23-28, col 29-34, and col 35-40; (2) Returns from borrowing (col 41-46, col 47-52, and col 53- 58) ; (3) Presents ("cadeaux") or rewards ("gratifications") received. To field worker had to distinguish presents or rewards received in hard cash (col 59-64) from presents or reward received in kind (col 65-70);4 (4) Withdrawal from savings (col 71-76); (5) Salary or wage received last month (col 77-82); (6) Expected Income (col 83-88). Here, we asked the household head about the monthly income they would expect to earn in order to allow them cope with the living expenses of their household. In part II, (Expenditures for Familial Enterprises, translated from french "Depenses qui ne constituent pas un achat pour 1'unite de consommation dans le menage"), we asked about expenses people interviewed made in order to run their small business. These are not expenditures for consumption in the household. The information gathered here includes: (1) Purchase of goods and services to be sold in the familial enterprise ("familial business"): col 89-94; (2) Repayment of debts taken for familial business (col 95-100); (3) Presents given to relatives or friends not living to the household (col 101-106, and col 107-112); (4) The amount of money saved in 4 For presents or rewards received in kind, the interviewed had to estimate in hard cash the price of this present or reward. 93 his/her account (col 113-118); (5) "Ristourne"5 (col 119- 124) . Part III of the module household budget collects data on expenditures made by household members for their consumption. The following information on household consumption expenses were asked: (1) Expenses on Food. Two types of information had to be collected. First the expenses in hard cash made in order to buy food (col 125-130) . Second, for households that consumed food produced by themselves, we estimated the prices of this food if those households had to buy it. This estimation was written in columns 131-136. Then we collected data on housing expenses. Four types of housing expenditures are distinguished. The first type includes the expenditures in housing investment (e.g., purchasing a house, buying cement, etc.) . This information was written in col 137-142. The second type includes expenses for rent, taxes, electricity bill, water bill, repair expenses, etc. (col 143-148). The third type are expenses on energy (col 149-154). The fourth type are expenses on furniture (col 155- 5 "Ristourne" (a French word) is a form of loan in which a group of two or more persons agree to put a certain amount of money together in the end of each month. This amount is given to one person in the group. The next month the same amount of money is collected but this time the money is given to another person in the group. The process will continue until all the members of that group will be serviced. The idea behind this operation is the allow each member of the group to build a capital in order to start a business or to buy a goods that he/she cannot buy from his/her monthly salary. "Ristourne" is an economic operation commonly practiced in urban areas of Zaire. It is a form of loan because the money one person receives will be reimbursed gradually. 94 160) . We also collected data on clothing expenses for men (col 161-166), for women (col 167-172), and for children (col 173- 178) . Finally, we collected data on miscellaneous expenses which include: transportation (col 179-184), education (col 185-190), health care expenses (col 191-196, col 197-202, and col 203-208), and others (col 209-214). G) Module VII: Immunization Status of Children Aaed 12-23 Months This module gathers information on immunization status of children who were still alive and living in the household surveyed. This information aims at evaluating the immunization coverage in urban areas of Zaire. This part was set up in the way that for each child included it should be possible to identify his/her characteristics, those related to his/her mother, and those related to the household where she/he lives. The following information are included: (1) Mother's identification number (col 23-24); (2) Child' identification number (col 25-2 6); (3) Availability of the child's Immunization card (col 27) ; (4) Child's date of birth: Day (col 28-29), Month (col 30-31), and Year (col 32-33); (5) Immunization scar (col 34) : This concerns BCG; (6) Immunization status and dates of inoculation of: BCG (col 35- 40), Polio 1 (col 41-46), Polio 2 (col 47-52), Polio 3 (col 53-58), DTCoq 1 (col 59-64), DTCoq 2 (col 65-70), DTCoq 3 (col 71-76), Measles (col 77-82). 95 H) Module VIII: Diarrhoea and Fever among the under Five Children This module concerns children under 5 years still alive and still living in the selected households. For diarrhoea, the survey intended to determine whether the child experienced diarrhoea the last 15 days before the interview. If the child had diarrhea the last 15 days of the survey, the following questions were asked. How often the following nutrients breastfeeding, water, solid food were given to the child? The second question wanted to know whether the sugar-salt solution was given to the child, where the child was brought for treatment, how many days elapsed before the child was brought there for treatment, what type of treatment the child had received, and whether the mother knows about the sugar-salt solution. In addition, module VIII collected data on fever status of the child the last 15 days before the interview. If the child had fever, the following information was collected. Where the child was brought for treatment, how many days elapsed before the child was brought there, what type of treatment the child had received, After how many days the first treatment was received. The next part of section 1 concerns the administrative organization of the field. 96 3.1.7. Organization of the survey Chart 3.1.1. shows the administrative organization of the field and the organization of the exploitation of the data. In this organization, the Technical team was the core of the enterprise. Its role was to devise, execute, and organize the survey in the field, and to link the field's activities to those in the computing center. This technical team was working under the supervision of the supervisor who represented the FONAMES and the Ministry of Health. The scientific committee of the survey was meeting frequently to discuss activities devised by the technical team, and propose new direction when necessity appeared. In each city, two field teams were operating. The first team was made up of interviewers whose role was to interview people living in the households of all selected compounds. A total of 130 interviewers were employed. The second team was made up of 26 verificators, two per city. 97 Chart 3.1.1.: The Organization of the Survey FONAMES-UNICEF r ¥ n ▼ SUPERVISOR SCIENTIFIC COUNCIL i ▼ TECHNICAL TEAM I ▼ FIELD TEAMS Source: Ngondo a Pitshandenge, Dr. Ruben Gamboa, N'zinga Luyinduladio, Niwembo Kinavwidi, 1988, p.26. The role of verif icators was to check the questionnaire filled in by interviewers. The verificators received a special training to allow them to detect the incoherences in the questionnaire filled. A questionnaire not correctly filled was sent back to the field worker, through the Head of the team, in order to be filled again. This latter had to go back in the household to ask for the missing information. In the field, both teams were under the supervision of the Head of the team. A total number of 13 Heads of team worked in this survey, one per city. The Heads of teams were selected and trained in Kinshasa. Only candidates with a university degree were chosen to head the field team. The ▼ COMPUTING TEAMS 98 majority of them had a degree in demography and/or had a large experience in survey supervision. This group worked in close collaboration with the TECHNICAL TEAM. At the end of the field operations, the 13 Heads of the teams worked in collaboration with the Technical team to codify the questionnaire. One year and a half was necessary to finish all the operations defined for the survey. Table 3.1.3. below gives the calendar of the operations. 99 Table 3.1.3.: Calendar of the Execution of the Survey ACTIVITIES PERIODS 1) Conception and preparation of the survey: definition of objectives, writing methodological documents, etc. January-April 1987 2) Seminar about the Survey April 1987 3) Recruitment and Training of Head of Teams April-May 1987 4) Identification and Establishment of of Sampling Frame at First Stage of Sampling; Stratification and Selection of Primary Sampling-Units in each city May-June 1987 5) Recruitment and Training of Interviewers and Verificators in each City June 1987 6) Census of Sampling-Units Selected at First Stage; Adjustment of Sampling Fraction and Selection of the Final Sample June 1987 7) Survey Interviews on the Field July-August 1987 8) Data Codification and Computing August-Oct. 1987 9) Analysis of Data Nov.-Dec. 1987 10) Report Writing Feb.-April 1987 Source: Ngondo a Pitshandenge, Dr. R. Gamboa, N'zinga Luyinduladio, and N. Kinavwidi, 1988, Table 1.03, p.28. 3.2. SOCIO-DEMOGRAPHIC CHARACTERISTICS OF THE POPULATION. HOUSING CONDITIONS This section will give a global picture of the characteristics of the population surveyed and describes the housing conditions in the 13 Cities surveyed. 100 3.2.1. Demographic Characteristics of the Population Six characteristics of urban population of Zaire are presented. These are (i) status of residence, (ii) composition of population, (iii) marital status, (iv) fertility, (v) level of education, and (vi) occupational status of the population. About the status of residence of the population, Table 3.2.1.indicates that the population surveyed (42,877 individuals) is made up of 97.2% residents, and only 2.8% of visitors. Surprisingly, Table 3.2.1. shows that the majority of residents are male while the majority of visitors are female. Table 3.2.1.: Distribution of Population bv Residence Status Gender Status of Residence Male Female Total % Residents 20,977 20,714 41,691 97.2 Visitors 482 704 1,186 2.8 Total 21,459 21,418 42,877 100 Table 3.2.2. shows the composition of the population by age and gender. It indicates that the urban population in the 13 Cities surveyed is extremely young: In effect, 60.2% of the population sampled was less than 20 year-old against 34.6% of persons aged 20-54 year-old, and only 3.9% aged 55 years and more. The mean age of the population is 20.09 years for male, 19.24 years for females, and 19.67 years for both. In 101 addition, this table shows that there is gender imbalance in the urban population of Zaire. In age groups less than 20 years, the composition by gender seems balanced except for age group less than 5 years where the sex-ratio is 97%, indicating a surplus of female population, which may be explained by male over mortality compared to female of the same age. In age group 20-55 year-old, there is an over-feminity between the ages 20 and 30, and an over masculinity between the ages 30 and 50. In the age group 55 and above, the age-sex composition is clearly dominated by male population. Table 3.2.2.: Composition of the Population bv Age and Gender AGE GROUP MALE FEMALE TOTAL SEX-RATIO (1) (2) (3) (4) (5) 0-4 3,662 3,770 7,432 97 5-9 3,480 3,432 6,912 101 10-14 3,109 3,043 6,152 102 15-19 2,648 2,688 5,336 99 20-24 1,824 2,162 3,986 84 25-29 1,452 1,706 3,158 85 30-34 1,191 1,153 2,344 103 35-39 944 905 1,849 104 40-44 672 625 1,297 108 45-49 597 556 1,153 107 50-54 514 519 1,033 99 55-59 503 321 824 157 60-64 219 136 355 161 65-69 189 111 300 170 70-74 62 33 95 188 75-79 30 28 58 107 80-84 15 8 23 188 85 + 19 11 30 173 Unknown 329 211 540 156 TOTAL 21,459 21,418 42,877 100 Source: Naondo a Pitshandenge et al., 1988, op. cit., Table 2.07, p.47. (The last column is added by me). 102 About marital status, Table 3.2.3. indicates that approximately half of the population 15 years and more who live in the 13 Cities are married. Single people represent about 40 percent. Other marital statuses make less than 10 percent of the population. The mean age at marriage for men is 26.7 years, and 21.7 for women. The latter statistics indicate that urban population in Zaire is beginning to postpone marriage at a later age compared to their rural fellows who, in 1975-77, were getting married, on average, at 24.1 and 19.5 years (Tabutin, 1982). However, in spite of the postponement of marriage at a later age, the intensity of marriage is still high. In effect, only about 3% of people of 55 and more are single, 2.6% for men and 2.4% for women. Table 3.2.3.: Distribution of Population fin %) bv Marital Status and Gender Single M. Mono M. Poly. Consens. Divo. Widowed Male 44.6 44.4 6.0 2.9 1.1 0.8 Female 35.2 43.7 9.2 5.5 4.3 2.0 Legend: M. Mono= Married Monogamously; M. Poly.= Married Polygamously; Consens.= Consensual; Divo.= Divorced Tables 3.2.4. and 3.2.5. indicate that fertility in the 13 Cities surveyed is very high. For example, Table 3.24. shows that each woman aged 45-49 years has an average of 6.91 children. The total fertility rate was about 6 children per woman in 1987 and 7.25 children per woman in 1983. The mean of women at birth of child for the retrospective period of last 12 months, is 30.06 years. High fertility in urban Zaire is not a new phenomena. It has been reported in previous studies by Romaniuk (1967, 1968) and Tabutin (1982) . What is new in the data displayed in Table 3.2.5. is the reduction of fertility in age groups 15-19 and 20-24 years, and at the same time the increase in fertility in age groups 25-29 an 30-34 years. All the figures presented in Tables 3.2.4. and 3.2.5. are unadjusted. Table 3.2.4.: Distribution of Women bv Age Group. Children Ever Born bv Aae Group, and Average Parity per Woman bv Age Age Group Children Ever Born at the Time Women __________________ Average Parity of Survey Both Sex per Woman 13-14 1124 17 0.0151 15-19 2,671 552 0.2067 20-24 2,129 2,746 1.2898 25-29 1,713 4,952 2.8908 30-34 1,152 5, 066 4.3976 35-39 901 5,369 5.9589 40-44 620 4,230 6.8226 45-49 550 3,799 6.9073 Total 10,860 26,731 2.4614 104 Table 3.2.5.: Distribution of Children Born in 1981-85. Children Born the last 12 months before the Survey. Period Fertility Rates (in thousand) in 1981-85 and in the last 12 months Period Fertility Rates Age Group at the time of Survey Children Time Born in 1981-85 Children Born the Last 12 Months In 1983 Last 12 Months 13-14 4 2 ---- ---- 15-19 194 233 17.2185 87.2332 20-24 1,656 545 186.9075 255.9887 25-29 2,245 522 352.9874 304.7286 30-34 1,484 290 318.1136 251.7361 35-39 1,089 178 307.6271 197.5583 40-44 509 50 185.0909 80.6452 45-49 226 23 83.0882 41.8182 50 + 25 6 ----- ----- Unknown 28 12 Total 7,470 2,248 For the period 1981-1985:b . Total fertility rate: 7.2552 children per woman . General fertility rate: 184.11 per thousand (1,481/8,044) . Crude birth rate: 49.20 per thousand (1,481/30,099) For the last 12 months: . Total fertility rate: 6.0985 children per woman . General fertility rate: 228.84 per thousand (2,228/9,736) . Crude birth rate: 52.40 per thousand Tables 3.2.6. and 3.2.7. give data on level of education of the population, male and female respectively. In both tables, I distinguish schooled population, i.e. people who had 6 Figures are centered in 1983. The number of births in 1983 is estimated at 1,481, the yearly average number of births occurred in the period 1981-85. The total population in 1983 is approximately equal to 30,099 (i.e., the 1987 population minus the population born between 1983 and 1987). The number of women aged 15-49 years in 1983 is estimated at 8,044. This corresponds to the female population aged 19-53 years in 1987. 105 been to school but were not attending it at the time of the survey, from school-attending population during the academic year 1986-1987. About schooled population, the data reveal that, for both sexes, not all the people aged 6-25 years had been schooled. The percents of population schooled are specifically weak between the ages 6 and 8 due probably to late school enrollment. Men appear to be advantaged than women, especially for the cohorts born before the independence of the country in 1960 where the percent of schooled population is around 86.1 for males against 61.5 for females. About the school-attending population during the academic year 1986-1987, the data also show weak percentages between the ages 6 and 9 years due probably to late school enrollment. Above the age 9 up to 16 years for boys and 14 years girls, the school-enrollment rates are above 90 percent. Here, again males are advantaged compared to females. The illiteracy rates by age in the urban population surveyed in 1987 are obviously high in the age groups 6-9, varying between 8-97% for boys and 10-97% for girls. As said before, these high rates reflect more the late school- enrollment than definitive illiteracy per se. In the age groups 9 to 25 years, the illiteracy rates vary around in the range 1.8-5.8% for males, and in the range 4.6-10.5% for females. Above 25 years, population made up in majority of cohorts born before the independence on June 30 1960, the 106 illiteracy rate is about 12.5 % for males and 36.5 % for females. For the whole population, the rate of illiteracy is 10.82% for males and 20.42% for females, statistics which do not include children born after 1980. In addition to gender inequality in education in urban areas of Zaire, Tables 3.2.8. and 3.2.9. show that high illiteracy rates are concentrated in neighborhoods lived by poor, especially in extended neighborhoods (strata 4) and eccentric neighborhoods (strata 6). People aged 15 years or more who live in those two areas have spent less time in school than those who live in rich neighborhoods. In neighborhoods lived by rich people (strata 0 and strata 3), the rates of illiteracy are low (4.7-5.9% for males and 10.2- 17.1% for females) and people seem to have stayed in school longer than in poor neighborhoods. 107 Table 3.2.6 • • • Level of Education Julv-Aucrust 1987 of Male in Year of Birth Age1 Total Popul. Schooled Pop. School Total % Total -Attend. % Illit. rate % 1982-87 -6 4,413 — — — — — — 1981 6 623 18 2.9 16 2.6 605 97.1 1980 7 786 515 65.5 507 64.5 271 34.5 1979 8 653 553 84.7 547 83.8 100 15.3 1978 9 667 614 92.1 605 90.7 53 7.9 1977 10 586 565 96.4 558 95.2 21 3.6 1976 11 633 605 95.6 601 94.9 28 4.4 1975 12 656 631 96.2 617 94.0 25 3.8 1974 13 678 663 97.8 647 95.4 15 2.2 1973 14 556 542 97.5 526 94.6 14 2.5 1972 15 711 698 98.2 670 94.2 13 1.8 1971 16 479 466 97.3 450 93.9 13 2.7 1970 17 554 535 96.6 487 87.9 19 3.4 1969 18 453 444 98.0 364 80.4 9 2.0 1968 19 451 437 96.9 353 78.3 13 2.9 1967 20 428 417 97.4 316 73.8 11 2.6 1966 21 369 355 96.2 246 66.7 14 3.8 1965 22 360 344 95.6 185 51.4 15 4.2 1964 23 351 337 96.0 151 43.0 14 4.0 1963 24 316 307 97.2 91 28.8 6 2.0 1962 25 1961 and 345 323 93.6 98 28.4 20 5.8 before 25+ 6,051 5,211 86.1 195 3.2 756 12.5 Unknown — “ 340 ” — “ " ■ 340 100 TOTAL — 21,459 14,580 88.7 a1,230 50.0 6,788 10.8“ * Schooled Pop.= Schooled Population; School-Attend.= School Attending Population; Illit.= Illiterate Population; Rate= Illiteracy Rate; f1) Age is given in years; (2) Children born after 1980 are not included in this percent. Source: Ngondo a Pitshandenge et al., op. cit., 1988, Annex 2.04A, p.158 108 Table 3.2.7.: Level of Education of Female in Julv-Auaust 1987 Year of Birth Age1 Total Popul. Schooled Pop. N % School N -Attend. % Illit. rate % 1982—87 -6 4,516 — — — — 4,516 — 1981 6 670 18 2.7 14 2.1 652 97.3 1980 7 637 467 73.3 458 71.9 270 42.4 1979 8 629 538 85.5 537 85.7 91 14.5 1978 9 650 581 89.4 579 89.1 69 10.6 1977 10 590 544 92.2 538 91.2 46 7.8 1976 11 665 617 92.8 602 90.5 48 7.2 1975 12 651 609 93.5 602 92.5 42 6.5 1974 13 570 538 94.4 523 91.8 32 5.6 1973 14 567 537 94.7 513 90.5 30 5.3 1972 15 627 598 95.4 549 87.6 29 4.6 1971 16 496 472 95.2 404 81.5 24 4.8 1970 17 517 489 94.6 365 70.6 28 5.4 1969 18 524 484 92.4 295 56.3 40 7.6 1968 19 524 479 91.4 225 42.9 45 8.6 1967 20 434 408 94.0 147 33.9 25 5.8 1966 21 418 391 93.5 113 27.0 27 6.5 1965 22 453 409 90.3 97 21.4 41 9.1 1964 23 469 432 92.1 58 12.4 37 7.9 1963 24 388 360 92.8 30 7.7 28 7.2 1962 25 440 394 89.5 24 5.5 46 10.5 1961 and before 25+ 5,768 3,550 61.5 102 1.8 2,104 36.5 Unknown 215 215 100 TOTAL - 21,418 12,915 79.5 6,785 41.7 8,485 20.42 Leaend: Schooled Pop.= Schooled population; School-Attend.= School Attending Population; Illit.= Illiterate population; Rate= Illiteracy rate; N= Number; f1) Age is given in years; (2) Children born after 1980 are not included in this percent. Source: Ngondo a Pitshandenge et al., op. cit., 1988, Annex 2.05A, p.159. 109 9 Table 3.2.8.: Educational Patterns bv Strata and Gender in Urban Areas in Zaire. 13 Cities. Population 15 years or more. (Percentages). YEAR OF SCHOOLING 0 1-3 4-6 7-9 10-12 13+ STRATA 0 4.7 2.2 MALE 13.1 24.8 36.2 19.0 1 9.3 6.5 22.8 26.2 31.0 4.2 2 7.8 5.0 19.8 28.3 31.7 7.4 3 5.9 4.1 17.3 29.6 37.1 6.0 4 9.2 4.8 22.5 27.1 31.7 4.7 6 16.8 10.2 23.9 23.8 23.0 2.3 STRATA 0 10.2 5.1 FEMALE 21.1 32.1 27.3 4.2 1 26.4 9.9 22.4 25.3 15. 3 0.7 2 20.0 10.0 25.0 29.0 14.7 1.3 3 17.1 6.5 25.8 36.3 13.3 1.0 4 24.2 10.5 29.3 25.3 10.0 0.7 6 37.6 9.9 24.1 17.4 9.7 1.3 (See definition strata's codes in Chapter 3, Table Annex 1). Table 3.2.9.: Average Number of Years of Schooling of Population over 15 years. STRATA GENDER 0 1 2 3 4 6 TOTAL Male . Mean 10.1 8.20 8.64 8.83 8.42 7.47 8.57 . Number 896 2,535 1,518 1,941 3,420 512 10,822 Female . Mean 8.3 7.0 6.99 7.24 6.54 6.57 7.17 . Number 875 2,684 1,518 1,848 3,429 536 10,890 Both sex . Mean 9.21 7.58 7.82 8.05 7.48 7.01 7.87 . Number 1,771 5,219 3,036 3,789 6,849 1,048 21, 7i; (See definition strata's codes in Chapter 3, Table Annex 1). 110 Table 3.2.10. shows the data on occupational status of the population. Because of the extreme youthfulness of urban population in Zaire, a large manpower is available. But, as it is emphasized by Houyoux and Kinavwidi (1986), jobs in urban Zaire are not abundant. These data reveal, for example, that the working population represents 14%, housewives 13.6%, population still going to school represents 35.1%, while the inactive population, made up essentially by children less than 6 years and others, represents 29.5 % of the population. The category "other" (which includes those who lost their jobs, those who never worked, retired, etc.) represents only 7.8 percent. The gender distribution of these percentages indicate that male labor force participation is 3 times higher than female labor force participation. Further, Table 3.2.10. reveals a high dependency ratio in urban Zaire. On average, one working adult supports about 6.14 persons. Table 3.2.10.: Distribution fin %1 of Urban Population bv Occupational Status and bv Gender in August 1987 Students Housewives Working Inactive Others Total Male 38.4 — 21.8 29.2 10.6 100 Female 31.7 27.1 6.2 30.1 4.9 100 Total 35.1 13.6 14 29.5 7.8 100 Source: Ngondo a Pitshandenge et al. , 1988, op. cit. 111 3.2.2. Socio-Economic Characteristics of the Population Here, I will present the data on income and purchasing power of the urban population surveyed. For the 13 cities studied, the indicators of income and purchasing power are valid for the two months of field research, i.e., July and August 1987. Table 3.2.11. gives the structure of the household revenues ("recettes") in urban Zaire. This table indicates that household revenues are mainly made up of family enterprises which, represent about 63% of the total revenue. The salary represents only 12.6% of the total household revenue. Another substantial part of the revenue comes from "loan" obtained from friends or other institutions. Unexpectedly, the social solidarity participates only a little in the household revenue (7.6%), giving some clue that solidarity among members of the extended family is disappearing in urban area in Zaire. Table 3.2.12. gives the structure of household consumption expenditures in July-August 1987. It shows clearly that almost half of the household income is expended for food (49.55%). Housing expenditures makes one-fourth of the total expenditure (26%) while clothing takes about 12 percent. On average, the expenditures for household consumption are six times higher than the average salary received monthly by each family. In addition, Table 3.2.12. shows that transportation, education, and health take little money from the household 112 budget, costing, on average, 8.78% of the total household budget. Table 3.2.11.: Structure of Household Revenues ("recettes1 ') Julv-Auqust 1987 fin Zaire Currency) Category of Returns Number of Households Average Returns (per month) % . Salary or wage 6453 3,160.11 12.6 . Family Enterprises (or family business) . Solidarity( ' 6464 15,721.39 62.9 6453 1,897.21 7.6 . Loan ("Emprunt") 6464 2,771.20 11.1 . Drawing out 6453 1,458.13 5.8 . Total Incoming 6453 25,022.35 100.0 (*) This return is made up of presents, rewards, etc., in money and in kinds received from family or friends. Another structure of the household consumption expenditures shown in Table 3.2.13. is the inequality in the distribution of income in urban Zaire. This table indicates that 99% of the households share only 68% of the total household consumption budget, meaning that less than 1% of the household are sharing 31% of the total household consumption budget. 113 Table 3.2.12.: Structure of Household Consumption Expenditures Julv-Auaust 1987 (in Zaire Currency) Category of Expenditure Average Expenditure (per month) % . Food 9,687.15 49.55 . Housing 5,100.97 26.09 . Clothing ("Habillement") 2,359.56 12.06 . Miscellaneous: - Transportation 549.9 2.8 - Health 630.39 3.2 - Education 536.44 2.7 (Total Miscellaneous 1,716.73 8.78) . Others 684.95 3 .50 . Total 19,549.36 100.00 Table 3.2.13.: Distribution of Income in Julv-Auaust 1987. PERCENTAGE OF HOUSEHOLDS PERCENTAGE OF TOTAL INCOME SHARED 10.00 0.45 20.00 1.29 25.05 1.82 40.03 4.10 50.03 6.46 75.00 18.40 80.00 22.95 90.00 37.08 95.00 49.36 99.004 68.84 0.996 31.16 -► Total number of households: 6,428 -► Total household consumption expenditure in July-August, 1987 (in Zaire currency): 104,296,305 -► Gini concentration ratio: + 0.7375 -► Index of Concentration: 58.39 % 114 3.2.3. Housing Conditions In this section, I will discuss the housing characteristics of the households identified in this survey. Specifically, five points will be discussed: the distribution of households by number of rooms, housing quality, housing ownership, housing comfort, and environmental condition of immediate habitat. Table 3.2.14. gives the distribution of households by the number of rooms. It shows that the housing conditions in the 13 cities in 1987 were not highly decent: About 70% of households live in the houses containing only 2 to 4 rooms (living room and bedroom included), about 7.3% of the households live in houses with only one room, only 21.7% of households live in the houses containing at least 5 rooms. Table 3.2.14.: Distribution of Households Tin %) bv Number of Rooms Number of Distribution of Rooms Households Available(in %) 1 7.3 2 27.0 3 26.1 4 17.9 5 10.8 6+ 10.9 Total 100 Source: Ngondo a Pitshandenge et al., 1988, op. cit., Annex 2.09A, p.163. 115 The second point of discussion concerns the quality of housing, that is the quality of materials used to build the ground, walls, and roofs of the houses which shelter the households surveyed. Table 3.2.15. describes these materials. It indicates that the majority of the households (54.8%) live in houses with dirt floor, 44.0% of households live in houses with cement floor, only 1% of households lived in houses with tile floor. The walls of houses of the households are made with adobe or clay bricks (42.6% of households), hard bricks (39.8% of households), adobe or clay walls (15.7% of households), wooden wall (1.3%), and grass (0.6%). In terms of roofing, a large majority of households live in houses with metal sheet (63.1%), 16.4% of them lived in houses with reused metal sheet, 13.7% in grass houses, and 2% in houses with tiles. These data indicate that a large majority of households identified in this survey are not secured from the inclemency of weather or from infectious diseases whose host is soil, air, or rain water. About the ownership of houses, Table 3.2.16. indicates that more than half of the households (53.3%) own the housing unit they live in, 32.3% are tenants of the houses they occupy. Households which are housed by employers represent only 7.3%, those housed free by a family member or a friend account only for 6.9% of the households. These data confirm two facts observed above. Firstly, they explain why renting or housing cost is not the first sector in the household 116 expenditures in urban areas. Secondly, they confirm the observation that, in urban areas where money is now the dominant way of exchange in daily life, African solidarity is becoming less and less practiced. Table 3.2.15.: Distribution of Households (in %) bv Types of Materials Used to Build the Ground. Walls, and Roofs. GROUND % WALLS % ROOFS % Dirt Floor 54.8 Hard Bricks 39.9 Metal Sheet 63.0 Cement Floor 44.0 Adobe Bricks 42.5 "Eternit'' 4.0 Tile Floor 1.0 Adobe Walls 15.7 Reused metal Sheet 16.5 Others 0.2 Grass VO • o Total 100 Wooden walls 1.3 Grass 13.7 Others 0.1 Tile 2.0 Total 100 Others 0.9 Total 100.0 (*) These are bricks made of cement or burned-building-bricks. Table 3.2.16.: Distribution of Households bv Ownership Status of the House Ownership Status % Owners 53.4 Tenants 32.3 Housed by Employers 7.3 Free (*) 6.9 Other 0.1 Total 100 (*) Households Housed by family members or friends. Five variables are used to assess the comfort of housing: 1) the location of latrine, 2) the mode of use of the latrine, 117 3) the distance between the latrine and the housing unit, 4) source of drinking-water, and 5) the distance between the housing and the source of drinking-water. The data related to latrine (Table 3.2.17.) indicate that very few households have latrine in the house (9.7%), the majority of them have the latrine in the compound but outside the housing unit (83.4%); 4.4% of the households use a latrine located outside the compound. A total of 2.5% of households have no latrine at all. Table 3.2.17. also indicates that of the 97.5% of households which have a latrine, 61% of them use a latrine that is individuals, i.e., a latrine used exclusively by the members of the family, but 36.8% of the households use a collective latrine, i.e., a latrine used by several households. It is clear that these 36.8% of households live in an uncomfortable type of housing when one considers only the latrine mode of use. Here, I have to remind that the average size of a household in the 13 Cities surveyed is 6.52 persons. Thus, one can imagine the difficulties that members of a household will endure when they have to share a latrine with two, three, or more other households per day because this will concretely mean that a collective latrine will be shared by 13, 19.56 or more people per day. Another indicator of comfort of housing is the distance separating the house and the latrine, especially for households which do not have a latrine within the house. Table 3.2.17. shows that about 40% of the households use a latrine 118 located at more than 10 meters from the housing unit. About 60% of the households live at less than 10 meters to the latrine. This latter observation has two implications. In households whose latrine is individual, its closeness to the house may not be so dangerous if I assume that individual latrines are cleaner than common ones. The second implication concerns the households whose latrine is common. In this case, its closeness to the house may be assumed to be dangerous for the users because common latrines are more likely to be unclean than individual ones. Table 3.2.17.: Distribution of Households bv Location. Mode of Use and Distance Between the Latrine to the Housincr Unit. LOCATION % MODE OF USE % DISTANCE(*) % •In the house 9.7 Individual 60.5 .Outside the House, Common 36.8 83.4 No Latrine 2.7 .Outside Compound 4.4 Total 100 •No latrine 2.7 .Total 100.0 < 5 meters 5-7 meters 7-10 meters > 10 meters Total 22.4 20.6 17.9 39.1 100 (*) This concerns only households which have latrine outside the home but inside the compound and those which use a latrine located outside the compound. They represent 543 households. Housing comfort can also be assessed by the principal source of supply of drinking-water and the distance separating the house from this source. About the source of drinking- water, data reveal (Table 3.2.18.) that only half of urban households in Zaire (50.8%) are connected to the modern water supply system. The rest of the households (49.2%) are not 119 linked to this system, they get their drinking-water from different sources: spring (11.1%), street fountain (21.3%), well/lift and force pump (14.6%), river or rain or pool, etc. (2.2%). On the issue of distance between the housing and the source of drinking-water, Table 3.2.18. shows that many households get their drinking water from a long distance. It is shown, for example, that only 14.7% of total households have this source installed in the house, 24.6% of them get drinking water from a distance of less 10 meters, about 34% of households get water at the distance averaging 10 to 80 meters, more than one-four of the households (26.8%) get the drinking water at a distance more than 80 meters. What these data imply is that urban residents in Zaire still face a high risk of getting infectious diseases from contaminated water. In fact, drinking water may be contaminated at different levels. It may be contaminated at the source of supply. Here, 49.2% of total households surveyed run the risk of getting the water polluted at the source. Further, drinking water may be infected while carrying it to the house. Here, 85.3% of households are at risk. 120 Table 3.2.18.: Distribution of Households bv Source of Supply of Drinking-Water and Distance of the Su ppIv and the House SOURCE OF SUPPLY % DISTANCE (in meters) % Water-conveying system 50.8 Less than 10 24.6 Spring 11.1 10-20 13.4 Street fountain 21.3 20-30 7.4 Well/lift and force pump 14.6 30-40 4.6 River or rain 40-50 1.8 or pool 2.2 50-80 6.7 80-100 15.3 Total 100 100 or more 11.5 In the house 14.7 Total 100 Other indicators of sanitation include the environmental decency of immediate habitat. Specifically, two indicators are included in this dissertation to assess this decency. One is the treatment given to human waste (feces and urine), another is the treatment given to household waste. The data indicate that 97.8% of households in urban areas of Zaire use a latrine as their principal place to deposit their feces and urine, the rest of households (2.2%) use the bush ("brousse") or river, etc., as their principal place to deposit feces and urine. The latter percentage corresponds roughly to the percentage of households which do not have a latrine (see Table 3.2.17.). Members of these households, regrettably, contribute in rendering unhealthy the immediate environment of the habitat. Even among the households which use latrine to deposit their human wastes, Table 3.2.17. shows that more than one-third of them use a common latrine, a condition which increases the 121 chance of spreading out the infectious diseases that originate from feces or urine since common latrines are more often unclean. The data on the treatment of household wastes also indicate that the immediate habitat where the households live in is unhealthy. For example, Table 3.2.19. shows that 35.1% of households throw away their waste on the ground (surface), 41.4% of them throw away their waste in a hole which is often not covered, finally only 13.6% of households use a garbage- can to throw away their waste. Unfortunately, Table 3.2.19. indicates that these garbage-cans are not usually covered either while in Table 3.2.20. it is shown that the distance separating the house from the place where household wastes are thrown is very short. The evidence indicates, for example, that 49.1% of households throw away their waste at less than 15 meters from the house, only 42.9% of them have their garbage-can or hole, etc., located 15 meters or more from the house. Table 3.2.19.: Distribution of Households bv Mode of Human and Household Wastes HUMAN WASTES % HOUSEHOLD WASTES % GARBAGE-CAN OR HOLE % Latrine Bush River Others Total 97.8 0.6 0.2 1.4 100.0 Garbage-can 13.6 Hole 41.4 Surface 35.1 Others 9.9 Total 100 Covered 3.7 Not Covered 51.3 Surface + Others 45.0 Total 100 Source: Ngondo a Pitshandenge et al., 1988, op. cit. 122 Table 3.2.20.: Distance between the House and the Place Where Household Wastes are Thrown DISTANCE (in meters) % Less than 5 19.7 5-10 14.6 10-15 14.8 15 or more 42.9 No Place 8.0 Total 100.0 Source: Ngondo a Pitshandenge et al. , 1988, op. cit. 123 Map 3.1.1.: The Official Map of Zaire in 1988 © T\ Anao • - HAUJ^ELE ZONGOC Dungu Libenge YaKoma Djugu^ EQUATEU Lukolela \ /■ ' \ W»lun«M! JA S n a D u n d , ( Vi Panel I MVmtfl*,' M AI-NDOM BE: baS d u n d u % % KINSHAS^X 1 SANKURUf D « k 6 M ibombo Bagata SUO-KIVU s ^ » • , ...KW ILU JP' • \BuJung_u Idiof L u k u la iP .X ;. KifrfvuftC • M a n im b * --:* _ • tfty liiB a n z a NgungU tManifnbfc 0 andadlS C M pI olo C fa d im 6a*^ojib* f b ^ Dimbeien CARTE OE [.’ORGANISATION ADMINISTRATIVE EN 1988 0_100_ ^ 00_ BJi0p km LEGENDE L im it* d * p ay * L im it* d* rdgloii . . . . . . . . . L im it* d * s o u s -rtg lo n ........................ L im it* d * zo n * MBANDAKA N o m d * V lll* ..^HjHAUT-i'Bukw.,* Santfoa / L O M A M I /w it a r' L V WIMWI I f Mitwabp LUALABA"', «-.^>|*l^*.v-/haut-SHABA V i. Lubudl % n M U i- t J j Oilolo VOLW EZId \ ^ ^ . . K u a n g a l WEZU r KambovM<.<* LUBlrfrfBASHI • MBUJI- TSHILENGE K0L’ 1 - Tahllanga 2 Kabeya Kamwanga 3 KatarxJa 4 Lupatapata 6 Mlabi ^\Sakanla« \ I V-N| L . d* SAINT MOULIN D’apris les documents de l'lnstitut Gtographique du Zaire et les textes actuellement en vigueur. Source: de Saint Moulin s.j., Leon, 1988, "Histoire de 1'organisation administrative du Zaire," ZAIRE AFRIQUE, 28e annee, No.224, p . 2 7 . 124 Table Annex 1: Criteria of Stratification of Primary Sampling Units (quarters, blocs, cellules, etc.l STRATA 0: "Upper Class Neighborhoods" It is composed of all high standing neighborhoods in terms of housing quality; it is composed of neighborhoods registered in the land-register; these localities are provided with hygienic installations, sewage system; they are connected to the system of water supply; they have electricity; houses are built in hard materials; roads are tarred or bituminous and accessible; residents belong to high class and high professional category. STRATA 1: "Old Neighborhoods:" This strata is made up of neighborhoods which were there in the beginning of the city. They are pre-colonial. These quarters are registered in the land-register, but they do not have hygienic installations; they are not connected to system of water supply; some roads are tarred, but all roads are accessible; residents have an average standard of life. STRATA 2: "New Neighborhoods:1 1 . This strata groups new neighborhoods created some time before independence in 1960, but they were not in the beginning of the city; these quarters are registered in the land-register; there are made of large compounds; usually they are provided with hygienic installations and water supply; some roads are tarred, but all roads are accessible by car; residents have average standard of living. STRATA 3: "Planned Neighborhoods:" These quarters are registered in the land-register; they are provided with hygienic installations, water supply, and electricity; roads are tarred and accessible; residents above middle-class. STRATA 4: "Extended Neighborhoods:" These quarters are not registered in the land-register; these are neighborhoods born after the independence; they are not usually linked to the water supply-system; they do not have electricity; most of the roads are not accessible and not tarred; houses and compounds are small. STRATA 6: "Eccentric Neighborhoods:" . These are neighborhoods located outside the center of the city; they are not allotted and do not have the basic hygienic installations; they are not provided with electricity or drinking water. They resemble the extended neighborhoods, the only difference is that they are isolated from the city; they do not have accessible roads. Legend: Strata 5 is made up of "Villages" which surround the city; while Strata 7 is made up of military bases. Both Strata were not included in the survey. 125 Table Annex 2: Distribution bv Strata and bv City of Primary Sampling Units (quarters, blocs, cellules) in 1987. STRATA 0 1 2 3 4 6 Total MATADI 17 140 28 13 — — 198 BOMA 22 163 7 14 13 56 275 BANDUNDU1 — — — — — — — KIKWIT 7 57 56 1 32 2 155 MBANDAKA 27 27 53 10 30 13 160 ZONGO — 46 41 — — — 87 KISANGANI 14 62 7 43 38 21 185 BUKAVU 23 — — 46 77 — 146 LUBUMBASHI 27 8 61 29 86 — 211 LIKASI 7 10 22 27 16 — 82 KOLWEZI 8 — — 39 9 4 60 KANANGA 5 36 15 16 29 — 101 MBUJI-MAYI 7 2 19 315 344 ALL 164 551 290 257 645 96 2,003 Legend: — means that the figure does not exist. (A) For this city, the supervisor did not send back the table. Source: Unpublished information gathered by Technical Team in 1987. 126 Table Annex 3: Distribution bv Sub-Strata and bv Citv of Total and Sampled Primary Sampling Units, and Sampling Fractions at 1st. 2nd Stages, and Global. Number of Primary Sampling Units Sampling Fractions Sub-Strata Total Sampled 1st. 2nd Global MATADI < 1,000 inhab1 139 17 1 8.18 1 4.42 1/36.14 1,000 or more 59 22 1 2.68 1 13.5 1/36.14 BOMA < 500 inhab 184 15 1 12.3 1 3.38 1/41.50 500-1,000 61 15 1 4.07 1 10.2 1/41.50 1,000 or more 30 20 1 1.5 1 27.7 1/41.50 BANDUNDU < 500 inhab 275 36 1 7.64 1 7.14 1/54.57 500 or more 39 13 1 3.00 1 18.2 1/54.57 KIKWIT < 1,000 inhab 80 14 1 5.71 1 10.5 1/59.97 1,000 or more 75 21 1 3.57 1 16.8 1/59.97 MBANDAKA < 1,000 inhab 89 6 1 14.8 1 3.81 1/56.56 1,000 or more 71 16 1 4.44 1 12.7 1/56.56 ZONGO < 20 inhab 46 17 1 2.71 1 7.63 1/20.68 2 0 or more 41 30 1 1.37 1 15.1 1/20.68 KISANGANI < 2,000 inhab 140 27 1 5.19 1 9.46 1/23.27 2,000 or more 45 21 1 2.14 1 7.76 1/23.27 BUKAVU < 1,000 inhab 79 11 1 7.18 1 4.39 1/31.51 1,000 or more 67 24 1 2.79 1 11.3 1/31.51 LUBUMBASHI < 1,000 inhab 27 2 1 13.5 /5.2 1/70.20 1,000-2,000 71 9 1 7.89 /8.9 1/70.20 2,000-3,500 65 14 1 4.64 1 15.2 1/70.20 3,500 or more 48 26 1 1.85 1 38.2 1/70.20 LIKASI < 2,500 inhab 42 7 1 6 1 9.12 1/54.74 2,500 or more 40 14 1 2.86 1 19.1 1/54.74 Legend: Inhab.= inhabitants 127 Table Annex 3 (continued) Number of Primary Sampling Units Sampling Fractions Sub-Strata Total Sampled 1st. 2nd Global KOLWEZI < 2,500 inhab1 38 8 1/4.75 1/9.23 1/43.85 2,500 or more 22 13 1/1.69 1/25.9 1/43.85 KANANGA 2,500 inhab 60 11 1/5.45 1/8.69 1/47.34 2,500 or more 51 25 1/2.04 1/23.2 1/47.34 MBUJI-MAYI < 1,000 inhab 101 6 1/15.8 1/4.42 1/69.67 1,000-2,000 134 16 1/8.33 1/8.36 1/69.67 2,000 or more 109 7 1/14.6 1/4.78 1/69.67 TOTAL 2,328 483 --- --- --- Legend: Inhab.= Inhabitants Source: Ngondo a Pitshandenge et al., 1988, op. cit., Annex 1.03A, p.145. (The last row is added by me). 128 Chapter 4 INFANT AND CHILD MORTALITY IN URBAN AREAS OF ZAIRE: LEVEL AND DETERMINANTS This chapter includes two sections. The first section estimates the level of infant and child mortality in urban areas in Zaire. The second section investigates the role of maternal education on infant and child mortality, using muIt ivar iate techn iques. 4.1. LEVEL OF INFANT AND CHILD MORTALITY IN URBAN AREAS OF ZAIRE IN 1987 To estimate the level of infant and child mortality in urban Zaire among the birth cohorts 1981-85, I will use direct and indirect procedures. Table 4.1.1. in appendix gives the descriptive statistics of direct estimations derived from the distribution of births and deaths among birth cohorts 1981-85 in all the 13 Zairian Cities surveyed in 1987. This table indicates that infant mortality in urban areas during the study period was 77.4 per thousand live births while the under five mortality rate was 93.2 per thousand live births. This Table also gives the distribution of births and death cross tabulated by socio-economic, demographic and contextual characteristics of mother, child and place of residence. The unique observation which can be drawn from Table 4.1.1. is that child mortality rates in urban areas of Zaire are not 129 constant along the socio-economic, demographic and contextual factors related either to the mother or to the child him/herself. For mother-related socio-economic factors, it is clear that child death rates among the birth cohorts 1981-1985 in urban areas in Zaire vary with education, occupational status, marital status, and place of delivery. About these factors, Table 4.1.1. clearly reveals that child mortality rate is increased by low education of mothers. In fact, the death rate among children of mothers with at least a secondary education, i.e. 7-9 years of schooling or 10 years or more, is 81 and 74 per thousand live births, respectively, while the correspondent rates for children whose mothers are illiterate or primary school educated (1-3 or 4-6 years of schooling) are 97, 114, and 104 per thousand live births, respectively. Further, Table 4.1.1. indicates that child death varies with mother's occupation: high risks of child death are associated with student mothers (137 deaths per thousand live births), mothers who never worked or who are inactive for physical disability or prostitutes (144 per thousand), while low mortality among children is associated with housewife and working mothers (child death rates are 89.8 and 92.5 per thousand, respectively). Childhood mortality in urban areas in Zaire also varies with marital status, mother's gestational status while pregnant, and the place where the delivery occurred. About marital status, Table 4.1.1. shows that high risks of child 130 death are associated with single mothers (136.7 per thousand) , mothers in consensual unions, divorced, widowed, etc., (144.2 per thousand). About mother's gestational status, two indicators are displayed in Table 4.1.1. First, is the length of pregnancy. Here, mothers who delivered premature babies (length of pregnancy less than 9 months) experienced a high risk of child death. This risk is twice higher than when the baby was delivered at 9 months; that is 177 against 88 per thousand. The second indicator of mother nutritional status during pregnancy is reflected by birth weight (in grams). Table 4.1.1. shows clearly that low birth weight (less than 2,500 grams) is associated with high risk of child mortality (115.5 per thousand) while low risks are associated with normal birth weight (2,500-3,499 grams and 3,500 grams or more), 82.9 and 74.2 per thousand, respectively. Finally, child mortality rate varies with the place where the delivery took place; when the mother delivered in health institutions (maternity, health centers, etc.), the risk of child death is the lower (84.6 per thousand) than when she delivered at home (142.9 per thousand). Again child death rates vary with the household income. Data on household pattern of expenditures in July-August 1987 allow the grouping of households in four classes of household income: (1) class below the first quartile (4,730 zaires), (2) class between 4,730 and 8,936 zaires, (3) class between 8,937 and 17,268 zaires, and (4) class of household income above 131 17,269 zaires. Mortality rates in those classes of income are 104.6, 96.6, 76.2, 95.9 per thousand, respectively. What these statistics reveal is that poor households have high risks of child death but this risk is not necessarily low in "rich" households. Here, it will be remembered that, in contrast to data on other socio-economic factors of the mother, which are related to the time of birth of the child, data on household income are not related to the time of child death; instead, they are related to the time of survey. Table 4.1.1. also indicates that physiological and social characteristics of children at the time of birth have an impact on their survival chance. Among these factors, the most important one is birth order. Clearly, our data reveal that the risk of dying among children born in urban areas in Zaire between 1981 and 1985 decreases with birth order. For example, first births have the highest risk of death (113.7 per thousand), while childhood mortality rate is 104.9 per thousand for birth order 2-3, and 84.1 per thousand for birth order 4-6, and 72.7 per thousand for birth order 7 or more. As shown in Table 4.1.1. there is not a substantive difference in childhood mortality by gender in urban areas of Zaire among birth cohorts 1981-85. In fact, the risk of dying in these cohort is 92 per thousand for males, and 93 per thousand for females. Direct estimations displayed in Table 4.1.1. also give variation in child death rates by the type of residence 132 (strata) of the neighborhood. They clearly show that neighborhoods located in areas where modern facilities are available, e.g., tarred routes, electricity, adduction in potable water, sewage systems and other hygienic installations, etc., represented in our data by "Strata 0" and "Strata 3," have low child mortality 46.5 and 75.04 per thousand (an average of 66.4). Child death rates in neighborhoods with average standing in terms of modern facilities are 101.3 per thousand for "Strata 1" and 84.0 pr thousand for "strata 2" (with an average of 94.9 %. for both) . In squatter neighborhoods, composed of "strata 4" and "strata 6," child death rates are 107.7 and 118.6 per thousand respectively (with an average of 109.1 for both). Table 4.1.2.: Proportion of Children Dead Among the Ever Born by Mother's Age and Years of Schooling. