University of Southern California Dissertations and Theses

Factors affecting child survival in Pakistan

(USC Thesis Other)

Factors affecting child survival in Pakistan

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. U M I 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 U M I a complete manuscript
and there are missing pages, these w ill be noted. Also, if unauthorized
copyright material had to be removed, a note w ill 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 overtaps.
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 UM I directly to order.
ProQuest Information and Learning
300 North Zeeb Road. Ann Arbor, M l 48106-1346 USA
800-521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FACTORS AFFECTING CHILD SURVIVAL IN PAKISTAN
Copyright 2000
by
Muhammad Arshad Mahmood
A Dissertation Presented to the
FACULITY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
in Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIOLOGY)
DECEMBER 2000
Muhammad Arshad Mahmood
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3041495
_______ (f t
UMI
UMI Microform 3041495
Copyright 2002 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
Muhammad Mahmood
under the direction of k X J S i 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 ate D ecember I8> 2000 .
DISSERTATION COMMITTEE
Chairperson
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Muhammad Arshad Mahmood Dr. David M. Heer
ABSTRACT
FACTORS AFFECTING CHILD SURVIVAL IN PAKISTAN
Using data from the Demographic and Health Survey conducted in Pakistan in
1991, this dissertation examines how the socioeconomic factors affect child survival
through environmental, demographic, nutritional and health care factors. The findings
do not fully support the hypothesis that socioeconomic factors affect child survival
through the proximate determinants as proposed by Mosley and Chen’s (1984)
framework, rather they also have their independent and direct effect in improving
child survival in Pakistan. However, the proximate determinants have stronger
influence on neonatal and post neonatal mortality than do the socioeconomic factors.
Among the socioeconomic factors, the father’s education has a stronger effect
on post neonatal mortality compared to the mother’s education. However, the mother’s
education does have an effect on improving child survival in rural areas. The
preceding birth interval and the survival status o f the older siblings are the most
important demographic determinants of neonatal and post neonatal mortality. The
survival o f children improves significantly when the mothers with shorter birth
intervals and women living in rural areas attend prenatal care services. Moreover,
babies delivered at private hospitals have lower odds of stunting compared to babies
delivered at home.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As the child grows, diarrhea has a greater negative impact on his/her
nutritional status. Boys are at higher risk of stunting compared to girls, but as boys
grow the odds o f stunting decreases relative to the girls.
The analysis documented a disturbing picture of mortality and stunting among
children in underprivileged segments of the population. The findings confirm the great
magnitude o f under-nutrition that continues to hamper the physical growth and mental
development of more than a half o f the Pakistani children. The analysis shows that the
causes of growth retardation in Pakistan are deeply rooted in poverty, unhygienic
household environments, non-utilization of health services and lack of education.
The dissertation proposes some policy implications starting from better
utilization of health services to the opening o f mixed primary schools in rural areas.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
I have been extremely fortunate to have Professor David M. Heer as chairman
of my dissertation committee and academic mentor throughout the period o f my
graduate studies. I have benefited immensely from the suggestions he gave me at
various occasions.
I gratefully acknowledge Professor Angela James, Professor Merill Silverstein
and Professor Wendy Mack, my dissertation committee members, for their valuable
suggestions, which encouraged me to make this study possible.
I want to express my appreciation for the fellowship support provided by The
William and Flora Hewlett Foundation and Fred H. Bixby during my four-year stay at
USC. I am particularly indebted to them for their financial and administrative support.
I am also grateful to Professor Dowell Myers for the financial support and for
providing me the opportunity to work for him during the year 2000.
My studies may not be completed without the support of my family. I wish to
express my special thanks to my wonderful wife— Farha. She sacrificed a lot to make
this study possible. My deep love to my daughter — Natasha and son —Zeehasham.
Finally, I am also grateful to Dr. Sultan S. Hashmi, my professional mentor in
the Ministry of Population Welfare who encouraged me to come to this program at the
University of Southern California..
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
Serial No. Contents Page
ACKNOWLEDGEMENTS ii
LIST OF TABLES x
LIST OF FIGURES xiv
1 CHAPTER I INTRODUCTION
1.1 Background and justification 1
1.2 Objectives 5
1.3 Background Information on Pakistan 6
1.3.1 Population and Geography 6
1.3.2 Geographic Regions 6
1.3.3 Urban Rural Population Distribution 7
1.3.4 Climate and Seasons 7
1.3.5 Religion and Culture 8
1.3.6 Fertility 9
1.3.7 Education 9
1.3.8 Health and Family Planning Services 10
1.3.9 Community Based Program 12
1.3.10 Child Survival Projects 13
1.3.11 Nutrition 14
1.3.12 Infant and Child Mortality 15
1.3.13 Housing 18
2 CHAPTER2 LITERATURE REVIEW
2.1 Socioeconomic Factors 19
2.1.1 Maternal Education 20
2.1.2 Father’s Education 26
2.1.3 Place of Residence (Urban-Rural) 27
2.2 Household Environmental and Hygiene Factors 28
2.2.1 Source of Water Supply and Sanitation 28
2.2.2 Household Density 34
2.3 Demographic Factors 35
2.3.1 Maternal Age 35
2.3.2 Birth Order 36
2.3.3 Preceding Birth Interval 37
2.3.4 Sex of the Child 40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
2.4 Nutritional / Dietary Factors 41
2.4.1 Breastfeeding 42
2.4.2 Maternal Malnutrition 47
2.5 Health Seeking Behavior 47
2.6 Nutritional Status of Living Children 50
2.6.1 Causes of Malnutrition 53
2.6.2 Diarrhea and Malnutrition 54
2.6.3 Malnutrition and Mortality 55
2.6.4 Different Approaches of Measuring Nutritional Status 58
2.6.5 Statistical Measures of Nutritional Status 60
2.6.5.1 Percentiles 60
2.6.5.2 Percentage of Median 61
2.6.5.3 Standard Deviations 61
2.6.6 Choice of Reference Population 62
2.6.7 Cut off Points 62
2.6.8 Anthropometric Indices 63
2.6.9.1 Height-for-Age 63
2.6.9.2 Weight-for-Height 65
2.9.6.3 Weight-for-Age 66
3 CHAPTER 3 DATA AND METHODOLOGY
3.1 Data Source 68
3.1.1 Sample Design 68
3.1.2 Questionnaires 70
3.1.3 Data Quality 70
3.1.4 Potential Effects of Data Errors 72
3.2 Conceptual Framework 75
3.2.1 Mosley and Chen Framework 75
3.2.2 Modified Analytical Framework 78
3.2.3 Framework to Study the Underlying
Causes o f Malnutrition and Death 79
3.2.4 Proposed Framework 8 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
3.3 Variables Used in the Analysis 84
3.3.1 Socioeconomic Variables 83
3.3.1.1 Index of Household Possessions 83
3.3.1.2 Parental Education 84
3.3.1.3 Place of Residence 85
3.3.2 Household Environment and Hygiene 86
3.3.2.1 Source of Drinking Water 86
3.3.2.2 Toilet Facilities 87
3.3.2.3 Housing Material 88
3.3.2.4 Exposure to Diarrhea 89
3.3.3 Demographic and Maternal Factors 90
3.3.3.1 Age of the Child 90
3.3.3.2 Preceding Birth Interval 91
3.3.3.3 Birth Order 91
3.3.3.4 Previous Death of Siblings 92
3.3.4 Nutritional / Dietary Intake Factors 93
3.3.4.1 Birth Weight / Birth Size 93
3.3.4.2 Breastfeeding 94
3.3.4.3 Supplementary Food 94
3.3.5 Health Seeking Behavior factors 95
3.3.5.1 Prenatal care 95
3.3.5.2 Place of Delivery 96
3.3.5.3 Birth Attendant 98
3.3.5.4 BCG Vaccination 98
3.4 Dependent Variables 99
3.4.1 Neonatal and Post neonatal Child 99
3.4.2 Nutritional Status (Stunting) 101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
3.5 Statistical Methodology 102
3.5.1 Proportional Hazards Analysis 102
3.5.2 Ordered Logistic Regression 106
3.5.3 Multinomial Logit Model 109
4 CHAPTER4 NEONATAL AND POST NEONATAL MORTALITY
4.1 Bivariate Analysis 110
4.1.1 Socioeconomic Variables 110
4.1.1.1 Parental Education 111
4.1.1.2 Index of Household Possessions 113
4.1.1.3 Place of Residence 113
4.1. 1.4 Region of Residence 114
4.1.2 Domestic Environment and Hygiene Factors 114
4.1.2.1 Source of Drinking Water 11 4
4.1.2.2 Toilet Facilities 115
4.1.2.3 Househo Id Construction Material 118
4.1.3 Demographic and Maternal Factors 118
4.1.3.1 Maternal Age 118
4.1.3.2 Preceding Birth Interval 119
4.1.3.3 Births During the Last Five Years 120
4.1.3.4 Sex of the Child 121
4.1.3.5 Previous Siblings Death 121
4.1.4 Dietary Factors 122
4.1.4.1 Premature Births 123
4.1.4.2 Birth weight/Birth-size 124
4.1.5 Health Seeking Behavior 126
4.1.5.1 Prenatal care 126
4.1.5.2 Place of Delivery 126
4.1.5.3 Delivery Attendant 129
4.1.5.4 BCG Vaccination 129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
4.2 Proportional Hazard Analysis 130
4.2.1 Neonatal Mortality 130
4.2.1.1 Socioeconomic Factors 130
4.2.1.2 Domestic Environment & Hygiene 133
4.2.1.3 Demographic Factors 135
4.2.1.4 Nutritional Factors 138
4.2.1.5 Health Seeking Behavior 139
4.2.1.6 Socioeconomic Factors and Proximate
Determinants 142
4.2.2 Post Neonatal Mortality 147
4.2.2.1 Socioeconomic Factors 147
4.2.2.2 Domestic Environment & Hygiene 150
4.2.2.3 Demographic Factors 152
4.2.2.4 Nutritional Factors 156
4.2.2.5 Health Seeking Behavior 159
4.2.2.6 Socioeconomic Factors and Proximate
Determinants 161
4.3 Conclusion 166
5 CHAPTERS NUTRITIONAL STATUS OF LIVING CHILDREN
5.1 Bivariate Analysis 170
5.1.1 Socioeconomic Variables 171
5.1.1.1 Parental Education 171
5.1.1.2 Index of Household Possessions 173
5.1.1.3 Place of Residence 173
5.1.1.4 Region of Residence 174
5.1.2 Domestic Environment and Hygiene 174
5.1.2.1 Source of Drinking Water 174
5.1.2.2 Toilet Facilities 175
5.1.2.3 Housing Construction Material 176
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
5.1.3 Demographic and Maternal Factors 177
5.1.3.1 Age of the Child 177
5.1.3.2 Maternal Age 178
5.1.3.3 Preceding Birth Interval 179
5.1.3.4 Birth Order 181
5.1.3.5 SexoftheChild 181
5.1.3.6 Previous Siblings Death 181
5.1.4 Nutritional Factors 182
5.1.4.1 Premature Births 182
5.1.4.2 Birth weight/Birth-size 183
5.1.4.3 Bottle Feeding with Nipple 183
5.1.4.4 Diarrhea During Two Weeks Before the Survey 185
5.1.5 Health Seeking Behavior 185
5.1.5.1 Prenatal care Received 185
5.1.5.2 Delivery Attendant 186
5.1.5.3 Place of Delivery 186
5.1.5.4 BCG Vaccination 188
5.1.5.5 Contraceptive Use 188
5.2 Logistic Regression Analysis 189
5.2.1 Socioeconomic Factors 189
5.2.2. Domestic Environment and Hygiene 191
5.2.3 Demographic Factors 194
5.2.4 Nutritional Factors 196
5.2.5 Health Seeking Behavior 199
5.2.6 Socioeconomic Factors and Proximate
Determinants 201
5.3 Multivariate Logit Model Analysis: Mortality and Stunting Together 210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Serial No. Contents Page
6 CONCLUSION 215
6.1 Socioeconomic Factors 216
6.2 Environment & Hygiene 218
6.3 Demographic Factors 219
6.4 Nutritional Factors 220
6.5 Health Seeking Behavior 221
Bibliography 226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table-3.1: Distribution of births included in the analysis
Table-3.2: Distribution of households by index of household
Possessions included in the sample and in this analysis
Table-3.3: Distribution of children by parental education
in the sample and in this analysis
Table-3.4: Distribution of children by place of residence
in the sample and in this analysis
Table-3.5: Distribution of households by source of drinking water
in the sample and in this analysis
Table-3.6: Distribution of households by toilet facilities
in the sample and in this analysis
Table-3.7: Distribution of households by construction material used
in the sample and in this analysis
Table3.8: Distribution of children by the length of preceding birth interval
in the sample and in this analysis
Table-3.9: Distribution of babies by their birth-size
in the sample and in this analysis
Table-3.10: Distribution of women by the status of prenatal care
in the sample and in this analysis
Table-3.11: Distribution of women by place of delivery
in the sample and in this analysis
Table-3.12: Distribution of women by birth attendant
in the sample and in this analysis
PAGE
75
84
85
86
87
88
89
91
94
96
97
98
X
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES (Continued) PAGE
Table-3.13: Distribution of children by BCG vaccination
in the sample and in this analysis 99
Table-3.14: Distribution of number of live births by age of children
in the sample and in this analysis 100
Table-4. l:Bivariate Relationship Between Socioeconomic Factors and percent of
Deaths during the Neonatal Post neonatal period, PDHS, 1990-91 112
TabIe-4.2:Bivariate Relationship Between Domestic Hygiene Factors and percent of
Deaths during the Neonatal Post neonatal period, PDHS, 1990-91 116
Table-4.3:Bivariate Relationship Between Demographic Factors and
Deaths during the Neonatal Post neonatal period, PDHS, 1990-91 117
TabIe-4.4:Bivariate Relationship Between Nutritional Factors and
Deaths during the Neonatal Post neonatal period, PDHS, 1990-91 124
Table-4.5:Bivariate Relationship Between Health Seeking Behavior and
Deaths during the Neonatal Post neonatal period, PDHS, 1990-91 128
Table-4.6: Hazard Rate Ratios Obtained from Proportional Hazard Models of Socioeconomic
Factors for Predicting the Neonatal Mortality, PDHS, 1990-91 132
Table-4.7: Hazard Rate Ratios Obtained from Proportional Hazard Models of Environmental
Factors for Predicting the Neonatal Mortality, PDHS, 1990-91 134
Table-4.8: Hazard Rate Ratios Obtained from Proportional Hazard Models of Demographic
Factors for Predicting the Neonatal Mortality, PDHS, 1990-91 137
Table-4.9: Hazard Rate Ratios Obtained from Proportional Hazard Models of
Nutritional Factors Predicting the Neonatal Mortality, PDHS, 1990-91 139
Table-4.10: Hazard Rate Ratios Obtained from Proportional Hazard Models of
Health Seeking Behavior for Predicting the Neonatal Mortality, PDHS, 1990-91 141
Table-4.11 :Hazard Rate Ratios obtained from Proportional Hazard Model for
Predicting Neonatal Mortality, PDHS, 1990-91 144
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES (Continued) PAGE
Table-4.12:Hazard Rate Ratios obtained from Proportional Hazard Model for
Predicting Neonatal Mortality, PDHS, 1990-91 145
Table-4.13: Hazard Rate Ratios Obtained from Proportional Hazard Models of Socioeconomic
Factors for Predicting the Post Neonatal Mortality, PDHS, 1990-91 148
Table-4.14: Hazard Rate Ratios Obtained from Proportional Hazard Models of Environmental
Factors Predicting the Post neonatal Mortality, PDHS, 1990-91 151
Table-4.15: Hazard Rate Ratios Obtained from Proportional Hazard Models of Demographic
Factors for Predicting the Post neonatal Mortality, PDHS, 1990-91 153
Table-4.16: Hazard Rate Ratios Obtained from Proportional Hazard Models of Nutritional
Factors for Predicting the Post neonatal Mortality, PDHS, 1990-91 157
Table-4.17: Hazard Rate Ratios Obtained from Proportional Hazard Models of Health Seeking
Behavior for Predicting the Post neonatal Mortality, PDHS, 1990-91 160
Table-4.18: Hazard Rate Ratios Obtained from Proportional Hazard Model Analysis for
Predicting Post neonatal Mortality, PDHS, 1990-91 163
Table-4.19: Hazard Rate Ratios Obtained from Proportional Hazard Model Analysis for
Predicting Post neonatal Mortality, PDHS, 1990-91 164
Table-5. l:Bivariate Relationship Between Socioeconomic Factors and height-for-age
Among Children aged 0-59 months, PDHS, 1990-91 172
Table-5.2:Bivariate Relationship Between Domestic Hygiene Factors and height-for-age
Among Children aged 0-59 months, PDHS, 1990-91 175
Table-5.3 :Bivariate Relationship Between Demographic Factors and height-for-age
Among Children aged 0-59 months, PDHS, 1990-91 180
Table-5.4:Bivariate Relationship Between Nutritional Factors and height-for-age
Among Children aged 0-59 months, PDHS, 1990-91 184
Table-5.5:Bivariate Relationship Between Health Seeking Behavior and height-for-age
Among Children aged 0-59 months, PDHS, 1990-91 187
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES (Continued) PAGE
Table-5.6: Odds Ratios obtained from the Ordered Logistic Regression Models of
Socioeconomic Factors for Predicting the Nutritional Status of Children 0-59 Months 190
Table-5.7: Odds Ratios obtained from the Ordered Logistic Regression Models of
Environmental Factors for Predicting the Nutritional Status of Children 0-59 Months 193
Table-5.8: Odds Ratios obtained from the Ordered Logistic Regression Models of
Demographic Factors for Predicting the Nutritional Status of Children 0-59 Months 195
Table-5.9: Odds Ratios obtained from the Ordered Logistic Regression Models of
Nutritional Factors for Predicting the Nutritional Status of Children 0-59 Months 198
Table-5.10: Odds Ratios obtained from the Ordered Logistic Regression Models of
Health Seeking Behavior for Predicting the Nutritional Status of Children 0-59 Months 200
Table-5.11: Odds Ratios Obtained from the Ordered Logistic Regression Analysis to Predict
the Probability of Stunting and Severely Stunting Among Children 0-59 Months 202
Table-5.12: Odds Ratios Obtained from the Ordered Logistic Regression Analysis to Predict
the Probability of Stunting and Severely Stunting Among Children 0-59 Months 203
Table-5.13: Predicted Probabilities of Severely Stunted, Moderately Stunted and
Normal Growth based on the Reduced Ordered Logistic regression Model,
Children 0-59 Months, PDHS, 1990-91 206
Table-5.14: Odds Ratios obtained from the Multinomial Logit Model
to Predict the Probability of Death, Severely Stunted and Moderately Stunted
of Children Aged 0-59 Months, PDHS, 1990-91 211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES PAGE
Figure-3.1: Mosley and Chen Framework to study child survival 77
in developing countries
Figure-3.2: Causes of malnutrition and death 80
Figure-3.3: Proposed Model for child survival in Pakistan 82
Figure-5.1: Predicted probabilities of stunting based on final ordered logistic regression
model by age of the child, taking mean values of all the independent variables 208
PDHS, 1990-91
Figure-5.2: Predicted probabilities of stunting based on final ordered logistic regression
model by age of the child, lower SES illiterate parents living in rural Sindh with
Wells water and no toilet facility, PDHS, 1990-91 208
Figure-5.3: Predicted probabilities of stunting based on final ordered logistic
regression model by age of the child, higher SES parents with 10 years of 209
education living urban Punjab with piped water and flush system
PDHS, 1990-91
xiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1
INTRODUCTION
1. Background and j ustiilcation
Ensuring the survival and well being o f children is a concern of families,
communities and nations throughout the world. Since the turn of the 20th century
infant and child mortality in more developed countries has steadily declined and,
currently, has been reduced to almost minimal levels. In contrast, although infant and
child mortality has declined in the past three decades in most less developed countries,
the pace of change and the magnitude of improvement vary considerably from one
country to another. Children are at risk of both mortality and morbidity.
In the past, emphasis was placed on infant and child mortality because of the
high rates of infant and child mortality in developing countries. But now that infant
and child mortality has been considerably lowered in most o f the developing countries,
the emphasis has shifted to improving the health and nutritional status of children.
Until now, childhood mortality accounts for about three-fourths of the total
deaths in Pakistan (Sathar, 1987a). By contrast, it accounts for only 2-3 percent of the
total deaths in North America and Northern Europe (Kent, 1991:5).
Many studies in developing countries show that infant mortality is lower for
countries with higher socio-economic development and higher for countries with
lower socio-economic development. Countries that have improved in economic
performance during the I970’s and early I980’s have achieved rapid progress in child
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
health as indicated by survival chances for children under five (Mosley and Cowley,
1991). In general, these achievements have been attributed to improvements in living
standards, government interventions such as social welfare programs, and public
health services (Huffman and Steel, 1995; Preston, 1985). Furthermore, the success of
better health achievers is measured by exercising the “political and social well” in
those countries (Caldwell, 1994a).
Per capita income in Pakistan is about two times higher than what it is for
Bangladesh or Nepal, but its infant mortality rate is the highest in South Asia (World
Bank, 1999). Pakistan Demographic and Health Survey revealed that more than 90
infants out of 1000 live births died before their first birthday, and half o f the Pakistani
children aged 0-4 years are experiencing malnutrition. Moreover, about one-third of
all children are severely malnourished (NIPS, 1992).
The problem of malnutrition is widespread in developing countries and
particularly severe in South Asian countries, where almost fifty percent of the
undernourished children of the world live (Carlson and Wardlaw, 1990: 12-13). Rural
populations are especially prone to malnourishment and malnutrition because they are
more likely to be poor (Tinger, 1998). The analysis o f Demographic and Health
Surveys (DHS) in 19 developing countries shows that children living in rural areas are
more likely to be malnourished (Sommerfelt and Stewart, 1994). A poor diet and
exposure to repeated illnesses are two of the major causes o f malnutrition in
developing countries (Mosley and Chen, 1984). When the child survives the neonatal
period, better child nutrition becomes an important part o f child health since nutrition
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
during childhood makes a major contribution to child development, growth, and
survival, ultimately influencing the human and social capital of a society. The role of
breastfeeding is very important in the post-neonatal period. The mother's milk not
only provides the complete nutritional requirements of the child but also provides
protection against infection (Jelliffe and Jelliffe, 1978). Pregnant women who receive
inadequate nutriment levels are likely to give birth to underweight babies who are
more likely to get infectious diseases, leading to early death. Furthermore, those who
survive but receive inadequate food in their early life are more likely to be exposed to
permanent stunting o f their physical growth.
The physical environment, socioeconomic status, diet, parental education and
infections determine a child’s growth (Sommerfelt and Stewart, 1994). In developing
countries, the first 2-3 years of growth are very important in determining their adult
stature, work capacity and productivity (Martorell et al., 1990; Satyanarayana et al.,
1980). The reason for this is that children stunted early in life continue to live in the
same deprived environment that precludes catch-up growth and, therefore, never catch
up to their full growth potential in adulthood (Martorell, 1995; Osmani, 1992;
Waterlow, 1993). Genetic factors have a much smaller role because the growth
potential in early childhood is similar across ethnic groups (Martorell, 1985).
Small stature in men leads to reduced work capacity and productivity
(Gopalan, 1983; Spurr, 1990; 1977). Underweight and stunting increases the risk of
morbidity and mortality, and has functional consequences later in life such as
diminished work capacity and increased obstetric risk for women (Osmani, 1999).
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The very first studies undertaken in India (Keilmann and McCord, 1978), Bangladesh
(Sommer and Loewenstein, 1975) and Papua New Guinea (Heywood, 1982)
established the basic findings that the risk of mortality is inversely related to
anthropometric indices of nutritional status. This is also confirmed in recent studies
from different regions o f the World, (Van Den Broeck et al., 1993; Pelletier et al.,
1994b; Yambi, 1988; V ellaetal., 1993; 1994).
In this context, improving child health and nutritional status to prevent
underweight and stunting has become very important. Therefore, improving the health
and nutritional status of children 0-4 years of age deserves high priority. Understanding
the factors that affect growth during these critical periods is essential for determining
appropriate intervention strategies to address the problem. However, the immediate
causes of growth faltering among childhood involve a complex web of biological, social,
and economic factors, which are location-specific and often poorly understood for
particular population groups, locations, and time periods (Anderson, Pelletier and
Anderman, 1995).
In the child survival literature, one of the most common findings is the
education-mortality relationship (Cleland and Van Ginneken, 1988; Caldwell, 1979;
Ware, 1984) even after controlling for the effects of economic factors or other
socioeconomic variables in developing countries (Visaria and Simons, 1997; Caldwell,
1991). However, the possible mechanisms or pathways through which the mother’s
education affects child survival is not well understood (Bicego and Boerma, 1992).
“There are still many missing links between education and child survivorship, with the
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
consequence that our understanding of the mechanism o f influence still remains quite
elusive” (Visaria and Simons, 1997).
Any integrated attempt to investigate the mechanism on the basis o f data
covering broader aspects of possible proximate determinants of child survival do not
exist in Pakistan. The Pakistan Demographic and Health Survey, 1990-91, for the first
time collected information on household characteristics and children's morbidity and
sickness care as well as anthropometric measurements of children 0-59 months of age.
Using these data, this study aims to understand the mechanism through which the poor
chances of child survival in Pakistan operate by employing appropriate
methodological tools and strategies for analyzing these data. To understand why infant
mortality is high, the nutritional status of the living children is also included. This
provides useful information on causal pathways through which the socioeconomic
status of parents influences the survival chances of children.
1.2 Objectives
o To identify the factors (proximate determinants such as, demographic
factors, environmental factors, nutritional factors and health seeking
behavior) associated with neonatal and post neonatal mortality,
o To assess child stunting as reflected in anthropometric indicators of
height measurements for children under 5 years old according to a set of
proximate determinants,
o To understand the mechanisms through which selected socio-economic
factors operate to influence child survival prospects.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13 Background information on Pakistan:
13.1 Population and Geography:
Pakistan is located at a strategic location in South Asia between China to the
north, India to the east, and Afghanistan and Iran to the west and northwest. The Arabian
Sea is to the south. Pakistan became independent on August 14, 1947 by partition of
British India. Originally, it consisted of two separate land areas located about 1,600 km
apart to the east and west of India, but the eastern portion seceded in 1971 and became
the independent nation of Bangladesh. The total land area of Pakistan is about 796,000
square kilometers.
Pakistan had a population of 130.6 million people as of March 1998
(Government of Pakistan, 2000), which positions it as the seventh most populous country
in the world. Pakistan’s population grew during the period 1981-1998 at an annual rate of
2.6 percent. The population of the area that now constitutes Pakistan was 16.6 million in
1901. Annual growth rates rose from 1 percent in the first three decades of the century to
2 percent in the next three decades, and peaks at slightly over 3 percent from the 1960s
through the early 1980s. Marked decline began during the 1990s and the present
population growth rate is estimated at about 2.3 percent (Government of Pakistan, 1999).
13.2 Geographic Regions (Provinces):
Pakistan consists of four provinces: Punjab, Sindh, North West Frontier Province
(NWFP), and Balochistan, and the Federally Administrated Tribal Areas (FATA) and
Islamabad Capital Territory. According to the population census of 1998, the proportion
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
by each province shows a slight variation compared to the 1981 census. The proportion
of NWFP, Sindh and Islamabad Capital Territory has increased from 13.1 to 13.4
percent, 22.6 to 23.0 percent and 0.4 to 0.6 percent, respectively, while the share of
Punjab has declined from 56.2 to 55.6 percent; FATA from 2.6 to 2.4 percent; and
Balochistan from 5.1 to 5.0 percent (Government of Pakistan, 2000).
1.3.3 Urban-Rural Population Distribution:
The Capital territory, Islamabad is the most urbanized area with an urban
population share of 65.6 percent and FATA is on the other extreme end with the share of
urban population amounting to only 2.7 percent, according to the population census of
1998. Among the provinces, Sindh is the most urbanized at 48.9 percent, followed by
Punjab, Balochistan and NWFP, 31.3 percent, 23.3 percent and 16.9 percent,
respectively. Overall, 32.5 percent of the population lives in urbanized areas o f the
country (Government of Pakistan, 2000).
1 J.4 Climate and Seasons:
Pakistan's climate is hot and dry, with cooler temperatures and greater rainfall
in mountain areas. There are four well marked seasons in Pakistan: Cold season
(December-March), Hot season (April-June), Monsoon season (July-September), and
Post-monsoon season (October-November).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 J.5 Religion and Culture:
All the provinces have their own language and culture. Pakistani, especially the
rural society, remains rigidly patriarchal. Women have no say in family decision
making and decisions about marriage, education, employment and health care are
usually made by men and not by the women themselves (Shah, 1984). Most women
in Pakistan are married within the relatives. If a woman marries a man of her choice
without the consent of the parents, she would never again be accepted by her family.
By contrast, Islam gives the option or right o f choice to the individual and
forbids parents from imposing their will on the children for marriage. The approval
and consent of the girl to marriage is a prerequisite for the validity of marriage in
Islam. She has the right to say yes or no. This Islamic practice holds no grounds for
parents when they try to marry off their daughters without their consent.
In Pakistan, the cultural norm is for women to stay home and take care of the
household chores. According to the prevalent misperceptions in Pakistan, it is not
common for women to join the labor force (NIPS, 1992). The working women are
supposed to bring disrepute to their husbands because the husband is thought to be
incapable of controlling the wife and too poor to provide their needs (Malik, 1997).
Therefore, men who are well-off see it as a part o f their 'honor1 to keep their women
economically inactive. Daughters are seen as an economic burden in the family and
there is no concept of considering the female infant as a future supporter of the family
( Malik; 1997).
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0 . 6 Fertility:
Fertility and mortality are the two crucial components o f national population
growth. Although different sources of vital statistics are occasionally varied, a crude birth
rate of 32 and crude death rate of 9 deaths per 1000 as o f 1998 is generally reported
(Government of Pakistan, 1999). Pakistan is far behind its neighbors in the transition to
lower fertility. While the reproductive health of women remains generally poor
compared with other countries, there are signs of some improvements. The total fertility
rate, which remained stagnant for many years, has finally begun to decline from 6.3 in
1975 to 5.3 in 1996-97. Dramatic changes in marriage patterns, but also a rise in the
contraceptive prevalence rate, are precipitating this decline in fertility in the 1990s
(Government of Pakistan, 1999). On average, Pakistani women breastfeed their children
for 20 months, and the median duration of amenorrhea following pregnancy is 6.3
months (NIPS, 1992)
1.3.7 Education
Like many other developing countries, the condition o f Pakistan’s education
sector is not encouraging. While Pakistan has seen some improvements in education
over the past two decades, progress remains slow. The estimated literacy rate is 47
percent: 59 percent for males and 35 percent for females (Government of Pakistan,
2000). According to World Bank, only 40 percent o f the population is literate,
compared to an average of 49 percent in South Asia and 53 percent in low-income
countries worldwide (WHO, 2000).
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gender disparities in education remain significant. A vast majority o f women
are deprived from the right o f attaining education on the argument that women have to
run the household affairs and, hence, need no formal education. Only around 62
percent of Pakistan's girls are enrolled in school, compared to 80 percent among boys
(WHO, 2000). Enrollment figures do not convey the true picture of the gender gap in
access to education because dropout rates vary and are significantly higher for girls
(Government of Pakistan, 2000). Male children are preferred to female children on the
basis that a male child is considered a source of numerical and economic strength for
the family. Moreover, boys are sent to schools and colleges, but the girls are usually
confined to the home. Gender disparities in schooling tend to be particularly marked
among the lowest income groups and in rural areas. In a situation of limited financial
resources, there are greater economic reasons for educating boys since this would lead
to better employment prospects. In contrast, most daughters are not expected to seek
jobs. Rather, they are expected to stay at home.
1.3.8 Health and Family Planning Services
There are two major sectors providing health care services to people, the
Government and the Private sectors. The private sector also includes Non-Govemmental
Organizations. According to the National Health Survey, both, men and women make 5-
6 contacts annually with either one of these two sectors (Pakistan Medical Research
Council, 1998). The government of Pakistan recognizes the need to improve women’s
and children’s overall health status by providing universal access to health care services
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
through the infrastructure of health care facilities. In the Government sector, the Ministry
of Health and the Ministry of Population Welfare separately provide health services
through their own infrastructure. The health care delivery system in Pakistan is strained
by the country’s high population growth rate. The health system involves a spectrum of
practices, including religious and faith healers, traditional healers, homoeopathic
dispensaries, modem hospitals and specialized clinics.
The medical and health establishments of the Ministry o f Health include 877
hospitals, 4,625 dispensaries, 5,152 Basic health units (BHS), 855 maternity & child
health centers (MCH centers), 530 rural health centers (RHC) and 263 TB centers in the
country (Government of Pakistan, 2000).
However, the conventional medical systems focus largely on the curative aspects
of disease. The government is encouraging the use of Western medicine, community
health workers and doctors, and large specialized hospitals are being established in the
cities. The major problem in the area of health continues to be inadequate services and
supplies. Most of the doctors are concentrated in the urban centers, and the situation with
regard to heath services and supplies in the rural area has been relegated to slow progress
(WHO, 1997). Most of the health providers, doctors, nurses and dispensers, are based in
city hospitals. Rural people are left with little or no treatment or have to resort to
traditional healers. Currently 68 per cent of the population lives in rural areas; they are
served by Basic Health Units (BHU), which refer cases to the District Hospitals.
According to the World Bank Report (1998), on average there were only 24
patients seen per day at a Basic Health Unit (BHU). These BHUs are designed to
1 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
provide health care services to up to 10,000 people of a specific geographic region.
Similarly, the government-managed family planning clinics provide services to only
5,000 population.
Population welfare program provides comprehensive family planning and
maternal and child health (MCH) services, both, in urban and rural communities of the
country through its own infrastructure by adopting the International Conference on
Population and Development (ICPD) program of action approach. The infrastructure of
population welfare program includes 1758 Family Welfare Centers (FWCs), 101
Reproductive Health Services Centers (RHS-A), 131 Mobile Service Unites (MSUs)
providing information and services throughout the country. Over 23000 Medical
Practitioners, 13300 Hakeems (Traditional healers) and 7300 Homoeopaths are also
associated with the program in order to strength service delivery, referral and
motivational activities of the program (Government of Pakistan, 1999).