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. Age Group YEARS OF SCHOOLING at the time _______________________________ of Survey Illit. 1-3 4-6 7-9 10+ ALL 15-19 (*) .09459 .10811 .12429 .06881 .04348 .09058 20-24 .12500 .12281 .12531 .07340 .05660 .09468 25-29 .12717 .11187 .10417 .08104 .07590 .09653 30-34 .12903 .12667 .09652 .08699 .06832 .10067 35-39 .12680 .12205 .11450 .07715 .07708 .10803 40-44 .14531 .11533 .11699 .07287 .06849 .11797 45-49 .14570 .13932 .12094 .07693 .10983 .13530 All .13622 .12383 .11124 .07896 .07234 .10830 Legend: Illit. =Illiterate; (*) This age group, figures are very small at each school level. 133 Table 4.1.2. gives the proportion of children dead among children ever born by age of mother and years of schooling in urban areas of Zaire for birth cohorts 1981-1985. It allows the computation of indirect estimations of infant and child mortality in urban areas of Zaire using the information on children ever born, children surviving, and parity of mothers by age of mother at the time of the survey. The techniques for computing these estimations are explained in detail in Manual X of the United Nations publication series (1983). From the data displayed in Table 4.1.2., the following probabilities of dying from birth to age X are computed in Coale-Demeny life table models. These probabilities are presented in Table 4.1.3. Table 4.1.3.: Probability of dying from birth to aae X fin years). cohort of births 1981-1985. Age: North South East West 1 92.583 89.559 96.396 95.072 2 95.296 99.473 99.793 99.642 3 93.161 99.045 97.822 97.387 5 100.762 103.938 102.354 102.574 It is worth noting that the indirect estimations of infant mortality, i<3 q' are obtained from proportions of children dead reported by women aged 15-19. As can be seen in Table 4.1.2. these latter proportions are computed from very 134 small numbers. In addition, it has been reported elsewhere that indirect estimations, especially 2^0' 3^0' obtained from these techniques from indirect are often unreliable. Thus, I will use the probability of dying from birth before the fifth birthday as an indicator of entry to estimate infant mortality in Coale-Demeny Life tables models. The results for these latter operations are shown in Table 4.1.4. below. Table 4.1.4. Estimated Levels of Infant Mortality using 5aQ as an Indicators of entry in Coale and Demenv Life Table Models. Life table Models: Levels Entry used (5q0) (per thousand) Estimated IMR (iq0) (per thousand) North South East West 17-18 19-20 18-19 17-18 100.762 103.938 102.354 102.574 65.06 78.09 67.945 73.88 Table 4.1.4. indicates that the estimated levels of infant mortality in urban areas of Zaire among birth cohorts 1981-87, using the indirect approaches, vary in the range 65- 78 per thousand, depending on the region of model chosen, while under five mortality rates varied in the range 100.8- 103.9 per thousand. 135 4.2. MATERNAL EDUCATION AS A DETERMINANT OF INFANT AND CHILD MORTALITY IN URBAN AREAS IN ZAiRE IN 1987 4.2.1. The Method To investigate the role of maternal education in infant and child mortality in urban areas of Zaire among children born in cohorts 1981-1985, I will use a logistic regression model. The logistic regression model is the log-odds, i.e., the log of the ratios of the probability of dying to the probability of surviving (p/(l-p)), that relates the dependent variable— here the ratio of the odds of dying to the odds of surviving— and a set of explanatory variables. It takes the following form ln(Pjy (1-pj^)) = « + 6;iXi;L + 62xi2 + fi3Xi3 + ••• + 6kxik Where p^ is the probability of dying for the ith individual, and « is the intercept parameter, R^, R2, S3, ••• 6^ are the parameter estimates or regression coefficients. Finally xil' xi2' xi3' *•* xik the set of k covariates or independent or explanatory variables. This relationship is also described as logit(p^). To perform the logistic regression, I will adopt the stepwise and forward strategy which consists in including the explanatory variables one by one. At each step, I record the 136 likelihood estimates, and the chi-square statistic. Since the logistic regression model has been extensively described in other works such as Halli & Rao (1992), Gujarati (1988), Aldrich & Nelson (1984), SAS/STAT User's Guide, Vol. 2, Version 6 (1990), etc., I shall concentrate here on the results of all the operations. Before presenting the results of these operations, I will first introduce the variables included in this dissertation. 4.2.2. The Variables Included in the Study and Results of Multivariate Regressions Table 4.2.1. describes the explanatory variables included in my study. A total number of 14 variables previously identified in the literature as capable of mediating the effect of maternal education on infant and child mortality have been included in my dissertation. Table 4.2.1. also describes the interaction effect between education and some socio-economic variables discussed in the literature review. Table 4.2.2 and 4.2.4. (in Appendix) give the results of the multivariate technique relating infant and child death in urban areas of Zaire in the birth cohorts 1981-1985 to the maternal education, controlling for social, demographic determinants of mothers and/or child, and environment factors. For infant death, Table 4.2.2. reveals that mother's schooling does not significantly affect death before one year in the urban areas surveyed, after controlling for other socio economic and demographic factors. In all the 16 models defined 137 does not significantly affect death before one year in the urban areas surveyed, after controlling for other socio economic and demographic factors. In all the 16 models defined in this table, education has not been shown significantly associated with infant death, except in models 5, 6, and 8, with p-values of 5% in the first two models, and 10% in model 8. On average, Table 4.2.2. indicates that a one-year increase in maternal schooling is associated with a decrease of infant mortality of 1.33 to 3.96 per cent, though the association is not significant in all the models. A positive association between infant mortality and maternal education is displayed in four models but these associations are not only very close to zero (0.03 to 0.9 per cent) but also they are not significant at all. In addition, Table 4.2.2. indicates that household income is not at all significantly related to infant death in urban areas in Zaire. But unlike maternal education, household income displays a positive association with the odds of dying in infancy, controlling for other maternal and/or infant factors. If maternal education and household income taken separately or taken together did not influence significantly the survival outcome in infancy of children born in cohorts 1981-85 in urban areas of Zaire, what then determine this outcome? A close look at of Tables 4.2.3. and 4.2.5., which are drawn from Table 4.2.2., reveals that there are four key 138 factors of infant mortality in urban areas of Zaire. These determining factors for infant death in urban Zaire are gender, age of the mother at birth of child, birth weight, length of pregnancy, and number of bedrooms in the housing unit of the household. (Table 4.2.3. displays the average values of odds ratio, a summary of all the 16 models of Table 4.2.2., while Table 4.2.5. gives the last model of the Table 4.2.2.). 139 Table 4.2.3.: Average Odds Ratios of Significant Factors of Infant Mortality in Urban in Zaire, 1987 FACTORS MEAN ODDS RATIO SIGNIFICANCE LEVEL (in %) DECLINE OR INCREASE (in %) GENDER:1 . (Male) 1. 000 (--) (--) . Female 0.788 10 -21.2 MOTHER AGE AT BIRTH (in years): . ( <20 ) 1.000 (--) (--) . 20-24 0.836 NS -16.4 . 25-34 0. 530 1 -47.0 . 35 + 0. 617 NS -38.3 BIRTH WEIGHT (in grams) . ( < 2,500) 1.000 (--) (--) . 2500-3499 0.436 0.1 -56.4 . 3500 + 0.443 0.1 -55.7 . Unknown 1.137 NS +13.7 LENGTH OF PREGNANCY: . (9 months) (1.000) (--) (--) . < 9 months 2.815 0.1 +181.5 COMMUNICABLE & RESPIRATORY DISEASES: . Number of 0.884 1 -11.6 bedrooms Source: Table 4.2.2. Legend: (****) p <= o.OOl; (***) p <= o.Ol; (**) p <= 0.05; ( ) p <= 0.10; N.S.=Not Significant. 1 Gender is not a strong significant factor of infant mortality. But its level of significance has been maintained in all the models. (See Table 4.2.2. in Appendix). Among these four factors, the most significant ones are associated with mother's nutritional status during pregnancy. Two proxy variables of nutritional status of the mother during pregnancy are included in this study. One is birth weight. Table 4.2.3. (also Table 4.2.5.) indicates that the odds of dying for babies born with normal birth weight (2,500-3.499 grams) as compared to those born with low birth weight (below 2,500 grams) is, on average, 0.436. This is statistically significant at 0.1 per cent. In addition, for babies born with weight above 3,500 grams ("overweight babies"), the odds of dying in infancy is, on average, 0.443, compared to babies born with low birth weight. Here again the difference is significant at 0.1 per cent. The other proxy of maternal nutritional status during pregnancy is captured by the length of pregnancy. Like the birth weight, Table 4.2.3. (also Table 4.2.5.) displays that prematurity (length of pregnancy less than 9 months)— compared to full-term pregnancy (9 months)— is significantly associated with high infant death; odds ratios vary from 2.737 to 2.935. On average, infants born premature are 2.815 times more likely to die before the first birthday than those born full-term. The difference is also significant at 0.1 percent. Mother's age at birth of the infant is another factor found significantly associated with infant death in urban areas in Zaire. Table 4.2.3. clearly indicates that the risk 141 of dying in infancy decreases when the age of mother at birth of infant increases. On average, infants born to mothers of age 25-34 years are 0.530 times less likely to die in infancy than infants born to teenager mothers. The difference is significant at 1 percent. For infants born to mothers whose age groups surround 25-34, i.e., 20-24 years and 35 years or more, their risk of dying before the first birthday is lower than for teenager mothers, but the differences are not always significant. The odds ratios are, on average, 0.836 and 0.617 for mothers 20-24 and 35 years or more, respectively. Thus, it is clear that in urban areas in Zaire, being born to a teenager mother presents for the baby the highest odds of dying before the first birthday. Table 4.2.5. below, which presents only the last model of the Table 4.2.2., is consistent with these comments. About infant gender, Table 4.2.3. shows that, on average, females are 0.788 less likely to die in infancy that males. But the difference is significant at 10 percent only. The only socio-economic factor which significantly influences infant death in urban areas in Zaire is the number of bedrooms available in the housing unit where the family lives. The evidence displayed in Table 4.2.3. (also in Table 4.2.5.) reveals that an increase in number of bedrooms by one- unit deflates the odds of dying before the infancy (which is 0.884) by 11.6 percent. This is highly significant at 1 142 percent. The number of bedrooms in the house is probably capturing the effect of infectious, respiratory and communicable diseases such as measles, which have been reported as strongly associated with the crowding within the household (Aaby, 1989). Chart 4.2.1. shows the significant paths from the proximate factors defined in Table 4.2.1. to infant mortality in urban Zaire. Chart 4.2.1.: From the Proximate Determinants to Infant Death in Urban Areas Zaire: The Significant Paths. INFANT DEATH BIOLOGICAL FACTORS ASSOCIATED WITH THE INFANT: . Female SOCIO-ECONOMIC FACTORS: . Number of bedroom in the house BIO-DEMOGRAPHIC FACTORS ASSOCIATED WITH THE MOTHER: . Age at birth . Nutritional status during pregnancy: * Birth weight * Length of pregnancy 143 Table 4.2.5.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Infant Mortality in Urban Areas of Zaire. The Last Model. VARIABLE NAME 17 INTERCEPT «JU *|f J U «JU ( 0.062 (0.708) GENDER . (Male) . Female 1 . 0 0 0 0.792 (0.130) MOTHER'S AGE AT BIRTH OF CHILD . (Less than 20) . 20-24 . 25-34 . 35 or more 1 . 0 0 0 0.909 ( o . i s u 0.600 (0.243) 0.703 (0.345) BIRTH WEIGHT (in grams) . (Less than 2,500) . 2,500-3,499 .3,500 or more .unknown 0.475 <0*19U ** 0.482 (0.218) 1.160 I (0.239) 1 LENGTH OF PREGNANCY: . (9 Months) . Less than 9 months 1.000_ I 4c 4c 4c 4c 1 3.012 (0.179) NUMBER OF BEDROOMS: 4c 4c 4c 0.888 (0.045) -2 LOG L CHI-SQUARE DF 2041.82 4c 4c 4c 4c 152.7 29 Source: Table 4.2.2.: Legend: (****) p <- q.qqi; (***) p <= 0.01; (**) p <= 0.05. The following factors -are not significant: mother's education, birth order, place of delivery, marital & occupational status, type of neighborhood of residence, quality of the house, and quality of the immediate environment of the house. 144 The evidence shown in Tables 4.2.3. and 4.2.5., and in Chart 4.2.1. reveals that in urban areas of Zaire many factors such as birth order, access to maternity hospital for delivery, marital and occupational status of mother, access to modernity, and even the sanitary conditions of immediate environment of the house, reported in the literature reviewed in Chapter 1 of this dissertation are not statistically significant. The latter factor— the sanitary conditions of immediate environment of the household— is a composite index which sums up the scores of seven variables: (1) localization of the latrine, (2) access to a latrine, (3) individual or collective use of the latrine,2 (4) source of water supply in the household, (5) distance of the household to this source, (6) management of wastes produced by the household, (7) care given to garbage-can or hole. The total score of this composite index is divided into three categories: category 1 when the score is in the class 1-6, category 2 when the score is 7-9, and category 3, when the score is in between 10-14. The fact that this composite index, which indicates the sanitary conditions of immediate environment of the household, is not a significant factor of infant mortality in urban areas of Zaire contradicts the studies by Pant (1991), DaVanzo, Butz 2 When a latrine is used by the members of one household, it is considered as individual. When a latrine is used by more than one household, it is considered as collective. 145 and Habicht (1983), Khan (1988), Jain (1985), and Meegama (1980),3 which have indicated that household characteristics, especially, source of water supply, access to toilet, access to electricity, etc., are significantly associated with infant mortality. However, the comparison of our results with those from the studies mentioned above is somewhat biased because previous studies have not used a composite index. To be consistent with the previous studies mentioned above, I included in the regression the individual variables which indicates the household conditions. Specifically, three variables have been considered in this new regression because they have been widely used in the studies reviewed in the literature (see Chapter 1). These variables are (1) source of water supply, (2) use of garbage-can, and (3) access to toilet. Table Appendix 4.2.2. indicates that all these variables, except the source of water supply, have no significant impact in lowering infant mortality in urban areas of Zaire. About the source of water supply, the evidence shows that its impact in lowering infant mortality in urban Zaire (with a p-value of 0.0824) is not as strong as it is in other countries mentioned in previous studies. However, our findings are consistent with the recent study by Majumber and Islam 3 The studies by DaVanzo, Butz and Habicht (1983), Khan (1988), Jain (1985), and Meegama (1980) were reviewed and cited by Pant (1991). 146 (1993), which has indicated that household water supply and toilet facility have no impact on the survival of the child in Bangladesh. Is maternal education significantly related to the death of the child before the fifth birthday in urban Zaire? In addition, what is the direction of this relationship? The evidence displayed in Table 4.2.4. (see Appendix) indicates that maternal education affects significantly child mortality in urban areas in Zaire (p-value < 0.1). Further, this evidence reveals that the relationship between maternal education and child mortality in urban Zaire is negative. All the models shown in Table 4.2.4. reveal that the change in the odds ratio corresponding to a one-year increase in mother's education is in the range of 0.939-0.967, with an average value of 0.950. The same evidence also shows that in urban Zaire household income is not significantly associated with child mortality, though the relationship is negative in every single model included. In addition, our results tell that the interaction maternal education-household income is not only insignificant but also its effect is not important in the association. The results presented in Tables 4.2.6. and 4.2.7. clearly indicate that unlike infant death which is determined mostly by demographic and/or biological factors, child mortality in urban areas of Zaire is dominated by social determinants. 147 (Table 4.2.6. gives the average values of odds ratio, the significance level, and the percent increase or decrease of significant factors of childhood mortality, while Table 4.2.7. presents the results of the last model of Table 4.2.4.) . Among these social determinants five are significant: (1) maternal education, (2) type of neighborhood of residence, (3) palce of delivery, (4) occupational status of mother, (5) and number of bedrooms in the house. The following factors have no significant impact in child death in urban Zaire, namely household income, child's gender, age of the mother at birth of child, birth order, birth weight, length of preganancy, marital status of mother, quality of house, conditions of immediate environment of the household, and all the interaction between maternal education and socio-economic variables. Both tables show that the single most important determinant of childhood mortality in urban Zaire is the type of neighborhood of residence where the family lives. It is clear that child mortality is lower in "high standing" neighborhoods (which include "upper class" and "planned" neighborhoods) and in "average standing neighborhoods" (which include extended and eccentric neighborhoods) as compared to squatter neighborhoods. However, the difference is only significant (at 0.1 percent) between high standing neighborhoods and squatter neighborhoods, indicating that, on 148 average, children who live in high standing neighborhoods are 0.547 times less likely to die before the fifth birthday than those who live in squatter areas. In other words, childhood mortality in high standing neighborhoods is 45.3 percent lower compared to squatter child mortality. Table 4.2.6.: Mean Odds Ratios of Significant Factors of Child Mortality in Urban Zaire, 1987 FACTORS MEAN ODDS RATIO SIGNIFICANCE LEVEL (in %) DECLINE OR INCREASE (in %) MATERNAL EDUCATION 0.950 1 -5 TYPE OF NEIGHBORHOODS. . (Squatter) (1.000) (--) (--) .High Standing N. 0.547 0.1 -45.3 . Average Standing 0.841 NS -15.9 PLACE OF DELIVERY: .(At Home) (1.000) (--) (--) •Health Inst 0.658 1 or 5 -34.2 OCCUPATIONAL STATUS OF MOTHER: . (Housewives) (1.000) (--) (---) . Students 1.584 NS +58.4 . Working Mothers 1.496 5 +49.6 . Others 1.518 NS +51.8 NUMBER OF BEDROOMS 0.954 10 4.6 Source: Table 4.2.4. Legend: **** p <= 0.001; *** p <= 0.01; ** p <= 0.05; * p <= 0.10; N.S.= Not Significant. 149 Table 4.2.7.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Child Mortality in Urban Areas of Zaire.4 VARIABLE NAME Model 17 INTERCEPT 0.201*** (0.594) EDUCATION 0.961** (0.017) PLACE OF DELIVERY: . (At home) . (At health institution) 1.000 0.686** (0.164) OCCUPATIONAL STATUS OF MOTHER: . (Housewife) . Student . Working . Others(k) 1.000 1.641 ( ° * 4 2 2 i * 1.526 (0.158) 1.548 (0.270) ACCESS TO MODERN FACILITY: . (Squatter neighborhood) . High standing neighborhood . Average standing neighborhood 1.000 4 c 4 t ' i c 0.564 (0.175) 0.850 (0.128) NUMBER OF BEDROOMS: 0.937* I (0.036) | -2 LOG L CHI-SQUARE DF 2804iU i 87.4 27 Source: Table 4.2.4. Legend: **** p <= 0.001; *** p <= 0.01; ** p <= 0.05; * p <= 0.10; N.S.=Not Significant. 4 Drawn from the last model of the Table 4.2.2. Here, only significant factors are presented. Contrary to my expectation, the effect of maternal education on child death rate is less pronounced in poor (or squatter) neighborhoods than in high status strata (high standing neighborhoods) or in average standing neighborhoods but in both cases the interaction education-type of neighborhoods is not significant. All the indicators of the household characteristics, which are supposed to reflect the quality of the house or the conditions of the immediate environment of the house, included in Table 4.2.4. are not significantly associated with child mortality. This particular finding is true for both the composite index of the household characteristics and individual variables of this index. What this particular result suggests to me is that factors, which are intended to capture the household characteristics, are redundant in presence of the "types of neighborhoods," variable which already includes quality of immediate environment where the household lives. In fact, high standing neighborhoods are in large majority inhabited by people who belong to high social classes. These neighborhoods possess all the public infrastructures (adduction in water, sewage service, electricity, tarred roads, etc.), while "Average standing" neighborhoods possess only some of these infrastructures. Squatter neighborhoods— located, in majority, in peripheral areas of the city— usually do not 151 possess those modern infrastructures. Another key factor of mortality before five years shown in Table 4.2.6. (and Table 4.2.7.) is the place of delivery. Contrary to my expectation, data from urban Zaire show that being born in health institutions (health center, maternity ward, etc.), in comparison to being born at home, is a significant mortality-inhibiting factor in childhood (1-4 years) rather than in infancy (less than 1 years). It seems to me that in urban Zaire the factor "place of delivery" captures the ability of the mother to access modern health in general rather than only the antenatal care. The third key factor of mortality before five is occupational status of mothers. According to the evidence, housewives surveyed have experienced lower child mortality than student mothers, working mothers, and others respectively, though the effect of mother occupational status is only significant for working mothers. Children of working mothers compared to those of housewives are 1.496 times more likely to die in the age range 1-4 years. The difference is significant at 5 percent. Finally, childhood mortality in urban Zaire is slightly determined by the number of bedrooms in the housing, a proxy for communicable, infectious and respiratory diseases. On average, the odds ratio equivalent to a one-unit increase in the number of bedroom is 0.954. This is significant at 10 152 percent only. Thus, the evidence from urban Zaire implies that the number of bedrooms in the house is not as important in childhood as it is in infancy. Chart 4.2.2. below shows the significant paths from the proximate factors defined in Table 4.2.1. to childhood mortality in urban Zaire. Chart 4.2.2.: From the Proximate Determinants to Child Death in Urban Areas Zaire: The Significant Paths. 1) MATERNAL EDUCATION OTHER SOCIO-ECONOMIC FACTORS: 2) Type of neighborhoods 3) Place of delivery 4) Occupational status of mothers: * Working mothers 5) Number of bedrooms in the house CHILD DEATH Though the evidence displayed in Table Appendix 4.2.4. shows that there is a lack of significant association between the marital status of mother and child mortality, it seems that single mothers have 1.203 times higher child mortality than women who are monogamously married. Likewise women in 153 polygamous marriage have 1.297 times higher childhood mortality than women in monogamous marriage single. While women who are either divorced, separated, or widowed (called category "others") have 1.268 times higher child mortality than women in monogamous marriage. The interaction between education and marital status of women, though not significant, is implying that the effect of schooling for single women or women in polygamous marriages is lower than for those who are married monogamously. 154 Table 4.2.1.: Variables Included in the Study VARIABLE AND ITS CATEGORIES SYMBOL USED TYPE OF VARIABLE Education EDUCAT3 Continuous Household Income HINCOME Continuous (Interaction Educ & Household Income) INTI Continuous Mother Age at Birth of Child (in years): . (Less than 20)* . 20-24 . 25-34 . 35 or more AGE20 24 AGE25 34 AGE35 Categorical (Dummy Variable) Birth Order: . (First birth)* . 2-3 . 4-6 . 7 or more RANG2 RANG4 RANG7 Categorical (Dummy Variable) Birth Weight (in grams): . (Less than 2,500)* . 2,500-3,499 . 3,500 or more . Unknown BWEIGHT2 BWEIGHT3 BWEIGHT9 Categorical (Dummy Variable) Length of Pregnancy: . (9 Months)* . Less than 9 months AGEP1 Categorical (Dummy Variable) Place of Delivery: . (At Home)* . Health Institution PLACE1 Categorical (Dummy Variable) Marital Status: . (Married Monogamously)* . Single . Married Polygamously . Others'1 MARITAL0 MARTTAL2 MARITAL3 Categorical (Dummy Variable) 155 Table 4.2.1. (Continued): VARIABLE NAME AND ITS CATEGORIES SYMBOL USED TYPE OF VARIABLE (Interaction between Education & Marital Status): . Educ*Single . Educ*Married Polyg. . Educ*Othersb INT2 INT3 INT4 Continuous Occupational Status: . (Housewife) . Student . Working . Others' OCCUPO OCCUP2 OCCUP4 Categorical (Dummy Variables) Access to Modern Facility: . (Squatter Neighborhood) . High Standing Neighborhood1 1 . Average Standing Neighborhood' ACCESS2 ACCESS 1 Categorical (Dummy Variables Index of the Quality of Housing: 1) Number of Rooms in the Housing Unit 2) Material of Construction of Wall, Roof, and Floor (composite index): . (Low: index 1-6)* . Average: index 7-9 . High: index 10-12 ROOMS LIVING 1 LIVING2 Continuous Categorical (Dummy Variables) 156 Table 4.2.1. (Continued) VARIABLE NAME AND ITS CATEGORIES SYMBOL USED TYPE OF VARIABLE Index of the Quality of Immediate Environment of the Housing Unit:' . (Low: index 1-6)* . Average: index 7-9 . High: index 10-12 HCOND2 HCOND3 Categorical (Dummy Variables) Water:* . (Not Connected to Piped Water)* . Piped water WATER Categorical (Dummy Variable) Trash:h . (Don’t Use a garbage-can/ . Use Garbage-Can TRASH Categorical (Dummy Variable) Access to Latrine: . (No)* . Yes SEWAGE Categorical (Dummy Varaible) O ’ ) For marital status, the category "others” includes consensual unions, divorced, separated, widowed. (') For occupational status, the category "others" includes never worked for physical inability, prostitutes, etc. (*) High standing neighborhood is made up of upper class neighborhoods (strata 0) and planned neighborhoods (strata 3). See Table Annex 1 for more information. O Average standing neighborhood is made up of old neighborhoods (strata 1) and new neighborhoods (strata 2 ). See Table Annex 1 for more information. 0 This is a composite index built from information on the location and use of latrine, source and distance of water supply, the way of handling household and human wastes in the household, etc. (*) Connected to piped water includes households which use water from "REGIDESO," the public company of distribution of "drinkable" water. O ’ ) Don't use garbage-can includes households which throw away household wastes in a hole, on ground (surface), or in a river, etc. 157 Chapter 5 FROM MATERNAL EDUCATION TO CHILD MORTALITY: THE PATHWAYS OF INFLUENCE The goal in this chapter is to examine the pathways through which maternal schooling influences the proximate determinants of child mortality defined in Table 4.2.1. and presented in Charts 4.2.1. and 4.2.2. Six sections are included in this chapter 5. The first section will relate maternal schooling and the gestational status of mother while pregnant, a determinant shown in chapter 4 to be strongly associated with infant mortality in urban areas in Zaire. The second section will examine whether maternal schooling has a significant effect on the use of health services for delivery. The third section examines the relationship between maternal education and child nutritional status. The fourth section investigates the net association between maternal schooling and the following factors: (1) health knowledge, (2) incidence of diarrhoea, (3) behavioral responses to diarrhoea, (4) incidence of fever, (5) behavioral responses to fever. The fifth section will investigate the association between education of mother and immunization status of her child. The sixth section will examine the relationship of maternal education and the occupational status of the mother, a determinant shown in Tables 4.2.6. and 4.2.7. strongly associated with child mortality in urban Zaire. 158 5.1. EDUCATION AND GESTATIONAL STATUS OF WOMEN DURING PREGNANCY IN URBAN ZAIRE The goal of this section is to examine the impact of maternal education on mother's gestational status after controlling for a number of other covariates. In accordance with the health transition theory, I expect that the higher the education of mother, the better her gestational status because educated mothers are able to flout traditional food taboos related to their nutrition while pregnant (Ware, 1984) . Two proxies of gestational status of mothers are used in this study. One is birth weight, another is length of pregnancy. Both proxies suggest that when a mother is better fed during pregnancy there is a higher chance for her to deliver a normal or an overweight baby and a higher chance to deliver a full term baby. For reasons of clarity, the two proxies are analyzed separately below. 5.1.1. Education of Mother and Birth Weight The variable birthweight derives from the question asked to all mothers about the birth weight of each child they have borne alive. To increase the reliability of the answer to this question, birth weights were grouped into four categories. Interviewed women had to choose the category which corresponds to her child's birth weight. These categories are "Less than 2,500 grams," "2,500-3499 grams," "3,500 and more," and the category "unknown birth weight." 159 A multinomial logistic regression is used to assess the relationship between maternal education and birth weight, controlling for socio-economic, demographic, and environmental factors. In this regression the dependent variable takes the value 1 for unknown birth weight, 2 for birth weight 3,500 grams or more (over weight) , 3 for birth weight in between 2,500 and 3,499 grams (normal weight), and the value 4 for birth weight less than 2,500 grams (low birth weight). Here, reference category is "low birth weight." Thus, I am examining the effects of explanatory factors on the log-odds of unknown birth weight versus low birth weight, the log-odds of over weight (birth weight i 3,500 grams) versus low birth weight, and the log-odds of normal weight (birth weight 2,500-3,499 grams) versus low birth weight. The logistic model for the log odds given K explanatory variables (or predictors) can be written as follows: in ((Pj_/ (1-Pi)) = ®i^ii ®2xi2 6kxik The global test of the model is given by the likelihood ratio chi-square. When the likelihood ratio is not significant, one concludes that the model fits the data. Otherwise, i.e. when the likelihood ratio is significant, the model does not fit the data. The decision-making for likelihood ratio in polytomous logistic regression is completely reversed for chi- square in a dichotomous logistic regression. I am going to use 160 the CATMOD procedure in SAS software. Table 5.1.1. indicates that maternal level of schooling significantly works to decrease the risk of unknown birth weight by 19.15 (p-value s 0.001) while it works to raise the odds of over weight baby and the odds of normal weight baby by 1.0167 and 1.0513, respectively. The odds ratio for normal weight is slightly significant at 10 percent, while the odds of overweight versus low birth weight is not significant. Birth weight is rather strongly and significantly enhanced by household income, controlling for other covariates. In fact, the mean value of the odds of being overweight is 1.2466 for one-unit increase in income. This is significant at 0.001 percent. Though the odds ratio of being normal weight is 1.069, it is not significant. The interaction of the education of the mother and household income is not statistically significant, though it suggests that educational effect is higher in households with low income. To summarize, Table 5.1.1. shows that the effect of maternal education on birth weight in not as great as the effect of household income. However, better educated mothers are more likely to know the birth weight of their children than the less educated ones. Beside education and household income, low birthweight in urban Zaire is influenced by four factors. One of them is the child gender. Table 5.1.1. indicates that female babies have higher birthweight than male babies. This is significant (at 5%) for overweight but not for normal weight, while unknown 161 birth weight is slightly significant (at 10%) for female compared with male babies. Why does the risk of low nutritional status rise when the mother is pregnant for male than for female babies? Is this because male foetuses cause more health problems to their mothers than female foetuses? This question cannot be answered in this study because of lack of relevant data on this issue. Another factor of birth weight displayed in Table 5.1.1. is mother's age at the birth of the child. This table shows, for example, that unknown birth weight is higher among teenagers than among women 20-24 years, and 25-34 years of age, but it is slightly higher among mothers of 35 years and over. But the difference is only statistically significant (at 5 percent) for women of 20-24 years compared with teenagers. Mothers of age 25-34 years are, on average, 0.8479 times less likely not to know the birth weight of their child than teenager mothers. The odds of normal birth weight (2,500-3,499 grams) is lower among women of 20-24 years and women of 25-34 years as compared with teenagers mothers, though the contrast is only statistically significant for women of 25-34 years. On average, the odds of normal birth weight (versus low birth weight) is 0.8955 for mothers aged 20-24 years, and it is 0.8408 for mothers aged 25-34 years. In other words, these odds indicate that normal birth weights (2,500-3,499 grams) are higher among teenagers than among mothers of age groups 20-24 and 25-34 years. This can be explained by the fact that 162 teenager mothers have reported significantly more cases of unknown birth weight than mothers of 25-34 years of age. In addition, the odds of overweight (3,500 grams and above) is significantly lower for women 20-24 years and women 25-34 years than for teenagers. The mean value of odds ratio for age group 20-24 is 0.8955. In other words, what these data are revealing is that "older mothers," especially those of 25-34 years of age, have higher risk than teenagers to bear low birth weight babies. Two possible explanations can be evoked here. Firstly, because unknown birth weight is concentrated on teenagers, this may deflate the cases of low birth weight babies that these very young mothers have borne between 1981- 1985. Secondly, these findings may reflect the fact that due to physiological depletion caused by repeated child bearing, women of 25-34 years of age are at higher risk of nutritional deficiency during pregnancy than very young mothers. Another determinant of birth weight is the marital status of the mother when pregnant. Table 5.1.1. shows that women who are in polygamous unions, compared with those in monogamous marriage, have significantly high risk of low birth weight. Likewise women who are divorced, separated, and widowed have higher risk of low birth weight than women monogamously married. Finally, the odds of unknown birth weight is statistically and significantly higher among single mothers than those who are in monogamous marriage. These findings may be reflecting the competition for household resources and 163 poverty. In polygamous unions, competition for resources may be stronger than in monogamous households, while for women who are divorced, widowed and separated, poverty may be the main explanation in a society where economic resources are controlled by men. Table 5.1.1. indicates that birth weight is also influenced by the occupational status of the mother during pregnancy. It clearly shows that working mothers have significantly lower risk of having low birth weight than are housewife mothers. Why do working women have lower odds of having low birth weight? Though this question cannot find a conclusive explanation in this dissertation because of the limitations of our data, I can mention two possible answers. One is that maybe working mothers are financially able to compensate for their own nutrition during pregnancy. Therefore, their risk to bear a baby with a low birth weight is reduced. Another answer is that these findings are may be reflecting the reporting errors since the odds of unknown birth weight is significantly higher among working women than among housewives. The next proxy for the gestational status while pregnant includes the length of the pregnancy. This is analyzed in the paragraph below. 5.1.2. Education of Mother and Premature Birth The variable premature birth is derived from the question asked to all mothers about the length (in months) of each 164 pregnancy they had. For pregnancies which resulted in live birth, I have created a dichotomous variable coded 1 for premature birth, and 0 for full-term birth. A dichotomous logistic regression is used to assess the relationship between maternal education and premature birth, controlling for socio economic, demographic, and environmental factors. The findings of this regression are displayed in Table 5.1.2. Table 5.1.2. clearly shows that the relationship between maternal education and premature birth is statistically significant (p-value < 0.001). Specifically, this table shows that higher educated mothers are, on average, 0.9385 times less likely to give birth to a premature baby than less educated mothers. This table also indicates that maternal education rather than household income is the most important factor of full-term birth. The interaction between education and household income is not significant at all, although it is positive. Two additional determinants play a significant role on prematurity (see Table 5.1.2.). First is birth order. This factor is strongly significant indicating that the risk of prematurity in urban areas of Zaire decreases when birth order increases. Second is marital status of the mother. The evidence shows that unwed women and women in polygamous marriage have higher risk of prematurity than those who married monogamously. But the difference is only slightly significant (at 10%) for single in comparison to women in 165 monogamous marriage. In the next section I will examine the role of maternal education on the use of maternity for delivery. 5.2. EDUCATION AND THE USE OF MATERNITY FOR DELIVERY In my data set, of about 7485 births which occurred between 1981 and 1985, 6369 (85%) of these birth took place in health institutions, only 1,116 births (15%) occurred outside maternity, at home, en route towards hospital, etc. To examine the net effect of maternal schooling on this factor, I will assume that educated mothers will use health institutions for delivery more than the uneducated ones because school integrates women to a new society in which health problems are solved using a modern medical approach. A logistic regression approach will be used to reach the goal set in this section. In this multivariate analysis, the dependent variable takes the value 1 when the delivery of the child took place in maternity or 0 when it took place outside maternity. The results of these operations are shown in Table 5.2.1. Table 5.2.1. indicates that the hypothesis formulated above is confirmed. Indeed, the use of maternity for delivery in urban Zaire is higher among educated women than uneducated women. On average, for example, one-year increase in maternal schooling raises the use of health services for delivery by 166 16.67 percent. In all the models presented in Table 5.2.1., the p-value is less than 0.1 per cent. This table also shows that household income does not significantly affect the use of maternity hospital for delivery. At first, the models indicate that the effect of income is negative; then its effect turns positive when one controls for the type of neighborhood where the parents live. About the latter factor, Table 5.2.1. indicates that the odds of using health institutions for delivery is enhanced for women who live in accessible neighborhoods of residence. For example women who reside in high standing neighborhoods are 2.9645 times more likely to use maternity hospitals for delivery than are women who live in squatting areas. Even women who reside in "average standing" neighborhoods are 2.882 times more likely to use this service for delivery than women in the squatting neighborhoods. The relationship is very significant at 1 percent in both cases. Four additional significant factors of use of maternity hospital are indicated in Table 5.2.1. The first factor is the age of mothers, especially the age above 34 years. Women of age 35 year or more are 3.969 times more likely to use a maternity than teenager mothers. The difference is significant at 5 percent. However, the significance of the difference disappears when the type of neighborhoods of residence is taken into account. The second of use of a maternity hospital is the birth order. Evidence indicates that the use of a 167 maternity is high for a first born child, it is significantly reduced for birth orders 2-3 and 4-6, then it rises for birth order 7 and more. It seems like Zairian mothers living in urban areas use maternity hospitals mostly for birth orders associated with a high risk of infant and/or maternal mortality, namely a first birth order and birth order 7 or more. The third significant factor of delivering the baby in health institutions includes marital status. Table 5.2.1. indicates that women in polygamous unions, compared with those married monogamously, have low use of maternity hospitals. The difference is significant at 5 per cent. Finally, delivering in a maternity hospital is influenced by the occupational status of women. It seems that housewives make higher use of a maternity than student women, working women and "others," respectively. But the significance level of differences of the marital categories is slim (10%). 5.3. EDUCATION OF MOTHER AND CHILD NUTRITIONAL STATUS Here, I assume that children of educated women will be better nourished than those of uneducated women. The reasons are that educated mothers have larger knowledge of nutritious foods for children than uneducated ones; they are able to control children nutritional status and to choose healthy food for their children. In addition, educated mothers will more 168 likely than the uneducated mothers avoid the practice of food discrimination which consists of feeding adult male in priority and children after that. However, because the access to food also depends on the ability to buy, I expect that household income will exert a significant impact on nutritional status of children. I also expect a positive interaction of mother education and household income since educated mothers have the tendency to live in higher income household. The dependent variable in this section is the mid-arm circumference (left arm). Mid-arm circumference is a technique often used to assess muscle wasting or poor muscle development (Jelliffe, 1966), an indication of protein-calorie malnutrition, especially for children in early childhood. In our data, arm circumference was measured at the mid upper left arm with fibre-glass tape provided by the UNICEF office in Zaire. The measurement was done only among children 12-59 months of age who, at the time of survey, were resident in the households identified. Children less than 12 months and those over 59 months were excluded in our data because it seems that arm circumference of those group of children is not always related to their nutritional status. About this exclusion, Jelliffe (1966, pp.75-77 and 176-179), states that at infancy the nutritional status of the child is largely influenced not only by breastfeeding and/or bottlefeeding, but also by maternal nutritional status during pregnancy, etc. 169 Above 59 months, the author states that arm circumference is largely influenced by the physical and other muscular exercises. Pre-school children (12-59 months of age) were included in the study because this is a period of high needs in food due to rapid growth of the body. The child in this age period is in transition in many aspects of his/her life. One change is that during this age period the child experiences a diet change. In urban Zaire, for example, previous study by Bakutuvwidi et al. (1985) has shown that children are completely weaned around 16 to 20 months of age. Another change which appears at this age concerns a change in immunity status. In effect, a child who reaches 12-59 months is no longer protected from maternal immunity acquired since he/she was a foetus. At the same time, his/her contacts with bacteria and other infectious agents increase. Another change which appears at this age concerns the change in the pattern of day care practice. Indeed, for many children this is the age period when they face for the first time a separation from mother. In our data, the mid-arm circumference was coded 1 when the circumference is less than 12.5 centimeters, it was coded 2 when the circumference is in between 12.5-13.5 centimeters, and it was coded 3 when the circumference is more than 13.5 centimeters. This way of reporting the data presents both advantages and inconveniences. It is advantageous because it 170 is easy to proceed in the field. In addition, it avoids errors due to misreading the score on the tape by the field workers. The main inconvenience of this way of reporting data is that I cannot compare the observed score of nutrition for each child with the standard of reference. Henceforth, these three intervals of mid-arm circumference will be called as follows (Frisancho, 1993): 1) score less than 12.5 centimeters will be called "severe protein-calorie malnutrition," 2) score in between 12.5-13.5 centimeters will be called "moderate protein-calories malnutrition," 3) score more than 13.5 centimeters will be considered as "good nutritional status." Table 5.3.1. gives the percentages of children 1-4 years (born between 1983 and 1986) by age and nutritional status. It shows that the prevalence of malnutrition in urban areas in Zaire is about 20 per cent, meaning that about one child aged 12-59 months over 5 suffers from some forms of malnutrition. This table also indicates that malnutrition tends to be greater among very young children 1-2 years of age compared with children 3-4 years. Further, this table indicates that moderate malnutrition is the category of malnutrition most prevalent in urban areas in Zaire. It includes about 15.1% of children 12-59 months of age. 171 Table 5 . 3 . 1 . : Distribution (in %) of Children 1 - 4 years b v Acre and Nutritional Status. Cohort 1 9 8 3 - 1 9 8 6 AGE < 12.5 Nutrition Status ARM CIRCUMFERENCE (in centimeters) 12.5-13.5 in • m H A TOTAL % 1 8.2 19.1 72.7 1109 100 2 5.2 14.9 79.9 1314 100 3 3.2 12.8 83.9 1061 100 4 2.2 13.3 84.5 924 100 TOTAL 213 667 3528 4408 % 4.5 15.1 80.4 100 To examine the impact of education on nutritional status of children born between 1983 and 1986 in urban Zaire, I have employed a multinomial logistic regression, using the CATMOD procedure in SAS Software. Although the dependent variable— the arm circumference of children 12-59 years of age— is ordinal, I have employed the CATMOD procedure for nominal dependent variable instead of the CATMOD procedure for ordinal dependent variable because I have no control whether or not there was a clear cut specification of classes in the field. The results of this regression are given in Table 5.3.2. What Table 5.3.2. reveals is that maternal education rather than household income is the factor significantly associated with malnutrition in urban areas of Zaire. The relationship education-child malnutrition is negative. For 172 example, children of more educated mothers are 0.9820 times less likely to suffer from "severe malnutrition" than children of less educated mothers. The difference is not significant, while children of mothers with high education are significantly (at 0.1%) 0.9231 times less likely to suffer from moderate malnutrition than children of less educated mothers. However, the effect of education is very significant (p-value < 0.001) only in the contrast "moderate malnutrition" against "good nutrition." The findings displayed in Table 5.3.2. indicate that the effect of household income is not significantly associated with nutritional status of children in urban Zaire, though the relationship is negative for the contrast moderate malnutrition against "good nutrition." Further, these findings imply that in urban Zaire, the educational effect is higher in households with higher income than those with low incomes, though the interaction maternal education and household income is shown to be not statistically significant. In addition, these findings confirm that birth weight, an indicator of the nutritional status of the mother during pregnancy, is not a significant factor of child nutrition after infancy, i.e., after one year of life. In effect, birth weight is shown in Table 5.3.2 to be not a significant factor of child nutrition after 12 months of life. However, length of pregnancy suggests a puzzling result that when the child was born prematurely, his/her nutritional status is "better" after infancy. The 173 association is very significant (p-value < 0.01), especially for moderate malnutrition. Further, Table 5.3.2. indicates that the child nutritional status after infancy is strongly influenced by the child's bio-demographic characteristics. The first demographic factor of the child shown in Table 5.3.2. is his/her age. In fact, children aged 18-35 and 36-59 months, respectively, have higher odds of malnutrition of both types (severe and moderate) than are children aged 12-17 months. For example, compared to children of 12-17 months, children of 18-35 months are, on average 1.4144 times more likely to suffer from severe malnutrition. In addition, children 18-35 months are, on average, 1.2071 times more likely to have moderate malnutrition. Toddler children (36-59 months of age) are, on average, 1.9601 times more likely than children 12-17 months to suffer from severe malnutrition, and they are 1.4319 times more likely to suffer from moderate malnutrition. In both cases of malnutrition— severe and mild— the differences of these categories to the reference one (12-17 months) are extremely significant at 0.1 percent. This finding can be understood by the fact that at 12-17 months some children are still breastfeeding, and as a result they have not yet completely lost the physiological immunity acquired at birth from their mothers. They have just entered the transitional period I have mentioned above. This finding implies that higher severe malnutrition in late childhood age (18-35 months and 36-59 months) may be the main reason why child mortality is high in urban areas of Zaire. Previous hospital and community studies have found that only severe (or extreme) malnutrition is associated with high risk of child mortality while the relationship between mild or moderate malnutrition and child mortality is inconclusive if not conflicting (Van Den Broeck, 1993). The second demographic factor of the child associated with malnutrition is gender. In all models presented in Table 5.3.2. it is indicated that female children have lower odds of both severe and moderate malnutrition. But the association is not significant. The third demographic characteristic of the child related significantly related to his/her malnutrition is the birth order. This is especially true for moderate malnutrition. For example, Table 5.3.2. indicates that for moderate malnutrition the odds ratios are, on average, 1.0857, 1.1781, 1.2299 for birth orders 2-3, 4-6, and 7 or more, respectively. The differences between order categories 4-6, 7 or more and the reference category are statistically significant at 5 percent in both cases. However, the relationship between birth order and child nutrition is not statistically significant in case of severe malnutrition, though the association is positive. Finally, Table 5.3.2. indicates that severe malnutrition is higher among children living in squatting neighborhoods than those who live in "average standing” neighborhoods, for 175 example. The corresponding odds ratio (0.6893) is significant at 5 percent. The reverse (finding) is shown for moderate malnutrition. Again this finding may explain why high standing and average standing neighborhoods considered individually have lower child mortality than squatting neighborhoods. The last model of Table 5.3.2. implies that educational effect in reducing severe malnutrition is higher in poor (squatting) neighborhoods than it is in average standing neighborhoods. This effect is significant at 10 per cent. But the educational effect is higher in very rich neighborhoods (high standing neighborhoods) than in poor ones, though this last effect is not statistically significant. In the section to follow, I will investigate the role of education on the medical knowledge, on diarrhoeal and febrile incidence, and on the behaviors associated with diarrhoea and malaria. 5.4. EDUCATION AND KNOWLEDGE OF SUGAR-SALT-SOLUTION, INCIDENCE OF DIARRHOEA AND FEVER, AND BEHAVIORAL RESPONSES TO DIARRHOEA OR FEVER This section investigates the impact of maternal education on knowledge of sugar-salt-solution, on incidence of diarrhoea and fever, and on the behavioral responses of mothers when their children had diarrhoea or fever. Diarrhoea and fever (especially malaria-related fever) are two important morbidity problems among children under five 176 in developing countries because of their predominant role in child mortality (Van Den Broeck, 1993; Hudelson, 1993; Gaminiratne, 1991; Tsui et al., 1988). In Zaire, numerous studies have reported that diarrhoea, malaria or fever are very prevalent among children. Tsui et al. (1988), for example, in a study in Bas-Zalre which included rural areas (53 villages of Songololo) and the city of Matadi, identified that the incidence of diarrhoea is around 20% in both urban and rural areas, with large variation by age of the child. For malaria, the incidence level was 25% in urban area and 21% in rural. Like the incidence of diarrhoea, large age variations were identified. The incidence of fever-not-associated with malaria among under five children surveyed was 9% in urban area and 11% in rural, with also some variation by age of child. In a study by Ngondo and Gamboa (1988), based on a representative sample of 3,165 households living in Kinshasa in 1986, the incidence of diarrhoea and fever, are reported of being 21.3% and 37.8% respectively. Here again, the distribution of diarrhoeal incidence varies with the age of the child: 75% of all cases of diarrhoea occurred among children under 3 years old. This study does not give the age distribution of febrile incidence, nor does it distinguish fever-related malaria from fever-unassociated with malaria. Another study by Bakutuvwidi et al. (1985), based on a representative sample of households included 4 cities and 2 rural areas. This study reports that in cities the incidence 177 of diarrhoea varies from 24.4 to 52.8 per cent, while in rural areas it varied between 38.9 and 60.2 per cent. The correspondent incidence rates for fever were in the range of 22.1-60.8 per cent in cities and 64.0 and 82.6 per cent in rural areas. Here again, data are based on mother's reports about her child health two weeks before the survey. In addition, reports on fever do not distinguish malaria-related from nonmalaria-related fever. Though malaria and diarrhea are leading causes of morbidity among children under fiver in Zaire, only malaria is reported as a predominant cause of death, while diarrhoea seems to play a modest role on child mortality (Van Den Broeck, 1993) . The latter findings about the mild role of diarrhoea as a cause of infant and child death in Zaire was also confirmed in Kasongo Project Team (1989), a study in eastern Zaire (Kivu province) where diarrhoea was found contributing to only 12% of deaths. This finding about diarrhoea in Zaire contrasts with the predominant role attributed to this cause of morbidity on infant and child mortality in Asian countries. Here, high incidence of diarrhoea is believed to enhance the association between nutritional status and child mortality. In Zaire, based on the study by Van Den Broeck et al. (1993) , the association between nutritional status and risk of mortality is only obvious for extreme malnutrition (marasmus or kwashiorkor), but it is not clear for mild and moderate mortality. 178 Three points— all related to my data set— are very important before getting into the subject matter. The first is that during the data collection there was no precise definition of diarrhoea and/or fever. The data used for this dissertation are based on the report of mother during the interview on whether or not her child had diarrhoea or fever the last 15 days from the date of the interview; if yes what she (the mother) did, and where she took her child. Here, the data reflect more mother's understanding of what is diarrhoea and/or fever. This type of data collection cannot rule out the possibility of both over-reporting (for example, some mothers may have reported infantile loose stools as diarrhoea) and, more importantly, this study cannot rule out under-reporting. As reported by Gaminiratne (1991), under-reporting related to morbidity statistics derived from cross-sectional data often occurs for multiple reasons: (1) lapses in recall, (2) child care during daytime is not always done by the mother, especially for working mothers, (3) the belief system prevent some families to report diarrhoea since suffering from this disease is seen as a cause for shame, (4) the reference period used is "too long". For example, Gaminiratne (1991) cites studies carried out in Bangladesh by Alam et al. (1989) and in Bolivia by DHS in-depth study of maternal and child health which have both evidenced that accuracy of reporting of diarrhoea declines with the increase in length of the reference period. According to this evidence, the best 179 reference period to use for diarrhoea is 48 hours. There are also authors such as van Ginneken (1991) who reports a length of recall period of 24 hours used in some DHS data. The second point to consider is that there is no way in my dissertation to have a clear cut distinction between fever and malaria. The way the question was asked during the interview (did your child suffer from fever the last two weeks before the survey?) does not help make this distinction. As mentioned before, there are some good reasons to make this distinction in Zaire. First because the incidence of fever unassociated with malaria among under five children is not as high as the incidence of malaria-related fever (Tsui et al., 1988) . Second, in a recent multi-round study in Bwamanda health zone (a rural area in northern Zaire), Van Den Broeck et al. (1993) have indicated that malaria is the number one killer among children five children, representing 26% of total deaths recorded in that zone. Finally, the data analyzed in this dissertation were collected in a single round cross-sectional survey. Thus, only information on health status of surviving children who were resident of the sampled households is available. 5.4.1. Incidence of Diarrhoea & Fever; Descriptive Statistics Table 5.4.1. drawn from our survey confirms the findings of previous studies mentioned above: the incidence of diarrhoea and fever among children under five-year-old is very 180 high in urban areas in Zaire, with incidence rates of 22.5% and 33.0%, respectively. In addition, the age patterns of incidence of diarrhoea and fever in our data set are similar to the age pattern found in previous studies in Zaire and elsewhere, particularly the study by Tsui et al. (1988) in Bas-Zalre, and the study by Gaminiratne (1991) in Sri-Lanka. Table 5.4.1.: Incidence (%^ of Diarrhoea and Fever bv Aae. Children 0-59 months. Urban Areas of Zaire. 13 Cities. FONAMES/UNICEF Survey. 1987. AGE (in months) DIARRHEA FEVER NUMBER of CHILDREN 0-3 9.6 15.4 468 4-6 27.8 38.7 313 7-9 37.4 48.4 289 10-11 44.7 40.4 208 12-17 37.5 42.3 624 18-23 32.5 38.2 545 24-35 20.1 33.8 1,078 36-47 15.5 29.8 956 48-59 7.2 24.0 666 ALL 22.5 33.0 5,147 For diarrhoea, for example, our data indicate that the incidence rate in urban areas in Zaire, is very low during the first three months of life before rising sharply at 4-6 months, the age of introduction of supplementary food. After this age interval, the rate rises again and reaches a peak at 10-11 years and then begins to fall. Meanwhile, the age pattern of incidence of fever is low at 0-3 months of age and rises sharply at 4-6. After 6 months of age, the incidence of 181 fever reaches the highest level at 7-9 months before starting to decline thereafter, though another rise is observed in between 12-17 months of age. Anyway, in all age intervals the incidence of fever is higher than that of diarrhoea, which is an indication that the level of fever identified in our data captures more the incidence of fever-related malaria than that of fever-unassociated with malaria. The age pattern of diarrhoea and fever found in our data corresponds to some biological and socio-behavioral forces which operate in child life from birth to age five. In the age interval 0-6 months, for example, the infant is still protected by maternal antibodies acquired since the he/she was in womb of the mother. Towards the end of the period (0-6 months), the introduction of supplementary food which, according to a study in rural and urban areas of Zaire by Bakutuvwidi et al. (1985), occurs around 3.4 to 5 months of age, operates to increase the risk of morbidity. In the age interval 7-12 months, the postpartum biological protection acquired from maternal antibodies is expected to be lost. This operates to raise in diarrhoeal (or febrile) incidence rate in this age interval. Due to the weaning process, which occurs in urban Zaire around 16.1 to 20.7 months of age (Bakutuvwidi et al. , 1985), high incidence rates are maintained after one year of age until the child will develop his/her own immunity to protect against diseases. Table 5.4.2. indicates that incidence rates of diarrhoea 182 (and in less extent of fever) among children under five years in urban Zaire varies by maternal education. For example, illiterate mothers experienced high risk of child morbidity while mothers with secondary school degree or more (7 years or more) experienced low risk particularly for diarrhoea. However, Table 5.4.2. is inconclusive about the child morbidity-maternal education relationship since it does not tell whether or not variations of morbidity by maternal education is still maintained after controlling for other variables. This justifies the use of multivariate logistic regression techniques to assess the net effect of maternal education on child morbidity. Table 5.4.2.: Incidence of Diarrhoea and Fever bv Mother's Education. Children Aged 0-59 Months Reported having Diarrhoea and Fever in the two weeks reference bv Education of Mother. MOTHER EDUCATION DIARRHOEA FEVER NUMBER OF CHILDREN Illiterate 23.3 36.5 800 1-3 years 23.1 34.1 481 4-6 years 24.3 32.8 1,507 7-9 years 21.4 30.9 1,526 10+ years 20.0 33.2 825 Total 22.5 32.9 5,147 183 5.4.2. Incidence of Diarrhoea & Behavioral Responses of Mothers in Urban Zaire; Multivariate Results Table 5.4.3. intends to address the question whether maternal education works to reduce significantly the incidence of diarrhoea among children living in urban areas of Zaire in 1987. Here, the dependent variable is whether or not the child had diarrhoea the last 15 days before the interview. This variable is coded 1 when the answer is "yes" and 0 when the answer is "no." I will use a logistic regression (with a stepwise forward procedure). The results indicate that maternal education is negatively associated with the risk of diarrhoea among under five children living in urban areas in Zaire. The association is significant (at 5%) only when one considers the bio demographic factors of the child, namely his/her age, his/her gender, his/her birth order and also when one considers the nutritional status of the mother during pregnancy. The mean value of the odds ratio of incidence of diarrhoea is 0.847 in this case. When one controls for the age of mother at birth of the child, her marital status, her occupational status, and the characteristics of the house and immediate environment, the effect of maternal schooling is no longer significant, though it remains negative. This implies that the effect of maternal education on child incidence of diarrhoea becomes a weak factor when other socio-economic and demographic factors of the mother and immediate environment of life are taken into 184 account. Besides maternal schooling, the risk of diarrhoea in urban areas of Zaire is determined by many other factors among which are some demographic factors of child and mother. For the child, the single most important factor is his/her age. For example, children of 4-6, 7-11, and 12-35 months of age have, on average, 2.7940 times, 5.3057 times, and 2.9116 times respectively, higher incidence of diarrhoea than toddler of 3 6-59 months of age. For all three age categories, the differences with the reference age (36-59 months) are statistically significant. However, for infants aged 0-3 months the odds of suffering from diarrhoea is 1.2821 times lower than for children aged 36-59 months. Here, the difference is not statistically significant. Another demographic factor of diarrhoea related to the child is birth order. What Table 5.4.3. indicates is that first born children are less likely to suffer from diarrhea than children of birth order above one. On average, children of birth order 2-3 are 1.0697 times more likely to experience diarrhoea than first born ones. The difference, however, is not statistically significant. For children of birth order 4-6 and 7 or more, the difference is significant only when one takes into account demographic characteristics of the mother. In this case, children of birth order 4-6 are significantly (at 10%) 1.3075 times more likely to experience diarrhoea than first born children. Again, when mother's demographic 185 characteristics are taken into the picture, children of birth order 7 or more are significantly (at 5%) 1.4364 times more likely to suffer from diarrhoea than first born children. The birth order of the child becomes a non-significant factor of diarrhoea in presence of factors pertaining to the type of neighborhood of residence, number of bedrooms in the house, and quality of the immediate environment of the housing unit. The maternal demographic factors of high risk of child diarrhoea include first her nutritional status while pregnant. Here, length of pregnancy and birth weight— two proxies of nutritional status— are very important factors. Table 5.4.3. shows that children who were born premature have significantly higher incidence of diarrhoea than children born full-term. On average, children born premature are 1.3722 times more likely to experience diarrhoea than children who were born after full-term pregnancy. The difference is significant at 5 percent. In addition, Table 5.4.3. indicates that in urban Zaire normal or overweight-born children have significantly lower risk of diarrhoea than children who were born with a low weight. This table displays that children born with birth in the range 2,500-3,499 grams (normal birth weight) are, on average, 0.7232 times less likely to have diarrhoea than those who were born with low birth weight. The difference is statistically significant at 5 percent. In addition, children born with birth weight 3,500 grams or more (over birth weight) are, on average, 0.7526 times less likely to suffer from 186 diarrhoea than those who were born with low birth weight. What these two factors (length of pregnancy and birth weight) are trying to convey is that when the nutritional status of mother during pregnancy is not qualitatively and/or quantitatively sufficient, the foetus (baby) is delivered prematurely with low birth weight. However, the baby who is born in that condition does not acquire solid maternal immunity to protect him/her against infectious diseases long after or even when he/she is born. Thus, these data imply that low birth weight and prematurity are significantly related to infant mortality because children born with low birth weight and those born prematurely have a high risk of suffering from diarrhoea. This seems to be true because I have shown early that the risk of diarrhoea in urban Zaire is highest among infant of 4-6 and 7-11 months of age. Another significant demographic factor of mother associated with high risk of diarrhoea in urban Zaire is mother's age. In effect, compared with children of teenager mothers, those of mothers aged 20-24 years are 0.7232 times less likely to experience diarrhoea. The difference is significant at 5 percent. Children of mothers aged 25-34 years are 0.6966 times less likely to suffer from diarrhoea than children from teenager mothers. The difference is very significant (at 1%). Children of mothers aged 35 or above as compared to children of teenager mothers have 0.7339 times lower risk of diarrhoea but the difference is not significant 187 when characteristics of the housing quality, type of neighborhood of residence, and conditions of immediate environment are taken into account. It appears to me that maternal age is capturing here the behavioral and experience of mother related to child rearing, though it may also capture some physiological factor. Table 5.4.3. also indicates that the occurrence of diarrhoea among children under five years in urban Zaire is associated with some socio-economic factors. One of these factors is household income. As expected, this factor is shown very significantly related to child risk of diarrhoea in Zaire (p-value < 0.01). Unexpectedly, the direction of relationship is positive, indicating that one-unit increase in household income raises the risk of diarrhoea by 11.79 percent (odds ratio=l.1251). Unfortunately, there is no way in these data to verify this finding since it is not possible to control for patterns of daytime child care when mothers are absent from home. In addition, it is difficult to interpret this finding since children of working mothers display lower risk of diarrhoea than children of housewives, though the difference is not significant in all models where occupational status of mother is included. Based on my own observation as a dweller in urban area in Zaire, I can speculate by stating that it is not uncommon in urban areas of Zaire, especially among well- to-do families, to see the mother give responsibility of child care to babysitter during daytime. I do think that the 188 variable occupational status included here does not capture the complex patterns of daytime child care in urban Zaire. Another determinant of the incidence of diarrhoea shown in Table 5.4.3. is the place of delivery used by the mother to bear the index child. As said before, this factor has two facets. First, it may be a proxy of use of health care by the mother while pregnant. Second, place of delivery can capture the socio-economic background of the mother, especially in case where care during delivery is not free. In that case the use of maternity hospital is partly determined by the ability to pay. Table 5.4.3. indicates that children whose mothers have delivered them in hospital had, on average, 0.7599 times lower incidence of diarrhoea than children whose mothers delivered at home. The difference is significant at 5 per cent. But this significance level is reduced (at 10%) when the index of quality of housing unit is included in the regression. Another socio-economic factor negatively associated with the incidence of diarrhoea in urban Zaire is the number of bedrooms that the house contains. According to the data displayed in Table 5.4.3., a one-unit increase in the number of bedroom decreases the risk of diarrhoea by 8.15% (odds ratio is 0.9218). This factor is very significant (p-value s 0.1) . In addition, this variable remains very significant even when the index of the quality of immediate environment is included. It is worth noting that the latter index is a 189 composite factor which includes the source of water and its distance, use and location of toilet, and use of garbage-can for household waste, etc. Another important question raised in this section is whether or not maternal education influences the mother's behaviors when her child is suffering from diarrhoea? From the health transition theory described in chapter 1, I would expect that when a child is suffering from diarrhoea, educated mothers will display higher use of sugar-salt-solution, higher use of "modern health facility," higher use of modern medicine to treat diarrhoea, and will be more prompt in seeking modern health care than the uneducated women. The main reason is because school integrates the person who had been socialized by it into a new society where disease is believed to be a physiological disorder (rather than a curse afflicted by spirits or gods) for which the solution is to be found in modern health care. Tables 5.4.4., 5.4.5., 5.4.6., and 5.4.7., which consider separately the different types of behaviors a mother may display when her child has diarrhoea attempt to test this hypothesis. The first behavioral response to investigate in this study concerns the use by mothers of sugar-salt-solution (SSS) for children who had diarrhoea fifteen days before the survey. As a remind, sugar-salt-solution is one of the simple and inexpensive strategies recommended by UNICEF (1986) to all countries members of the United Nations in order to prevent 190 children with diarrhoea to be dehydrated. In Zaire, the responsibilities of making this procedure to be known to women is left to health centers, maternity wards, and centers for immunization. Anytime that a woman enters into contact with one of these health institutions, she is informed about the role of SSS in curving diarrhoea dehydration, about how to make this solution properly, and about how to administer it. To examine whether educated mothers are more likely to use the SSS when their children suffer from diarrhoea, I construct a dependent variable called "use of SSS." This variable takes the score 1 when the interviewed mother used this solution. It takes the score 0 when the interviewed mother did not use this solution. Let us recall that, about 5,147 children aged less than 5 years were identified in this survey, 1157 of whom had diarrhoea (22.5%). The unit of analysis in this section is the dyad child-mother. A logistic regression technique is used to examine the association between education and use of SSS. Table 5.4.4. shows that education of the mother is one of the key factors which increases her use of SSS when her child has diarrhoea. It is shown that a one-year increase in maternal schooling corresponds to an average rise of 4.36 percent in the use of sugar-salt-solution (average value of odds ratio is 1.0446). Educational effect is significant at 5 percent. But this level of significance declines at 10 percent when factors pertaining to type of neighborhoods of residence, 191 quality of housing unit of the household, and quality of immediate environment of the house, are taken into account. Household income does not play a key role in the association between maternal education and use of SSS, nor does the interaction between education and household income play a key role. Three additional factors played a key role in the use of SSS by mothers whose children had diarrhoea in urban areas in Zaire. The first factor is the age of child. Table 5.4.4. indicates that the use of SSS by mothers is very low when the child is 0-3 months of age compared with toddler (36-59 months of age), but the difference is not significant. However, the use of SSS by mothers rises significantly for children of 4-6 months, 7-11 months or 12-35 months of age. In effect, SSS is used 2.3985 times more for infants 4-6 months than for children 36-59 months. In addition, SSS is used 1.7946 times more for infants aged 7-11 months of age than for toddler (36- 59 months). Finally, this solution is used 1.71997 times more for children 12-35 months than for toddlers. In these three cases, the differences are very significant (at 1%). The second factor which affects the behavior of the mothers concerning the use of SSS is her marital status at the time of the outbreak of diarrhoea. Our data reveal that single mothers in urban areas in Zaire are 0.1455 less likely than women in monogamous marriage to use SSS. The difference is strongly significant at 0.01 percent. Other categories of 192 marital status are not significantly related to mother's use of SSS. The third factor which slightly influences the use of SSS in urban areas of Zaire is the occupational status of the mother. For example, working women are 1.488 times more likely than housewives to utilize SSS when the child has diarrhoea. However, the difference between the two occupational categories is significant at 10 per cent only. The next mother's behavioral response to be investigated in this section is the use of modern health when the child has an episode of diarrhoea. In order to determine the net effect of maternal schooling on the use of health facility in case of diarrhoea, I have constructed a dependent variable called "use of modern facility" from the question "where did you take (name) the child to treat the diarrhoea?" This is a precoded question whose possible answers were: (1) nowhere, (2) the child was taken to a traditional healer, (3) the child was taken to a health center/dispensary, (4) I (the mother) consulted the local pharmacy (drug store), (5) the child was taken to private doctor, and (6) the child was taken to the hospital. Answers number 3, 4, 5, and 6 are grouped under the category "modern health facility." This category is coded 1. The answers number 1 and 2 form the reference category which is coded 0. A logistic regression will be used here in order to investigate this association. Table 5.4.5. reveals that in urban Zaire, maternal 193 education is not a key factor which works to enhance the use of modern health facility in case of diarrhoea. In addition, the direction of the relationship between maternal education and use of modern health facility becomes negative when demographic factors of the mother (age, marital and occupational statuses) are controlled for. Further, Table 5.4.5. indicates that household income, which is one of the factors of access to modern health services, is not a significant factor in this study. Its impact is negative before the inclusion of mother's demographic factors. After that inclusion, household income becomes positively related to the use of health facility in case of diarrhoea, but it remains not significant. Further, Table 5.4.5. also reveals that the use of health services when the child has a diarrhoeal episode is determined by previous behaviors of the mother in regard to modern health, by her age, her global access to modernity as it is captured by the type of neighborhood where she lives. It is clear that mothers who have used SSS, or mothers who have previously delivered in maternity ward are among the highest users of health facility in case of diarrhoea. The odds ratio in both cases are 3.044 and 1.5592, respectively. Another previous behavior which significantly determines the use of health facility for child diarrhea in urban Zaire is whether or not mothers had changed their "water-feeding practice." Table 5.4.5. indicates that mothers who reduced or 194 stopped giving water to the child when the latter had diarrhoea have used significantly (p-value s 0.01) health services more than mothers who increased or maintained the same quantity of water to the child. This can be interpreted by the fact that reducing or stopping the use of water when the child has a diarrhoeal episode may lead to quick dehydration of the child. As a consequence, this behavior may result in increase in use of health facility. This suggests that mothers should be informed about not to stop or decrease the use of drinking water when a child has diarrhoea. About the age of the mother, Table 5.4.5. indicates that the use of health facility in case of child diarrhea is higher for mothers of 20-24 years, 25-34, and 35 or more than for teenager mothers but the difference is significant only for mothers 20-24 years (p-value < 0.05). These latter mothers have used modern health services for child diarrhoea 1.6872 times more than teenager mothers. The study also shows that mothers who live in high standing neighborhoods or in average standing neighborhoods have used health facility more when their children had diarrhoea than those who live in squatting neighborhoods. But the difference is very significant only between average standing neighborhoods and squatting neighborhoods (p-value s 0.001). This seems logical because most health facilities in urban Zaire are located in rich neighborhoods or in average standing ones. Rare are the health facilities located in the 195 squatting neighborhoods, except for the health facilities run by confessional institutions (Luyinduladio, 1994). Finally, the use of health facilities in case of child diarrhea is significantly (at 10%) 0.4593 times less likely for very young children (0-3 months of age) than for toddler (36-59 months of age). Another maternal behavior to investigate in this section includes the use of modern drugs by the mother for her child who has diarrhoea. The basic question here is whether educated mothers in urban Zaire use modern drugs more than the uneducated ones. A mixed message is displayed in Table 5.4.6. On one hand, maternal education is shown significantly associated with the increase in use of modern drug when one controls only for characteristics related to the child (age, gender, birth order) and to the household income. In this case, the average value of the odds ratio is 1.0652. On the other hand, when socio-demographic characteristics related to the mother (age of mother, place of delivery, marital status, occupational status) and socio-environmental factors of the housing unit are taken into account, the significant effect of education starts from decreasing then disappears completely, though the association remains positive. At the same time, when the effect of maternal education disappears, household income becomes a significant factor which increases the use of drug by an average of 16.27 percent (odds ratio is 1.1767) in case of child diarrhoea (p-value i 0.05). Four additional determinants of the use of modern medicine for child diarrhoea are shown in Table 5.4.6. One of these factors is the age of child. It is revealed that the use of drug to treat child diarrhoea is significantly lower for very young kids (0-3 months) than for toddler (36-59 months). The use of drugs for child diarrhoea is higher for children aged 4-6 months, 7-11 months, and 12-35 months than for toddler children (36-59 months), though the difference is only significant for the kids 7-11 months compared with children 36-59 months of age. This finding is consistent with the age pattern of incidence of diarrhoea shown in Table 5.4.1. Another factor associated with high the use of modern medicine to combat child diarrhoea is the place of delivery. Evidence shows that a mother who delivered the index child in the maternity hospital is, on average, 1.9441 times more likely to use modern drugs when the index child has diarrhoea than a mother who delivered the index child at home. The accessibility of the neighborhood of residence is another significant factor of high use of modern medicine for child diarrhoea. Evidence displayed in Table 5.4.6. indicates that mothers who live in rich (high standing) neighborhoods are 1.686 times more likely to use modern drugs to combat diarrhoea than mothers who live in squatting areas of the cities. The difference is significant at 5 percent. Finally, Table 5.4.6. shows that the birth order of the child tends to deflate the use of modern medicine when the 197 child suffers from diarrhea. However, the significance level of this factor is low (only 10%) and it appears only for birth order 7 and above. The fourth behavioral change to examine in this section concerns the quickness of the mother in seeking modern health services soon after the symptoms of diarrhoea appear. Here, the goal is to find evidence to the question: are educated women more prompt than the uneducated ones in seeking modern health services when their children suffer from diarrhoea? In this study, a health-seeking behavior is called prompt when the child was brought to the health service within 24 hours. In this case the dependent variable is coded 1. A health- seeking behavior is called not prompt when the health service was sought within more than 24 hours. The dependent variable is this case is coded 0. Table 5.4.7. gives the results of the logistic regression of promptness in seeking health services when the child has diarrhoea. It is clear that mother's education is not a key factor of promptness in seeking health services. Rather this behavioral response is significantly determined by household income. Evidence shows that, on average, a one-unit increase in household income increases the promptness in seeking health services by 16.25 percent (odds ratio is 1.1765). This is significant at 5 percent. Note that a one-year increase in education of mother rises the promptness by 1.07 percent only (odds ratio=l.0108) . But the educational effect is not 198 significant at all. Another socio-economic factor associated with promptness in seeking health services is the type of neighborhood of residence. Parents who live in squatting neighborhoods bring their child to health services late compared with those who live in highly accessible neighborhoods (high standing neighborhoods and average standing neighborhoods). About the type of neighborhood of residence, Table 5.4.7. shows children whose parents live in high standing neighborhoods are, on average, 2.336 times more likely to be brought to health services within 24 hours of the start of diarrhoea than children from squatting neighborhoods. This is significant at one percent. Even children from average standing neighborhoods are brought to the health services 2.0204 times earlier than those from squatting areas. This is also significant at 1 percent. Three other determinants of promptness are shown in Table 5.4.7. The first is the birth order. It appears that medical service is sought more quickly among high birth order than for first born children. The difference is significant only for birth orders 2-3 and 7 or above compared to first birth order. In both cases the p-value is in the range 5-10 percent. In addition, the evidence shows that children of birth order 2-3 are 1.7163 times more likely to be brought earlier to health services than are first born children. The difference is significant in the range 5-10 percent. 199 The second determinant of quickness is the marital status of mother. Here again, evidence indicates that unmarried women are more prompt in seeking medical service in case of diarrhoea than monogamously married women. The difference is significant (at 5%) only between single mothers and women in monogamous marriage. Children of single mothers are 3.7307 times more likely to be brought to health services earlier than children of mothers in monogamous marriage. Finally, mother's promptness in seeking care in case of diarrhoea is determined by her age. Here, only mothers of age group 20-24 years respond relatively quickly to child's diarrhoea. There are some indications that women of age 35 years or more seek health services late compared with teenagers. But the difference between these two groups is not significant. The next part of this section 5.4. deals with the issue of whether or not maternal education is a key factor in child malaria in urban areas in Zaire. After responding to this question, I will then examine the role of mother's education on her health-seeking behaviors. 5.4.3. Incidence of Fever and Behavioral Responses to Fever: A Multivariate Analysis Table 5.4.8. below presents the results of the multivariate analysis of the incidence of fever among children under five years in urban areas in Zaire. Incidence of fever 200 is defined by a dichotomous variable which takes the score 1 when a child suffered from fever the last 15 days from the day of the interview. It takes the score 0 when the child did not suffer from fever during this period. A logistic regression is applied here because the dependent variable is dichotomous. As expected Table 5.4.8. indicates that the higher the education of mothers the lower is the incidence of fever among children 0-59 months in urban Zaire. On average, one-year increase in mother's schooling reduces the incidence of malaria by 2.04 percent (odds ratio is 0.9798). The effect of mother's schooling on fever is statistically significant (p- value < 0.01). However, the significance level of mother's schooling disappears when the index of quality of housing (especially material of construction of the house) is included in the model. What Table 5.4.8. also shows is that household income is not a significant factor of incidence of fever, though the association of both variables is negative. There is even a change in the direction of the association between income and incidence of fever when the types of neighborhood of residence is included in the regression. The interaction between education and household income is negative, but not significant. This indicates that the effect of mother's schooling is stronger at low income. The same observation is made when one considers the interaction between education and type of neighborhood of residence. Clearly, the effect of 201 mother's education on incidence of fever is stronger in squatting neighborhoods than in rich neighborhoods. Five additional key determinants of incidence of fever are displayed in Table 5.4.8. The first factor is the age of the child. For example, model 20 indicates that fever is 0.509 times less likely to hit children 0-3 months than toddler children (3 6-59 months of age). But the incidence of fever in urban Zaire is 1.817 times, 2.389 times, and 1.687 times more likely to catch children of aged 4-6 months, 7-11 months, and 12-35 months, respectively, than to catch toddler (36-59 months of age). The second important factor of malaria shown in this Table 5.4.8. is the type of neighborhood of residence. Children who live in high standing neighborhoods are 0.677 times less likely to experience malaria than children living in the squatting neighborhoods. But children living in average standing neighborhoods are 1.421 times more likely to suffer from fever than children living in the squatting areas. This finding suggests to me that fever in this study is more likely to reflect malaria than another type of fever. Two reasons help me think that way. First, in urban Zaire the sewage systems built to drain used water are found only in upper class or planned neighborhoods and in average standing areas, but not in squatting areas. (In average standing neighborhoods, most of sewage disposal systems were built during the colonial period). The second reason is that due to 202 the policy of structural adjustment imposed to Zaire by the World Bank and IMF (Lututala, Kintambu, and Mvudi, 1993), urban areas have witnessed a break down in public sewage systems and in other social services. (This is a macro- economic policy taken by the government, a policy which consists of cutting down public expenditures in health, education, and agriculture). In very rich neighborhoods, inhabited in majority by top government officials and high executive officers of private companies, I can assume that the damage in social services (including the sewage system) was negligible compared with the same damage in other neighborhoods inhabited in majority by workers who are not in managerial positions. The third significant factor of malaria shown in Table 5.4.8. is the quality of house (evaluated from the material of construction of the house). It is clearly displayed that children who live in high quality houses are 0.783 times less likely to suffer from fever (malaria) than children who live in low quality houses. The fourth important factor of malaria in urban Zaire are the conditions of the immediate environment of the house. For example, Table 5.4.8. displays that when the immediate environment of the house is highly clean (index 10-12), children are 0.787 times less likely to suffer from fever (malaria). The fifth key determinant of malaria includes demographic 203 characteristics of mothers, especially her age, marital status, and occupation. These three factors are significant (at 5%) . For example, children of mothers 20-24 years are 0.808 times less likely to suffer from malaria than children of teenager mothers. To sum up, what Table 5.4.8. is revealing is that maternal education has a significant effect on the occurrence of malaria to children under 5 years, but its the importance of education disappears when characteristics to the condition of the house, to the type of neighborhoods of residence, and to the conditions of immediate environment of the house are taken into account. The next part of this section deals with the behavioral responses triggered by fever. Specifically, I want to know how education affects the health-seeking behaviors of a mother when her child suffers from fever/malaria. Three types of behaviors are considered in this section: (1) use of modern health services (Table 5.4.9.), (2) use of modern medicine (Table 5.4.10.), and (3) promptness in seeking modern health care (Table 5.4.11.). For those three behaviors, I have two expectations First, I expect that mothers with a higher level of schooling will use modern health services or modern drugs to a greater extent than mothers of low level of education. Second, I expect that mothers with higher education will be more prompt in seeking modern health care than mothers with less education. 