1.3.9 Community Based Program:
Community Health Workers and village-based family planning workers are the
two major new initiatives introduced by the Ministry of Health and Ministry of
Population Welfare, respectively. The village-based family planning workers scheme
under the Ministry of Population Welfare was launched in 1992 and the Prime Minister’s
Program for family planning and primary health care in the ministry of health was
initiated in 1994 just after the International Conference on Population and Development
(ICPD). Around 44,000 female health workers and 12000 village-based female family
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
planning workers have been inducted in the health and family planning program to
provide local health and family planning services mostly in rural areas (Government of
Pakistan, 1999).
1.3.10 Child Survival Projects:
After the promise of “Health for all by the year 2000” was made by 124
countries in 1978 Alma Ata Conference’s declaration (WHO, 1981), many countries,
including Pakistan, realized very soon that primary health care was too broad to
implement with given resources and, thus, started fragmented approaches to control
major killers of children. The first two major components of child survival programs that
began in the 1980s were the Diarrhoeal Disease Control Program (CDD) and the
Expanded Program on Immunization (EPI) with technical and financial assistance from
the international donors.
After the withdrawal of USAID funding in 1990, these projects came under
severe pressure to seek out new donors. Currently, an expanded program of
immunization against six diseases (tuberculoses, diphtheria, tetanus, pertusis, measles
and poliomyelitis) is under execution, along with a program of production/ distribution of
oral re-hydration salt (ORS) packets under the social action program (SAP). The
objective of the EPI program is to reduce mortality among infants, children and women,
resulting from the said diseases. The progress of the program is evidenced by the fact
that, in the current year 1999-2000, about 4 million children have been immunized and
23 million ORS packets have been distributed (Government of Pakistan, 2000).
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1J.11 Nutrition
Malnutrition is a serious health problem in Pakistan. Infants, young children and
women are identified as high risk groups. Malnutrition affects adult women more than
men, and it contributes to a vicious cycle of poor growth from generation to generation.
Malnutrition in women is the result of inadequate food intake due to poverty (Tinger,
1998). In Pakistan, per capita daily calorie intake is estimated at 2715 calories for 1999-
2000. The intake of daily protein per capita is 71.03 grams. The national food intake
balance sheet of six major food items including pulses (edible seeds, peas beans etc.),
sugar, milk, meat, egg and edible oil, shows an improvement in milk, meat and edible oil
over the last year, while it has declined in pulses and sugar (Government of Pakistan,
2000). A study conducted in one area of Pakistan shows that the mean consumption of
calories, protein, total fat, sugar and cholesterol was higher among the urban children
whereas the average daily intakes of carbohydrates, fiber and starch was higher among
the rural children (Hakim et al., 1997). Pregnant women receive 87 percent of the
recommended calories and lactating women 74 percent, and protein intake for these
women is around 85 percent of recommended levels (Federal Bureau of Statistics, 1995).
The government of Pakistan has exhibited a very high sensitivity to this problem
and has proposed a variety of remedial measures including food fortification, mass media
nutrition education, new weaning foods and village level food processing (Government
of Pakistan, 1999).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.12 Infant and Child Mortality
Since Pakistan’s independence in 1947, several attempts have been made to
estimate mortality indices through direct and indirect approaches. The direct estimates of
mortality are available from three demographic survey series: the Pakistan Growth
Estimate (PGE) experiment 1962-1965; the Pakistan Growth Surveys PGS-1, 1968-71
and PGS-2, 1976-1979; and the Pakistan Demographic Survey (PDS) 1984-1997.
The first systematic attempt to study the level of fertility and mortality in Pakistan
was through the PGE, a dual-record survey conducted from 1962 to 1965. Two systems,
a cross-sectional (CS) and a longitudinal (LR) system, were used to collect data on births
and deaths from a national sample. The mortality estimates provided by CS seemed to be
too low compared to those from the LR system, so a controversy developed as to which
were the true estimates of mortality. The first Population Growth Survey (PGS-1),
undertaken during 1968-1971, was designed to improve the methodology of PGE.
During the period 1976-79, a repeat survey PGS-2 similar to PGS-l, was undertaken to
yield estimates of vital rates for Pakistan including provinces and urban-rural areas. The
analysis of PGS (1 and 2) rates also suggests substantial under-enumeration of deaths
during the period 1968-71 and 1976-79 (United Nations, 1986). The PDS series initiated
in 1984, showed a sudden increase in infant mortality rates in Pakistan, which perhaps
was due more to the adjustments made by the generating agency, than the reflection of
the actual phenomenon (Afzal et al., 1988:638).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Utilizing the pregnancy history information collected from two retrospective
surveys, the Pakistan Fertility Survey (PFS), 1975 and Pakistan Labor Force and
Migration Survey (PLMS), 1979, Alam and Cleland (1984) and Sathar (1985) calculated
infant mortality rates. Their estimates indicate that infant mortality rates based on the
PGE were lower than those based on pregnancy history data from the retrospective
surveys, the PFS and the PLMS. This presumably reflects under-enumeration of deaths
in the PGE. Comparison between the PFS and the PLMS shows lower infant mortality in
PLMS (Sathar, 1985). One reason for the lower rates obtained in PLMS was that the
deaths in the PLMS may have been under-reported because the evaluation of the PFS
data quality indicated that the births and deaths were more accurately reported (Booth
and Shah, 1984). The difference in levels of infant mortality rate can partly be attributed
to sampling variability because the PLMS sample was twice the size of PFS’ sample
(Sathar, 1985).
Indirect estimates of infant and child mortality were obtained by applying the
Trussed variant of the Brass method based on the West Model Life Table from four
sample surveys: the National Impact Survey (NIS), the Household, Education and
Demographic survey (HED), the PFS and the PLMS and the ten per cent count of 1981
Population Census of Pakistan are also available. Like the estimates obtained from direct
measures, indirect estimates were also not consistent. In addition to the underlying
assumptions of the methodology used, the derived estimates tend to vary from one survey
to another, owing to the variation of age misreporting and under-enumeration of children
ever bom and dead. Moreover, different sample sizes and sampling errors are bound to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
influence the comparability of results. The high rates from the PFS indicate that deaths
were more completely reported (Martin et al., 1983).
However, the available studies show that infant mortality had declined from
1906, when it was about 228 per 1000, until the 1960s, when it stabilized at fairly high
levels of about 125-140 per 1000 live births (United Nations, 1986; Alam and Cleland,
1984). There has also been a decline over the past two decades due to the rapid
expansion of immunization (Tinger, 1998). The infant mortality rates ranged from 112 to
102 deaths per 1000 live births during the period between 1984 and 1994, respectively,
reported by the PDS, 1984-94 (Government of Pakistan, 2000). The figures produced by
the PDS, 1997 is 84 for the country and 73 and 89 per 1000 live births for urban and
rural, respectively. The infant mortality rates reported by other agencies range between
95 and 91 deaths per 1000 live births for the period 1997 to 1999 (UNFPA, 1997;
Population Reference Bureau, 1999). Although infant mortality at around 90 per 1000
live births, is declining, it still remains high by all standards compared with an average of
75 per 1,000 in South Asian countries and 68 per 1,000 in low-income countries
worldwide (WHO, 2000).
Maternal mortality has declined slightly to about 420 deaths per 100,000 live
births and, given the high fertility rate, this implies an exceedingly high life time risk for
women dying from maternal causes, about one in 35 women (WHO, 2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.3.13 Housing:
Housing conditions are also not encouraging. About one-third of the households
live in houses with cement and baked bricks with T irons. The degree of crowding is also
such that, on average, five persons sleep in one room. Half of the population lives in one
room housing units. The housing development program is under increased pressure in
light of population growth, and the situation is being further aggravated by the rural to
urban migration over and above the existing high rate of natural increase. It is estimated
that there were 2459 Kachi Abadies (Shanti towns) in the country in 1992-93, with a
population of 5.8 million (Government of Pakistan, 1993).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
LITERATURE REVIEW
This chapter reviews theoretical and empirical literature on determinants of child
health and survival in developing countries. Based on this review, specific pathways are
selected to model the effects of socioeconomic factors on child health and survival.
Theoretical guidance is provided by the Mosley and Chen (1984) framework, which
defines child health and survival as a function of proximate determinants that are
influenced by socioeconomic factors.
2.1 Socioeconomic Factors
The inverse relationship between socio-economic variables of the parents and
infant and child mortality is well established by several studies (Muhuri, 1995; Forste,
1994; Hobcraft et al., 1984; Caldwell, 1979; Sathar, 1985; 1987) and it holds true
irrespective of the overall level of mortality in the national populations (Ruzicka, 1989).
Education has been one of the key concepts used as a variable for explaining health. The
influence of parental education on infant and child health and mortality has proved to be
universally significant (Bicego and Boerma, 1993; Caldwell et al., 1990. The father’ s
education, mother’ s education and their work status each have independent effects upon
child survival in developing countries (Sandiford et al., 1995; Forste, 1994; Caldwell et.
al., 1983).
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Caldwell, who found that maternal education was the most important determinant
of child mortality in Nigeria, suggested that increased education strengthens the mother’s
role in the family’s decision-making process and that she may re-allocate family
expenditures from older members to younger members, as well as from male to female
children (Caldwell, 1979). Trussell and Preston, in their study of Sri Lanka and Korea,
found that the mother’s and father’s education are negatively associated to about the
same extent with child mortality (Trussell and Preston, 1982).
Economic conditions of the household also help in explaining the variation in
infant and child mortality. The nature of housing, diet, access to and availability of water
and sanitary conditions as well as medical attention all depend on the economic
conditions of the household. For example, poor families may reside in crowded,
unhygienic housing and, thus, suffer from infectious disease associated with inadequate
and contaminated water supplies and with poor sanitation (Esrey and Habicht, 1986).
2.1.1 Maternal Education:
Mother's education has been found in some developing countries to have
greater impact on child survival than the father’s because mothers normally have
more direct responsibility for child care (Caldwell, 1979; Caldwell and
McDonald, 1981). Maternal education has been given much attention from social
scientists. It is generally treated as a socioeconomic variable at the micro level,
while it is treated as an indicator of development or the social welfare of a
country at the macro level. In other words, maternal education is often viewed
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from a “development perspective” in which it is seen as a component of the
positive process of modernization. Thus, education is expected to provide
knowledge and skills beyond necessary literacy which will encourage women to
participate not only in developing activities, but also in building human capital by
promoting child health and welfare (Caldwell, 1979). Hence maternal education
has been increasingly suggested as one of the routes leading to further
improvements in child survival and child health in developing countries
(Caldwell, 1979; 1986; Cleland and Van Ginneken, 1988).
Caldwell (1979) and Caldwell and McDonald (1981) suggested that
maternal education is linked with lower infant and child mortality through a
number of ways. Three broad pathways of influence, linking maternal education
to child survival, that result in greater utilization of modem health services have
been suggested. First, reduction in fatalism in the face of a child's illness and a
change in the traditional balance of family relationship that gives the mother
more influence in decision-making about childcare and treatment in care of
illness treatment (Caldwell, 1979: 408-412). Second, maternal education also
helps in understanding the 'modem world' such as the choices which she may
make in utilizing what is available in the community to their advantage
(Caldwell, 1990; Barrera, 1990; Goodbum et al., 1990). Third, educated mothers
may be able to make independent decisions regarding their own and their
children’s health leading to greater utilization of modem health facilities
(Caldwell, 1979; Caldwell, 1986).
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Educated mothers are also more likely to earn higher incomes and to
marry educated men, therefore enabling the children to live in a better
environment, and allowing to provide better food and better health services
relative to their uneducated counterparts (Schultz, 1984). However, Ware (1984)
argued that it is difficult to separate the effect o f income and education in matters
of childcare. She pointed out that in a situation where there is enough food to go
around, the decisions, irrespective of the mother’ s education, will be different
from those in a situation where difficult choices have to be made about who will
get how much from the very limited food available.
At the macro-social view, Caldwell relates the high levels of female
autonomy and education to other features of society. For example, he discusses
the effects o f religion on the demand for education and notes that one condition
for unusual educational advances is “a basic reverence for enlightenment or
education” (Caldwell 1986a).
Caldwell also argued that “countries that advance most rapidly in this
area are those in which parents achieve as much satisfaction seeing their
daughters at school as they do their sons” (Caldwell, 1986). When he compares
educated and uneducated mothers in a single society this perspective of
education, as a reflection of societal attitudes, is replaced by education as a
force that changes individuals. For instance, in a South India study he argued
that “the woman who has been to school knows that the school expects her to
take action and that she should not be bound by deference to traditional
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
decision-making patterns and excessive female modesty when children’s
health is at stake. Strangely enough, this is a view usually shared by her
parents-in-law” (Caldwell 1989a).
In their research o f South India, Caldwell and colleagues found a
multiplicity o f reactions between the mothers of sick children and the health
system, which explained why educated mothers obtained more help from that
system. The educated mothers were more likely to take children for treatment
at the modem health centers and were likely to do so with less delay. These
mothers spent a longer period with the doctor because they were willing and
able to describe symptoms more fully and because the doctor listened to them
at greater length. They followed the prescribed treatment more exactly and
were more likely to persist with it. They were much more likely to report back,
and to do so sooner, if the child’s condition was not improving (Caldwell et
al., 1990).
Their greater willingness to employ modem medicine, and to spend a
larger proportion of family income for it, arose because their education made
them feel that they had a responsibility for taking an initiative because they
were more likely to believe that the doctor and the health center embodied a
scientific truth about the world (Caldwell et al., 1990). Ample research on
child health demonstrates that maternal education is significantly associated
with childhood mortality and nutritional status o f living children. Caldwell and
others suggested that maternal education is linked with lower infant and child
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mortality through a reduction in fatalism in the face of a child's illness and a
change in the traditional balance of family relationship that gives the mother
more influence in decision-making about child care and treatment in care of
illness (Caldwell, 1986; Koenig et al., 1991). Maternal education also helps in
understanding the 'modem world' such as the choices which she may make in
practicing health care, nutrition, hygiene, preventive care, and disease
treatment.
Being educated often enhances the ability of women to express
themselves and communicate effectively with health workers (Karki and Levine,
1993). Furthermore, they are likely to be better treated by health workers
(Maclean, 1974). By seeking an explanation in terms of the higher-quality health
care that educated mothers may be able to evoke, it becomes possible to explain
why, for instance, the difference in childhood survival related to maternal
education persists in urban areas where there is a much greater availability of
health services (Cleland and Van Ginneken 1989; Bicego and Boerma 1993).
However, the association between mother’s education and child mortality
becomes more significant when the child is older than the infancy period
(Cleland and Van Ginneken, 1988) because biological factors, rather than
childcare practices, play an important role in determining mortality among
newborn children.
Bicego and Boerma (1993) argued that in the post-neonatal period,
mortality risk is almost two times more sensitive to the effect of maternal
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
education as is neonatal mortality risk. They argued that under-nutrition is the
presumable causal pathway through which mother’s education exerts an impact
on the survival prospects of the children during the post-neonatal period.
However, the association between stunting and maternal education tends to
decrease when the economic status of the household is controlled, which suggests
that the relationship between maternal education and stunting reflects more the
differentials in economic status than the differentials in education. The
relationship between education and underweight is more pronounced than the
education-stunting relationship (Bicego and Boerma, 1993).
Another study based on DHS 25 developing countries data found that 1-3
years of maternal schooling reduces child mortality by 15 percent and this effect
increases as mothers gain more education (Hobcraft et al., 1993). This effect may
either be through some socioeconomic factors such as maternal work or through
proximate determinants, such as use of health care, but relatively little attention
has been directed towards examining these in explaining differentials in child
health. The exact mechanisms through which education influences child health
are not yet clear (Cleland and Van Ginneken, 1988; Levine et al., 1994). The
Bangladesh study reveals that the mother’s education has a greater impact on
child survival than the father’s education. The authors argued that the influence
of the father’s education is largely due to the socioeconomic status (Majumder et
al., 1993).
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.1.2 Father’s Education:
In some Asian and Pacific countries, the father's education was also
significant in reducing infant and child mortality (Ruzicka, 1989). Although the
father’s education is found to be less influential than the mother’s education in
reducing infant mortality in some studies (D’Souza and Bhuiya, 1982), it is more
strongly correlated with household income and general socioeconomic conditions
that may directly influence the parents’ preferences and opportunities in choice of
consumption goods and child care arrangements. Mensch and colleagues in their
study suggest that the effect o f the father’s education may be operating on child
mortality primarily through economic channels; and the maternal education
influence, on the other hand, may be more closely related to child-care practices
(Mensch et al., 1985). Another study on the covariates o f infant and child
mortality in the Philippines, Indonesia, and Pakistan, found that the father’s
education makes a difference only at older ages of children (Martin et al. 1983).
Using data of World Fertility Surveys of 28 countries, a study shows that, in
terms of statistical significance, the father’s education is almost as important as
the mother’s education; but in terms of the magnitude of effect, the mother’s
education is a more decisive factor (Hobcraft et al., 1984).
The father’s education shows up in the World Fertility Survey (WFS) and
DHS as an important factor in increasing child survival. Indeed, in some
countries it appears to be more important than the mother’s education, or even
more important. The Fathers’ education is so strongly interrelated with their
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
incomes that any attempt to control for income would not be wholly successful.
Therefore, demographers mostly ignore the findings (Caldwell, 1994). They are
also aware that mothers are usually more sensitive to children’s illnesses than are
fathers, and this is undoubtedly correct as measured by the likelihood of first
noticing that something is wrong (Caldwell et al. 1989a). Another study shows
that the estimated net effect o f higher maternal education is about twice as large
as that of paternal education (O’Toole et al. 1991).
2.1 J Place of Residence (Urban-Rural):
Another important factor of variation in infant and child mortality is the
place of current residence of the parents in developing countries including
Pakistan (Martin et al., 1983: 428; Mahmood, 1992; Alam and Cleland, 1984:
208; Hobcraft et al., 1983: 202-204; Trussell and Hammerslough, 1983: 21). The
analysis of the Pakistan Contraceptive Survey data also shows that children
whose parents live in rural areas have higher infant and child mortality. The
lower mortality in urban areas may be mainly due to the better health conditions
and access to medical facilities available in urban areas (Mahmood, 1992).
In Ghana, a study indicates the existence of large urban-rural differentials
in infant and child mortality, which the author argued may partly be explained in
terms of access to and availability of health care facilities or it may reflect the
educational differences between the urban and rural areas (Tawiah, 1989).
Bicego and Boerma (1993) also found urban-rural differentials in child health and
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
argued that the advantage of maternal education is more pronounced in rural
areas than in urban areas.
However, another study in Sri Lanka reveals that children of parents
living in urban areas and estates have child mortality 14 percent and 40 percent
higher, respectively, than the children of parents living in rural areas. It appears
that living in rural areas seems to provide the healthiest environment for children
in this country (Cochrane, 1980).
2.2 Household Environmental and Hygiene Factors:
2.2.1 Source of Drinking Water and Sanitation:
Improvements in household environmental conditions, especially the role
of piped water and sanitation facilities, in increasing the chances of survival is
clear in United States (Gaspari and Woolf, 1985) and Europe (Preston and Van
de Walle, 1978) in the nineteenth century. In 1964, Johnson reported that two of
the most important causes of poor health among people of developing countries
rest on the inadequate provision for sewage disposal and contamination of
drinking water (Johnson, 1964). The importance of clean drinking water in the
dwelling is greatest to child survival when breastfeeding is stopped and the child
is weaned. The protection against infection conferred by breastfeeding is no
longer available and the child is more exposed to infections. The improvement in
the quality of water and the provision of sanitation facilities for safe disposal of
human excreta is particularly important for the health of children.
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Access to piped water in the household is likely to be o f direct benefit in
lowering child mortality by reducing exposure to water related diseases like
diarrhea, cholera and typhoid (Merrick, 1985). These are directly transmitted
when water contaminated by faeces or urine is drunk or used in the preparation of
food (Poppel and Heijden, 1997). Access to safe water may also reduce the
impact of disease that is transmitted by the fecal oral route by leading to better
hygiene through washing hands and cooking utensils and cleaning floors (Esrey
and Habicht, 1986). According to UNICEF, the benefits of improved water
supply occurs not only by a reduction in diarrhea, but also by increased time
spent in food procurement and feeding activities when water collection times are
reduced (UNICEF, 1997).
Because so many of the major infectious agents o f disease are spread
through faeces and urine, hygienic disposal of waste is vitally important. Thus,
improvements in water quality alone, without improvements in hygiene and
sanitary conditions, might not lead to better health (Poppel and Heijden, 1997).
Excreta disposal probably play a more important role in determining children’s
health in developing countries than does water quality, especially where the
prevalence of diarrhea is high (Esrey and Habicht, 1986).
Several studies in developing countries also demonstrated that infant and
child mortality is closely associated with water supply and sanitation facilities
(Timaeus and Lish, 1995; Gubhaju et al., 1991; Merrick, 1985). For example,
infant mortality and household environment were found to be strongly associated
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in urban Nepal (Gubhaju et al., 1991). The risk of death was higher among
infants bom to households which used drinking water from a river or a lake than
among infants bom to households using piped water after controlling for other
socioeconomic and demographic factors. They also found that the risk of death
was higher among infants bom to households which did not have toilet facility
than among their counterparts bom to households which had such facilities
(Gubhaju et al., 1991).
A DHS comparative study of 17 developing countries demonstrated a
significant relationship between environmental and domestic hygiene with
respect to the prevalence of diarrhea, which is a leading cause of child mortality
(Boerma at el., 1991). In Brazil, Merrick (1985) found that access to piped water
plays an important role in reducing early childhood mortality during the period of
1970 to 1976, but the role of toilet facilities was found to be relatively minor.
Toilet facilities are associated with lower child mortality in Sri Lanka (Trussell
and Hummerslough, 1983), and with postnatal mortality in Malaysia (Da Vanzo
et al., 1983). Child mortality and morbidity have been strongly associated with
household environmental conditions in urban areas of Egypt, Brazil and Thailand
(Timaeus and Lish, 1995). After controlling for socioeconomic status variables,
environmental conditions of households were strongly associated to mortality in
Egypt and Brazil and to diarrhea prevalence in Brazil, Thailand and Ghana
(Timaeus and Lish, 1995).
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The United Nations study found deviations in different areas of Africa.
In Ghana, for example, better housing structure does not improve the survival
status of children. In Sierra Leone, the child mortality in houses with water taps
in the household is higher than that associated with water supply from public
taps. In Sudan, the mortality levels were higher for children where the pipe water
was available in the compound compared to the village piped water (United
Nations, 1985). Piped water in some areas is more subject to the risk of
contamination relative to the water from wells or streams, particularly where the
water-supply system is not well maintained (Defo, 1997).
Housing quality affects child mortality in Costa Rica (Haines and Avery,
1982). The risk of pneumonia, a major killer among children under five, was
found to be strongly associated with the quality of housing in the Philippines
(Johnson and Nelson, 1984). In Bangladesh, Rahman et al. (1985) studied the
impact of the source of water supply, presence of toilet, and persons per room and
the risk of infant mortality and found that the risk of post-neonatal mortality in
the households having toilet facilities was 3.12 times lower than those without
toilet facilities, and 1.5 times higher in the households with 10 or more persons
than in smaller households. On the other hand, some studies found that water
supply and toilet facilities have no effect on mortality after controlling for
socioeconomic variables (United Nations, 1985). It seems that the effect of
household environment on health is conditioned by several other characteristics
and behaviors of the household and the community (Woldemicael, 2000). The
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
effect of water supply and toilet facilities on child health may vary between
individuals or localities depending on the household income (Timaeus and Lish,
1995).
Some studies suggested that the effect of household environment on
health is conditioned on parental education (Esrey and Habicht, 1986) and also on
child-feeding practices (Butz et at., 1984), as well as or cultural beliefs and
practices (Togoe,1995). After controlling for maternal education, economic
status, as well as religion and ethnicity, Habtemariam (1994) found a negligible
effect of water supply and toilet facilities on child mortality in urban Ethiopia.
However, other studies have shown that the effect of poor housing
environment on child mortality is greater for children of the poor socioeconomic
groups than those of the better-off, suggesting that this latter group have the
resources to protect their children from infectious diseases that lead to death
(Timaeus and Lish, 1995). Some studies also show that the effect of poor
housing environment on child mortality is greater for poor children than those of
higher socioeconomic groups. It suggests that the persons of higher
socioeconomic conditions have the resources to protect their children from
infectious diseases that lead to death.
Butz et al. (1984) argued that the effects of breastfeeding and water
supply were strongly interactive and that they changed systematically during the
course of infancy. Infants bom into households with piped water experienced
significantly lower neonatal mortality compared with infants in houses using
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
other water sources. They found that the presence of piped water was associated
with significantly lower mortality only for infants who were breastfed little or not
at all. They also found, that in nearly all instances, the presence of toilets was
more important than the presence of piped water. As a result, breastfeeding
declines were less harmful in households with either toilets or piped water, and
much less harmful in households with both.
Merrick (1985), who studied the role of water supply in child mortality
differentials in Brazil in 1970 and 1976, stressed the need to take into account
variables associated with the real consumption o f this particular service by
households. A household may not consume a service such as water because the
supply network is not geographically accessible or because that household
chooses not to do so for reasons of price or income limitations.
The fact that the studies assessing the child health effects of water
supply have resulted in contradictory outcomes is also partly due to the
methodological deficiencies of many of these studies. Lindskog and collegues
(1987) argued that the design and method o f data collection had an effect on
the results. They concluded that positive effects o f water supply were shown in
those studies, which were based on ‘weaker’ methodologies. Several well-
designed studies failed to show clear-cut results, probably because a change
was multi-factorial: an intervention was only one o f several influential factors
(Lindskog et al., 1987).
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The nutritional status o f children can be associated with household
sanitation and hygiene, but there are inconsistencies reported in different studies.
For example, having a pit latrine versus no toilet facility was not significant,
whereas having access to piped water, including distinctions between in-house
taps and public standpipes, was found to be significantly associated with child
heights in three of the urban areas but in none of the five rural areas of five
African DHS countries (Bateman et al., 1993). In Ghana, a study found that child
height and weight was significantly lower in households with pit latrines than in
households, with no such facility (Annan, 1985). In the Philippines, children are
taller in communities in which excreta is not found in public places (Barrera,
1990).
2.2.2 Household Density:
The level of crowding in a house is not only a proxy for general social
standing; it may also have a direct effect on mortality through the transmission of
infectious diseases (Bernhardt, 1992). The study in Cameroon suggests that the
hazards of health arise mainly from overcrowding, unhygienic methods of
construction, inadequate and impure water supplies, and the arrangements of
cooking and storage of food (Defo, 1997).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.3 Demographic Factors:
Maternal factors, which are biological attributes of birth, such as the age of
mother at the time of childbirth, birth order and birth interval (Forste, 1994; Rutstein,
1984), have significant effects on child survival. Infant and child mortality are also
affected by the sex of the child, and infants bom to mothers who have lost a child are at
greater risk of dying during infancy (Cleland and van Ginneken, 1988). Moreover, some
of the studies found that, within a family, deaths of infants are correlated (Curtis et al.,
1993; Das Gupta, 1990; Gubhaju, 1985; Majumder, 1989).
2 J .l Maternal Age:
An extensive literature exists on the relationships between age of the
mother and the risk of infant mortality. The analysis of 29 World Fertility Survey
countries shows that the risk of mortality is higher during the first year of life of
children bom to young mothers and gradually decreases to reach a minimum for
women 25-30 years old in most of the developing countries (Rutstein, 1984),
except Pakistan and Bangladesh where it is at a minimum for women 35-39 years
and thereafter rises again (Alam and Cleland, 1984; Al-Kabir, 1984; Rutstein,
1984). Teenage births are at a higher risk of death relative to births of women 20-
29 years of age, and the risk is even higher in countries that have enjoyed some
success in reducing infant mortality rates (McDevitt et al., 1996).
Teenagers are also prone to pregnancy-related complications (Akther et
al., 1996). Mothers who are extremely young can also be socially and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
psychologically immature to take on many of the responsibilities involved in
infant care (Balk, 1997).
2 .3.2 Birth Order:
Child mortality shows different levels by birth order, starting high with
earlier birth order, then falling and rising again with higher birth order. Nutrition
and economic factors may explain the differential in mortality rates by birth
order. The higher mortality for first births may be due to the mother’s
inexperience in childcare, or due to specific conditions that make child bearing
difficult. By contrast, the children of higher birth order are more likely to be bom
to older mothers who are physically worn out and these children face competition
from older siblings for food and other resources (Rutstein, 1984). The higher
infant mortality risk for first births may also be explained by the fact that a higher
proportion of younger women have first births (Gubhaju, 1986).
The distribution of household resources depends on the number of
children in the household. Hence, a large number of children indicates a smaller
share of scarce resources. Lack of adequate resources to meet the nutritional
requirements of growing children may result in under-nutrition. This relationship
between higher birth order and stunting was found to be positive among children
in the Philippines (Horton, 1988).
A mother’s poor health and poor nutritional status may also have post
natal consequences, like impaired lactation (Retherford et al., 1989; Gubhaju,
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1986), and render her unable to give adequate care to her children. First-born
children o f very young mothers are at risk of dying, while infants because o f their
mother's physical immaturity (Gubhaju, 1986).
23 3 Preceding Birth Interval:
Another determinant of the maternal factors affecting child survival is
birth interval. A short interval has traditionally been viewed as a risk factor for
poor pregnancy outcomes, particularly infant mortality in developing countries
(Winikoff, 1983). It has been observed in several studies that the death risks of
an index child whose birth closes a short birth interval are higher than those
experienced by an index child whose birth closes a longer birth interval (Hobcraft
et al, 1985; Palloni and Tienda, 1986). It has been observed that children bom
within the preceding interval of 18 months experienced higher mortality risks
during infancy than those bom in an interval of two to three years (Cleland and
Sathar: 1984; Winikoff, 1983).
Previous studies have concentrated on three main mechanisms for the
birth interval relationship with mortality. First, maternal depletion syndrome,
which postulates that short repeated pregnancies, close spacing, and extended
periods of breastfeeding resulted in a reduced quality o f mothers’ milk and a
general decline in maternal health, evidenced by progressive weight loss and a
prematurely aged appearance (Jelliffe, 1976; Jelliffe and Jelliffe, 1978). The child
succeeding the short interval is, thus, disadvantaged as result of fetal malnutrition
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and a compromised intrauterine environment. A mother with repeated
pregnancies, especially at short intervals, does not have sufficient time for
recovery, both, physically and nutritionally, and is more likely to have pregnancy
losses and babies of lower birth weight (Jelliffe, 1966). This also may be due to
the amount of pressure on the mother for the care of the children.
Secondly, competition among siblings is considered a plausible
mechanism in the association between birth intervals and child survival: the
newborn child has to compete with other young siblings for household resources
and the mother's care. The situation may have a bearing on the nutrition of the
youngest child (Winikoff, 1983; Boerma et al., 1992). If there are two or more
young children close in age within a family, it is necessary for them to compete
for resources and for maternal attention. Boerma and Bicego (1992) suggested
that this could have an impact on the nutritional status of the index child, on
incidence of morbidity, and on higher fatality from illness as well as accidents.
Finally, it is also suggested that there is higher exposure to infectious
disease for the younger child (Boerma and Bicego, 1992). The older sibling of a
short birth interval reaches an age in which infectious disease is particularly
prevalent just as the younger child is vulnerable because immunity acquired from
the mother has declined while the infant has not yet fully acquired its own.
Investigators have questioned whether "maternal depletion syndrome" is
due to childbearing patterns (short birth intervals) or to inadequate food intake
(Winikoff and Castle, 1992). Postpartum stress may influence births following a
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
short interval because the care of an infant or very young child may place such a
physical and/or emotional strain on the mother that it interferes with the growth
of the fetus or the length of the subsequent pregnancy.
Similarly, breast-feeding could potentially be a confounding factor since
it affects, both, child survival and the length of the birth interval. Children with
short preceding birth intervals are less likely than others to have ever been breast
fed (Retherford et al., 1989). However, Palloni (1989), using data on breast
feeding collected in the World Fertility Survey for Peru, found that breast-feeding
is probably not the main mechanism through which the birth interval effect
operates.
A recent study in Bangladesh suggests that the excess infant and child
mortality associated with short birth intervals, however, is relatively modest, and
the basis for a health rationale for spacing births remains hampered by the lack of
a clear understanding of the mechanisms underlying the association between
short spacing and child mortality (Rahman, 1998). Cleland and Sathar’s analysis
of Pakistan WFS data found that children bom two or less years since the last
birth were twice as likely to die than those bom four or more years later. This
interval effect persisted whether the prior child died or not, suggesting to these
researchers a biological mechanism, such as maternal depletion, rather than
competition between siblings as the intervening mechanism (Cleland and Sathar,
1984).
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Studies also noted that the birth interval is associated with child growth.
In rural Bolivia, a large difference in stunting was found by the length of the
preceding birth interval. The risk of stunting was higher among children who
were bom within 24 months o f the preceding birth (Boerma and Bicego, 1990).
Similar results were found for Asian and Latin American countries (Mueller,
1984). Desai confirmed this relationship in her study of child nutrition in Africa
and Latin America (Desai, 1992).
2 .3.4 Sex of the Child:
In most contemporary populations, male mortality is higher at all ages
than female mortality (Heligman, 1983). This pattern has also been observed in
other populations in the past (Preston, 1976; Johansson, 1990) as well as in some
contemporary populations outside South Asia (Makinson, 1986). Apparently, this
is attributed to the male’s higher degree of biologically-based susceptibility to
disease and the higher prevalence of poor health habits and risky behavior,
especially among adolescents and adult men (Waldron, 1986). In many parts of
South Asia, the picture is quite different. After the neonatal period, female
mortality rates are higher than those for males throughout childhood and often
throughout adulthood until the end of the reproductive period (D’Souza and
Bhuiya, 1982; Dyson and Moore, 1983; Das Gupta, 1987; Basu, 1989).
Evidence from studies in India and Bangladesh indicates that higher
female mortality rates in childhood after the neonatal period result from
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
preferential treatment by family members for sons (Chen et al. 1981; Miller 1981;
Sen and Sengupta 1983; Bhatia 1983; Das Gupta 1987; Basu 1989; Freed and
Freed 1989).
The observation that male mortality rates are higher than those for
females in the neonatal period is consistent with this explanation, since most
causes of death in the first month of life are either beyond families’ immediate
control or are not due to sex-specific treatment of children. After the neonatal
period, environmental factors that are under control of the family such as
nutritional intake, exposure to disease, breastfeeding, parental time and attention,
and use of health-care services become predominant.
Surprisingly, Chen et al. (1981) show that, despite differences in diet and
other forms of care, the prevalence of illnesses, including diarrhea, is roughly
equal among girls and boys aged 0-4 years in Bangladesh. However, boys are
twice as likely than girls to receive treatment for an episode of diarrhea, even
though the treatment and transport to the clinic were provided free by the project
in Bangladesh. Das Gupta (1987) also showed that families in her 1984 study
spent less on medication and clothing for girls than for boys.
2.4 Nutritional/Dietary Factors
Calories, protein, and the micro-nutrients available not only to the child but also
to the mother influence the survival of the child. Birth weight and quality of breast-milk
are affected by maternal nutrition during pregnancy and lactation, respectively.
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.4.1 Breastfeeding:
Breastfeeding has numerous bio-demographic, social, and economic
effects. It affects the health and nutritional status of, both, the mother and child.
The role of breastfeeding is very important in the post-neonatal period. Energy
intakes for children under two years of age are determined more by feeding
practices such as the frequency o f feeding and the energy-density of foods offered
and less by household food availability (Brown and Begin, 1993). Due to the
small stomach size, young children need to have frequent, high energy-density
meals in order to consume sufficient amounts of nutrients for optimal growth.
Breastfeeding also benefits the mother by helping the uterus to retract,
which reduces postpartum blood loss and, by delaying the return of the menses,
thereby preventing subsequent closely spaced pregnancy (Jelliffe and Jelliffe,
1978). Therefore, effects of breastfeeding are critical determinants of child
survival and reproductive health in developing countries.
Based on the empirical findings of studies in developing countries,
breastfeeding practices have at least three known mechanisms by which
breastfeeding contributes to infant health and survival. First, breastfeeding
contains the optimal combination of nutrients, which suits the baby’s metabolic
structure. Second, breastfeeding allows the mother to pass on immunities that she
herself acquired to the baby. Third, breastfed children receive less of other food
and liquid, which may be contaminated with disease-causing agents (Cabigon,
1997; Yoon et al., 1996; Palloni and Tienda, 1986; Jelliffe and Jelliffe, 1978).
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The beneficial effect of breastfeeding to the health of infants was
suggested early in this century (Jelliffe, 1955) and later confirmed in several
studies that it had a protective effect against diarrhea in infancy (Feachem and
Hoblinsky, 1984; Habicht et al., 1986; Huttly et al., 1987; Victoria et al., 1987;
Briend et al., 1988; Brown et al., 1989). The protective effect of breastfeeding
tends to be stronger during infancy than after the age 1 year (Cabigon, 1997;
Majumber, 1991). This effect is more pronounced in populations with higher
infant mortality and poor socioeconomic conditions as well as for children living
in rural areas of poor families and of less educated mothers (Goldberg et al.,
1984; Palloni and Tienda, 1986; Ratherford et al., 1989). Breastfeeding also
provides important protection against infectious diseases (Jelliffe and Jelliffe,
1978; Feachem and Koblinski, 1984; Jason et al., 1984; Cunninghan et al., 1991;
Victoria, 1996), which account for over two-thirds of the 12 million annual
deaths in children younger than 5 years in less developed countries (Murray and
Lopez, 1996).
A set of guidelines for infant feeding in developing countries is
recommended by the WHO and UNICEF. The basic recommendations are:
initiation of breastfeeding within about one hour of birth; infants should get only
breast milk up to 4-6 months of age; starting at age 4-6 months in addition to
breast milk, adequate and appropriate supplementary foods should be given;
breastfeeding continue in combination with supplementary foods up to the second
birthday or beyond (WHO/UNICEF, 1994, WHO, 1991).
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Infants put to the breast later than the second day after birth have the least
chance of receiving the mother’s colostrums and are also the most likely to be
given prelacteal feed. Withholding of breastfeeding for the first three to five days
after birth make the child vulnerable to morbidity and mortality (Bhatia, 1981:
70). In some cultures colostrums is regarded as a harmful substance that is
capable of causing diarrhea, pneumonia or other illness (Jelliffe, 1962). In other
cultures colostrums may be viewed as simply inadequate and are withheld until
the point at which the mother judges her milk is mature (Mata, 1978). A study in
Honduras shows that the delayed initiation of breastfeeding may also result in the
newborn being provided with other sources o f nourishment that can introduce
infection and delay the milk arrival (Perez-Escamilla et al., 1996). In Malaysia it
was found that during the first month of life, infants who were breastfed
throughout their first week of life had neonatal mortality 16 per 1000 lower than
those not breastfed (DaVanzo et al., 1983).
After the first six months, the child breast milk alone does not satisfy the
energy, protein, and micronutrient requirements of most infants. Therefore,
infants need extra nutrition and it has been found that the weaning period is the
most critical period for the child with respect to mortality and morbidity
(Winikoff, 1983). However, during the transitional period when complementary
foods are being introduced, on-demand and frequent breastfeeding should
continue to ensure that infants receive all the benefits of breastfeeding.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Exclusive breastfeeding (breast milk as the only source o f infant food or
liquid), meets the nutritional requirements of the infants (Cohen et al., 1994).
Exclusive breastfeeding also satisfies fluid needs even in hot and humid climates
(Sachdev et al., 1991). It also protects against illness for about the first six
months of life (Huffman and Combest, 1990). A study in Brazil by Victora and
colleagues found that exclusive breastfed infants are 14 times less likely to die
from diarrhea compared to infants fed by formula milk, and 4 times less likely to
die compared with partially breastfed infants (Victora et al., 1987).
Early introduction o f supplementary foods, before 3 months of age, is
reported to lead to lasting growth deficits (Piwoz et al., 1994). On the other hand,
consumption of supplementary food from 3 months on may have no long-term
adverse effects on child growth (Rowland et al., 1985; Neumann and Harrison,
1994). In the Philippines, Barrera (1991) reports lower weight-for-age in
children supplemented either before 4 months or after 6 months, but no
association with the total duration of breastfeeding.
The child’s ability to digest food other than the mother’s milk and resist
disease-causing agents increases with age. The benefits of breastfeeding in
combination with supplemental food continue well beyond the first year of life.
Even with a high prevalence of malnutrition and poor sanitation, breastfeeding
substantially enhances child survival up to 3 years of age (Briend et al., 1988;
Molbaketal., 1994).
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Bangladesh, breastfeeding was associated with lower mortality up to
the age o f 3 years in malnourished children (Briend et al., 1988) and a lower
prevalence and severity of bloody and chronic diarrhea (Clemens et al., 1986;
Victoria et al., 1987). Other studies have associated prolonged breastfeeding with
reduced food intake and malnutrition (Dettwyler, 1987; Victoria et al., 1984;
Brakohiapa et al., 1988; Michaelsen, 1988). Because of the association between
the low state of nutrition and prolonged breastfeeding, it has been suggested that
breastfeeding after 18 months o f age may be detrimental (Brakohiapa et al.,
1988).
A study which analyzed the impact of breastfeeding on diarrheal disease
and survival in children above one year of age in Guinea-Bissau (Molbak et al.,
1994) shows that there was a higher rate of diarrhea in weaned than in breastfed
children above one year of age; the relative risk was 1.4 in one year old children
and 1.7 in two year olds. When children who had a younger sibling were
excluded, there was no effect on these results. Their results show that during the
second year of life the incidence decreased from three to two episodes per 100
days at risk in breastfed children, whereas the incidence was higher and declined
more slowly in the weaned children (Motbak et al., 1994).
Motbak and colleagues suggested hat the beneficial effects of
breastfeeding are not restricted to infancy. Though children who are partially
breastfed after infancy may have a lower state of nutrition than the weaned
children, the benefit in terms of lower morbidity may be more important for child
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
survival in places with a high morbidity from diarrhea and with high mortality
(Motbak et al., 1994). A recent analysis of 37 DHS countries on breastfeeding
shows that exclusive breastfeeding is widely practiced in many countries in the
first month of life, but declines with each additional month. After 5 months of
age, exclusive breastfeeding declined to single digits in some countries and
between nonexistent in most countries (Haggerty and Rutstein, 1999). The range
of exclusive breastfeeding is greatest in sub-Saharan Africa, from I percent in
Niger to 90 percent in Rwanda. In Asia, it ranges from 26 percent in Pakistan to
89 percent in Nepal.
2.4.2 Maternal Malnutrition:
Maternal malnutrition also affects the infant's nutritional status, as well as
its survival status, through its effect on breast milk and duration of breastfeeding
(Wray, 1978). Pregnant women who receive inadequate nutriments are likely to
have underweight babies, and are more likely to get infectious diseases leading to
early death. Those who survive but receive inadequate food in their early life are
more likely to be exposed to the permanent stunting (Bender and Smith, 1997).
2.5 Health Seeking Behavior
Health seeking behavior includes, both, preventive and curative measures.
Preventive measures include immunization against preventive diseases such as
tuberculosis, polio, measles, neonatal tetanus and smallpox, whereas curative measures
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
include the care and types of treatment undertaken for specific conditions, both modem
and traditional (Mosley and Chen, 1984: 28). “Preventive measures” can also be
measured by the numbers of antenatal visits by medical personal and whether the birth
was attended by medical personnel as well as the place of delivery. Since education is
believed to enhance women’s assertiveness, confidence, ability to deal with others within
and outside the home, and willingness to seek services (Caldwell, 1979 Caldwell et al.,
1983), maternal education is seen as an important determinant in the use of health care.
Studies from Bangladesh and India suggest that excess female mortality is caused
by gender differentials in the treatment of illness, rather than differentials in food intake
or health status in the home (Basu, 1989; Fauveau et al., 1991). Becker and colleagues
(1993) reported in the study of the rural Philippines, that each year of maternal education
increased the odds o f receiving some prenatal care by six percent and the odds of
receiving adequate care by 19 percent. They defined prenatal care as a three-category
variable showing no care, some care, and adequate care. The study found stronger effects
of the husband’s education in the urban Philippines. A comparative study of three DHS
countries showed larger net effects of maternal education on prenatal care in Bolivia and
Egypt after adjusting for the effects of various factors (Sommerfelt and Stewart, 1991).
Bicego and Boerma (1991) found that relative risk of receiving no prenatal care was
higher among uneducated women compared to educated women in developing countries.
With respect to receiving tetanus injections during pregnancy, Sommerfelt and
Stewart (1991) found only small differences by maternal education in Kenya and Egypt,
while they found substantial differences in Bolivia. Bicego and Boerma (1991) found a
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stronger relationship between receiving tetanus toxoid during pregnancy and maternal
education in rural areas than in urban areas. They also suggested that educated women in
urban areas perceived tetanus injections as unnecessary while, in contrast, educated rural
women availed themselves of such facilities. Educated mothers are more likely to deliver
their babies in clinics or hospitals compared to uneducated mothers.
In Pakistan, an average of 92 percent of the infants are bom under the medical
supervision of a Dai (midwife) or some relatives. They are usually breast-fed for at least
one year and probably half of them do not get any supplementary food until age one.
The rate of contraceptive use is very low and the infant and child mortality very high.
The analysis of the PDHS shows that the infant and child mortality is lowest for the
children whose mothers had received prenatal care or whose delivery was attended by
medical personnel. On the other hand, the infant and child mortality was recorded highest
for those whose mothers did not receive antenatal care and delivered their babies at home
and were attended by Dais, Traditional Birth Attendants (TBAs) or some relatives (NIPS,
1992). The PDHS also found that three-fourths of the pregnant women received no
ante-natal care at all and only 30 percent saw a medically qualified person. The vast
majority of deliveries (85.2 per cent) took place at home and most of them were attended
by Dais or TBAs (Mahmood, 1992).
Proper medical attention and hygienic conditions during delivery can reduce
the risk of infections and facilitate management of complications that can cause death
or various illnesses for the mother or the newborn child (Mitra et al., 1997).
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In urban slums of Karachi, the largest city of Pakistan, Acute Respiratory
Infection (ARI) is the second leading cause o f death for children under two years. One
of the associated causes of death is malnutrition. In Pakistan and many other cultures,
infants are not taken out of the house even when ill for fear o f exposing them to
further risks (Bang et al. 1990). PDHS data shows that one in six children had suffered
from the symptoms o f ARI during the two weeks preceding the survey. Among those
two-thirds were taken to a health facility or health provider for treatment.
2.6 Nutritional Status of Living Children:
The nutritional status of an individual refers to the availability of energy and
nutrients to the cells of the body relative to their requirements. Malnutrition is a result of
an imbalance in energy or nutrients at the cellular level, including deficiencies and
excesses (Pelletier, 1998). Growth Faltering (height relative to age) among children in
developing countries is highly prevalent (Dewey, 1998; Wells et al, 1993). Although
globally there has been a reduction in the prevalence of underweight among children
under five years (UNFPA, 1998), the prevalence of stunting is still alarmingly high.
Nearly forty percent of pre-school children are moderately stunted (UNFPA, 1998).
Malnutrition is common and makes an important contribution to the burden of morbidity
and mortality of infants and children in developing countries (Pelletier, 1998). Several
studies have recorded an increase in the prospective risk of death in growth-retarded
children, with a curvilinear relationship that varies slightly in the various regions of the
world.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Punjab, Kielmann and McCord (1978) found that children between 1-36
months of age with weight-for-age between 70 and 80 percent of the reference had a risk
of death four times higher than children with weight-for-age at or above 80 percent. In a
Bangladesh study, at the age o f 13-23 months, children with height-for-age below 85
percent of the reference or with weight-for-age below 65 percent of the reference had a
mortality between three and seven times higher than normal children (Chen et al., 1980).
The adverse effects of growth faltering on children’s health and development
have been widely studied. Growth-retarded children have greater susceptibility to
infections (Black et al., 1984; Rivera and Ruel, 1997), higher mortality rates (Martorell et
al, 1994; Vella et al., 1993), and increased likelihood of impaired cognitive development
and reduced school performance (Pollitt, 1992; Sigman et al., 1989).
Severe stunting can cause a variety of problems throughout life (Osmani, 1999).
For instance, the severely stunted children will be more susceptible to recurrent
infections and will also be more likely to suffer retardation in cognitive development.
Children who are stunted early in life and who continue to reside in the same deprived
environment never catch up to their full growth potential (Martorell et al., 1994) in
adulthood. They will have lower physical capacity to work during their adulthood
(Spurr, 1998). They will also be more susceptible to morbidity and premature mortality
in comparison with adults of normal height (Osmani, 1999). Moreover, stunted girls who
survive the harsh environmental conditions and end up as short adults have increased
obstetric risks due to their small stature, and are more likely to have poor pregnancy
outcomes (Leslie, 1991; Martorell et al., 1981).
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Research has identified two vulnerable periods during which growth faltering
commonly occurs (Brown and Begin, 1993). First, during intra-uterine development,
genetic, endocrine and maternal factors can affect prenatal growth. Maternal factors
affect not only birth weight and birth length, but also growth afterward (Dewey, 1998;
Kusin et al, 1992). Mothers who suffer serious under-nutrition during pregnancy tend to
produce babies who carry a life-long disadvantage in terms of both physical and mental
health (Osmani, 1999). These mothers also produce lower birth-weight babies, who tend
to die sooner than babies of normal birth-weight (Pan American Health Organization,
1997). This association constitutes an intergenerational effect because low-weight is
related to growth deficits in the same individuals at later ages (Pan American Health
Organization, 1997).
Second, growth faltering occurs during the weaning period, which generally
coincides with the second 6 months of life. The relatively poor quality of complementary
foods is one factor implicated in growth faltering during this period (Dickin et al, 1990;
Mata et al, 1972; Moy et al, 1994). Understanding the factors that affect growth during
these critical periods is essential for determining appropriate intervention strategies to
address this problem.
The impact of genes on growth has been studied and it is concluded that, in
malnourished populations, the explanatory power of genetic factors is likely to be very
small relative to the large effects which poverty produces (Martorell, 1985; Rona et al.,
1995).
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.6.1 Causes of Malnutrition:
The imbalance of energy and nutrients at the cellular level can arise from
inadequate nutrient intake and/or disease, such as diarrhea, measles, malaria and
other parasitic infections which may affect nutrient absorption, transport, storage
and utilization. The causes of malnutrition and ill health are the state of
household food insecurity, the inadequacy of health services, the inadequacy of
maternal and childcare, and an unhealthy environment (UNICEF, 1990). This
causative role of childcare highlights the fact that malnutrition can arise even in
the presence of household food security and adequate availability of health
services. Conversely, as revealed by earlier research, good nutritional status can
be maintained even in the face of food insecurity, poor sanitation and the
associated conditions of poverty (Zeitlin et al., 1990), which shows that food,
health and care when considered individually, represent a necessary but not
sufficient explanation for the existence of malnutrition in the household and
community (Pelletier, 1998). However, these underlying causes are subject to the
availability of and control over resources as modified by political and economic
forces (UNICEF, 1990).
However, poverty is related to most of the immediate, underlying and
basic causes of malnutrition. The stunted growth is associated with poor living
conditions (Gopalan, 1986; WHO, 1995; Keller, 1991). According to Gopalan
(1986), stunting is a reflection of the attributes of the poverty syndrome. The
poverty syndrome involves low levels of income, inadequate diets, low-paid
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
occupations, poor levels o f education, poor housing conditions, large family size,
lack of environmental sanitation, and high prevalence of morbidity (Gopalan,
1986). In a particular context, any of these factors can be identified as immediate,
underlying, and basic causes of malnutrition.
2.6.2 Diarrhea and Malnutrition:
It is a fact that infections cause a deterioration in the nutritional status of
children. Infectious diseases are considered to play a major role in the causation
of malnutrition in childhood (Asenso-Okyere et al., 1997). Infections account for
a lot of the growth deficits in children in poor communities. Wasting has been
associated with widespread disease, diarrhea, and epidemics of measles (WHO
Working Group, 1986; Keller, 1991). Disease can slow growth by negatively
affecting energy balance through decreased nutrient intakes, increased nutrient
malabsorption, and increased nutrient requirements and expenditures. While
temporary growth faltering commonly results from acute morbidity, catch-up
growth may eliminate the deficit unless another disease episode quickly ensues
(Briend et al., 1989). Cumulative growth retardation is commonly found to be
associated with chronic diarrhea and dysentery, but not with short-lived episodes
of acute watery diarrhea, at least after infancy (Mata, 1978; Black et al., 1984).
A study which intensively followed a cohort of Mayan Indian children
from birth to 3 years, found that diarrheal diseases were very frequent and were
strongly associated with diminished food intake and growth faltering (Mata,
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1978). The same relationship was also found in the study o f Guaymi Indians in
Costa Rica (Mata, 1980). Other studies in developing countries also found that
diarrheal diseases were a major cause of growth retardation in young children in
The Gambia (Rowland, et al., 1977). Similarly, a decrease in expected weight
gain was found in Bangladeshi infants during the period o f high diarrheal disease
prevalence (Black et al., 1984a). Thus, these findings suggest that diarrhea
prevalence is associated with a significant decrement in both linear growth and
weight gain. However, the magnitude of growth faltering associated with
diarrhea may depend on the age of the child, the season, the etiologic agent,
dietary intake, food preparation and feeding practices (National Academy Press,
1992).
2.6J Malnutrition and Mortality:
Childhood mortality arises from the same causes as malnutrition itself
(Pelletier, 1994), which emphasized the interaction between disease and nutrient
intake but is explicit about certain pathways leading to malnutrition and mortality
(Pelletier, 1998).
In most developing countries the question is whether mortality risk is also
increased in mildly to moderately undernourished children who are not in a
hospital. If so, is it only because there is a progression towards more severe
malnutrition, or mild to moderate malnutrition in itself increases the risk of death
from non-nutritional causes such as infections (Pelletier, 1999)?
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
There have been studies that did not show an increased mortality risk in
moderate under-nourishment (Chen et al., 1980; Kasongo Project Team, 1986;
Smedman et al., 1987). Other studies have shown a gradual increase in the risk of
death with worsening nutrition (Lindskog et al., 1988; Kielmann and McCord;
1978; Sommer and Lowenstein; 1975; Heywood; 1982).
In Malawi, anthropometric assessment shows that the mortality risk was
related to weight-for-height and height-for-age. However, mortality among
children living in households with piped water tended to be lower than mortality
for those using traditional water sources but the difference was not statistically
significant (Lindskog et al., 1988). A cross-sectional anthropometry study in
Bangladesh indicated that severely malnourished children experienced
substantially higher mortality risk than normal, mild, and moderately
malnourished children (Chen et al., 1980).
Indonesian children of less than two years of age with a low height-for-
age were at greater risk of dying than children of the same age who were not
stunted. This risk declined with increasing age. The mortality risk associated with
mild wasting also declined with increasing age. However, the risk of dying
among moderately to severely wasted children increased with increasing age.
These results suggest that stunting (lower height-for-age), rather than wasting
(lower weight-for-age), puts younger children at greater risk of death, but among
older children, wasting carries a greater relative mortality risk over an 18-month
period (Katz et al., 1989)
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Bangladesh, each of the anthropometric indicators are significant
predictors of mortality even after controlling for several variables such as
maternal age, parity, religion, education, number of cows, type of floor are and
mother’s height and weight (Cogill, 1982). The relation between season and
mortality showed that mortality rates were highest during the main harvest of the
wheat crop, reflecting the effects of food scarcity, relative child neglect, and
climate on child deaths among those already underweight (Kielmann and
McCord, 1978).
Another analysis o f the United Republic of Tanzania and Malawi showed
that the effects of the anthropometric variables on mortality persist after
controlling for socioeconomic variables (Yambi, 1988, Pelletier, 1994c). Another
study reported that the relative risk estimates of mortality across socioeconomic
groups with relation to other groups of populations combined found elevated
relative risks of mortality for anthropometric indicators. This relation also holds
for the mild-to-moderate malnutrition groups (Vella el al., 1994).
After reviewing 28 research studies, from Asia and Africa relating
anthropometric indicators of nutritional status at one point in time to the
subsequent survival status of individual children aged 6-59 months, Pelletier
concluded that, with the exception of two studies from Zaire (Kasongo Project
Team, 1983; Van Broeck et al., 1993), all other studies show a significant
relationship between anthropometric indicators and subsequent mortality
(Pelletier, 1994a). In this analysis mortality increased by 5.9 percent for every 1
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
percent decrement in weight-for-age (Pelletier, 1994b). In these studies, weight-
for-age as a percentage of the median scale was used to measure nutritional
status. The results showed remarkable consistency in the relationship between
malnutrition and mortality across populations, despite the differences in ecology,
disease, culture, child-age ranges and absolute levels of mortality (Pelletier, 1998;
Bairagi, 1981; Ewbank and Gribble, 1993).
Thus, the evidence suggests that the relationship between anthropometric
data and mortality is not confounded by socioeconomic factors (Pelletier, 1998).
Furthermore, mortality is not only high in severely malnourished children, but is
also elevated among children with mild-to-moderate malnutrition (Pelletier,
1999), relative to adequately nourished children.
In the assessment of nutritional status: choice of index which best
represents the nutritional status; choice of reference population to compare the
observed anthropometrical values o f individual children; choice of best statistical
measure to express the nutritional status of children; and what is the suitable
cut-off points for the identification of children who are at risk, are very important
(WHO, 1979).
3.6.4 Different Approaches of Measuring Nutritional Status
There are three main approaches to measure under-nutrition in children
(Floud et al., 1990). First, nutritional status is measured by comparing food
intake to food requirements. Comparing measured food intake to neutrient
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
requirements for living and working in different conditions (Acheson et al., 1980;
Sukhatme and Margen, 1982), nutritionists suggested this is an impractical
method to directly measure nutritional status (Gorstein et al., 1994; Payne, 1987).
Pelletier (1998) argued that measurement of dietary intake is difficult over a short
period of time and almost impossible over a long period of time. Although
dietary intake is a contributor to nutritional status it is not a direct measure of
status, because it does not capture the effects of disease on the individual.
Clinical examination o f certain deficiency of certain vitamins and
minerals is the second approach of measuring under-nutrition. This is a useful
approach to determine severe under-nutrition such as kwashiorkor and edema.
However, this approach cannot be used to compare the nutritional status of
different populations (Floud et al., 1990).
The third approach in measuring under-nutrition is through
anthropometric measurements (Gorstein et al., 1994; WHO, 1995; WHO
Technical Report, 1995). The status of child nutrition is often examined by using
anthropometric measures on weight and height of children by the indices
weight-for-height weight-for-age and height-for-age. Each of these indices
provides somewhat different information about the child’s nutritional status
(Waterlow et al., 1977; Sommerfelt and Arnold, 1998). The basic perception
behind this approach is that children suffering from inadequate nutrition, whether
due to insufficient food intake or a disease that interferes with the body’s ability
to convert food into energy, either grow more slowly or stopped growing
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
altogether. The WHO recommended that children’s height and weight be related
to age and that weight also be related to height (WHO, 1986). Although these
indices are related, each has a specific meaning in terms of the process or
outcome of growth impairment (Dibley et al., 1987; WHO, 1995). In a normal
(well nourished) population, boys are taller and heavier than girls of the same
age. Therefore, the child’s sex is taken into consideration when making
comparisons with the reference population (WHO, 1986).
2.6.5 Statistical Measures of Nutritional Status
There are three different statistical measures to express the comparative
anthropometrical values to choose from, they are "percentiles of the reference
population", "percentage of median" or "number of standard deviation units".
2.6.5.1 Percentiles:
The percentile refers to the position of an individual on a given reference
distribution. Percentiles rank a child’s height or weight against a reference
population of the same age and sex so that a child who is 95th percentile indicates
that only 5 percent of the reference healthy children are taller or heavier. This
method is commonly used in clinics because the interpretation is straightforward
(WHO, 1995). This method is also useful to track the child’s growth pattern but,
because averages and standard deviations cannot be derived from percentiles, it is
an inappropriate measure for population-based assessments. Another
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
disadvantage of this method is that there may be a lack o f change in percentile
values near the extremes of the reference distribution when, in reality, there is a
substantial change in individuals weight or height (WHO, 1995).
2.6.5.2 Percentage of Median
The percentage of median is an expression of the observed child’s height
or weight as a percentage of the reference population’s median height or weight
for a given age and sex. The nutritional status inference is inconsistent for the
different age groups because the proposed cut-off points for percent of median
are different for each of the three commonly used indices. Therefore, this
expression allows only to report the means and standard deviations based on
distribution of the sample. For instance, to approximate a cut off points o f -2 Z-
score are different for height-for-age (90% of the median) and for low weight-for-
height (80% of the median) (Dibley et al., 1987a; 1987b; WHO, 1995).
2.6.53 Standard Deviations
WHO recommended that the child’s weight and height be expressed in
terms of Standard Deviation units that a child’s measurement deviates from the
median value of the reference population for children of the same age and sex.
Such standard deviation units are called “Z-Scores”. The distribution of the
international reference population recommended for use by WHO have been
developed by the US Centers for Disease Control based on data from the US
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
National Center for Health Statistics. These scores are standardized into Z-scores
and conventionally compared with international references. Z-scores are
calculated as (Dibley et al., 1987b):
Z-score = (Actual individual anthropometric value - median value
of reference population) / Standard deviation of reference population.
2.6.6 Choice of Reference Population:
This measure has been developed by the World Health Organization
(WHO), based on the data for North American children o f the same sex and age
provided by the United States National Child Health Statistics and Center for
Diseases Control (NCHS/CDC). There are several other anthropometric
reference data available, such as, Harvard reference growth data, the U.K.
reference data, the Canadian reference data, and the NCHS growth reference
standards. The NCHS is now considered the most suitable for use as an
international standard by the WHO and others (Waterlow et al., 1977:490).
2.6.7 Cut-off points:
Children who fall more than two standard deviations below the median
value of nutrition status for an international reference population, recommended
by the World Health Organization, are considered to be malnourished and
children who are more than three standard deviations below the reference
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
population are considered to be severely malnourished (Dibley et al., 1987a;
1987b; Waterlow et al. 1977: 494). In a healthy population, only 0.5 percent of
children would be found below -3 standard deviation points. It has also been
found that a cut-off point of -2 standard deviation correlated well with morbidity
and mortality (Graitcer et al., 1 981). However, there is always a margin o f error
and risk of misclassification because of biological differences among individuals
(Beaton, 1986).
2.6.8 Anthropometric Indices:
2.6.8.1 Height-for-Age:
Height-for-age is an indicator of past and also continuing or
long-term problems in inadequate diet and high morbidity (Sommerfelt
and Arnold, 1998). It is considered as a proxy for the health o f a
population (Tanner, 1982) and this measure is referred to stunting. This
term “stunting” is used by human biologists to denote retardation in
height growth relative to the standard for well-nourished people
(Martorell and Habicht, 1986). It results from recurrent and chronic
illness, especially when illenesses are not treated properly and from
inadequate food intake. Each time the child experiences inadequate
nutrition, the child will stop growing or grow more slowly. On the other
hand, when the child’s nutritional status improves, normal growth
resumes, but the child is unlikely to “catch up” to the height that would
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have been attained if the incidence of poor nutrition had not occurred.
Each time a child experiences poor nutrition, it falls further behind in
height relative to a child with no such experience. Thus, Height-for-age
measures a child’s cumulative nutritional status since birth.
Anthropometric data around the world suggest that a strong
relationship exists between the heights of children and adults in different
populations and the measures of socioeconomic development (WHO,
1995). It is observed that in a population, if the socioeconomic conditions
develop, heights of children and adults also increase. Stunting in
developing countries is a result of economic deprivation (Gopalan, 1986;
Keller, 1991; WHO, 1995). Poor environmental sanitation and unsafe and
inadequate water supplies place children at risk of long-term growth
failure or chronic under-nutrition (Bogin, 1999).
The interpretation of low height-for-age prevalence in a
community depends on its magnitude. When the prevalence of low
height-for-age is high, it may be safely assumed that most short children
are stunted. When the prevalence of low height-for-age is low, the causes
of short stature may lie either in genetic variation or factors relative to the
individual child (WHO, 1995; Tomkins, 1995).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.6.8.2 Weight-for-Height
Weight-for-height, indicates a deficit in tissue and fat-mass
(WHO, 1986) and is often referred to “Wasting” (Martorell and Habicht,
1986). This measure is independent of age of the child (Keller, 1991;
Gibson, 1990) and is therefore, useful in populations where the age of the
child is not available. Weight-for-height measures body mass in relation
to height. A child with a weight-for-height Z-score below -2SD is
considered very thin, i.e., weight is low in relation to the child’s height. In
the international reference population only a very small percentage of
children (2.3 percent) have a Z-Score lower than -2 standard deviations.
This percentage reflects the degree of wasting in a population. This is
mainly a result of insufficient food intake in the recent past. Wasting is
also seen immediately after illness episodes such as, after diarrhea.
Waterlow et al. (1977: 491) proposed weight-for-height as a
useful indicator to distinguish between children who are currently
experiencing malnutrition and those who have suffered malnutrition in
the past. Weight-for-height reveals nothing about past episodes of
malnutrition if the child is receiving adequate nutrition. Their weight will
be in the normal range for a child of their height. However, a stunted
child will have low weight compared with a healthy child o f the same age
because the stunted child’s weight has adjusted to its shorter stature. A
child currently experiencing malnutrition would also have lower weight-
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for-age. Therefore, weight-for-age is less useful than the other two indices
weight-for-height and height-for-age because it does not distinguish
between short-term and long-term malnutrition.
2.6.8J Weight-for-Age:
Weight-for-age being a reflection o f both linear and soft tissue is a
composite indicator of long-term (wasting) and short-term (stunting)
malnutrition (Waterlow, 1972: 566). In the short term, when severe
stunting is accompanied by low weight for height, there is a higher risk of
infection and greater severity of infection (Osmani, 1992). In the long
term, severe malnutrition, when combined with other adverse conditions,
can result in impaired mental development (Colombo et al., 1988).
Pelletier (1998) suggests that weight-for-age is most useful as an indicator
of current or recent nutritional status among young rather than older
children (Pelletier, 1998), because at older ages weight-for-age tends to
reflect stunting that occurred earlier in the individual’s life (Beaton et al.,
1990).
Because this index is a composite of height for age and of weight
for the height of a child, the interpretation is difficult (WHO, 1995). For
instance, height-for-age and weight-for-age are highly correlated.
Anderson (1979) found in surveys from five developing countries that the
correlation between these two indices ranged from 0.60 to 0.87 for
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children under five years o f age (Anderson, 1979). Thus, when the
prevalence of wasting in a population is low, weight-for-age and height-
for-age provide the same information (WHO, 1995). However, when the
prevalence of wasting in a population is high, the prevalence o f weight-
for-age does not necessarily parallel the prevalence of height-for-age
(WHO, 1995).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3
DATA AND METHODOLOGY
This chapter presents the data source, methodology and the definitions of
variables used in the analysis. This chapter is organized into four parts which describes
the data source and quality, conceptual framework, variables used, and the methodology
used in the analysis.
3.1 Data Source
The Pakistan Demographic and Health Survey (PDHS), conducted from
November 1990 to March 1991, is used to address the objectives of the study. The
primary objective of the PDHS was to provide national as well as provincial-level data
on population and health in Pakistan. The National Institute of Population Studies
(NIPS), on behalf of the Government of Pakistan and Macro International Inc. in
Calverton, Maryland, carried out all the activities in collaboration with the Federal
Bureau of Statistics, Statistics Division, Government of Pakistan. The funding was
provided by the United States Agency for International Development (USAID) and the
Government of Pakistan (Government of Pakistan, 1992).
3.1.1 Sample Design:
The geographic coverage of the PDHS was extended to the whole of Pakistan
excluding the Federally Administrated Tribal Areas (FATA), military restricted areas in
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the cantonments, and sparsely populated and highly inaccessible areas. The sample
represents all ever-married women aged 15-49 years living in private households in all
urban and rural areas of Punjab, Sindh, NWFP and Balochistan. The federal district of
Islamabad was included in the Punjab. The population covered by the sample represents
96 percent of the population of Pakistan.