204 Table 5.4.9. gives the results of the logistic regression of use of modern health services (outside the home) on maternal education, controlling for other covariates. Here, the dependent variable takes the value 1 if the mother used modern health service when the child suffered from fever; it takes the value 0 if the mother did not use modern health services or used traditional services. This table illustrates that mother's level of schooling is a significant factor of the use of health services in case of fever. A one-year increases in women's education is associated with a 3.99 percent rise in the use of health services (odds ratio is 1.0407). At first, the level of significance of education is very high (with p-value < 0.01). But when the birth order of child is included in the model, this significance level is reduced (it becomes only 5 percent). The same situation happens when one controls for the place of birth of delivery. Table 5.4.9. also reveals that the effect of education on the use of modern health services for fever is reduced when household income increases, though the interaction between income and education is not statistically significant. Household income appears in Table 5.4.9. to reduce significantly (at 10%) the use of modern health services. For example, a one-unit increase in log-income reduces the use of health services by 8.37 percent (odds ratio equals 0.9198). This is surprising at first, but it makes sense when one remembers that in urban Zaire, only families with high income 205 can afford to have a family pharmacy at home. Therefore, the negative relationship displayed in Table 5.4.9. between income and use of modern health care when the child suffers from fever may be reflecting the fact that families with high income a possess small pharmacy at home and may not use health services outside the home unless the fever is life- threatening. Table 5.4.9. also indicates that the use of health services significantly (at 1%) rises by 30.93% (odds ratio equals 1.3625) when the child who suffers from fever also suffers from diarrhoea. Five other factors are shown in Table 5.4.9. to be importantly related to the use of health services when the child has fever. One factor is the birth order of child. The evidence illustrated in Table 5.4.9. indicates that the use of health services when the child has fever decreases with the birth order of the child. The use of health services is about 0.6608 times lower for children of birth order 2-3 than for first born children. The difference between these two categories is significant at 5 percent. Again, the use of health services is about 0.5609 times lower for children of birth order 7 or more than for first born. Here again, the difference is significant at 5 percent. Another factor associated with the use of health services by the mother is the child's age. The effect of age is low for younger infants 0-3 months of age, but it rises after the 206 first three months of life. At 7-11 months, for example, the age effect on the use of health service is 1.6068 times higher than for toddler (36-59 months of age). This difference is significant at 5 percent. Further, the use of health services by mothers for children who have fever is determined by the mother's age. Clearly Table 5.4.9. indicates that the use of health services is 1.5887 times more likely for mothers 20-24 years, 1.9045 times more likely for mothers 35 years or more than for teenager mothers. In both cases the difference is statistically significant at 5 percent. Another factor which enhances the use of modern services is the mother's occupational status. For example, working mothers are, on average, 1.5066 times more likely to use modern health services than housewives, while mothers who are still students are 2.3735 times more likely to use modern health services than housewives. The difference for working mothers is significant at 5 percent, but it is only 10 percent for students. The most important socio-economic factors associated with a higher use of modern health services for the child who suffers from malaria are captured by mother's previous contact with health institutions (place of delivery) and by type of neighborhoods of residence. Mothers who had previously been in contacted with one type of health service (e.g., maternity hospital) are, on average, 1.3 625 times more likely to use 207 modern health services to treat child fever than mothers who had not been exposed before to modern service. This is very significant at 1 percent. The same observation was made for diarrhoea (see Table 5.4.5.), although for this disease the impact of the place of delivery was not significant. Children whose mothers reside in "average standing neighborhoods" are 1.4951 times more likely to use the health services outside the home than those whose mothers live in the squatting areas. However, use of modern health services for child fever is not significantly different for mothers living in very rich areas (high standing neighborhoods) compared with mothers who live in the squatting areas. The latter observation is consistent with the negative association found between household income and use of modern services (Table 5.4.9.). This confirms that families with high income which, in majority, inhabit in rich neighborhoods, can afford to own a small pharmacy at home. Therefore, they may not seek health services outside the home unless the sickness is life-threatening. The next behavioral response to child fever to be examined in this section is about the use of modern medicine (drugs). Here again, use of modern medicine is a dichotomous variable which takes the score 1 if the child was given modern drugs to treat fever and the score 0 if the child was not treated or was given traditional medicine. Again, let me remind that the aim in this part of the section is to investigate the role of maternal education on the use of 208 modern drugs. Table 5.4.10. gives the results of the logistic regression of use of modern medicine on mother's level of schooling. This table displays a mixed message about the effect of maternal education on the use of modern medicine for a sick child. On the one hand this Table 5.4.10. indicates that the mother's level of schooling is an important factor of the use of modern drugs by mothers. This is true when one controls only for demographic variables of the child only (age, gender, and birth order). In this case, a one-year increase in mother's schooling increases the use of modern drugs by 6.89 percent (odds ratio=l.0713) . Here, the mother's education is a significant factor at 5 percent. On the other hand Table 5.4.10. indicates that the impact of the mother's education on the use of drugs for the child is not significant but its effect is positive. This is true when one includes in the model mother's demographic factors (namely her marital status, her occupation status, etc.) , type of neighborhoods of residence, and housing and environmental quality. Household income is not at all an important factor of the use of modern drugs. The use of modern medicine to treat fever (malaria) is also determined by the dynamic of infection the child is suffering from. For example, Table 5.4.10. shows that children who have suffered from both fever and diarrhoea, at the same time, are about 0.5939 times less likely to be given drugs 209 aimed at treating fever. This suggests that mothers establish some priority in the treatment of both diseases. In addition, this table indicates that the use of modern drugs to treat fever is determined by the age of the child. Here, evidence shown in Table 5.4.10. reveals that the odds of using modern drugs to treat fever is 0.3287 times lower for very young children 0-3 months of age than for toddler 36-59 months of age. The difference is significant (at 1%). Further, the use of modern drugs to treat fever is 1.7828 times and 2.2845 times greater for children of birth order 2-3 and 4-6, respectively, than for first born children. Besides birth order, the use of modern drugs to treat fever is significantly greater in rich neighborhoods than in poor ones. For example, Table 5.4.10. indicates that children who live in rich neighborhoods are 2.402 times more likely to be given modern medicines than children living in squatting neighborhoods, while children who live in "average standing areas" are 2.2816 times more likely to be treated with modern drugs than children of mothers who live in squatting areas. In addition, this evidence indicates that the use of modern drugs to treat child fever is affected by the age of mother, especially the age group 20-24 years. Children of mothers aged 20-24 years have higher odds (2.1067) of being given modern drugs to treat fever than children of teenager mothers. The difference is significant at 5 percent. Furthermore, the use of modern drugs by the mother for 210 the benefit of the child is also determined by her previous contact with health service. The evidence shows that mothers who had delivered the index child in the hospital are, on average, 1.7657 times more likely to use modern drug to treat fever than mothers who delivered the index child at home. However, the difference is significant at 10 percent only. Finally, Table 5.4.10. indicates that the use of modern drugs is determined by marital status of the mother. This table shows mothers who are in polygamous marriage are 0.4067 times less likely to use modern medicine for the sick child compared with women who are in monogamous marriage. The difference between the two marital categories is significant at 5 percent. The last behavioral response to be discussed in this section concerns mother's promptness in seeking modern treatment when the child has fever. Here again, the central question is whether the mother's level of schooling affect promptness in seeking modern treatment. Promptness is defined as a dichotomous variable which takes the score 1 when the child is given the modern treatment within 24 hours. It takes the score 0 when the mother gave the treatment to the child after one day of child sickness. A logistic regression is employed to examine the relationship between mother's education and promptness in seeking treatment. Table 5.4.11. gives the results of this multivariate logistic regression. Table 5.4.11. clearly exhibits that maternal education, 211 household income, and the interaction of both variables do not significantly affect the promptness in seeking modern treatment when the child has fever; it is rather the type of neighborhood of residence. The effect of household income is almost zero in presence of the type of neighborhood of residence, meaning that this latter factor has captured better the standard of living than does income alone. What is shown in Table 5.4.11. is that children who live in high standing neighborhoods are given modern treatment 1.6432 times quicker than children who live in squatting areas. The difference is significant at one percent. Children who live in neighborhoods of average standing are given treatment, on average, 1.6396 times quicker than those who live in squatting areas. It seems that the type of neighborhood of residence is reflecting two factors of life in urban areas of Zaire. First, it might reflect the opportunity to have access to both modern infrastructure and high income. Second, it might reflect the possibility to have access to modern information and to a new life style which can modify the belief system. Four other factors affect significantly the quickness of mothers in seeking medical care for their sick child. One factor, which is significant at 5%, consists of the birth order of the child. Children of birth orders 4-6 and 7 or more are given medical care quicker than first born children. The mean values of their odds ratio are 1.6411 and 1.9343, respectively. Another key factor, significant at 5%, consists 212 of the age of the mother at birth of the child. Surprisingly, mothers aged 20-24 years are 0.6342 times less likely to give modern treatment to their children within 24 hours than are teenager mothers. In the same manner, mothers aged 25-34 years are about 0.6069 times less likely to treat the child within 24 hours than are teenager mothers. In both cases, the difference is significant at 5 percent. At the present stage of knowledge, I do not have any logical explanation for this finding. Another factor which determines the quickness of the mother in seeking modern treatment for the sick child is the gender of the child. Evidence shown in Table 5.4.11. tells that girls are, on average, 0.7684 times less likely then boys to be given modern treatment within 24 hours of the start of fever. The difference between the two gender groups is significant in the range 5-10 percent. Finally, this evidence shows that unwed mothers, especially widowed, separated, and divorced, are more prompt than married women in seeking modern treatment for their sick child. On average, women who are widowed, separated, and divorced, are 2.9777 times more prompt than women in monogamous marriage. The difference is significant at 1 percent. In section 5.5. below, I will examine the role of education on the completeness of immunization of children 12- 23 months of age. 213 5.5. MATERNAL EDUCATION AND IMMUNIZATION OF CHILDREN 12-23 MONTHS OF AGE The most cited explanations given to justify the low risk of mortality among children of educated mothers compared with the uneducated is that schooling probably "enhances knowledge about effective ways to prevent, recognize, and treat childhood illnesses" (Cleland, 1989, p. 411). In this dissertation, I will investigate this explanation by hypothesizing that if a higher level of maternal education results in improved child survival, then health services that effectively prevent fatal childhood diseases must be used to a greater extend by mothers with higher education than those with little or no education (Streatfield, Singarimbun, and Diamond, 1990, p.447) . To be specific, this section suggests that the higher the education of the mother, the more likely she will have children who are fully immunized. The data to test this hypothesis are given by the information gathered by FONAMES/UNICEF survey conducted in 1987. A special questionnaire was included in this survey to assess the coverage of immunization among children 12-23 months of age identified, i.e., children who belong to birth cohorts 1985 and 1986. For each child aged 12-23 months, information on immunization status was collected for four types of vaccines: (1) Bacilli Calmett Guerin (BCG), (2) Measles, (3) Poliomyelitis (three doses: Poliol, Polio2, Polio3), and (4) Diphtheria, Pertussis, and Tetanus (three doses: DPT1, DPT2, 214 DPT3). Before discussing my main hypothesis of this section, I will show, first, some of the descriptive statistics concerning the immunization status of children 12-23 months in urban ares of Zaire. 5.5.1. Immunization of Children 12-23 months: Cross-tabulation Table 5.5.1. presents the distribution of children 12-23 months by types of vaccine and by knowledge of immunization dates by the mother in urban Zaire in July-August 1987. What is shown is that of the 1881 children aged 12-23 months identified in the survey, about 8 percent had never been immunized of any vaccine while 92 percent of them had been immunized of at least one vaccine. When immunization rates are classified by type of vaccine, this table indicates that the coverage rate is the highest for all vaccines given at birth. This corresponds roughly with the prevalence of birth in the maternity among these children (i.e. 89%). For example, a high prevalence of immunization is recorded for BCG, 93%, which is given at birth compared with 90% for measles which is given at 9 months in Zaire.1 For POLIO and DPT, Table 5.5.1. shows that immunization rates decline with the number of dose. As a reminder, the three doses of Poliomyelitis and Diphtheria- 1 The latest date to receive the measles injection vaccine is 9 months of age. In an earlier paper, I came across to some studies which suggest that measles vaccine be given as early as 6 months just after the child loses the immunity acquired at birth because many younger children in Zaire are catching measles. The only problem is that measles vaccine is not effective before 9 months (Luyinduladio, 1990). 215 Pertussis-Tetanus are given at 2, 3, and 4 months of age for the 1st, 2nd, and 3rd doses, respectively. Table 5.5.1.: Distribution of Children 12-23 months bv Types of Vaccine. Immunization Status and bv knowledge of Immunization Dates J Types 1 I Vaccine Children not Immunized Children Immunized Number of Children N. % Dates Known Dates Unknown <**> BCG 129 6.85 878 874 93.15 1881 MEASLES 188 10.00 718 975 90.00 1881 P0LI01 149 7.92 867 865 92.08 1881 P0LI02 158 8.40 819 904 91.60 1881 P0LI03 160 8.51 788 933 91.49 1881 DPT1 148 7.87 860 873 92.13 1881 DPT2 154 8.19 821 906 91.81 1881 DPT3 164 8.72 769 948 91.28 1881 ALL**) — 8.31 --- --- 91.69 1881 (*) Average percentage. (**) % of children immunized (date known + date unknown). On the basis of data presented in Table 5.5.1., one can conclude that the immunization coverage in urban Zaire is very high. This finding is very difficult to verify now by lack of reliable studies for comparison. However, the study on contraceptive prevalence, based on data collected in 1982-84 in urban areas (it includes 4 cities only) and in 2 well- equipped rural areas (Vanga and Nkara), also reports high rates of coverage of immunization among children under five years identified in the sample (Bakutuvwidi et al., 1985). For example, this study reports that the percentage of non 216 immunized children in urban areas surveyed in 1982-84 was about 6 percent, except in Kananga and Kisangani cities where this percentage reached 19 and 23%, respectively. However, the latter study is not itself reliable. In addition, this study covers all children under 5 years, an interval too large to be considered in questionnaire about immunization coverage. In any case, I suspect that the immunization coverage displayed in Table 5.5.1. are probably overestimating the reality. Two reasons allow me to think that way. First, information about the immunization status of the children 12- 23 months of age was not always taken from the immunization card because only 54.4% of children in the sample for whom their mothers exhibited the card to our field workers. As a reminder, our survey identified a total of 1881 children 12-23 months of age in the 13 Cities. However, the lack of immunization card cannot be equated to never been immunized because in Zaire, especially in urban areas, many health centers keep the immunization card to allow health officials to be able to control the child's immunization status. Unfortunately our data did not explore this fact. (The National survey on immunization conducted in Zaire in 1991 included a question on the status of immunization card. This survey was financed by UNICEF, USAID, WORLD HEALTH ORGANIZATION, and ROTARY INTERNATIONAL (Hurley et al., 1993). However, these data will not be analyzed in this dissertation since I received them late in November 1994 from the Center 217 for Disease Control). The second reason why I suspect that our survey overestimates the rates of immunization coverage in urban Zaire is the fact that dates of immunization are not known in many cases. Although not remembering the dates of immunization may be affected by lapse of memory, it may also imply that the child was not, in fact, immunized. Maybe the mother wanted just to cover up the shame associated with not having got the time to go to the health center to get her child immunized since the injection is free. Table 5.5.2. gives the distribution of rates of coverage of immunization among children 12-23 months by maternal education, classes of incomes, mother's age at birth of child, birth order, gender, length of pregnancy, birth weight, mother marital status at birth, mother occupational status at birth of child, strata, and age of child. This table also gives the percentage of children born in the hospital and the percentage of children for whom interviewed mothers have presented an immunization card. Two conclusions can be drawn from Table 5.5.2. First, there are large variations in immunization coverage rates by mother's or child socio-economic, demographic, and environmental factors for all the 5 dependent variables considered here (completeness of immunization, prevalence of immunization in DPT, POLIO, measles, and BCG). Considering, for example, the education of the mother, its is shown that 218 only mothers with 10 years of education or more have the highest level of immunization coverage. Below 10 years of education, the differences in prevalence are not large. This finding confirms the study by Streatfield et al. (1990) which also illustrates that there is a threshold effect since the rates of immunization are considerably greater for mother with at least secondary education (10 years or more). As is the case for education, there is also a threshold effect for household income: only high income household (^ 17,269 zaires currency in July-August 1987) have the highest level of coverage rates. These findings also indicate that teenager mothers have higher rates of immunization prevalence than any other age category. Here, I suspect that the data may be distorted by the types of reporting errors I mentioned above. Another reason why I suspect reporting error is the fact that teenager mothers had the highest rate of delivery at home compared with any other age group. However, the place of delivery may not be a good indicator of guality of data on immunization coverage since it is not uncommon in urban Zaire that the baby (and his/her mother) who is delivered at home be brought to maternity hospital for two reasons: (1) to receive the anti tetanus vaccines, and (2) to receive a "birth certificate" ("Certificat de naissance" in French) which allows the parents to seek a "birth attestation" ("attestation de naissance") later. A "birth certificate" is only delivered by maternity 219 while a "birth attestation" is delivered by the Officer of vital statistics service. Table 5.5.2. also indicates that first born children are more immunized than children of high birth order. Further, children 18-23 months of age have the highest rates of immunization compared with children 12-17 months. In addition, this table reveals that mothers who live in squatting neighborhoods (strata 4 + strata 6) have the lowest coverage rate for child immunization. Children born with low birth weight (less than 2,500 grams) are less likely to be immunized compared with other birth weight categories. Working mothers display higher prevalence of immunization than other occupational categories. However, the prevalence of immunization for all the vaccine does not vary by gender nor by length of pregnancy. Having shown the variations in immunization rates by maternal or child socio-economic, and environmental factors, the next part of the section will examine the hypothesis formulated above. Specifically, I will explore the relationship between education and the likelihood for the child 12-23 months of age to be immunized, controlling for other covariates. 5.5.2. Education and Child Immunization; A Multivariate Logistic Regression Here, the dependent variable is the completeness of immunization. This variable takes the score 1 when the child 220 is fully immunized, i.e., when he/she received one injection of measles, one injection of BCG, three injections of poliomyelitis, three DPT injections. It takes the score 0 when the child had never been immunized or was incompletely immunized. Table 5.5.3. presents the results of the logistic regression of completeness of immunization on maternal education, controlling for other covariates. This table clearly indicates that maternal education is not a significant factor that enhances the odds of being fully immunized in urban areas in Zaire, though the association is positive (odds ratio=l.0310). There are studies that had previously found similar results. For example, in a study using data from Indonesia, Streatfield et al. (1990) observed that the influence of formal education on the odds of a child's being immunized vanishes when mothers have correct knowledge of the vaccine functions. The authors showed that knowledge of the function of immunization is the most significant factors of the child's being immunized. Table 5.5.3. also indicates that household income is not an important factor that affects the odds of the child's being immunized. This is not surprising because immunization injections for all the vaccines were free of charge in Zaire at the time of survey. Four variables are shown in Table 5.5.3. to be significantly associated with the odds of being fully 221 immunized. Firstly is the age of child. This factor has a linear effect. In effect, children aged 18-23 months are 1.442 times more likely than very young children (12-17 months of age) to be fully immunized. But the effect of this covariate is not significant. Secondly is marital status of mother, especially mothers who are divorced, separated, widowers, or in consensual unions (category "others"). Children born from these mothers are 9 times more likely than children of women in monogamous marriage to be fully immunized (p-value £ 0.05). In this category of marital status, the effect of schooling is lower than the same effect for women in monogamous marriage. The interaction between education and marital status is highly significant (p-value i 0.01). The third covariate relatively important in the relationship education-completeness of immunization is made up of mothers who never work or who are physically handicapped (this is the occupational category called "others"). Children of these women are three times less likely than children of housewives women to be fully immunized (p-value < 0.05). Finally, the single most important factor of completeness of immunization in urban Zaire is the type of neighborhood of residence of the family. Surprisingly, the odds of the child's being fully immunized is 0.3433 times lower for children living in high standing neighborhoods than for children living in squatting neighborhoods. The difference is very significant (at 1%). This finding confirms the conclusion reached in 222 section 5.4. of this dissertation about the use of sugar-salt- solution to prevent diarrhoea. It was indicated in this section 5.4. that the use of preventive remedy is lower in rich neighborhoods than in poor neighborhoods. This finding also confirms the observation made by Van Lerberghe and Pangu (1988) in their brilliant study which critically examines different health policies implemented in Africa from the colonization period up today. Both authors observed that today rich people in Africa are enjoying technology or hospital- based health system while poor people are being given the services defined in the primary health strategy. In a recent analysis of data on health zones in Zaire, I reached roughly the same conclusion after I examined rural and urban disparities in health care (Luyinduladio, 1994) . I observed that the primary health care policy is more effective in rural areas than in urban areas. Further, I observed that two different health policies are working side by side in Zaire. One is the primary health care strategy which offers 7 types of services (including immunization and health education). This policy is more functional in rural health zones than in urban ones. The second policy, more prevalent in urban health zones, offers a technology-based strategy. In this policy, the accent in on the hospitals, which have more health infrastructures and personnel, but offer less defined in the primary health care activities. The disparity between neighborhoods in terms of 223 immunization displayed in Table 5.5.3. is just confirming the fact that the policy of Primary Health Care is an option for the poor. Rich people, who not only live in "clean" areas and eat "well," but also benefit from more "advanced" type of health care, which is centered on hospital. However, the observation that children in squatting neighborhoods in urban Zaire are more fully immunized than children from rich neighborhoods is puzzling when one considers the fact that child mortality in urban areas in Zaire (see Chapter 4) is significantly higher in squatting neighborhoods than in rich neighborhoods. Based on evidence provided in this dissertation and on studies reviewed in chapter 1, I can suggest two explanations for this puzzle. The first explanation is that child mortality is high in poor neighborhoods (compared to rich ones) because severe or extreme malnutrition is higher in squatting neighborhoods than in rich neighborhoods (Table 5.3.2.), while mild or moderate malnutrition is significantly higher in rich neighborhoods than in poor ones (see Table 5.3.2.). A recent study in rural Zaire based on a representative sample of children under five has indicated that risk of child mortality associated with kwashiorkor is 135 per thousand, in case of extreme marasmus this risk is 23 per thousand, but child mortality is only 1 per thousand in case of moderate or mild malnutrition (Van Den Broeck et al., 1993). The second explanation for the puzzle is that because of the effect of synergy which exists between 224 measles, diarrhoea, and malnutrition, I think that children who live in poor neighborhoods lose the protective function of measles injections due to severe malnutrition. In the next section I will investigate the association between maternal education and women's occupational status. 5.6. MATERNAL EDUCATION AND OCCUPATION STATUS OF MOTHERS The study by Farah and Preston (1982) based on evidence from the Sudan has indicated that working mothers had higher child mortality that housewives. The evidence presented in my study (chapter 4) is consistent with this finding. It shows that childhood mortality is determined by the occupational status of the mother. However, the evidence displayed in chapter 5 of my dissertation does not fully support Farah and Preston's (1982) explanation that children of working mothers are more deprived of adequate child care. It does only indicate that children of working mothers have a higher risk of severe malnutrition than those of housewife mothers. But does education influence really the occupational status of a women? In this section, the goal is to investigate the effect of education on women's occupation, assuming that educated mothers are more likely to work outside the home even when they have preschool children. To do so I will use a 225 multinomial logistic regression of occupational status of mothers with preschool children in 1987. In this regression, the dependent variable takes the score 1 for mothers who are housewives, the score 2 for mothers who are in the occupational group named "other" which includes inactive, students, or never worked. The dependent variable takes the score 3 when the mother is working. In this regression, I will exactly investigate the effect of explanatory factors on the log-odds of being a housewife mother versus being a working mother, the log-odds of being in the category "other" versus being a working mother. Of 4,731 women with pre-school children identified in the survey, 71.7% were housewives, 12.4% were in the category "other," and 15.9% were working outside the home. Table 5.6.1. indicates that educated mothers are, on average, less likely to be housewives rather than working mothers. On average, the odds ratio of being a housewife rather than a working mother is 0.8948. This odds ratio is significant at 0.1 percent. In addition, Table 5.6.1. shows that educated mothers are more likely to be working outside the home rather than being in category "other." Here again, the difference is strongly significant (at 0.1%). Two additional factors determine the occupational status of women with preschool children in urban areas in Zaire. They include (1) the age of the mother at the birth of the child, (2) and the type of neighborhood of residence. Table 5.6.1. 226 shows no indication that the number of children a mother has borne is an important factor. About the age of mother, Table 5.6.1. shows that mothers of age 20-24, 25-34, and 35 years or more are 1.2119 times, 1.5599 times, and 2.1332 times, respectively, more likely to be housewives rather than working mothers as compared with teenagers. The significance level of these differences are 5% for age group 20-24 years, and 0.1% for both age groups 25-34 and 35 years or more. Table 5.6.1. also indicates that women with preschool children living in rich neighborhoods of urban Zaire, compared with those who live in sguatting areas, are 0.9366 less likely to be housewives rather than to be working mothers. But the difference is not significant. The comparison of average standing neighborhoods with sguatting areas yields a different result. Here, women with preschool children are 1.4434 times more likely to be housewives rather than working outside the home. The difference between the two types of neighborhoods is significant at 1 percent. In other words, the latter result means that mothers who live in poor neighborhoods are more likely to work outside the homes than are mothers in average standing neighborhoods. The evidence shown in Table 5.6.1. partially supports Farah and Preston (1982) 's explanation that women's work is an indicator of economic stress in the household. But this 227 evidence indicates that this is true only for women of poor neighborhoods. This evidence makes me believe that the main reason why woman's work outside the home is shown in Table 4.2.3. having a high risk of child mortality in urban areas in Zaire is because working outside the home is very much associated with working in the informal sector. To say it correctly working outside the home for poor women is very much associated with economic stress in the household. To summarize, this chapter has found that (1) education moderately increases the odds for mother to deliver a baby with normal weight; (2) education reduces the odds of giving birth to a premature baby; (3) education increases the odds of mother to deliver in maternity; (4) children of educated mothers are less likely to be moderately malnourished; (5) educated mothers use the sugar-salt-solution more than the uneducated mothers; (6) for malaria, educated mothers use health services more than uneducated mothers; (7) educated mothers are more likely to work outside the home than uneducated ones. Chart 5.1.1. below shows the pathways of the effect of maternal education on infant and child mortality. It gives the net effect of the relationship between maternal education and the factors believed to play a substantial role in childhood survival. 228 Chart 5.1.1. E D U C A T I O N -> -> — > (-) — > (+) — > (+) -> -> -> ->(+/ Pathways of the Effect of Maternal Education on Infant and Child Mortality. PROXIMATE VARIABLES -) +) -) +) -) :+) -> (+) -) -) WORKING STATUS: . Housewife/Working . Other/Working BIRTH WEIGHT: Unknown/Low bweight Over /Low bweight Normal/Low bweight LENGTH of PREGNANCY: . Premature birth PLACE of DELIVERY: | , Health institutions _______________________I CHILD NUTRIT. STATUS| . Severe Mai. vs Good . Moderate vs. Good I DIARRHOEA: Incidence Behavioral response to diarrhoea: (1) Use of S.S.S. (2) Use of Health Services (3) Use of Modern Drugs (4) Promptness P.E. NET EFFECT Odds S.L.' ' - 0.1112 -0.0868 -0.0182 -0.0799 -0.1660 -0.0105 0.8948 0.9169 **** **** -0.1916 0.8257 **** 0.0166 1.0167 NS 0.0500 1.0513 * -0.0635 0.9385 **** +0.1667 1.1814 **** 0.9820 NS 0.9231 **** 0.8470 ** 0.9895 NS +0.0436 1.0446 **/* +0.04085 -0.01341 1.0417 NS 0.9867 NS +0.0631 1.0652 ** +0.0399 1.0407 NS +0.01075 1.0108 NS These are mean values. (b) P.E.= Parameter Estimates; Odds= Odds Ratio; S.L.= Significance Level: **** p ^ 0.001; *** p i 0.01; ** p £ 0.5; * p < . 0.10; NS not significant 229 Chart 5.1.1. (Continued) PROXIMATE VARIABLES E D U C A T I O N -> (-) -> ( + ) '> ( + ) ->(+/-) (3) Promptness (+) COMPLETENESS OF IMMUNIZATION P.E. NET EFFECT Odds S.L.' ' FEVER: Incidence Behavioral response (1) Use of Health Services (2) Use of Modern Drugs 0.0204 0.9798 *** 0.0171 0.9830 NS +0.0399 1.0407 ** +0.0689 1.0713 ** +0.0323 1.0328 NS -0.0051 0.9949 NS +0.0189 1.0191 NS +0.0305 1.0310 NS These are mean values. (b) p.E.= Parameter Estimates; Odds= Odds Ratio; S.L.= Significance Level **** p £ 0.001; *** p £ 0.01; ** p £ 0.5; * p £ 0.10; NS not significant. 230 TABLE 5.5.2.: Percentage of Children 12-23 Months who Have Never Received anv Vaccine: Percentage who received all the vaccine. Immunization Prevalence against DPT. Polio. x Measles, and BCG: Percentage of children 12-23 months whose Birth Took Place in Hospital Setting; and Percentage of children with a Immunization Card bv Socioeconomic & Biological Characteristics of Mother and Child, and bv Types of Neighborhood. MOTHER'S EDUCATIONAL LEVEL Mother's Education NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT1 POLIO1 MEASLES ILLITERATE 6.3 89.9 90.8 91.8 90.8 1-3 5.8 86.9 91.2 92.0 86.9 4-6 4.9 90.2 92.4 91.9 91.4 7-9 5.9 88.9 91.5 91.3 90.5 I 10 + 2.7 93.2 96.8 96.8 94.1 MOTHER'S EDUCATIONAL LEVEL (continued) I Mother's Education BCG BORN IN HOSP. CARD NUMBER OF CHILDREN — ILLITERATE 93.7 75.8 34.3 207 1-3 93.4 81.8 50.4 137 . 4-6 94.3 81.3 53.1 407 7-9 93.8 89.2 60.4 389 + o H 96.4 96.4 67.6 222 LEGEND: Hosp.= Hospital; Card= Immunization Card ( ■ * ■ ) For DPT and POLIO, the figures represent the percentage of children who received the three doses. 231 Table 5.5.2. (continued) CLASSES OF HOUSEHOLD INCOME Household Income (in zaires) NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES < 4,730 8.5 86.0 88.5 87.5 87.7 4,730-8,936 6.1 88.8 91.4 91.1 89.8 8,937-17,268 3.9 89.5 93.1 93.9 90.3 1 > 17,268 3.6 93.3 95.1 95.1 94.6 CLASSES OF HOUSEHOLD INCOME (continued) Household Income (in zaires) BCG BORN IN HOSP. CARD NUMBER OF CHILDREN < 4,730 90.4 86.0 43.8 816 4,730-8,936 93.3 83.7 49.8 313 8,937-17,268 95.9 87.0 58.0 362 > 17,269 96.2 84.4 54.1 390 LEGEND: Hosp.= Hospital; Card= Immunization Card 232 Table 5.5.2. (continued) AGE OF MOTHER AT TIME OF SURVEY AGE OF MOTHER (in years) NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES < 20 4.4 91.2 93.9 93.9 92.1 20-24 6.1 90.5 92.1 92.3 91.5 25-29 5.6 89.4 92.5 91.9 90.3 30-34 4.0 91.6 93.6 93.2 93.2 35-39 5.3 87.9 92.1 92.6 88.4 40-49 2.8 87.5 90.3 91.7 90.3 AGE OF MOTHER (continuec i) AGE OF MOTHER (in years) BCG BORN IN HOSP. CARD NUMBER OF CHILDREN < 20 94.7 81.6 57.9 114 20-24 93.1 85.4 58.5 378 25-29 94.2 85.5 53.3 360 30-34 95.6 84.8 56.0 250 35-39 94.2 88.9 49.5 190 40-49 97.2 80.6 38.9 72 I LEGEND: Hosp.= Hospital; Card= Immunization Card 233 Table 5.5.2. (continued) BIRTH ORDER OF CHILD 1 Birth Order NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES 1 3.2 92.1 94.9 94.9 93.1 I 2-3 6.4 90.0 91.9 91.9 91.0 ! 4-6 4.9 89.7 92.5 92.0 91.0 I 7 + 5.1 89.0 91.8 92.4 90.1 BIRTH ORDER OF CHILD (continued) Birth Order BCG BORN IN HOSP. CARD NUMBER OF CHILDREN | 1 95.8 88.9 58.3 216 2-3 93 .4 85.3 55.7 409 4-6 94.3 81.3 57.1 387 7 + 94.6 87.8 47.3 353 I jEGEND: Hosd.= HosDital; Card= Immunization Card CHILD GENDER Gender NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES 1 MALE 5.9 90.0 92.6 92.3 90.9 | FEMALE 4.4 90.0 92.4 92.7 91.3 234 Table 5.5.2. (continued) __________________ CHILD GENDER (continued) Gender BCG BORN IN HOSP. CARD NUMBER OF CHILDREN MALE 93.4 86.0 54.8 662 FEMALE 95. 3 84.7 53.9 700 I AGE OF CHILD | Age of child (in months) NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES 1 12-17 5.1 87.8 91.5 91.1 89.4 18-23 5.0 90.8 92.9 92.9 92.1 Unknown 9.8 86.3 87.8 87.8 87.4 AGE OF CHILD (continued) Age of child (in years) BCG BORN IN HOSP. CARD NUMBER OF CHILDREN 12-17 93.8 68.9 57.8 650 18-23 94.6 62.2 54.0 780 Unknown 89.6 51.2 30.4 451 LEGEND: Hosp.= Hospital; Card= Immunization Card 235 Table 5.5.2. (continued) MOTHER'S OCCUPATIONAL STATUS 1 Mother's I Occupational Status NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES STUDENT 12.5 84.4 87.5 87.5 84.4 HOUSEWIFE 5.2 90.1 92.6 92.6 91.0 WORKING 2.6 92.9 95.5 95.5 92.9 OTHERS 6.7 84.0 88.0 88.0 90.7 MOTHER'S OCCUPATIONAL STATUS (continued] Mother's Occupational Status BCG BORN IN HOSP. CARD NUMBER OF CHILDREN STUDENT 87.5 90.6 59.4 32 HOUSEWIFE 94.6 85.0 54.9 1,103 WORKING 97.4 84.5 53.5 155 | OTHERS 88.0 88.0 46.7 75 PLACE OF BIRTH NEVER 1 Place BEEN ALL of IMMUNI VACCIN DPT POLIOM MEASLES Birth -ZED E YELITE 1 AT HOME 5.1 92.9 93.9 93.9 92.9 MATERNITY 5.0 89.7 92.4 92.4 90.9 236 Table 5.5.2. (continued) PLACE OF BIRTH (continued) Place of Birth BCG BORN IN HOSP. CARD NUMBER OF CHILDREN AT HOME 94.9 — — 38.4 198 I MATERNITY 94.5 — 57.2 1,164 STRA*]PA OF RES][DENCE Strata of Residence NEVER BEEN IMMUNI -ZED ALL VACCIN E DPT POLIOM YELITE MEASLES 0 6.3 82.0 88.3 88.3 85.2 1 4.5 92.1 94.6 94.3 93.0 2 8.9 90.0 90.0 88.7 95.2 3 8.7 83.7 88.2 86.5 86.0 4 5.0 92.7 93.9 93.7 93.7 1 6 ... 14.3 77.1 78.1 80.0 78.1 STRATA OF RESIDENCE (continued) . Strata of Residence BCG BORN IN HOSP. CARD NUMBER OF CHILDREN _ 0 91.4 71.1 51.6 128 1 94.6 65.2 50.5 442 2 95.2 73.2 52.8 231 3 90.2 60.1 53.7 356 4 94.7 57.4 46.4 619 6 85.7 44.8 42.9 105 LEGEND: Hosp.= Hospxtal; Card= Immunization Card; For STRATA, see Table Annex 1 (Chapter 3) about the meaning of codes. 237 Chapter 6 CONCLUSIONS AND RECOMMENDATIONS This study has followed two objectives. The first objective was the analysis of the effect of maternal education on childhood mortality. Along with this objective, I assessed the relative contribution of economic, demographic, biological, and environmental factors in explaining the education-child survival relationship. The second objective was the investigation of the pathways through which maternal education exerts its impact on child survival. Along with this second objective, I examined a certain number of issues discussed in the health transition theory in relation to child survival in the Third World. For example, I investigated whether educated mothers compared to uneducated are most likely to be better nourished during pregnancy. If yes, is this because of education or is this a reflection of economic status? In addition, I examined whether education increases mothers' awareness about effective ways to prevent, recognize, and treat childhood ailments? Further, I investigated whether maternal schooling has an impact on child nutritional status? If the latter issue is true what is the role of economic and other factors on education-child nutrition relationship. Our study has been justified for practical and theoretical reasons. The practical reason to pursue this research is that the study of the education-mortality 238 association is still relevant because this topic offers "an excellent entry point for the study of general household determinants of health” (Cleland, 1989, p.415). Therefore, a deep understanding of mother's education and its role in child welfare will bear fresh understandings of the determinants of health care utilization, household decision-making and intra household resource allocation, etc., (Cleland, 1989). Three theoretical reasons pushed me to pursue this research. The first reason is that many authors such as Bicego and Boerma (1993) have observed that in the majority of the studies done previously on this topic data on household wealth or income are not always available. Therefore, this missing factor in the majority of previous studies make it difficult to respond adequately to the question to what extent is the observed education-child mortality relationship merely a function of education's link to economic status? (Bicego and Boerma, 1993). In my study I assessed the role of household wealth in the maternal education-child mortality relationship by including (1) the household income measured directly through the information on household consumption expenditures and (2) the index of living conditions, measured by the material of construction of wall, ground, and roof. The second theoretical reason for pursuing this topic is the need to re-assess whether the education-child mortality association is in fact a causal one and, if so, to examine new evidence to see if the theory of education-conditioned use of 239 modern health services and change in pattern of family formation is conclusive. This is a relevant reason because the evidence relating education to the use of health services, on the one hand, and to change in pattern of family formation, on the other, is still inconclusive, making it very difficult to determine, with minimum doubt, the question what behaviors serve to mediate the education advantage on childhood mortality. The third theoretical reason for pursuing this research is that no study on this topic has been done using the data from Zaire. Therefore, my study, which intended to examine the household determinants of child health, was designed in a way to enlighten policy-makers in setting a new child health policy, in particular, and a responsible reproduction health policy, in general. 6.1. THE FINDINGS OF THE STUDY The evidence analyzed in Chapter 4 has indicated that infant and child mortality are still high in urban Zaire. Direct estimations, show that infant mortality an<* under five mortality (5^0) are 77•4 Per thousand live births and 93.2 per thousand live births, respectively. Indirect estimations yield figures which are closer to direct estimations, especially for infant mortality. Infant mortality in urban areas of Zaire varies in the range 65-78 per thousand 240 live births, while under five mortality varies in the range 100.8-103.9 per thousand live births. Our data clearly suggest that in urban areas in Zaire maternal education is not a significant factor associated with infant mortality, though the relationship is negative indicating that a one-year increase in mother's schooling corresponds to a decline in infant mortality of 1.33 to 3.96 percent. In addition, evidence displayed in Table 4.2.2. shows that household income does not have a significant impact on infant mortality in urban areas in Zaire. What the evidence has shown is that infant mortality in urban Zaire is determined by demographic and/or biological factors which include: 1) One factor related to the infant him/herself, namely the gender of the infant. According to Table 4.2.2. female infants have lower risk of dying than males, with odds ratio for varying in the range 0.773-0.795. But the difference is not very significant (only 10%) . 2) Two factors are related to the mother. The first mother- related factor is her the age at birth of the infant. Here, infants born from mothers of 25-34 years of age are, on average, 0.530 times less likely to die before the first birthday than infants born from teenager mothers. The significance level of the difference varies in the range of 1- 5 percent. The second mother-related factor is her gestational status. The two variables used to capture this status— length 241 of pregnancy and birth weight— have shown a strong and significant association with infant death in Zairian cities. For example, the data indicate that infant born prematurely (less than 9 months) are, on average, 2.815 times more likely to die before the first birthday than infants born full-term. The difference is very significant (at 0.1 percent). The second variable with respect to gestational status significantly related to infant mortality in urban areas in Zaire is birth weight. It is shown that babies born with normal weight (2,500-3,499 grams) are, on average, 0.436 times less likely to die in infancy than low birth weight babies. The difference is very significant (0.1%). Similarly, babies born with weight 3,500 grams or more are 0.443 times less likely to die in infancy than low birth weight infants. The latter difference is also strongly significant at 0.1 percent. 3) The only social factor strongly associated with infant mortality in urban areas in Zaire is the number of bedrooms available in the housing unit where the family lives. Evidence has indicated that a one-unit increase in bedroom corresponds with a decline in infant mortality of 11.6 percent (odds ratio equals 0.884). This factor is significant at 1 percent. As I mentioned above, this factor is probably capturing the effect of infectious and respiratory diseases such as measles, etc. Chart 4.2.1. illustrates what has been said about the significant factors and determinants of infant mortality in urban areas of Zaire. 242 After infancy, especially in the age period 1-4 years, the evidence from urban Zaire has indicated that mortality is dominated by socio-economic factors rather than by demographic and/or biological factors. The effect of maternal schooling on childhood mortality is negative and statistically significant. On average, a one-year increase in mother's level of education deflates child mortality by 5 percent. The p-value shows that maternal education is a very significant factor, with significance level varying around 1 percent. These results imply that mothers with primary school education (6 years of schooling in Zaire) will experience child mortality 30 percent lower than uneducated mothers, and mothers with 10 years of schooling will experience child mortality 50 percent lower, while mothers who completed at least the secondary school will experience child mortality at least 60 percent lower than uneducated women. Besides maternal education, my dissertation has revealed that there are four additional socio-economic factors which are significantly associated with child mortality in urban Zaire. The first one is the place of delivery. According to the evidence, children born in health institutions (health center, maternity hospital, etc.) have, on average, child mortality 34.2 percent lower than those born at home. The difference is very significant, with significance level varying between 1 and 5 percent depending on the models considered (see Table 4.2.3.). Thus, in urban Zaire, the place 243 of delivery may be a proxy of the ability of the mother to access modern health services rather than a proxy of the quality or type care received during the delivery process. This because if the latter explanation was true then the place of delivery would be strongly associated with infant mortality than childhood mortality. The second socio-economic factor associated with child mortality in Zairian cities is the type of neighborhood of residence. Clearly children whose parents live in high standing neighborhoods where modern facilities (covered road, electricity, sewage system, piped water, health services, stores, etc.) are available, experience, on average, childhood mortality 45.3 percent lower than children whose parents are dwellers in squatter neighborhoods. The difference is very significant with p-value 4 0.001. In addition, children whose parents live in average neighborhoods, which possess some modern facilities such as piped water, stores, health facilities but not usually a good sewage system, experience— on average— child mortality only 15.9 percent lower than children from squatter neighborhoods. Here, the difference is not statistically significant. The type of neighborhood of residence is the most significant factor of child mortality in urban areas in Zaire. Our study has indicated that rich neighborhoods by itself captures the income, the quality of immediate environment, the access to health services and other modern facilities. In case of health services, for example, I 244 indicated, in a previous study, that health services and health personnel (physicians, pharmacists, nurses, etc.) in Kinshasa, the capital of the country, are concentrated in high standing neighborhoods where reside the high strata of Zairian society (Luyinduladio, 1990). The third significant socioeconomic factor of child mortality in urban Zaire is the mother's occupational status. Chapter 4 of this dissertation has shown that children of working mothers experience, on average, child mortality 50 percent higher than housewives. The difference is significant at the 5 percent level. The fourth determinant of child mortality shown in this study is the number of bedrooms the household possesses. However, our study reveals that the impact of the number of bedrooms is not strongly significant (at 10% only) in the age period 1-4 years as it is in infancy. Chart 4.2.2. illustrates what has been said about the significant determinants of child mortality in urban areas of Zaire. In chapter 5, I have investigated the impact of maternal education on the factors depicted in the literature as constituting the proximate determinants through which socioeconomic factors exert their influence on infant and child mortality. Our goal was to really determine whether education plays a key role on these proximate factors. Before pursuing further I shall mention here the limitations of our study in assessing the impact of education. 