A stratified two-stage sampling technique was used for all the four provinces.
Enumeration blocks demarcated by the Federal Bureau of Statistics for their sampling
frame for urban areas and Village/Mouzas for rural areas, constituted the primary
sampling units. Their selection was made on the basis of probability proportionate to the
number of households determined by listing or quick count survey. During the first stage
408 primary sampling units (225 urban and 183 rural) were selected.
The secondary sampling unit was households, which were selected by systematic
sampling with a random start. The urban population represents 32 percent o f the national
population but was considered to be more heterogeneous than the rural population, so 40
per cent of the sample was collected in urban areas. Hence, the sampling fractions for
urban and rural areas were different. A total o f 8,019 secondary sampling units, 3,384
urban and 3,227 rural households, were selected. Finally, a woman was defined as
eligible if she was between the age of 15 and 49, ever married, and present in the selected
household the night before the interview. About 6904 women were identified to
interview, among those 6611 were interviewed face to face. The individual response rate
was 95.5 percent in urban and 97.1 percent in rural areas (Government of Pakistan,
1992).
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.1.2 Questionnaires:
Three types of questionnaires were used in this survey: the Household
Questionnaire, the Woman’s Questionnaire and the Husband’s Questionnaire. The
contents of these Questionnaires were based on the DHS Model B Questionnaire. These
Questionnaires were designed for countries with a low level of contraceptive use. After
the finalization o f these Questionnaires with related ministries, the Questionnaires were
translated into regional languages (Punjabi, Sindhi and Pushto). The data for the children
came from the Woman’s Questionnaire. All the ever-married women aged 15-49 who
slept in the household the night before the household interview were interviewed.
Eligible women were asked questions about the topics: Background characteristics;
Reproductive history; Knowledge and use of contraceptives; Pregnancy and
breastfeeding; Vaccinations and the health of children; Marriage; Family size
preferences; and Husband’s background. In addition, the interviewing teams also
measured the height and weight of all respondents’ children who were less ihan 5 years
old.
3.13 Data Quality:
The fieldwork was carried-out by several interviewing teams. Each team
consisted of one field supervisor from the Federal Bureau of Statistics, one field editor,
three female interviewers, one male interviewer and one driver. The completed
questionnaires were checked and examined by the field editors in the field to ensure that
all necessary information was collected; otherwise, the information was collected by
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
revisiting the households. Throughout the fieldwork, survey staff maintained close
contact with all the teams through direct communication or spot-checking. The objective
was to provide support in the field and advice to enhance data quality and consistence as
well as the efficiency of interviewers.
A total of 6611 ever married women were interviewed and complete reproductive
histories collected to facilitate detailed analysis o f fertility and infant mortality. For each
child that a woman had bome, she was asked to report the date of birth, date of death (if
the child was dead) and sex. If exact dates were not available only the year or the month
was recorded and the age were also estimated from the other information provided by the
women such as their age at marriage and ages of her other children. In addition, detailed
information on 6407 live births regarding feeding practices, children's morbidity,
sickness care, immunization, birth attendance were asked of women for the births which
occurred during the five years prior to the survey. Anthropometric data on 4037
surviving children age 1-59 months were collected.
The measurement of children’s height and weight was undertaken after the
children’s shoes and clothes were removed. The interviewers were trained in weighing
and measuring the height or length of the child according to the guidelines in the United
Nations Manual “How to Weigh and Measure Children” (United Nations, 1986). They
were trained how to measure the children’s height (within 5 millimeters) using a
measuring board and how to weigh children (within 100 grams) using a hanging spring
balance scale. The height of a child under 24 months of age was actually recumbent
length, measured with the child lying down on an adjustable wooden measuring board as
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
recommended by the WHO. The same board was used to measure the standing height of
older children.
A total number of 4037 children, which represents about 80 percent of the 5776
living children, were weighed and measured. Other than the reason of absence of the
index child from home, culture was the most common reason for non-measuring of
children given by the mothers in two provinces, Balochistan and NWFP. In both
provinces, women did not want strangers to weigh or measure their children
(Government of Pakistan, 1992).
3.1.4 Potential effects of data errors:
Data collected from retrospective fertility surveys may suffer from various types
of errors, which bias the demographic measures. These errors arise from sources, such as
lack of knowledge among the respondents, misinterpretation of the questions, memory
lapse or poor interaction between respondent and interviewer. Data errors are particularly
problematic for estimates of dates and age at certain events, which occurred further back
in time (Palloni, et al., 1986). Errors due to misreporting can be of two types: omission
of events; and misreporting of dates of events.
Both, births and deaths are important events relevant to the study of infant and
child mortality as well as nutritional status of living children. Omission of these events is
likely to occur due to recall problems, especially among older and illiterate women. This
is likely for births which occurred in the distant past, particularly if the child did not
survive. But ignoring these cases would cause downward biases in childhood mortality
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
rates because this information is more likely to be missing for children who died than
children who are still alive (Sullivan, 1990). These omissions may also be affected by the
gender o f the child, which is most likely to occur in societies where sex preferences
prevail.
Curtis (1995), while evaluating the data quality o f DHS, estimated the differences
between the infant and child mortality rates for the period immediately prior to the WFS
and found that the PDHS rates were 30 percent below the WFS rates, suggesting under
reporting of the dead children in the distant past in PDHS (Curtis, 1995). The re
interview survey conducted in Pakistan following the DHS survey also found evidence of
omission of infant deaths throughout the birth histories in the DHS (Curtis and Arnold,
1994). The evaluation of the Pakistan DHS shows omission o f date of births of dead and
living children in birth history data but the omission is uniform for both surviving and
dead children, therefore, it does not present significant problem in estimating the
childhood mortality (Curtis, 1995).
However, in order to minimize the recall error, this analysis is limited to a period
of 5 years prior to the survey. Since the data on maternal health care and maternal
breastfeeding practices were only available for births in the five years preceding the
survey, limiting the sample to this age range also allows use of these independent
variables for all the cases.
A major advantage of using the data from reproductive histories to analyze the
infant mortality is that each child is treated as a unit of analysis. However, including
more than one child from one family also creates a problem. Including these children in
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the analysis could cause a spurious relationship and also violate the assumption of
independence in the modeling techniques commonly used in the analysis. If the data
includes individuals who share common circumstances that influence the outcome,
significant association may be artificially identified owing to the underestimation of
standard errors.
The PDHS data include a total of 6492 births that occurred 5 years preceding the
survey to 4061 women. To avoid the violation of the independence assumption, only the
last births are included in the analysis. Multiple births are also excluded because they
experience a higher risk of death associated with their multiplicity, which could distort
the results (Curtis et al., 1993). Births occurring during the month of interview are also
excluded because their exposure to neo-natal death is censored.
Thus, this analysis is restricted to singleton births, bom 1-59 months before the
survey. To include the survival status o f the older siblings in the analysis, only women
are included who have at least two births. There are many reasons that deaths tend to
cluster in families, such as premature delivery or intrauterine growth retardation.
These are likely to be repeated in other pregnancies as well. Also those children share
risks associated with family behavior and child care practice, such as infant feeding,
use of health services and general standards of domestic environment and hygiene.
Das Gupta (1990) called them parental competence.
Table-3.1 depicts the number and percentage of births included in this analysis
by each category of reaching at that number.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.1: Distribution of births included in the analysis
Description Number o f Births Percentage
Total Sample 6492 100.00%
Number o f women 4061 62.55%
Women with First order births 634 15.61%
Women having at least two births 3427 84.39%
Women having twins 55 1.60%
Women having Singleton 3372 98.40%
Births during the month of the Interview 70 2.08%
Number o f Births included in the analysis 3302 97.92%
Effective Sample 3302 50.86%
3.2 Conceptual framework
3.2.1 Mosley and Chen framework:
A framework proposed by Mosley and Chen (1984) for the study of child
survival in developing countries is a very valuable contributions. This framework
represents a complex theoretical paradigm expressed in a simple way, which helps to
identify particular determinants of child health and survival. It combines what they refer
to as two separate approaches of research into the subject: the social science perspective
and the biomedical perspective. It was assumed that child death is rarely an outcome of
one disease episode, rather growth faltering and ultimately mortality in children are the
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cumulative consequences o f multiple disease processes including their biosocial
interactions. This framework is based on the proposition that socio-economic factors do
not directly affect the outcome variable but rather must operate through proximate
variables to affect child survival (Mosley and Chen, 1984).
Mosley and Chen describe their aggregation of variables as “intermediate or
proximate” in the sense that influences on infant mortality are mediated through one or
another of these behaviorally mediated intermediate variables. They have grouped these
proximate variables into five classes, each of which represent or contain proximal
correlates to the respective category. They suggest that these proximate determinants are
influenced by the socioeconomic variables.
The Mosley and Chen (1984) framework is presented in Figure-3.1.
The Mosley and Chen (1984) framework identifies the following 14 proximate
determinants of child survival which are grouped into five major categories in the
framework:
o Maternal Factors: Age of the mother at the time o f birth; Birth order; and
Birth interval.
o Environmental contamination: The four categories representing the main
routes whereby infectious agents are transmitted to the human host are: Air;
food/water/fingers; Skin/soil/ inanimate objects; Insect vectors.
o Nutritional deficiency: Calories; Protein; Micro-nutrients;
o Injury: Accidental; Intentional;
o Personal illness control: Personal preventive measures; Medical treatment;
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure-3.1: Mosley and Chen Framework to Study Child Survival in
Developing Countries
Maternal
Factors
Environmental
Contamination
Nutrient
Deficiency
Injury
Control
Source: Mosley and Chen (1984).
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This framework provides an overall paradigm from which modules can be
extracted for particular research interests. However, this framework does not include
the cultural factors, which is an important determinant of, both, infant and child
mortality and nutritional status of living children. In populations, the custom of early
weaning could lead to serious problems among the poorer groups, where breastfeeding
is crucial to the reduction o f infant and child mortality. Another criticism of this
framework is that it combines variables with quite different levels of “proximate-ness”
to health outcomes and some of the most important determinants are missing from the
framework, such as breastfeeding and birth weight (van Norren and Vianen, 1986).
3.2.2 Modified Analytical framework:
Another analytical framework developed by van Norren, Boerma, and
Sempebwa (1989) distinguished four biological risk factors affecting the infection-
malnutrition syndrome and its outcome (survival or death). These four risk factors are:
exposure to infection, susceptibility to infections, constitution at birth, and nutrition
intake. The key to the conceptual framework is the identification of a set of
intermediate variables, which link the social and biological system together. In their
framework, the following risk factors are said to be affected by the five groups of
behavioral variables:
o Curative care variables: Oral re-hydration drugs,
o Exposure determinants; Personal hygiene; nutritional hygiene; water
supply and sanitation; housing; crowding; feeding; traditional practices.
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o Susceptibility determinants; Breastfeeding; immunization; nutritional,
o Constitution at birth determinants; Reproductive pattern; health and
nutrition in pregnancy; immunization in pregnancy; prenatal care
o Nutritional variables. Breastfeeding; weaning; post-weaning nutrition.
There is also some criticism of this framework, such as, in this framework,
constitution at birth is relevant to health outcomes chiefly because of susceptibility to
disease and an argument can be made that it should be placed at a less proximate level
(Cebu Study Team, 1992). Moreover, nutritional intake can influence susceptibility to
infection, infection can influence nutritional intake, and infections and intakes may
synergistically influence child growth (Grosse, 1996).
3.2.3 Framework to Study the Underlying Causes of Malnutrition and
Death:
The study o f underlying causes of malnutrition and death has been simplified
by the United Nations Children’s Fund, UNICEF in 1990, which provided the
foundation for the International Conference on Nutrition (1992). As shown in the
figure-3.2, the causes of malnutrition and death are insufficient household food
security, insufficient health services and unhealthy environment, inadequate maternal
and child health care (UNICEF, 1990). This causative role o f childcare highlights the
fact that malnutrition can arise even in the presence of household food security and
adequate availability of health services. This shows that food, health and care, when
considered individually, represent a necessary but not sufficient explanation for the
existence of malnutrition in households and communities (Pelletier, 1998).
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure-3.2: Causes o f Malnutrition and Death
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.4 Proposed Framework:
Bicego and Boerma (1993) used the conceptual framework proposed by
Mosley and Chen (1984) and Van Norren and Van Vianen (1989) with modifications
based on the limitations and structure of the DHS data. Their analysis focused on
maternal education as a determinant of childhood mortality and nutritional status with
control for economic status and the particular role o f health services use and family
formation pattern in mediating the relationship. They used the physical sanitary
facilities as conditioned by economic in the education-health outcome relationship.
They also used the anthropometric indices as the nutritional status of living children.
Rural-urban residence was used as a proxy for physical access to modem health
facilities.
Keeping in view the aforementioned frameworks, the model proposed here
uses not only the survival status of the children but also the nutritional status of the
living children. Like the Mosley and Chen (1984) framework, the socioeconomic
variables affect the outcome (survival status, and the nutritional status of living
children) through the four proximate determinants (data on Injury is not available in
the DHS) namely, demographic factors, environmental factors, nutritional factors and
health seeking behavior factors. Three dependent variables are used: survival status
(dead and alive) and nutritional status of survivors (normal growth; stunted; and
severely stunted based on height-for-age). The resultant classification for each child is
normal growth; stunted, severely stunted, and dead. The proposed model for this study
is illustrated in Figure-3.3.
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure-3: Proposed Model for Child Survival in Pakistan
Socioeconomic
F idon:
Mother's
Education
Father’s
Education
Standard o f
living index
Place of
Residence
Region o f
Residence
Denuwanhlc
Factors:
1.Mothers’Age;
2.Binh Interval;
3.Birth Order;
4.Sex o f the child
S.Siblings Death
6.Children < 5
1.Drinking Water
2.Sanitary Facility;
3.Housing Material
Environmental
Factors:
1. Prenatal Care
2.Place o f Delivery
3.Delivery Attendant
4.BCG Vaccine
5.Contraceptive use
Health seeking
Behavior:
1. Breastfeeding
2. Birth Weight
3. Premature Births
4. Starting age o f
Supplements
Nutritional Factors:
I .Survival Status
1. Dead
2. Alive
2 .Slunting
1. Normal Growth
2. Stunted
3. Severely Stunted
3.Flnal Outcome
Status
1. Dead
2. Severely Stunted
3. Stunted
4 .Normal Growth
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33 Variables Used in the Analysis:
The PDHS, 1990-91 provides data for each of the proximate determinants. For
the purpose of this study the important variables identified on the basis of available
literature review are categorized under the groups presented in the model. The statistical
methodology used for this research is explained after the variable definitions.
Independent variables included in the analysis are described in the sections 3.3.1
through 3.3.5 and dependent variables are defined in the section 3.4.
3.3.1 Socioeconomic Variables:
It is assumed that the socioeconomic conditions of the families remained the
same during the five years period before the survey. Therefore, the household conditions
when the children were bom assumed to be the same as when the data were collected.
33.1.1 Index of Household Possessions:
In the absence of income data, the index of household possessions was used as a
proxy for family wealth. The household resources are indicated by possession of
household goods such as, refrigerator, television, radio, bicycle and motorcycle. From
these indicators, an index was created by summing-up these scores to represent the
economic position of the household by assigning an equal score to each of these items. A
maximum score of 5 indicates a high socioeconomic status and a very low
socioeconomic status by having a 0 score. Then it was categorized into three: low (0
score), medium (1-2 score) and high (3-5 score) socioeconomic status. The following is
the percent of households by possession of durable goods:
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.2: Distribution of households by index of household possessions
included in the sample and in this analysis
Items Sample
Percent
Study
Percent
Low 37.0 37.9
Medium 48.8 47.7
High 14.2 14.4
Number of households 5429 3302
3.3.1.2 Parental Education:
The mothers’ education is important in studying the survival status as well as the
nutritional status of living children. The mothers’ education affects the outcome through
different ways as studied by many researchers. It is commonly assumed that greater
empowerment and autonomy are inevitable consequences of schooling and that these are
the main pathways that link education to better child survival.
In Pakistan, female literacy is very low. Female education has generally lagged
behind male education due to a comparatively unfavorable societal attitude. The major
bottleneck to formal education is lack o f schools for females, especially in rural areas.
Since education is sex-segregated in the country, female teachers are reluctant to work in
isolated rural areas or even in urban slum areas (Government of Pakistan, 2000). Table-
3.3 shows the distribution of parental educational levels in the DHS sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.3: Distribution of children by parental education
in the sample and in the analysis
Education Sample
Percent
Study
Percent
Maternal Education
No Education 79.2 79.7
Primary 13.5 13.1
Secondary and above 7.3 7.2
Father’s Education
No Education 46.5 48.7
Primary 17.4 17.3
Secondary 31.9 29.8
College 4.2 4.2
33.13 Place of Residence:
The place of residence of the parents affects, both, the survival status and
nutritional status of the living children in developing countries. The urban areas are
mostly equipped with a better infrastructure for health services than rural areas.
Availability and accessibility to public health services at the community level play a
crucial role in child survival prospects. In most developing countries, availability of
primary health services is mostly concentrated in urban cities where people with higher
socioeconomic status tend to reside. In rural areas, health centers are generally
characterized by poor services and staffing. In Pakistan, both the public and private
sector are providing medical facilities in the country but the private sector has
concentrated in the urban areas (Government of Pakistan, 2000).
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In general, in developing countries, people from lower social class are inclined to
reside in rural areas. Thus, urban/rural residence is used as a socioeconomic variable to
capture the effect of differentials in utilization in health services on child survival. Table-
3.4 shows the distribution of children by their place of residence.
Table-3.4: Distribution of children by place o f residence
in the sample and in the analysis
Place of Residence Sample
Percent
Study
Percent
Major Urban 17.4 17.1
Other Urban 13.1 13.0
Rural 69.5 69.9
Total 100 100
3.3.2 Household Environment and Hygiene:
3.3.2.1 Source of Drinking Water:
Safe and clean drinking water is important for human health, especially for
children whose immune systems are still maturing. Even if contaminated water is used
for washing or bathing, it increases the chances o f catching infection. When
supplementary foods are prepared for the infants and children, use of clean water is very
important. If the water is contaminated, it is more likely that the child will get infections
and diarrhea. It has been reported that poor or contaminated sources of drinking water
are associated with an increased number of deaths due to diarrhea. The access to piped
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
water in the household is likely to be of most direct benefit in lowering early childhood
mortality by reducing exposed to water-borne diseases, particularly diarrhea and
dysentery (Merrick, 1985).
In this analysis, five sources of water supply are included: piped water in to the
residence; piped water on to the property; public tap; well and surface water (such as,
river or rain). Table-3.5 shows the distribution of households by their source of drinking
water:
Table-3.5: Distribution of households by source of drinking water
in the sample and in the analysis
Source of drinking water Sample
Percent
Study
Percent
Piped into residence 18.3 9.0
Piped onto property 9.2 10.5
Public tap 7.1 7.7
Well 55.3 54.5
River, canal and others 10.1 8.3
Total 100 100
3.3.2.2 Toilet Facilities:
Toilet facilities are very important for the health and hygiene for, both, children
and adults. Proper use of flush toilets can reduce risk o f parasite infection, morbidity and
poor nutrition status o f children.
In Pakistan, the majority of children are not diapered, but are allowed to urinate
and defecate freely on the dirt floor (most of the houses do not have cemented floor),
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
after which the mother or the older sister sweeps the excrement out into the yard. It is
assumed that households with flush toilets are hygienically better than those which have
other source of toilet facilities.
Three different sources of toilet facilities are included in this analysis. Flush
system; latrine (pit and bucket); no facilities (they do not have any facility at home, use
nearby crops or bushes). Table-3.6 shows the distribution of toilet facilities in the
households.
Table-3.6: Distribution of households by toilet facilities
in the sample and in the analysis
Sanitation facility Sample
Percent
Study
Percent
Flush 26.5 27.1
Bucket & Pit Latrine 20.6 18.8
No and other facilities 52.9 54.1
Total 100 too
3 J.2 .3 Housing Material:
Material used in building the house was also included in the survey. The
construction shows the hygienic conditions of the building. Houses built of baked bricks
(cemented) are supposed to be hygienically better than the houses constructed from mud
or unbaked bricks. Table-3.7 shows the construction distribution of the households.
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.7: Distribution of households by construction material used
in the sample and in the analysis
Housing Material Sample
Percent
Study
Percent
Baked bricks, cement 37.8 38.5
Unbaked bricks, mud
and others 62.2 61.5
Total 100 100
3.3.2.4 Exposure to Diarrhea:
Exposure to diarrhea is defined as a dichotomous variable based on whether or
not a child experienced an episode of diarrhea during the last two weeks before the
survey. The question asked the mothers, “Has (name of the child) had diarrhea in the last
two weeks?” If the response was yes, than the follow-up question was “Has (Name) had
diarrhea in the last 24 hours?” and a second follow-up question was “For how many days
(has the diarrhea lasted/did the diarrhea last)?”
The dummy takes the value of 1 if child had diarrhea and 0 if child did not have
diarrhea during the past two weeks before the survey.
Diarrhoeai disease is a major cause of child morbidity and mortality in Pakistan
(Government of Pakistan, 1992). The DHS sample shows that about 15 percent of the
children aged 1-59 reported diarrhea during the preceding 2 weeks of the survey.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.3 Demographic and Maternal Factors:
3 J J .l Age of the Child:
The child’s age at the time of the survey is measured in terms of months.
Nutritional status varies as children get older, therefore, age-square and age-cube
variables are also included to evaluate the possible curvilinear relationship between age
and nutritional status of the children. If the infant is breastfed, weight gain is usually very
good during the first 4 to 6 months of life. This weight gain during this period may be
superior to that of Western bottle-fed babies, in some communities an individual may
never be better nourished throughout life as at the age of 4 to 6 months (Jelliffe, 1968).
Even if the breastfeeding is usually continued after 6 months of life in developing
countries, the quantity is no longer sufficient for the larger children. At this age the child
also is losing the immunity which was passed to him through the placenta from his
mother and will begin to be susceptible to various infections (Jelliffe, 1968). Moreover,
the second year of the child’s life is the most dangerous phase of growth. Even if the
breastfeeding is continued during this period, the amount of protein supplied in this way
is small. In some children, the weight may actually decrease. It is by no means unusual to
see a child of 18 months who has reverted to his original weight of 6 months old (Jelliffe,
1968).
Children over 3 years acquire a certain degree of resistance to various infections
and is able to obtain and digest a wider range of the family diet. Under these conditions
the child may remain below the standard weight and height for years, but starts growing
slowly (Jelliffe, 1968).
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
333.2 Preceding Birth Interval:
It is a well established fact that infant and child mortality varies by the length of
the preceding birth interval. Therefore, three different dummies of preceding birth
intervals are included in this analysis: 18 months and less; 19-35 months; 36-47 months;
and over 48 months.
Table-3.8: Distribution of children by the length o f preceding birth interval
in the sample and in the analysis
Preceding Birth Interval
Moths
Sample
Percent
Study
Percent
18 and less 18.7 12.7
1 9 -35 48.3 35.8
35 - 47 15.5 29.2
48 and over 17.5 22.4
Total 100 100
33 J.3 Birth Order:
Birth order is included in this analysis because birth order not only tells us the
rank o f the child in the family but also tells us something about the number of children in
the family, in households with limited resources, its distribution depends on the number
of children in the household. The larger number of children indicates a smaller share of
scarce resources. Because the children have to compete for their share, the nutritional
requirements for growing children may result in malnutrition.
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In most of the developing countries, higher mortality risks were found for the
first-bom children compared to birth order two through six (Sullivan et al., 1994;
Rutstein, 1984; Hobcraft et al., 1985). The positive relationship between higher birth
order and stunting is found in developing countries (Horton, 1988; Strauss, 1990).
3.3.3.4 Previous Death of Siblings:
The idea that children of certain families have biological traits which predispose
them to high mortality is supposed by evidence that the effects o f sibling deaths on the
health of the subsequent child are considerably significant even after controlling for
socioeconomic, biological and behavioral factors (Curtis et al., 1991; Das Gupta, 1990;
Hobcraft et al., 1985). The survival status of the previous child is observed to be
associated with the length of the birth interval as well as with breastfeeding status.
In this analysis, the survival status of the previous child is included to see if that
has any effect on the survival status of the index child in Pakistan. To include this
variable in the analysis, we need at least two births. Therefore, this variable is obviously
not applicable the first order bom children in the analysis.
A previous death is defined as a dichotomous variable indicating if death
occurred coded as I, elsewhere coded as 0.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.4 Nutritional Status /Dietary Intake:
3.3.4.1 Birth Weight/ Birth Size:
In Pakistan, a large proportion of births occur at home and it was difficult to
obtain the exact birth weights o f these babies. In the PDHS, a question asked was: was
the baby very large, large, medium, small or very small at the time of the birth. This is
clearly far from a birth weight, but it would be meaningful if the mother herself evaluates
her baby as being very large or very small than average. Two dummy variables are
created, one for very small and the second for very large keeping small, medium and
large as reference.
Premature birth with its consequent low birth weight is one of the outcomes of
that has been described as the “maternal depletion syndrome”. It is known to be
associated with maternal age (<20 and >35), short birth interval (Carlaw et al., 1983),
high birth order (Hobcraft et al., 1985) and a higher risk of death. The prevalence of low
birth weight reflects the health and social status of women and the communities into
which children are bom (WHO, 1987).
The birth weight of the infant is one of the most important predictor of survival
and is also a strong predictors of growth (National Academy Press, 1992). Small infants
with less supervision are more likely to ingest pathogens if they live in an unhygienic
environment, and are at greater risk of injury. Low birth weight poses a greater threat to
a child bom in conditions prevailing in slums or villages of developing countries than to
a child bom in a society with adequate medical services and a satisfactory physical and
cultural environment (Kimm, 1979). Babies bom very big are abnormal babies
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Yerushalmy et al., 1965) and their risk of mortality during neonatal period is also higher
than the normal size bom babies (Rees et al., 1999). Table-3.9 shows the distribution of
babies by their birth size.
Table-3.9: Distribution of babies by their birth-size
in the sample and in the analysis
Birth Weight/Birth Size Sample
Percent
Study
Percent
Very small 6.3 6.5
Medium 91.8 91.4
Very large 1.9 2.1
Total 100 100
3.3.4.2 Breastfeeding:
Breastfeeding is used as a time-variant covariate for the analysis of post neonatal
mortality. However, for the small number of children who died in the first month of life
and who did not have the chance to breastfeed, death is not associated with lack of
breastfeeding. The dummy is coded as 1 for all infants who died in the first month along
with all other children who breastfed during the first year of their life. This procedure
will produce a small conservative bias in estimating the effect o f breastfeeding on child
survival.
3 J.4 J Supplementary Food:
The period when the children’s health is most in danger starts at three months and
lasts until they can feed themselves, at about three years old (Khan, 1984). During this
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
period several feeding practices can have an influence on the child’s nutrition. First, is
the age at which food supplements are introduced into the child’s diet and second, is the
method of preparation of the supplementary food.
The prevalence of bottle-feeding with nipples is another possible source of
contamination of milk and provides an indication o f inappropriate feeding practice
(Phillips et al., 1969). Studies in developing countries found bottles and nipples were
contaminated by different pathogens (Elegbe et al., 1982; Black et al., 1989). Studies also
show that virtual elimination of feeding bonles resulted in considerable reduction in the
rate o f diarrhea (Feachem, 1984). Hence, the relationship between weaning food
contamination and diarrhea can be inferred from available evidence.
Improved cleaning of bottles and nipples may reduce the risk of contamination of
milk or formula by boiling of bottles and nipples. However, proper use of these
procedures is often difficult because poor households may have only one feeding bottle
which is constantly in use.
In this analysis, supplementary food is also used as a time-varying variable.
3.3.5 Health Seeking Behavior:
3.3.5.1 Prenatal Care:
Regular prenatal care is needed to help detect and manage some pregnancy-
related complications and to educate women about danger signs, potential complications,
and where to seek help (WHO, 1994). Prenatal care beginning early in the first trimester
of pregnancy and continuing on a regular basis is important to the health of both mother
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and infant. Early prenatal care provides an opportunity to offer preventive care that will
benefit the infant as well as the mother such as, counseling on hygiene, breastfeeding,
nutrition, family planning, tetanus toxoid immunization and iron and folate
supplementation. Prenatal care also benefits treatment of existing diseases that may be
aggravated by pregnancy. Prenatal care helps to prevent complications during pregnancy
and labor. Also, there is a negative association between timing of prenatal care and low
birth-weight. If a woman's pregnancy goes to term, she may typically have anywhere
from 10 to 14 prenatal visits. In PDHS, if the respondent reported prenatal care was
taken, then the question regarding the number of visits were also asked. This variable is
used to predict both mortality and nutritional status of living children. Table-3.10
presents the data on prenatal care taken.
Table-3.10: Distribution o f women by the status of prenatal care
in the sample and in the analysis
Prenatal Care Sample
Percent
Study
Percent
Taken 12.6 12.4
Not taken 87.4 87.6
Total 100 100
33.5.2 Place of Delivery:
It is a fact that deliveries at medical institutions are safer, both, for mother and the
child. Hence, a higher child survival rate is expected for these deliveries. The
assumption is that mothers who have delivered in hospitals will also be aware of better
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
child health care practices. These include the importance of breastfeeding, hygiene,
nutrition, immunization etc. These babies also have higher chances of getting the BCG
vaccine compared to the babies delivered at home. Moreover, hospitals are assumed to
maintain safe delivery environment, and newborns may have fewer chances of
contamination. Hence, these children should be better off than children bom at home.
Although delivery services are almost free in government hospitals /clinics, some
people prefer private clinics/doctors for several reasons. First, it is generally believed
that anything worthwhile or valuable will cost money. The medical services that must
be paid for are seen as better than government services. In addition, practitioners who
charge for these services are expected to be more polite and attentive and to devote more
care and concern to patients.
Three dummies are included to represent this variable: one for delivery at a
governmental hospital; second a dummy for delivery at a private hospital; and the third
dummy for if delivery occurred at other than respondents’ home keeping at respondents
home as reference. Table-3.11 shows the distribution of births by place of delivery.
Table-3.11: Distribution of births by place of delivery
in the sample and in the analysis
Place of Delivery Sample
Percent
Study
Percent
At Home 83.9 85.4
Other Home 2.4 2.1
Government Hospital 6.4 5.7
Private Hospital/Clinic 7.3 6.8
Total 100 100
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33.53 Birth Attendant:
A birth attended by trained medical staff such as physicians, nurses, or family
planning workers would be safer for both mother and the newborn infant. These
medically trained personnel are more likely to use sterilized equipment for the delivery
compared to the non-trained personnel. Hence, this variable is categorized 1 if the birth
was attended by the trained medical staff and 0 elsewhere. Table-3.12 shows the
distribution of births by type o f birth attendant.
Table-3.12: Distribution of women by birth attendant
in the sample and in the analysis
Birth Attendant Sample Study
Percent Percent
Trained Personal 35.2 34.6
Untrained 64.8 65.4
Total 100 100
33.5.4 BCG Vaccination:
Vaccination against tuberculosis prevents a disease that carries an increased risk
of dying for children in developing countries. According to the Global Tuberculosis
Control WHO Report 2000, in Pakistan about 1.5 million people suffer from TB, and
more than 210,000 new cases occur each year. Table-3.13 shows the distribution of
babies by the status of BCG vaccination:
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.13: Distribution o f children by BCG vaccination
in the sample and in the analysis
BCG Vaccine Sample Study
Percent Percent
Had BCG 62.0 61.4
Did not have BCG 38.0 38.6
Total 100 100
3.4 Dependent Variables:
3.4.1 Neonatal and Post neonatal Mortality
Mortality is specified as neonatal and post neonatal mortality. The dependent
variable used in the hazard model analysis is neonatal and post neonatal survival time. It
is measured as the duration starting from the infant birth to death, if the event occurred or
from the infant birth to the survey date for censored data. Table-3.14 shows the
distribution of births and their survival status during the 1-59 months before the survey.
Keeping in view the heaping of age of death at 12 months, the post neonatal
mortality period is collapsed for l-l 1 and 12-23 months to avoid biases introduced by
any relationships between age at death misreporting and the explanatory variables.
Further, only the inclusion of women who had at least two births (excluding first order
births), the small number of deaths at older ages and the heavy right-censoring precludes
statistically relevant results for a separate child model. Therefore, two observational
periods are identified and modeled separately in this analysis: zero completed months
(neonatal period); and the period between 1 and 23 months o f child age (Post neonatal
mortality).
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.14: Distribution of number of live births by age of children
in the sample and in the analysis
Child Age Total Sample Identified for the Study
in Months Alive Dead Total Alive Dead Total
0 83 309 392 0 101 101
1 154 50 204 113 10 123
2 145 34 179 122 6 129
3 126 35 162 101 5 105
4 132 15 147 105 5 110
5 121 8 129 92
2
94
6 115 25 141 85 6 91
7 120 4 124 104 0 104
8 113 14 127 93 6 98
9 96 20 116 78 12 90
10 86 16 101 68 2 69
1 1 78 3 82 64 2 66
12 121 48 170 95 22 117
13 158 I 159 117 I 118
14 140 4 145 90 4 95
15 125 0 125 99 0 99
16 101 5 105 86
2
88
17 83 0 83 67 0 67
18 111 7 118 84
2
86
19 86 0 86 68 0 68
20 73 0 73 57 0 57
21 69 0 69 55 0 55
22 82 1 84 53 1 54
23 66 0 66 48 0 48
24 75 22 97 37 6 43
25 109 0 109 67 0 67
26 90 0 90 49 0 49
27 107 0 107 49 0 49
28 97 0 97 57 0 57
29 103 0 103 52 0 52
30 97 0 97 39 0 39
31 93 0 93 44 0 44
32 93 0 93 46 0 46
33 95 0 95 41 0 41
34 85 0 85 43 0 43
35 99 0 99 56 0 56
36 77 9 86 38
2
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-3.14 continued
Child Age Total Sample Identified for the Study
in Months Alive Dead Total_________ Alive Dead Total
37 99 0 99 31 0 31
38 93 0 93 33 0 33
39 102 0 102 34 0 34
40 100 0 100 34 0 34
41 74 0 74 22 0 22
42 114 0 114 26 0 26
43 100 0 100 31 0 31
44 108 0 108 36 0 36
45 118 0 118 40 0 40
46 84 0 84 18 0 18
47 97 0 97 25 0 25
48 62 0 62 21 0 21
49 63 0 63 26 0 26
50 92 0 92 17 0 17
51 78 0 78 18 0 18
52 88 0 88 18 0 18
53 73 0 73 10 0 10
54 84 0 84 14 0 14
55 85 0 85 22 0 22
56 86 0 86 20 0 20
57 87 0 87 12 0 12
58 91 0 91 17 0 17
59 76 0 76 14 0 14
Total 5860 632 6492 3104 198 3302
3.4.2 Nutritional Status (Stunting):
A child with a height-for-age Z-score below -2SD is considered stunted
and a Z-Score lower than -3 standard deviations is considered severely stunted.