245 The data used in this study come from a cross-sectional survey in which some key variables such as nutritional status of children, morbidity status of children the last 15 days before the survey, immunization status of children 12-23 months of age, etc., are known only for surviving children, especially those who were still resident in the households surveyed. Because of these limitations I could not apply a linear structural equations method (or a strict path analysis) which would allow the estimation of the total, direct and indirect effects of education on infant and child mortality. As expected, the evidence displayed in Chapter 5 has confirmed that maternal schooling reduces the likelihood of low birth weight, but the association is not as much significant as it would be expected (see Chart 5.1.1.). Instead, the data reveal that mother's education significantly improves the quality of reporting birth weight. The evidence displayed in Table 5.1.1. also suggests that the effect of maternal education on birth weight is not as strong as the effect of household income. In higher income households, it seems that the effect of education on birth weight is weak. But this effect is stronger in low income families than in high income families. Premature births are significantly reduced (p-value i 0.01%) as the mother is educated. As displayed in Chart 5.1.1., on average, a one-year increase in maternal schooling corresponds to a decline in prematurity risk of 6.35 percent. 246 This implies that 6 years of schooling will decrease prematurity by 38.1 percent, while 12 years of education will correspond to a decline of 76.2 percent. Education of the mother is also positively associated with the use of health services for delivery. Here, the data indicate that a one-year increase in mother's education increases the use of health institutions for delivery by 16.7 percent. Here, household income is not a key factor for the use of maternal health services. Rather the type of neighborhood of residence is the most significant factors of the use of maternal health services. Maternal schooling does have a positive impact in improving the nutritional status of preschool children. For example, Chart 5.1.1. indicates that a one-year increase in mother's education reduces the odds of severe malnutrition by 1.8 percent. This implies that 6 years of schooling by mothers will be accompanied with 10.92 percent reduction in severe malnutrition of their children, and 48 percent reduction in moderate malnutrition. The result reveals that household income is not an important factor of child nutrition in urban areas in Zaire. This study has also indicated that the education of the mother is negatively associated with the occurrence of diarrhoea among children under five years. The association is significant when one controls only for children's demographic and physiological factors (age, gender, birth order) and 247 mother's nutritional status while pregnancy. But the association is no longer significant— though it remains negative— when the age of the mother, her marital and occupational status, indicators of housing quality, and the type of neighborhoods of residence, are taken into account. Higher incidence of diarrhoea among children of age 4-6 and 7-11 months, among those born premature, and among those born with low birth weight appears to be the main reason why in urban areas in Zaire infant mortality is higher for children born with low birth weight and those born premature. In addition, diarrhoea appears to be the principal reason children of mothers aged 25-34 have lower infant mortality than infants of teenager mothers. The incidence of diarrhoea appears to be higher among households with high income than among poor households. Our data sets do not have enough information to explain this puzzle. This study has also investigated the effect of maternal education on mother's behaviors with respect to child diarrhoea. The appropriate evidence has indicated that: 1) Mother's education has a positive and significant impact on the use of sugar-salt-solution, a simple and cheap technology aims at preventing dehydration of the child who is suffering from diarrhoea. The level of significance of education is in the range 5-10 percent; 2) Maternal education competes with household income in its effect on the use of modern drugs to treat diarrhoea. When one 248 controls only for characteristics related to the child, the impact of maternal education appears positive and significant (5%). In this case, household income is not significant. But when one controls for factors related to the mother and the type of neighborhood, the effect of education becomes insignificant, at the same time household income appears significant and positive. 3) Maternal education is not a key factor for the use by the mother of health service when the child has diarrhoea. In addition, the mother's use of health service in this situation is not significantly determined by household income. Here, the most important factor is the previous contact of the mother with modern health services. 4) Maternal education is also not a key factor that stimulates the mother to promptly seek modern health services. Rather, the evidence indicates that promptness in seeking modern treatment for the child in urban Zaire is significantly determined by income. About the effect of maternal education on the incidence of fever associated with malaria among preschool children identified in the survey, this study has revealed mixed results. Maternal education is a significant factor which works to reduce incidence in child fever by 2.04 percent. At the same time maternal education is not a significant factor of low incidence in fever when variables pertaining to the quality of immediate environment of housing unit and the type 249 of neighborhood of residence are taken into account. This study has also investigated the effect of maternal education on mother's behavior in case her child has fever. On this issue three observations are worth mentioning: 1) Maternal education is definitely a significant and positive factor of the use of modern health services when the child has malaria. A one-year increase in education of mother significantly (at 5%) increases the use of modern health service for malaria by about 4 percent, meaning that 6 years of mother's schooling will increase the use of health services by 24 percent, while 10 years of mother's education will rise the use of health services for child malaria by 40 percent. This is significant at 5 percent. 2) Education of the mother is a significant and positive factor of the use of modern drugs, but education is no longer significant (though its effect remains positive) when mother's demographic factors, housing factors, and type of neighborhood of residence are taken into account. 3) It is not maternal education nor household income which significantly affects the quickness in seeking modern treatment, but it is the type of neighborhood of residence. Clearly, children who live in rich neighborhoods are brought quickly to health services than children from squatter neighborhoods. 250 6.2. CONGRUENCE OF THE FINDINGS OF THIS STUDY WITH PREVIOUS STUDIES Our study has indicated that mother's education has no significant impact on infant mortality. This result is consistent with the general pattern found in the literature (reviewed in Chapter 1 of this study) related to the topic using data from other Third World countries. This general pattern suggests that "education-mortality relationship is stronger in childhood than in infancy" (Cleland and van Ginneken, 1988, p.1359). However, our result is in contradiction with some particular studies such as Victoria et al. (1992) which, after using the evidence from Brazil, has shown that the impact of maternal education on infant mortality is significant after adjusting for confounders. My study also suggests that in infancy the risk of dying decreases with the age of the mother at birth of the child. For example, the evidence produced in Table 4.2.3. shows that the risk of infant death is 16.4%, 47.0%, and 38.3% lower for mothers of ages 20-24, 25-34, and 35 or more, respectively, than for teenager mothers. This finding is consistent with a recent study on Liberia by Ahmad et al. (1991). However, my finding is in sharp contradiction with the studies by Hobcraft et al. (1983, 1984, 1985) which, using the WFS data from a wide array of developing countries, have reported that children born to very young or very old mothers experience 251 excess mortality compared to those born to mothers in the intermediate age groups. In addition, our results are not partially congruent with the study by Trussell and Hammerslough (1983) which, using the WFS data from Sri-Lanka, has indicated an excess infant (and child) mortality among children born to teenager mothers and to mothers over 35 years. After infancy, our study has shown that the impact of maternal education on child survival is maintained significant and positive even after controlling for the other socio economic factors (household income, number of bedrooms, index of housing standing, etc.), for factors related to the pattern of family formation, and for the conditions of the immediate environmental of the houses, etc. Specifically, our study has found that mothers with primary school education (i.e., six years of schooling in Zaire) will experience child mortality 30 percent lower than uneducated mothers, and mothers with 10 years of schooling will experience child mortality 50 percent lower, while mothers who completed at least the secondary school (i.e., twelve years of schooling in Zaire) will experience child mortality at least 60 percent lower than uneducated women. These results corresponds roughly with those presented by Caldwell and McDonald (1981) and Caldwell (1979), using data from Ibadan in Nigeria. In effect, relying on Nigerian evidence, the two authors indicated that Children with mothers with primary schooling to have mortality 20-50 percent lower than those with 252 uneducated mothers, and for those with mothers with secondary and tertiary education to be 30-60 percent lower and 60-90 percent lower respectively (Caldwell and McDonald, 1981, p.85). The importance of mother's schooling in lowering child mortality (independently from the other socioeconomic factors) found in our dissertation is also consistent with many other studies such as the study by Wolfe and Behrman (1982, 1987) in Nicaragua, the study by Farah and Preston (1982) in the Sudan, the study by Cochrane et al. (1982) , which is a review of literature, the study by Victoria et al. (1992), etc. Our finding is particularly consistent with a Bicego and Boerma (1993) who, after using the data of Demographic and Health Surveys conducted over the period 1987-1990 in 17 countries,1 indicated that the pattern of family formation does little to explain the advantage of the education of the mother in child survival. In that subject matter our study is also congruent with previous finding by Hobcraft et al. (1984, 1985) , Cleland and van Ginneken (1988). However, it is worth noting that our study included only two variables of the pattern of family formation: the age of the mother at birth of the child and the birth order. It did not include the birth interval between children. 1 Are included four Latin America and Caribbean countries (Bolivia, Colombia, Dominican Republic, and Guatemala), eight countries of Sub-Saharan Africa (Burundi, Ghana, Kenya, Mali, Senegal, Togo, Uganda, and Zimbabwe), three North African countries (Egypt, Morocco, Tunisia) and two Asian countries (Thailand and Sri-Lanka). For more information read (Bicego and Boerma, 1993, p.1210). 253 Although our study agrees with Caldwell and McDonald's (1981) study on the level of impact that the education of the mother has on reducing child death in the household, it differs from both authors in the order of importance of the factors associated with the decline of child mortality. In my study, the evidence displayed in Table 4.2.6. indicates that the greatest impact in reducing child mortality in urban areas of Zaire is exerted by the type of neighborhood of residence, while Caldwell and McDonald (1981) have indicated that education has the greatest impact in reducing child mortality. In our study, the evidence establishes that children whose parents live in rich neighborhoods have, on average, 45.3 percent lower child mortality than children who live in squatter areas. The difference is very significant (at 0.1%). This determinant is followed by the place of delivery, which indicates that a child whose delivery took place in a maternity hospital has, on average, 34.2 percent lower risk of dying in childhood period than a child whose delivery took place at home. Maternal education is in third position in the business of reducing child mortality. In these particular findings of our study are also in total contradiction with the studies by Farah an Preston (1982) in the Sudan, Trussell and Hammerslough (1983) in Sri-Lanka, O'Toole and Wright (1991) in Burundi, Majumber and Islam (1993) in Bangladesh, etc., which have confirmed the statement that maternal education in the Third World is the key factor in lowering child mortality. 254 In addition, our research has indicated that children of working women have 58.4 percent higher child mortality than children of housewives. The difference is significant (at 5%) . The same result was found by Farah and Preston (1982) using the evidence from the Sudan. Both authors found that children of working women in Sudan have higher mortality than children of housewives. One of the explanations suggested by the two authors is that children of working mothers were more deprived of adequate child care. Though our study shows that maternal education in urban areas in Zaire influences positively and significantly the likelihood for the mother to work outside the home rather than being a housewife, the evidence analyzed in Chapter 5 does not fully agree with Farah and Preston's explanation that children of working mothers are more deprived of adequate child care. This evidence indicates that children of working mothers compared to those of housewives: 1) Are 1.470 times more likely to be brought to modern health service to be treated for diarrhoea. At first the difference between working mothers and housewives is significant, but later the difference is no longer significant when the type of neighborhoods of residence in taken into account; 2) Are 1.558 times more likely to be brought earlier to modern health services when they are suffering from diarrhoea. Here again, the difference becomes not significant when the type of neighborhood of residence is taken into account (Table 255 5.4.7.); 3) Are 1.167 times more likely to suffer from malaria but the difference becomes insignificant when the type of neighborhood of residence is controlled for (Table 5.4.9.); 4) Are 1.507 more likely to be brought to modern health services to treat malaria. The difference is significant at 5 percent (Table 5.4.10.). The assessment of the explanations suggested by the health transition theory about the pathways through maternal education exerts an impact on child mortality has yielded mixed findings. For example, about the education-gestational status relationship, our findings agree, on the one hand, that educated mothers have a significantly lower odds of giving birth to premature babies than the uneducated women. Here, household income is not an important factor at all. On the other hands, our findings disagree with the health transition theory when one considers the birth weight factor. Here, the relationship between maternal education and birthweight of the baby is not statistically significant, though it is positive. In contrast, our finding indicates that household income is the key factor for reducing low birth weight in the household. A mixed message is also found in our study on the issue of the use of modern health services. In the first place, our study is consistent with the assumption that educated mothers will use health institutions more than the uneducated ones because school integrates women to a new society in which 256 health problems are solved using a modern medical approach. This assumption is specifically confirmed for the association maternal education-use of maternity hospital for delivery. Indeed, our study indicates that mother's education is the key determinant of the use of maternity hospital for delivery. Household income is not at all a significant factor. However, maternal education is not the greatest factor in this relationship. The impact of the type of neighborhood of residence compete with that of education. This assumption is also confirmed for the association maternal education-use of health services for child malaria. Here, our findings indicate that the higher the education of the mother the higher her use of modern health services when her child suffers from malaria. This net effect of education is significant. However, the results also indicate that the impact of education is reduced, but still remains significant, when household income increases. In addition, the results indicate that the greatest impact which trigger the mother's use of modern health services when the child has malaria is exerted by two factors: mother's previous contact with health institutions and the type of neighborhood of residence. In contrast, our findings are not consistent with the assumption mentioned above concerning the use of modern health services by the mother when the child has diarrhoea. In this case household income is also not a key factor. Here, our study has found three important determinants of the use of 257 health services when a child has diarrhoea. These are by order of importance (1) type of neighborhood of residence, (2) previous contact with health institutions, and (3) the age of mother. Our study also yields a mixed message on the issue concerning prevention. Let us recall that the theory of health transition assumes that female education increases her capacity in health management. The latter is defined as the "behavior that prevents sickness from occurring or limits the damage once it does occur" (Caldwell, 1990, p.58). Two indicators are included in our study to test this assumption: use of sugar-salt solution and completeness in immunization. The first factor, the use of SSS, is consistent with the theory. The use of sugar-salt-solution is a simple and inexpensive strategy recommended to mothers in order to prevent dehydration among children who are suffering from diarrhoea. Maternal education is indeed a key factor which enhances the use of SSS. But the greatest factor of the use of SSS is the marital status of the mother, with women in monogamous marriage more likely than single mothers to use SSS. About the completeness of immunization, our study does not confirm the assumption stated by the health transition theory. In this subject matter, our finding indicates that maternal education in urban areas of Zaire is not a key factor that enhances that odds of a child being completely immunized, though the association is positive. This result is congruent 258 with the study by Streatfield et al. (1990) which has used data from Indonesia. The latter study has evidenced that the knowledge of the vaccine functions is the key factor of the child's being immunized in Indonesia. Our findings are completely in disagreement with the health transition theory on the issue of promptness in seeking health modern health care when the child suffers from diarrhoea or malaria. From this theory, one would imply that educated mothers will be more prompt in seeking medical care when her child is sick because this theory assumes that educated mother believes in that institution to begin with, because education gives women the power and confidence to take decision-making into their hands, and because education increases positively the nature of mother-child relationship. What the evidence collected in urban Zaire tells us is that when the child has diarrhoea or malaria the education of his/her mother is not a significant factor which boosts the mother's quickness in seeking modern health care. In case of diarrhoea, this maternal behavior is significantly promoted by the household income, the type of neighborhood of residence (with mothers from squatter neighborhoods less prompt than mothers in rich areas) , and by the marital status (with unmarried women more prompt in seeking health care than women in monogamous marriage) . In case of malaria-related fever, the promptness in seeking modern health care is not triggered by the household income, rather the type of neighborhood of 259 residence (mothers living in squatter areas are less prompt than mothers from rich areas). The type of neighborhood is an indicator of access to modern infrastructure, to high income, to modern information, and access to a new life style which can modify the belief system of people who live in there. About the nutritional status of children of 12-59 months of age, our findings are consistent with the health transition theory which implies that children of educated mothers would be better nourished than those of uneducated women because the former have a larger knowledge of nutritious foods for children and because educated mothers would be more likely than the uneducated ones to avoid the practice of food discrimination within the household. Our study confirms that mother's education is a key force which works to reduce child malnutrition. It shows that the education-child malnutrition association is negative for the two types of malnutrition, severe and moderate malnutrition. However, the impact of education is significant only for moderate malnutrition not for severe malnutrition. Household income is not a significant factor of child nutritional status. These results are consistent with obtained by Wolfe and Behrman (1982, 1987) in Nicaragua which shows strong significance between mother' schooling and child nutrition, while income is not a significant factor. However, our results contradict the work of World Bank (1980) and Ward and Sanders (1980) v/hich reported that generalized increase in income will improve 260 health and nutrition of children in developing countries. Besides maternal education, our findings have indicated three additional factors of malnutrition among children 12-59 months of age. The first factor is the age of the child. Here, the study has shown that children aged 18-35 and 36-59 months are significantly more likely than children aged 12-17 months to be severely and moderately malnourished. Considering that the longitudinal study by van den Broeck (1993) in the Northern Zaire has indicated that only severe (or extreme) malnutrition is associated with high risk of child mortality while the relationship between moderate malnutrition and child mortality is inconclusive if not conflicting, I can imply that higher severe malnutrition in late childhood (18-35 and 36-59 months of age) than in late infancy (12-17 months of age) observed in our study is probably the main reason child mortality is high in the 13 Cities surveyed. The second additional factor includes the birth order of the child. Our study has shown that children of birth orders 4-6 and 7 or more have a higher risk of moderate malnutrition than first born children. The difference is statistically significant at 5 percent. The third additional factor of child malnutrition shown in this study is the type of neighborhood of residence. Here, it is shown that severe malnutrition is significantly higher among children who live in the squatter neighborhoods than among those who live in average standing areas. In contrast, moderate malnutrition is significantly 261 higher among children who live in average standing areas than those who live in squatter neighborhoods. The findings of this study also yields a mixed message concerning the relationship between mother's schooling and the incidence of diarrhoea and fever among children under five years in 1987. About diarrhoea, our findings indicate that the higher the education of the mother the lower her children will suffer from this ailment. The relationship between the two variables is significant only when one controls for the child's age, birth order, birth weight, and length of pregnancy. Maternal education is no longer a significant factor when one control for the mother's age at birth of the child, mother's marital and occupational status, type of neighborhood of residence, etc. About fever, our study has revealed that education has effectively a significant impact on the incidence of fever. But the significant level of mother' schooling on fever disappears when the index of housing quality, the type of neighborhoods, etc., are controlled for. Here, household income is not a significant factor of fever, though its impact is negative. Let me recall that the theory of health transition assumes that female education increases her capacity in health management. This implies that the incidence of diarrhoea and fever would be lower among children of educated mothers than among children of uneducated ones. 262 6.3. SUGGESTIONS FOR FUTURE RESEARCHES In light of our study, I would make four recommendations for future researches. The first suggestion stems from the fact that our study has shown that the type of neighborhood of residence is a crucial determinant of child survival. Thus, I recommend that future researches in child or maternal health should include the understanding of how the immediate environment (be it physical or social) determines the health of children. For example, future researches should include the understanding of inequality in the distribution between the residential settings of the basic social infrastructures (e.g., sewage and water supply systems; availability of accessible roads, health services, and of collective garbage collection; etc.) and the impact of this inequality on health at the household level. In addition, in the light of the results provided in our study, I suggest a longitudinal study based on reduced sample size in order really to assess the pathways through which status enhancement by women influences child and maternal health. Only through this type of study would research be able to determine these pathways by allowing the computation of the total, direct, indirect, and spurious effects of education on child health, using the linear structural equation modelling. 263 In addition, only through longitudinal data would crucial information which is missing in cross-sectional type of data (e.g., nutritional status of children dead, their immunization status, their morbidity history before death, etc.) be collected and analyzed. Further, I suggest the intensive use of qualitative approach in the data collection to allow a deep understanding of the determinants of health both at household and community levels. At household level qualitative data would provide rich insights on parents'(especially mother's) health management capability, child rearing practices, parents' behavioral responses to sickness, and on household decision-making and intra-household resource allocation, etc. At community level, qualitative data would allow the understanding of social beliefs, attitudes, and obstacles to health, and social resources to fight diseases. Finally, I suggest that more research be done on the evaluation of uses of health services (e.g. prenatal care and post-natal, immunization services, etc.), and on the impact of these services on maternal and child health. 6.4. POLICY RECOMMENDATIONS Our study suggests four policies alternatives aimed at reducing infant and child mortality in Zaire. The first recommendation consists of increasing the efforts for 264 education, of women particularly, for four reasons. First, education helps women to increase their control over their own destiny, especially on fertility issues. Second, education improves their financial status by increasing their labor force participation in modern sector. Therefore, education increases the household income. Third, education of the mother specially helps to reduce the likelihood of premature and low birth weight. Fourth, and more important, education increases women integration into the global system of the world in which knowledge and skills are the basic weapon for understanding and applying health, political, technical information. While education should continue to benefit as a national priority in social policy, greater attention should be directed towards the day care practices of children of working women. Here, I advocate the promotion of accessible and affordable private institutions which would specialize in taking care of preschool children when their parents are at work. To avoid abuses, however, maximum attention should be directed towards the functioning of these institutions. The second policy recommendation I will make concerns the health policy. I suggest that promotional campaigns using both the media and exiting health facilities be increased in order to make mothers aware on the importance of monitoring their own nutrition during pregnancy, the use of prenatal care services on the survival of the baby to be borne. In addition, mothers should be made aware on why and how to monitor the 265 nutritional status of their children. However, the success of this policy recommendation depends on the appropriate policy measures which aim at insuring accessibility and affordability of foods, maternal and child care services, and ante/postnatal care facilities for all pregnant women. I also suggest that the policy of primary health care be given additional supports in Zaire so that simple and cheap health strategies be boosted in order to reduce infant and child mortality, and to improve health of the population in general. However, reinforcing the strategy of primary heath care should not prevent the adoption of a national health policy which aims at reducing the unequal distribution of health personnel and resources both among individuals and residential areas. The third recommendation I make concerns the promotion of family planning in Zaire in order to reduce fertility, especially to reduce fertility among teenagers, group at risk of high infant mortality. The reduction in fertility will be advantageous for three reasons. First, it will alleviate the overcrowding of bedrooms, factor which, in turn, helps to reduce the spread of infectious, respiratory, and other communicable diseases such as measles. Second, it will reduce the malnutrition of preschool children in the household by relieving the competition for food. Third, at aggregate level, fertility reduction will alleviate the burden of educational cost. The strategy of family planning should be consistent with 266 the national and local plans of development, it has to be devised, promoted, and financed by people who are the beneficiary. To avoid abuses, the strategy of family planning should not respond to the needs of foreign corporations to increase the market of their contraceptives. But family planning should be a national weapon, devised and implemented by nationals, set in order to respond to a national problem which is the maintaining of high fertility in Zaire. The fourth recommendation I will make in the study concerns the need for a national socio-economic and development plan aimed at reducing rural-urban inequalities in the distribution of basic infrastructures of development (schools, hospitals, services, etc.) by integrating these two areas. A national development plan will also aim at reducing inequality among neighborhoods within urban areas, at reducing inequality in income between the haves and the haves-not. Finally, it will aim at reducing the rural exodus which brings into the squatter neighborhoods of major cities new settlers who generally are trapped in the cycle of poverty. 267 REFERENCES Aaby, Peter, 1989, "Social and Behavioural Factors Affecting Transmission and Severity of Measles Infection," In J. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.826-842. Ahmad, Omar B., Isaac W. Eberstein, and David F. Sly, 1991, "Proximate Determinants of Child Mortality in Liberia," Journal of Biosocial Science. Vol.23, pp.313-326. Aiken, Leona S., and Stephen G. West, 1991, Multiple Regression: Testing and Interpreting Interactions. Sage Publications, Newbury Park, London, New Delhi. Alchkar, Ahmad, and William J. Serow, 1988, "The Socioeconomic Determinants of Mortality: An International Comparison," Genus, Vol.44, Nos.3-4, pp.131-151. Aldrich, John H. and Nelson, Forrest D. (1984) "Linear Probability, Logit, and Probit Models." Sage University Paper series on Quantitative Applications in the Social Sciences, 07-045. Beverly Hills and London: Sage Pubns. Allison, Paul D., 1977, "Testing for Interaction in Multiple Regression," American Journal of Sociology. Vol.83, No.l, pp.144-153. Allison, Paul D., (1984) "Event History Analysis. Regression for Longitudinal Event Data." Sage University Paper series on Quantitative Applications in the Social Sciences. 07-046. Beverly Hills and London: Sage Pubns. Antoine, Philippe and F. G. Mbodji, 1991, "La mortalite des enfants au Senegal," Etude de la Population Africaine/African Population Studies. No.5, pp.18-35. Bakutuvwidi, Makani, et al., 1985, Planification familiale. fecondite et sante familiale au Zaire 1982-1984. Rapport sur les resultats d'une enouete regionale sur la prevalence contraceptive. Institut National de la Statistique. Westinghouse Public Applied Systems. Bankole, Akinrinola and David 0. Olaleye, 1991, "The Effects of Breastfeeding on Infant and Child Mortality in Kenya," Paper Prepared for Presentation at the Demographic and Health Survey World Conference. August 5-7, Washington, D.C. 268 Barbieri, Magali, 1991, "The Socio-Economic and Cultural Context of Infant and Child Mortality in Sub-Saharan Africa," Paper Prepared for Presentation at the Demographic and Health Surveys World Conference. August 5-7, 1991, Washington, D.C., 15 Pages + Annexes. Beenstock, Michael and Patricia Sturdy, 1990, "The Determinants of Infant Mortality in Regional India," World Development. Vol.18, No.3, pp.443-453. Behrman, J.C., and B.L. Wolfe, 1987, "How Does Mother's Schooling Affect Family Health, Nutrition, Medical Care Usage, and Household Sanitation," Journal of Econometrics. 36, pp.185-204. Bicego, George T. and J. Ties Boerma, 1991, "Maternal Education and Child Survival: A Comparative Analysis of Survey Data from 17 Countries," Social Science and Medicine. Vol.36, No.9, 1993. Blalock Jr., M. Hubert, 1979, Social Statistics. Revised Second Edition, McGraw-Hill Book Company. van Den Broeck, Jan, Roger Eeckels, and Jacques Vuylsteke, 1993, "Influence of Nutritional Status on Child Mortality in Rural Zaire," Lancet. Vol.341, No.8859, pp.1491-1495. Caldwell, John C., 1979, "Education as a Factor in Mortality Decline: An Examination of Nigeria Data," Population Studies. Vol.33, No.3, pp.395-413. Caldwell, John C., 1981, "Maternal Education as a Factor in Child Mortality," World Health Forum. Vol.2, No.l, pp.75- 78. Caldwell, John C. and Peter McDonald, 1981, "Influence of Maternal Education on Infant and child Mortality: Levels and Causes," International Population Conference. Manila, Vol.2, IUSSP, pp.79-96. Caldwell, John C., P. Reddy, and P. Caldwell, 1983, "The Social Component of Mortality Decline: An Investigation in South India Employing Alternative Methodologies," Population Studies. Vol.37, pp.185-205. Caldwell, John C., and P. Caldwell, 1985, "Education and Literacy as Factors in Health," in S.B. Halstead, J.A. Walsh, and K.S. Warren (eds.), Good Health at Low Cost. The Rockefeller Foundation, New York, pp. 269 Caldwell, John C., 1986, "Routes to Low Mortality in Poor Countries," Population and Development Review. Vol.12, No.2, pp.171-220. Caldwell, John C. et al., 1989, "Introductory Thoughts on Health Transition," In J. Caldwell et al. (eds.), What We Know About Health Transition; The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.xi-xiii. Caldwell, John C. et al., 1989, "Cultural, Social and Behavioural Determinants of Health and their Mechanisms: A Report on Related Research Programs," In J. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.534-541. Caldwell, John C., 1990, "Cultural and Social Factors Influencing Mortality Levels in Developing Countries," Annals of the American Academy of Political and Social Science. Vol.510, pp.45-59. Caldwell, John C., 1993, "Health Transition: The Cultural, Social, and Behavioural Determinants of Health in the Third World," Social Science and Medicine. Vol.36, No. 2, pp.125-135. Casterline, John B., Elizabeth C. Cooksey, and Abdel F.E. Ismail, 1989, "Household Income and Child Survival in Egypt," Demography. Vol.26, No.l, pp.15-35. Cebu Study Team, 1991, "Underlying and Proximate Determinants of Child Health: The Cebu Longitudinal Health and Nutrition Study," American Journal of Epidemiology. Vol.133, No.2, pp.185-201. Chen, Lincoln C., 1983, "Child Survival: Levels, Trends, and Determinants," in Rodolfo A. Butalao and Ronald D. Lee (eds.), Determinants of Fertility in Developing Countries, Vol.l, Supply and Demand for Children, Academic Press, New York, pp.199-232. Cleland, John G., and Jerome F. van Ginneken, 1988, "Maternal Education and Child Survival in Developing Countries: The Search for Pathways of Influence," Social Science and Medicine. Vol.27, No.12, pp.1357-1368. Cleland, John, 1989, "The Idea of the Health Transition," In J. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.xviii-xix. 270 Cleland, John G., 1989, "Maternal Education and Child Survival: Further Evidence and Expectations," in J. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.400-419. Cleland, John, George Bicego, and Greg Fegan, 1992, "Socioeconomic Inequalities in Childhood Mortality: The 1970s to the 1980s," Health Transition Review. Vol.2, No.l, pp.1-18. Cochrane, S.H., D.J. O'Hara, and J. Leslie, 1980, "The Effects of Education on Health," World Bank Staff Working Paper No.405. World Bank, Washington, D.C. Cochrane, S.H., J. Leslie, and D.J. O'Hara, 1982, "Parental Education and Child Health: Intracountry Evidence," Health Policy and Education. 2, pp.213-250. Cornia, Giovanni Andrea, 1984, "Analyses des etudes transversales et chronologiques sur les facteurs influengant le bien-etre des enfants," in R. Jolly et G.A. Cornia (eds.), L'impact de la recession mondiale sur les enfants. pp.38-64. Cornia, Giovanni Andrea, 1984, "Synthase et interpretation des donnees," in R. Jolly et G.A. Cornia (eds.), L'impact de la recession mondiale sur les enfants. pp.387-406. Dankert, G., and J. van Ginneken, 1991, "Birth Weight and Other Determinants of Infant and Child Mortality in Three Provinces of China," Journal of Biosocial Science. Vol.20, No.4, pp.477-489. Dargent-Molina, Patricia, 1994, "Association Between Maternal Education and Infant Diarrhea in Different Household and Community Environments of Cebu, Philippines," Social Science and Medicine. Vol.38, No.2, pp.343-350. Defo, Barthelemy Kuate, 1992, "Mortality and Attrition Processes in Longitudinal Studies in Africa: An Appraisal of the Iford Surveys," Population Studies. Vol.46, pp.327-348. Easterlin, Richard A., and Eileen M. Crimmins, 1985, The Fertility Revolution. A Supply-Demand Analysis. University of Chicago Press, Chicago. Ebomoyi, Ehigie et al., 1991, "Birthweight and Sociobiological Factors in Ilorin, Nigeria," Journal of Biosocial Science. Vol.23, pp.417-423. 271 Ewbank, D., R. Henin, and J. Kekovole, 1986, "An Integration of Demographic and Epidemiologic Research on Mortality in Kenya," In Determinants of Mortality Changes and Differentials in Developing Countries. Population Studies No.94, United Nations Department of International Economic and Social Affairs, New York. Ewbank, Douglas, 1988, "Health in Africa," in Etienne van de Walle, Patrick 0. Ohadike, and Mpembele D. Sala-Diakanda (eds.), The State of African Demography. International Union for the Scientific Study of Population, pp.85-101. Ewbank, Douglas C., and James N. Gribble (eds.), 1993, Effects of Health Programs on Child Mortality in Sub- Sahara Africa. Population Dynamics of Sub-Sahara Africa, National Academy Press, Washington, D.C. Farah, Abdul-Aziz, and Samuel H. Preston, 1982, "Child Mortality Differentials in Sudan," Population and Development Review. Vol.8, No.2, pp.365-383. Flegg, A.T., 1982, "Inequality of Income, Illiteracy and Medical Care as Determinants of Infant Mortality in Underdeveloped Countries," Population Studies. Vol.36, No.3, pp.441-458. Galway, Katrina, Brent Wolff, and Richard Sturgis, 1987, Child Survival: Risks and the Road to Health. The Demographic Data for Development Project, Institute for Resource Development/Westinghouse. Gaminiratne, K.H.W., 1991, "Socio-Economic and Behavioral Determinants of Diarrhoeal Morbidity in Sri Lanka," Paper prepared for presentation at the Demographic and Health Survey World Conference. August 5-7, 1991, Washington, D.C. Garenne, Michel, and Francine van de Walle, 1985, "Knowledge, Attitudes and Practices Related to Child Health and Mortality in Sine-Saloum, Senegal," International Population Conference. Florence, Vol.4, IUSSP, pp.267- 277. Gbesemete, P. Kwame, and Dick Jonsson, 1993, "A Comparison of Empirical Models on Determinants of Infant Mortality: A Cross-National Study on Africa," Health Policy. Vol.24, pp.155-174. 272 van Ginneken, Jeroen K., 1991, "Childhood Diarrhoea Morbidity and Treatment Patterns: A Comparison of Results in Demographic and Health Surveys with Diarrhoeal Diseases," Paper prepared for presentation at the Demographic and Health Surveys World Conference. August 5-7, 1991, Washington, D.C. Grosse, Robert N., and Christopher Auffrey, 1989, "Literacy and Health Status in Developing Countries," Annual Review of Public Health. Vol.10, pp.281-297. Gubhaju, Bhakta, K. Streatfield, and A.K. Majumder, 1991, "Socioeconomic, Demographic and Environmental Determinants of Infant Mortality in Nepal," Journal of Biosocial Science. Vol.23, pp.425-435. Halli, Shiva S., and K. Vaninadha Rao, 1992, Advanced Techniques of Population Analysis. The Plenum Series on Demographic Methods and Population Analysis, Plenus Press, New York and London. Heer, David M., and Jill S. Grigsby, 1992, Society and Population. Third Edition, Prentice Hall, Englewood Cliffs, New Jersey. Hill, Althea, and Kenneth Hill, 1988, "Mortality in Africa: Levels, Trends, Differentials and Prospects," in Etienne van de Walle, Patrick O. Ohadike, and Mpembele D. Sala- Diakanda (eds.), The State of African Demography. International Union for the Scientific Study of Population, pp.67-84. Hobcraft, J.N., J.W. McDonald, and S.O. Rutstein, 1983, "Child-Spacing Effects on Infant and Early Child Mortality," Population Index. Vol.49, No.4, pp.585-618. Hobcraft, J.N., J.W. McDonald, and S.O. Rutstein, 1984, "Socioeconomic Factors in Infant and Child Mortality: A Cross-National Comparison," Population Studies. Vol.38, pp.193-223. Hobcraft, J.N., J.W. McDonald, and S.O. Rutstein, 1985, "Demographic Determinants of Infant and Early Child Mortality: A comparative Analysis," Population Studies. Vol.39, pp.363-385. Houyoux, J. and N. Kinavwidi, 1986, Kinshasa 1975. ICHEC. Huldelson, M. Patricia, 1993, "ORS and Treatment of Childhood Diarrhea in Managua, Nicaragua," Social Science and Medicine. Vol.37, No.l, pp.97-103. 273 Jejeebhoy, Shireen J., 1991, "Women's Status and Fertility: Successive Cross-Sectional Evidence from Tamil Nadu, India, 1970-80," Studies in Family Planning. Vol.22, No.4, pp.217-230. Jelliffe, Derrick B., 1966, The Assessment of the Nutritional Status of Community. World Health Organization, Geneva, 271 pages. Kanji, Gopal K., 1993, 100 Statistical Tests. Sage Publications, Ltd, London, Newbury Park, New Deli. Kim, Tai-Hun, 1988, "Changing Determinants of Infant and Child Mortality: On the Basis of the Korean Experience, 1955- 73," Journal of Biosocial Science. Vol.20, No.3, pp.345- 355. Koch-Weser, Dieter and Alfred Yankauer, 1991, "What Makes Infant Mortality Rates Fall in Developing Countries," American Journal of Public Health. Vol.81, No.l, pp.12- 13. Korte, R., T. Rehle, A. Merkle, "Strategies to Maintain Health in the Third World," Tropical Medicine and Parasitology. Vol.42, No.4, pp.428-432. van Lerberghe, Wim, and K.A. Pangu, 1988, "Les Politiques de Sante," In D. Tabutin (ed.), Population et Societes en Afriaue au Sud du Sahara. Editions L'Harmattan, Paris, pp.335-367. LeVine, Robert A., 1980, "Influence of Women's Schooling on Maternal Behavior in the Third World," Comparative Education Review. Vol.24, No.2, Part 2, pp.S78-S105. LeVine, Robert A., Sarah E. LeVine, Amy Richman, F. Medardo Tapia Uribe, Clara Sunderland Correa, Patrice M. Miller, 1991, "Women's Schooling and Child Care in the Demographic Transition: A Mexican Case Study," Population and Development Review. Vol.17, Vol.3, pp.459-496. Lindenbaum, Shirley, M. Chakraborty, and M. Elias, 1985, The Influence of Maternal Education on Infant and Child Mortality in Bangladesh. Special Publication No.23, International Centre for Diarrhoeal Disease Research, Bangladesh, Dhaka. 274 Lindenbaum, Shirley, 1989, "Maternal Education and Health Care Processes in Bangladesh: The Health and Hygiene of the Middle Classes," In J.C. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Vol.I, Canberra, pp.426-440. Lututala, M., M. Kintambu, and M. Mvudi, 1993, "Zaire," in Aderanti Adepoju (ed.), The Impact of Structural Adjustment on the Population of Africa. The Implications for Education. Health and Employment. UNFPA, Heinemann, and James Currey, pp.130-143. Luyinduladio, N'zinga, 1978, Le raieunissement de la population de Kinshasa: ses consequences demoaraphioues et socio- economiaues. M&noire de Licence en Demographie, Universite Nationale du Zaire, Campus de Kinshasa, Faculty des Sciences iconomigues. Luyinduladio, N'zinga, April 1994, Health Care Delivery in Zaire: Organization. Rural Versus Urban Inequalities. Paper Written for Sociology 599: Seminar in Medical Sociology, Department of Sociology, University of Southern California, 54 pages. Luyinduladio, N'zinga, 1989, Morbidity and Mortality in Urban and Rural Areas in Zaire. Whv Are There Differences?. Paper Presented in Demography 777: Seminar in African Demography, University of Pennsylvania, 38 pages + annexes. Luyinduladio, N'zinga, 1990, Cause Pattern of Child Morbidity and Mortality From Hospital Data in Kinshasa. Zaire. 1982-86. Master Paper Presented in Demography 707/708, University of Pennsylvania, 48 pages + Annexes. Majumder, Abdul Kashem, and S.M. Shafiqul Islam, 1993, "Socioeconomic and Environmental Determinants of Child Survival in Bangladesh," Journal of Biosocial Science. Vol.25, No.3, pp.311-318. Martin, Teresa Castro, 1994, "Women's Education and Fertility in Latin America: Exploring the Significance of Education for Women's Lives," DHS Working Papers. No.10, 23 pages. Mason, Karen Oppenheim, 1986, "The Status of Women: Conceptual and Methodological Issues in Demographic Studies," Sociological Forum. Vol.l, No.2, pp.284-300. 275 Mbake, C., and E. van de Walle, 1987, Socio-economic Factors and Access to Health Services as Determinants of child Mortality. IUSSP Seminar on Mortality and Society in Sub- Saharan Africa, Cameroon. Millard, V. Ann, 1994, "A Causal Model of High Rates of Child Mortality," Social Science and Medicine. Vol.38, No.2, pp.253-268. Mosley, Henry W., 1984, "Child Survival: Research Policy," In W. Henry Mosley and Lincoln C. Chen (eds.), Child Survival. Strategies for Research. Population and Development Review. Supplement to Volume 10, pp.3-23. Mosley, Henry W., and Lincoln C. Chen, 1984, "An Analytical Framework for the Study of Child Survival in Developing Countries," In Henry W. Mosley and L.C. Chen (eds.), Child Survival. Strategies for Research. Population and Development Review. Supplement to Volume 10, pp.25-45. Mosley, Henry W., 1985, "Will Primary Health Care Reduce Infant and Child Mortality? A Critique of Some Current Strategies, With Special Reference to Africa and Asia," In Jacques Vallin and Alan D. Lopez (eds.), Health Policy. Social Policy and Mortality Prospects. Institut National d'lStudes Demographiques (INED), International Union for the Scientific Study of Population (IUSSP), Ordina Editions, pp.103-137. Mpando, Ladislas R.S., 1994, Determinants of Neonatal and Postneonatal Mortality: The Case of Malawi. 