The preliminary analysis o f PDHS shows that about 50 percent of the children
are stunted and thirty percent of all the children are severely stunted in Pakistan.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5 Statistical Methodology:
Both bivariate and multivariate statistical techniques are used in this
analysis. Cross-tabulation with chi-square test is used for the bivariate analysis.
For multivariate analysis proportional hazards analysis is employed for the analysis
of neonatal and post neonatal mortality as it takes in to account the censoring. For
the analysis of stunting, ordered logistic regression as well as multivariate logit
techniques are employed. Following is a brief description of each technique.
3.5.1 Proportional Hazards Analysis:
In this analysis, Cox-regression is used to estimate the effects of
covariates on the neonatal and postnatal mortality. This regression is also called
a proportional hazards model. Proportion hazard models have been widely used
in demographic research when the time of occurrence of an event is observed.
In any study in which subjects are followed over time until an event of interest
occurs, it is not always possible to follow every subject until the event is observed.
Subjects may drop out of the study and be lost to follow-up, or be deliberately
withdrawn or the end of the data collection period may arrive before the event is
observed to happen. For such subjects all that is known is that the time to the event
was at least as long as the time to when the subject was last observed. Under such
circumstances the observed time to the event is censored.
Life-table-analysis or event-history-analysis, generally allows for censored
data. There are several techniques to deal with such types of data, such as, life
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
table techniques, discrete-time logistic regression and Cox-regression. Life-table-
analysis models survival-time-distributions and compares distributions of different
groups. Discrete-time logistic regression models the probability of an event within
a given period as a function of one or more covariates. Cox regression, models the
hazard probability as a function of a baseline hazard rate and a parameter vector.
In the life-table-analysis, when assessing the effects of socio-economic or
other covariates on the dependent variable, we have to compute separate life tables
for each category of a covariate. When attempting to evaluate the effects of several
covariates simultaneously, the number of cells may become too large and the
number of observations in each cell too small for reasonable analysis. In this case
the discrete time logistic models should be transformed into a proportional hazard
regression model. As the time intervals become finer, the dependent variable for
this analysis turns into a ratio of hazards.
The proportional hazard model derives its name from the relationship that
exists between the hazard functions for two individuals. As the ratio of the hazard
functions for two individuals is constant in time, their hazards are proportional, and
the model is called proportional hazards model. This technique also
accommodates the set of time-dependent covariates in the analysis (Allison, 1995).
The independent variables, length of breastfeeding may vary for each individual at
risk, is taken as time-dependent covariates. By introducing time-depending
covariates, not only is it necessary to consider the matter of coding the time-
invariant covariates, it is necessary to consider the coding of time-dependent
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
covariates also. The interpretation of these coefficients depends on their correct
coding.
The ratio o f hazard functions or instantaneous failure rates generalizes the
epidemiological concept of relative risk, that is, a ratio of rates for two groups.
Though the regression coefficients /?, determine the effect of explanatory variable
Xi on the hazard function, it is easier to interpret exponential of /?, as a relative
risk. Therefore, the results are presented in the form of relative risks in each table
for neonatal and post neonatal mortality.
In this analysis partial likelihood estimation technique is employed to get
the hazards of each independent variables without specifying the baseline hazard
function Xo(t) (Allison, 1995). The partial likelihood method is based on the fact
that the likelihood function for data arising from the proportional hazard model can
be classified into two main factors:
1). Factor containing information only about the coefficient /?,;
2). Factor containing information about the function \j(t).
The partial likelihood is a product of a likelihood contribution for all events
that are observed to occur.
«-iiA
1 = 1
Where PL is the partial likelihood and / is the total number of events in the
sample. While the ordinary likelihood function is a product o f likelihood for all
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
individuals in the sample, the partial likelihood is a product of the likelihood for all
the events that re observed (Allison, 1995).
The PL function also can be written solely as a function of parameters for
covariates such that:
i = l
Where hi(tj is the value of the hazard function for the jth subject at time /,
is the time at which the ith subject had either the event or the censoring; and oi is a
dummy variable that takes 1 when the ith subject had an event and 0 if the jth
observation was censored. This partial likelihood function is actually a product of
conditional probabilities (Yamaguchi, 1990).
The proportional hazards models are fit to test the hypotheses that the
socioeconomic variables work through the proximate determinants to effect
neonatal and post neonatal mortality. First the dependent variables are regressed on
socioeconomic variables, and then the proximate determinants and interactions are
added one by one to observe the change in the coefficients. However, to get the
final parsimonious model, all the non-significant variables are excluded from the
model.
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5.2 Ordered Logistic Regression:
To assess the levels of child stunting or height-for-age, ordered logistic
regression is employed. This is also called the cumulative logit model (Allison,
1999). A child with a height-for age Z-score below -3SD is considered severely
stunted (coded as 3) and a Z-Score lower than -2 standard deviations is
considered severely stunted (coded as 2). Children with normal growth are
coded as I.
The basic principle of the method is similar to ordinary logistic regression.
The LOGISTIC procedure of the SAS program procedure is used to fit the model
with ordinal response variable. This procedure fits a parallel lines regression
model that is based on the cumulative distribution probabilities of the response
categories. In this case, the response has three possible outcomes: 3=severely
stunted, 2=stunted, and l=normal growth. This procedure produces one set of
regression coefficients but different intercepts (Allison, 1995).
The model for this study is mathematically defined as:
log
( F,
Where = £ P,
= aj + P *, j =1>2
im
m = l
And f5x = filxil + + P kxik
Fij is the probability that individual i is in the jth category or lower. This
model produces one set of coefficients and two different intercepts for each of the
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
equation for severely stunted and stunted. In this model the explanatory variables
predict the probability o f being in a lower category, in this case stunted or severely
stunted. The model then is specified as a set of 2 equations, since there are two
independent contrasts that can be constructed. For example, it can be considered
that the log odds of being in category 3 versus category I; or the log odds of being
in category 2 versus category 1, expressed as a linear function of the predictors,
this becomes as follows:
where fix = f3vr, + + ........+ fit xk . The explanatory variables predict
the probability o f being in a lower category than in the higher category (Allison,
1999).
The transformation of the logistic regression parameter estimates (logit
coefficients) to the probabilities are calculated by using the following equations as
suggested by Allison (1995):
log//(p,+p2) = log
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The sum of the 3 probabilities should be 1.0 to check the correctness of the
calculations.
The ordered logistic regression models are fit to test the hypotheses that the
socioeconomic variables work through the proximate determinants to effect
stunting. First the dependent variables are regressed on socioeconomic variables,
and then the proximate determinants and interactions are added one by one to
observe the change in the regression coefficients. However, to get the final
parsimonious model, all the non-significant variables are excluded from the model.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5.3 Multinomial Logit Model:
Combining both mortality and stunting together to make one dependent
variables with four categories, multinomial logit analysis is employed by using
the SAS, CATMOD procedure. This procedure provides the maximum
likelihood estimates (Allison, 1999).
The child death is coded as 1, severely stunted condition as 2, stunted as
3 and a child with normal growth as 4. As in the dichotomous case, we model
the log odds of being in a category of interest as a linear function of the
explanatory variables.
CATMOD procedure produces three different sets o f parameter, one for
each separate log odds (i.e log odds of dead versus normal growth, log odds of
severely stunted versus normal growth, and log odds of stunted versus normal
growth) (Allison, 1999). The interpretation o f the coefficients is exactly the
same as for the bivariate regression. Thus, according to the value of odds, it is
possible to find which variables have a stronger effect on the changing status of
the child from normal growth to stunting, severely stunted or to dead.
log & = a 3 + /?3 X j, +....+ /?3 *x3 *
\P*J
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4
NEONATAL AND POST NEONATAL MORTALITY
This chapter presents the bivariate and multivariate analysis of neonatal and
post neonatal mortality. The bivariate analysis is based on the cross-tabulation o f the
proportion dead during the neonatal and post neonatal period by all the explanatory
variables and the proportional hazards model is also used to analyze the bivariate
effect of explanatory variables on the neonatal and post neonatal mortality. The
multivariate analysis is based on the proportional hazards models by including all the
explanatory variables in a series of multivariate models. The bivariate analysis is
presented in section 4.1 and the multivariate analysis in section 4.2. The results o f the
proportional hazard models are presented in the form o f hazard rate ratios.
4.1 BIVARIATE ANALYSIS
The cross-tabulation of proportion dead by the explanatory variables is used to
evaluate the association between these variables by applying the chi-square test. Most
of these independent variables are also associated with each other, therefore,
multivariate analysis is later applied to observe the independent effect of each
individual variable by keeping the effect of the other variables constant.
The results of the cross tabulation are explained for each o f the variables
included in this analysis.
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.1 Socioeconomic Variables:
4.1.1.1 Parental Education:
Table-4.1 shows an association between the mother’s educational level and the
incidence of neonatal mortality. It shows that the proportion of neonatal deaths is
highest for babies of uneducated mothers relative to mothers with primary or higher
than primary education. The proportion dead for women with primary education is less
than 1 percent, compared to 1.81 percent for the babies of mothers who have more
than primary education. However, bivariate analysis does show increasing the
proportion dead by maternal education during post neonatal period but the association
is not statistically significant. This shows that mothers with no education have a higher
proportion (3.0 percent) of post neonatal deaths compared to the women who have
primary (2.1 percent), or more than primary education (1.3 percent).
Table-4.1 also shows that neonatal deaths are more frequent among children
whose fathers have primary education compared to children whose fathers have no
education. However, children of the fathers who have secondary or higher education
have a lower proportion of neonatal deaths. The proportion of neonatal deaths is 2.74
for children of fathers with no education compared to 2.03 percent of neonatal deaths
among children of fathers with more than secondary education.
The association between fathers’ education and post neonatal mortality is
highly significant (p < 0.001). Earlier analyses of WFS in the Philippines, Indonesia
and Pakistan show that father’s education only makes a difference for children of older
ages (Martin etal., 1983).
I ll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.1: Bivariate Relationship Between Socioeconomic Factors and percent of deaths
during the Neonatal and Post neonatal Period, PDHS, 1994-91
Socioeconomic Variables % Neonatal Number % Postnatal Number
Deaths (N) Deaths (N)
Maternal Education 3302 3201
No education 3.51 2630 3.03 2538
Primary 0.95 434 2.09 430
Greater than Primary 1.81 238 1.33 233
Chi-Square 9.213** 3.144
Fathers Education
No education 2.74 1598 3.66 1554
Primary 5.06 567 3.54 538
Secondary 2.55 977 1.34 952
Higher 2.03 139 0.21 136
Chi-Square 9.600* 16.134***
Index of Household Possessions
Lower 3.89 1313 3.16 1262
Medium 2.91 1436 2.76 1396
Higher 1.44 552 1.95 544
Chi-Square 8.262** 3.190
Place of Residence
Rural 3.45 2308 3.02 2228
Urban 2.13 994 2.22 973
Chi-Square 4.085* 1.603
Region of Residence
Punjab 3.19 1998 2.77 1934
Sindh 3.32 719 3.66 694
NWFP 2.21 4.63 1.37 452
Balochistan 1.25 1.23 3.14 121
Chi-Square 3.127 5.368
• • • p < 0.001, • • p < 0.01, • p < 0.05. + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.1.2 Index of Household Possessions:
The index of possession o f household goods has a statistically significant
association with neonatal mortality (p < 0.01). The proportion of deaths is higher
among children living in households that have a lower index score compared to the
proportion of neonatal deaths among children living in households that have medium
or higher index score o f household possessions. These results show that about 4
percent of the children died during the neonatal period among mothers with a lower
index, compared to 2.91 percent and 1.4 percent with the medium and the higher index
scores, respectively. The same pattern is observed in the case of post neonatal
mortality but the association is not statistically significant. The proportion of post
neonatal deaths among children o f lower index score households is 3.16 percent
compared to 2.76 percent and 1.95 percent of post neonatal deaths among the children
o f mothers with medium and higher index of household possessions, respectively.
4.1.1.3 Place of Residence:
As expected, differentials in neonatal mortality are also found by place of
residence. Table-4.1 shows that the relationship is statistically significant (p < 0.05).
About 3.45 percent of neonatal deaths are reported in rural areas, compared to 2.13
percent in urban areas. For post neonatal mortality, the same pattern is observed but
the differences are less pronounced and are not statistically significant.
Earlier findings o f different studies show different urban rural mortality
patterns in developing countries. Cleland et al. (1992) found that in a sample o f 12
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
developing countries, better survival among children in urban areas over those in rural
areas increased between the mid-1970s and the mid-1980s from a factor o f 1.4 to a
factor of 1.6. However, urban-rural differences in child mortality have been found to
be insignificant once effects of parental education and household socioeconomic
factors have been controlled in studies of many different settings, including Pakistan
(Martin et al., 1983; Hobcraft et al., 1984).
In some countries, urban mortality is higher than in rural areas after controlling
for other variables (Trussell and Preston, 1982; Mensch et al., 1985; Martin et al.,
1983). Earlier studies of child mortality in developing countries hypothesized that
differences in health facilities and medical care were important factors accounting for
the urban advantage (Johnson, 1964; Davis, 1973).
4.1.1.4 Region of Residence:
Table-4.1 also presents the differentials in the association of neonatal and post
neonatal mortality by region of residence. The highest proportion of neonatal deaths
is observed in the province of Sindh followed by Punjab, NWFP and Balochistan.
4.1.2 Domestic Environmental and Hygiene Factors:
4.1.2.1 Source of Drinking Water:
Available research suggests that among environmental factors, household
sanitation conditions, measured in terms of provision of drinking water, waste disposal
and type of housing, play an important role in bringing down post neonatal mortality.
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.2 presents the association between source of drinking water and neonatal
mortality. Among the available sources of drinking water, the proportion of neonatal
deaths is highest among the households that use drinking water from the wells. In
Pakistan, more than 50 percent of the population depends on wells for drinking water.
Table-4.2 shows that piped water is the best source of drinking where the neonatal
mortality ranged from 1.5 percent for piped in to the house to 2.33 percent for public tap.
Households depending on river, canal, or rain as the source of drinking water have
slightly higher neonatal mortality.
A significant proportion of post neonatal deaths is attributed to water-borne
diseases, therefore, provision of potable water itself can bring down infant mortality in
Pakistan. The bivariate analysis shows that the families who depend on wells for drinking
water are at the highest risk of both neonatal and post neonatal mortality. The analysis
shows the association between source of drinking water and neonatal and post neonatal
mortality is statistically significant (p < 0.05).
4.1.2.2 Toilet Facilities:
The advantage o f the flush system in the household is clearly demonstrated by
the bivariate analysis (Table-4.2). The proportion of neonatal deaths in the household
equipped with flush system is only 1.67 percent, whereas the households without any
toilet facilities have the highest (3.68 percent) proportion o f neonatal deaths followed
by the households with a bucket or pit as toilet facilities. More than 50 percent of the
households in this analysis do not have any toilet facilities.
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The analysis shows the same association between the uses o f toilet facilities
and post neonatal mortality but the association is less pronounced and is not
statistically significant.
These results are in accordance with the previous studies. Several studies in
developing countries demonstrated that post neonatal mortality is closely associated
with water supply and sanitation facilities (Timaeus and Lush, 1995; Merrick, 1985).
Table-4.2: Bivariate Relationship Between Domestic Hygiene Factors and percent of deaths
during the Neonatal and Post neonatal Period, PDHS, 1990-91
Domestic Hygiene Factors % Neonatal Number % Postnatal Number
Deaths (N)__________ Deaths______ (N)
Source of Drinking Water 3302 3201
Piped into residence 2.02 628 1.26 615
Piped onto property 1.50 344 2.38 339
Public tap 2.33 255 2.95 249
Surface 2.40 275 1.51 269
Well 3.91 1800 3.57 1729
Chi-Square 10.433* 11.081*
Toilet Facility
Bush 3.68 1787 3.02 1721
Bucket 3.23 621 2.97 601
Flush 1.67 894 2.18 879
Chi-Square 8.289** 1.639
Housing Construction Material
Unbaked bricks 3.32 2031 2.96 1963
Baked bricks 2.62 1271 2.48 1238
Chi-Square 1.286 0.647
p < 0.001, ** p < 0.01, • p < 0.05. + p < 0.10
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-43: Bivariate Relationship Between Bi-Demographic Factors and percent of deaths
during the Neonatal and Post neonatal Period PDHS, 1990-91
Bio-demographic % Neonatal Number % Postnatal Number
Deaths (N) Deaths (N)
Maternal Age at the time of the Birth
15-19 2.74 112 0.73 109
20-29 2.10 1901 2.70 1861
30-49 4.48 1289 3.09 1231
Chi-Square 14.652*** 2.178
Preceding Birth Interval 3302 3200
Less than 18 months 8.28 418 4.44 383
18-35 3.09 1181 3.76 1144
36-47 2.00 964 1.60 945
48 and over 1.48 739 1.90 728
Chi-Square 49.052*** 14.856***
Birth Order
2nd 3.61 596 2.29 574
3rd-5th 2.94 1571 2.61 1525
6th or over 2.92 1135 3.27 1102
Chi-Square 0.757 1.648
Births during the last 5 years
1 1.83 1448 2.73 1421
2 3.72 1458 2.76 1404
3 and more 5.05 396 3.05 376
Chi-Square 82.596*** 0.117
Sex of the Child
Girl 3.22 1627 3.14 1574
Boy 2.89 1675 2.43 1627
Chi-Square 0.296 1.462
Previous Death (Sibling)
No 1.93 2534 1.58 2485
Yes 6.74 768 6.94 716
Chi-Square 46.124*** 59.051***
• • • p <0.001, • * p < 0 .0 1 ,» p < 0 .0 5 .+ p <0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.2 .3 Household Construction Material:
The analysis also shows that household construction material is associated with
the neonatal and post neonatal mortality. However, the association is not statistically
significant. The reason for such a difference could be because infants living in the
constructions of unbaked brick houses are more exposed to the risk o f respiratory and
other infectious diseases. It may also be due to the socioeconomic conditions of the
family which reflects the type of housing.
4.1.3 Demographic and Maternal Factors:
4.1.3.1 Maternal Age:
Table-4.3 shows that maternal age at the time of the birth o f the index child is
statistically significantly associated with neonatal mortality. The results show a J-
shaped pattern of neonatal mortality by age of the mother. Teenager mothers have the
higher proportion o f neonatal deaths (2.74 percent), whereas, the proportion of
neonatal deaths is the least (2.1 percent) for mothers in the 20-29 age-group. Mothers
in their thirties and forties have the highest proportion o f neonatal deaths. It is worth
mentioning here that in this analysis only those women are included who have at least
two live births and one o f them under 5 years of age. Therefore, it is highly likely that
most of the children o f teenager mothers who are at the highest risk of mortality are
not included in this analysis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, this is a well documental pattern in the neonatal mortality. Several
studies have reported that children bom to teenage mothers experience greater health
problems and mortality risks than those of older mothers (Cochrane and Farid, 1989;
Haaga, 1989). These findings support the earlier research. Young maternal age can
increase children’s health risks, both, for physiological and behavioral reasons.
Teenage mothers are more likely to have higher infant mortality (McDevitt et al,
1996) because they have more chances of pregnancy-related complications (Akther et
al., 1996). These young mothers may be physically less mature and less able to handle
the demands of pregnancy, childbirth, and child care.
4.1.3.2 Preceding Birth Interval:
Table-4.3 shows that the length of preceding birth intervals is significantly
associated with neonatal and post neonatal mortality in Pakistan. The result shows that
the higher the birth interval, the lower the proportion of neonatal deaths. The proportions
of neonatal deaths found are the highest (8.28 percent) if the preceding birth interval is
less than 18 months. Longer birth intervals, 36-47 months shows 2.00 percent and the
birth interval of more than 48 months shows 1.48 percent of neonatal deaths. The same
pattern is observed for post neonatal mortality, except for the interval 48 months and
above, where the proportion dead is slightly higher for babies bom after a 48 month birth
interval.
This pattern of preceding birth interval and infant and child mortality is well
documented in demographic research (Hobcraft et al, 1985; Trussell and Hammerslough,
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1983; Cleland and Sathar, 1984). Table-4.3 confirms the results of earlier studies.
Previous studies reported maternal depletion syndrome as the mechanism which
postulates that short intervals between births do not allow the mother sufficient time in a
non-pregnant, non-lactating state in order to replenish her nutritional state (Jelliffe,
1964). A mother with repeated pregnancies, especially at short intervals, does not have
sufficient time for recovery, both physically and nutritionally, and is more likely to have
pregnancy losses and babies of lower birth weight (Da Vanzo et al, 1984).
4.1 J J Births During the Last Five Years:
The number of children under five years of age in the family also has a
statistically significant association with neonatal mortality. Mothers who do not have
any other birth than the index child during the period five years preceding the survey,
have the smallest proportion of neonatal deaths compared to mothers who have more
births. The proportion of neonatal deaths is only 1.83 percent for mothers who did not
have any other births during the last five years compared to 5.05 percent of the neonatal
deaths for women who have 3 or more births during the five years preceding the survey.
The same pattern of mortality is observed for post neonatal period but the association is
less pronounced and is not statistically significant.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1J.4 Sex of the Child:
There are no statistically significant gender differences observed in the bivariate
association of neonatal and post neonatal mortality. Unexpectedly, the results show that
boys have lower neonatal mortality than girls. Earlier research in sex differentials in
mortality reported that boys are biologically weaker than girls at the time of birth and
experience higher neonatal mortality in developed countries. The observation that male
mortality rates are higher than those for females in the neonatal period is consistent with
this explanation, since most causes of death in the first month of life are either beyond
the families’ immediate control, or are not due to sex-specific treatment of children. After
the neonatal period, environmental factors that are under control of the family such as
nutritional intake, exposure to disease, breastfeeding, parental time and attention, and use
of health-care services become predominant. However, in this analysis, lower neonatal
and post neonatal mortality for boys than girls is observed.
4.1 .3.5 Previous Sibling Death:
Several studies have reported that siblings’ death tends to be associated with the
mortality of the index child even after controlling for other socioeconomic and
demographic factors (Curtis et al., 1993; Das Gupta, 1990; Hobcraft et al., 1985; Miller
et al., 1992). This analysis also found that the proportion of neonatal mortality is very
high (6.74 percent) for women who experienced earlier child death compared to the
babies of women who never experienced a child death earlier (1.93 percent). The
influence of the survival status of the preceding child on the mortality risk of the index
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
child has been explained in terms of the existence or lack of sibling competition for
maternal attention and household resources (Hobcraft, 1987; Koeing et al., 1990).
In her analysis, Das Gupta (1990) found that the probability o f a child’s death
was significantly increased if the child has siblings who died in childhood. She argued
that the women who had experienced multiple child deaths were also often less
resourceful and differed in use of basic child health care (Das Gupta, 1990). The strong
association observed for immediate pairs of siblings may indicate that familial mortality
effects change over time or with maternal age, so that conditions experienced by siblings
close in age are more similar than those for siblings bom farther apart (Zenger, 1993).
However, Guo (1993) argues that a family’s environment is likely to remain similar
throughout the time when all children are bom and raised. Table-4.3 also shows a highly
significant association (p < 0.001) between previous siblings death and post neonatal
mortality in Pakistan.
4.1.4 Dietary Factors:
For neonatal mortality, breastfeeding and supplementary foods variables were
not analyzed. Since all the babies were breastfed for at least one month, or if they died
they were breastfed until death, there would also be no variation on neonatal mortality
as a result of the effect of breastfeeding. It is also possible that if the baby was sick or
weak, it may not be breastfed. In this case, the lack of breastfeeding is not a case, but a
consequence of childhood illness during the neonatal period. For post neonatal
mortality, continuous breastfeeding and starting age of supplementary food are
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
included in the analysis as a time-varying covariate in the proportional hazards
analysis. Premature births and baby-size at birth are included in this analysis and the
results are presented below.
4.1.4.1 Premature Births:
Premature births are at a very high risk of deaths if the proper health facilities
are not made available on time. Table-4.4 shows that babies bom prematurely have a
statistically significant (p < 0.001) higher percentage of neonatal mortality compared
to full-term bom babies. The results show that about one-third (31.1 percent) of
premature births end up with neonatal mortality compared to only one death in forty
full-term births (2.6 percent).
Premature babies continue to die at a higher rate even after the neonatal period.
The bivariate analysis shows that more than 10 percent of premature babies died in
their post neonatal period compared to only 2.7 percent of full-term bom babies. This
is a highly statistically significant (p < 0.001) association. These premature bom
babies do not develop the proper sucking and results in poor stimulation and normal
lactation (Schofield and Hill, 1998). Therefore, the effect of the reduced volume of
breast-milk is that the infants do not get the protective effect of breast milk, which
provides immediate immunity after birth. Hence, these babies are more likely to die in
the post neonatal period.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.4.2 Birth-Weight/ Birth-Size:
The birth weight was not recorded for most o f the babies at the time of the
birth but baby-size was reported by the mothers in the questionnaire. The babies were
categorized into three different birth-sizes for this analysis: very small, normal, and
very large. The analysis shows that normal sized babies have the lowest percent of
death during the neonatal period, compared with very small and very large babies. It is
observed that nearly one-fourth (23.55 percent) of the very-large babies bom could not
survive during the neonatal period. The normal size babies have the lowest proportion
(2.47 percent) of neonatal deaths compared to 4.67 percent of very-small sized bom
babies.
Table-4.4: Bivariate Relationship Between Nutritional Factors and percent of Deaths
during the Neonatal and Post neonatal Period, PDHS, 1990-91
Nutritional Factors % Neonatal Number % Postnatal Number
Deaths (N) Deaths (N)
Premature Birth 3302 3201
Full-term 2.63 3253 2.70 3167
Pre-term 31.06 49 10.18 34
Chi-Square 132.247*** 6.954***
Birth-weight/Birth-size
Very-small size 4.67 215 2.68 204
Normal size 2.47 3019 2.79 2944
Very-large size 23.55 68 2.52 52
Chi-Square 102.174*** 0.022
• • • p < 0 .0 0 l,* * p < 0 .0 1 . • p < 0 .0 5 ,+ p <0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Babies bom after a long gestational but low birth weight or those of short
gestation with high birth weight are both abnormal (Yerushalmy, et al., 1965). In a
study of children bom to African American Adolescents in the United States, higher
relative risk of mortality for heavier than optimal infants is observed during the
neonatal period (Rees et al., 1999). Moreover, diabetes or glucose intolerance in
pregnancy is a risk factor for both high birth weight and diabetes in this population
(McCance et al., 1994). Higher prevalence of diabetes in the offspring of women who
had diabetes during pregnancy is observed, which is associated with high birth weight
(Phillips et al., 1998).
The survival status changes after the neonatal period for these babies. It is
observed that the proportion of deaths after the neonatal period is slightly higher for
normal-born babies compared to the very-small and very-big sized bom babies. This
may be that very-big and very-small sized babies who were in the critical conditions
died during their neonatal period and those who survived are taken care of.
Low birth weight, which is an important cause of infant mortality, can be
attributed to the nutritional status of the mother during pregnancy. Since there is a
high percentage of low birth weight babies in Pakistan, maternal and child health
services emphasized the provision of iron and folic acid tablets to pregnant women to
protect them against nutritional anemia.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.5 Health Seeking Behavior:
4.1.5.1 Prenatal Care:
Although one of the major responsibilities of the health program in Pakistan is to
provide prenatal care to pregnant women, the proportion o f women who received this
care is still very low. Only one in eight women received this opportunity in Pakistan.
Table-4.5 depicts that only 1.8 percent o f the babies died during the neonatal period of
mothers who received prenatal care during their pregnancies compared to 3.23 percent of
the neonatal deaths for those whose mothers did not go for prenatal care. However, this
association is not statistically significant. The same non-significant pattern is observed
during the post neonatal period.
4.1.5.2 Place of Delivery:
The proportion of neonatal deaths is higher among babies delivered in
government hospitals than those bom at the respondents’ homes. It is a common
observation that many pregnant women are rushed to hospitals when complications
related to the delivery arise. These last minute complicated deliveries in hospitals are at
higher risk of mortality if not attended on time. This odd finding may also reflect the
selectivity of births in the government hospitals. The proportion of neonatal deaths in
government hospitals is 5.17 percent compared to 2.95 percent for babies delivered at
respondents home.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
On the other hand, the proportion of neonatal deaths is lowest for babies
delivered at private hospitals or clinics. This may reflect the fact that people of higher
socioeconomic conditions prefer private hospitals or clinics than government hospitals to
deliver their babies. The reasons for this may be that it is generally believed that
anything worthwhile will cost money, the medical services that must be paid for are seen
as better than government services where these services are mostly free. It may also be
that the practitioners who charge for these services are expected to be more attentive and
to devote more care and concern to patients (World Bank, 1996).
The highest proportion of neonatal deaths is observed for babies delivered at
“other homes” (not at the respondent homes). These deliveries at other homes may be
delivered by the relatives or untrained dais and are moved to homes o f nearby relatives.
These may also be that the complicated cases are taken at the last moment to other
homes. Babies delivered at “other homes” have no post neonatal mortality, while babies
delivered at their own homes have the highest proportion of post neonatal mortality
followed by government hospitals (2.10 percent) and private hospitals (1.94 percent).
However, the differences are not statistically significant.
The proportion of neonatal deaths of women who delivered their babies at their
homes is only 2.95 percent, which is almost half than the proportion of neonatal deaths to
babies delivered at government hospitals. Babies delivered at home are most likely to be
attended by untrained dais and relatives as a result seem to face the least risk of infant
death. This may be only due to that the complicated cases are referred to the hospitals.
The doctors are more likely to save the babies than are the traditional birth attendants.
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.5: Bivariate Relationship Between Health Seeking Behavior and percent of Deaths
during the Neonatal and Post neonatal Period, PDHS, 1990-91
Health Seeking Behavior % Neonatal Number % Postnatal Number
Deaths (N)__________ Deaths______ (N)
Prenatal Care
No
Yes
Chi-Square
3.23
1.81
2.432
2892
409
1.130
2.90
1.75
2799
402
Place of Delivery
Own home 2.95 2824 2.95 2741
Other home 7.02 68 0.00 62
Government Hospital 5.17 187 2.10 177
Private Hospital/Clinic 1.38 223 1.94 220
Chi-Square 8.665* 3.001
Delivery Attendant
Un-trained 3.32 2158 3.13 2086
Medical Trained 2.55 1144 2.13 1115
Chi-Square 1.502 2.688+
BCG
No 7.75 1274 4.56 1175
Yes 0.10 2028 1.75 2026
Chi-Square 154.790*** 21.749***
Contraceptive Use
Ever Used 1.31 443 2.46 438
Never Used 3.32 2858 2.83 2763
Chi-Square 5.228* 0.193
• • • p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.1.5 J Delivery Attendant:
To ensure better child survival, it is essential that delivery should be conducted
in hygienic conditions and with the assistance of a trained medical practitioner. The
differences in neonatal mortality according to the delivery attendant are not statistically
significant. However, it shows a slightly higher proportion of neonatal deaths for the
deliveries attended by untrained persons compared to the deliveries attended by the
medically trained persons. The proportion of post neonatal mortality is also high (3.13
percent) for those delivered by untrained attendants compared to (2.13 percent) for those
attended by medically trained personnel.
4.1.5.4 BCG (Bacilli Calmette-Guerin) Vaccination:
BCG vaccine is given soon after the child’s birth to protect against tuberculosis.
It is the only vaccine children could have received during their first month of life. The
bivariate analysis shows a highly statistically significant association between BCG
vaccination and neonatal mortality. More than 60 percent of the babies received BCG
vaccination but only 0.10 percent of them died during their neonatal period, compared to
7.75 percent o f those babies who did not receive BCG vaccination died during their first
month of life. It is also that the sick children at birth are not given the BCG or any
vaccination. Hence, these sick children have higher chances of mortality. The use of
BCG not only benefits during the neonatal period but the same association remains
highly significant in their post neonatal mortality. The chi-square analysis shows a highly
significant association between receiving BCG and post neonatal mortality.
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to the Global Tuberculosis Control WHO Report 2000, in Pakistan
about 1.5 million people suffer from TB, and more than 210,000 new cases occur each
year. It also reports that 97 percent of the children are immunized against the BCG
compared to 67 percent in 1995. The TB Association of Karachi conducted a study that
asked 100 doctors to write a prescription for a TB patient whose case details and body
weight were given. Only one physician in seven was able to prescribe effective treatment
(WHO, 2000). The irony is that efficient diagnosis would in most cases serve only to
identify a disease that would not be treated effectively or cured in any event.
The utilization of BCG vaccination not only shows the prevention of
tuberculosis, it also shows the parental enthusiasm of utilizing the preventive measures
for their children. This vaccine is given to the children at birth, therefore, children who
received this vaccine are much more likely to get full immunization later.
4.2 PROPORTIONAL HAZARD ANALYSIS:
In this section, results of proportional hazard analyses of neonatal and post
neonatal mortality are presented. The exponential function o f each regression
coefficient shown in the tables represents the relative risk rates.
4.2.1 Neonatal Mortality:
4.2.1.1 Socioeconomic Factors:
Table-4.6 shows the hazard rate ratios and the significance level obtained from
the proportional hazard models. The proportional hazard model shows that mothers’
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
education has an inverse effect on neonatal mortality. It shows that each year increase
in the mothers’ education decreases the hazards o f neonatal mortality by 11 percent
without adjustment for other independent variable. This effect is statistically
significant (P < 0.01). However, in the multivariate model, after adjustment for other
socioeconomic variables, the effect o f maternal education disappeared.
Table-4.6 shows that the effect o f the fathers’ education on neonatal mortality
is not significant both in bivariate and multivariate analysis.