1992. Master Thesis, University of Southern California, Department of Sociology. Nelder, J.A., 1974, "Log Linear Models for Contingency Tables: A Generalization of Classical Least Squares," Applied Statistics. Vol.23, No.3, pp.323-329. Ngondo, a Pitshandenge, and Dr. Ruben Gamboa, 1988, Morbidite et Mortality Infantiles et Juveniles a Kinshasa. Niveaux et Determinants. Departement de la Sante et des Affaires Sociales, Fonds National M§dico-Sanitaire (Fonames), et Fonds des Nations Unies pour l'Enfance (Unicef). Ngondo, a Pitshandenge, Dr. Ruben Gamboa, N'zinga Luyinduladio, N. Kinavwidi, 1988, Morbidity et Mortalite Infantiles et Juveniles dans les Grandes Villes du Zaire. Niveaux. Tendances et Determinants. Departement de la Sante et des Affaires Sociales, Fonds National Mfidico- Sanitaire (Fonames), et Fonds des Nations Unies pour l'Enfance (Unicef). 276 Omideyi, Adekunbi K., 1993, "The Emergence of the Nuclear Family: Its Effect on Women's Roles and Fertility Change in Nigeria,” Genus. Vol.49, Nos.1-2, pp.201-216. Orubuloye, I.O., and J.C. Caldwell, 1975, "The Impact of Public Health Services on Mortality: A Study of Mortality Differentials in a Rural Area of Nigeria,” Population Studies, Vol.29, No.2, pp.259-272. O'Toole, Josephine, and Robert E. Wright, 1991, "Parental Education and Child Mortality in Burundi,” Journal of Biosocial Science. Vol.23, No.3, pp.255-262. Palloni, Alberto, 1981, "Mortality in Latin America: Emerging Patterns,” Population and Development. Vol.7, No.4, pp.623-649. Pant, Prakash Dev, 1991, "Effect of Education and Household Characteristics on Infant and child Mortality in Urban Nepal,” Journal of Biosocial Science. Vol.23, No.3, pp.437-443. Pebley, A.R., and P.W. Stupp, 1987, "Reproductive Patterns and Child Mortality in Guatemala," Demography. Vol.24, No., pp.43-60. Preston, Samuel H., 1975, "The Changing Relation Between Mortality and Level of Economic Development," Population Studies. Vol.29, No.2, pp.231-248. Preston, Samuel H., 1985, "Resources, Knowledge, and Child Mortality: A Comparison of the U.S. in the Late Nineteenth Century and Developing Countries Today," International Population Conference. Florence, Vol.4, IUSSP, pp.373-386. Puffer, R.R. and C.M. Serrano, 1973, Patterns of Mortality in Childhood. Pan American Health Organization, Washington, D.C. Romaniuk, Anatole, 1967, La fecondite des populations conaolaises. Mouton, Paris. Romaniuk, Anatole, 1968, "The Demography of the Demographic Republic of the Congo," In W. Brass et al. (eds.), The Demography of Tropical Africa. Princeton University Press, pp.241-339. Rozenzweig, Mark R., and T. Paul Schultz, 1982, Child Mortality and Fertility in Colombia: Individual and Community Effects," Health Policy Education. 2, pp.305- 348. 277 Rutstein, S.O., 1984, "Socioeconomic Differentials in Infant and Child Mortality," WFS Comparative Studies. No.40, ISI, Voorburg. SAS/STAT User's Guide, 1990, Volume 2, GLM-VARCOMP, Version 6, Fourth Edition, 2nd printing, SAS Institute Inc. SAS/STAT User's Guide, 1988, Release 6.03 Edition, 4th printing, SAS Institute Inc. Schlotzhauer, D. Sandra, and Ramon C. Littell, 1987, SAS System for Elementary Statistical Analysis. SAS Institute Inc. Schultz, Paul T., 1984, "Studying the Impact of Household Economic and Community Variables on Child Mortality," In W. Henry Mosley and Lincoln C. Chen (eds.), Child Survival; Strategies for Research. Population and Development Review, Supplement to Volume 10, pp.215-235. Schultz, Paul T., 1985, "Household Economic and Community Variables as Determinants of Mortality," In International Population Conference Florence. Vol.2, IUSSP, Lidge, pp.225-236. Seoane, Nicole, and Michael C. Latham, 1971, "Nutritional Anthropometry in the Identification of Malnutrition in Childhood," Journal of Tropical Pediatrics and Environmental Child Health. Vol.17, pp.98-105. Streatfield, Kim, M. Singarimbun, 1988, "Social Factors Affecting Use of Immunization in Indonesia," Social Science and Medicine. Vol.27, No.11, pp.1237-1245. Streatfield, Kim, M. Singarimbun, and I. Diamond, 1990, "Maternal Education and Child Immunization," Demography. Vol.27, No.3, pp.447-455. Tambashe, Oleko, and D. Shapiro, 1991, Employment. Education, and Fertility Behavior: Evidence from Kinshasa. Substantive Report, University de Kinshasa, D£partement de Demographie. Tawiah, E.O., 1989, "Child Mortality Differentials in Ghana: A Preliminary Report," Journal of Biosocial Science. Vol.21, pp.349-355. Tekge, Belgin, and Frederic C. Shorter, 1984, "Determinants of Child Mortality: A Study of Squatter Settlements," In W. Henry Mosley and Lincoln C. Chen (eds.), Child Survival: Strategies for Research. Population and Development Review. A Supplement to Volume 10, pp.257-280. 278 Trussell, James, and S.H. Preston, 1982, "Estimating the Covariates of Childhood Mortality from Retrospective Reports of Mothers," Health Policy and Education. Vol.3, pp.1-36. Trussell, James, and Charles Hammerslough, 1983, "A Hazards- Model Analysis of the Covariates of Infant and Child Mortality in Sri-Lanka," Demography. Vol.20, No.l, pp.l- 26. Tsui, Ong Amy, Julia DeClerque, and Nlandu Mangani, 1988, "Maternal and Sociodemographic Correlates of Child Morbidity in Bas-Zalre: The Effect of Maternal Reporting." Social Science and Medicine. Vol.26, No.7, pp.701-713. Unicef, 1991, La Situation des Enfants dans le Monde 1992. 100 pages. United Nations, 1982, Levels and Trends of Mortality Since 1950. Department of International Economic and Social Affairs, New York. United Nations, 1983, Manual X. Indirect Techniques for Demographic Estimation. Department of International Economic and Social Affairs, Population Studies No.81, ST/ESA/SER.A/81, New York. United Nations, 1985, Socio-Economic Differentials in Child Mortality in Developing Countries. UN Department of International Economic and Social Affairs, New York. United Nations, 1993, "Women's Education and Fertility Behaviour," Population Newsletter. No.56, pp.1-6. Uyanga, Joseph, 1990, "Economic Development Strategies: Maternal and Child Health, Social Science and Medicine. Vol.31, No.6, pp.649-659. Victora, Cesar G., Sharon R.A. Huttly, Fernando C. Barros, Cintia Lombardi, and J. Patrick Vaughan, 1992, "Maternal Education in Relation to Early and Late Child Health Outcomes: Findings from a Brazilian Cohort Study," Social Science and Medicine. Vol.34, No.8, pp.899-905. van de Walle, Etienne, 1989, "How Do We Define the Health Transition," In J. Caldwell et al. (eds.), What We Know About Health Transition: The Cultural. Social and Behavioural Determinants of Health. Volume I, Canberra, pp.xiv-xv. 279 Ward, J.O. and J.H. Sanders, 1980, "Nutritional Determinants and Migration in the Brazilian Northeast: A Case Study of Rural and Urban Ceara," Economic Development and Cultural Change, Vol.29, pp.141-163. Ware, Helen, 1984, "Effect of Maternal Education, Women's Roles, and Child Care on Child Mortality," In W. Henry Mosley and L.C. Chen (eds.), Child Survival: Strategies for Research. A Supplement to Volume 10, Population and Development Review, pp.191-214. Wolfe, Barbara L., and Jere R. Behrman, 1982, "Determinants of Child Mortality, Health, and Nutrition in a Developing Country," Journal of Development Economics. 11, pp.163- 193. Wolfe, Barbara L., and Jere R. Behrman, 1987, "Women's Schooling and Children's Health," Journal of Health Economics. Vol.6, pp.239-254. World Bank, 1980, World Development Report 1980. Washington, D.C. 280 TABLE APPENDIXES Table 4.1.1.: Distribution of Births and Deaths, and Mortality Rates. Birth Cohorts 1981-85. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. TOTAL BIRTH TOTAL DEATHS MORTALITY RATE (per thousand) MOTHER'S EDUCATION Illiterate 1356 132 97.3 1-3 766 87 113.6 4-6 2224 231 103.9 7-9 2089 169 80.9 10+ 1019 75 73.6 Unknown 16 2 CLASSES OF HOUSEHOLD INCOME (in zaires, July-August 1987) < 4,730 1673 175 104.6 4730-8936 1750 169 96.6 8937-17268 1968 150 76.2 > 17,269 2115 203 95.98 MOTHER'S AGE AT BIRTH OF CHILD -20 years 1264 157 124.2 20-24 2318 226 97.5 25-29 1721 164 95.3 30-34 1124 73 64.9 35-39 639 53 82.9 40-49 293 17 61.4 (*) 50+ 21 — — Unknown 90 6 — 283 Table 4.1.1. (continued) BIRTH ORDER 1 1346 153 113.7 2-3 2269 238 104.9 4-6 2176 183 84.1 7+ 1679 122 72.7 CHILD GENDER Male 3703 342 92.36 Female 3740 348 93.05 Unknown 27 6 — LENGTH OF PREGNANCY < 9 months 475 84 176.7 . 9 months 6792 596 87.8 . Unknown | 203 16 78.9 BIRTH WEIGHT < 2500 grams 1 701 81 115.5 2500-3499 3427 284 82.9 3500 + 2183 162 74.2 Unknown 1159 169 145.8 MOTHER'S OCCUPATIONAL STATUS Students 117 16 136.7 Housewife 6169 554 89.8 Working 865 80 92.5 Others 312 45 144.2 Unknown 6 1 — 284 Table 4.1.1. (continued) PROFESSIONAL CATEGORY OF WORKING MOTHERS Employee 296 22 74.3 Independent 555 58 102.7 Not in work force 6552 615 93.9 Others 46 1 — Unknown 14 MOTHER'S MARITAL STATUS Single 323 44 136 Married Monogamously 5596 486 86.8 Married Polygamously 894 84 93.9 Consensual Union 474 54 113.9 Others 183 28 153.0 PLACE OF DELIVERY At home( )1 1108 158 142.9 Health institutions 6362 538 84.6 YEAR OF BIRTH AND AGE IN 1987 Year Age 1981 5-6 1276 139 108.9 1982 4-5 1476 141 95.5 1983 ' 3-4 1526 135 88.5 1984 2-3 1514 150 99.1 1985 1-2 1678 131 78.1 1986 0-1 1672 107 63.99 (***) Table 4.1.1. (continued) STRATA 0 581 27 46.5 (*) 1 1629 165 101.3 2 953 80 84.0 3 1346 101 75.04 4 2590 279 107.7 6 371 44 118.6 (*) AGE OF THE CHILD 0-1 7470 578 77.4 0-5 -.Trr~ _ 7470 696 93.2 icye \ ( ' Including deliveries which occurred on the way to maternity hospital, ... in the car, etc. i c y c ( ) At the time of survey some children of this cohort did not reached one year-old. Table 4.2.2.: Odds Ratio and Standard Error for Logistic Regression of Infant Mortality. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 VARIABLES 1 2 3 4 INTERCEPT 0.053“ “ 0.078"” 0.075"* 0.082**" (0.103) (0.520) (0.889) (0.531) EDUCATION 0.978 0.982 0.988 0.983**" (0.0106) (0.016) (0.143) (0.017) HOUSEHOLD 0.957 0.961 0.961 INCOME (0.056) (0.097) (0.057) INTERACTION: Educ*Hous Income 0.999 (0.015) GENDER: . (Male) 1.000 . Female 0.780" (0.119) -2 LOG L 2581.303 2512.492 2512.490 2457.798 CHI-SQUARE 1.904 2.057 2.059 6.158 DF 1 2 3 3 Legend: «*♦» p <=6.001: »»» a <=0.01; «» p <=0.5: » p <=0.16 286 Table 4.2.2. (continued) VARIABLES 5 6 7 8 INTERCEPT 0.132— 0 .1 3 3 "" 0.166*“ 0.106“ (0.543) (0.545) (0.566) (0.590) EDUCATION 0.965“ 0.961“ 0.996 0.995 (0.018) (0.018) (0.019) (0.020) HOUSEHOLD 0.974 0.983 1.006 1.016 INCOME (0.058) (0.058) (0.052) (0.061) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 0.775" 0.773 0.795* 0.788’ (0.120) (0.120) (0.123) (0.126) MOTHER AGE AT BIRTH OF CHILD: (in years) ■ ( < 20 ) 1.000 1.000 1.000 1.000 . 20-24 0.727" 0.823 0.793 0.833 (0.151) (0.168) (0.170) (0.176) . 25-34 0.386"" 0.536"" 0.480*“ 0.527“ * (0.165) (0.224) (0.229) (0.234) . 35 or more 0 .4 6 6 "" 0.711 0.582* 0.584 (0.219) (0.315) (0.322) (0.333) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 . 2-3 0.851 (0.167) 0.891 (0.170) 0.872 (0.176) . 4-6 0.650 (0.223) 0.699 (0.229) 0.727 (0.234) . 7 + 0.590 (0.287) 0.719 (0.291) 0.756 (0.298) BIRTH WEIGHT (in grams) .(< 2,500) .2,500-3,499 . 3,500 + . Unknown 1.000 0.353— (0.178) 0.365— (0.196) 0.991 (0.217) 1.000 0.449— (0.189) 0.455— (0.209) 1.163 (0.202) LENGHT OF PREGNANCY: . (9 months) . < 9 months 1.000 2.779— (0.171) -2 LOG L 2394.448 2389.954 2262.403 2146.232 CHI-SQUARE 44.470"" 48.963""" 111.804— 130.996— DF 6 9 12 13 287 Table 4.2.2. (continued) VARIABLES 9 10 11 12 INTERCEPT 0.107” ” 0.088**” 0.087” ” 0.091” ” (0.624) (0.635) (0.636) (0.635) EDUCATION 0.995 0.994 1.000 0.999 (0.020) (0.020) (0.023) (0.020) HOUSEHOLD 1.016 1.029 1.026 1.026 INCOME (0.061) (0.062) (0.062) (0.062) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 0.788* 0.791* 0.788* 0.790* (0.126) (0.126) (0.127) (0.126) MOTHER AGE AT BIRTH OF CHILD: (in years) . ( < 2 0 ) 1.000 1.000 1.000 1.000 . 20-24 0.834 0.840 0.842 0.857 (0.176) (0.177) (0.177) (0.178) . 25-34 0.526*” 0.532“ * 0.534“ * 0.553” (0.234) (0.235) (0.235) (0.236) . 35 or more 0.584 0.594 0.599 0.628 (0.333) (0.334) (0.335) (0.335) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 1.000 . 2-3 0.872 0.902 0.896 0.898 (0.176) (0.178) (0.179) (0.179) . 4-6 0.728 0.768 0.763 0.763 (0.234) (0.237) (0.237) (0.238) . 74- 0.757 0.794 0.791 0.790 (0.298) (0.300) (0.300) (0.301) BIRTH WEIGHT (in grams) .(< 2,500) 1.000 1.000 1.000 1.000 .2,500-3,499 0.449**” 0.451**“ 0.452**” 0.450**” (0.189) (0.189) (0.190) (0.210) . 3,500 4 0.454“ ” 0.456” ” 0.457**“ 1.150 (0.209) (0.210) (0.210) (0.229) . Unknown 1.152 1.173 1.179 2.769*“ * (0.228) (0.229) (0.230) (0.172) LENGTH OF PREGNANCY: . (9 months) 1.000 1.000 1.000 1.000 . < 9 months 2.776**” 2.737“ “ 2.742”” 0.972 (0.171) (0.171) (0.172) (0.194) 288 Table 4.2.2. (continued) VARIABLES 9 10 11 12 PLACE OF DELIVERY: .(At Home) 1.000 1.000 1.000 1.000 .Maternity 0.986 0.978 0.979” * 0.639*“ (0.194) (0.194) (0.194) (0.159) MARITAL STATUS: . (Married 1.000 1.000 1.000 Monog.) . Single 1.419 1.586 1.390 (0.255) (0.455) (0.337) . Married 0.962 1.024 0.980 Poly gam. (0.208) (0.332) (0.209) . Others*0 1.284 1.494 1.317 (0.204) (0.384) (0.213) INTERA CTION: (Edu*MMon) 1.000 Edu*Single 0.981 (0.065) Edu*MarPol 0.988 (0.054) Edu‘ Others 0.973 (0.059) OCCUPATIONAL STATUS: .Housewife. Student 1.000 0.918 . Working (0.466) 0.671* . Others0 ’ * (0.240) 1.105 (0.326) -2 LOG L 2146.004 2142.999133.973” 2142.710 2139.629 CHI-SQUARE 130.969” “ 134.263” ” 137.3“ ” DF 14 17 20 20 ' This includes: consensual unions, divorced, separated, widowers; This includes: Never worked, inactives, etc. 289 Table 4.2.2. (continued) VARIABLES 13 14 15 16 INTERCEPT 0.093“ * 0.071“ * 0.072“ * 0.066“ * (0.653) (0.668) (0.672) (0.689) EDUCATION 0.998 1.044 1.046 1.045 (0.020) (0.029) (0.029) (0.030) HOUSEHOLD 1.024 1.027 1.061 1.069 INCOME (0.063) (0.063) (0.065) (0.065) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 0.789* 0.794* 0.792* 0.797* (0.127) (0.128) (0.128) (0.129) MOTHER AGE AT BIRTH OF CHILD: (in years) . (Less than 20) 1.000 1.000 1.000 1.000 . 20-24 0.857 0.887 0.902 0.903 (0.178) (0.180) (0.180) (0.181) . 25-34 0.554” 0.572“ 0.597” 0.588” (0.237) (0.241) (0.241) (0.243) . 354- 0.628 0.650 0.701 0.683 (0.336) (0.339) (0.340) (0.343) BIRTH ORDER: .(1st birth) 1.000 1.000 1.000 1.000 . 2-3 0.897 0.858 0.845 0.840 (0.179) (0.181) (0.181) (0.182) . 4-6 0.761 0.741 0.734 0.716 (0.238) (0.241) (0.241) (0.243) . 7 + 0.786 0.783 0.809 0.821 (0.301) (0.304) (0.304) (0.307) BIRTH WEIGHT: (in grams) .( < 2,500) 1.000 1.000 1.000 1.000 . 2,500-3,499 0.444“ * 0.483“ * 0.475“ * 0.477“ * (0.190) (0.194) (0.195) (0.197) . 3,500+ 0.449“ * 0.477“ * 0.470“ * 0.475“ * (0.211) (0.215) (0.217) (0.217) . Unknown 1.150 1.174 1.167 1.155 (0.229) (0.235) (0.235) (0.237) 290 Table 4.2.2. (continued) VARIABLES 13 14 15 16 LENGTH OF PREGNANCY: .(9 months) 1.000 1.000 1.000 1.000 . < 9 months 2.770” “ 2.874” ” 2.964**“ 3.047**“ (0.172) (0.174) (0.175) (0.179) PLACE OF DELIVERY: .(At home) 1.000 1.000 1.000 1.000 . Maternity 0.972 0.919 0.939 0.956 (0.196) (0.197) (0.197) (0.199) MARITAL STATUS .(Married Mono) 1.000 1.000 1.000 1.000 . Single 1.396 1.454 1.534 t.529 (0.338) (0.340) (0.342) (0.344) . Married Polygamously 0.981 0.861 0.926 0.951 . Other**’ (0.209) (0.222) (0.223) (0.224) 1.321 1.363 1.385 1.367 (0.215) (0.216) (0.216) (0.216) OCCUPATIONAL STATUS: .(Housewife) 1.000 1.000 1.000 1.000 . Student 0.915 0.925 0.963 0.979 (0.466) (0.469) (0.471) (0.473) . Working 0.673* 0.626* 0.628* 0.636* (0.241) (0.251) (0.252) (0.253) . Others'*’ ’ 1.106 1.084 1.173 1.192 (0.326) (0.328) (0.330) (0.332) ACCESS TO MODERN FACILITY .(Squatting neighborhood) 1.000 1.000 1.000 1.000 .High stand, neighborhood 1.029 1.668 1.714* 1.843* . Average standing neig (0.168) (0.320) (0.321) (0.334) 0.987 1.360 1.400 1.420 (0.153) (0.265) (0.266) (0.268) This includes: consensual union, divorced, separated w idow ed;( b > This includes: Never worked, inactives, et , c. 291 Table 4.2.2. (continued) VARIABLES 13 14 15 16 INTERACTION .(Edu*Squat) 1.000 1.000 1.000 . Edu*High stand neigh 0.907“ 0.909“ 0.911* .Edu*Average stand neigh (0.048) (0.048) (0.050) 0.938 0.942 0.945 (0.041) (0.041) (0.041) NUMBER OF 0.888“ * 0.881“ * BEDROOMS (0.044) (0.044) INDEX OF QUALITY OF THE HOUSING UNIT', material of construction .(Low: 1-7) 1.000 .Avrg: 7-10 1.068 (0.177) .High: 10-12 0.843 (0.225) -2 LOG L 2139.570 2094.675 2086.152 2046.804 CHI-SQUARE 137.4*“ * 137.66**“ 145.10“ " 147.9**“ DF 22 24 25 27 292 Table 4.2.2. (continued) VARIABLE 17 18 19 20 INTERCEPT 0.062'**' 0.077“ " 0.076“ “ 0.072“ “ (0.708) (0.659) (0.658) (0.686) EDUCATION 1.053' 1.015 1.015 1.014 (0.030) (0.021) (0.021) (0.021) HOUSEHOLD 1.075 1.076 1.078 1.078 INCOME (0.066) (0.063) (0.063) (0.063) GENDER . (Male) 1.000 1.000 1.000 1.000 . Female 0.792' 0.771" 0.770“ 0.771“ (0.130) (0.126) (0.126) (0.126) MOTHER’S AGE AT BIRTH OF CHILD . ( < 20) 1.000 1.000 1.000 1.000 . 20-24 0.909 0.817 0.819 0.820 (0.181) (0.176) (0.176) (0.176) . 25-34 0.600“ 0.561“ 0.564“ 0.564“ (0.243) (0.235) (0.235) (0.235) . 35 or more 0.703 0.626 0.632 0.632 (0.345) (0.334) (0.335) (Q.335) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 1.000 ,2nd-3rd birth 0.844 0.919 0.916 0.914 .4th-6th birth (0.182) (0.178) (0.178) (0.178) .7th and above 0.713 0.786 0.780 0.779 (0.243) (0.237) (0.238) (0.238) 0.821 0.920 0.906 0.903 (0.308) (0.298) (0.300) (0.299) BIRTH WEIGHT (in grams) .(<2500)* 1.000 1.000 1.000 1.000 .2500-3499 0.475“ " 0.442“ " 0.443"“ 0.443““ (0.197) (0.191) (0.191) (0.191) .3500 or more 0.482“ “ 0.471“ “ 0.465““ 0.466“ “ .unknown (0.218) (0.210) (0.210) (0.210) 1.160 1.216 1.192 1.196 (0.239) (0.228) (0.229) (0.229) 293 TABLE 4.2.2. (continued) VARIABLE 17 18 19 20 LENGTH OF PREGNANCY: (9 months) 1.000 1.000 1.000 1.000 . < 9 months 3.012~* 3.092“ “ 3.103“ “ 3.104“ “ (0.179) (0.171) (0.171) (0.171) PLACE OF DELIVERY: .(home) 1.000 1.000 1.000 1.000 .(health institution) 0.954“ 1.042 1.044 1.048 (0.199) (0.194) (0.194) (0.195) MARITAL STATUS OF MOTHER: .(Married monogamousiy) 1.000 1.000 1.000 1.000 .Single 1.502 1.487 1.530 1.536 (0.347) > (0.338) (0.339) (0.340) .Married polygamously 0.937 0.967 0.960 0.961 (0.224) (0.213) (0.213) (0.213) . Others'*’ 1.343 1.356 1.377 1.375 (0.217) (0.210) (0.211) (0.211) OCCUPATIONAL STATUS OF MOTHER: (Housewife) 1.000 1.000 1.000 1.000 .Student 1.022 1.024 0.995 0.991 (0.476) (0.453) (0.455) (0.455) .Working 0.637” 0.642” 0.636” 0.634* (0.253) (0.242) (0.242) (0.242) .Others'*’ ’ 1.203 1.120 1.118 1.114 (0.332) (0.328) (0.329) (0.330) 294 TABLE 4.2.2. (continued) VARIABLE 17 18 19 20 ACCESS TO MODERN FACILITY: . (Squatting neighborhood) 1.000 1.000 1.000 1.000 . High standing 1.932* 1.115 1.130 1.129 neighborhood (0.337) (0.185) (0.185) (0.185) . Average standing neighborhood 1.435 1.087 1.100 1.102 (0.269) (0.154) (0.155) (0.155) INTERACTION: INT 5 0.910* (0.050) INT 6 0.942 (0.042) NUMBER OF BEDROOMS: 0.888*** 0.886*** 0.887*** 0.886*** (0.045) (0.043) (0.043) (0.043) INDEX OF QUALITY OF HOUSING: Material o f construction .(Low: 1-7) 1.000 1.000 1.000 1.000 •Avrg: 7-10 1.118 1.073 1.082 1.074 (0.179) (0.173)0.913 (0.173) (0.175) .High: 10-12 0.920 (0.223) 0.931 0.925 (0.229) (0.223) (0.224) INDEX OF THE QUALITY OF IMMEDIATE ENVIRONMENT OF THE HOUSING UNIT: Propriety of the housing unit: .(Low: 1-7) 1.000 .Avrg:7-10 0.974 (0.176) .High: 10-12 0.702 (0.207) 295 Table 4.2.2. (continued) VARIABLE 17 18 19 20 WATER:* .(Not Connected to Piped Water) 1.000 1.000 1.000 .Piped Water 0.796* 0.790* 0.789* (0.137) (0.137) (0.137) TRASH:" .(D on't Use Garbage Can) 1.000 1.000 .Uses Garbage Can 0.778 0.780 (0.215) (0.215) ACCESS TO LATRINE: •(No)' 1.000 .Yes 1.069 (0.232) -2 LOG L 2041.824 2141.652 2140.223 2140.139 CHI-SQUARE 152.66**“ 162.14— 163.57*- 163.65*— DF 29 26 27 28 296 Table 4.2.4.: Odds Ratio and Standard Error for Logistic Regression of Child Mortality. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987 VARIABLES 1 2 3 4 INTERCEPT 0.0079"""* 0.159“ “ 0.021” 0.146“ “ (0.087) (0.459) (0.753) (0.465) EDUCATION 0.949""“ 0.953“ ” 0.906 0.953“ ” (0.014) (0.015) (0.127) (0.015) HOUSEHOLD 0.926 0.902 0.927 INCOME (0.050) (0.083) (0.050) INTERACTION: 1.005 Educ*Hous Inc. (0.014) — GENDER: . (Male) 1.000 . Female 1.145 (0.104) -2 LOG L 3101.305 3024.070 3023.914 3021.422 CHI-SQUARE 13.597 15.009 15.165 16.719 DF 1 2 3 3 ^egend: ***»p < = 0 .0 0 1 : *** A I I o © « » ■ o A I I ).5; * p < = 0 .1 0 297 Table 4.2.4. (continued) VARIABLES 5 6 7 8 INTERCEPT 0.180"” " 0 .1 6 5 "" 0.151— 0.157— (0.477) (0.483) (0.510) (0.514) EDUCATION 0.941"- 0.939— 0.951“ 0.950— (0.015) (0.015) (0.016) (0.016) HOUSEHOLD 0.931 0.938 0.931 0.929 INCOME (0.051) (0.051) (0.051) (0.051) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 1.149" 1.149 1.131 1.137 (0.104) (0.104) (0.105) (0.105) MOTHER AGE AT BIRTH OF CHILD: (in years) . ( < 2 0 ) 1.000 1.000 1.000 1.000 . 20-24 0.847 0.850 0.845 0.854 (0.151) (0.167) (0.167) (0.167) . 25-34 0.878 1.014 0.996 0.979 (0.144) (0.193) (0.193) (0.194) . 35 or more 0 .5 4 5 - 0.727 0.723 0.713 (0.208) (0.274) (0.274) (0.275) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 . 2-3 1.078 (0.166) 1.070 (0.166) 1.076 (0.167) . 4-6 0.912 (0.200) 0.904 (0.201) 0.918 (0.202) . 7 + 0.725 (0.239) 0.729 (0.240) 0.758 (0.245) BIRTH WEIGHT (in grams) .(< 2,500) .2,500-3,499 . 3,500 + . Unknown 1.000 1.075 (0.199) 0.962 (0.210) 1.604" (0.217) 1.000 1.074 (1.000) 0.948 (0.212) 1.561" (0.218) LENGHT OF PREGNANCY: . (9 months) . < 9 months 1.000 1.112 (0.204) -2 LOG L 3002.690 2998.680 2966.989 2924.700 CHI-SQUARE 27.194 31.205 43.269 42.517 DF 6 9 12 13 298 Table 4.2.4. (continued) VARIABLES 9 10 11 12 INTERCEPT 0.250“ * 0.227“ * 0.223“ * 0.214*“ (0.537) (0.546) (0.546) (0.548) EDUCATION 0.953“ * 0.955*“ 0.967* 0.950“ * (0.016) (0.016) (0.019) (0.016) HOUSEHOLD 0.924 0.922 0.918 0.926 INCOME (0.051) (0.052) (0.052) (0.052) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 1.128 1.128 1.119 1.121 (0.106) (0.106) (0.106) (0.106) MOTHER AGE AT BIRTH OF CHILD: (in years) . ( < 2 0 ) 1.000 1.000 1.000 1.000 . 20-24 0.856 0.851 0.851 0.848 (0.168) (0.168) (0.168) (0.168) . 25-34 0.986 0.963 0.970 0.937 (0.196) (0.195) (0.195) (0.196) . 35 or more 0..736 0.721 0.726 0.697 (0.276) (0.276) (0.277) (0.277) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 1.000 . 2-3 1.063 1.094 1.083 1.104 (0.168) (0.196) (0.169) (0.169) . 4-6 0.905 0.964 0.951 0.972 (0.202) (0.204) (0.205) (0.204) . 7 + 0.756 0.811 0.806 0.815 (0.242) (0.244) (0.244) (0.244) BIRTH WEIGHT (in grams) .(< 2,500) 1.000 1.000 1.000 1.000 .2,500-3,499 1.068 1.065 1.063 1.084 (0.200) (0.200) (0.200) (0.200) . 3,500 + 0.941 0.927 0.931 0.944 (0.212) (0.213) (0.213) (0.213) . Unknown 1.200 1.210 1.206 1.224 (0.238) (0.238) (0.239) (0.239) LENGHT OF PREGNANCY: . (9 months) 1.000 1.000 1.000 1.000 . < 9 months 1.089 1.063 1.053 1.058 (0.205) (0.205) (0.206) (0.206) 299 Table 4.2.4. (continued) VARIABLES 9 10 11 12 PLACE OF DELIVERY: .(At Home) 1.000 1.000 1.000 1.000 .Maternity 0 .6 3 4 - 0 .6 3 5 - 0 .6 3 8 - 0.639— (0.158) (0.158) (0.158) (0.159) MARITAL STATUS: . (Married Monog.) 1.000 1.000 1.000 . Single 1.354 1.819 1.007 (0.243) (0.382) (0.300) . Married 1.285 1.715*" 1.258 Poly gam. (0.156) (0.234) (0.156) . Others'"’ 1.390" 1.282 1.228 (0.175) (0.328) (0.185) INTERA CTION: (Edu*MMon) 1.000 Edu*Single 0.943 (0.061) Edu*MarPol 0.934 (0.043) Edu*Others 1.013 (0.049) OCCUPATIONAL STATUS: .(Housewife) 1.000 . Student 1.521 (0.417) . Working 1.460“ (0.155) . Others"’ ’ 1.535 (0.266) -2 LOG L 2916.336 2910.2231356.634 2906.752 2902.825 CHI-SQUARE 50.521— 60.105— 64.032— DF 14 17 20 20 This includes: consensual unions, divorced, separated, widowers, etc. < w This includes: Never worked, inactives, etc. 300 Table 4.2.4. (continued) VARIABLES 13 14 15 16 INTERCEPT 0.242" 0.260" 0.242" 0.266" (0.562) (0.568) (0.566) (0.578) EDUCATION 0.957"* 0.942“* 0.959" 0.962" (0.016) (0.023) (0.017) (0.017) HOUSEHOLD 0.920 0.918 0.939 0.934 INCOME (0.053) (0.053) (0.054) (0.054) GENDER: . (Male) 1.000 1.000 1.000 1.000 . Female 1.121 1.120 1.105 1.120 (0.106) (0.106) (0.107) (0.107) MOTHER AGE AT BIRTH OF CHILD: (in years) . (Less than 20) 1.000 1.000 1.000 1.000 . 20-24 0.871 0.854 0.875 0.882 (0.169) (0.170) (0.170) (0.171) . 25-34 0.981 0.963 0.981 0.995 (0.197) (0.197) (0.198) (0.199) . 35 + 0.723 0.711 0.725 0.734 (0.278) (0.279) (0.281) (0.283) BIRTH ORDER: .(1st birth) 1.000 1.000 1.000 1.000 . 2-3 1.104 1.108 1.087 1.102 (0.170) (0.170) (0.171) (0.171) . 4-6 0.988 0.993 0.980 0.975 (0.205) (0.206) (0.206) (0.207) . 7 + 0.839 0.856 0.887 0.883 (0.245) (0.246) (0.246) (0.248) BIRTH WEIGHT: (in grams) .( < 2,500) 1.000 1.000 1.000 1.000 . 2,500-3,499 1.093 1.089 1.073 1.054 (0.200) (0.201) (0.201) (0.201) . 3,500+ 0.945 0.939 0.919 0.904 (0.213) (0.214) (0.214) (0.215) . Unknown 1.197 1.205 1.187 1.154 (0.240) (0.241) (0.241) (0.242) 301 Table 4.2.4. (continued) VARIABLES 13 14 15 16 LENGTH OF PREGNANCY: .(9 months) 1.000 1.000 1.000 1.000 . < 9 months 1.059 1.059 1.092 1.091 (0.206) (0.206) (0.207) (0.209) PLACE OF DELIVERY: .(At home) 1.000 1.000 1.000 1.000 . Maternity 0.684“ 0.698“ 0.686” 0.692“ (0.161) (0.162) (0.162) (0.163) MARITAL STATUS .(Married Mono) 1.000 1.000 1.000 1.000 . Single 1.048 1.040 1.056 1.073 (0.301) (0.301) (0.302) (0.303) . Married Polygamously 1.209 1.185 1.203 1.185 (0.157) (0.158) (0.161) (0.163) . Other'" 1.254 1.252 1.257 1.244 (0.186) (0.186) (0.186) (0.187) OCCUPATIONAL STATUS: .(Housewife) 1.000 1.000 1.000 1.000 . Student 1.524 1.534 1.595 1.621 (0.417) (0.417) (0.418) (0.419) . Working 1.478“ 1.477“ 1.519*” 1.517*" (0.156) (0.157) (0.157) (0.158) . Others'" 1.449 1.450 1.547 1.535 (0.266) (0.266) (0.268) (0.269) ACCESS TO MODERN FACILITY .(Squatting neighborhood) 1.000 1.000 1.000 1.000 •High stand, neighborhood 0.528“ “ 0.396” * 0.553**” 0.558**” (0.158) (0.313) (0.159) (0.173) Average standing neig 0.987 0.738 0.852 0.862 (0.817) (0.200) (0.126) (0.128) This includes: consensual union divorced, separated , widowed, etc. ; lw This includes: Never worked, inactives, etc. 302 Table 4.2.4. (continued) VARIABLES 13 14 15 16 INTERACTION .(Edu*Squat) . Edu*High stand neigh .Edu*Average stand neigh 1.000 1.053 (0.045) 1.024 (0.032) NUMBER OF BEDROOMS 0.934* (0.036) 0.934* (0.036) INDEX OF QUALITY OF THE HOUSING UNIT: material of construction .(Low: 1-7) .Avrg: 7-10 .High: 10-12 1.000 0.915 (0.139) 0.893 (0.180) -2 LOG L CHI-SQUARE DF 2885.130 81.707“ " 22 2873.561 81.914— 24 2853.750 83.072— 23 2809.386 83.20— 25 303 Table 4.2.4. (continued) VARIABLE 17 18 19 20 INTERCEPT 0.201*" 0.267“ “ 0.267“ 0.247" (0.594) (0.578) (0.578) (0.603) EDUCATION 0.961“ 0.962“ 0.962“ 0.9 6 1 " (0.017) (0.017) (0.017) (0.017) HOUSEHOLD 0.944 0.934 0.934 0.934 INCOME (0.054) (0.054) (0.054) (0.054) GENDER . (Male) 1.000 1.000 1.000 1.000 . Female 1.118 1.123 1.123 1.124 (0.107) (0.107) (0.107) (0.107) MOTHER’S AGE AT BIRTH OF CHILD . ( < 20) . 20-24 1.000 1.000 1.000 1.000 0.886 0.883 0.883 0.884 . 25-34 (0.171) (0.171) (0.171) (0.171) 0.993 0.998 0.998 0.998 . 35 or more (0.199) (0.199) (0.199) (0.199) 0.729 0.739 0.739 0.738 (0.284) (0.283) (0.283) (0.284) BIRTH ORDER: .(1st Birth) 1.000 1.000 1.000 1.000 ,2nd-3rd birth 1.095 1.102 1.102 1.101 ,4th-6th birth (0.172) (0.171) (0.171) (0.171) ,7th and above 0.961 0.974 0.975 0.973 (0.208) (0.207) (0.207) (0.207) 0.870 0.879 0.879 0.876 (0.248) (0.247) (0.248) (0.248) BIRTH WEIGHT (in grams) .(<2500)* 1.000 1.000 1.000 1.000 .2500-3499 1.063 1.054 1.054 1.056 (0.202) (0.201) (0.201) (0.202) .3500 or more 0.920 0.899 0.900 0.902 .unknown (0.215) (0.215) (0.215) (0.215) 1.194 1.153 1.153 1.161 (0.245) (0.242) (0.242) (0.243) 304 TABLE 4.2.4. (continued) VARIABLE 17 18 19 20 LENGTH OF PREGNANCY: (9 months) 1.000 1.000 1.000 1.000 . < 9 months 1.110 1.089 1.089 1.092 (0.210) (0.209) (0.209) (0.210) PLACE OF DELIVERY: .(home) 1.000 1.000 1.000 1.000 .(health institution) 0.686" 0.691" 0.691" 0.694" (0.164) (0.163) (0.163) (0.164) MARITAL STATUS OF MOTHER: .(Married monogamously) 1.000 1.000 1.000 1.000 .Single 1.053 1.077 1.077 1.082 (0.303) (0.303) (0.303) (0.304) .Married polygamously 1.177 1.180 1.180 1.181 (0.163) (0.163) (0.163) (0.163) . Others1 ’1 1.221 1.250 1.249 1.249 (0.187) (0.187) (0.187) (0.187) OCCUPATIONAL STATUS OF MOTHER: (Housewife) 1.000 1.000 1.000 1.000 .Student 1.641 1.621 1.621 1.616 (0.420) (0.419) (0.420) (0.420) .Working 1 .5 2 6 - 1.519— 1.519— 1.512— (0.158) (0.158) (0.158) (0.159) .Others( b ) 1.548 1.538 1.538 1.532 (0.270) (0.269) (0.269) (0.270) 305 TABLE 4.2.4. (continued) VARIABLE 17 18 19 20 ACCESS TO MODERN FACILITY: . (Squatting neighborhood) 1.000 1.000 1.000 1.000 . High standing 0.564"” "" 0.554'*’’ 0.554"""" 0.553"""" neighborhood (0.175) (0.174) (0.174) (0.174) . Average standing 0.850 0.859 0.859 0.860 neighborhood (0.128) (0.128) (0.129) (0.129) NUMBER OF 0.937" 0.935" 0.935" 0.935* BEDROOMS: (0.036) (0.036) (0.036) (0.036) INDEX OF QUALITY OF HOUSING: Material o f construction .(Low: 1-7) 1.000 1.000 1.000 1.000 .Avrg: 7-10 0.901 0.907 0.907 0.899 (0.141) (0.142)0.881 (0.143) (0.144) .High:10-12 0.892 (0.186) 0.881 0.875 (0.183) (0.186) (0.187) INDEX OF THE QUALITY OF IMMEDIATE ENVIRONMENT OF THE HOUSING UNIT: Propriety of the housing unit: .(Low: 1-7) 1.000 .Avrg:7-10 1.345" (0.159) .High: 10-12 1.147 (0.182) 306 Table 4.2.4. (continued) VARIABLE 17 18 19 20 WATER:* .(Not Connected to Piped Water) 1.000 1.000 1.000 .Piped W ater 1.032 1.032 1.030 (0.116) (0.116) (0.116) TRASH:* .(D on't Use Garbage Can) 1.000 1.000 .Uses Garbage Can 1.004 1.006 (0.175) (0.175) ACCESS TO LATRINE: .(No)* 1.000 .Yes 1.091 (0.201) -2 LOG L 2804.948 2811.021 2811.020 2810.829 CHI-SQUARE 87.4” ” 83.720” ” 83.721“ “ 83.912“ ” DF 27 26 27 28 307 Table 5.1.1.: Parameter Estimates and Standard Error (in Parenthesis) for Multinomial Logistic Regression of Birth Weight; Children Born in Birth Cohorts 1981- 1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES P1/P4 P2/P4 P3/P4 MODEL L.RATIO INTERCEPT 1.336’ '"** 1.041**** 1.431""’* 1 EDUCATION <°-08n*** -0.191 (0.083) 0.015 (0.079) 0.025 76.2** (0.014) (0.012) (0.012) INTERCEPT 0.599 -1.156*** 0.741** 2 EDUCATION -0.194 (0.398) 0.006 (0.378) 0.023 H_INC0ME (0.015) 0.082 (0.048) (0.013) 0.242 (0.043) (0.012) 0.077 (0.041) " i t 4 e 4 e 4 c 13707.3 INTERCEPT 0.475 -1.574** -0.1922 3 EDUCATION (0.726) -0.222 (0.728) 0.082 (0.694) 0.189 H INCOME (0.126) 0.096 (0.112) ^ k 0.288 (0.106) 0.180 INTERACTION: (0.080) (0.079) (0.076) Educ*H_income 0.003 (0.014) -0.008 (0.012) -0.018 (0.012) 4 c “ i t 13701.4 INTERCEPT 0.624 -1.114*’ ”' 0.770 4 EDUCATION -0.195 (0.399) 0.006 (0.340) 0.023 H INCOME (0.015) 0.080 (0.013) ^ x 4 r 0.238 (0.012) 0.074 GENDER: (0.048) (0.043) (0.041) . (Male) . Female 0.000 0.094 0.000 • J e 4 t 0.122 0.000 0.043 (0.049) (0.044) (0.042) 15110.9 Legend: PI =Birth weight unknown; P2 =Birth weight 3,500 grams or more; P3 =Birth weight 2,500-3,499 grams; P4 =Less than 2,500 grams. **** p <=o.001; *** p <=0.01; ** p <=0.05; * p <=0.10 L.RATIO =Likelihood Ratio; H INCOME =Household Income 308 Table 5.1.1. (continued) INTERCEPT 0.794’ ' 1 -1.077 0.914 5 (0.467)(0.421) (0.401) EDUCATION -0.195 0.006 0.024 (0.015) (0.013)(0.012) H_INC0ME 0.079 0.238--- 0.073 (0.048) (0.043) (0.041) GENDER: . (Male) 0.000 °-000*** 0.000 . Female 0.091 0.122 0.043 (0.050) (0.044) (0.042) CHILD YEAR OF BIRTH: . 1981 -0.098 -0.009 -0.041 5 (0.079) (0.070) (0.067) . 1982 -0.058 -0.027 -0.052 (0.076) (0.068) (0.065) . 1983 -0.077 0.002 -0.044 (0.075) (0.067) (0.064) . 1984 -0.039 -0.019 -0.087 (0.077) (0.067) (0.064) 16960.8 . (1985) 0.000 0.000 0.000 INTERCEPT 1.042** -0.675 1.154 6 (0.483) (0.435) -0.199 0.006 (0.414) EDUCATION 0.020 (0.015) 0.226 (0.013) H_INCOME 0.073 0.065 (0.049) (0.044) (0.042) GENDER: . (Male) . Female 0.000 0.095* °-000*** 0.127 0.000 0.049 (0.050) (0.045) (0.043) CHILD YEAR OF BIRTH: . 1981 -0.124 -0.042 -0.057 (0.080) (0.071) (0.068) . 1982 -0.072 -0.042 -0.064 (0.078) (0.068) (0.065) . 1983 -0.081 0.002 -0.041 (0.076) (0.067) (0.064) . 1984 -0.051 -0.027 -0.090 (0.077) (0.068) (0.065) . (1985) 0.000 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000 0.000 -0.190*** 0.000 -0.136 * . 20-24 -0.157 . 25-34 -0.199 -0.339 -0.199 (0.071) (0.062) (0.059) . 35+ -0.010 -0.117 -0.017 16845.3 309 Table 5.1.1. (continued) INTERCEPT 0.932* -0.514 1.161 (0.516) (0.460) (0.437) EDUCATION -0.196 0.008 0.022 (0.015) (0.013) H_INCOME 0.032 0.208 0.045 (0.049) (0.044) (0.042) GENDER: . (Hale) 0.000 0-000*** 0.127 0.000 7 . Female 0.098* 0.051 (0.050) (0.045) (0.043) CHILD YEAR OF BIRTH: . 1981 -0.117 -0.041 -0.052 (0.080) (0.071) (0.068) . 1982 -0.058 -0.036 -0.056 (0.078) (0.068) (0.065) . 1983 -0.072 0.006 -0.034 (0.076) (0.067) (0.064) . 1984 -0.049 -0.027 -0.089 (0.078) (0.068) (0.065) . (1985) 0.000 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000 °-000*** -0.168 0.000 -0.114** . 20-24 -0.117 . 25-34 (0.073) -0.133 -0.307 -0.166 (0.072) (0.063) (0.060) . 35+ 0.061 -0.085 0.018 (0.089) (0.081) (0.076) MARITAL STATUS1 OF MOTHER: .(Married Mono) 0.000 0.000_ 0.234** 0.000_ 0.196** . Single 0.459 . Married Pol -0.314 (0.097) -0.253 (°.°871 -0.166 . Others (0.087) (0.082) (0.080) 16812.3 0.229 -0.074 0.087 (0.093) (0.075) (0.072) INTERCEPT 0.851 -0.551 1.019 8 <°-56§L* -0.194 (0.515) (0.491) EDUCATION 0.003 0.018 (0.018) <°*01£L* (0.015) H_INCOME 0.032 0.209 0.046 (0.050) (0.044) (0.042) GENDER: . (Male) 0.000 0'000*** 0.125 0.000 . Female 0.098 0.050 (0.050) (0.045) (0.043) 310 Table 5.1.1. (continued) CHILD YEAR OF . 1981 . 1982 . 1983 BIRTH: - 0.121 (0.080) -0.062 (0.078) -0.076 (0.076) -0.041 (0.071) -0.037 (0.068) 0.005 (0.067) -0.054 (0.068) -0.058 (0.065) -0.037 (0.064) -0.089 (0.065) 0.000 0.000 -0.116** <°-059L * -0.168 (0.060) 0.024 (0.076) 0.000 0.337** (0.161) -0.355 (0.153) 0.299 (0.142) 0.000 0.047 (0.046) -0.067 (0.044) 0.071 (0.042) 1.066 9 <°-469l* 0.032 (0.013) 0.041 (0.043) CHILD YEAR OF BIRTH: . 1984 -0.050 -0.028 (0.078) (0.068) . (1985) 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000^ 0.000 . 20-24 -0.121 -0.169*** (0.073) (0.064) . 25-34 -0.137 -0.307 (0.072) (0.063) . 35+ 0.065 -0.082 (0.089) (0.081) MARITAL STATUS OF MOTHER: .(Married Mono) 0.000 0.000 . Single 0.612*** 0.281 . Married Pol -0.551 -0.320 (0.156) (0.157) . Others<a' 0.406 0.003 (0.158) (0.143) INTERACTION: (Edu*Married_M.) 0.000 0.000 . Educ*Single 0.066 0.017 (0.063) (0.050) . Edu*Married_P -0.112 -0.024 (0.050) (0.045) .Edu*Others 0.067 0.027 (0.053) (0.043) INTERCEPT 0.917 -0.604 (°,572i**(°*494) EDUCATION -0.186 0.018 (0.016) (°-°lU** H_INCOME 0.026 0.203 (0.050) (0.045) GENDER: . (Male) °*000** . Female 0.098 (0.050) 0.000 0.000 0.126*** 0.050 (0.045) (0.043) 16794.9 311 Table 5.1.1. (continued) CHILD YEAR OF BIRTH: . 1981 -0.124 -0.047 -0.058 (0.080) (0.072) (0.068) . 1982 -0.058 -0.038 -0.058 (0.078) (0.069) (0.065) . 1983 -0.075 0.005 -0.036 (0.076) (0.068) (0.064) CHILD YEAR OF BIRTH: 9 . 1984 -0.042 -0.025 -0.086 (0.078) (0.068) (0.065) . (1985) 0.000 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000 . 20-24 -0.131* 0.000 -0.182*** 0.000_ -0.129** . 25-34 (0.074) (0-06U** (0.060) -0.162 -0.336 -0.197 (0.073) (0.064) (0.061) . 35+ 0.019 -0.128 0.026 (0.090) (0.082) (0.077) MARITAL STATUS OF MOTHER: .(Married Mono) 0.000 . Single 0.635 0.000 0.274** 0.000 0.233** . Married Pol (0.120) (o. m ) -0.323 -0.263 -0.177 . Others (0.087) (0.082) (0.080) 0.259 -0.087 0.072 (0.097) (0.078) (0.075) OCCUPATION STATUS OF MOTHER: .(Housewife) 0.000 0.000 0.000 . Student 0.099 0.096 0.092 . Working (0.246) 0.239 0.245 0.255 (0.078) (0.065) (0.062) . Others -0.418 -0.173 -0.167 (0.148) (0.133) (0.127) 16782.9 CNTERCEPT 0.670 1.227** 0.599 10 (0.514) (0.486) EDUCATION -0.188 0.013 0.029* (0.016) (0-01U** 0.215 (0.014) H_INCOME 0.044 0.058 (0.051) (0.046) (0.044) GENDER: . (Male) 0.000 0.000 0.000 . Female 0.080 0.102 0.028 (0.052) (0.046) (0.044) 312 Table 5.1.1. (continued) CHILD YEAR OF BIRTH: . 1981 -0.139 -0.044 -0.064 (0.081) (0.073) (0.070) . 1982 -0.072 -0.040 -0.069 (0.079) (0.070) (0.066) . 1983 -0.090 0.011 -0.045 (0.078) (0.069) (0.066) CHILD YEAR OF BIRTH: 10 . 1984 -0.064 -0.041 -0.110 (0.079) (0.070) (0.067) . (1985) 0.000 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000 0.000 0.000 . 20-24 -0.088 -0.152 -0.092 . 25-34 (0.075) -0.137 -0.295 -0.163 (0.074) (0.067) (0.063) . 35+ 0.048 -0.075 0.012 (0.092) (0.085) (0.080) MARITAL STATUS OF MOTHER .(Married Mono) 0.000 0.000 0.000 . Single 0.634 0.222 0.168 . Married Pol (°-15§L* (0.126) (0.116) -0.294 -0.259 -0.174 . Others (0.089) (0.083) (0.081) 0.231 -0.141 0.010 (0.099) (0.081) (0.078) OCCUPATION STATUS OF MOTHER: .(Housewife) 0.000 0.000 0.000 . Student 0.081 0.045 0.067 . Working <°-24n** 0.252 (0-177U** 0.222 (0-162L** 0.234 (0.080) (0.067) (0.063) . Others -0.464 -0.175 -0.157 (0.155) (0.141) (0.134) LENGTH OF PREGNANCY: . (9 Months) . < 9 Months 0.000 ** 0.201 0.782 0.000 • I p J f 0.563 (0.078) (0.083) (0.068) 16199.3 INTERCEPT 1.152* -0.828 0.846 11 (0.599) -0.179 (0.527) (0.499) EDUCATION 0.012 0.028 (0.016) (0.014) H_INCOME 0.030 0.178 0.039 (0.052) (0.047) (0.045) 313 Table 5.1.1. (continued) GENDER: . (Male) 0.000 0.000 0.000 . Female 0.091 0.106 0.031 (0.052) (0.046) (0.044) CHILD YEAR OF BIRTH: . 1981 -0.143 -0.046 -0.064 (0.082) (0.073) (0.070) . 1982 -0.071 -0.040 -0.069 (0.080) (0.070) (0.067) CHILD YEAR OF BIRTH: 11 . 1983 -0.093 0.009 -0.045 (0.078) (0.069) (0.066) . 1984 -0.053 -0.039 -0.109 (0.080) (0.070) (0.067) . (1985) 0.000 0.000 0.000 AGE OF MOTHER AT BIRTH: . (< 20 Years) 0.000 0.000_ -0.160** 0.000 . 20-24 -0.114 -0.093 (0.076) <0-067i***<°*062>** -0.302 -0.160 . 25-34 -0.191 (0.076) (0.067) (0.063) . 35+ -0.011 -0.081 0.015 (0.093) (0.085) (0.080) MARITAL STATUS OF MOTHER: .(Married Mono) 0.000 0.000 0.000 . Single 0.557 0.192 0.159 . Married Pol (°*12£U (O.ne) -0.262 -0.260 -0.180 . Others (0.089) (0.084) (0.082) 0.161 -0.172 -0.003 (0.101) (0.082) (0.078) OCCUPATION STATUS OF MOTHER: .(Housewife) 0.000 0.000 0.000 . Student 0.064 0.035 0.062 . Working (0.252) 0.182 <°*177U* <°-162i*** 0.194 0.217 (0.067) (0.064) . Others -0.431 -0.184 -0.173 (0.156) (0.142) (0.136) LENGTH OF PREGNANCY: . (9 Months) . < 9 Months 0.000 ^ 0.191 °-000**** 0-000**** 0.775 0.559 (0.079) (0.083) (0.068) TYPE OF NEIGHBORHOODS: .(Squatting N.) 0.000 0.000 0.000 . High Stand. 0.479 0.202 0.099* (0.061) (0.056) (0.053) . Average Stand. 0.383 0.016 -0.045 (0.072) (0.062) (0.060) 314 Table 5.1.2.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Premature Birth. Cohorts 1981-1985. Urban Areas of Zaire, 13 cities. Fonames/Unicef Survey, 1987. VARIABLE 1 2 3 INTERCEPT 0.084"*” 0.210” 0.280“ " (0.083) (0.715) (0.439) EDUCATION 0.966“ " 0.935 0.943” ” (0.013) (0.119) (0.014) HOUSEHOLD INCOME 0.905 0.932 (0.079) (0.047) INTERACTION 1.004 Educ*H_Income (0.012) AGE OF MOTHER AT BIRTH: .(Less than 20) . 20-24 1.000 0.834 . 25-34 (0.126) 0.492“ ” . 35 and above (0.132) 0.307” ” (0.209) -2 LOG L 3508.597 3396.308 3310.089 CHI-SQUARE 6.775"” 10.598” 65.666**** DF 1 3 -egend: • • • • p < =0.001: *•* p < =0.01; '** p < =0.05; * p < = =0.10. 315 Table S. 1.2. (continued) VARIABLE 4 5 6 INTERCEPT 0.283“ “ 0.234“ " 0.232“ " (0.442) (0.452) (0.452) EDUCATION 0 .9 3 6 "" 0 .9 3 6 "" 0.939“ “ (0.014) (0.014) (0.017) HOUSEHOLD INCOME 0.949 0.957 0.956 (0.047) (0.048) (0.048) AGE OF MOTHER AT BIRTH: . (Less than 20) . 20-24 1.000 1.000 1.000 1.058 1.063 1.061 . 25-35 (0.139) (0.140) (0.140) 0.916 0.914 0.916 . 35 and over (0.176) (0.177) (0.177) 0.732 0.737 0.734 (0.275) (0.275) (0.276) BIRTH ORDER: . (1 year) 1.000 1.000 1.000 . 2-3 years 0.705“ “ 0.738" 0.739" (0.135) (0.137) (0.137) . 4-6 years 0.451"“ 0 .4 8 5 "" 0 .4 8 3 "" (0.177) (0.180) (0.180) . 7 yrs and over 0.329"“ 0 .3 5 5 "" 0 .3 5 4 "" (0.233) (0.236) (0.236) MARITAL STATUS OF MOTHER: . (Married Mono) 1.000 1.000 . Single 1.632" 1.829“ (0.191) (0.319) . Married Poly 1.101 1.346 (0.156) (0.242) . O thers 1.271 1.006 (0.156) (0.304) INTERACTION: . (Educ*Married Mono) . Single 1.000 0.979 . Married Polygamously (0.048) 0.955 . Others* (0.043) 1.041 (0.045) -2 LOG L 3282.802 3275.259 3272.755 CHI-SQUARE 92.953"" 100.496"" 103.000"" DF 8 11 14 Lecend: ***• d < =0.001: o < = 3.6); *• p < =0.05; * p < =0.10; (*) This includes: consensual unions, divorced, separated, widows, etc. 316 Table 5.1.2. (continued) VARIABLE 7 8 9 INTERCEPT 0.228"” 0.195“ “ 0.225"“ (0.453) (0.473) (0.479) EDUCATION 0.930“ ” 0.932” ” 0.902” ” (0.014) (0.015) (0.023) HOUSEHOLD INCOME 0.959 0.972 0.974 (0.048) (0.049) (0.049) AGE OF MOTHER AT BIRTH: . (Less than 20) . 20-24 1.000 1.000 1.000 1.062 1.057 1.040 . 25-35 (0.141) (0.141) (0.142) 0.899 0.891 0.878 . 35 and over (0.178) (0.179) (0.179) 0.715 0.706 0.697 (0.277) (0.277) (0.278) BIRTH ORDER: . (1 year) 1.000 1.000 1.000 . 2-3 years 0.758” 0.763” 0.771" (0.138) (0.138) (0.138) . 4-6 years 0.495"“ 0.503” ” 0.507” ” (0.180) (0.181) (0.181) . 7 yrs and over 0.360” ” 0.370”” 0.374” ” (0.236) (0.237) (0.237) MARITAL STATUS OF MOTHER: . (Married Mono) 1.000 1.000 1.000 . Single 1.645” 1.605' 1.572" (0.246) (0.248) (0.248) . Married Poly 1.089 1.086 1.072 (0.156) (0.157) (0.157) . Others* 1.265 1.228 1.222 (0.163) (0.164) (0.164) (*) This includes: consensual unions, divorced, separated, widows, etc. 317 Table S. 1.2. (continued) OCCUPATIONAL STATUS OF MOTHER: . (Housewife) . Student . Working . Others'1 1.000 1.726* (0.316) 1.272 (0.152) 0.711 (0.274) 1.000 1.723* (0.316) 1.244 (0.153) 0.706 (0.275) 1.000 1.683* (0.316) 1.248 (0.154) 0.707 (0.274) ACCESS TO MODERN FACILITY: . (Squatting Neighborhood) . High standing Neighborhood 1.000 1.000 . Average Standing Neighborhood 0.907 0.767 (0.136) (0.252) 1.136 0.845 (0.116) (0.190) INTERACTION . Educ*High_Standing 1.037 Neighborhoods (0.038) . Educ*Aver_Standing 1.063“ Neighborhoods (0.031) -2 LOG L 3266.462 3263.520 3375.755 CHI-SQUARE 109.293““ 112.235**“ 116.182“ “ DF 14 16 18 -egend: O This includes: Never worked, inactives, etc. 318 Table 5.2.1. Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Maternity Hospital for Delivery. Women who Gave Birth between 1981 and 1985. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLE 1 2 3 IN TER C EPT 2.756""’ 3.174*" 4 .7 0 5 " " (0.052) (0.445) (0.304) EDUCA TIO N 1.161— 1.302"* 1.181**" (0.009) (0.084) (0.010) HO USEHO LD INCOM E 0.984 0 .9 3 0 " (0.048) (0.032) INTERACTION: . Educ*H income 0.988 (0.009) AGE O F M O TH ER A T BIRTH (in years) . (Less than 20) 1.000 . 20-24 0.910 (0.102) . 25-35 1.072 (0.099) . 35 and over 1.533**" (0.129) -2 LOG L 6044.149 5852.992 5770.057 CHI-SQUARE 2 60.341"“ 260.729**“ 281.773**" DF 1 3 5 .egend: ♦*** p < = 0 .0 o i: *** p < = B'.Oi; ** p < = 6 .0 5 ; * p < = 6 .1 0 ; H _incom e= Household Income; 319 Table 5.2.1. (continued) VARIABLE 4 5 6 IN TERCEPT 5.717**“ 4.360“ “ 4.4444**“ (0.309) (0.315) (0.315) EDUCATION 1.181“ “ 1.179**“ 1.173**“ (0.010) (0.010) (0.012) H O USEHO LD INCOM E 0.925“ 0.949 0.949 (0.047) (0.033) (0.033) AGE O F M O TH ER A T BIRTH (in years) . (Less than 20) 1.000 1.000 1.000 . 20-24 1.056 1.074 1.071 (0.113) (0.113) (0.113) . 25-35 1.174 1.206 1.197 (0.132) (0.132) (0.133) . 35 and over 1.397* 1.436“ 1.431“ (0.174) (0.174) (0.175) BIRTH ORDER: . (1 year) 1.000 1.000 1.000 . 2-3 years 0.701*“ 0.721*“ 0.720“ * (0.116) (0.117) (0.117) . 4-6 years 0.711“ 0.752“ 0.754“ (0.137) (0.138) (0.138) . 7 yrs and over 1.012 1.066 1.070 (0.161) (0.162) (0.163) M A RITA L STATUS O F M OTHER: . (M arried M ono) 1.000 1.000 . Single 1.362 1.429 (0.188) (0.274) . M arried Poly 0.711**“ 0.590**“ (0.096) (0.143) . Others* 1.420“ 1.602“ (0.139) (0.232) INTERACTION: . (Educ*M arried M ono) . Educ*Single 1.000 0.988 . Educ*M arried Polygamously (0.048) . Others* 1.051* (0.028) 0.975 (0.045) -2 LOG L 5749.790 5725.618 5721.512 CHI-SQUARE 302.041**“ 326.213**“ 330.319**“ D F 8 11 14 L eeend: (*) This includes: consensus unions, divorced, separated, widows; **** p < = 0 .0 0 1 ; **» p < = o .0 1 ; ** 320 Table 5.2.1. (continued) VARIABLE 7 8 9 IN TER C EPT 4.536**” 1.429 1.464 (0.316) (0.328) (0.330) EDUCATION 1.171“ “ 1.156**“ 1.153— (0.011) (0.012) (0.015) HO USEHO LD INCOM E 0.950 1.033 1.032 (0.033) (0.034) (0.034) AGE O F M O TH ER A T BIRTH (in years) . (Less than 20) 1.000 1.000 1.000 . 20-24 1.062 0.985 0.981 (0.114) (0.115) (0.116) . 25-35 1.192 1.020 1.018 (0.133) (0.135) (0.135) . 35 and over 1.420“ 1.218 1.215 (0.175) (0.178) (0.178) BIRTH ORDER: . (1st birth) 1.000 1.000 1.000 . 2-3 0.716*“ 0.729“ * 0.731 (0.117) (0.119) (0.119) . 4-6 0.750“ 0.771* 0.771 (0.138) (0.140) (0.140) . 7 and over 1.067 1.091 1.090 (0.163) (0.165) (0.165) M ARITA L STATUS O F M OTHER: . (M arried Mono) 1.000 1.000 1.000 . Single 1.669“ 1.343 1.334 (0.228) (0.229) (0.229) . M arried Poly 0.588**“ 0.649“ * 0.648 (0.141) (0.146) (0.146) . Others* 1.509*” 1.257 1.255 (0.145) (0.148) (0.148) INTERACTION . Educ* M arried 1.052* 1.049* 1.049* Polygamously (0.028) (0.029) (0.029) (*) This includes: consensual unions, divorced, separated, widows; 321 Table 5.2.1. (continued) OCCU PA TIO N A L STATUS O F M O TH ER: . (Housewife) . Student . W orking . O thersb 1.000 0.609 (0.326) 0.964 (0.115) 0.788 (0.200) 1.000 0.572* (0.333) 0.821* (0.118) 0.925 (0.203) 1.000 0.558* (0.333) 0.828 (0.118) 0.925 (0.203) ACCESS TO M ODERN FACILITY: . (Squatting N eighborhood) . High Slanging N eighborhood 1.000 1.000 . A verage Standing Neighborhood 2.819**“ 3.117**“ (0.097) (0.167) 3.039**“ 2.734**“ (0.087) (0.130) INTERACTION . Educ*High_Standing 0.981 N eighborhood (0.027) . Educ*A verage_Standing 1.027 N eighborhood (0.024) -2 LOG L 5719.030 5489.979 5487.714 CHI-SQUARE 332.801— 561.852**“ 564.117**“ D F 15 17 19 Legend: This includes N ever w orked, inactives, etc. 322 Table 5.3.2.: Parameter Estimates and Standard Error (in Parenthesis) for Multinomial Logistic Regression of Nutritional Status of Children 12-59 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES P1/P3 P2/P3 MODEL L.RATIO INTERCEPT EDUCATION -2.884**** (0.166) -0.018 -1. /**** <°-08§L* -0.073 1 46.34* (0.024) (0.014) INTERCEPT -3.487’ '’ '’ '’ ' (0.750) EDUCATION -0.019 (0.025) HOUSEHOLD INCOME 0.067 (0.080) --L. (0.432) 4 g J f J p -0.074 (0.014) -0.009 (0.047) 2 2 3552.9 INTERCEPT -2.973’ '’ ' -0.681 3 (1.454) (0.771) EDUCATION -0.107 -0.179 (0.215) (0.122) HOUSEHOLD INCOME 0.0112 -0.070 (0.157) (0.084) EDUC*INCOME 0.009 0.011 (0.023) (0.013) 3 3551.9 INTERCEPT -3.438™** -1.248*** 4 (0.783) (0-452U** EDUCATION -0.002 -0.064 (0.025) (0.014) HOUSEHOLD INCOME 0.067 -0.009 (0.081) (0.047) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.098 0.139 (0.151) (0.086) 3,500 + 0.102 0.049 (0.158) (0.089) UNKNOWN -0.271 -0.111 (0.171) (0.101) 3744.6 Legend: . P1/P3 means "Severe Malnutrition" versus "Good" . P2/P3 means "Moderate Malnutrition" vs. "Good" . L.RATI0= Likelihood Ratio **** p <= 0.001 ; *** p <= 0.01 ; ** p <= 0.05 ; * p <= 0.10; 323 Table 5.3.2. (continued) INTERCEPT J f J f -3.347 4 > 4 -0.919 5 (0.808) (0.467) _ -0.061 (0.015) EDUCATION -0.008 (0.026) HOUSEHOLD INCOME 0.071 (0.082) -0.014 (0.048) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.115 (0.153) 0.079 (0.091) 3,500 + 0.114 (0.162) -0.027 (0.094) UNKNOWN -0.246 (0.172) -0.157 (0.105) LENGTH PREGNANCY: .(9 MONTHS) . < 9 MONTHS 0.000 -0.143 0.000 „ -0.283 3638.5 INTERCEPT 4 ? 4 f -3.407 -0.912* 6 (0.807) (0.468) t -0.061 (0.015) EDUCATION -0.007 (0.026) HOUSEHOLD INCOME 0.076 (0.082) -0.015 (0.048) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.117 (0.153) 0.079 (0.091) 3,500 + 0.103 (0.162) -0.026 (0.094) UNKNOWN -0.247 (0.173) -0.157 (0.105) LENGTH PREGNANCY: .(9 MONTHS) 0.000 0.000^ -0.283 . < 9 MONTHS -0.133 GENDER: . (MALE) 0.000 0.000 . FEMALE -0.145 -0.012 (0.087) (0.049) 3737.7 Legend: . P1/P3 means "Severe Malnutrition" versus "Good" . P2/P3 means "Moderate Malnutrition" vs. "Good" - . L.RATIO= Likelihood Ratio **** p <= o.OOl ; *** p <= 0.01 ; ** p <= 0.05 ; * p <= 0.10 324 Table 5.3.2. (continued) INTERCEPT -3.579 (0.813) -0.971** (0.470) 4f * 4 * « i » * i f 7 EDUCATION -0.015 (0.026) -0.065 (0.015) HOUSEHOLD INCOME 0.072 (0.083) -0.018 (0.049) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.112 (0.154) 0.076 (0.091) 3,500 + 0.103 (0.162) -0.026 (0.094) UNKNOWN -0.243 (0.175) -0.156 (0.105) LENGTH PREGNANCY: .(9 MONTHS) 0.000 0.000 . 4 e 4e -0.299 (0.090) . < 9 MONTHS -0.163 (0.166) GENDER: . (MALE) 0.000 0.000 . FEMALE -0.134 (0.088) -0.008 (0.050) AGE OF CHILD: . (12-17) 0.000 0.000 . 18-35 MONTHS 0.349**** " i t 4 t 0.188 . 36-59 MONTHS (0.066) ~ — — fcjdctc 0.677 0.355 (0.118) (0.069) 3805.6 INTERCEPT -3.713**** 4c 4c 4 c -1.285 8 EDUCATION (0.829) -0.016 (0.027) (0.481) 9c 4c 4c 4 c -0.076 (0.015) HOUSEHOLD INCOME 0.073 (0.083) -0.006 (0.049) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.104 (0.154) 0.076 (0.091) 3,500 + 0.091 (0.162) -0.035 (0.094) UNKNOWN -0.258 (0.175) -0.161 (0.106) 325 Table 5.3.2. (continued) LENGTH PREGNANCY: 8 .(9 MONTHS) 0.000 °-000*** -0.265 (0.091) . < 9 MONTHS -0.155 (0.168) GENDER: . (MALE) 0.000 0.000 . FEMALE -0.136 (0.088) -0.007 (0.050) AGE OF CHILD: . (12-17) 0.000 0.000 _ . 18-35 MONTHS 0.350**** 0.190 . 3 6-59 MONTHS (0*10H*** 0.679 (0.118) 0.363 (0.069) BIRTH ORDER: . (1st. Born) 0.000 0.000 . 2-3 0.082 (0.126) 0.092 (0.071) . 4-6 0.191 0.184 (0.132) (°-o7n** 0.232 . 7 + 0.035 (0.137) (0.082) 3807.3 INTERCEPT • ! * J f -3.645 4 > 4 f -1.262 9 (0.877) (0.508)t 9 EDUCATION -0.017 (0.027) “ i t & 4 t -0.076 (0.015) HOUSEHOLD INCOME 0.072 (0.085) 0.008 (0.050) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.098 (0.155) 0.069 (0.091) 3,500 + 0.078 (0.164) -0.039 (0.095) UNKNOWN -0.267 (0.176) -0.175 (0.106) LENGTH PREGNANCY: .(9 MONTHS) 0.000 0*000*** -0.266 (0.091) . < 9 MONTHS -0.155 (0.168) 326 Table 5.3.2. (continued) GENDER: 9 . (MALE) 0.000 0.000 9 . FEMALE -0.139 (0.088) -0.012 (0.050) AGE OF CHILD: . (12-17) . 18-35 MONTHS 0‘000**** 0.354 0.000 0.185 . 36-59 MONTHS 0.681 (0.118) 0.354 (0.069) BIRTH ORDER: . (1st. Born) 0.000 0.000 . 