The index of the household possessions shows a significant effect on neonatal
mortality. The babies bom to mothers of higher index have statistically significantly
lower neonatal mortality than babies bom to mothers of lower index of the household
possessions. The hazard of neonatal mortality o f babies of higher household index is
only 38 percent of those of lower household index. These results are in accordance to
the earlier research, which shows that the nature of housing, diet, access to and
availability o f water and sanitary conditions as well as medical attention all depends
on the economic conditions of the household (Esrey and Habicht, 1986). Hence,
babies bom to higher socioeconomic status mothers are more likely to survive than the
babies of lower socioeconomic status mothers.
Babies bom to mothers of medium index also have lower neonatal mortality
compared to the children bom to mothers of lower index of the household. However,
the difference is not statistically significant. At the multivariate level, the index of
household possessions does not show any significant effect on neonatal mortality.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-1.6: Hazard Rate Ratios Obtained from Proportional Hazard Models o f Socioeconomic Factors for
Predicting the Neonatal Mortality, PDHS, 1990*91
Independent Model-1 Model-2 Model-3 Model-4 Modle-5 Modle-6
Variables
Years of Maternal Education 0.892 *
Years o f Education 0.981
0.916
1.005
Index o f Household Possessions
L o w e r
M e d i u m
H i g h e r
R e f e r e n c e
0.755
0.384 **
R e f e r e n c e
0.811
0.565
Place o f Residence
R u r a l
U r b a n
R e f e r e n c e
0.618 *
R e f e r e n c e
0.793
Region of Residence
N W F P & B a l o c h i s t a n
P u n j a b
S i n d h
R e f e r e n c e
1.592
1.754
R e f e r e n c e
1.760 +
2.080 *
-2 Log Likelihood 8.235 1.11
Degrees o f Freedom 1 I
8.049
2
4.242
1
3.005
2
16.7
7
• • • p < 0 . 0 0 1 , • • p < 0 . 0 1 , • p < 0 . 0 5 . + p < 0 . 1 0
A significant difference in neonatal mortality is also observed by place of
residence at bivariate level as shown in table-4.6. The hazard analysis shows
significantly lower neonatal mortality for babies bom to the residents of urban areas
compared to the babies bom to the residents of rural areas. It shows that the hazard of
neonatal mortality for children living in urban area is almost 38 percent lower than the
children living in rural areas. However, the difference in neonatal mortality by place
of residence is not significant when included in the multivariate model.
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The survival during the neonatal period also varies by region o f residence. The
risk of neonatal mortality is significantly higher for babies bom in the provinces of
Punjab and Sindh in the multivariate analysis.
4.2.1.2 Domestic Environment and Hygiene:
Improved household environmental conditions, especially the source of
drinking water and sanitation, play a major role in the decline o f childhood mortality.
In this analysis, using “wells” as a source of drinking water as the reference category,
the proportional hazard analysis shows that all other sources have a negative effect on
neonatal mortality. Having access of piped water in the house, or piped water on the
property are both associated with a significantly lower risk o f neonatal mortality. Use
of public-tap water and surface water has non-significantly lower risks of neonatal
mortality compared to the families depending on well water for drinking. These
findings are in accordance with the earlier research that says that piped water is much
better than the well or surface water.
Improved toilet facilities in the household provide better hygiene. The hazard
analysis shows that babies bom to mothers living in households with a flush facility
have significantly (P < 0.01) lower neonatal mortality than babies bom to mothers
living in households without any toilet facility. Table-4.7 shows that at the bivariate
level, the risk of neonatal mortality is 55 percent lower for children living in
households with flush facilities compared to the children bom to mothers living in
houses without any toilet facility. Babies bom to mothers living in households with
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pit or bucket as a toilet facility also have lower neonatal mortality than the children
bom to mothers living in houses without any toilet facility but the difference is not
statistically significant.
The effect of housing construction also shows that babies bom to mothers
living in houses constructed with baked bricks have a lower risk of neonatal mortality
compared to babies bom to mothers living in un-baked bricks /mud constructed
houses. However, the effect is not statistically significant.
Table-4.7: Hazard Rate Ratios Obtained from Proportional Hazard Models of Environmental Factors
for Predicting the Neonatal Mortality, PDHS, 1990-91
Independent
Variables
Model-1 Model-2 Model-3 Model-4
Source o f Drinking Water
P i p e d i n t o r e s i d e n c e
P i p e d o n t o p r o p e r t y
P u b l i c t a p
S u r f a c e
W e l l
0 . 5 1 5 *
0 . 3 8 3 *
0 . 5 9 6
0 . 6 1 2
R e f e r e n c e
0 . 6 7 4 0
0 . 4 7 0 0
0 . 6 6 4 0
0 . 6 1 0 0
R e f e r e n c e
Toilet Facility
F l u s h
B u c k e t
B u s h
0 . 4 5 2 * *
0 . 8 7 7
R e f e r e n c e
0 . 5 0 0 0 +
0 . 9 6 4 0
R e f e r e n c e
Housing Construction Material
U n b a k e d b r i c k s
B a k e d b r i c k s
R e f e r e n c e
0 . 7 9
R e f e r e n c e
1 . 2 7 7 0
- 2 L o g L i k e l i h o o d
D e g r e e s o f F r e e d o m
1 0 . 7 6 2
4
9 . 0 1 1
2
1 . 2 7 2
I
1 4 . 6 0 4
6
• p < 0.001, ** p < 0.01, • p < 0.05, + p < 0.10
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To evaluate the effect o f household environmental and hygiene factors on
neonatal mortality, when all the three variables (source o f drinking water, use of toilet
facility, and type o f construction variables) are included in the multivariate analysis,
show that availability of the flush toilet facility in the household marginally decrease
the neonatal mortality.
4.2.1.3 Demographic Factors:
As expected the proportional hazard analysis shows that younger mothers and
older mothers have higher relative risks of neonatal mortality. The relative risk of
neonatal mortality for children bom to mothers of less than 20 years is 18 percent
higher than the babies bom to mothers aged 20-29, but the difference is not
statistically significant. The relative risks o f neonatal mortality are 40 percent higher
for babies of mothers aged over 30 years age than the babies bom to mothers aged 20-
29 years in the bivariate model. In the multivariate model, the pattern of neonatal
mortality is changed by age of the mother. The risk of neonatal mortality is lower for
children o f mothers aged 15-19. This pattern is different than the U-shaped
relationship between age of mother at birth of the index child and neonatal mortality
observed across developing countries. The reason for this difference is that mothers
who have only one child are not included in this analysis, as mentioned earlier in
chapter 3. Younger mothers (15-19) are more likely to have first births, which are at
higher risk o f mortality. The exclusion of those babies from the analysis reduced the
risk o f neonatal mortality for this age-group.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.8 also shows that the presence of a previous birth interval has a
significant effect on neonatal mortality. The mortality risk of babies bom within 18
months o f a previous child is more than 4 times higher than the babies bom during the
36-47 month birth interval. This difference is highly statistically significant (p <
0.001). Babies bom within the interval of 18-35 months also have 1.5 times higher
risk o f neonatal mortality than the babies bom to during 36-47 months interval. This
difference is not statistically significant. The analysis shows that the hazard risk of
neonatal mortality for babies bom after 48 months are lower than the babies bom
within 36-47 months interval. However, the difference is not statistically significant.
The same pattern is observed in the multivariate model. It means that the risk of
neonatal mortality decreases with increasing length of the preceding birth interval.
As observed in the chi-square analysis in the previous section, the hazard
analysis shows no statistically significant association between the sex of the child and
neonatal mortality. Table-4.8 shows that the risk of neonatal mortality is lower for
boys than girls. However, earlier research in other countries shows that boys are more
likely to die during the neonatal period compared to girls due to biological reasons.
The analysis also shows that birth order higher than 5 has a lower risk of
neonatal mortality than the birth order 3-5 in the bivariate model. The same pattern is
observed in the multivariate model but the difference is statistically significant (p <
0.01). However, the higher the numbers of children below five years age are in the
family, the higher the risk of neonatal mortality is recorded in this analysis, both, in
the bivariate and multivariate models.
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.8: Hazard Rate Ratios Obtained from Proportional Hazard Models o f Demographic Factors for
Predicting the Neonatal Mortality, PDHS, 1990-91
Independent Model-1 Model-2 Model-3 Model-4 Model-5 Model-6 Model-7
Variables
Maternal Age at the time of the Birth
1 5 - 1 9 1 . 1 7 9
2 0 - 2 9 R e f e r e n c e
3 0 - 4 9 1 . 3 9 6 * *
0 . 7 6
R e f e r e n c e
2 . 8 9 2 * * *
Preceding Birth Interval
L e s s t h a n 1 8 m o n t h s
1 8 - 3 5
3 6 - 4 7
4 8 a n d o v e r
4 . 1 3 1 '
1 . 5 4 1
R e f e r e n c e
0 . 6 9 8
««»
2 . 8 4 8 * * *
1 . 3 2 5
R e f e r e n c e
0 . 6 1 5
Birth Order
^ n d
2 id
6 11 1 a n d a b o v e
1 . 2 2 8
R e f e r e n c e
0 . 9 9 5
1 . 3 8 2
R e f e r e n c e
0 . 4 6 8 * *
Sex o f the Child
G i r l
B o y
R e f e r e n c e
0 . 9
R e f e r e n c e
0 . 9 8 0 0
Number of Children Under 5
Previous Sibling Death
1 . 6 8 5 * * *
3 . 4 9 0 * * *
1 . 5 5 7 * * *
3 . 9 0 9 * * *
- 2 L o g L i k e l i h o o d 6 . 7 7 2
D e g r e e s o f F r e e d o m 2
3 7 . 6 5 4
2
0 . 7 0 2
2
0 . 2 9
1
1 3 . 5 9 3 7 . 4 1
1
9 4 . 2 2 4
9
• • • p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Babies of mothers who have experienced earlier child death have a
significantly higher risk of neonatal mortality. The risk o f death during the neonatal
period for these babies is 3.5 times higher than the babies bom to mothers who did not
experience a previous child death. This relationship is highly statistically significant.
In summary, when all of these demographic variables are included in the
proportional hazard analysis, it shows that all of these significantly contribute to the
risk of neonatal mortality. Mothers older than 30 years, and with a birth interval within
18 months, have higher hazards o f neonatal mortality. The numbers of siblings under
age 5 and previous sibling death in the family are also positively associated with
neonatal mortality. However, it shows that there is no gender difference in neonatal
mortality.
4.2.1.4 Nutritional Factors:
Table-4.9 shows that the risk of neonatal mortality is 1.9 times higher for
small-size bom babies and 9.5 times higher for very-large bom babies compared to the
normal sized babies. These very-large babies are more likely to be the offspring of
women who had diabetes during pregnancy which is associated with high birth weight
(Phillips et al., 1998). The higher neonatal mortality for high birth weight babies was
also found in the children bom to African American Adolescents in the United States
(Rees etal., 1999).
The risk o f death of prematurely bom babies is almost 12 times higher than
that o f full-term babies. Premature births are biologically weak and more likely to
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
contribute to neonatal mortality in developing countries. Premature birth with its
consequent low birth weight is one of the outcomes o f what has been described as the
“maternal depletion syndrome”.
Table-4.9: Risk Ratios Obtained from Proportional Hazard Models o f Nutritional
Factors Predicting the Neonatal Mortality, PDHS, 1990-91
Independent
Variables
Model-1 Model-2 Model-3
Premature Birth
F u l l - t e r m
P r e - t e r m
R e f e r e n c e
1 1 . 8 1 5 * * *
R e f e r e n c e
1 1 . 7 1 4
Birth-weight/Birth-size
V e r y - s m a l l s i z e
N o r m a l s i z e
V e r y - l a r g e s i z e
1 . 8 9 •
R e f e r e n c e
9 . 5 2 1 0 * * •
1 . 1 1 5
R e f e r e n c e
9 . 2 8 2 • • •
-2 Log Likelihood
Degrees of Freedom
4 5 . 3 0 1
I
4 2 . 8 1 2
2
8 5 . 7 1 4
3
• • • p < 0.001, • • p < 0.01, • p < 0.05, -r p < 0.10
4.2.1.5 Health Seeking Behavior:
In table-4.10, results of the proportional hazard model shows that prenatal care
is negatively associated with neonatal mortality. As the number o f prenatal care visits
increases, the risk of neonatal mortality decreases. As seen in the bivariate analysis in
the previous section, children bom in government hospitals have a higher risk of
neonatal mortality relative to children bom in their own houses. The reason may be
that mothers who anticipated problems in delivery are more likely to go to government
hospitals to deliver their babies compared to mothers who did not have any
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
complications. The hazard analysis shows that babies delivered at government
hospitals have the relative risk o f 1.75 times higher (p < 0.05) o f neonatal mortality
compared to the babies delivered at home at bivariate level, but it has almost 4 times
higher hazards of neonatal mortality at multivariate level (p < 0.001). On the other
hand, although the health services at private hospitals are very costly in Pakistan, those
mothers who want good care prefer to visit a private facility for their delivery. The
relative risks of neonatal mortality are lower for those babies delivered at private
hospitals than the deliveries at home, but the difference is not statistically significant.
As observed earlier, babies delivered in houses other than the mother’s are at a
significantly higher risk of neonatal mortality. The bivariate analysis shows that the
risk ratios are more than twice for babies bom at other homes compared to the babies
bom at the respondents home (RR=2.35, p < 0.05). In the multivariate model, the risk
ratios are even more pronounced (RR=2.90, p < 0.05).
The hazard analysis shows that deliveries attended by medical personnel have
a lower risk of neonatal mortality than those delivered by untrained personnel, but the
difference is not statistically significant. Mothers who immunized their babies after
birth for BCG vaccine have a significantly lower risk of deaths during neonatal period
compared to babies who did not receive the BCG vaccine at birth (RR=0.013, p <
0.001). One reason for such difference may be that babies who were not healthy at the
time o f birth most likely not to be given BCG at birth. Therefore, only the healthy
babies are more likely to get BCG in the first month. Those babies who received the
BCG in the first month are also more likely to receive full immunization.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.10: Risk Ratios Obtained from Proportional Hazard Models o f Health Seeking Behavior
for Predicting the Neonatal Mortality, PDHS, 1990-91
Independent Model-1 Model-2 Model-3 Model-4 Model-5 Model-6
Variables_________________________________________________
Prenatal Care
N o R e f e r e n c e
Y e s 0 . 9 2 3 +
R e f e r e n c e
0 . 9 8 8
Delivery Attendant
U n - t r a i n e d
M e d i c a l T r a i n e d
R e f e r e n c e
0 . 7 6 8
R e f e r e n c e
0 . 8 1 8
Place of Delivery
O w n H o m e
G o v e r n m e n t H o s p i t a l
P r i v a t e H o s p i t a l / C l i n i c
O t h e r H o m e s
R e f e r e n c e
1 . 7 5 4 +
0 . 4 6 8
2 . 3 8 0 +
R e f e r e n c e
3 . 9 5 2 * * *
1 . 5 0 4
2 . 9 0 2 *
BCG
N o
Y e s
R e f e r e n c e
0 . 0 1 3
« * *
R e f e r e n c e
0 . 0 1 3 * * *
Contraceptive Use
N o
Y e s
R e f e r e n c e R e f e r e n c e
0 . 3 9 6 * 0 . 7 0 3
-2 Log Likelihood
D e g r e e s o f F r e e d o m
3 . 5 6 4 1 . 4 9 9
1 2
7 . 5 0 3
3
1 6 8 . 7 8
1
6 . 2 5 4
1
1 8 2 . . 0 7
7
• • • p < 0.001, **p< 0.01. • p < 0.05, p < 0.10
Children born to mothers who used contraceptives have a lower risk of
neonatal mortality. At the bivariate level, it is observed that children o f mothers who
had used contraceptives have a 60 percent lower risk o f neonatal mortality compared
to the children of mothers who did not use contraceptive methods. One of the
objectives of the family planning program in Pakistan is to provide contraceptives to
mothers for birth spacing. Those women who used contraceptives for birth spacing,
and babies bom within a larger birth interval, have lower neonatal mortality.
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In sum, when all the health care variables are included in the proportional
hazard model, it is observed that babies delivered at government hospitals have almost
4 times higher risk of neonatal mortality compared to babies delivered at home. The
babies delivered at other homes also have almost 3 times higher risk o f neonatal
mortality compared to the babies delivered at respondents own home. BCG vaccine
and contraceptive use both have negative effect on neonatal mortality.
4.2.1.6 Socioeconomic Factors and Proximate Determinants:
Table-4.11 depicts the multivariate results o f the proportional hazard models in
relation to neonatal mortality. The results are shown in the form o f hazard risk ratios
and the significance level. Among the socioeconomic factors in model-1, children
living in the provinces of Punjab (RR=1.76, p < 0.10) and Sindh (RR=2.08, p < 0.05)
have significantly higher neonatal mortality compared to the children living in NWFP
and Balochistan provinces (Reference). When the environmental factors are added in
model-2, it is observed that the neonatal mortality difference in living in Punjab
disappeared. However, residence in the Sindh province still has a significantly higher
neonatal mortality. When demographic variables are included in the model, it is seen
that the maternal age over 30 years, previous birth intervals less than 18 months,
number of children under age 5 years, and previous sibling death have positive
significant effects in increasing the risk of neonatal mortality.
In model-4, when the baby-size and premature births are included, both
premature births and baby-size at birth have significant effects on neonatal mortality.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Premature bom babies have 13.6 times higher risk o f neonatal mortality than full-term
bom babies. Babies of very-large size also have more than 8.5 times higher risk of
neonatal mortality than normal-sized bom babies. Another interesting change to note
is that the significance of higher neonatal mortality in the province of Sindh was
spurious. The higher neonatal mortality in Sindh was due to higher proportion of
premature births and babies of very large-sized at birth. The neonatal mortality in the
province of Punjab is slightly increased, however, it is not significant.
Including health seeking behavior variables in model-5, it is seen that babies
who had BCG vaccine at birth are at a significantly lower risk of neonatal mortality
than those who did not receive BCG immunization at birth. The utilization o f other
preventive health services, such as prenatal care and delivery attended by a medical
personnel, did not show any significant effect on neonatal mortality. However, babies
delivered at government hospitals maintained a higher risk o f neonatal mortality than
babies delivered at home (p < 0.10). Again, including the health seeking variables in
the model, it is observed that the effect of Sindh reduced and the Punjab increased.
In this analysis, it is observed that demographic, nutritional and health seeking
behavior factors are responsible for higher neonatal mortality in Pakistan. The higher
neonatal mortality observed in Sindh was due to the higher proportion of premature
births and the lower utilization of health services.
To test if any behavioral factors may also affect the outcome of pregnancy
other than the biological mechanisms operating in the prenatal period, the interactions
between demographic and health care variables are included in model-6 (table-4.12).
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.11: Hazard Rate Ratios obtained from Proportional Hazard Model for Predicting
Neonatal Mortality, PDHS, 1990-91
Independent Variables Model-1 Model-2 Model-3 Model-4 Model-S
Years o f Maternal Education 0.920 0.927 0.945 0.897 0.941
Years o f Paternal Education 1.005 1.008 1.008 1.009 1.010
Household Possessions Index: Lower (Ref)
Higher 0.565 0.612 0.508 0.478 0.557
Medium 0.811 0.813 0.851 0.807 0.882
Place o f Residence: Urban 0.793 1.009 1.192 1.243 1.388
Province o f Residence:NWFP&Baiochistan (Ref)
Punjab 1.760 + 1.655 1.304 1.455 1.830
Sindh 2.080 • 2.059 • 2.090 - 1.347 1.294
Source o f Drinking W ater Well (Ref)
Pipe in to the house 0.707 0.648 0.743 0.666
Pipe on to the Property 0.534 0.554 0.526 0.646
Public Tap 0.670 0.562 0.649 0.715
Surface 0.605 0.557 0.577 0.488
Latrine Facility: No Facility (Ref)
Bucket or Pit 1.827 1.791 1.596 1.300
Flush 1.585 1.516 1.584 1.038
Construction Baked Bricks 1.335 1.048 1.181 0.936
Mother Age: 20-29 (Ref)
15-19 0.677 0.746 0.822
30-49 2.771 • • • 2.419 • • • 2.354 • • •
Previous Birth Interval: 36 - 47 (Ref)
Less than 18 months 3.147 • • • 2.992 • • • 3.698 • • •
18-35 1.445 1.187 1.513
48 and more 0.607 0.689 0.783
Birth O rder 2nd (ref)
3rd - 5th 1.557 1.454 1.213
6th and above 0.442 • • • 0.523 • • 0.555 *
Sex o f the Child:Boy 0.979 0.937 1.054
Number o f Children under 5 1.561 • • 1.598 •* 1.468 •
Sibling Death 3.400 2.783 • • • 2.484 • • •
Premature Birth 13.648 • • • 6.442 • • •
Baby-size at Birth:Average (Ref)
Small-size 0.974 0.889
Big-size 8.557 4.601 • • •
Prenatal care visits 0.982
Place o f Delivery: At Home (Ref)
At Government Hospital 2.546 t
At Private Hospital 1.921
At Home other than Mother's 1.812
Delivery Attended by Medical Personal 0.692
BCG Vaccine received 0.017
Contraceptive Use 1.057
Log likelihood
16.7 23.68 124.8 198.95
324.3
D eg rees of F reed o m
7 14 24 27
34
p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.12:Hazard Rate Ratios obtained from Proportional Hazard Model for
Predicting Neonatal Mortality, PDHS, 1990-91
Independent V ariables Model-6
Years o f Maternal Education 0.933
Years o f Paternal Education 1.012
Household Possessions Index: Lower (Ref)
Higher 0.623
Medium 0.894
Place o f Residence: Urban 1.05
Province o f Residence:NWFP&Balochistan (Ref)
Punjab 2.101 +
Sindh 1.448
Source o f Drinking W ater Well (Ref)
Pipe in to the house 0.736
Pipe on to the Property 0.695
Public Tap 0.782
Surface 0.721
Latrine Facility: No Facility (Ref)
Bucket or Pit 1.896
Flush 1.205
Construction Baked Bricks 0.928
Mother Age: 20-29 (Ref)
15-19 0.71
30-49 2.514 • • •
Previous Birth Interval: 36 - 47 (Ref)
Less than 18 months 10.114 • • •
18-35 3.246 • •
48 and more 0.741
Birth O rder 2nd (ref)
3 rd - 5 th 1.356
6th and above 0.462 • • •
Sex o f the Child: Boy 1.102
Number o f Children under 5 1.496 •
Previous Sibling Death 7.901 • • •
Premature Birth 6.566 • • •
Baby-size at Birth: Average (Ref)
Small-size 0.981
Big-size 4.714 • • •
Prenatal care visits 0.699 •
Place o f Delivery: At Home (Ref)
At Government Hospital 2.843 •
At Private Hospital 2.154
A t Home other than Mother's 2.146
Delivery Attended by Medical Personal 0.739
BCG Vaccine received 0.016 • • •
Contraceptive used 0.867
Interactions
Urban & Prenatal Care 1.493 •
Space < 18 & Prenatal Care 0.16 • • •
Space 18-35 & Prenatal Care 0.238 • •
Log likelihood 346.325
D e g re e s o f Freedom 37
• • • p < 0.001, •* p < 0.01, • p < 0 .0 5 , + p < 0 .1 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It is hypothesized that pregnant mothers with short spacing o f births still have
very young children and may not take prenatal care services that would otherwise be
the case. On the other hand, pregnant women with longer birth intervals are more
likely to attend prenatal care services which ultimately results in a healthy child birth
(Boerma and Bicego, 1992). When the interactions between prenatal care and
previous birth intervals are included in the model, several changes in the significance
of other variables are also noted.
First of all, this analysis supports the argument that, if mothers with shorter
previous birth intervals have used prenatal care, their babies are significantly more
likely to have better survival chances during the neonatal period than those mothers
with the same short birth interval who did not receive prenatal care for the index child.
The hazard analysis shows that the babies of short previous birth intervals have
statistically significantly improved chances of survival during the neonatal period if
their mothers had attended the prenatal care. It is also observed that babies delivered
at government hospitals have significant higher neonatal mortality than babies
delivered at respondents own home even after controlling for all other variables.
The interaction model also shows that babies of mothers living in rural areas,
and who received prenatal care, have significantly higher chances of survival during
neonatal period than their counterparts living in urban areas. The results clearly
demonstrate the importance of prenatal care in rural areas. If mothers go for prenatal
care to check-up and to help detect and manage some pregnancy-related complications
and to educate women about danger signs, potential complications, and where to seek
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
help (WHO, 1994). Early prenatal care provides an opportunity to offer preventive
care that will benefit the infant as well as the mother such as, counseling on hygiene,
breastfeeding and nutrition.
A higher neonatal mortality in Punjab is also observed compared to NWFP and
Balochistan. This shows that women with short birth intervals are less likely to utilize
prenatal care in Punjab compared to other provinces and, also, rural women are less
likely to utilize the prenatal care in Punjab compared to the other provinces.
4.2.2 Post neonatal Mortality:
4.2.2.1 Socioeconomic Factors:
Table-4.13 presents the hazard rate ratios obtained from the proportional
hazard analysis for predicting the post neonatal mortality by socioeconomic variables.
The bivariate analysis shows a non-significant effect of maternal years of education on
post neonatal mortality. The bivariate model shows that the fathers’ education has a
negative statistically significant (RR=0.912, P <0.001) effect on post neonatal
mortality. It shows that each year increase in the fathers’ education decreases the risk
ratios o f post neonatal mortality by 9 percent. The index of household possessions
shows a non-significant effect on post neonatal mortality. Hazard analysis shows that
lower post neonatal mortality is observed for urban babies compared to the babies of
rural residents but the effect is not significant.
Unlike the neonatal mortality, which is determined mostly by demographic,
nutritional and health seeking behavior factors, the proportional hazard model shows
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that one of the important determinant o f post neonatal mortality is the father’s
education. When all the socioeconomic variables are included in the multivariate
hazard model, the fathers’ years o f schooling holds its importance in reducing post
neonatal mortality after controlling for other socioeconomic variables. It is also
observed that the post neonatal mortality is significantly higher for the children living
in the province of Sindh (RR=2.2l, p < 0.05) compared to the children living in
NWFP and Balochistan.
Table-4.13: Hazard Rate Ratios Obtained from Proportional Hazard Models of Socioeconomic Factors for
Predicting the Post Neonatal Mortality, PDHS, 1990-91
Independent Model-1
Variables
Model-2 Model-3 Model-4 Modle-5 Modle-6
Years of Maternal Education 0 . 9 3 6 1.012
Years of Paternal Education 0 . 9 1 2 * * • 0 . 9 1 2 • •
Index o f Household Possessions
Lower
Medium
Higher
R e f e r e n c e
0 . 5 5 1
0 . 8 8 2
R e f e r e n c e
0.931
1.031
Place o f Residence
Rural
Urban
R e f e r e n c e
0 . 7 2 9
R e f e r e n c e
0.816
Region of Residence
NW FP & Balochistan (Ref)
Punjab
Sindh
R e f e r e n c e
1 . 5 7 2
2 . 0 9 7
R e f e r e n c e
1.702
2.414 •
-2 Log Likelihood
Degrees of Freedom
3 . 1 2 5
1
1 3 . 2 5 1
1
3 . 0 1 5
2
1 . 7 2 1
1
4 . 3 1 6
2
1 9 . 3 6
7
• • • p < 0.001, • • p < 0.01, • p < 0.05. + p < 0.10
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Contrary to the expectations, the effect o f maternal education on post neonatal
mortality is not significant. Even the sign of the relationship between maternal
education and post neonatal mortality reversed after introducing other socioeconomic
variables in the proportional hazard model. Similar results have been observed in
Pakistan by using other data (Agha, 1995). Agha (1995) found that, by including the
fathers’ education in the multivariate model, the effect of maternal education becomes
non-significant. The attenuation in the effect of maternal education suggests that other
socioeconomic variables are stronger predictors of post neonatal mortality. However,
among other socioeconomic variables, only region o f residence shows a significant
difference in post neonatal mortality, whereas, index of household possessions, and
place of residence have not shown any significant effect on post neonatal mortality.
In the United Nations and WFS comparative studies, no significant net effect
of maternal education on child mortality was found in Nepal, Jamaica, Sudan, Haiti,
Guyana, Trinidad and Tobago, Syria and Fiji (Cleland and van Ginneken, 1988).
Cleland and van Ginneken (1988) concluded that the fewer number of educated
women in Nepal and the data quality of other countries, except for the Caribbean, are
responsible for these exceptional results.
On the other hand, a study on the covariates of infant and child mortality in the
Philippines, Indonesia, and Pakistan, found that the father’s education makes a
difference only at older ages of children (Martin et al. 1983). In this analysis the post
neonatal period included 1-23 months of age, so the finding that father’s education
shows as a significant predictor of post neonatal mortality is not unexpected. In
1 4 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
general, the father’s education is strongly interrelated with their incomes (Caldwell et
al., 1989). There was no such variable included in PDHS to measure the income of
the household. The lower post neonatal mortality observed may be due to the fact that
the variation in the father’s education associated with class and status differences and,
perhaps for these reasons, the father’s education rivaled the explanatory effectiveness
o f the mother’s education.
4.2.2.2 Domestic Environment and Hygiene:
Similar to the neonatal mortality, proportional hazard analysis shows that
babies bom to women living in households connected with piped water have
significantly lower post neonatal mortality than the babies bom in households
depending on water from well. The other sources of drinking water do not
significantly influence post neonatal mortality compared to households with wells
water. Table-4.14 shows that children of mothers living in houses connected with
piped water have 65 percent lower hazards of post neonatal mortality (RR=0.35, p <
0.01) in the bivariate model and in the multivariate model, the children of mothers
connected to piped water have 68 percent lower risk of post neonatal mortality.
As expected, the hazard analysis shows that households with buckets as source
of latrine have higher post neonatal mortality than the households with flush system,
however, the difference is not statistically significant. The improved toilet facilities in
the household provide better hygiene. It also shows that households without any toilet
facilities have a higher post neonatal mortality compared to the households with flush.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.14: Hazard Rate Ratios Obtained from Proportional Hazard Models o f Environmental
Factors Predicting the Post neonatal Mortality, PDHS, 1990-91
Independent Variables Model-1 Model-2 Model-3 Model-4
Source of Drinking Water
P i p e d i n t o r e s i d e n c e
P i p e d o n t o p r o p e r t y
P u b l i c t a p
S u r f a c e
W e l l
0 . 3 4 7 * *
0 . 6 6 3
0 . 8 3 4
0 . 4 0 7
R e f e r e n c e
0 . 3 2 * *
0 . 6 1 7
0 . 7 9 4
0 . 4 1 2
R e f e r e n c e
Toilet Facility
B u s h
B u c k e t
F l u s h
R e f e r e n c e
0 . 9 6 9
0 . 7 2 6
R e f e r e n c e
1.111
1 . 0 9 7
Housing Construction Material
U n b a k e d b r i c k s
B a k e d b r i c k s
R e f e r e n c e
0 . 8 3 6
R e f e r e n c e
1 . 0 4 9
- 2 L o g L i k e l i h o o d
D e g r e e s o f F r e e d o m
1 2 . 6 9 8
4
1 . 5 6 4
2
0 . 6 5 4
1
1 2 . 9 4 5
7
• • • p < 0.001, • • p < 0.01, • p < 0.05, + ■ p < 0.10
The non-significant effect o f baked bricks construction is observed during the
post neonatal mortality compared to babies bom in un-baked bricked constructed
houses.
The multivariate analysis, when all the three environmental variables are
included, shows that only babies bom in the houses using piped water in to their
houses have statistically significant negative effect on post neonatal mortality
compared to the babies bom in households with well water. Toilet facility and
construction material do not show significant influence in post neonatal mortality.
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.2.3 Demographic Factors:
The hazard analysis shows that babies o f younger mothers have lower risks of
post neonatal mortality than the infants bom to mothers aged 20-29 years. However,
the difference is not statistically significant. Post neonatal mortality risks are
significantly higher for mothers aged 30-49 than post neonatal mortality of babies
bom to mothers aged 20-29. Table-4.15 shows that children o f mothers 30-49 years
old have 2.38 times higher risk of post neonatal mortality than the children of mothers
in 20-29 years of age (RR=2.38, p < 0.001) in the multivariate model.
The analysis of WFS countries shows that the risk of post neonatal mortality is
higher for babies bom to young mothers and gradually decreases to reach a minimum
for women 25-30 years old in most of the developing countries (Rutstein, 1984: 27)
except Pakistan and Bangladesh where it is at a minimum for women 35-39 years and
thereafter rises again (Aiam and Cleland, 1984: 194; AI-Kabir, 1984: 15; Rutstein,
1984: 27). Teenage births are at a higher risk o f death than births o f women aged 20-
29 years, and even the risk is higher in countries that have enjoyed some success in
bringing down the infant mortality rates (McDevitt et al., 1996).
The effect of previous birth intervals on post neonatal mortality is also
significant, following the same pattern observed for neonatal mortality. In the bivariate
model, the hazard o f post neonatal mortality o f babies bom within the birth interval of
less than 18 months is significantly higher (RR=2.53, P < 0.001) compared to the
babies bom during the 36-47 months interval. Babies bom within the interval of 18-
35 months also have significantly higher (RR=2.33, p < 0.01) post neonatal mortality
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
than the babies bora within the 36-47 months birth interval. The analysis shows that
the hazard risk o f post neonatal mortality for babies bom after 48 months are also
higher than those bora within 36-47 months interval.