2-3 0.060 (0.131) 0.070 (0.074) . 4-6 0.174 (0.139) 0.153 (0.077) . 7 + 0.016 (0.145) 0.199 (0.086) MARITAL STATUS OF MOTHERS: . (Married Mono.) 0.000 0.000 . Single -0.172 (0.178) -0.111 (0.108) . Married Polyg. 0.045 0.072 . Others (0.140) (0.084) 0.168 -0.071 (0.174) (0.084) 3802.5 INTERCEPT -3.316**** -1.873*** 10 (0.922) <°-58n*** -0.074 (0.015) 10 EDUCATION -0.021 (0.027) HOUSEHOLD INCOME 0.071 (0.085) 0.011 (0.050) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.086 (0.155) 0.075 (0.091) 3,500 + 0.066 (0.164) -0.034 (0.095) UNKNOWN -0.271 (0.176) -0.174 (0.106) This includes: consensualunions,divorced, separated, widows 327 Table 5.3.2. (continued) LENGTH PREGNANCY: .(9 MONTHS) . < 9 MONTHS GENDER: . (MALE) . FEMALE AGE OF CHILD: . (12-17) . 18-35 MONTHS . 36-59 MONTHS BIRTH ORDER: . (1st. Born) . 2-3 . 4-6 . 7 + MARITAL STATUS OF . (Married Mono.) . Single . Married Polyg. . Others OCCUPATION STATUS . (Housewife) . Student Working Others 0.000 -0.149 (0.168) 0.000 -0.140 (0.088) 0.000 0.348**** (0-102L** 0.673 (0.119) 0.000 0.033 (0.134) 0.151 (0.139) -0.007 (0.148) MOTHERS: 0.000 0.010 (0.245) 0.045 (0.141) 0.215 (0.182) OF MOTHER: 0.000 -0.482 (0.277) 0.028 (0.143) -0.071 (0.235) 0.000 Jp -0.270 (0.091) 0.000 -0.013 (0.050) 0.000 0.189 (0.066) 0.360 (0.070) 0.000 0.086 (0.0861 0.168 (0.077) 0.215 (0.086) 0.000 -0.262* (0.138) 0.073 (0.084) -0.098 (0.085) 10 10 0.000. 0.631 (0.282) -0.035 (0.080) 0.129 (0.138) ** 3791.3 INTERCEPT -3.704 -1.912*** 11 (0.947) -0.070 (0.016) 11 EDUCATION -0.018 (0.028) HOUSEHOLD INCOME 0.111 (0.088) 0.005 (0.050) 328 Table 5.3.2. (continued) BIRTH WEIGHT: (<2,500) 0.000 0.000 2,500-3,499 0.076 0.069 (0.155) (0.092) 3,500 + 0.052 -0.037 (0.164) (0.095) UNKNOWN -0.312 -0.159 (0.178) (0.107) LENGTH PREGNANCY: .(9 MONTHS) 0.000 0.000 -0.278*** . < 9 MONTHS -0.151 (0.168) (0.091) GENDER: . (MALE) 0.000 0.000 . FEMALE -0.143 -0.010 (0.089) (0.051) AGE OF CHILD: . (12-17) 0.000^ 0.000 . 18-35 MONTHS n a 0.343 0.188*** . 36-59 MONTHS (0.102) 9 c 9 c 9 c 9 c 0. 667 0.359 (0.119) (0.070) BIRTH ORDER: . (1st. Born) 0.000 0.000 . 2-3 0.034 0.082 (0.134) (0.074) VO 1 * ! • • 0.145 0.157 (0.142) (0.078) . 7 + -0.015 0.196 (0.148) (0.086) MARITAL STATUS OF MOTHERS: . (Married Mono.) 0.000 0.000 -0.268** . Single 0.074 (0.248) (0.139) . Married Polyg. 0.052 0.078 . Others (0.141) (0.084) 0.269 -0.104 (0.185) (0.086) Includes: Consensual unions, Divorced, Separated, Widowed * * * * p < = 0.001; *** p <= 0.01; ** p <= 0.05 ; * p <= 0.10 Table 5.3.2. (continued) OCCUPATION STATUS OF MOTHER: . (Housewife) 0 . 0 0 0 0 . 0 0 0 11 . Student -0.507 (0.277) 0.618 (0.281) 11 . Working 0.060 -0.036 . OthersW (0.145) (0.081) -0.090 (0.237) 0.136 (0.138) TYPES OF NEIGHBORHOODS: .(Squatting Neigb.)0.000 0 . 0 0 0 0.156** (0.069) . High Standing -0.006 <° - 1 2 2 l * . Average Stand. -0.226 0.058 (0.105) (0.060) 3780.1 INTERCEPT V ) | i K -3.656 -1.999**** 12 (0.952) -0.087 (0.023) 12 EDUCATION 0.029 (0.045) HOUSEHOLD INCOME 0.106 (0.088) 0.007 (0.050) BIRTH WEIGHT: (<2,500) 0 . 0 0 0 0 . 0 0 0 2,500-3,499 0.075 (0.156) 0.068 (0.092) 3,500 + 0.053 (0.165) -0.038 (0.095) UNKNOWN -0.315 (0.178) -0.155 (0.107) LENGTH PREGNANCY: .(9 MONTHS) 0 . 0 0 0 °*000*** -0.275 (0.092) . < 9 MONTHS -0.145 (0.169) GENDER: . (MALE) 0 . 0 0 0 0 . 0 0 0 . FEMALE -0.133 (0.089) -0.011 (0.051) AGE OF CHILD: . (12-17) 0 . 0 0 0 „ 0 . 0 0 0 _ . 18-35 MONTHS 0.336*** 0.189*** . 36-59 MONTHS ( 0 * 1 0 2 U * * 0.661 (0.119) ( 0 * 0 6 6 L * * 0.361 (0.070) * * * * p <= 0.001; *** p <= 0.01; ** p <= 0.05 ; * p <= 0.10 330 Table 5.3.2. (continued) BIRTH ORDER: . (1st. Born) 0.000 0.000 . 2-3 0.033 0.081 (0.134) (0.074) . 4-6 0.142 0.157 (0.143) (0.078) 0.192 . 7 + -0.009 (0.150) (0.087) MARITAL STATUS OF MOTHERS: . (Married Mono.) 0.000 0.000 . Single 0.061 -0.262** (0.250) (0.139) . Married Polyg. 0.043 0.081 . Others (0.141) (0.084) 0.263 -0.104 (0.184) (0.086) OCCUPATION STATUS OF MOTHER: . (Housewife) 0.000 0.000 ** 0.622 . Student -0.538* (0.278) (0.281) . Working 0.069 -0.036 . Others (0.145) (0.081) -0.106 0.138 (0.238) , . (0.138) TYPES OF NEIGHBORHOODS:' ' .(Squatting Neigb•)0.000 0.000 . High Standing 0.074 0.208 (°*276i* (0.136) . Average Stand. -0.518 0.142 INTERACTION: (0.104) (0.072) . Educ*Squatting 0.000 0.000 . Educ*High_Stand 0.015 0.020 . Educ*Aver Stand -0.106 0.033 Includes: Consensual unions, Divorced, Separated, Widowed Includes: Invalid, Never worked, etc. (c) High Standing= (STRATA 0 + STRATA 3); Average Standing= (STRATA 1 + STRATA 2); Squatting neighborhoods= (STRATA 4 + STRATA 6); **** p <= o.OOl; *** p <= 0.01; ** p <= 0.05 ; * p <= 0.10 331 Table 5.4.3.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Incidence of Diarrhoea among Children Aged 0-59 months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES M O DEL 1 M O DEL 2 M O D EL 3 M O DEL4 INTERCEPT 0.323**" (0.064) 0 .1 1 6 "" (0.310) 0 .0 5 5 "" (0.577) 0 .0 5 9 " " (0.342) EDUCATION 0 .9 8 2 " 0. 009) 0.979" (0.010) 1.118 (0.088) 0.972*" (0.011) H.INCOME* 1.119 (0.033) 1.212"* (0.062) 1.117"* (0.036) INTERACTION: Educ*Hincome 0.986 (0.009) — AGE O F CHILD: (in months) 0 -3 4-6 7-11 12-35 (36 o r more) 0.801 (0.184) 2.789**" (0.157) 5.212**" (0.128) 2.832**" (0.095) 1.000 -2 LOG L CHI-SQUARE D F 5470.393 3.832" 1 5300.014 14.74“ " 2 5297.673 17.08*"* s 4558.511 273.8**" 6 -eeend: (*) H .incom e = Household Income; **** p < = 0 .0 0 1 ; p < = 0 .0 1 ; ** p < = 0.05; * p < = 0 .1 0 332 Table 5.4.3. (continued) VARIABLES M O DEL 5 M ODEL 6 M O D EL 7 M O D EL 8 INTERCEPT 0.057“ “ 0.057“ “ 0.053**“ 0.068— (0.352) (0.357) (0.374) EDUCATION 0.971“ " 0.975“ 0.978“ 0.98* (0.011) (0.011) (0.011) (0.012) H.INCOM E 1.117*“ 1.113“ * 1.113“ * 1.115*“ (0.036) (0.036) (0.036) (0.037) AGE O F CHILD: 0-3 months 0.801 D.800 0.801 0.790 (0.184) (0.184) (0.186) (0.182) 4-6 months 2 .79"“ 2.802**“ 2.797**“ 2.79“ “ (0.157) (0.157) (0.159) (0.159) 7-11 months 5.217“ “ 5.228**“ 5.209**“ 5.200“ “ (0.128) (0.128) (0.130) (0.130) 12-35 months 2.828“ “ 2.822“ “ 2.805*“ * 2.786**“ (0.095) (0.095) (0.097) (0.097) (36 o r more) 1.000 1.000 1.000 1.000 GENDER: (Male) 1.000 1.000 1.000 1.000 Fem ale 1.069 1.069 1.078 1.072 (0.073) (0.073) (0.075) (0.075) BIRTH ORDER: (1st. Bom ) 1.000 1.000 1.000 2-3 0.983 0.972 0.976 (0.112) (0.114) (0.114) 4-6 1.021 1.041 1.049 (0.112) (0.114) (0.114) 7 + 1.104 1.127 1.133 (0.121) (0.123) (0.123) PREM ATURITY: . (9 months) 1.000 1.000 . < 9 months 1.443“ 1.370“ BIRTH W EIGHT:" . ( < 2,500) 1.000 . 2500-3499 0.721" (0.130) . 3,500 + 0.748" (0.138) . Unknown 0.821 (0.163) -2 LOG L CHI-SQUARE D F (°) In grams 333 Table 5.4.3. (continued) VARIABLES M O D EL 9 M O DEL 10 M O DEL 11 M O D EL 12 INTERCEPT 0.076— 0.102” ” 3.097**“ 0.098” ” [0.379) (0.399) (0.411) (0.412) EDUCATION 0.983 0.985 3.985 0.982 (0.012) (0.012) (0.012) (0.014) H.INCOM E 1.115” 1.114” * 1.118*” 1.119*“ (0.037) (0.037) (0.037) (0.037) AGE O F CHILD: 0-3 months 0.777 0.776 3.768 0.768 (0.188) (0.188) (0.189) (0.189) 4-6 m onths 2.831*“ * 2.817**“ 2.823**“ 2.814**“ (0.160) (0.161) (0.161) (0.161) 7-11 months 5.300“ “ 5.311**“ 5.303“ “ 5.300— (0.131) (0.131) (0.132) (0.132) 12-35 months 2.819“ “ 2.834**“ 2.828— 2.834**“ (0.097) (0.097) (0.097) (0.097) (36 o r more) 1.000 1.000 1.000 1.000 GEND ER: (Male) 1.000 1.000 1.000 1.000 Fem ale 1.076 1.073 1.077 1.080 (0.075) (0.075) (0.075) (0.076) BIRTH ORDER: (1st. Bom ) 1.000 1.000 1.000 1.000 2-3 1.123 1.111 1.109 1.113 (.125) (0.126) (0.128) (0.129) 4-6 1.315* 1.301* 1.317* 1.325* (0.147) (0.148) (0.151) (0.151) 7 + 1.446“ 1.426“ 1.453” 1.463“ (0.176) (0.176) (0.180) (0.180) PREM ATURITY: . (9 m onths) 1.000 1.000 1.000 1.000 . < 9 months 1.354“ 1.351“ 1.345“ 1.339“ (0.151) (0.151) (0.151) (0.152) BIRTH W EIG H T:b . ( < 2,500) 1.000 1.000 1.000 1.000 . 2500-3499 0.730“ 3.726“ 0.724” 0.724” (0.131) (0.131) (0.131) (0.132) . 3,500 + 0.761“ 0.759“ 0.754“ 0.753“ (0.139) (0.139) (0.139) (0.139) . Unknown 0.833 0.681“ 0.683“ 0.684“ (0.164) (0.184) (0.185) (0.185) -2 LOG L 4388.111 4380.551 4377.884 4374.208 CHI-SQUARE 287.6“ ” 293.6**” 296.2**“ 299.9— D F 18 21 24 (°) In grams 334 Tabic 5.4.3. (continued) VARIABLES M O D EL 9 M O DEL 10 M O D EL 11 M O D EL 12 AGE O F M O TH ER AT BIRTH O F CHILD: . ( < 20 YEARS) 1.000 . 20-24 YEARS 0.710"“ 1.000 1.000 1.000 (0.128) 0.715*“ 0.710“ * 0.703“ * . 25-34 YEARS 0.671*“ (0.129) (0.129) (0.129) (0.150) 0.677*“ 0.670*“ 0.661*“ . 35 + 0.679“ (0.150) (0.151) (0.151) (0.196) 0.693* 0.683* 0.675“ (0.197) ; < M 9 S ) (0.198) PLACE o f DELIVERY: . (At Home) . Hospital 1.000 1.000 1.000 0.733“ 0.736“ 0.737“ (0.135) (0.135) (0.135) M ARITAL STATUS o f M OTHER: .(M arried M onogam ously) . Single 1.000 1.000 0.987 1.387 . M arried Poly (0.186) (0.353) 1.021 0.932 . O thers' (0.120) (0.223) 1.209 0.947 (0.133) (0.263) INTERACTIONS: .(Edu*M M on) 1.000 ,Edu*Single 0.932 (0.051) .Edu*M arPol 1.017 (0.033) .Edu*Others 1.041 (0.037) -2 LOG L 4388.111 4380.551 4377.884 4374.208 CHI-SQUARE 287.6“ “ 293.6“ “ 296.2“ “ 299.9“ “ D F 17 18 21 24 (°) This includes: Consensual unions, D ivorced, Separated, W idow ers, etc. **** p < ==0.001; *** p < = 0 .0 1 ; ** p < = 0.05; * p < = 0 .1 0 335 Table 5.4.3. (continued) VARIABLES M O D EL 13 M O DEL 14 M O DEL 15 M O D EL 16 INTERCEPT 0.097**** 0.128“ * 3.124“ * 3.113*“ (0.421) (0.428) (0.436) EDUCATION 0.987 0.986 3.989 3.988 (0.012) (0.012) (0.018) (0.013) H.IN CO M E 1.117— 1.092“ 1.093“ 1.094“ (0.037) (0.038) (0.038) (0.038) AGE O F CHILD: 0-3 months 0.764 0.758 3.758 3.781 (0.189) (0.189) (0.189) (0.192) 4-6 months 2.801**** 2.760**** 2.755*“ 2.789— (0.161) (0.161) (0.161) (0.164) 7-11 months 5.301**** 5.267*“ 5.282“ * 5.487*“ (0.131) (0.132) (0.132) (0.134) 12-35 months 2.835**** 2.818*“ 2.282*“ 2.938“ * (0.097) (0.097) (0.098) (0.010) (36 o r more) 1.000 1.000 1.000 1.000 GENDER: (Male) 1.000 1.000 1.000 1.000 Fem ale 1.082 1.081 1.082 1.086 (0.076) (0.076) (0.076) (0.076) BIRTH ORDER: (1st. Born) 1.000 1.000 1.000 1.000 2-3 1.119 1.110 1.110 1.090 (0.129) (0.129) (0.129) (0.130) 4-6 1.318* 1.294* 1.294* 1.296* (0.151) (0.152) (0.152) (0.153) 7 + 1.450“ 1.410* 1.408* 1.324 (0.181) (0.181) (0.181) (0.184) PREM ATURITY: . (9 months) 1.000 1.000 1.000 1.000 . < 9 months 1.337* 1.348“ 1.353” 1.379” (0.152) (0.152) (0.152) (0.155) BIRTH W EIG H T:b . ( < 2,500) 1.000 1.000 1.000 1.000 . 2500-3499 3.721“ 0.712“ 0.711“ 0.730” (0.131) (0.131) (0.132) (0.134) . 3,500 + 3.753** 3.739” 3.739“ 0.756” (0.139) (0.140) (0.140) (0.142) . Unknown 0.681“ 0.670” 0.674” 0.707* (0.185) (0.185) (0.186) (0.189) -2 LOG L 4373.730 4363.565 4362.860 4248.772 CHI-SQUARE 300.4**** 310.6*“ 311.3*“ 315.1“ * D F 24 26 28 28 (°) In grams 336 Table S.4.3. (continued) VARIABLES M O D EL 13 M O DEL 14 M O D EL 15 M O DEL 16 AGE O F M OTHER: 1.000 . ( < 20 YEARS) 0.719“ * 1.000 1.000 1.000 . 20-24 YEARS (0.130) 0.739“ 0.740“ 0.725“ 0.692“ (0.130) (0.130) (0.132) . 25-34 YEARS (0.152) 0.716“ 0.717“ 0.705“ 0.714* (0.153) (0.153) (0.155) . 35 + (0.199) 0.739* 0.739 0.770 (0.200) (0.200) (0.202) PLACE o f DELIVERY: . (At Home) . Hospital 1.000 1.000 1.000 1.000 0.730*“ 0.770* 0.769* 0.783* (0.134) (0.136) (0.136) (0.138) M ARITAL STATUS o f M OTHER: .(M arried M onogam ously) . Single 1.000 1.000 1.000 1.000 0.975 1.053 1.054 1.028 . M arried Poly (0.232) (0.232) (0.232) (0.235) 1.024 1.031 1.033 1.046 . O thers0 (0.120) (0.121) (0.121) (0.122) 1.261* 1.352“ 1.352“ 1.352“ (0.138) (0.140) (0.140) (0.141) OCCUPATION O F M OTHER: .(Housewife) 1.000 1.000 1.000 1.000 . Student 1.107 1.095 1.083 1.064 (0.294) (0.294) (0.294) (0.300) . W orking 0.808* 0.838 0.842 0.830 (0.127) (0.128) (0.128) (0.129) . Others 0.805 0.790 0.791 0.823 (0.217) (0.216) (0.216) (0.218) TYPE O F NEIGHBORHOODS: . (Squatting) 1.000 1.000 1.000 .High Standing 0.955 1.086 1.009 N eighborhoods (0.096) (0.202) (0.106) .Av. Standing 0.751“ * 0.736* 0.766*“ Neighborhoods (0.094) (0.172) (0.096) -2 LOG L CHI-SQUARE DF O This includes: C onsensual unions, D ivorced, Separated, W idow ers, etc. 337 Table 5.4.3. (continued) VARIABLES M O D EL 15 M O D EL 16 INTERACTION: .Edu*Squatting 1.000 •Edu*High_Stan 0.980 (0.028) .Edu*Av Stand 1.004 (0.025) HOUSING QUALITY: M aterial o f Construction .(Low:Index 1-7) 1.000 . Average: 7-10 1.053 (0.113) . High: 10-12 0.968 (0.134) (°) This includes: Consensual unions, D ivorced, Separated, W idow ers, etc. Table 5.4.3. (continued) VARIABLES M O D EL 17 M O DEL 18 INTERCEPT 0.114*“ * 0.106**“ (0.438) (0.450) EDUCATION 0.991 0.991 (0.013) (0.013) H.INCOM E 1.117*** 1.121*“ (0.039) (0.039) AGE O F CHILD: 0-3 months 0.778 0.778 (0.192) (0.192) 4-6 months 2.769**“ 2.784**“ (0.165) (0.165) 7-11 months 5.486“ “ 5.496**“ (0.134) (0.134) 12-35 months 2.920— 2.923**“ (0.100) (0.099) (36 o r more) 1.000 1.000 GENDER: (Male) 1.000 1.000 Fem ale 1.091 1.089 (0.077) (0.077) BIRTH ORDER: (1st. Bom ) 1.000 1.000 2-3 1.054 1.054 (0.131) (0.131) 4-6 1.257 1.252 (0.153) (0.154) 7 + 1.328 1.324 (0.184) (0.184) PREM ATURITY: . (9 months) 1.000 1.000 . < 9 months 1.424“ 1.429“ (0.156) (0.156) BIRTH W EIGH T:6 . ( < 2,500) 1.000 1.000 . 2500-3499 0.730“ 0.727“ (0.134) (0.135) . 3,500 + 0.759* 0.759“ (0.142) (0.142) . Unknown 0.714* 0.717* (0.189) (0.189) -2 LOG L 4238.369 4237.543 CHI-SQUARE 325.4— 326.3**“ D F 29 31 O In grams Table 5.4.3. (continued) VARIABLES M O DEL 17 M O DEL 18 AGE O F M O TH ER .( < 20 years) 1.000 1.000 . 20-24 years 0 .7 3 6 " 0.736“ (0.132) (0.132) . 25-35 years 0.7 3 1 " 0 .7 3 0 " (0.155) (0.155) . 354- 0.835 0.835 (0.204) (0.204) PLACE o f DELIVERY . (At Home) 1.000 1.000 . Hospital 0.794’ 0.792* (0.138) (0.138) MARITAL STATUS o f M OTHER: .(M arried M onogamously) 1.000 1.000 . Single 1.052 1.040 (0.236) (0.236) . M arried Polygamously 1.101 1.099 (0.123) (0.123) . O thers' 1.351“ 1.343“ (0.141) (0.141) OCCUPATION O F M O TH ER . (Housewife) 1.000 1.000 . Single 1.116 1.116 (0.301) (0.301) . W orking m other 0.836 0.839 (0.130) (0.130) . Others 0.881 0.889 (0.219) (0.219) TYPE O F NEIGHBORHOOD . (Squatting Neighborhood) 1.000 1.000 . High Standing 1.057 1.061 Neighborhoods (0.107) (0.108) . A verage Stand. 0.7 9 4 " 0.791“ Neighborhoods (0.097) (0.097) INDEX O F QU ALITY O F NEIGHBOR. .(Low: Index 1-7) 1.000 1.000 . A verage: 7-10 1.048 1.052 (0.113) (0.114) . High: 10-12 0.970 0.983 (0.134) (0.137) -2 LOG L 4238.369 4237.543 CHI-SQUARE 3 2 5 .4 "" 3 2 6 .3 "“ D F 29 31 ..egend: Neighbor. = Neighborhood Table 5.4.3 (continued) VARIABLES M O DEL 17 M O DEL 18 NUM BER O F BEDROOMS 0.921“ * 0.922*“ (0.026) (0.026) QUALITY O F IM M ED IATE ENVIRO NM ENT: . (Low: index 1-7) 1.000 . Average: 7-10 1.073 (0.115) . High: 10-12 0.998 (0.126) -2 LOG L 4238.369 4237.543 CHI-SQUARE 325.4“ “ 326.3**“ D F 29 31 ^egend: *"»*• p < =O.Q01: ♦*» d < = 0 .0 l: • • d < = 0.05: * o < = 0 .1 0 Table 5.4.4. Odds Ratio and Standard Error (in parenthesis) for the Logistic Regression of Use of Sugar-Salt- Solution by Mothers whose Children are aged 0-59 Months. Urban Areas Zaire, 13 Cities, Fonames/UNICEF Survey, 1987. VARIABLE M O D EL 1 M O D EL 2 M O D EL 3 IN TER C EPT 0.696“ * 0.429 0.630 (0.115) (0.539) (1.027) EDUCA TIO N 1.056“ * 1.057*“ 0.985 (0.017) (0.017) (0.160) H O USEHO LD 1.051 1.009 INCO M E (0.057) (0.110) INTERACTION: 1.008 Edu*H _Incom e (0.017) -2 LOG L 1566.698 1524.893 1524.700 CHI SQUARE 10.114*“ 11.526“ * 11.719*“ D F 1 2 3 H _Incom e= H ousehold Income; 342 Table 5.4.4. (continued) VARIABLE M O D EL 4 M O DEL 5 M O D EL 6 IN TER C EPT 0.365* 0 .1 7 9 - 0.187“ (0.560) (0.602) (0.603) EDUCATION 1.062“ * 1.057— 1.057— (0.018) (0.019) (0.019) HO USEHO LD 1.068 1.106 1.108* INCO M E (0.059) (0.062) (0.062) D EHY DRATION RISK:1 .(Low) 1.000 1.000 1.000 High 0.968 1.000 1.009 (0.151) (0.162) (0.162) AGE O F CHILD IN M ONTHS: .(3 6 + ) 1.000 1.000 .0-3 0.750 0.749 (0.379) (0.379) .4-6 2.223— 2 .2 1 3 - (0.293) (0.293) .7-11 1.692** 1.699** (0.228) (0.228) .12-36 1 .6 6 0 - 1.667“ (0.189) (0.189) GEND ER: .(M ale) 1.000 •Female 0.884 (0.133) -2 LOG L 1415.277 1281.263 1280.406 CHI SQUARE 13.386— 2 8 .5 4 7 * - 29.404— D F 3 7 8 L e g e n d :1 L ow risk when the m other gave the child lots o f w ater w hile the child had <iarrhea; high risk means the m other stopped giving w ater to the child o r gave very little w ater when the child had diarrhea. **** p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 343 Table S.4.4. (continued) VARIABLE M O D EL 7 M O DEL 8 M O DEL 9 IN TER C EPT 0.159” * 0.148*” 0.161” * (0.623) (0.655) (0.665) EDUCA TIO N 1.061” * 1.052“ 1.051” (0.020) (0.021) (0.021) H O USEHO LD INCOM E 1.108 1.107 1.107 (0.062) (0.063) (0.063) DEHY DRATION RISK: .(Low) 1.000 1.000 1.000 •High 1.015 0.999 0.990 (0162) (0.163) (0.164) AGE O F CHILD IN M ONTHS: .(3 6 + ) 1.000 1.000 1.000 .0-3 0.755 0.758 0.777 (0.380) (0.381) (0.384) .4-6 2.206” * 2.237*“ 2.289*” (0.293) (0.295) (0.298) .7-11 1.730” 1.754“ 1.729” (0.228) (0.229) (0.230) .12-36 1.653” * 1.678*“ 1.670*” (0.190) (0.191) (0.191) GENDER: .(M ale) 1.000 1.000 1.000 .Fem ale 0.887 0.873 0.862 (0.133) (0.135) (0.135) BIRTH ORDER: .(1st) 1.000 1.000 1.000 .2-3 1.227 1.257 1.270 (0.206) (0.208) (0.219) .4-6 1.033 1.051 1.032 (0.206) (0.207) (0.258) .7 + 1.308 1.288 1.161 (0.269) (0.220) (0.313) Legend: **** p < =0.001: *** p < =0.01: * " » p < =6.05: * p < =6.16 344 Table 5.4.4. (continued) VARIABLE M O D EL 7 M O D EL 8 M O DEL 9 BIRTH W EIG H T; in gram s: .(< 2 5 0 0 ) 1.000 1.000 .2500-3499 1.352 1.314 (0.228) (0.229) .3 5 0 0 + 1.044 1.014 (0.240) (0.241) . unknown 0.821 0.801 (0.288) (0.290) M O TH ER ’S AG E A T BIRTH: .( < 2 0 yrs) 0.908 .20-24 (0.222) 1.000 .25-34 (0.260) 1.111 .3 5 + (0.350) -2 LOG L 1277.744 1271.037 1265.295 CHI SQUARE 32.066“ 38.773“ 37.669"* DF 11 14 17 345 Table 5.4.4. (continued) VARIABLE M O D EL 10 M O DEL 11 M O D EL 12 INTERCEPT 0.150” * 0.190” 0.190” (0.694) (0.725) (0.725) EDUCATION 1.050” 1.046“ 1.050* (0.021) (0.021) (0.025) HO USEHO LD INCOM E 1.107 1.106 1.104 (0.063) (0.064) (0.065) DEHY DRATION RISK:1 .(Low) 1.000 1.000 1.000 ■High 0.987 1.021 1.023 (0.165) (0.167) (0.167) AGE O F CHILD IN M ONTHS: •(36+) 1.000 1.000 1.000 .0-3 0.775 0.787 0.778 (0.384) (0.385) (0.386) .4-6 2.297” * 2.371"* 2.375” * (0.297) (0.301) (0.302) .7-11 1.731” 1.816“ 1.804” (0.230) (0.232) (0.233) .12-36 1.666"* 1.718“ * 1.710“ * (0.192) (0.193) (0.193) GENDER: .(Male) 1.000 1.000 1.000 .Fem ale 0.862 0.861 0.860 (0.135) (0.137) (0.137) BIRTH ORDER: .(1st) 1.000 1.000 1.000 .2-3 1.270 1.150 1.154 (0.219) (0.228) (0.299) .4-6 1.037 0.939 0.938 (0.258) (0.267) (0.268) .7 + 1.171 1.069 1.070 (0.313) (0.323) (0.323) 1 Low risk= the mother gave the child lots of water while the child had diarrhea; high risk means the mother stopped giving water to the child or gave very little water when the child had diarrhea. Table 5.4.4. (continued) VARIABLE M O D E L 10 M O D EL 11 M O D EL 12 BIRTH W EIG H T (in grams) .(< 2 5 0 0 ) 1.000 1.000 1.000 .2500-3499 1.311 1.239 1.235 (0.229) (0.232) (0.233) .3 5 0 0 + 1.013 0.976 0.974 (0.241) (0.245) (0.246) . unknown 0.844 0.807 0.800 (0.326) (0.330) (0.331) M O TH ER ’S A G E AT BIRTH: .( < 2 0 yrs) 1.000 1.000 1.000 .20-24 0.907 0.818 0.821 (0.222) (0.228) (0.228) .25-34 0.995 0.909 0.917 (0.260) (0.265) (0.260) .35 + 1.104 1.019 1.026 (0.350) (0.356) (0.357) PLA CE O F DELIVERY: .(Home) 1.000 1.000 1.000 .Health Institutions 1.087 1.125 1.121 (0.235) (0.236) (0.236) M ARITAL STATUS O F M O TH ER: .(M arried M onogamously) 1.000 1.000 .Single 0.266"* 0.203* (0.408) (0.881) .M arried Polygamously 0.794 0.889 (0.212) (0.400) .Others 1.177 1.303 (0.234) (0.454) INTERACTION: ,(Edu*M arried Mono) 1.000 .Edu*Single 1.046 (0.126) .Edu*M arried Poly 0.980 (0.060) .Edu*Others 0.984 (0.062) -2 LOG L 1265.168 1250.494 1250.179 CHI SQUARE 37.796*" 5 2 .4 6 9 "" 52.784"* D F 18 21 24 347 VARIABLE M O DEL 13 M O DEL 14 M ODEL 15 IN TER C EPT 0 .1 8 7 " 0 .2 0 0 " 0 .2 1 2 " (0.726) (0.736) (0.745) EDUCATION 1.038' 1.033 1.024 (0.022) (0.022) (0.033) H O USEHO LD INCO M E 1.109 1.100 1.098 (0.064) (0.065) (0.065) DEHY DRATION RISK:2 .(Low) 1.000 1.000 1.000 •High 1.018 1.035 1.049 (0.167) (0.168) (0.169) A G E O F CHILD ( in M onths) .(3 6 + ) 1.000 1.000 1.000 .0-3 0.806 0.794 0.790 (0.389) (0.390) (0.390) .4-6 2.469” 2.432” 2.430” (0.303) (0.304) (0.305) .7-11 1.855” 1.851” 1.842” (0.234) (0.235) (0.235) .12-35 1.750” 1.756” 1.756” (0.194) (0.195) (0.195) GENDER: .(M ale) 1.000 1.000 1.000 •Female 0.856 0.863 0.863 (0.138) (0.138) (0.138) BIRTH ORDER: .(1st) 1.000 1.000 1.000 .2-3 1.187 1.170 1.184 (0.229) (0.230) (0.231) .4-6 0.980 0.948 0.949 (0.270) (0.271) (0.272) .7 + 1.132 1.084 1.095 (0.326) (0.328) (0.328) 2 Low risk= the mother gave the child lots of water while the child had diarrhea; high risk means the mother stopped giving water to the child or gave very little water when the child had diarrhea. 348 Table 5.4.4. (continued) VARIABLE M O D EL 13 M O DEL 14 M O DEL 15 BIRTH W EIG H T (in grams) .( < 2500) 1.000 1.000 1.000 .2500-3499 1.202 1.194 1.192 (0.235) (0.236) (0.236) .3500+ 0.963 0.955 0.965 (0.247) (0.247) (0.248) . Unknown 0.744 0.765 0.752 (0.332) (0.334) (0.334) M O TH ER ’S AGE AT BIRTH: .( < 2 0 years) 1.000 1.000 1.000 .20-24 0.818 0.837 0.837 (0.229) (0.230) (0.230) .25-34 0.860 0.885 0.882 (0.269) (0.271) (0.272) .35 + 0.937 0.942 0.947 (0.361) (0.363) (0.363) PLA CE O F DELIVERY: .(Home) 1.000 1.000 1.000 .Health Institutions 1.110 1.128 1.137 (0.236) (0.241) (0.242) M ARITA L STATUS O F M OTHER: .(M arried M onogam ously) 1.000 1.000 1.000 .Single 0.115*"“ 0.123“ “ 0.121**“ (0.546) (0.549) (0.548) .M arried Polygamously 0.807 0.822 0.812 (0.212) (0.213) (0.214) .Others* 0.998 1.041 1.035 (0.243) (0.247) (0.247) OCCU PATION : .(Housewife) 1.000 1.000 1.000 .Student 2.914* 2.861* 2.991* (0.581) (0.586) (0.591) .W orking 1.459 1.495* 1.503* (0.230) (0.232) (0.233) .Others'1 2.703“ 2.741“ 2.740“ (0.470) (0.472) (0.470) 349 Table 5.4.4. (continued) VARIABLE M O DEL 13 M O DEL 14 M O D EL 15 ACCESS TO M ODERN FA CILITY : .(Squatting) 1.000 1.000 .High Standing 0.862 0.908 Neighborhood (0.173) (0.312) .A vrg Standing 1.265 0.915 N eighborhood (0.172) (0.367) INTERACTION: .Edu*Squatting 1.000 .Edu*High_Stand. 0.990 N eighborhood (0.046) .Edu* Average 1.052 Standing Neigh. (0.052) -2 LOG L 1242.090 1237.755 1236.341 CHI SQUARE 60.873“ 65.208“ 66.623“ D F 24 26 28 Legend: High Stand. = High Standing Neighborhood 350 Table S.4.4. (continued) VARIABLE M O D EL 16 M O DEL 17 M O DEL 18 IN TERCEPT 0.201 ~ 0.202** 0.170** (0.761) (0.761) (0.786) EDUCATION 1.043’ 1.042 1.043’ (0.023) (0.023) (0.023) HO USEHO LD INCOM E 1.103 1.095 1.102 (0.066) (0.067) (0.067) DEHY DRATION RISK:5 .(Low) 1.000 1.000 1.000 •High 1.027 1.023 1.033 (0.171) (0.171) (0.172) AGE O F CH ILD (in months) .(3 6 + ) 1.000 1.000 1.000 .0-3 0.751 0.750 0.749 (0.403) (0.403) (0.403) .4-6 2.702*** 2.698*** 2.729*** (0.315) (0.315) (0.316) .7-11 1.874*** 1.878*** 1.888*** (0.241) (0.241) (0.242) .12-36 1.796*** 1.800*** 1.813*** (0.201) (0.201) (0.202) GENDER: .(M ale) 1.000 1.000 1.000 .Female 0.839 0.838 0.832 (0.141) (0.141) (0.141) BIRTH ORDER: •(1st) 1.000 1.000 1.000 .2-3 1.188 1.212 1.230 (0.234) (0.238) (0.238) .4-6 1.022 1.038 1.039 (0.276) (0.278) (0.279) .7 + 1.208 1.217 1.228 (0.337) (0.337) (0.338) 3 Low risk= the mother gave the child lots of water While the child had diarrhea; high risk means the mother stopped giving water to the child or gave very little water when the child had diarrhea. Table 5.4.4. (continued) VARIABLE M O D E L 13 M O DEL 14 M O D EL 15 BIRTH W EIG H T (in grams) .(< 2 5 0 0 ) 1.000 1.000 1.000 .2500-3499 1.217 1.220 1.203 (0.242) (0.242) (0.243) .3 5 0 0 + 0.931 0.934 0.933 (0.253) (0.253) (0.253) . Unknown 0.746 0.743 0.736 (0.341) (0.341) (0.342) M O TH ER ’S A G E AT BIRTH: .( < 2 0 years) 1.000 1.000 1.000 .20-24 0.829 0.820 0.822 (0.233) (0.234) (0.234) .25-34 0.824 0.809 0.799 (0.276) (0.278) (0.278) .3 5 + 0.878 0.848 0.841 (0.370) (0.376) (0.377) PLA CE O F DELIVERY: .(Home) 1.000 1.000 1.000 .Health Institutions 1.168 1.157 1.143 (0.245) (0.246) (0.246) M A RITA L STATUS O F M O TH ER: .(M arried M onogam ously) 1.000 1.000 1.000 .Single 0.131*"“ 0.130**“ 0.128“ “ (0.550) (0.550) (0.551) .M arried Polygamously 0.860 0.845 0.834 (0.216) (0.218) (0.219) .Others* 1.020 1.028 1.018 (0.249) (0.250) (0.251) OCCU PATION : .(Housewife) 1.000 1.000 1.000 .Student 2.608 2.610 2.627 (0.600) (0.600) (0.600) .W orking 1.480* 1.475* 1.456* (0.236) (0.236) (0.237) .Othersb 2.727“ 2.694“ 2.715“ (0.474) (0.474) (0.474) -eeend: M ar. M o n o = M arried M onoeam ouslv: M ar. Polv. = M arried Polygamously; (*) This includes: consensus union, divorced, separated, widows, etc.; (*) This includes: N ever w orked, inactives, etc. 352 Table 5.4.4. (continued) VA RIABLE M O DEL 16 M O D EL 17 M O D EL 18 ACCESS TO M O DERN FACILITY: .(Squatting) 1.000 1.000 1.000 .High Standing 0.927 0.914 0.892 N eighborhoods (0.176) (0.179) (0.180) .A verage Standing 1.577“ 1.555 1.579“ N eighborhoods (0.190) (0.192) (0.194) IND EX O F HOUSING: .(Low: 1-7) 1.000 1.000 1.000 .Avrg: 7-10 0.882 0.883 0.878 (0.199) (0.199) (0.202) .High: 10-12 0.606“ 0.607“ 0.610“ (0.239) (0.239) (0.244) BEDROOM S 1.025 1.029 (0.047) (0.047) IN D EX O F IM M ED IA TE ENVIRO N M EN T: .(Low: 1-7) 1.000 .A verage: 7-10 1.222 (0.208) .High: 10-12 1.046 (0.227) -2 LOG L 1199.405 1199.135 1197.700 CHI SQ U A R E 71.643“ “ 71.913“ “ 73.348“ “ D F 28 29 31 ~egend: Avrg standing neighborhood = A verage Standing N eighborhoods; **** p < = 0 .0 0 1 ; •* * p < = 0 .0 1 ; ** p < = 0 .0 5 ; • p < = 0 .1 0 353 Table 5.4.5.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Modern Services in Case of Diarrhoea by Mothers whose Children Are Aged 0-59. Urban Areas of Zaire, 13 Cities. Fonames- Unicef Survey, 1987. VARIABLES M odel 1 M odel 2 M odel 3 M odel 4 IN TER C EPT 1.128 1.098 0.312 0.794 (0.112) (0.525) (0.979) (0.553) EDUCA TIO N 1.018 1.017 1.284 1.001 (0.017) (0.017) (0.154) (0.017) H INCOM E 1.003 1.149 0.991 (0.056) (0.105) (0.058) EDU*H INCOM 0.975 (0.016) USE O F SSS . (No) 1.000 . Yes 3.222*"* (0.126) -2 LOG L 1635.5 1595.5 1593.1 1470.3 CHI-SQUARE 1.106 0.949 3.298 91.26— DF 1 2 3 3 .egend: H m com e= H ousehold Income; Edu* Hincome= Interaction betw een education and Household incom e; SSS Sugar-Salt-Solution; •*** p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 354 Table 5.4.5. (continued) VARIABLES M odel 5 M odel 6 M odel 7 M odel 8 IN TER C EPT 0.872 0.790 0.779 0.871 (0.593) (0.629) (0.631) (0.649) EDUCA TIO N 1.000 1.000 1.000 1.000 (0.019) (0.020) (0.020) (0.021) H INCOM E 0.979 0.998 0.997 0.996 (0.063) (0.065) (0.065) (0.065) USE O F SSS 3.062” 2.986” 2.990” 3.017” (0.134) (0.141) (0.141) (0.141) D EHY DRATION RISK:(«®) .(Low) 1.000 1.000 1.000 1.000 . High 1.629*“ 1.767*“ 1.762“ * 1.754*“ (0.165) (0.177) (0.177) (0.177) CHILD AGE: (months) .0-3 0.451“ 0.452“ 0.445“ (0.379) (0.379) (0.380) . 4-6 1.304 1.306 1.300 (0.314) (0.314) (0.314) . 7-11 1.109 1.108 1.095 (0.240) (0.240) (0.240) . 12-35 0.891 0.890 0.890 (0.193) (0.193) (0.194) .(36 + ) 1.000 1.000 1.000 GENDER: . (Male) 1.000 1.000 1.000 . Fem ale 1.036 1.036 1.034 (0.139) (0.139) (0.139) BIRTH ORD ER .(1st bom ) 1.000 . 2-3 0.839 (0.216) . 4-6 0.959 (0.214) . 7 + 0.871 (0.229) -2 LOG L 1308.9 1192.5 1192.4 1191.5 CHI-SQUARE 82.3” 85.03” 85.1” 86.5” D F 4 8 9 12 .eeend: (qra) see m eaning in Table 5.4.4. 355 Table 5.4.5. (continued) VARIABLES M odel 9 M odel 10 M odel 11 M odel 12 IN TER C EPT 0.742 0.568 0.434 0.442 (0.659) (0.672) (0.697) (0.699) EDUCATION 0.998 0.987 0.988 0.984 (0.022) (0.023) (0.023) (0.021) H INCO M E 0.996 0.992 1.012 1.011 (0.066) (0.066) (0.065) (0.065) USE O F SSS 3.045“ 3.028“ 3.039“ 3.060“ (0.142) (0.143) (0.144) (0.144) DEHY DRATION RISK :(*°) .(Low) 1.000 1.000 1.000 1.000 • High 1.778*"" 1.722"“ 1.692“ " 1.692*“ (0.180) (0.180) (0.181) (0.182) CHILD AGE: (months) .0-3 0.450 0.445“ 0.449“ 0.444“ (0.380) (0.383) (0.385) (0.388) . 4-6 1.256 1.285 1.299 1.293 (0.320) (0.321) 0.322) (0.323) . 7-11 1.045 1.061 1.052 1.064 (0.242) (0.243) (0.244) (0.244) . 12-35 0.850 0.846 0.839 0.841 (0.196) (0.196) (0.197) (0.197) •(36 + ) 1.000 1.000 1.000 1.000 GENDER: . (Male) 1.000 1.000 1.000 1.000 . Fem ale 1.014 1.011 1.015 1.011 (0.140) (0.141) (0.142) (0.142) BIRTH ORD ER .(1st bom ) 1.000 1.000 1.000 1.000 . 2-3 0.733 0.746 0.806 0.809 (0.230) (0.231) (0.238) (0.238) . 4-6 0.880 0.909 1.001 1.003 (0.271) (0.273) (0.281) (0.286) . 7 + 0.791 0.813 0.892 0.888 (0.328) (0.329) (0.337) (0.338) -2 LOG L 1179.4 1174.01 1171.4 1168.8 CHI-SQUARE 92.5“ 97.9“ 100.5“ 103.1“ D F 15 16 19 22 ^eeend: f88) see m eaning in Table 5.4.4. 356 Tabic 5.4.5. (continued) VARIABLE M odel 9 M odel 10 M odel 11 M odel 12 M O TH ER ’S AGE . ( < 20 ) 1.000 1.000 1.000 1.000 . 20-24 1.721“ 1.701“ 1.725“ 1.758“ (0.232) (0.232) (0.235) (0.236) . 25-34 1.238 1.191 1.208 1.229 (0.270) (0.271) (0.273) (0.275) . 35 + 1.410 1.355 1.405 1.440 (0.364) (0.365) (0.369) (0.370) PLA CE O F DELIV ERY: . (At Home) 1.000 1.000 1.000 . Healt institution 1.575“ 1.546“ 1.547“ (0.196) (0.197) (0.197) M ARITAL STATUS: .(M arried M ono) 1.000 1.000 . Single 1.367 1.096 (0.374) (0.733) . M arried P. 0.798 0.515 (0.222) (0.434) . Others 1.252 1.818 (0.250) (0.503) INTERACTION: .(Edu*M _mon) 1.000 .Edu*Single 1.040 (1.040) . Edu*M _pol. 1.080 (0.047) . Edu*Others 0.944 (0.068) -2 LOG L 1179.4 1174.01 1171.4 1168.8 CHI-SQUARE 92.5**** 97.9**** 100.5“ “ 103.1**“ D F 15 16 19 22 .eeend: M m o n = M arried M onoeam ouslv: M pol. = M arried Polygamously; **** p < = ( >.001; *** p < = 0 .0 1 ; * * p < = 0 .0 5 ; * p < = 0 .1 0 357 Table 5.4.5. (continued) VARIABLES M odel 13 M odel 14 M odel 15 M odel 16 IN TER C EPT 0.410 0.279* 0.321 0.324 (0.699) (0.712) (0.729) (0.729) EDUCATION 0.982 0.984 0.978 0.980 (0.023) (0.023) (0.024) (0.024) H INCO M E 1.017 1.048 1.043 1.059 (0.067) (0.068) (0.069) (0.070) USE O F SSS 2.975**“ 3.020**** 3.058**** 3.072**“ (0.144) (0.146) (0.148) (0.148) DEHY DRATION RISK: .(Low) 1.000 1.000 1.000 1.000 . High 1.696*“ 1.617— 1.615— 1.626*“ (0.182) (0.183) (0.185) (0.185) CHILD AGE: (months) .0-3 0.463** 0.497* 0.480* 0.479* (0.387) (0.389) (0.398) (0.398) . 4-6 1.331 1.314 1.227 1.231 (0.323) (0.324) 0.330) (0.330) . 7-11 1.062 1.070 1.100 1.096 (0.245) (0.247) (0.251) (0.251) . 12-35 0.837 0.824 0.859 0.856 (0.198) (0.199) (0.204) (0.204) ■(36 + ) 1.000 1.000 1.000 1.000 GENDER: . (Male) 1.000 1.000 1.000 1.000 . Fem ale 1.006 1.016 1.020 1.023 (0.142) (0.143) (0.145) (0.145) BIRTH ORD ER .(1st bom ) 1.000 1.000 1.000 1.000 . 2-3 0.823 0.835 0.844 0.799 (0.238) (0.239) (0.242) (0.246) . 4-6 1.034 1.094 1.079 1.027 (0.283) (0.284) (0.288) (0.290) . 7 + 0.939 0.993 0.894 0.870 (0.339) (0.341) (0.348) (0.348) -2 LO G L 1165.6 1153 1125.4 1123.7 CHI-SQUARE 106.3— 118.9— 116.8**** 118.6“ “ D F 22 24 26 27 358 Table 5.4.5. (continued) VARIABLE M odel 13 M odel 14 M odel 15 M odel 16 M O TH ER ’S AGE • ( < 20 ) 1.000 1.000 1.000 1.000 . 20-24 1.724” 1.648“ 1.590” 1.637“ (0.236) (0.2372) (0.240) (0.241) . 25-34 1.155 1.055 1.011 1.065 (0.277) (0.280) (0.283) (0.285) . 354- 1.310 1.194 1.241 1.366 (0.373) (0.376) (0.381) (0.388) PLA C E O F DELIV ERY: . (At Hom e) 1.000 1.000 1.000 1.000 . Health 1.568” 1.346 1.324 1.348” institutions (0.197) (0.204) (0.206) (0.207) M ARITA L STATUS: .(M arried M ono) 1.000 1.000 1.000 1.000 . Single 0.797 0.707 0.673 0.685 (0.517) (0.521) (0.521) (0.521) . M arried Polygamously 0.800 0.799 0.826 0.863 (0.222) (0.224) (0.226) (0.229) . O thers 1.107 0.985 0.997 0.977 (0.258) (0.262) (0.263) (0.264) OCCU PATION STATUS: .(Housewife) 1.000 1.000 1.000 1.000 . Students 2.555 2.523 2.435 2.423 (0.595) (0.594) (0.594) (0.594) . W orking 1.576' 1.433 1.428 1.441 (0.244) (0.247) (0.249) (0.249) . O thers 1.628 1.665 1.714 1.755 (0.478) (0.482) (0.482) (0.428) TY PE O F NEIGHBOR HOOD .(Squatting) 1.000 1.000 1.000 . High Standing 1.399* 1.301 1.349 Neighborhoods (0.179) (0.197) (0.199) . A verage Standing 1.884” ” 1.829“ * 1.895“ ” Neighborhoods (0.182) (0.185) (0.187) -2 LOG L 1165.6 1153 1125.4 1123.7 CHI-SQUARE 106.3**” 118.9“ “ 116.8**” 118.6” ” DF 22 24 26 27 359 Table 5.4.5. (continued) VARIABLE M odel 15 M odel 16 QU ALITY O F HOUS ING: .(Low: 1-7) 1.000 1.000 . Average: 7-10 0.944 0.944 (0.204) (0.205) . High: 10-12 1.087 1.080 (0.247) (0.247) N U M BER O F BED 0.937 ROOMS (0.049) -2 LOG L 1125.4 1123.7 CHI-SQUARE 116.8“ “ 118.6“ “ D F 26 27 360 Table 5.4.6. The Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Modern Medicine to Treat Diarrhoea. Urban Areas of Zaire, 13 Cities. Fonames/UNICEF Survey, 1987. VARIABLE 1 2 3 4 IN TER C EPT 3.797“ “ 1.381 1.365 0.743 (0.145) (0.737) (1.280) (0.798) EDUCA TIO N 1.074“ * 1.071“ * 1.074 1.069“ * (0 .0 2 4 ) (0.024) (0.215) (0.025) H O U SEH O LD 1.114 1.115 1.156* IN CO M E (0.079) (0.139) (0.084) INTERACTION Edu*H Inc. 1.000 (0.023) AGE O F CH ILD IN M ONTHS: • (36 + ) 1.000 .0-3 0.408“ (0.379) .4-6 1.669 (0.390) .7-11 2.163“ (0.319) .12-35 1.457* (0.226) -2 LOG L 972.796 957.981 957.981 844.281 CHI SQUARE 9.384“ * 11.137*“ 11.137“ 30.040**“ D F 1 2 3 6 ^egend: H in c= Household Income; * •** p < = =6.001; *** p < = 0 .0 1 ; * • p < = 0 .0 5 ; * p < = 0.10 361 Table S.4.6. (continued) VARIABLE 5 6 7 8 IN TER C EPT 0.759 1.053 1.054 0.691 (0.802) (0.833) (0.857) (0.867) EDUCA TIO N 1.069“ * 1.060“ 1.067“ 1.047* (0.025) (0.026) (0.027) (0.028) HO USEHO LD 1.156’ 1.168* 1.168* 1.161* INCOM E (0.084) (0.085) (0.086) (0.087) AGE O F CH ILD : .(36 + ) 1.000 1.000 1.000 1.000 .0-3 0.408“ 0.402“ 0.382“ 0.378“ (0.379) (0.380) (0.387) (0.390) .4-6 1.666 1.684 1.512 1.623 (0.390) (0.391) (0.395) (0.399) .7-11 2.166“ 2.128“ 2.069“ 2.157“ (0.320) (0.320) (0.323) (0.325) .12-35 1.459* 1.502* 1.422 1.437 (0.226) (0.227) (0.230) (0.232) GENDER: .(M ale) 1.000 1.000 1.000 1.000 .Fem ale 0.957 0.952 0.929 0.922 (0.177) (0.178) (0.180) (0.181) BIRTH ORD ER (1st) 1.000 1.000 1.000 .2-3 0.631 0.582 0.600 (0.302) (0.330) (0.335) .4-6 0.733 0.779 0.834 (0.305) (0.384) (0.390) .7 + 0.546* 0.487* 0.505 (0.310) (0.432) (0.438) AG E O F M OTHER: .( < 2 0 yrs) 1.000 1.000 .20-24 1.269 1.242 (0.316) (0.321) .25-34 0.842 0.772 (0.350) (0.355) .3 5 + 1.570 1.447 (0.465) (0.472) PLA C E O F DELIVERY: .(Home) 1.000 .Health 2.068“ * Institution (0.222) - 2 LOG L 844.218 839.716 827.796 817.661 CHI SQUARE 30.102**“ 34.605**“ 41.438**“ 51.574**“ D F 7 10 13 14 362 Table S.4.6. (continued) VARIABLE 9 10 11 12 INTERCEPT 0.664 0.660 0.651 0.517 (0.892) (0.893) (0.890) (0.898) EDUCATION 1.047 1.035 1.043 1.041 (0.028) (0.033) (0.028) (0.029) HO USEHOLD 1.169* 1.176* 1.168’ 1.188* INCOM E (0.088) (0.089) (0.088) (0.088) AGE O F CHILD (in months) .(36 + ) 1.000 1.000 1.000 1.000 .0-3 0.385** 0.390** 0.393** 0.404** (0.393) (0.395) (0.394) (0.395) .4-6 1.630 1.642 1.671 1.614 (0.399) (0.400) (0.400) (0.401) .7-11 2.178** 2.203** 2.198** 2.203** (0.326) (0.328) (0.327) (0.328) .12-35 1.438 1.452 1.433 1.423 (0.232) (0.233) (0.232) (0.233) GEND ER .(M ale) 1.000 1.000 1.000 1.000 .Fem ale 0.917 0.915 0.908 0.917 (0.182) (0.182) (0.182) (0.183) BIRTH ORD ER .(1st) 1.000 1.000 1.000 1.000 .2-3 0.604 0.607 0.609 0.605 (0.342) (0.343) (0.343) (0.343) .4-6 0.839 0.842 0.853 0.855 (0.401) (0.401) (0.402) (0.402) .7 + 0.503 0.497 0.514 0.511 (0.449) (0.449) (0.451) (0.452) AGE AT BIRTH OF CHILD: .(< 2 0 yrs) 1.000 1.000 1.000 1.000 .20-24 1.244 1.250 1.260 1.252 (0.322) (0.323) (0.323) (0.323) .25-34 0.777 0.782 0.765 0.737 (0.356) (0.358) (0.358) (0.360) .3 5 + 1.481 1.496 1.451 1.399 (0.475) (0.476) (0.477) (0.478) .egend: **** p < = 0.001 *** p < = 0 .0 1 ; • • p < = 0 .0 5 ; • p < = 0 .1 0 363 Table 5.4.6. (continued) VARIABLE 9 10 11 12 PLA C E O F DELIVERY: .(Home) 1.000 1.000 1.000 1.000 .Health 2.055*“ 2.050*“ 2.083**“ 1.873“ * Institution (0.223) (0.718) (0.223) (0.230) M ARITAL STATUS: .(M arried M ono) 1.000 1.000 1.000 1.000 .Single 0.947 0.759 0.690 0.673 (0.480) (0.791) (0.592) (0.593) .M arried Polyg. 0.849 0.660 0.844 0.856 . Others* (0.272) (0.461) (0.272) (0.273) 0.991 0.896 0.904 0.853 (0.382) (0.591) (0.328) (0.333) INTERACTION: .(Educ*M ar M onogam .) 1.000 .Single 1.046 (0.137) .M arried Poly 1.051 (0.075) .Others 1.018 (0.087) O C C U PA TIO N A L STATUS O F M O TH ER: .(Housewife) 1.000 1.000 .Student 1.593 1.561 (0.812) (0.815) .W orking 1.335 1.275 (0.334) (0.336) ,Othersb 1.556 1.584 (0.565) (0.565) ACCESS TO M ODERN FACILITY: .(Squatting Neighborhood) 1.000 .High Standing 1.603“ Neighborhoods (0.238) .A ver Standing 1.398 N eighborhoods (0.228) -2 LOG L 817.299 816.773 815.863 811.190 CHI SQUARE 5 1.936'“ * 52.461**“ 53.371**“ 58.044**“ D F 17 20 20 22 ^eeend: (*) This includes: consensual union, divorced, separated, w idow ers, etc.; (") This includes: never worked, inactives, etc.; High Stand. = High Standing; A ver Stand. = A verage Standing. **** p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 364 Table 5.4.6. (continued) VARIABLE 13 14 15 16 IN TER C EPT 0.524 0.669 0.675 0.483 (0.911) (0.927) (0.927) (0.965) EDUCA TIO N 1.038 1.040 1.042 1.040 (0.039) (0.030) (0.030) (0.030) HO USEHO LD 1.188* 1.191* 1.197” 1.211“ INCO M E (0.088) (0.089) (0.090) (0.091) AGE O F C H ILD IN M ONTHS: (36 + ) 1.000 1.000 1.000 1.000 .0-3 0.405** 0.372“ 0.372“ 0.380“ (0.395) (0.402) (0.402) (0.402) .4-6 1.612 1.477 1.482 1.522 (0.401) (0.406) (0.406) (0.407) .7-11 2.200“ 2.153“ 2.156“ 2.228“ (0.328) (0.332) (0.332) (0.334) .12-35 1.419 1.415 1.414 1.450 (0.233) (0.240) (0.240) (0.241) SEX .(M ale) 1.000 1.000 1.000 1.000 .Fem ale 0.919 0.919 0.920 0.921 (0.183) (0.186) (0.186) (0.187) BIRTH ORD ER .(1st) 1.000 1.000 1.000 1.000 .2-3 0.603 0.567 0.558 0.560 (0.343) (0.355) (0.358) (0.359) .4-6 0.858 0.788 0.776 0.766 (0.403) (0.415) (0.417) (0.418) .7 + 0.511 0.462* 0.458 0.454* (0.452) (0.467) (0.468) (0.469) AGE A T BIRTH O F CHILD: .( < 2 0 yrs) 1.000 1.000 1.000 1.000 .20-24 1.251 1.392 1.406 1.418 (0.324) (0.331) (0.332) (0.332) .25-34 0.733 0.766 0.779 0.775 (0.361) (0.364) (0.367) (0.367) .35 + 1.393 1.576 1.636 1.606 (0.479) (0.486) (0.497) (0.498) *egend: **** p < = 0.001 *♦* p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 365 Table 5.4.6. (continued) VARIABLE 13 14 15 16 PLA CE O F DELIVERY: .(At Home) 1.000 1.000 1.000 1.000 .Health 1.874” 1.840” 1.853*” 1.828** Institution (0.232) (0.235) (0.236) (0.237) M A RITA L STATUS: .(M arried M ono) 1.000 1.000 1.000 1.000 .Single 0.674 0.665 0.667 0.647 (0.593) (0.598) (0.597) (0.596) .M arried Polyg 0.858 0.833 0.850 0.844 . Others* (0.274) (0.276) (0.281) (0.282) 0.854 0.825 0.816 0.808 (0.333) (0.337) (0.338) (0.338) OCCU PATION AL STATUS O F M OTHER: .(Housewife) 1.000 1.000 1.000 1.000 .Student 1.538 1.550 1.544 1.541 (0.815) (0.820) (0.820) (0.819) .W orking 1.275 1.245 1.249 1.226 (0.336) (0.339) (0.339) (0.340) .Othersb 1.584 1.609 1.638 1.690 (0.565) (0.570) (0.572) (0.571) A CCESS TO M ODERN FA CILITY: .(Squatting Neighborhood) 1.000 1.000 1.000 1.000 .High Stand. 1.709 1.706** 1.731** 1.707** Neighborhood (0.476) (0.264) (0.236) (0.271) .A ver Stand. 1.295 1.408 1.428 1.399 Neighborhood (0.385) (0.233) (0.236) (0.238) Leeend: H ieh Stand. = H ieh Standing: A ver Stand. = A ***«• p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; • • p < = 0 .0 5 ; * p verage Standing; < = 0 .1 0 366 Table 5.4.6. (continued) VARIABLE 13 14 15 16 INTERACTION . (Educ*Squatting) 1.000 . Ed*H Stand. 0.990 . Ed*Av Stan. (0.071) 1.015 (0.061) IND EX O F QUALITY O F HOUSING UNIT: .(Low: 1-7) 1.000 1.000 1.000 .Average: 0.717 0.718 0.682 7-10 (0.264) (0.264) (0.269) •High: 10-12 0.821 0.810 0.761 (0.327) (0.327) (0.335) N U M BER O F ROOMS 0.977 0.979 (0.063) (0.064) IN D EX O F IM M EDIATE ENVIRONM ENT: .(Low: 1-7) 1.000 .Average: 1.415 10-12 (0.260) .High: 10-12 1.368 (0.291) -2 LOG L 811.069 785.194 785.054 783.264 CHI SQUARE 58.165*"* 62.044*"* 62.184**" 63.974**" D F 24 24 25 27 ^egend: E d*Squatt= Interaction Education & Squatting Neighborhoods; Ed*H Stand. = Interaction Ex ucation & High Standing Neighborhoods; Ed*Av_Stan. = Interaction Education & A verage Standing N eighborhoods; **** p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; * • p < = 0 .0 5 ; • p < = 0 .1 0 367 Table 5.4.7.: The Odds Ratio and Standard Error (in parentheses) for the Logistic Regression of Promptness in Seeking Modern Health Care in case of Diarrhoea. Urban Areas of Zaire, 13 Cities. FONAMES- UNICEF Survey, 1987. VARIABLE I 2 3 4 IN TER C EPT 0.611 0.193 0.276 0.200 (0.156) (0.732) (1.423) (0.735) EDUCATION 1.033 1.027 0.965 1.022 (0.023) (0.023) (0.213) (0.024) HO USEHO LD 1.137 1.095 1.134 INCOM E (0.077) (0.151) (0.077) INTERACTION: 1.007 (Edu*Income) (0.023) USE O F SSS: •(No) 1.000 .Yes 1.038 (0.167) -2 LOG L 889.860 963.460 863.374 854.386 CHI SQUARE 2.307 4.321 4.408 3.835 DF 1 2 3 3 .egend: **** p < = 0 .0 0 1 : *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 ; S S S = Sugar-Salt-Solution. 368 Tabic 5.4.7. (continued) VARIABLE 5 6 7 8 IN TER C EPT 0.177 0.168 0.172 0.114 (0.769) (0.813) (0.814) (0.853) EDUCA TIO N 1.023 1 . 0 2 2 1 . 0 2 1 1.026 (0.025) (0.026) (0.026) (0.028) H O USEHO LD 1.155* 1.138 1.141 1.147* INCO M E (0.080) (0.082) (0.082) (0.083) USE O F SSS: .(No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yes 1.006 0.945 0.942 0.903 (0.175) (0.184) (0.184) (0.186) DEHY DRATION : .(Low Risk) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .High Risk 0.718* 0.728 0.737 0.741 (0 . 2 0 1 ) (0 .2 1 0 ) (0 .2 1 1 ) (0 .2 1 1 ) A G E O F C H ILD (in months) .(36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 1.169 1.171 1.208 (0.580) (0.580) (0.585) .4-6 1.378 1.367 1.403 (0.374) (0.374) (0.376) .7-11 1.407 1.414 1.453 (0.304) (0.304) (0.306) .12-35 1.396 1.395 1.416 (0.257) (0.257) (0.259) GEN D ER .(M ale) 1 . 0 0 0 1 . 0 0 0 .Fem ale 0.903 0.899 (0.178) (0.179) BIRTH ORDER: .(First) 1 . 0 0 0 .2-3 1.637 (0.278) .4-6 1.428 (0.275) .7 + 1.550 (0.290) -2 LO G L 784.616 727.300 726.974 723.462 CHI SQUARE 7.603 8.315 8.641 12.153 D F 4 8 9 1 2 369 Table 5.4.7. (continued) VARIABLE 9 1 0 1 1 1 2 IN TER C EPT 0.091 0.092 0.042 0.043 (0.878) (0.896) (0.947) (0.953) EDUCATION 1 . 0 1 2 1 . 0 1 2 1.015 0.996 (0.028) (0.029) (0.029) (0.035) HO USEHO LD 1.149' 1.149* 1.191** 1.197** INCO M E (0.084) (0.084) (0.086) (0.087) U SE O F SSS: •(No) 1.000 1.000 1.000 1.000 .Yes 0.900 0.900 0.950 0.962 (0.188) (0.188) (0.191) (0.192) DEHYDRATION: .(Low Risk) 1.000 1.000 1.000 1.000 .High Risk 0.734 0.735 0.660* 0.645** (0.213) (0.214) (0 .2 2 0 ) (0.223) AGE O F CHILD: (in months) (36 + ) 1.000 1.000 1.000 1.000 .0-3 1.283 1.283 1.292 1.276 (0.589) (0.589) (0.599) (0.601) .4-6 1.444 1.443 1.419 1.434 (0.378) (0.379) (0.383) (0.384) .7-11 1.444 1.444 1.372 1.362 (0.308) (0.308) (0.312) (0.313) .12-35 1.458 1.458 1.451 1.447 (0.261) (0.261) (0.265) (0.266) GENDER: .(M ale) 1.000 1.000 1.000 1.000 .Fem ale 0.901 0.901 0.938 0.927 (0.181) (0.181) (0.183) (0.184) BIRTH ORDER: .(First) 1.000 1.000 1.000 1.000 .2-3 1.463 1.462 1.834** 1.887** (0.289) (0.289) (0.306) (0.309) .4-6 1.168 1.166 1.591 1.595 (0.330) (0.331) (0.351) (0.352) .7 + 1.623 1.621 2.186* 2.161 (0.408) (0.408) (0.426) (0.429) 370 Table 5.4.7. (continued) VARIABLE 9 1 0 1 1 1 2 AGE O F M O TH ER AT BIRTH: .( < 2 0 yrs) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 1.674 1.675* 1.809 1.899** (0.306) (0.306) (0.316) (0.321) .25-34 1.622 1.625 1.724 1.798 (0.354) (0.355) (0.365) (0.368) .35 + 0.957 0.960 1 . 0 1 0 1.056 (0.479) (0.480) (0.490) (0.493) PLA C E O F DELIVERY: .(At home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institutions 0.984 0.978 0.978 (0.272) (0.277) (0.278) M ARITAL STATUS: .(M arried M ono.) 1 . 0 0 0 1 . 0 0 0 .Single 5.293*** 2.060 (0.514) (0.914) .M arried Polyg. 1.180 0.730 (0.297) (0.599) .Others* 1.509 1.388 (0.297) (0.566) INTERACTION ,(Edu*M arried Mono) 1 . 0 0 0 .Edu*Single 1.182 (0.137) .Edu*M Polyg. 1.082 (0.086) .Edu*Others 1.014 (0.079) -2 LOG L 715.384 715.380 703.038 700.833 CHI SQUARE 17.421 17.425 29.767 31.972* DF 15 16 19 2 2 .eeend: M M = M a m e d M onoeam ouslv: M P olv= M arried Polygamously 371 Table 5.4.7. (continued) VARIABLE 13 14 15 16 IN TER C EPT 0.038 0.018 0.018 0.014*“ (0.957) (1.007) ( 1 .0 2 2 ) (1.046) EDUCATION 1.003 1.004 1.0006 1 . 0 1 0 (0.030) (0.030) (0.049) (0.032) HO USEHO LD I o C 4 1.270*“ 1.271“ 1.283“ INCO M E (0.087) (0.090) (0.090) (0.091) USE O F SSS: (No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yes 0.951 0.966 0.971 0.879 (0.192) (0.196) (0.196) (0.204) DEHY DRATION : .(Low Risk) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .High Risk 0.658* 0.664* 0.654* 0.662* (0 .2 2 1 ) (0.224) (0.225) (0.227) AG E O F CHILD (in months) (36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 1.330 1.614 1.605 1.238 (0.602) (0.610) (0.612) (0.645) .4-6 1.471 1.507 1.495 1.576 (0.385) (0.391) (0.391) (0.408) .7-11 1.369 1.425 1.408 1.413 (0.314) (0.318) (0.320) (0.327) .12-35 1.433 1.468 1.454 1.423 (0.267) (0.271) (0.272) (0.281) SEX .(M ale) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Fem ale 0.924 0.956 0.951 1.006 (0.184) (0.187) (0.188) (0.191) BIRTH ORDER: .(First) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 1.814 1.799* 1.778* 1.838* (0.309)* (0.315) (0.315) (0.318) .4-6 1.564 1.649 1.661 1.695 (0.354) (0.362) (0.363) (0.368) .7 + 2 . 2 1 2 2.349* 2.355* 2.493 (0.431) (0.440) (0.441) (0.454) 372 Table 5.4.7. (continued) VARIABLE 13 14 15 16 AGE O F M O TH ER AT BIRTH: .( < 2 0 years) 1.000 1.000 ’ 1.000 1.000 .20-24 1.812* 1.677 1.664 1 . 6 6 6 (0.319) (0.325) (0.326) (0.328) .25-34 1.704 1.561 1.532 1.507 (0.369) (0.377) (0.377) (0.381) .3 5 + 0.913 0.749 0.731 0.738 (0.496) (0.507) (0.509) (0.516) PLA C E O F DELIVERY: .(At home) 1.000 1.000 1.000 1.000 .Health Institutions 0.982 0.809 0.803 0.772 (0.278) (0.290) (0.291) (0.293) M A RITA L STATUS: .(M arried Mono) 1.000 1.000 1.000 1.000 .Single 4.516** 3.880** 3.915*" 3.604** (0.646) (0.647) (0.647) (0.651) .M arried Polygamously 1.187 1.237 1.238 1.190 (0.298) (0.302) (0.303) (0.305) .Others 1.400 1.230 1.245 1.249 (0.310) (0.315) (0.315) (0.318) O CCU PA TIO N A L STATUS O F M OTHER: .(Housewife) 1.000 1.000 1.000 1.000 .Student 1.815 1.921 1.858 1.815 (0.603) (0.606) (0.607) (0.616) .W orking 1.599* 1.517 1.504 1.612 (0.284) (0.288) (0.289) (0.293) .Others 0.973 1.032 1.048 1.092 (0.555) (0.555) (0.555) (0.559) ACCESS TO M ODERN FA CILITY : .(Squatting) 1.000 1.000 1.000 .High Standing 2.160*** 2.720** 2.169*** N eighborhoods (0.237) (0.496) (0.260) .A verage Standing 2.097*** 1 . 8 6 6 2.108*** Neighborhoods (0.235) (0.451) (0.240) Legend: H ieh Stand. Neighborh. = High Standing Neighborhood; Aver. Stand = A verage Standing 373 Table 5.4.7. (continued) VARIABLE 13 14 15 16 INTERCATION: .(Edu*Squatting) 1 . 0 0 0 .Edu*H igh Standing N. 0.964 (0.070) .Edu*A verage Standing 1 . 0 2 1 N. (0.065) IN D EX O F QU ALITY O F HOUSING: .(Low : index 1-7) 1 . 0 0 0 .Average: 7-10 1 . 2 1 1 (0.279) .High 10-12 1.229 (0.327) -2 LOG L 699.433 685.251 684.541 663.002 CHI SQUARE 3 3.371' 47.554"“ 48.263“ * 48.419*“ D F 2 2 24 26 26 x e e n d : E d*Sauat=Interaction Education & Squatting Neighborhood: E d*H _Stand=Interaction Education & High Standing N eighborhood; Ed*Av_Stan = Interaction Education & A verage Standing Neighborhood; **** p < = 0 .0 0 1 ; *** p < = a 0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 374 Table 5.4.8.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Prevalence of Fever among Children 0-59 Months. Urban Areas __________ of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLE 1 2 3 4 IN TER C EPT 0.544*"“ (0.057) 0.627* (0.277) 0.484 (0.512) 0.576* (0.299) EDUCA TIO N 0.983“ (0.008) 0.984* (0.009) 1.030 (0.078) 0.977“ (0.009) H O USEHO LD INCO M E 0.983 (0.030) 1 . 0 1 1 (0.056) 0.968 (0.032) IN TERACTION: Edu*H_Inc 0.995 (0.008) AGE O F C H ILD (in months) .(36 + ) .0-3 .4-6 .7-11 .12-36 1 . 0 0 0 0.501**“ (0.149) 1.698**” (0.137) 2.230**“ (0.115) 1.594**” (0.076) -2 LOG L CHI SQUARE D F 6514.205 4.067“ 1 6276.967 4.239 2 6276.606 4.599 3 5566.506 138.8— 6 ^egend: **** p = 0 .0 0 1 : p < = 0 .0 1 ; ** p <=0.05; * p < =0.10; H_Inc = Household Inc<sme 375 Table 5.4.8. (continued) VARIABLE 5 6 7 8 IN TER C EPT 0.517“ 0.547“ 0.630 0 . 6 6 6 (0.311) (0.302) (0.307) (0.311) EDUCATION 0.995 0.980“ 0.975*“ 0.975“ (0.017) (0 .0 1 0 ) (0 .0 1 0 ) (0 .0 1 0 ) HO USEHOLD 0.968 0.969 0.972 0.966 INCOM E (0.032) (0.032) (0.032) (0.032) AGE O F CHILD .(36 m onths o r more) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 0.606* 0.501**” 0.500**“ 0.521**“ (0.286) (0.149) (0.149) (0.149) .4-6 2 5 7 9 — 2.374“ * 1.703**“ 1.715**“ (0.284) (0.272) (0.137) (0.139) .7-11 2.545“ “ 2.228**“ 2.232**“ 2.303“ “ (0.228) (0.1150) (0.115) (0.117) .12-36 1.800“ “ 1.589**“ 1.601— 1.628**“ (0.148) (0.076) (0.076) (0.077) INTERACTION: .Edu*Age0_3 0.967 ( - ) (0.043) .Edu*Age4_6 0.932* 0.945 (0.042) (0.040) .E d u * A g e7 _ ll 0.977 ( - ) (0.033) .E du*A gel2_36 0.979 ( - ) (0 .0 2 2 ) GENDER: .(M ale) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .(Fem ale) 1.059 1.056 1.068 (0.065) (0.065) (0.065) BIRTH ORDER: .( 1 st) 1 . 0 0 0 1 . 0 0 0 .2-3 0.851* 0.842* (0.096) (0.098) .4-6 0.834* 0.820“ (0.097) (0.098) .7 + 0.858 0.834* (0.105) (0.107) LEN GTH O F PREGNANCY: . (9 months) 1 . 0 0 0 . < 9 months 0.834 (0.142) -2 LOG L 5563.273 5563.757 5561.841 5425.180 CHI SQUARE 142.0**“ 141.525**“ 143.442**“ 143.069**“ D F 9 8 1 0 1 1 376 Table 5.4.8. (continued) VARIABLE 9 1 0 1 1 1 2 INTERCEPT 0.855 0.873 0.793 0.687 (0.328) (0.332) (0.352) (0.363) EDUCATION 0.974“ 0 .9 7 7 " 0.975“ 0 .9 7 6 " (0 .0 1 0 ) (0 .0 1 0 ) (0 .0 1 0 ) (0 .0 1 0 ) HOUSEHOLD 0.956 0.957 0.958 0.968 INCOM E (0.033) (0.033) (0.033) (0.033) AGE O F CHILD (in months) (36 + ) 1.000 1.000 1.000 1.000 .0-3 0.523"“ 0.520*“ 0.521*“ 0.511“ * (0.150) (0.150) (0.150) (0.150) .4-6 1.710“ * 1.701“ 1.721*“ 1.722*“ (0.139) (0.140) (0.140) (0.140) .7-11 2.305— 2.335*“ 2.336— 2.317*“ (0.117) (0.118) (0.118) (0.118) .12-36 1.625"“ 1.631— 1.632*“ 1.622*“ (0.077) (0.077) (0.077) (0.078) GENDER: .