Table-4.15: Hazard Rate Ratios Obtained from Proportional Hazard Models o f Demographic F actors
for Predicting the Post neonatal Mortality, PDHS, 1990-91
Independent Model-1
Variables
Model-2 Model-3 Model-4 Model-5 Model-6 Model-7
Maternal Age at the time of the Birth
1 5 - 1 9 0 . 2 9 8
2 0 - 2 9 R e f e r e n c e
3 0 - 4 9 1 . 7 0 4 * *
0 . 1 9 7
R e f e r e n c e
2 . 3 7 5 • • •
Preceding Birth Interval
L e s s t h a n 1 8 m o n t h s
1 8 - 3 5
3 6 - 4 7
4 8 a n d o v e r
2 . 5 2 8 0 * * *
2 . 3 3 3 0 • •
R e f e r e n c e
1 . 2 4 3 0
1 . 8 9 7
2 . 1 1 6 * *
R e f e r e n c e
1 . 0 3 8
Birth Order
2 n d
3 r d - 5 t h
6 t h a n d a b o v e
0 . 9 2 4
R e f e r e n c e
1 . 2 0 4
1 . 4 1 6
R e f e r e n c e
0 . 5 9 8 •
Sex o f the Child
G i r l
B o y
R e f e r e n c e
0 . 7 7 8
R e f e r e n c e
0 . 8 2 5 0
Number of Children Under 5 1 . 3 5 * 1 . 4 3 9 •
Previous Sibling Death 3 . 9 6 7 * • • 4 . 6 1 4 * * *
- 2 L o g L i k e l i h o o d 9 . 2 1 5
D e g r e e s o f F r e e d o m 2
1 4 . 7 2 8 0 . 9 5 7
2
1 . 3 9 5 3 . 5 5 8 4 0 . 6 4
1 1
7 3 . 4 7 5
1 0
• • • p < 0.001, * * p < 0.01, * p < 0 .0 5 , + p < 0.10
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This U-shaped pattern of previous birth interval and post neonatal mortality is
observed in developing countries including Pakistan. It has been observed that
children bom within an interval o f less than 18 months experienced higher mortality
risks during infancy than those bom in an interval of two to three years (Hobcraft et
al., 1985: 371; Cleland and Sathar: 1984: 414; Winikoff, 1983: 233).
Previous studies have concentrated on three main mechanisms for this
relationship. First, maternal depletion syndrome, which postulates that short repeated
pregnancies, close spacing, and extended periods of breastfeeding resulted in a
reduced quality of mothers’ milk and a general decline in maternal health, evidenced
by progressive weight loss and a prematurely aged appearance (Jelliffe, 1976; Jelliffe
and Jelliffe, 1978). A mother with repeated pregnancies, especially at short intervals,
does not have sufficient time for recovery, both physically and nutritionally, and is
more likely to have pregnancy losses and babies o f lower birth weight (Da Vanzo et
al, 1984). This also may be due to the amount o f pressure on the mother for the care of
the children.
Secondly, competition among siblings is considered a plausible mechanism in
the association between birth intervals and child survival: the newborn child has to
compete with another young sibling for household resources and the mother's care.
The situation may have a bearing on the nutrition o f the youngest child (Winikoff,
1983; Boerma et al., 1992). Boerma and Bicego (1992) suggested that this could
impact on the nutritional status o f the index child, on incidence of morbidity, and on
higher mortality from illness as well as accidents.
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Finally, it is also suggested that there is higher exposure to infectious disease
for the younger child (Boerma and Bicego, 1992). The older sibling o f a short birth
interval reaches an age in which infectious disease is particularly prevalent just as the
younger child is particularly vulnerable because immunity acquired from the mother
has declined, but the infant has not yet fully acquired its own immunity.
In the multivariate model, only children bom within the previous birth interval
of 18-35 months have significantly higher (RR=2.12, p < 0.01) post neonatal mortality
compared to the children bom within 36-47 months birth intervals
The hazard analysis shows a lower hazard of post neonatal mortality for boys
than for girls, but the difference is not statistically significant. During the neonatal
period boys experience higher neonatal mortality due to the biological disadvantages
(Preston, 1976), during post neonatal period that biological disadvantage no more
persists and equal post neonatal mortality is expected in non-sex-discrimination
populations. However, lower post neonatal mortality for boys is also observed in
WFS and DHS countries. Evidence from studies in India and Bangladesh, indicates
that higher female mortality rates in childhood after the neonatal period, result from
preferential treatment by family members for sons (Chen et al. 1981; Miller 1981; Sen
and Sengupta 1983; Bhatia 1983; Das Gupta 1987; Basu 1989; Freed and Freed 1989).
Babies o f mothers who have experienced an earlier child death continue to
have significantly higher risks of post neonatal mortality. The hazard analysis shows
that risk of post neonatal mortality for those babies who have already lost a sibling are
significantly higher than the babies who did not experience a sibling death earlier.
1 5 5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This relationship is highly statistically significant. Sibling deaths tend to be correlated
due to the same risks associated with the home environment and with their mother’s
health and reproductive behavior (Zenger, 1993; Curtis et al., 1993; Das Gupta, 1990).
The proportional hazard analysis o f demographic determinants shows that
there are several variables contributing significantly not only to the neonatal mortality
but also significantly to post neonatal mortality.
After controlling for other demographic factors, maternal age o f older than 30
years (RR=2.38, p < 0.001) and previous birth interval of 18-35 months have higher
hazards of post neonatal mortality (RR=1.90, p < 0.001) per child. The number of
siblings under age 5 in the family is also positively associated with post neonatal
mortality (RR=1.44, p > 0.05). However, the hazard analysis shows that there is no
significant gender difference in post neonatal mortality. Previous sibling death also
significantly increases the risk of post neonatal mortality (RR=4.61. p < 0.001).
4.2.2.5 Nutritional Factors:
Breastfeeding and supplementary food are very important determinants of
infant and child mortality in developing countries. It is observed from the bivariate
hazard analysis that, as the length of continuous breastfeeding increases, the survival
of the child during the post neonatal period increases (RR=0.59, p < 0.01). The age of
starting supplementary food is also observed having a highly significant positive effect
on post neonatal mortality (RR=4.35, p < 0.001). The later the start o f supplementary
food, higher the post neonatal mortality is observed.
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.16: Hazard Rate Ratios Obtained from Proportional Hazard Models o f Nutritional Factors
for Predicting the Post neonatal Mortality, PDHS, 1990-91
Independent M odel-t M odel-2 Model-3 M odel-4 Model-S Model-6
Variables
Premature Birth
F u l l - t e r m Reference Reference
P r e - t e r m 4.23 • • 4.515 • •
Birth-weight/Birth-size
V e r y - s m a l l s i z e 1.1 0.886
N o r m a l s i z e Reference Reference
V e r y - l a r g e s i z e
0.7780 0.716
Breastfeeding
0.5890 + 0.448 • •
Age o f nou-Supplementary food
4.354 • • • 4.695 • • •
Feeding with Nipple
0.914 0.84
-2 Log Likelihood 4 . 6 0 3 0 . 0 8 9 2 . 7 9 6 2 8 . 4 0 . 1 5 1 3 9 . 0 2 2
Degrees o f Freedom
1 2 1 1 1 6
• • • p < 0.001. • • p < 0.01, • p < 0.05, + p < 0.10
Based on the empirical findings of studies in developing countries,
breastfeeding practices have at least three known mechanisms by which it contributes
to infant health and survival. First, Breastfeeding contains the optimal combination of
nutrients, which suits the baby’s metabolic structure. Second, breastfeeding allows the
mother to pass on immunities that she herself acquired to the baby. Third, breastfed
children receive less of other food and liquid, which may be contaminated with
disease-causing agents (Briend et al., 1988, Cabigon, 1997; Yoon et al., 1996; Jelliffe
and Jelliffe, 1978). This effect is more pronounced in populations with higher infant
mortality and poor socioeconomic conditions as well as for children living in rural
areas o f poor families and o f less educated mothers (Goldberg et al., 1984; Palloni and
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tienda, 1986; Ratherford et al., 1989). Breastfeeding also provides important
protection against infectious diseases (Jelliffe and Jelliffe, 1978; Feachem and
Koblinski, 1984; Jason et al., 1984; Cunninghan et al., 1991; Victoria, 1996), which
account for over two-thirds o f the 12 million annual deaths in children younger than 5
years in less developed countries (Murray and Lopez, 1996).
In the bivariate analysis, bottle-feeding does not show any significant effect on
post neonatal mortality. It was thought that bottle-feeding with nipples may have a
negative effect on post neonatal mortality if the bottle or the nipple are not properly
cleaned.
As observed for neonatal mortality, the risk o f post neonatal mortality for
premature bom babies in the bivariate analysis is significantly higher compared to the
post neonatal mortality o f full-term bom babies. Table-4.16 shows that the risk of
post neonatal mortality for pre-term bom babies are 4.23 times higher compared to the
hazards o f post neonatal mortality for full-term bom babies (RR=4.23, p < 0.01) in the
bivariate model. The risk ratio slightly increases in the multivariate model (RR=4.52,
p < 0.01). The disadvantage of birth-size seen in neonatal mortality is not observed in
the post neonatal mortality.
The proportional hazard model of nutritional factors shows that premature bom
babies continue to die at higher mortality during the post neonatal period even after
controlling for other nutritional factors. Longer period of breastfeeding and early start
of supplementary food also have lower post neonatal mortality.
1 5 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Premature birth, with its consequent low birth weight, is one of the outcomes
of what has been described as the “maternal depletion syndrome”. It is known to be
associated with maternal age (<20 and >35), short birth interval (Carlaw et al., 1983),
high birth order (Hobcraft et al., 1985) and a higher risk o f death. The prevalence of
low birth weight reflects the health and social status o f women and the communities
into which children are bom (WHO, 1987). Worldwide, the birth weight of the infant
is one of the most important predictors of survival and is also a strong predictor of
growth (National Academy Press, 1992). Small infants with less supervision are more
likely to ingest pathogens if they live in an unhygienic environment, and are at greater
risk o f injury. Low birth weight poses a greater threat to a child bom in conditions
prevailing in slums or villages of developing countries than to a child bom in a society
with adequate medical services and a satisfactory physical and cultural environment
(Kimm, 1979).
4.2.2.4 Health Seeking Behavior:
The bivariate hazard analysis shows that prenatal care is negatively associated
with post neonatal mortality. As the number of visits of prenatal care increases, the
risk of post neonatal mortality decreases. However, the difference is not significant.
The bivariate hazard analysis also shows that babies bom at medical institutions have
better chances o f survival in the post neonatal period. An advantage is observed for
babies delivered by the medical personnel compared to those delivered by untrained
personnel but is not statistically significant.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Babies who received BCG vaccine at birth continue having better chances of
survival during the post neonatal period. The hazard analysis shows babies who
received BCG have highly significant lower post neonatal mortality. The children
who received BCG vaccine have a 67 percent lower risk o f post neonatal mortality
than the children who did not receive BCG vaccine. Women who used contraceptives
have a non-significant lower post neonatal mortality than those who never used
contraceptives in the bivariate model but they have higher post neonatal mortality in
the multivariate model (Table4.l7).
Table-4.17: Hazard Rate Ratios Obtained from Proportional Hazard Models of Health Seeking
Behavior for Predicting the Post neonatal Mortality, PDHS, 1990-91
Independent
Variables
Model-1 Model-2 Model-3 Model-4 Model-S Model-6
Prenatal Care
N o
Y e s
R e f e r e n c e
0 . 9 3 6
R e f e r e n c e
0 . 9 6 9 0
Delivery Attendant
U n - t r a i n e d
M e d i c a l T r a i n e d
R e f e r e n c e
0 . 6 7 9 0
R e f e r e n c e
0 . 7 7 0 0
Place o f Delivery
O w n h o m e
G o v e r n m e n t H o s p i t a l
P r i v a t e H o s p i t a l / C l i n i c
R e f e r e n c e
0 . 7 1 4 0
0 . 6 8 9 0
R e f e r e n c e
1 . 2 2 8 0
1 . 3 1 5 0
BCG
N o
Y e s
R e f e r e n c e
0 . 3 2 0 0
**•
R e f e r e n c e
0 . 3 2 8 0
Contraceptive Use
N o
Y e s
R e f e r e n c e
0 . 7 7 6
R e f e r e n c e * * *
1 . 0 6 8 0
-2 Log Likelihood
Degrees o f Freedom
2 . 2 5
I
2 . 7 5 4
I
1 . 0 5 6
2
2 8 . 3 2 2
1
0 . 6 4 9
1
2 9 . 7 2 9
6
*** p < 0.001, ** p < 0 .0 1. * p < 0.05, + p < 0.10
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.2.6 Socioeconomic Factors and Proximate Determinants:
Table-4.18 shows the results of proportional hazard analysis in model-1
through model-5. Model-1, which includes all the socioeconomic variables shows that
fathers’ education has a negative significant effect on post neonatal mortality
(RR=0.9l, p < 0.01). The children of parents living in the province of Sindh
(RR=2.4l, p < 05) also shows a significant higher post neonatal mortality compared to
the children living in the provinces of NWFP and Balochistan.
When environmental variables are added in model-2, along with the
socioeconomic variables, it is observed that children living in households connected
with “piped water in to the house” (RR=0.34, p < 0.05) and households depending on
surface water have reduced risks of post neonatal mortality compared to the children
living in households with a well source of drinking water.
In model-3 when the batch o f demographic variables are added to
socioeconomic and environmental variables, older maternal age, shorter previous birth
interval, lower birth order, and the previous sibling death in a family all have a
positive significant effect on post neonatal mortality. However, the fathers’ education
and piped water inside the households still maintain their significance even after
controlling for all other socioeconomic, environmental, and demographic variables in
the model.
In the next model when the nutritional factors, including baby-size and
premature births are added, premature births have higher risks o f post neonatal
mortality even after controlling for other variables. Breastfeeding (RR=0.35, p< 0.01)
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and age o f non-supplementary food (RR=4.07, p< 0.001) are highly significant in
predicting the post neonatal mortality.
Including health seeking behavior variables in model-5, it is observed that only
babies who had BCG vaccine at birth are at a significantly lower risk o f post neonatal
mortality than those who did not receive BCG immunization at birth. The utilization
of other preventive health services, such as prenatal care or delivery attended by
medical personnel does not show any significant effect on post neonatal mortality.
In the analysis where all the variables are included in the multivariate model, it
is observed that among the socioeconomic variables, the fathers’ education (RR=0.91,
p < 0.01) has significantly reduces post neonatal mortality. Residence in the province
of Sindh is also associated with higher post neonatal mortality. Among the
environmental variables, piped water connected in to the house maintained its
importance to significantly improve the survival status of children during the post
neonatal period. Maternal age over 30 years (RR=2.55, p < 0.001), birth intervals of
less than 18 months (RR=2.38, p < 0.05), previous birth interval 18-35 months
(RR=2.40, p < 0.01), and siblings death (RR=4.49, p < 0.001), are associated with
higher post neonatal mortality.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.18: Hazard Rate Ratios Obtained from Proportional Hazard Model Analysis for Predicting
Post neonatal Mortality, PDHS, 1990*91
Independent Variables Model-1 Model-2 Model-3 M odel4 Model-5
Years o f Maternal Education 1.012 1.017 1.033 1.029 1.069
Years o f Paternal Education 0.912 • • 0.909 • • 0.917 • • 0.923 • 0.912 • •
Household Possessions Index: Lower (Ref)
Higher 0.931 0.905 0.731 0.691 0.848
Medium 1.031 0.990 0.870 0.841 1.026
Place o f Residence: Urban 0.816 0.812 0.825 0.813 0.895
Province o f Residence: NWFP&Balochistan (Ref)
Punjab 1.702 1.684 1.354 1.247 1.361
Sindh 2.414 * 2.706 • • 2.850 " 2.729 • 2.770 •
Source o f Drinking Water. Well (Ref)
Pipe in to the house 0.339 • 0.301 •* 0.297 • • 0.296 •
Pipe on to the Property 0.737 0.667 0.660 0.605
Public Tap 0.770 0.685 0.665 -r 0.666
Surface 0.369 + 0.324 • 0.359 0.310 +
Latrine Facility: No Facility (Ref)
Bucket or Pit 1.371 1.409 1.502 1.5 6 1
Flush 1.436 1.609 1.819 1.869
Construction Baked Bricks 1.148 1.120 1.141 1.063
M other Age: 20-29 (Ref)
15-19 0.170 0.173 0.128 -r
3 0 4 9 2.575 2.721 2.545
Previous Birth Interval: 36 - 47 (Ref)
Less than 18 months 2.062 + 2.082 + 2.378 *
18-35 2.358 • • 2.260 • • 2.386 • •
48 and more 0.920 0.937 0.997
Birth O rd er 3 4 (ref)
2“’ 1.596 1.705 1.875
6th and above 0.519 * 0.519 • 0.527 •
Sex o f the Child: Boy 0.868 0.809 0.822
Number o f Children under 5 1.413 + 1.428 + 1.462 •
Previous Sibling Death 4.448 4.284 • • • 4.492 • • •
Premature Birth 3.667 • 4.940 • •
Baby-size at Birth: Average (Ref)
Small-size 0.801 0.676
Big-size 0.528 0.406
Breastfeeding 0.349 • • 0.346 • •
Age at Supplementary food started 4.066 • • • 4.143
Feeding with Nipple 0.831 0.860
Prenatal care visits 0.956
Place o f Delivery: At Home (Ref)
At Government Hospital 1.140
At Private Hospital 0.953
Delivery Attended by Medical Personal 0.805
BCG Vaccine received 0.304 • • •
Contraceptive used 1.874
Log likelihood 19.36 31.14 104.95 140.51 172.09
D e g re e s o f Freedom 7 14 2 4 30 36
• • • p < 0.001, •* p < 0.01. • p < 0.05, + p < 0.10
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-4.19: Odds Ratios Obtained from Proportional Hazard Model Analysis Tor Predicting
Post Neonatal Mortality, PDHS, 1990-91
Independent Variables Model-6 Model-7
Years o f Maternal Education 0.944 0.955
Years o f Paternal Education 0.911 • • 0.915 *
Household Possessions Index: Lower (Ref)
Higher 0.026 • • • 0.032 • • •
Medium 0.974 1.040
Place o f Residence: Urban 0.713 1.009
Province o f Residence: NW FP&Balochistan (Ref)
Punjab 1.525
Sindh 2.695 • 2.074 •
Source o f Drinking W ater W ell (Ref)
Pipe in to the house 0.314 •• 0.357 •
Pipe on to the Property 0.877 0.983
Public Tap 0.746 0.709
Surface 0.336 • 0.338 •
Latrine Facility: No Facility (Ref)
Bucket or Pit 1.723
Flush 2.203
Mother Age: 20-29 (Ref)
15-19 0.129 0.126 i-
30-49 2.787 • • • 2.811
Previous Birth Interval: 36 - 47 (Ref)
Less than 18 months 6.413 • • • 6.364 • • •
18-35 5.357 • • • 5.393 *•*
48 and more 0.864 0.850
Birth O rder 3-4 (ref)
2«i
1.939 2.089
6th and above 0.486 • • 0.482 • •
Sex o f the Child: Boy 0.867
Number o f Children under 5 1.248 1.239
Previous Sibling Death 10.863 •** 11.047 • • •
Premature Birth 5.457 • • 5.262 • •
Breastfeeding 0.349 • • 0.367 • •
Age at Supplementary food started 4.207 • • • 4.033 • • •
Feeding with Nipple 0.957 1.033
Prenatal care visits 0.951 0.951
At Government Hospital 1.244
At Private Hospital 1.038
Delivery Attended by Medical Personal 0.762
BCG Vaccine received 0.337 • • • 0.341 • • •
Contraceptive used 1.203
Interactions
Prenatal Care & Spacing Less than 18 Months 0.208 • 0.206 •
Prenatal Care & Spacing 19-35 Months 0.251 • 0.242 • •
Higher Index & Under 5 Children 4.728 • • • 4.552 • • •
Mother Education & Urban Residence 1.298 + ■ 1.289 ^
Mother Education & Feeding with Nipple 0.873 - 0.873 -
Log likelihood 197.047 190.832
D eg rees o f F reedom 38 30
p < O.OOt, • • p < 0.01, • p < 0.05. + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As mentioned earlier, breastfeeding proved its importance during the first two
years o f a child’s life. Hazard models show the significant effect (RR=0.346, p < 0.01)
o f continued breastfeeding on reducing post neonatal mortality. Breastfeeding reduces
post neonatal mortality by more than 62 percent, keeping constant the effect of all
other variables. The age of start giving supplementary food is also very important in
predicting the post neonatal mortality. The proportional hazard model shows that as
the age of non-supplementary food is delayed, the risk of post neonatal mortality
increases by 4 fold if the supplementary food started later in childhood (RR=4.14, p <
0.001 ).
When the interactions between prenatal care and previous birth intervals are
included in the next model, several changes in the significance of other variables are
also noted. The advantage is that attending prenatal care for mothers with shorter birth
intervals still improves the survival status during the post neonatal period. The
interaction model shows that if mothers with shorter previous birth intervals have used
prenatal care, their babies are significantly less likely to die during the post neonatal
period than those mothers with the same short birth interval who did not receive
prenatal care for the index child. The same significant impact of prenatal care for short
birth interval babies was also observed for neonatal mortality.
The interaction between bottle feeding with nipple and maternal education is
added to test the hypothesis that educated mothers are more likely to clean the bottle
and nipple before giving it to the child than their uneducated counterparts. The result
o f this interaction provided the evidence that nipple improves the survival status of the
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children o f educated mothers. In this analysis, the nipple feeding shows the non
significant negative effect on post neonatal mortality in the additive model but in the
interaction model it shows a negative significant (RR=0.87, p < 0.10) effect on post
neonatal mortality for educated mothers.
Earlier research indicates that bottle-feeding with nipples is a possible source
of contamination of milk and provides an indication of inappropriate feeding practice
(Phillips et al., 1969). Studies in developing countries found bottles and nipples were
contaminated by different pathogens (Elegbe et al., 1982; Black et al., 1989).
The interaction model also shows that infants of higher socioeconomic status
enjoy higher survival chances during the post neonatal period. However, if the
numbers of children under 5 years are more than 1 the chances of post neonatal
mortality significantly increases for these children of higher index households
(RR=4.55,p< 0.001).
4.3 Conclusion:
On the basis of the results of, both, bivariate and multivariate analyses as
presented in this chapter, the following conclusions can be drawn:
Contrary to the common perception that maternal education is a more
important determinant of infant mortality than paternal education, the results of PDHS
suggest that fathers’ education is a better predictor of post neonatal mortality than
maternal education. Maternal education is associated with reduced post neonatal
mortality in rural areas. Bicego and Boerma (1993) also found urban-rural differentials
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in child health and argued that advantage of maternal education is more pronounced in
rural areas than in urban areas.
Children o f mothers with a higher index o f household possessions are also at
significantly lower risk o f post neonatal mortality. However, the proportional hazard
analysis shows that among the socioeconomic factors only maternal education at
bivariate level and region of residence affect neonatal mortality in Pakistan.
The results of proportional hazard model analysis also show that families
living in households connected with piped water in their houses have significantly
lower post neonatal mortality than those families who depend on wells for drinking
water. The results are indicative of the importance of safe drinking water for
improving post neonatal mortality in Pakistan. The results do not find evidence of
improved child survival in households who have flush toilet facilities than those who
do not.
Among the demographic variables, it is observed that children of older women
(30-49) are exposed to significantly higher neonatal and post neonatal mortality. On
the other hand, babies o f the youngest mothers are at the lower risk o f both neonatal
and post neonatal mortality. The explanation for this is that, in this analysis, the index
children are not the first order babies of these teenage mothers. In the earlier research
it was observed that children of younger mothers are at higher risk of infant and child
mortality because in most of the cases they were the first order babies. Children bom
with shorter previous birth intervals are also at significantly higher risk of neonatal
and post neonatal mortality. A short birth interval does not give the mother sufficient
1 6 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
time to recuperate from the birth and to replenish her stores o f nutrients used during
pregnancy, especially in conditions of malnutrition.
The analysis highlighted the advantage of prenatal care to prevent these
neonatal and post neonatal deaths if these mothers avail the prenatal care services
when they got pregnant after shorter intervals. The hazard analysis also shows that
risk o f both neonatal and post neonatal mortality for those babies who has already lost
own sibling are significantly higher than the babies who did not experience a previous
sibling death. This may be that the sibling deaths tend to be correlated due to the same
risks associated with the home environment and with their mother’s health and
reproductive behavior.
Among the nutritional factors, longer breastfeeding, and early start of
supplementary foods are associated with higher survival during the post neonatal
period. Premature bom babies are at higher risk of death both during the neonatal and
post neonatal periods. The analysis also shows that large-sized babies are at
significantly higher risk of neonatal deaths but if they survived the neonatal period
then these big-sized bom babies have better chances of survival than average-sized
bom babies.
Immunization of BCG at birth is an important predictor o f lower neonatal and
post neonatal mortality in Pakistan. It is also observed that children living in the
province of Punjab have higher neonatal mortality than children living in the other
provinces whereas, children living in the province of Sindh have significantly higher
post neonatal mortality than the children living in any other province. Earlier estimates
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o f infant mortality differentials by province based on different surveys show higher
infant mortality in Punjab compared to other provinces (United Nations, 1986).
Hence, on the basis of these results, it can be suggested that the rise in the
parental education, improvements in the quality of water supply and motivation of
mothers to utilize the health services for prenatal and post natal care including
immunization are the most important steps to be taken to reduce the differentials in
neonatal and post neonatal mortality in Pakistan.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5
NUTRITIONAL STATUS OF LIVING CHILDREN
The nutritional status of living children as observed by the anthropometric
indices is presented in this chapter. Both bivariate and multivariate, analyses are
applied to the data. The bivariate analysis is based on the cross-tabulation of the
proportion of children aged 0-59 months identified as malnourished according to
height-for-age, subdivided as either stunted or severely stunted. For the multivariate
analysis, the ordered logistic regression analysis is employed and the results are
presented in the form of odds ratios. The predicted probabilities are also presented in
graphics to illustrate the effect of independent variables by age of the child. The last
section in this chapter also includes the analysis of the multinomial logit model where
dead and stunting is taken together in one dependent variable.
5.1 BIVARIATE ANALYSIS
For the bivariate analysis, cross tabulation o f the proportion of children
who have height-for-age below -3 standard deviations (severely stunted) and
below -2 standard deviations (stunted) are presented. The Chi-square test is
also employed to observe the statistical significance o f the relationship
between the dependent and the independent variables.
The results of the cross tabulation are explained for each of the
independent variables included in the analysis.
1 7 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.1 Socioeconomic Variables:
5.1.1.1 Parental Education:
Table 5.1 shows the proportion o f severely stunted and stunted children
by the mothers’ educational level. More than half of the children o f uneducated
mothers experience stunting (moderate malnutrition) compared to one-third of
children of mothers who have primary education and less than one-fifth of
children o f mothers who have greater than primary education. The association
between the mothers’ education and moderate malnutrition is highly significant
(P < 0.001). The relationship between mothers’ education and severe stunting
(severely malnutrition) is even more pronounced. The proportion of severely
stunted children bom to mothers with no education is significantly higher than
the proportion of severely stunted children bom to mothers with primary or
greater than primary education. The results shows that about 3 1 percent o f the
children of non-educated mothers are identified as severely stunted compared
to 18.7 percent and 5.6 percent of children of mothers with primary and greater
than primary education, respectively.
Table 5.1 also shows the association of nutritional status o f children by
fathers’ education. Like the mothers’ education, the fathers’ education also
shows a highly statistically significant association (p < 0.001) with both
stunting and severe stunting. More than one-third of children of non-educated
fathers compared to one-fifth of secondary educated and only one-tenth of
higher than secondary education o f fathers are identified as severely stunted.
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.1: Bivariate Relationship Between Socioeconomic Factors and height-for-age
among Children Aged 0-59 Months, PDHS, 1990-91
Socioeconomic Variables % Severely % Stunted Number
Stunted_______________________(N)
Maternal Education 2221
No education 30.95 52.10 1714
Primary 18.73 32.29 326
Greater than Primary 5.62 18.61 181
Chi-Square 66.733*** 104.744***
Fathers Education
No education 34.23 56.92 985
Primary 27.76 45.63 396
Secondary 19.79 36.89 724
Higher 10.36 19.28 110
Chi-Square 60.513*** 102.730***
Index of Household Possessions
Lower 35.87 56.45 827
Medium 25.27 45.67 981
Higher 13.83 28.35 414
Chi-Square 70.775*** 87.930***
Place of Residence
Rural 31.77 52.08 1487
Urban 17.60 35.08 994
Chi-Square 50.046*** 57.072***
Region of Residence
Punjab 23.72 40.72 1335
Sindh 31.96 53.48 506
NWFP 30.32 56.04 326
Balochistan 44.98 64.52 55
Chi-Square 24.311*** 46.906***
p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.1.2 Index of Household Possessions:
As expected, children o f higher index households experience lower
proportions o f malnutrition than the children of medium and lower indices.
Table 5.1 shows that 56.45 percent of the children of lower index households
compared to 28.35 percent of the children of higher index households
experience moderate malnutrition. Moreover, 35.87 percent of the children of
lower index households, compared to 13.83 percent of the children of higher
index households, are in the state of severe malnutrition. The association
between index o f possessions and stunting and index o f possessions with
severely stunting are both highly statistically significant (p < 0.001).
5.1.1.3 Place of Residence:
Table 5.1 shows that the prevalence of stunting and severely stunting in
rural areas is significantly higher than urban areas (p < 0.001). It is observed
that more than half (52.1 percent) of the children living in rural areas are
experiencing stunting compared to more than two-third (35.1 percent) of the
urban children. Almost two-third (31.8 percent) of the rural children compared
to less than one-fifth (17.6 percent) of the urban children are experiencing
severely stunting malnutrition.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.1.4 Province of Residence:
The differentials in nutritional status o f children is also observed by the
region of residence. Table-5.1 shows that the prevalence o f stunting is the
highest in the province o f Balochistan where almost two-third (64.5 percent) of
the children are experiencing stunted followed by 56.0 percent in NWFP, 53.5
percent in Sindh and 40.7 percent in the province of Punjab. The provincial
differentials are statistically significant (p < 0.001). The proportion of severely
stunted children were also the highest in Balochistan (45 percent) followed by
Sindh (32 percent), NWFP (30 percent) and Punjab (24 percent).
5.1.2 Domestic Environmental and Hygiene:
5.1.2.1 Source of Drinking Water:
Having access to piped water, including distinctions between in-house
taps and public standpipes, was found to be significantly associated with
children’s heights in several populations (Bateman et al., 1993). Table-5.2
shows the relationship between the source of drinking water and nutritional
status of children under 5 years o f age. It is observed that the lowest
proportion of stunted children and severely stunted children is from the
households who have connected piped water into their houses and the
proportion is highest for the children from the households who depend on the
surface water. The importance o f clean drinking water in the dwelling is
greatest to child survival when breastfeeding is stopped and the child is
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
weaned. Access to piped water in the household is likely to be of direct benefit
in lowering child mortality by reducing exposure to water related diseases, like
diarrhea, cholera and typhoid (Merrick, 1985). Children were more likely to
be stunted in households without piped water (Victora et al., 1986).
It is observed that 65.68 percent o f the children are stunted and 40.71
percent of these children are severely stunted who live in households with
surface source o f drinking water. As tap water is more likely to be infected,
the results show that households depending on public tap water have the
second highest prevalence of stunting and severe stunting.
Table-5.2: Bivariate Relationship Between Domestic Hygiene Factors and height-for-age
among Children Aged 0-59 Months, PDHS, 1990-91
Domestic Hygiene Factors % Severely % Stunted Number
Stunted
(N)
Source of Drinking Water
????
Piped into residence 19.02 36.57 444
Piped onto property 20.27 38.22 268
Public tap 30.12 54.52 179
Surface 40.71 65.68 187
Well 29.11 47.82 1144
Chi-Square 41.685*** 11.081*
Toilet Facility
Bush 33.43 53.03 1140
Bucket 30.95 54.86 392
Flush 14.42 30.84 690
Chi-Square 82.334*** 98.555***
Housing Construction Material
Unbaked bricks 33.05 53.62 1281
Baked bricks 18.97 36.71 941
Chi-Square 54.435*** 62.32
• • • p < 0.001, • • p < 0.01. * p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.2.2 ToUet Facilities:
Because so many o f the major infectious agents of disease are spread
through faeces and urine, hygienic disposal of waste is vitally important.
Table-5.2 shows that the lowest proportion of stunted children are in the
households that have a flush system in the house. The highest proportion of
stunted children is found in houses that have no toilet facility. The association
between stunted children and availability of toilet facilities is highly
statistically significant (p < 0.001). The relationship between severely stunted
children and type of toilet facilities available shows the same pattern and is
statistically significant (P < 0.001).
5.1.2.3 Housing Construction Material:
The material used in the construction of the houses also shows a
statistically significant association with the prevalence o f malnutrition. Table-
5.2 shows that 53.62 percent o f the children who live in unbaked constructed
houses are stunted compared to 36.71 percent o f the children living in the
houses constructed of baked bricks. The reason for such a difference could be
because children in houses constructed from unbaked bricks are more exposed
to the risk o f infectious diseases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.3 Demographic and Maternal Factors:
5.1.3.1 Age of the Child:
In most developing countries, children decelerate in growth between 4
to 6 months after birth and continue decreasing in position relative to the
reference population through at least 18 or 24 months o f age (Martorell and
Habicht, 1986; Waterlow, 1988). Table-5.3 shows that child’s age is
significantly associated with height-for-age. It is observed that about 27
percent of the children of less than 1 year are experiencing stunting and the
prevalence of stunting sharply increases with increasing age of the child up to
64.42 percent among children aged 36-47 months and then slightly decreases
to 63.51 percent among 48-59 months children.
This age pattern of stunting is because the non-breast milk foods are
introduced to the child’s diet, introducing a wide variety of bacteria and viruses
into the child’s intestinal tract (Martorell and Habicht, 1986). This slows down
child growth for one to two years until the child builds up resistance to these
types of infections. At this age the child is also losing the immunity, which was
passed to him through the placenta from his mother and will begin to be
susceptible to various infections (Jelliffe, 1968). Moreover, the second year of
child’s life is the most dangerous phase o f growth. Even if the breastfeeding is
continued during this period, the amount of protein supplied in this way is
small. In some children, the weight may actually decrease (Jelliffe, 1968).
177
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
After the second year o f the child’s life, the growth slows down and
genetic and environmental factors become more deterministic (Karlberg et al.,
1994). In developing countries, the prevalence of stunting starts to rise after
the third month o f age (Nabarro et al., 1988). The process o f stunting usually
slows down at age 3, after which mean heights run parallel to the reference
population (WHO, 1995). Children over 3 years acquire a certain degree of
resistance to various infections and are able to obtain and digest a wider range
of the family diet. Under these conditions the child may remain below the
standard weight and height for years but does start growing slowly (Jelliffe,
1968).