(Male) 1.000 1.000 1.000 1.000 .(Female) 1.079 1.078 1.080 1.087 (0.066) (0.066) (0.066) (0.066) BIRTH ORDER: .( 1 st) 1.000 1.000 1.000 1.000 .2-3 0.843* 0.867 0.872 0.887 (0.098) (0.107) (0.107) (0.109) .4-6 0.816“ 0 .7 6 7 " 0 .7 7 3 " 0 .8 0 4 . (0.099) (0.127) (0.127) (0.130) .7 + 0.830* 0 .7 0 7 " 0 .7 1 1 " 0.746 (0.108) (0.153) (0.153) (0.157) LEN GTH O F PREGNANCY: .(9 months) 1.000 1.000 1.000 1.000 . < 9 months 0.810 0.816 0.814 0.810 (0.145) (0.145) (0.145) (0.145) .e e e n d : Edu*AeeO 3 = Interaction Education & Child’s A ee. etc.: * **• p < = 0 .0 0 1 ; * * • p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 < = 0 .0 5 ; * p < = 0 .1 0 377 Table 5.4.8. (continued) VARIABLE 9 1 0 1 1 1 2 BIRTH W EIGH T (in grams) .( < 2500) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2500-3499 0.786“ 0.790“ 0.781“ 0 .7 8 1 " (0.118) (0.119) (0.119) (0.119) .3500 and + 0.999 1 . 0 0 1 0.992 0.989 (0.124) (0.124) (0.125) (0.125) .Unknown 0.792 0.795 0.840 0.849 (0.148) (0.149) (0.166) (0.167) M O TH ER ’S AGE AT BIRTH: . ( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 0.848 0.848 0.847 (0 . 1 1 2 ) (0 . 1 1 2 ) (0.113) .25-34 1.046 1.041 1.037 (0.131) (0.131) (0.131) .35 + 1.205 1.193 1.185 (0.173) (0.173) (0.174) PLA CE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.113 1.113 (0.124) (0.124) M O TH ER ’S M ARITAL STATUS: .(M arried M ono) 1 . 0 0 0 .Single 0.959 (0.160) .M arried Polyg. 0.971 (0.107) .Others* 1.356“ * (0.115) -2 LOG L 5400.997 5381.847 5378.0 5370.448 CHI SQUARE 156.0*“ * 162.9**" 164.3“ “ 1 71.8"“ D F 14 17 18 2 1 -eeendT1**** d < =0.001: ** p < = 0 .0 1 ; ** p < = 0 .0 $; * p < = 0 . 10 378 Table 5.4.8. (continued) VARIABLE 13 14 15 16 INTERCEPT 0.690 0 . 6 8 6 0 .4 7 9 ' 0 .4 8 9 ' (0.363) (0.363) (0.377) (0.383) EDUCATION 0.968'"" 0.966“ 0.975“ 0 .9 7 1 ' (0 .0 1 2 ) (0 .0 1 1 ) (0 .0 1 1 ) (0.016) H O USEHOLD 0.971 0.971 1.007 1.007 INCOM E (0.033) (0.033) (0.034) (0.034) AGE O F CHILD (in months): •(36 + ) 1.000 1.000 1.000 1.000 .0-3 0.512“ “ 0.517“ “ 0.513“ “ 0.513“ “ (0.150) (0.151) (0.152) (0.152) .4-6 1.729“ “ 1.742“ 1.791“ “ 1.788“ “ (0.140) (0.140) (0.142) (0.142) .7-11 2.310“ “ 2.311“ “ 2.403“ “ 2.407“ “ (0.118) (0.118) (0 . 1 2 0 ) (0 . 1 2 0 ) .12-36 1.630“ “ 1.624“ “ 1.664*“ 1.664“ “ (0.077) (0.078) (0.079) (0.079) GENDER: .(Male) 1.000 1.000 1.000 1.000 .(Female) 1.091 1.083 1.072 1.073 (0.066) (0.066) (0.067) (0.067) BIRTH ORDER: •(1 st) 1.000 1.000 1.000 1.000 .2-3 0.900 0.899 0.906 0.906 (0 . 1 2 0 ) (0 . 1 1 0 ) (0 . 1 1 1 ) (0 . 1 1 1 ) .4-6 0.818 0.822 0.849 0.851 (0.131) (0.131) (0.132) (0.132) .7 + 0.7 5 3 ' 0 .7 5 8 ' 0.812 0.813 (0.157) (0.157) (0.159) (0.159) LENGTH OF PREGNANCY: .(9 months) 1.000 1.000 1.000 1.000 . < 9 months 0.807 0.807 0.816 0.815 (0.145) (0.146) (0.147) (0.147) Legend: p < = 0 .0 0 1 : * •* p < = 0 .0 1 : ** p < =0.05; * P "< = 0.iff- 379 Table S.4.8. (continued) VARIABLE 13 14 15 16 BIRTH W EIGH T (in grams) .(< 2 5 0 0 ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2500-3499 0.780“ 0 .7 8 0 " 0 .7 9 7 ' 0.796* (0.119) (0.119) (0 . 1 2 1 ) (0 . 1 2 1 ) .3500 0.989 0.988 1.018 1.019 (0.125) (0.125) (0.127) (0.127) .Unknown 0.843 0.841 0.324 0.825 (0.167) (0.167) (0.169) (0.169) M O TH ER ’S AGE AT BIRTH: . ( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 0.841 0.839 0 .8 2 0 ' 0 .8 1 9 ' (0.113) (0.113) (0.115) (0.115) .25-34 1.024 1 . 0 0 1 0.990 0.987 (0.132) (0.133) (0.134) (0.134) .354- 1.166 1.126 1.123 1 . 1 2 1 (0.174) (0.175) (0.177) (0.178) PLA CE O F DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.119 1.129 1.074 1.078 (0.124) (0.124) (0.126) (0.126) M O TH ER ’S M ARITAL STATUS .(M arried Mono) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Single 0.883 0.785 0 .6 8 4 ' 0 .6 8 0 ' (0.320) (0 .2 0 0 ) (0.204) (0.205) .M arried Polygam. 0.945 0.962 0.924 0.925 (0.198) (0.107) (0.108) (0.108) .Others 0.903 0 . 8 6 6 0.778 0.798 (0.227) (0.227) (0.229) (0.232) INTERACTION: ,Edu*Single ,Edu*M .arried Polyg ,Edu*Others 1.014 (0.044) 1.004 (0.030) 1.069“ 1.063' 1.060’ 1.055' (0.032) (0.032) (0.032) (0.033) ^eeend: £du*Single = Interaction Education & Single; Edu*M .Poly in te r a c tio n Education & M arried Polygamously; Edu*Others =Interation Education & Others; **** p < = 0 .0 0 1 ; *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 .1 0 380 Table 5.4.8. (continued) VARIABLE 13 14 15 16 OCCU PATION AL STATUS O F M OTHER: .(Housewife) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Student 1.266 1.341 1.336 (0.266) (0.271) (0.271) .W orking 1.256 1.182 1.185 (0.105) (0.106) (0.107) .Othersb 1.364' 1.379* 1.380* (0.182) (0.184) (0.184) ACCESS TO M ODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 .High Standing 1.368— 1.274 Neighborhoods (0.080) (0.148) .A verage Standing 0.579*” * 0.596*“ Neigborhoods (0.093) (0.194) INTERACTION: .Edu*H Stand 0.996 (0.027) .Edu*A Stand 1 . 0 1 2 (0 .0 2 2 ) -2 LOG L 5366.004 5359.275 5270.598 5270.114 CHI SQUARE 176.3*” * 183.0“ “ 271.7*“ * 272.2“ “ DF 24 25 27 29 -egend: High Stand = H gh Standing Neighborhood; A verg Stand = A verage Standing Neighborhood; E du*H _Stand=Interaction & High Standing Neighborhood; Edu*A_Stand in te r a c tio n & A verage Standing Neighborhood. 381 Table 5.4.8. (continued) VARIABLE 17 18 19 2 0 INTERCEPT 0.442“ 0.443“ 0.447“ 0.473* (0.389) (0.400) (0.389) (0.400) EDUCATION 0.983 0.982 0.984 0.987 (0 .0 1 2 ) (0.024) (0.014) (0 .0 1 2 ) HO USEHO LD 1.008 1.008 1.015 1.017 INCOM E (0.035) (0.035) (0.035) (0.035) AGE O F CHILD (in months) .(36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 0.511“ “ 0.511— 0.510“ “ 0.509— (0.154) (0.154) (0.154) (0.154) .4-6 1.813“ “ 1.811“ * 1.807“ “ 1.817— (0.144) (0.144) (0.144) (0.144) .7-11 2.416*“ * 2.425“ “ 2.413“ “ 2.389**“ (0 . 1 2 1 ) (0 . 1 2 1 ) (0 . 1 2 1 ) (0 . 1 2 1 ) .12-36 1.695“ “ 1.695“ “ 1.690**“ 1.687**“ (0.080) (0.080) (0.080) (0.080) G ENDER: .(M ale) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .(Female) 1.073 1.075 1.073 1.067 (0.068) (0.068) (0.068) (0.068) BIRTH ORDER: .( 1 st) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 0.913 0.913 0.904 0.896 (0 . 1 1 2 ) (0 . 1 1 2 ) (0.113) (0.113) .4-6 0.855 0.855 0.847 0.838 (0.134) (0.134) (0.134) (0.134) .7 + 0.821 0.818 0.824 0.819 (0.161) (0.161) (0.161) (0.161) LEN GTH OF PREGNANCY: .(9 months) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . < 9 months 0.845 0.842 0.855 0.855 (0.150) (0.150) (0.150) (0.150) 382 Table 5.4.8. (continued) VARIABLE 17 18 19 2 0 BIRTH W EIGH T (in grams) .(< 2 5 0 0 ) 1.000 1.000 1.000 1.000 .2500-3499 0.858 0.857 0.858 0.860 (0.123) (0.123) (0.123) (0.123) .3500 + 1.066 1.067 1.067 1.075 (0.129) (0.129) (0.129) (0.129) .Unknown 0.876 0.879 0.878 0.883 (0.172) (0.172) (0.172) (0.172) M O TH ER ’S AGE AT BIRTH: . ( < 2 0 years) 1.000 1.000 1.000 1.000 .20-24 0.803' 0 .8 0 4 ' 0 .8 0 7 ' 0 .8 0 8 ' (0.116) (0.116) (0.116) (0.116) .25-34 0.972 0.975 0.985 0.992 (0.135) (0.136) (0.136) (0.136) .35 + 1.131 1.137 1.164 1.179 (0.179) (0.179) (0.180) (0.181) PLACE OF DELIVERY: .(Home) 1.000 1.000 1.000 1.000 .Health Institutions 1.106 1.104 1.113 1.119 (0.127) (0.127) (0.127) (0.128) M O TH ER ’S M ARITAL STATUS: .(M arried Mono) .Single 1.000 1.000 1.000 1.000 .M arried Polyg 0 .6 5 1 " 0 .6 5 0 " 0 .6 6 0 " 0 .6 5 4 " (0.207) (0.207) (0.207) (0.208) .Others* 0.954 0.953 0.971 0.967 (0.109) (0.109) (0 . 1 2 0 ) (0 . 1 1 0 ) 0.766 0.778 0.761 0.765 (0.230) (0.230) (0.230) (0.230) INTERACTION: .Edu*Others 1.057' (0.032) 1.054' 1.058' 1.056’ (0.032) (0.032) (0.032) ^eeend: Edu*Others = Interaction education & other m arital status 383 Table 5.4.8. (continued) VARIABLE 17 18 19 2 0 OCCU PATION AL STATUS OF M OTHER: .(Housewife) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Student 1.239 1.238 1.257 1.265 (0.276) (0.277) (0.277) (0.277) .W orking 1.134 1.143 1.138 1.130 (0.108) (0.108) (0.108) (0.108) .Others 1.435' 1.436* 1.4 6 8 " 1.474" (0.185) (0.186) (0.186) (0.187) ACCESS TO M ODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .High Standing 1.414'**' 1.413**" 1.435**“ 1.421“ “ Neighborhoods (0.081) (0.082) (0.082) (0.082) .A verage Standing 0.654'**' 0.652**" 0.665**“ 0.677**“ Neigborhoods (0 . 1 0 0 ) (0 . 1 0 0 ) (0 . 1 0 1 ) (0 . 1 0 2 ) IND EX O F HOUSING: .(Low: index 1-7) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Average: 7-10 0.976 0.925 0.975 1.004 (0.098) (0.162) (0.098) (0.099) .High :10-12 0 .7 4 4 " 0.849 0.743” 0.783“ (0.117) (0 .2 1 0 ) (0.117) (0 . 1 2 0 ) INTERACTION: .INT13 1 . 0 1 0 (0.027) .IN T14 0.982 (0.031) BEDROOMS: 0.970 0.973 (0 .0 2 2 ) (0 .0 2 2 ) IND EX O F ENVIRONM ENT: .(Low : index 1-7) 1 . 0 0 0 .Average: 7-10 0.898 (0 . 1 0 0 ) .High :10-12 0.787“ (0 . 1 1 1 ) -2 LOG L 5153.342 5151.995 5151.461 5146.392 CHI SQUARE 274.9*"* 276.2**“ 276.8**“ 281.8**“ DF 29 31 30 32 .eeend: In tl3 = Interaction Education & A verage 1 ouslng Index In tl4 = Interaction Education & High Housing Index 384 Table 5.4.9.: Odds Ratios and Standard Error (in parenthesis) for the Logistic Regression of the Use of Modern Health Services for Child Fever. Urban Areas of Zaire, 13 Cities FONAMES/UNICEF Survey, 1987. VARIABLE 1 2 3 4 INTERCEPT 1.332*“ 3.078“ 2.767 3.100“ (0.092) (0.459) (0.816) (0.461) EDUCATION 1.039*“ 1.041*“ 1.062 1.043“ * (0.014) (0.014) (0.128) (0.014) HOUSEHOLD 0.911* 0.921 0.899“ INCOME (0.049) (0.089) (0.050) INTERACTION .Edu*H Inc 0.998 (0.014) DIARRHEA .(No) 1 . 0 0 0 . Yes 1.393*“ (0 . 1 1 1 ) -2 LOG L 2281.952 2193.563 2193.538 2184.437 CHI SQUARE 7.51— 10.39*“ 10.41“ 19.51**“ DF 1 2 3 3 Legend:**** p < = 0 .0 0 k *** p < = 0 .0 1 : ** p < =0.05; * p < = 0 .1 0 ; H In c= Household Income 385 Table 5.4.9. (continued) VARIABLE 5 6 7 8 INTERCEPT 2 .9 6 8 " 3.043" 3.055" 2.900" (0.495) (0.499) (0.511) (0.521) EDUCATION 1.043*" 1.043"* 1.041" 1.040" (0.015) (0.015) (0.016) (0.016) HOUSEHOLD INCOME 0.897" 0.897" 0.898" 0.899" (0.053) (0.053) (0.053) (0.053) DIARRHEA •(No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yes 1.303" 1.304" 1.301" 1.336“ (0.118) (0.118) (0.118) (0.119) AGE OF CHILD (in months): .(36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 0.645* 0.645* 0.646* o.63 r (0.263) (0.262) (0.262) (0.264) .4-6 1.426 1.422 1.430 1.383 (0.233) (0.233) (0.234) (0.237) .7-11 1.573" 1.575" 1.560" 1.543" (0.192) (0.192) (0.193) (0.195) .12-36 1.041 1.041 1.055 1 . 0 2 1 (0.129) (0.129) (0.129) (0.130) GENDER: .(Males) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Females 0.960 0.959 0.959 (0.107) (0.107) (0.108) BIRTH ORDER .( 1 st) 1 . 0 0 0 1 . 0 0 0 .2-3 0.786 0.642" (0.161) (0.179) .4-6 0.909 0.701 (0.163) (0.218) .7 + 0.788 0.535" (0.175 (0.259) -egend:*■*'** p < = 6 .0 0 1 : •** p < = 0 .0 1 : ** p < = 0.05; * p < = 0:i0- " 386 Table 5.4.9. (continued) VARIABLE 5 6 7 8 AGE OF MOTHER AT BIRTH: . ( < 2 0 years) .20-24 .25-34 .35 + 1 . 0 0 0 1.653— (0.184) 1.544“ (0.216) 2.054“ (0.288) -2 LOG L 1980.556 1980.408 1977.251 2004.712 CHI SQUARE 29.9” 30.09” 33.25” 44.25” DF 7 8 1 1 14 -egend:**** p < = 6 .0 0 1 : *** p < = 0 .0 1 : ** p < =6.05; * p < = 0.10 387 Table 5.4.9. (continued) VARIABLE 9 1 0 1 1 1 2 INTERCEPT 1.955 1.978 2.015 1.083 (0.535) (0.554) (0.556) (0.556) EDUCATION 1.029* 1.030* 1.040“ 1.035* (0.017) (0.017) (0.019) (0.018) HOUSEHOLD INCOME 0.896“ 0.893“ 0.889“ 0.895“ (0.054) (0.055) (0.055) (0.055) HAD DIARRHEA .(No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yea 1.374*“ 1.370“ * 1.370*“ 1.366*“ (0 . 1 2 0 ) (0 . 1 2 0 ) (0 . 1 2 1 ) (0 . 1 2 1 ) AGE OF CHILD (in months): .(36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 0.653“ 0.650 0.660 0.684 (0.265) (0.267) (0.266) (0.267) .4-6 1.442 1.448 1.438 1.472 (0.238) (0.239) (0.239) (0.240) .7-11 1.559“ 1.579“ 1.588“ 1.605“ (0.195) (0.196) (0.197) (0.197) .12-35 1.017 1.027 1.019 1.028 (0.131) (0.131) (0.132) (0.132) GENDER: .(Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Female 0.962 0.963 0.957 0.961 (0.108) (0.109) (0.109) (0.109) BIRTH ORDER ( 1 st) 1 . 0 0 0 0.645“ 1 . 0 0 0 1 . 0 0 0 .2-3 0.654“ (0.184) 0.627“ 0.645“ (0.180) 0.705 (0.185) (0.185) .4-6 0.707 (0.223) 0.680* 0.701 (0.219) 0.535“ (0.224) (0.225) .7 + 0.532“ (0.265) 0.522“ 0.537“ (0.260) (0.267) (0.267) Legend:**** p < =6.001: *** p < = 0 .0 1 ; ** p < = 0.05; * p < = 0 .1 0 388 Table 5.4.9. (continued) VARIABLE 9 1 0 1 1 1 2 AGE OF MOTHER AT BIRTH: . ( < 2 0 yeare) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 1.638— 1.604“ 1.625“ * 1.648*“ (0.185) (0.186) (0.186) (0.187) .25-34 1.546“ 1.513* 1.538“ 1.498* (0.217) (0.218) (0 .2 2 0 ) (0 .2 2 1 ) .35 + 2 .0 2 9 " 1.977“ 2 .0 2 0 “ 1.901“ (0.290) (0.291) (0.292) (0.295) PLACE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.729 1.749**“ 1.709**“ 1.740“ “ (0.161) (0.162) (0.163) (0.163) MARITAL STATUS: ,(Married_M) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Single 0.877 0.869 0.629 (0.267) (0.523) (0.336) .Married Polyg. 1.170 1.028 1.152 (0.179) (0.320) (0.179) .Others 1.123 2.310“ 2 .2 2 0 “ (0.184) (0.382) (0.387) INTERACTION: .Edu*Single 0.999 (0.073) .Edu*M_Poly 1.026 (0.051) .Edu*Others 0.892“ 0.882“ (0.052) (0.052) OCCUPATION: .(Housewife) 1 . 0 0 0 .Student 2.465* (0.481) .Working 1.571“ (0.180) .Others 1 . 2 2 1 (0.287) -2 LOG L 1948.994 1947.541 1941.855 1932.611 CHI SQUARE 55.72**“ 57.17— 62.86— 72.10**“ DF 15 18 2 1 2 2 ^eeend: Married M = Married Monogamously; Mar. Poly = Married Polygamoualy; ♦♦♦♦ p < =0.6oi; ♦** p < -0.01; * * p < = 0 .0 5 ; * p < = 0 .1 0 389 Table 5.4.9. (continued) VARIABLE 13 14 15 INTERCEPT 1.251 1.213 1.621 (0.569) (0.578) (0.584) EDUCATION 1.038" 1.044* 1.041“ (0.018) (0.026) (0.019) HOUSEHOLD INCOME 0.920 0.921 0.912* (0.056) (0.006) (0.056) HAD DIARRHEA .(No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yes 1.399*" 1.395“ * 1.392"* (0 . 1 2 2 ) (0 . 1 2 2 ) (0.123) AGE OF CHILD: (in months) .(36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 0.721 0.718 0 . 6 8 8 (0.268) (0.268) (0.272) .4-6 1.526* 1.524* 1.563* (0.241) (0.241) (0.244) .7-11 1.650" 1.658" 1.702"* (0.198) (0.198) (0 .2 0 0 ) .12-35 1.049 1.052 1.041 (0.132) (0.133) (0.134) GENDER: .(Males) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Females 0.970 0.970 0.956 (0 . 1 1 0 ) (0 . 1 1 0 ) (0 . 1 1 1 ) BIRTH ORDER .( 1 st) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 0.660" 0.655" 0.657“ (0.188) (0.187) (0.187) .4-6 0.727 0.720 0.699 (0.226) (0.227) (0.228) .7 + 0.570" 0.563“ 0.546“ (0.269) (0.270) (0.272) .egend: **** p < = 0 .0 0 1 : *** p < = 0 .0 1 : ** p < =6.65; * P <=0.10 Table 5.4.9. (continued) VARIABLE 13 14 15 AGE OF MOTHER AT BIRTH: . ( < 2 0 years) .20-24 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1.533" 1.543" 1.543" .25-34 (0.189) (0.189) (0.190) 1.386 1.399* 1.417" .3 5 + (0.223) (0.224) (0.225) 1.773* 1.785* 1.809" (0.296) (0.297) (0.298) PLACE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.586 1.568*" 1.558"* (0.166) (0.168) (0.169) MARITAL STATUS: .(Married Mono) .Single 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 0.561" 0.557* 0.560’ .Married Poly. (0.339) (0.340) (0.340) 1.153 1.155 1.166 .Others* (0.180) (0.180) (0.181) 2.024* 2.084* 1.991* (0.390) (0.395) (0.387) INTERACTION: .Edu*M arital Status 0.880" 0.876" 0.879" (0.052) (0.054) (0.052) OCCUPATION: .(Housewife) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Student 2.505* 2.484* 2.255* (0.482) (0.482) (0.484) .Working 1.484" 1.486" 1.488" (0.182) (0.182) (0.184) .Others 1.246 1.247 1.245 (0.288) (0.288) (0.289) -egend: *♦♦» p < =0.661: p < =0.01: ** p < =0.05; * p < =6.10 Table 5.4.9. (continued) VARIABLE 13 14 15 ACCESS TO MODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 •High Standing 1.127 1.386 1.061 Neighborhoods (0.158) (0.322) (0.172) .Average Standing 1.512— 1.516* 1.476*“ Neighborhoods (0.130) (0.235) (0.133) INTERACTION: .Edu*H Stand 0.967 Neighborh (0.046) ,Edu*A Stand 1 . 0 0 0 Neighbor (0.035) INDEX O F HOUSING: .(Low: index 1-7) 1 . 0 0 0 .Average: 7-10 0.816 (0.155) •High: 10-12 0.919 (0.194) -2 LOG L 1922.141 1921.534 1889.325 CHI SQUARE 82.57**“ 83.18— 83.64**“ 24 26 26 Table 5.4.9. (continued) VARIABLE 16 17 INTERCEPT 1.650 1.408 (0.585) (0.598) EDUCATION 1.042“ 1.041“ (0.019) (0.019) HOUSEHOLD INCOME 0.917 0.921 (0.057) (0.057) HAD DIARRHEA .(No) 1 . 0 0 0 1 . 0 0 0 .Yes 1.386"* 1.390*“ (0.123) (0.124) AGE OF CHILD (in months) .(36 + ) 1 . 0 0 0 1 . 0 0 0 .0-3 0.689 0.691 (0.272) (0.272) .4-6 1.569* 1.590* (0.244) (0.245) .7-11 1.700“ * 1.739— (0 .2 0 0 ) (0 .2 0 1 ) .12-36 1.040 1.048 (0.134) (0.134) GENDER: .(Males) 1 . 0 0 0 1 . 0 0 0 .Females 0.959 0.959 (0 . 1 1 1 ) (0 . 1 1 1 ) BIRTH ORDER .( 1 st) 1 . 0 0 0 1 . 0 0 0 .2-3 0.654“ 0.656“ (0.187) (0.188) .4-6 0.694 0.685* (0.229) (0.229) .7 + 0.547“ 0.536“ (0.272) (0.273) Legend: **** p < =0.001: *** p < = 0 .0 1 : ** p < =6.05; * p < =0.10" Table 5.4.9. (continued) VARIABLE 16 17 AGE O F MOTHER AT BIRTH: . ( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 .20-24 1.551” 1.556” (0.191) (0.191) .25-34 1.432 1.433 (0.225) (0.225) .3 5 + 1.856” 1 .8 6 6 ” (0.300) (0.301) PLACE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.570*” 1.556— (0.169) (0.170) MARITAL STATUS: .(M arried Mono) 1 . 0 0 0 1 . 0 0 0 .Single 0.565* 0.555* (0.340) (0.340) .Married Polyg. 1.185 1.179 (0.182) (0.182) .Others 1.956* 1.945* (0.388) (0.388) INTERACTION: .Edu*M arital Status 0.881” 0.880” (0.052) (0.052) OCCUPATION: .(Housewife) 1 . 0 0 0 1 . 0 0 0 • Student 2.270* 2.277* (0.485) (0.484) .Working 1.4% ” 1.517” (0.184) (0.185) .Others 1.273 1.289 (0.291) (0.291) leeend: **** o < = 6 .6 0 1 : *** p < = 6 .0 1 ; ** p < = 0.05; * p < = 0 . Table 5.4.9. (continued) VARIABLE 16 17 ACCESS TO MODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 .High Standing 1.074 1.081 Neighborhoods (0.173) (0.174) .Average Standing 1.492” 1.480” Neighborhoods (0.134) (0.134) INDEX O F HOUSING: .(Low: Index 1-7) 1 . 0 0 0 1 . 0 0 0 .Average: 7-10 0.814 0.811 (0.155) (0.157) .High: 10-12 0.917 0.927 (0.194) (0.199) BEDROOMS 0.973 0.975 (0.037) (0.037) INDEX OF ENVIRONMENT: .(Low: Index 1-7) 1 . 0 0 0 .Average: 7-10 1 . 2 1 0 .High: 10-12 (0.154) 1.069 (0.177) -2 LOG L 1888.766 1886.853 CHI SQUARE 84.20“ 8 6 . 1 1 “ DF 27 29 Legend: **** p < =0.001: *** p < = 0 .0 1 ; ** p < I I o a « •o A I I © © Table 5.4.10.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Use of Modern Drugs in Case of Fever Among Mothers whose Children are Aged 0-59 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES MODEL 1 MODEL 2 MODEL 3 MODEL 4 INTERCEPT 9.894*” * 5.718“ “ 42.676“ * 5.727* (0.166) (0.899) (1.445) (0.899) EDUCATION 1.071“ 1.068” 0.692 1.066” (0.027) (0.028) (0.248) (0.028) H INCOM E' 1.061 0.852 1.081 (0.098) (0.156) (0.098) EDUHNCOM E 1 1.048* (0.027) DIARRHEA: • (No) 1 . 0 0 0 . Yes 0.647“ (0.204) -2 LOG L 813.4 787.8 784.7 783.4 CHI-SQUARE 6.5“ 6 .6 “ 9.7“ 11.03“ DF . _ , j . i u . . . n m . r —r r r r 1 .-L. IJ . ----- 1. 2 3 3 Leeend:'H INCO M E= Household Income: '‘Interaction Education and Household Income: p < = 0 .0 0 1 : *** p < = 0.01; ** p < = 0.05; * p < = 0 .1 0 396 Table 5.4.10. (continued) VARIABLES MODEL 5 MODEL 6 M ODEL 7 MODEL 8 INTERCEPT 4.495 4.495 3.135 2.985 (0.966) (0.974) (0.978) (1.014) EDUCATION 1.075“ 1.075“ 1.082“ 1.076“ (0.029) (0.029) (0.031) (0.032) H INCOME* 1.113 1.113 1.094 1.079 (0.103) (0.103) (0.103) (0.105) DIARRHEA: • (No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yes 0.554“ " 0.554"“ 0.551“ * 0.589“ (0 .2 2 0 ) (2 2 0 ) (0 .2 2 0 ) (0.224) CHILD AGE: .0-3 Months 0.347“ * 0.347"“ 0.347“ 0.334“ * (0.407) (0.407) (0.410) (0.414) .4-6 Months 1.098 1.098 1.064 1.098 (0.451) (0.451) (0.453) (0.483) . 7-11 1 . 1 2 1 1 . 1 2 1 1 . 1 0 1 1.095 (0.373) (0.373) (0.374) (0.388) . 12-35 1.079 1.079 1.047 0.965 (0.265) (0.265) (0.266) (0.270) . (36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 1 . 0 0 0 1 . 0 2 2 1.015 (0 .2 1 0 ) (0 .2 1 1 ) (0.215) BIRTH ORDER .( 1 st) 1 . 0 0 0 1 . 0 0 0 . 2-3 1.916“ 1.542 (0.299) (0.335) . 4-6 2.046“ 1.929 (0.300) (0.432) . 7 + 1.642 1.674 (0.307) (0.501) -2 LOG L 705.3 705.3 698.6 678.9 CHI-SQUARE . 21.4“ " 21.4*“ 28.1*“ 31.54“ * DF 7 8 1 1 14 *H INCOME = Household ncoine; **** p < =0.001; *** p < =0.01; ** p < =0.05; * p < = 0 . 1 0 397 Table S.4.10. (continued) VARIABLES M ODEL 8 MODEL 9 M ODEL 10 MODEL 11 MOTHER AGE: . ( < 2 0 years) 1 . 0 0 0 . 20-24 2.233" (0.350) . 25-34 1.256 (0.409) . 35 + 1.290 (0.552) INTERCEPT 1.862 1.244 1.366 (1.027) (1.062) (1.075) EDUCATION 1.059* 1.054 1.024 (0.033) (0.033) (0.037) H INCOME 1 1.074 1.109 1.118 (0.105) (0.108) (0.109) DIARRHEA: • (No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yes 0.612" 0.609" 0.593" (0.226) (0.227) (0.227) CHILD AGE: .0-3 months 0.353" 0.332"* 0.313"* (0.419) (0.422) (0.425) .4-6 1.184 1.188 1 . 2 2 1 (0.486) (0.487) (0.490) . 7-11 1.119 1.109 1.157 (0.390) (0.392) (0.394) . 12-35 0.960 0.943 0.964 (0.271) (0.273) (0.274) . (36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 1.006 1.014 1.018 (0.216) (0.216) (0.217) -2 LOG L 678.9 672.6 668.4 783.4 CHI-SQUARE 31.5— 3 7 .9 "" 42.1"* 11.03" DF 14 15 18 3 H _IN COM E= Household Income; **** p < =0.001; *** p < = 0.01; ** p < =0.05; * p < = 0.10 398 Table 5.4.10. (continued) VARIABLES Model 9 Model 10 Model 11 BIRTH ORDER . ( 1 St.) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 2-3 1.589 1.724 1.739 (0.336) (0.346) (0.347) . 4-6 1.956 2.194’ 2.241* (0.435) (0.444) (0.449) . 7 + 1.695 1.959 1.956 (0.505) (0.516) (0.520) MOTHER AGE: . ( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 20-24 2.169“ 2.131“ 2.156“ (0.351) (0.354) (0.353) . 25-34 1.235 1.215 1.225 (0.411) 0.418) (0.421) . 35+ 1 . 2 2 1 1.165 1.214 (0.557) (0.562) (0.567) PLACE OF DELIVERY: . (At Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Health Institutions 2.023“ * 2 .0 1 2 * “ 1.928“ (0.268) (0.271) (0.275) M OTHER’S MARITAL STATUS: . (Married Monog) 1 . 0 0 0 1 . 0 0 0 . Single 1.137 0.854 (0.479) (0.884) .Married Polygamously 0.982 0.374“ (0.343) (0.489) . Others 2.299 2.118 (0.451) (0.880) INTERACTION:' . ( IN TO ) 1 . 0 0 0 . INT2 1.056 (0.130) . INT3 1.289“ (0 . 1 1 2 ) . INT4 1 . 0 2 1 (0.125) -2 LOG L 672.549 668.387 662.382 CHI-SQUARE 37.9“ “ 42.05“ “ 48.056“ “ DF 15 18 21 INTO=education*Married Monogamy; INfl'2= Education*S»gle; IN I'S = Education*Marricd Polygamously; INT4=Education*Others; **** p < = 0.001; *** p < = 0.01; ** p < = 0 .0 5 ; * p < = 0 .1 0 399 Table 5.4.10. (continued) VARIABLES MODEL 12 MODEL 13 MODEL 14 MODEL 15 INTERCEPT 1.267 0.535 0.594 0.985 (1.069) (1.088) (1.097) ( 1 . 1 2 1 ) EDUCATION 1 . 0 2 2 1.015 1 . 0 0 1 1.034 (0.035) (0.035) (0.043) (0.037) H INCOME 1.119 1 .2 0 2 “ 1.196“ 1 .2 0 1 ” (0.108) (0.109) (0.109) (0.109) DIARRHEA: • (No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yea 0.594“ 0.587“ 0.587“ 0.609“ (0.228) (0.230) (0.230) (0.233) AGE OF CHILD . 0-3 Months 0.323” “ 0.339“ 0.330“ ” 0.312“ ” (0.425) (0.428) (0.430) (0.431) . 4-6 1.256 1.375 1.359 1 . 2 2 1 (0.492) (0.497) (0.497) (0.503) . 7-11 1.176 1.190 1.197 1.118 (0.395) (0.398) (0.399) (0.402) . 12-35 0.963 0.998 0.995 0.939 (0.274) (0.276) (0.276) (0.281) . (36 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 CHILD GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 1.026 1.043 1.043 1.031 (0.218) (0.219) (0.219) (0 .2 2 2 ) BIRTH ORDER: . ( 1 at. ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 2-3 1.784” 1.852” 1.876” 1.863” (0.346) (0.350) (0.351) (0.356) . 4-6 2.284” 2.403“ 2.454“ 2.520“ (0.449) (0.449) (0.450) (0.464) . 7 + 1.986 2.069 2.098 2.372 (0.521) (0.520) (0.520) (0.542) M OTHER’S AGE: . ( < 2 0 years ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 20-24 2.177“ 1.986“ 1.970* 2.104“ (0.354) (0.359) (0.359) (0.367) . 25-34 1 . 2 1 2 1.094 1.077 1.035 (0.421) (0.424) (0.426) (0.434) . 35 + 1.185 1.096 1.076 1 . 0 2 0 (0.570) (0.570) (0.572) (0.584) Legend: H Income =Householc Income; **** p < = 0 .0 0 1 ; * * * p < = 0 .(>1; * * p < = 0 .0 5 ; * i 3 < = 0 .1 0 400 Table 5.4.10. (continued) VARIABLES MODEL 12 MODEL 13 MODEL 14 MODEL 15 PLACE O F DELIVERY . (At Home) 1.000 1.000 1.000 1.000 . Health Institutions 1.939“ 1.595* 1.614* 1.662* (0.276) (0.281) (0.283) (0.289) MARITAL STATUS: . (Married Mono.) 1.000 1.000 1.000 1.000 . Single 0.780 0.700 0 . 6 8 6 0.801 (0.620) (0.611) (0.612) (0.628) . Married Polyg. 0.367“ 0.350“ 0.354“ 0.377“ (0.487) (0.493) (0.491) (0.499) . Others 2.069 1.735 1.684 1.739 (0.464) (0.473) (0.475) (0.483) INTERACTION: ,(Edu*Married_M) 1.000 1.000 1.000 1.000 . Edu*Married Pol 1.291“ 1.302“ 1.300“ 1.284“ (0.113) (0.114) (0.113) (0.113) M OTHER’S OCCUPATION: . (Housewife) 1.000 1.000 1.000 1.000 . Student 3.972 3.544 3.386 2.904 (1.136) (L124) (1.124) (1.127) . Working 1.532 1.430 1.460 1.338 (0.392) (0.396) (0.397) (0.398) . Others 1.323 1.324 1.335 1.216 (0.580) (0.568) (0.570) (0.584) TYPE OF NEIGHBORHOOD: .(Squatting N.) 1.000 1.000 1.000 . High Standing 2.450“ 2.854 2.333“ Neighborhoods (0.349) (0.727) (0.375) . Average Standing 2.399“ “ 1.822 2.441“ * Neighborhoods (0.264) (0.425) (0.272) INTERACTION :* . (INTO) . INT5 . INT 6 1.000 0.978 (0.104) 1.057 (0.069) * ’ INTO=Educ*Squatting; INT5=Educ*High Standing Neighborhoods; 1NT6 = duc’ Average Stand mg Neighborhoods 401 Table 5.4.10. (continued) VARIABLES MODEL 15 MODEL 16 MODEL 17 INDEX OF QUALITY OF HOUSING: .(Low: index 1-6) 1 . 0 0 0 . Average: 7-9 0.402” (0.349) . High: 10-12 0.514 (0.455) INTERCEPT 0.832 0.879 (1.129) (1.158) EDUCATION 1.029 1.025 (0.037) (0.037) HOUSEHOLD INCOME 1.186 1.178 (0 . 1 1 0 ) (0 . 1 1 0 ) HAD DIARRHEA: • (No) 1 . 0 0 0 1 . 0 0 0 . Yes 0.622“ 0.614" (0.234) (0.234) CH ILD ’S AGE: . 0-3 months 0.303” 0.299” (0.431) (0.432) . 4-6 months 1.176 1.133 (0.503) (0.503) .7-11 months 1 . 1 1 0 1.078 (0.402) (0.853) . 12-35 months 0.933 1.078 (0.282) (0.404) . ( 3 6 + ) 1 . 0 0 0 1 . 0 0 0 GENDER OF CHILD: . (Male ) 1 . 0 0 0 1 . 0 0 0 . Female 1.015 1 . 0 2 2 (0 .2 2 2 ) (0 .2 2 2 ) BIRTH ORDER: . ( 1 St. ) 1 . 0 0 0 1 . 0 0 0 . 2-3 1 .8 8 6 * 1.876* (0.357) (0.359) . 4-6 2.576" 2.665“ (0.465) (0.467) . 7 + 2.330 2.420 (0.542) (0.545) Table 5.4.10. (continued) VARIABLE MODEL 16 MODEL 17 AGE OF MOTHER: . ( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 . 20-24 years 2.083” 2.073“ (0.369) (0.370) . 25-34 years 0.988 0.979 (0.435) (0.437) . 35 + 0.924 0 . 8 8 6 (0.587) (0.589) PLACE O F DELIVERY . ( At Home ) 1 . 0 0 0 1 . 0 0 0 . Hospital 1.597 1.604 (0.289) (0.290) MARITAL STATUS OF MOTHER: .(Married Monog.) . Single 1 . 0 0 0 1 . 0 0 0 0.792 0.789 . Married Polyg. (0.624) (0.630) 0.340” 0.348” . Others (0.505) (0.506) 1.756 1.816 (0.481) (0.483) INTERACTION: . (Educ*Married_M) 1 . 0 0 0 1 . 0 0 0 . Educ*Married_Poly 1.293“ 1.297“ (0.114) (0.114) M OTHER’S OCCUPATION . (Housewife) 1 . 0 0 0 1 . 0 0 0 . Student 2.732 2.780 (1.129) (1.136) . W orking 1.285 1.267 (0.399) (0.401) . Others 1.059 1.066 0.584) (0.588) TYPE OF NEIGHBORHOOD .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 . High Standing 2.247“ 2.182” Neighborhoods (0.377) (0.380) .Average Standing N. 2.363“ * 2.451*“ (0.275) (0.277) 403 Table 5.4.10. (continued) VARIABLE M ODEL 16 MODEL 17 QUALITY OF HOUSING: . (Low: index 1-6) 1 . 0 0 0 1 . 0 0 0 . Average: 7-9 0.407** 0.385*** (0.349) (0.359) . High: 10-12 0.516 0.464 (0.456) (0.470) NUMBER OF BEDROOMS 1.135 1.133 (0.080) (0.081) QUALITY OF IMMEDIATE ENVIRONMENT: . (Low: index 1-6) 1 . 0 0 0 . Average: 7-9 0.993 (0.313) . High: 10-12 1.366 (0.369) -2 LOG L 628.5 627.1 CHI-SQUARE 73.4**** 74.8**** DF 27 29 404 Table 5.4.11.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Promptness in Seeking Health Care in Case of Fever. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLE MODEL 1 MODEL 2 MODEL 3 INTERCEPT 1.362” 1.394 0.987 (0.117) (0.578) (1.091) EDUCATION 0.999 0.994 1.054 (0.017) (0.017) (0.160) HOUSEHOLD INCOME 1 . 0 0 2 1.040 (0.062) (0.119) INTERACTION 0.994 .(Edu*H_Inc) (0.017) -2 LOG L 1458.712 1388.824 1388.686 CHI SQUARE 0 . 0 0 1 0.141 0.279 DF 1 2 3 -egend:**** p < =0.001: *** p < = 0 .6 l: ** p < =0.05: * p < = 0 .1 0 ; H_Inc = Household Income. 405 Table 5.4.11. (continued) VARIABLE MODEL 4 MODEL 5 MODEL 6 INTERCEPT 1.361 1.670 1.999 (0.579) (0.621) (0.628) EDUCATION 0.991 0.987 0.990 (0.017) (0.018) (0.018) HOUSEHOLD INCOME 1.017 0.989 0.984 (0.063) (0.066) (0.066) DIARRHEA • (No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yea 0.752" 0.797 0.800 (0.133) (0.143) (0.144) AGE OF CHILD (in months) .(36+ ) 1 . 0 0 0 1 . 0 0 0 .0-3 1.181 1.199 (0.379) (0.380) .4-6 0.824 0.794 (0.269) (0.271) .7-11 0.936 0.938 (0 .2 2 2 ) (0 .2 2 2 ) .12-36 1.136 1.128 (0.166) (0.167) GENDER: .(Male) 1 . 0 0 0 .Female 0.756“ (0.134) -2 LOG L 1384.264 1265.290 1260.915 CHI SQUARE 4.701 5.705 10.080 DF 3 7 8 Cegend:**** p < = 0 .0 0 1 : *** p < = 0 .0 1 ; ** p < = 0 .0 5 ; * p < = 0 . 1 0 406 Table 5.4.11. (continued) VARIABLE MODEL 7 MODEL 8 M ODEL 9 INTERCEPT 2.005 2.496 1.955 (0.644) (0.657) (0.677) EDUCATION 1 . 0 0 0 1.004 0.999 (0.019) (0 .0 2 0 ) (0 .0 2 0 ) HOUSEHOLD INCOME 0.967 0.962 0.959 (0.067) (0.067) (0.067) DIARRHEA . No 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yes 0.796 0.797 0.812 (0.144) (0.144) (0.145) AGE OF CHILD (in months) • (36+ ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 1.173 1.181 1.219 (0.382) (0.384) (0.385) .4-6 0.842 0.860 0.879 (0.273) (0.274) (0.275) .7-11 0.960 0.969 0.978 (0.223) (0.224) (0.224) .12-36 1.145 1.153 1.152 (0.168) (0.168) (0.168) GENDER: .(Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Female 0.763“ 0.754“ 0.750“ (0.135) (0.135) (0.136) BIRTH ORDER: .( 1 st) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 0.940 1.076 1.078 (0.193) (0.208) (0.208) .4-6 1.153 1.445 1.457 (0.195) (0.249) (0.250) .7 + 1.325 1.656 1.681 (0.218) (0.315) (0.316) MOTHER 'S AGE AT BIRTH OF CHILD: .(< 2 0 years ) 1 . 0 0 0 1 . 0 0 0 .20-24 0.678* 0 . 6 6 8 (0.228) (0.228) .25-34 0.647 0.643* (0.267) (0.267) .35 + 0.674 0.654 (0.363) (0.364) Legend:**** n < =0.001: p < = 0 .0 1 : ♦♦ p < = 0.05: * p < = 0.10 407 Table 5.4.11. (continued) VARIABLE M ODEL 7 MODEL 8 M ODEL 9 PLACE OF DELIVERY: .(Home) 1 . 0 0 0 .Health Institutions 1.401 (0.224) -2 LOG L 1257.577 1254.265 1251.997 CHI SQUARE 13.419 16.730 18.998 DF 1 1 14 15 .egend:**** p < =6.661: +*+ p < =6.61: +* o < =6.657 • p < = 0 . 1 0 408 Table 5.4.11. (continued) VARIABLE MODEL 10 M ODEL 11 MODEL 12 INTERCEPT 1.262 1.271 1.373 (0.701) (0.704) (0.708) EDUCATION 1 . 0 0 0 1.008 1 . 0 2 1 (0 .0 2 0 ) (O.Q23) (0 .0 2 2 ) HOUSEHOLD INCOME 0.981 0.978 0.965 (0.069) (0.069) (0.070) DIARRHEA • (N o ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Yes 0.810 0.808 0.822 (0.146) (0.146) (0.147) AGE OF CHILD (in months): .(36+ ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 1.145 1.160 1.118 (0.386) (0.388) (0.388) .4-6 0.857 0.860 0.844 (0.277) (0.278) (0.278) .7-11 0.936 0.955 0.933 (0.227) (0.228) (0.228) .12-35 1.127 1.115 1.106 (0.169) (0.170) (0.170) GENDER: .(Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Female 0.773" 0.771* 0.764“ (0.137) (0.137) (0.137) BIRTH ORDER: .( 1 st) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 1 . 2 0 0 1.173 1.169 (0.215) (0.216) (0.217) .4-6 1.684“ 1 .6 6 6 “ 1.659" (0.257) (0.259) (0.260) .7 + 1.996“ 1.970“ 1.972“ (0.323) (0.324) (0.326) AGE OF MOTHER AT C H ILD ’S BIRTH: .(Less than 20 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 0.692 0.706 0.681 (0.231) (0.232) (0.234) .25-34 0.661 0.669 0.660 (0.270) (0.271) (0.274) .35 + 0.665 0.687 0.691 (0.226) (0.368) (0.372) Legend:**** p < =0.001: ♦♦♦ o < = 0.01: «» p < =0.05: * p < = 0.10 409 Table 5.4.11. (continued) VARIABLE MODEL 10 MODEL 11 M ODEL 12 PLACE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1.373 1.329 1.372 (0.226) (0.228) (0.227) MARITAL STATUS OF MOTHER: .(M arried Mono) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Single 2.0 7 1 " 2.407 2.352* (0.355) (0.743) (0.449) .Married Poly 1.241 0.830 1.271 (0 .2 2 1 ) (0.408) (0 .2 2 2 ) .Others* 1.572" 3.338"* 3.422*" (0 .2 2 2 ) (0.455) (0.453) INTERACTION .(Edu*Married Mono) 1 . 0 0 0 1 . 0 0 0 .Edu*Single 0.976 (0 . 1 0 0 ) .Edu*Married Poly 1.075 (0.061) ,Edu*Others 0.889* 0 . 8 8 8 * (0.061) (0.061) OCCUPATIONAL SATTUS OF MOTHER: .(Housewife) 1 . 0 0 0 .Student 0.552 (0.526) .Working 0.745 (0 .2 0 1 ) .Others* 1 1.172 (0.363) -2 LOG L 1243.973 1237.732 1235.201 CHI SQUARE 27.022* 33.263" 35.974" DF 18 2 1 2 2 .eeend:**** d < = 0 .0 0 l: 4"** d < = 0 .0 1 : ** d < = 0 .0 5 : * d < = 0 .1 0 410 Table 5.4.11. (continued) VARIABLE MODEL 13 MODEL 14 MODEL 15 INTERCEPT 0.780 0.800 0.772 (0.732) (0.744) (0.750) EDUCATION 1 . 0 2 0 1 . 0 1 2 1.025 (0 . 0 2 2 ) (0.033) (0.023) HOUSEHOLD INCOME 1.007 1.005 1.009 (0.071) (0.071) (0.072) DIARRHEA .(No) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Yes 0.804 0.819 0.776* (0.149) (0.150) (0.151) AGE OF CHILD (in months) .(3 6 + ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .0-3 1.174 1.188 1.436 (0.391) (0.391) (0.410) .4-6 0.887 0.898 0.905 (0.281) (0.282) (0.283) .7-11 0.958 0.952 0.960 (0.230) (0.230) (0.233) .12-35 1.142 1.143 1.173 (0.172) (0.172) (0.175) GENDER: .(Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Female 0.776* 0.777* 0.779* (0.138) (0.139) (0.140) BIRTH ORDER: .( 1 st) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .2-3 1.214 1.234 1.214 (0.219) (0 .2 2 0 ) (0 .2 2 0 ) .4-6 1.765“ 1.794“ 1.874“ (0.263) (0.264) (0.266) .7 + 2.078“ 2.097“ 2.287“ (0.329) (0.331) (0.337) AGE OF MOTHER AT CH ILD ’S BIRTH: .(Less than 20 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 0.609 0.606 0.573“ (0.239) (0.240) (0.242) .25-34 0.578 0.578 0.550“ (0.279) (0.280) (0.283) .3 5 + 0.618 0.629 0.552 (0.375) (0.378) (0.381) ^eeend :*"*** d < = 0 .0 0 1 : **♦ d < = 0.01; * * p < = 0 .0 5 ; * p < = 0 . 1 0 411 Table 5.4.11. (continued) VARIABLE MODEL 13 MODEL 14 MODEL 15 PLACE O F DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Health Institution 1.223 1.247 1.231 (0.231) (0.232) (0.232) MARITAL STATUS OF MOTHER: .(Married Mono) .Single 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 2.088 2.163 2.091 .Married Polyg. (0.450) (0.452) (0.452) 1.325 1.329 1.310 .Others* (0.224) (0.224) (0.224) 3.140“ 2.857 3.380“ * (0.454) (0.461) (0.460) INTERACTION ,(Edu*Married Mono) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Edu*Single .Edu*Married Polyg. .Edu*Others 0.891' 0.905 0.893* (0.061) (0.062) (0.062) OCCUPATIONAL STATUS OF MOTHER: .(Housewife) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .Student 0.551 0.554 0.466 (0.528) (0.528) (0.542) •Working 0.721 0.716 0.713 (0.204) (0.204) (0.209) .Othersb 1.247 1.232 1.186 (0.362) (0.362) (0.366) ACCESS TO MODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .High Standing 1.875*“ 1.073 1.807“ Neighborhoods (0.206) (0.414) (0.224) .Average Standing 1.631“ * 1.781 1.621“ Neighborhoods (0.162) (0.299) (0.166) INTERACTION: .(Edu*Squatting) .Edu*High Standing Neighborhoods .Edu*Average Standing Neighborhoods 1 . 0 0 0 1.092 (0.058) 0.986 (0.042) ^egend:***1 * p < =0.001: p < =0.01: ♦♦ p < =0.05: ♦ p < = 0.10 412 Table 5.4.11. (continued) VARIABLE MODEL 13 MODEL 14 MODEL 15 INDEX OF QUALITY OF HOUSING: .(Low: Index 1-7) 1 . 0 0 0 .Average: 7-10 0.937 (0.192) .High: 10-12 1.081 (0.237) -2 LOG L 1 2 2 2 . 0 2 1 1218.613 1190.301 CHI SQUARE 48.974“ * 52.382*“ 53.643“ * DF 24 26 26 413 Table 5.4.11. (continued) VARIABLE M ODEL 16 M ODEL 17 INTERCEPT 0.757 0.622 (0.751) (0.771) EDUCATION 1.024 1.024 (0.023) (0.023) HOUSEHOLD INCOME 1.005 1.016 (0.072) (0.073) DIARRHEA • (No) 1 . 0 0 0 1 . 0 0 0 . Yes 0.779* 0.778*** (0.152) (0.152) AGE OF CHILD (in months) .(3 6 + ) 1 . 0 0 0 1 . 0 0 0 .0-3 1.439 1.490 (0.410) (0.410) .4-6 0.902 0.917 (0.284) (0.285) .7-11 0.962 0.990 (0.233) (0.235) .12-36 1.177 1.190 (0.175) (0.176) GENDER: .(Male) 1 . 0 0 0 1 . 0 0 0 .Female 0.779* 0.779* (0.140) (0.141) BIRTH ORDER: •(1 st) 1 . 0 0 0 1 . 0 0 0 .2-3 1.224 1.232 (0 .2 2 1 ) (0 .2 2 2 ) .4-6 1.885** 1.839** (0.266) (0.267) .7 + 2.277** 2.186** (0.337) (0.338) AGE OF MOTHER AT CH ILD ’S BIRTH: .(Less than 20 yeaxs) 1 . 0 0 0 1 . 0 0 0 .20-24 0.573** 0.577** (0.242) (0.243) .25-34 0.548** 0.554** (0.283) (0.284) .35 + 0.546 0.557 (0.383) (0.384) Table 5.4.11. (continued) VARIABLE MODEL 16 MODEL 17 PLACE OF DELIVERY: .(Home) 1 . 0 0 0 1 . 0 0 0 .Health Institutions 1 . 2 2 1 1.190 (0.233) (0.234) MARITAL STATUS OF MOTHER: .(M arried Mono) 1 . 0 0 0 1 . 0 0 0 .Single 2.076 2.030 (0.452) (0.453) .Married Polyg. 1.292 1.264 (0.226) (0.227) .Others* 3.419*“ 3.319“ * (0.461) (0.460) INTERACTION: .Edu*Others 0.891* 0.892* (0.062) (0.062) OCCUPATIONAL STATUS OF MOTHER: .(Housewife) 1 . 0 0 0 1 . 0 0 0 .Student 0.468 0.473 (0.542) (0.544) .Working 0.710 0.720 (0.209) (0 .2 1 0 ) ,Othersb 1.168 1 . 2 2 0 (0.367) (0.368) ACCESS TO MODERN FACILITY: .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 .High Standing 1.791*“ 1.840“ * Neighborhoods (0.225) (0.227) .Average Standing 1.608“ * 1.565“ * Neighborhoods (0.167) (0.168) INDEX OF QUALITY OF HOUSING: .(Low: Index 1-7) 1 . 0 0 0 1 . 0 0 0 .Average: 8-9 0.944 0.968 (0.193) (0.196) .High: 10-12 1.088 1.157 (0.237) (0.243) Table 5.4.11. (continued) VARIABLE MODEL 16 MODEL 17 NUMBER OF BEDROOMS 1 . 0 2 2 1.026 (0.048) (0.048) INDEX O F TH E OUALITY OF IMMEDIATE ENVIRONMENT: . (Low: index 1-6) 1 . 0 0 0 . Average: 7-9 1 . 2 0 2 (0.196) . Hieh: 10-12 0.908 (0.227) -2 LOG L 1190.091 1187.089 CHI SQUARE 53.853“ " 56.854“ " DF 29 Table 5.5.3.: Odds Ratio and Standard Error (in parenthesis) for Logistic Regression of Completeness of Immunization Among Children 12-23 Months. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES MODEL I MODEL 2 MODEL 3 MODEL 4 INTERCEPT 7.844“ 2.181 13.719" 2.301 (0.170) (0.830) (1.569) (0.978) EDUCATION 1.023 1.013 0.730 1.008 (0.025) (0.026) (0.242) (0.031) H INCOME1 1.158 0.946 1.143 (0.091) (0.171) (0.106) EDU*INCOMEl 1.036 (0.026) CHILD’S AGE (in months) .(12-17) . 18-23 1 . 0 0 0 1.436’ (0.216) -2 LOG L 8 8 8 . 2 850.4 848.6 636.6 CHI-SQUARE 0.795 3.2 5.04 4.712 DF 1 2 3 3 Legend: ‘h INCOME = Household Income-. 'Interaction Education and Household Income'. *♦♦♦ p < =0.001'. o < =0.01: ■p <=0.05; *p <=0.10; 417 Table 5.5.3. (continued) VARIABLES MODEL 5 MODEL 6 MODEL 7 MODEL 8 INTERCEPT 1.990 2.060 3.202 3.542 ( 1 .0 1 ) (1.045) (1 . 1 2 2 ) ( 1 .0 2 2 ) EDUCATION 1.008 1.003 1.019 1 . 0 2 2 (0.031) (0.032) (0.033) (0.033) H INCOME1 1.161 1.170 1.159 1.158 (0.108) (0.108) (0 . 1 1 2 ) (0.113) CHILD’S AGE (in months) • (12-17) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 18-23 1.402 1.404 1.405 1.404 (0.217) (0.217) (0.219) (0 .2 2 0 ) GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 1.031 1.033 1.034 1.032 (0.217) (0.217) (0.219) (0.219) BIRTH ORDER . ( 1 st. ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 2-3 0.913 0.912 0.984 (0.335) (0.337) (0.369) . 4-6 1.139 1.195 1.249 (0.351) (0.354) (0.440) . 74- 0.741 0.821 0.810 (0.387) (0.351) (0.487) TYPE OF NEIGHBORHOOD .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 .High Standing 0.347“ “ 0.350“ “ Neighborhoods (0.284) (0.285) .Average Standing 0.663 0.671 Neighborhoods (0.289) (0.290) -2 LOG L 636.571 629.233 614.336 613.444 CHI-SQUARE 4.712 6 . 8 8 2 1 .8 ~ 22.29“ DF 3 7 9 1 2 ^eeend: ‘H INCOME = Household Income; High Stand. Neighbor. = High Standing Neighborhood: Avrg Stand. N eighbors Average Standing Neighborhood; **** p < =0.001; *** p < =0.01; **p < = 0 .0 5 ; * p < = 0 .1 0 418 Table 5.5.3. (continued) VARIABLES MODEL 8 MODEL 9 MODEL 10 MODEL 11 MOTHER AGE: (in years) .( < 2 0 ) 1 . 0 0 0 . 20-24 0.768 (0.402) . 25-34 0.867 (0.464) . 35 + 0.944 (0.568) INTERCEFT 3.897 4.276 4.098 (1.180) (1.191) (1.191) EDUCATION 1 . 0 2 2 1.046 1.059 (0.033) (0.040) (0.036) H INCOME1 1.154 1.131 1.139 (0.115) (0.116) (0.116) CHILD’S AGE: On months) . (12-17) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 18-23 1.403 1.442' 1.440 (0 .2 2 0 ) (0 .2 2 2 ) (0 .2 2 2 ) GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 1.028 1.028 1.033 (0.219) (0 .2 2 2 ) (0 .2 2 1 ) -2 LOG L 613.444 613.272 602.5 600.6 CHI-SQUARE 22.29" 22.46' 33.2" 35.2“ DF 1 2 15 18 19 H INCOME = Household Income; **** p < =0.001; *** p < =0.01; **p < = 0 .0 5 ; * p < = 0 .1 0 419 Table 5.5.3.(continued) VARIABLES Model 9 Model 10 Model 11 BIRTH ORDER . ( 1 St.) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 2-3 0.943 0.861 0.813 (0.385) (0.393) (0.397) . 4-6 1.194 1.095 1.044 (0.457) (0.466) (0.469) . 7+ 0.775 0.715 0.700 (0.505) (0.513) (0.515) TYPE OF NEIGHBORHOODS: .(Squatting Neiboighoods) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . High Standing 0.353“ " * 0.323“ “ 0.322“ “ Neighborhoods (0.286) (0.290) (0.290) . Average Standing 0.679 0.644 0.624 Neighbortioods (0.294) (0.296) (0.297) MOTHER AGE: . (< 20 Years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 20-24 0.755 0.803 0.768 (0.404) (0.409) (0.411) . 25-34 0.850 0.940 0.904 (0.468) (0.473) (0.477) . 35 + 0.920 1.016 0.967 (0.572) (0.574) (0.583) MARITAL STATUS OF MOTHER: . (Married Monogamously) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Single 0.828 0.381 1.375 (0.466) (0.765) (0.595) .Married Polygamously 1.030 0.851 1.004 (0.399) (0.709) (0.400) . Others'* 1 0.963 7.619“ 9.290“ (0.354) (0.958) (0.961) INTERACTION: . (INTO ) 1 . 0 0 0 1 . 0 0 0 . INT2 1.142 — (0 . 1 2 0 ) — . INT3 1.035 — (0 . 1 1 2 ) — . INT4 0.742“ 0.736“ (0.113) (0.113) Legend: INTO =Interaction Education & Married Monogamouslv: INT2 = Interaction Education & Single; 1NT3 = Interaction Education Sc Married Polygamously; INT4 = Interaction Education & Other MariUil Status. 420 Table S.5.3. (continued) VARIABLES MODEL 12 MODEL 13 MODEL 14 INTERCEPT 7.455 11.859 14.168 (1.250) (1.298) (1.334) EDUCATION 1.058 1.062 1.067 (0.039) (0.039) (0.040) H INCOME 1.108 1.098 1.095 (0.118) (0.118) (0.119) CHILD’S AGE: . (12-17 months) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 18-23 1.363 1.352 1.352 (0.229) (0.229) (0.229) CHILD GENDER: . (Male) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . Female 0.967 0.953 0.946 (0.228) (0.228) (0.229) BIRTH ORDER: . ( 1 St. ) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 . 2-3 0 . 8 8 6 0.881 0.877 (0.408) (0.409) (0.409) . 4-6 0.977 0.974 0.965 (0.480) (0.482) (0.483) . 7+ 0.759 0.735 0.735 (0.527) (0.529) (0.529) TYPE OF NEIGHBORHOODS .(Squatting Neighborhoods) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .High Standing 0.385*” 0.387“ * 0.394“ * Neigborhoods (0.316) (0.322) (0.323) .Average Standing 0.648 0.656 0.674 Neigborhoods (0.302) (0.302) (0.305) MOTHER’S AGE: .( < 2 0 years) 1 . 0 0 0 1 . 0 0 0 1 . 0 0 0 .20-24 0.890 0.915 0.920 (0.422) (0.423) (0.423) . 25-34 0.980 1.030 1.047 (0.484) (0.486) (0.486) . 35 + 0.920 0.989 1 . 0 1 0 (0.589) (0.588) (0.588) -2 LOG L 573.885 571.198 570.781 CHI-SQUARE 29.509 32.013* 32.43 DF 2 1 23 24 421 Table 5.5.3. (continued) VARIABLES MODEL 12 MODEL 13 MODEL 14 MARITAL STATUS: . (Married Mono.) 1.000 1.000 1.000 . Single 1.542 1.489 1.476 (0.634) (0.636) (0.636) . Married Polyg. 0.980 0.945 0.932 (0.401) (0.403) (0.404) . Others 8.669“ 9.208“ 9.157“ (0.964) (0.963) (0.959) INTERACTION: .(Edu*Married_M) 1.000 1.000 1.000 . Edu*Others 0.732“ 0.725“ 0.725“ (0.113) (0.113) (0.113) MOTHER’S OCCUPATION: . (Housewife) 1.000 1.000 1.000 . Student 0.911 0.986 0.998 (0.882) (0.887) (0.887) . Working 1.147 1.154 1.145 (0.386) (0.387) (0.387) . Others 0.368* 0.362“ 0.364“ (0.518) (0.520) (0.519) INDEX OF QUALITY OF HOUSING: .(Low: Index 1-6) 1.000 1.000 1.000 . Average: 7-9 0.617 0.640 0.649 (0.378) (0.380) (0.381) . High: 10-12 0.620 0.664 0.667 (0.405) (0.411) (0.412) -2 LOG L 573.885 571.198 570.781 CHI-SQUARE 29.509 32.013* 32.43 DF 21 23 24 422 Table 5.6.1.: Parameter Estimates and Standard Error (in Parenthesis) for Multinomial Logistic Regression of Occupational Status of Mothers with Preschool Children. Urban Areas of Zaire, 13 Cities. Fonames/Unicef Survey, 1987. VARIABLES P1/P3 P2/P3 MODEL L.RATIO INTERCEPT 2.107**** 0.010 1 EDUCATION (0.086)_ w 4 c 4 c 4 c -0.100 (0.116) -0.039 109.9**** (0.012) (0.016) INTERCEPT 1.785**** - 1 . O U 4 * * * * 2 EDUCATION -0.124 (0.012) -0.095 (0.018) AGE OF MOTHER: . (< 20 Years) 0 . 0 0 0 0 . 0 0 0 4 c 4c 4 c 4 c 0.735 (0.096) . 20-24 Years 0.152* (0.087) „ . 25-34 Years 9 c 4 c 4 c 4 c 0.374 1.284 . 35+ Years (0.082) 4b 4 c 4 c 4 c 0.608 1.791 4 c 4 c 4 c 4c 185.8 (0.089) (0.129) INTERCEPT 1.557’ '**’ ' -1. t > /4**** 3 EDUCATION (0.214) x -0.119 (0.013) < ° - 2 5 5 U * * -0.105 (0.018) AGE OF MOTHER: . (< 20 Years) . 20-24 Years 0 . 0 0 0 4 c 4c 0.193 0.645 . 25-34 Years (0.089) 0.512 ( ° * 0 9 9 i * * * 0.959 . 35+ Years < ° ‘ 0 9 5 l * * * 0.786 (0.109) 1.286 (0.158) NUMBER OF CHILDREN-■EVER-BORN: . (Childless) 0 . 0 0 0 0 . 0 0 0 . 1-2 0.147 (0.178) 0.235 . 3-4 0.014 (0.181) 0.459 ( S - 2 0 ^ * * * . 5 + -0.0670 0.808 4 c 4c 394.6 (0.183) (0.212) **** p <= o.OOl; *** p <= 0.01; ** p <= 0.05; * p <= 0.1; P1/P3 <==> HOUSEWIVES versus WORKING WOMEN; P2/P3 <==> OTHERS versus WORKING WOMEN; L.RATIO= LIKELIHOOD RATIO 423 Table 5.6.1. (continued) INTERCEPT 1.227 -i. 4 EDUCATION -0.109 (0.013) (0.287) _ f U -0.097 (0.018) AGE OF MOTHER: . (< 20 Years) . 20-24 Years 0.0 0 0 0.182** 0. 0 0 0 0.624 . 25-34 Years (0-092U** 0.491 0.910 . 35+ Years (0.098) w 4 k 0.771 (0.112) (0.114) 1.244 (0.164) NUMBER OF CHILDREN--EVER-BORN: . (Childless) 0 . 0 0 0 0 . 0 0 0 . 1-2 0.136 (0.179) 0.216 (0.191) . 3-4 -0.002 0.461 . 5 + (0.183) -0.084 (0.185) (0.203) ^ it ' i t 4t 0.821 (0.215) USE OF MODERN CONTRACEPTIVE: . (No) . Yes 0. 0 0 0 ' i t 4 c " t e 4 e 0.305 0. 0 0 0 0.138 528.5 INTERCEPT 0.129 -0.492 5 EDUCATION (0.277) A t 4 e r f c J p -0.110 (0.013) (0.362) ^ -0.099 (0.020) AGE OF MOTHER: . (< 20 Years) . 20-24 Years 0.0 00 • j k 4 > 0.215 0 . 000 4t 4t 4t 0.524 . 25-34 Years (0.093) 0.526 (0.107) 0.769 . 35+ Years (0.099) ^ » J j f 0.805 (0.114) <°-i2n*** 1.072 (0.176) NUMBER OF CHILDREN--EVER-BORN: . (Childless) 0. 000 0. 000 . 1-2 0.132 (0.181) 0.178 (0.206) . 3-4 0.068 (0.186) 0.151 (0.219) . 5 + 0.015 (0.189) 0.408 (0.234) 424 Table 5.6.1. (continued) USE OF MODERN CONTRACEPTIVE: . (No) . Yes 0 . 0 0 0 ^ ^ k «JU 0.252 0 . 0 0 0 k k 0.296 (0.093) (0.146) MARITAL STATUS OF MOTHER: . (Married Mono.) . Single 0 . 0 0 0 ^ ^ kkkk 0.802 0 . 0 0 0 -1.278**** . Married Poly 0.043 0.234 . Others (0.067) ft k k k 0.372 -0.723 (0.067) (0.087) INTERCEPT 0.592 -0.338 (0.382) ^ (0.477) EDUCATION it k k k -0.108 kkkk -0.141 (0.016) (0.031) 1087.5 AGE OF MOTHER: . (< 20 Years) . 20-24 Years . 25-34 Years . 35+ Years 0.000 4 > *4* 0.218 (0.093) ^ 0.527 (0.099) 4 * Jf 4r 0.805 (0.114) NUMBER OF CHILDREN-EVER-BORN: . (Childless) 0.000 . 1-2 0.143 (0.182) . 3-4 0.083 (0.187) . 5 + 0.027 (0.190) USE OF MODERN CONTRACEPTIVE: (No) 0.000 Yes 0.2641 (0.093) MARITAL STATUS OF MOTHER: . *** (Married Mono.) Single Married Poly Others 0.000 0.205 (0.256) 0.003 (0.129) k k k 0.507 (0.137) 0 . 000 0.517**** 0.761 (0.124) ftkkk 1.091 (0.176) 0.000 0.172 (0.207) 0.142 (0.220) 0.410 (0.235) 0.000 0.297 (0.146) ** 0.000 _ k k k k -1.347 (0.257) 0.120 (0.259) -0.462 (0.174) Table 5.6.1. (continued) INTERACTION: . (Educ*Married_M.) 0.000^ • I f 4g -0.196 0.000 . Educ*Single -0.005 (0.067) (0.063) . Educ*Married_P. -0.013 -0.053 (0.037) (0.088) . Educ*Others 0.040 0.084 (0.038) (0.049) INTERCEPT 0.709 -0.307 EDUCATION (0.358) ^ -0.104 (0.421) k e 4f -0.114 (0.014) (0.023) AGE OF MOTHER: . (< 20 Years) . 20-24 Years 0.000 0.192** 0.000 •fc ' f g 4f 0.489 (0.094) (0.108) . 25-34 Years » * * * 0.497 * 9 c 9 c 9 c 9 c 0.731 (0.100) ^ (0.124) . 35+ Years 0.765 w 9 c 9 c 9 c 1.043 (0.115) (0.176) NUMBER OF CHILDREN-EVER-BORN: . (Childless) 0.0 0 0 0. 0 0 0 . 1-2 0.145 0.165 (0.182) (0.207) . 3-4 0.097 0.149 (0.187) (0.221) 0.419 . 5 + 0.053 (0.190) (0.236) USE OF MODERN CONTRACEPTIVE: . (No) 0.0 0 0 " i t "k 0.0 00 . Yes 0.241 0.260 (0.093) (0.146) MARITAL STATUS OF MOTHER: . (Married Mono.) 0.00 0 0 . 0 00 • I f « J f c « 4 f -1.471 . Single 0.145 (0.261) (0.257) . Married Poly 0.059 0.262 . Others (0.068) 0.3i7**** -0.782* (0.068) (0.089) INTERACTION: . (Educ*Married_M.) . Educ*Single 0.00 0 0 . 0 00 -0.196*** -0.026 TYPE OF NEIGHBORHOODS: * A» 0.219 .High Standing Neig 0.084 . Average Stand. N. (0-06H*** 0.327 (0.098) llr 4r 0.363 (0.052) (0.080) 1067.5 1808.2 426 Table 5.6.1. (continued) I N T E R C E P T 0.818 -0.149 (0.369) (0.445) EDUCATION -0.115 -0.103 (0.023) (0.035) AGE OF MOTHER: . (< 20 Years) 0.000 0.000 . 20-24 Years 0.192** 0.489**** (0-094L** (0*108L** . 25-34 Years 0.494 0.727 (0.100) (0.124) . 35+ Years 0.763 1.040 (0.115) (0.176) NUMBER OF CHILDREN-EVER-BORN: . (Childless) 0.000 0.000 . 1-2 0.141 0.159 (0.182) (0.207) . 3-4 0.094 0.146 (0.187) (0.221) . 5 + 0.050 0.414 (0.190) (0.236) USE OF MODERN CONTRACEPTIVE: . (NO) 0.000 0.000 . Yes 0.236** 0.258* (0.093) (0.146) MARITAL STATUS OF MOTHER: . (Married Mono.) 0.000 0.000 . Single 0.141 -1.474**** (0.261) (0.257) . Married Poly 0.056 0.258 , v (0.068) (0.160) . O t h e r s 0.317 * -0.783**** (0.068) (0.089) INTERACTION: . (Educ*Married_M.) 0.000 0.000 . Educ*Single -0.198*** -0.026 (0.068) (0.061) TYPE OF NEIGHBORHOODS: . (Squatting Neighb.) 0.000 0.000 . High Standing Neig -0.066 0.025 (0.138) (0.23.71 . Average Stand. N. 0.442 0.366 (0.100) (0.150) INTERACTION: . Educ*High_Stand. -0.039 -0.056 (0.036) (0.058) . Educ*Average_Stand. 0.039 -0.003 (0.029) (0.044) **** p <= 0.001; *** p <= 0.01; ** p <= 0.05; 1801.4 <= o.l; 427

Linked assets