5.1.3.2 Maternal Age:
The age of the mother at childbirth has also been shown to influence
child health where risks of low birth-weight or congenital abnormalities are
highest for the youngest and oldest mothers. Children bom to relatively young
mothers are more likely to be malnourished and to suffer from other health
problems than are children bom to older mothers. Table-5.3 shows that 15-19
year old mothers have the highest proportion (57.3 percent) of stunted children
and the children o f mothers older than 29 years are also experiencing a higher
proportion o f stunting (49.3 percent). Children o f mothers aged 20-29 years
have the lowest proportion o f stunting (43.9 percent). The same significant (p
< 0.001) pattern is observed for severely stunted children.
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.3 J Preceding Birth Interval:
Table-5.3 also presents the association between the preceding birth
interval and nutritional status of children under 5 years o f age. It shows that the
birth interval is significantly associated with stunting (p < 0.05) and highly
significantly associated with severely stunted children (p < 0.001). As the
preceding birth interval increases the proportion of stunting and severe stunting
decreases. More than half (51.34 percent) of the children bom within 18
months of the preceding birth interval are suffering from stunting, and more
than one-third (34.8 percent) o f these children are severely stunted.
Close spacing of births may have a negative effect on prenatal growth
or aggravate competition for resources and maternal time. It has been argued
that the former effect is dominant and is largely restricted to children under
two years of age (Boerma and Bicego, 1992). Shorter birth intervals are
significantly related to child height in studies in Ethiopia (Groenewold and
Tilahun, 1990) and Brazil (Huttly et al., 1992). Analyses o f DHS African
countries have reported significantly lower height-for-age in children bom
within two to three years of the preceding birth in Burundi, Uganda, Senegal,
Mali, Ghana, and Togo (Sommerfelt, 1991 Boerma and Bicego, 1992).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5J:Bivariate Relationship Between Demographic Factors and height-for-age
among Children Aged 0-59 Months, PDHS, 1990-91
Demographic Factors % Severely
Stunted
% Stunted Number
(N)
Age of the Child 2221
l-l 1 Months 8.69 27.03 748
12-23 Months 31.35 54.95 630
24-35 Months 38.39 61.53 426
36-48 Months 42.81 64.42 265
49-59 Months 40.81 63.51 152
Chi-Square 209.243*** 116.695***
Maternal Age at the time of the Birth
15-19 48.55 57.25 74
20-29 25.49 43.87 1276
30-49 37.61 49.33 871
Chi-Square 19.072*** 9.804***
Preceding Birth Interval
Less than 18 months 34.80 51.34 238
18-35 28.64 47.04 819
36-47 27.79 48.23 640
48 and over 20.31 41.18 525
Chi-Square 20.521*** 9.083*
Birth Order
2nd 22.30 42.63 412
3rd-5th 25.20 42.23 1044
6th or over 32.24 54.28 765
Chi-Square 16.968*** 28.750***
Sex of the Child
Girl 25.66 45.92 1086
Boy 28.45 46.97 1135
Chi-Square 2.196 0.244
Previous Death (Sibling)
No
Yes
Chi-Square
24.91
35.04
19.480***
44.09
55.09
18.219***
1744
477
• • • p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.3.4 Birth Order:
Table-5.3 also presents the relationship between birth order and
malnutrition o f children under 5 years of age. A birth order effect could result
from maternal depletion, discrimination against later-born children, or resource
dilution (Horton, 1988). The table shows that the higher the birth orders, the
higher the prevalence o f malnutrition among the children. The table shows that
42.63 percent of children of the 2n d birth order children experience stunting
compared to 54.28 percent of the children of 6,h or higher birth order. This
relationship between birth order and stunting is highly statistically significant
(p < 0.001). The same pattern is observed for severely stunted children (P <
0.001).
5.1.3.5 Sex of the Child:
Contrary to a common perception, the analysis shows that 45.92
percent of the girls compared to 46.97 percent o f boys are stunted. The
proportion of boys who fell into the severely stunted category is also larger
than that of girls. However, the chi-square test shows that this difference is not
statistically significant.
5.1.3.6 Previous Sibling Death
Table-5.3 shows a statistically significant (p < 0.001) difference
between the proportions of stunted children whose family experienced a
181
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sibling death compared to the children who did not. It is observed that SS.10
percent of the children are stunted who experienced a sibling death compared
to 44.1 percent of the children who did not lose a sibling. The same highly
statistically significant (p < 0.001) pattern is observed for severely stunted
children.
5.1.4 Nutritional Factors:
5.1.4.1 Premature Births:
Table-5.4 shows an association between premature births and
malnutrition. Among the children who were bom prematurely, 63.59 percent
are identified as experiencing stunting compared to 46.27 percent of full-term
birth children. This association is marginally statistically significant (p < 0.10).
The same pattern is observed for severe stunting and premature births but the
relationship is more pronounced and is statistically significant (p < 0.05).
Children bom pre-term catch up in growth after infancy (Adair, 1989) unless
they experienced fetal growth retardation as well (Mata et al., 1975).
Symmetric fetal growth retardation, in which length and height are
proportionately reduced, results in permanent growth retardation, while
newborn infants low in weight but with normal lengths experience rapid catch
up growth (Adair, 1989).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.4.2 Birth-Weighty Birth-Size:
The most powerful predictor o f child height is usually size at birth
(Fields and Smith, 1994; Mata, 1978). Most risk factors for low birth-weight
do not appear to be associated with growth retardation later in childhood
(Sampson et al., 1994). Table-5.4 also presents the association between birth
size and malnutrition. It shows that normal size babies are less likely to be
stunted or severely stunted compared to very-small size or very-large size
babies. It is found that 64.12 percent o f the very small size bom babies are
experiencing stunting compared to 49.40 percent and 45.26 percent bom as
very-large and normal size bom babies, respectively. The association between
birth size and stunting is highly statistically significant (p < 0.001). The same
highly statistically significant (p < 0.001) pattern is observed for birth-size and
severe stunting. The results are inconsistent with the earlier studies that show
children who are bigger at birth are likely to remain bigger in preschool age
years (Frisancho et al., 1994).
5.1.4 J Bottle Feeding with Nipple:
Bottle-feeding with nipples is a possible source of contamination of
milk and provides an indication o f inappropriate feeding practice (Phillips et
al., 1969). Studies in developing countries found that bottles and nipples were
contaminated by different pathogens (Elegbe et al., 1982; Black et al., 1989).
Table-5.4 also presents the proportion o f children experiencing malnutrition
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and whether they were fed by nipple. The proportion experiencing stunting is
significantly higher for those who are fed by nipple compared to those who
were never fed by nipple. This may be because improved cleaning of bottles
and nipples are used, which reduces the risk of contamination of milk or
formula by boiling of bottles and nipples.
Table-5.4:Bivariate Relationship Between Nutritional Factors and height-for-age
among Children Aged 0-59 Months, PDHS, 1990-91
Nutritional Factors Severely Stunted Number
Stunted of Cases
Premature Birth 2222
Full-term 26.88 46.27 2198
Pre-term 46.36 63.59 24
Chi-Square 4.518* 2.834+
Birth-weight/Birth-size
Very-small size 47.20 64.12 135
Normal size 25.68 45.26 2058
Very-large size 33.44 49.40 28
Chi-Square 30.376*** 18.280***
Nipple Feeding
No 30.97 52.41 1464
Yes 19.59 34.96 758
Chi-Square 32.728*** 61.165***
Diarrhea During Two Weeks Before The Survey
No 26.09 45.78 1795
Yes 31.28 49.31 427
Chi-Square 4.703* 1.731
• • • p < 0.001, • • p < 0.01. • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.4.4 Diarrhea in the Two Weeks Before the Survey:
Malnutrition is usually the result o f a combination of inadequate dietary
intake and infection (UNICEF, 1998). The adverse effects of diarrhea on the
growth of children have been well documented (Mata et al., 1977; Martorell et
al., 1970; Rowland et al., 1976). Diarrheal disease is the leading cause of infant
and childhood mortality in Pakistan, where it accounts for an estimated
300,000 deaths per year (Lambert, 1986).
The proportion of stunted children who had an episode of diarrhea
during the two weeks before the survey is 49.31 percent, compared to 45.78
percent for those children who did not have an episode of diarrhea. This
difference is not statistically significant. The same pattern is observed for
severely stunted children, but the difference is more pronounced (p < 0.05).
5.1.5 Health Seeking Behavior:
5.1.5.1 Prenatal Care Received:
Prenatal care has long been endorsed as a means to identify mothers at risk of
delivering a pre-term or growth-retarded infant and to provide an array of
available medical, nutritional, and educational interventions intended to reduce
the determinants and incidence o f low birth weight and other adverse
pregnancy conditions and outcomes. Early prenatal care provides an
opportunity to offer preventive care that will benefit the infant as well as the
mother such as counseling on hygiene, breastfeeding and nutrition. The
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
advantage o f prenatal care is clearly demonstrated in table-5.5, it shows a
significant difference (p < 0.001) in proportion o f stunting by prenatal care.
Table-5.5 shows a prevalence of 48.64 percent of stunting among children
whose mothers did not receive prenatal care compared to 32.84 percent among
children whose mothers did take prenatal care. The association between
prenatal care and severe stunting is even more pronounced by the status of
prenatal care
5.1.5.2 Delivery attendant:
The bivariate analysis shows a highly statistically significant
association between the delivery attendant and malnutrition. Among the
deliveries attended by untrained personnel, a higher proportion experiences
stunting, as well as severe stunting state o f malnutrition. The analysis found
that 51.63 percent compared to 37.31 percent o f children experiencing stunting
were delivered by untrained and trained personnel, respectively. The
proportion is also significantly higher for severely stunted children for children
delivered by untrained attendants compared to the children delivered by trained
personnel. The association is statistically significant (p < 0.001).
5.1.5.3 Place of Delivery:
Table 5.5 also shows the proportion of stunted and severely stunted
children by the place of delivery. The lowest proportion of stunting and severe
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stunted children is observed for those mothers who gave birth at private
hospitals. As mentioned earlier these women who choose to deliver their
babies in private hospitals may be of higher economic status than women who
do not.
Table-5.5: Bivariate Relationship Between Health Seeking Behavior and height-for-age
among Children Aged 0-59 Months, PDHS, 1990-91
Health Seeking Behavior % Severely % Stunted Number
Stunted of Cases
Prenatal Care
No 29.10 48.64 1915
Yes 14.54 32.84 307
Chi-Square 28.407*** 26.567***
Delivery Attendant
Un-trained 30.97 51.63 1420
Medical Trained 20.22 37.31 802
Chi-Square 29.987*** 42.224***
Place of Delivery
Own home 29.48 49.50 1868
Other home 44.02 55.27 48
Government Hospital 11.18 31.39 125
Private Hospital/Clinic 8.80 23.14 181
Chi-Square 58.985*** 59.360***
BCG
No 30.22 48.51 752
Yes 25.49 45.41 1469
Chi-Square 5.634* 1.930
Contraceptive Use
Never Used 27.89 47.88 1897
Ever Used 22.39 38.12 324
Chi-Square 4.243* 10.618***
• • • p < 0.001. • • p < 0.01, • p < 0.05,1- p <0.10
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The highest proportion o f stunting and severely stunted children is
found in children whose mothers delivered babies at homes, 55.27 percent at
“other homes” and 49.5 percent for own home. The association is highly
statistically significant (p < 0.001).
5.1.5.4 BCG Vaccination:
The analysis shows no significant association between stunting and the
status of BCG vaccine. However, the association between severely stunting
condition and BCG vaccine is statistically significant (p < 0.05). It is observed
that 30.2 percent of the children who did not received BCG vaccine compared
to 25.5 percent of those who received BCG vaccine are experiencing severe
condition of malnutrition.
5.1.5.5 Contraceptive Use:
Table-5.5 shows a statistically significant association between
contraceptive use and nutritional status of children. For mothers who used
contraceptives, the proportion of stunting is significantly lower than those
children whose mothers did not use contraceptives. It may be that children
bom to mothers with longer birth intervals used contraceptives. It may also be
that educated mothers are more likely to use contraceptives in Pakistan and
these educated mothers are more likely to give better care to their children.
188
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To see the independent effect of each independent variable on the
nutritional status of children, multivariate analysis is presented in the next
section.
5.2 LOGISTIC REGRESSION ANALYSIS
In this section, results of logistic regression analysis are presented. The model
used in this analysis is the cumulative logit model as explained earlier. Although there
is a single set o f regression coefficients, there is a different intercept for each of the
equations. Therefore, the model predicts the probability o f being in a severely stunted
or stunted state rather than in a normal growth category.
5.2.1 Socioeconomic Factors:
As expected, maternal education is negatively associated with height-
for-age of the children. As the number of years o f maternal schooling
increases, the odds of stunting decrease. In the bivariate logistic regression,
each year increase in maternal schooling reduces the odds of stunting
(OR=0.86, p < 0.001) by 14 percent. When other socioeconomic variables are
included in the model, model-6 shows that the each year increase in maternal
education decreases the odds o f stunting by 9 percent (OR=0.91, p < 0.001).
Bicego and Boerma (1991) found more than a doubling o f risk of stunting
among children o f mothers with no education, compared with children of
mothers with at least secondary education in an analysis o f 17 DHS countries.
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.6 shows that stunting of children is strongly related to the
fathers’ education (p < 0.001). It shows that each year increase in the fathers’
education decreases the odds of being stunted by 9 percent (OR=0.91, p <
0.001) at bivariate level. The fathers’ education maintains its significance even
when the mothers’ education and other socioeconomic variables are included
in the model. Model-6 shows that each year increase in the fathers’ education
decreases the odds of stunting by 3 percent (OR=0.97, p < 0.01).
Table-5.6: Odds Ratios obtained from the Ordered Logistic Regression Models o f Predicting the Nutritional
Status of Children Aged 0-59 Months, PDHS, 1990-91
Independent
Variables
Model-1 Model-2 Model-3 Model-4 Modle-S Modle-6
Intercept
I n t e r c e p t 1
I n t e r c e p t
- 0 . 7 9 8 8
0 . 0 8 5 1
- 0 . 6 1 8 7
0 . 2 6 5 3
- 0 . 6 0 7 5 - 0 . 7 7 7
0 . 2 7 2 2 0 . 0 9 1 8
- 0 . 4 6 3 3
0 . 4 4 0 3
- 3 . 2 4 9 3
- 2 . 1 9 4 3
Maternal Education
Fathers Education
0 . 8 5 8 '
0 . 9 1 0 '
***
0 . 9 2 3 * * *
0 . 9 5 6 * * *
0 . 9 0 5 * * *
0 . 9 6 8 * *
Index of Household Possessions
L o w e r
H i g h e r
M e d i u m
R e f e r e n c e
0 . 2 9 6 * * *
0 . 6 3 4 * * *
R e f e r e n c e
0 . 5 9 5 * * *
0 . 7 8 4 * *
R e f e r e n c e
0 . 5 3 8 * * *
0 . 7 8 4 *
Place o f Residence
R u r a l
U r b a n
R e f e r e n c e
0 . 4 8 7 * * *
R e f e r e n c e
0 . 7 9 8 *
R e f e r e n c e
0 . 7 6 0 *
Age o f the Child
A g e
A g e - S q u a r e
A g e - C u b e
1 . 3 3 6 * * *
0 . 9 9 2 * * *
1 . 0 0 0 * * *
-2 Log Likelihood
Degrees o f Freedom
1 2 7 . 0 5
I
1 1 4 . 4 1 2
I
1 0 0 . 7 6 4 6 6 . 2 3 1
2 1
1 8 2 . 3 4 1
5
5 9 1 . 9
8
• • • p <0.001, * * p < 0 .0 1 . * p < 0 .0 5 , + p < 0 .l0
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As expected, table-5.6 also shows that the risk of stunting in children
bom to the mothers with the higher index o f household possessions are 70
percent lower (OR=0.30, p < 0.001) than to the risk o f stunting compared to
the children bom to mothers with lower index of household possessions in the
bivariate model and 46 percent (OR=0.54, p < 0.001) in the multivariate
model. The odds (OR=0.78, p < 0.05) o f children with medium index are also
78 percent less to the risk of stunting compared to the children of lower
possessions index. Children living in urban areas have the advantage of better
nutritional status compared to the children living in rural areas. The bivariate
logistic regression analysis shows that rural children are at least twice as likely
to be stunted than urban children (OR=0.48, p < 0.001). When all other
socioeconomic variables are controlled, the odds are 24 percent lower of urban
children than rural children to be stunted (OR=0.76, p < 0.05).
5.2.2 Domestic Environment and Hygiene:
The logistic regression shows that the source o f drinking water, toilet
facility and housing construction, each, has a significant effect on stunting both
in bivariate and multivariate analysis.
The bivariate logistic regression model shows that the children o f the
parents who have piped water in the residence and piped water on the property
are at lower risk of stunting (OR=0.61, p < 0.001) compared to the children of
parents with a well as a source of drinking water. The children of those
1 9 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
households depending on surface water have a significantly higher risk of
stunting (OR=1.87, p < 0.001). The results o f some studies have shown that the
availability of water supply inside the house have a considerable influence on
child survival (Victora et al., 1988; Shaffer et al., 1978).
In the multivariate model, when toilet facility and housing construction
are included, the odds of stunting among children in households with surface
water are 50 percent higher than the households with wells (OR=1.50, p <
0.01). Moreover, the odds o f stunting of children with public tap are 66 percent
higher (OR=1.66, p < .01) than the children depending on a well as the source
of drinking water. This highly significant effect o f tap water use on stunting
may be that the tap water is contaminated by sewage outflows or the tap water
is not chlorinated to any particular standard. These results are in accordance
with the findings of Sri Lanka where tap water is associated with lower child
survival (Patel, 1980). According to United Nations (1998), Pakistan needs to
address a vast array of problems regarding water usage, for example forty
percent o f urban deaths are caused by water-bome diseases.
Table-5.7 also depicts the effect of toilet facilities on stunting. It shows
that the odds o f stunting for children of households equipped with a flush toilet
facility are 60 percent lower than the households without any toilet facility
(OR=0.40, p < 0.001). In Bangladesh, it was found that poor environmental
sanitation was related to lower child survival while the effects o f type o f water
source were not statistically significant (Rahman et al., 1985).
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.7: Odds Ratios obtained from the Ordered Logistic Regression Models o f Environmental Factors
for Predicting the Nutritional Status of Children Aged 0-59 Months, PDHS, 1990-91
Independent Model-1 Model-2 Model-3 Model-4 Model-5
Variables
Intercept
I n t e r c e p t I
I n t e r c e p t
- 0 . 9 3 2 2
- 0 . 0 6 4 1
- 0 . 7 3 1 8
0 . 1 5 1 8
- 0 . 7 1 7 9
0 . 1 5 3 4
- 0 . 8 0 2
0 . 0 9 0 9
- 3 . 4 9 3 9
- 2 . 4 5 5 5
Source of Drinking Water
P i p e d i n t o r e s i d e n c e 0 . 6 1 0
P i p e d o n t o p r o p e r t y 0 . 6 5 6
P u b l i c t a p 1 . 2 0 1
S u r f a c e 1 . 8 7 4
W ell Reference
**•
• *
1 . 1 3 8
1 . 0 8 1
1 . 5 9 4 * * *
1 . 7 1 2 * * *
Reference
1 . 0 7 2
1 . 0 1 6
1 . 6 6 1 * *
1 . 5 0 2 « *
Reference
Toilet Facility
F l u s h
B u c k e t
B u s h
0 . 3 7 7 * * *
0 . 9 9 7
Reference
0 . 4 3 8 * * *
0 . 9 9 0
Reference
0 . 3 9 8 • * *
0 . 9 8 6
Reference
Housing Construction Material
U n b a k e d b r i c k s
B a k e d b r i c k s
Reference Reference Reference
0 . 4 9 3 * * • 0 . 7 8 2 * 0 . 7 5 8
Age of the Child
A g e
A g e - S q u a r e
A g e - C u b e
1 . 3 2 9 * • *
0 . 9 9 2 • * *
1.000 ***
- 2 L o g L i k e l i h o o d
D e g r e e s o f F r e e d o m
6 0 . 4 1 9
4
1 1 3 . 4 0 6
■ >
7 1 . 6 6 5
1
1 3 8 . 7 7
7
5 4 7 . 7 5 3
10
• • • p < 0.001. • • p < 0.01, * p < 0.05, + p < 0.10
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The housing construction material also shows a significant effect on
stunting (p < 0.001). In the bivariate model, the odds of stunting (OR=0.49, p <
0.001) for children living in baked brick constructed houses are 50 percent
lower than the children living in unbaked brick constructed houses. In the
multivariate model, the effect is reduced to 24 percent but it is still significant
(OR=0.76, p < 0.05).
5.2.3 Demographic Factors:
Maternal age at birth less than 20 years has a significant association
with stunting in the bivariate model (0R=2.17, p < 0.001) compared to
maternal age at birth 20-29 years. In the multivariate model, children of
younger mothers are significantly at higher risk of stunting (OR=1.93, p <
0.01) and children of older mothers are significantly less likely to be stunted
(OR=0.79, p < 0.05) compared to the children of mothers aged 20-29
(reference category).
Table-5.8 shows that a long preceding birth interval of at least four
years is associated with significantly better height-for-age (OR=0.72, p <
0.001) compared to the children bom within a 36-47 months birth interval. In
contrast to a long preceding birth interval, children bom within 18 months of
the preceding birth appear to be at no disadvantage in growth status compared
to the children bom within 36-47 months o f the preceding birth.
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.8: Odds Ratios obtained from the Ordered Logistic Regression Models o f Demographic Factors for
Predicting the Nutritional Status o f Children Aged 0-59 Months, PDHS, 1990-91
Independent Model-1 Model-2 Model-3 Model-4 Model-5 Model-6 Model-7
Variables
In tercept
Intercept 1
In tercept
-1.066
-0.214
-0.935
-0.082
-0.831
0.0255
-1.029
-0.18
-1.1
-0.24
-1.592
-0.716
-4.0112
-3.0229
Maternal Age at the time o f the Birth
15-19 2.170
20-29 Reference
30-49 1.115
2.451 • • •
Reference
0.831 -
1.930 ”
Reference
0.785 *
Preceding Birth Interval
Less than 18 months
18-35
36-47
48 and over
1.229
0.984
Reference
0.729 • • •
1.153
0.961
Reference
0.678 • • •
1.004
1.030
Reference
0.815 +
Number o f Siblings
No sibling
1-2
3-4
5 & more
Reference
0.706
0.804
1.151
Reference
1.408
1.763 •
2.657 • • •
Reference
1.339
1.508
2.186 **
Sex o f the Child
Girl
Boy
Reference
1.078
Reference
1.067
Reference
1.107
Previous Siblings Death 1.584 • • • 1.739 • • • 1.373 "
Age o f the Child
A ge
A ge-S q u are
A ge-C ube
1.304 ***
0.993 ” *
1.000 •**
-2 L og L ikelihood
D egrees o f F reedom
11.42
2
14.907
2
24.225
2
0.864
I
2 2.06
1
81.515
10
4 30.72
13
• • • p < 0.001, • • p < 0 .0 1, • p < 0.05, + p < 0.10
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.8 also shows that having more than 5 siblings has a significant
effect on stunting (OR=2.19, p < 0.01). It is common for children to compete
for parental resources, which include nurturing and food, among other more
durable family resources. This competition results in a negative effect on the
health of a child, who is neither old enough to compete with older siblings, nor
breast-fed (Mozumder, et al., 2000).
Contrary to earlier research in South Asian countries (Mozumder, et al.,
2000), the logistic regression analysis shows no evidence of sex differentials in
stunting either in bivariate or in multivariate analysis in Pakistan. However, if
the sibling has an earlier death in the family, the analysis shows that the index
child experiences significantly higher risks of stunting (OR=l.37, p < 0.01).
5.2.4 Nutritional Factors:
Table 5.9 shows that prematurely bom children continue to be
significantly stunted compared to full-term children in bivariate logistic
regression (OR=2.20, p < 0.05) while in multivariate logistic regression the
association becomes very weak (OR=2.0, p < 0.10). However, very small-sized
babies experience a significantly higher risk o f stunting compared to normal
sized babies in the bivariate (OR=2.39, p < 0.001) and multivariate logistic
regression analysis (OR=2.90, p < 0.001).
The analysis also shows that children o f mothers who started feeding
with nipples have a significantly lower risk of stunting (OR=0.49, p < 0.001).
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tables-5.9 also shows that children who experienced an episode of
diarrhea during the two weeks before the survey are at higher risk of stunting.
The odds o f stunting for children who had diarrhea are 36 percent higher than
those children who did not have a diarrhea episode during the two weeks
before the survey (OR=1.36, p < 0.01). Diarrhea significantly increased the
likelihood o f stunting in Filipino children (Adair, et al., 1997). These
investigators found the effect of morbidity was similar in younger and older
infants, despite the fact that diarrhea was more prevalent in the first years.
In another study in Pakistan, the prevalence o f diarrhea varied with the
age o f children; 6 to 24 months old children showed the highest prevalence.
Furthermore, children who were undernourished were more likely to have
diarrhea (Shah et al., 1997).
Diarrhea is an important cause o f malnutrition. This is because nutrient
requirements are increased during diarrhea, whereas nutrient intake and
absorption are usually decreased. Each episode of diarrhea can cause weight
loss and growth faltering. Moreover, if diarrhea occurs frequently, there may
be too little time to "catch up" on growth between episodes, therefore, they are
more likely to become undernourished than children who experience fewer or
shorter episodes o f diarrhea. To prevent growth faltering, good nutrition must
be maintained both during and after an episode o f diarrhea. This can be
achieved by continuing to give considerable amounts of nutritious foods
throughout the episode and during the period o f recovery.
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Among the illiterate mothers, the notion that feeding should be
reduced or stopped during diarrhea reflects a common belief that giving food
will cause stool output to increase and thus make the diarrhea worse, but this is
not usually the case.
Table-5.9: Odds Ratios obtained from the Ordered Logistic Regression Models o f Nutritional Factors
for Predicting the Nutritional Status o f Children Aged 0-59 Months, PDHS, 1990-91
Independent Model-1
Variables
Model-2 Model-3 Model-4 Model-5 Model-6
Intercept
I n t e r c e p t 1 - 0 . 9 9 9 6
l n t e r c e p t 2 - 0 . 1 4 9 9
- 1 . 0 5 2 7
- 0 . 1 9 5 1
- 0 . 7 8 3
0 . 0 8 4 5
- 1 . 0 2 6
- 0 . 1 7 7
- 0 . 8 8 3 3
- 0 . 0 0 4 4
- 3 . 6 2
- 2 . 6 0 2
Premature Birth
F u l l - t e r m R e f e r e n c e
P r e - t e r m 2 . 2 0 3 *
R e f e r e n c e
1 . 9 4 4 +
R e f e r e n c e
2 . 0 3 8 +
Birth-weight/Birth-size
V e r y - s m a l l s i z e
N o r m a l s i z e
V e r y - l a r g e s i z e
2 . 3 8 6 * * •
R e f e r e n c e
1 . 2 8 0
2 . 3 3 8 * * •
R e f e r e n c e
1 . 2 8 4
2 . 8 9 5 • * *
R e f e r e n c e
1 . 0 3 9
Bottle Feeding with Nipple
Y e s
N o
0 . 5 0 3 ” *
R e f e r e n c e
0 . 4 9 4
R e f e r e n c e
0 . 4 8 7 *”
R e f e r e n c e
Diarrhea during two Weeks before the Survey
Y e s
N o
1 . 2 0 1 -t-
R e f e r e n c e
1 . 2 1 0 +
R e f e r e n c e
1 . 3 5 5 ”
R e f e r e n c e
Age o f the Child
A g e
A g e - S q u a r e
A g e - C u b e
1 . 3 2 1 • * *
0 . 9 9 3 * ”
1 .0 0 0 *”
-2 Log Likelihood 4.073
Degrees o f Freedom 1
26.805
2
61.077
1
39.02
1
94.87
5
501.1
8
• • • p < 0.001, * * p < 0 .0 1 ,» p < 0 .0 5 ,+ p <0.10
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.2.5 Health Seeking Behavior:
Table-5.10 depicts the relationship between the health seeking behavior
variables with stunting. The bivariate (OR=0.87, p < 0.001) and multivariate
logistic regression analysis (OR=0.91, p < 0.001) shows highly significant
lower odds of stunting for children whose mothers received prenatal care
compared to the children whose mothers did not receive prenatal care.
The place of delivery appears to be very important for predicting
stunting. Table 5.10 shows that the significantly lower odds of stunting for
children whose mothers delivered their babies at hospital {government hospital
(OR=0.43, p < 0.001) and private hospitals (OR=0.29, p < 0.001)} for bivariate
models and only private hospitals for multivariate level (OR=0.48, p < 0.001).
The vaccination o f BCG at birth also has a significant positive effect on
stunting. In the bivariate model, the odds of stunting are 15 percent lower for
children who received a BCG vaccine at birth compared to the children who
did not receive BCG vaccine at birth. The effect of BCG vaccination even
appears to be stronger at the multivariate level (OR=0.69, p < 0.001).
However, when the age of the child was not included in the multivariate
model, the BCG did not show any significant effect on stunting, which
indicates that the effect of BCG is confounded by age o f the child.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.10: Odds Ratios obtained from the Ordered Logistic Regression Models o f Health Seeking
Behavior for Predicting the Nutritional Status o f Children Aged 0*59 Months, PDHS, 1990-91
Independent Model-1 Model-2 Model-3 Model-4 Model-5 Model-6 Model-7
Variables
Interceprt
I n t e r c e p t ) - 0 . 8 2 5
I n t e r c e p t 2 0 . 0 4 6
Prenatal Care
N o R e f e r e n c e
Y e s 0 . 8 7 0 * * *
- 0 . 7 9 9
0 . 0 6 3 5
- 0 . 8 7 3
- 0 . 0 0 4
- 0 . 8 8 3
- 0 . 0 3 3
- 0 . 9 4 1
- 0 . 8 9 6
- 0 . 7 5 2 4
0 . 1 2 6 1
R e f e r e n c e
0 . 9 1 1 • • •
- 3 . 3 6 7 9
- 2 . 3 3 9 4
R e f e r e n c e
0 . 9 1 4 * **
Delivery Attendant
U n - t r a i n e d
M e d i c a l T r a i n e d
R e f e r e n c e
0 . 5 6 0
R e f e r e n c e
0 . 8 0 4 *
R e f e r e n c e
0 . 8 6 8
Place o f Delivery
O w n h o m e
G o v e r n m e n t H o s p i t a l
P r i v a t e H o s p i t a l / C l i n i c
R e f e r e n c e
0 . 4 2 9 * * *
0 . 2 9 3 * * •
R e f e r e n c e
0 . 5 2 2
0 . 8 0 4
R e f e r e n c e
0 . 6 9 6
0 . 4 8 4 " *
BCG
N o
Y e s
R e f e r e n c e
0 . 8 4 9 *
R e f e r e n c e
0 . 9 9 2
R e f e r e n c e
0 . 6 8 6 *”
Contraceptive Use
N o
Y e s
R e f e r e n c e
0 . 6 9 0 * *
R e f e r e n c e
1 . 0 2 8
R e f e r e n c e
0 . 7 8 6 +
Age o f the Child
A g e
A g e - S q u a r e
A g e - C u b e
1 . 3 4 7 * **
0 . 9 9 2 ” *
1 . 0 0 0 • * *
-2 Log Likelihood 8 3 . 0 7 2
Degrees o f Freedom 1
4 5 . 4 7 2
I
7 2 . 0 3 2
2
3 . 6 4
1
9 . 8 2 3
1
1 0 5 . 8 4 5
6
5 3 1 . 6 6 2
9
• • • p < 0.001, •* p < 0.01, * p < 0.05, + p < 0.10
200
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.2.6 Socioeconomic Factors and Proximate Determinants:
In the previous section, it is observed that stunting significantly varies
by age of the child, so the age of the child is included in all the models along
with its square and cubic terms. It is seen here in multivariate analysis (Table-
5.11) that all the three age variables are significant (p < 0.001). Model-1
shows the effects o f socioeconomic variables on stunting. It shows that the
mothers’ education, fathers’ education as well as the index of the household
possessions are significantly associated with stunting. It also shows
significantly lower risk of stunting in urban children. Children living in the
province of Punjab have significantly lower stunting and children living in the
province of the Sindh have significantly higher stunting compared to the
children living in the provinces of NWFP and Balochistan.
Model-2 includes the household environmental and hygiene variables
along with the socioeconomic variables. It shows that the children living in
households with public tap as a source of drinking water have significantly
higher odds of stunting compared to households depending on wells as a
source of drinking water. Households with flush toilet facilities are better
nourished compared to their counterparts living in households without any
toilet facilities. The fit o f the Model-2 based on the score test for the
proportional odds assumption shows (Chi-square=l4.17 with 17 DF and p <
0.66), hence, a good fit.
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.11: Odds Ratios Obtained from the Ordered Logistic Regression Analysis to Predict the Probability
of Stunting and Severely Stunting Among Children 0-59 Months, PDHS, 1990-91
Independent Variables____________ Model-1____________ Model-2___________ Model-3 Model-4
Intercepts
Intercept 1 -3.124 • • • -3.148 • • • -3.465 -3.763 *•*
Intercept 2 -2.044 • • • -2.059 • • • -2.362 -2.640
Age o f the Child
Age o f the Child 1.350 • • • 1.356 • • • 1.358 • • • 1.364 • • •
Age-square 0.992 • • • 0.992 • • • 0.992 0.991
Age-cube 1.000 • • • 1.000 *•* 1.000 • • • 1.000 • • •
Mother's Education 0.913 • • • 0.924 • • • 0.928 • • • 0.932 • • •
Father's Education 0.971 • • 0.977 • 0.972 • 0.973 •
Socioeconomic Status:Lower (Ref)
Higher 0.505 • • • 0.591 •* 0.573 **• 0.567 • • •
Medium 0.790 • 0.845 0.820 - 0.798 *
Place o f Residence:Urban 0.636 0.824 0.788 0.812
Province o f Residence:NWFP & Balochistan (Ref)
P unjab 0.615 • • • 0.590 • • • 0.589 • • • 0.619 • • •
S indh 1.422 • 1.431 • 1.501 • • 1.428 •
Source o f Drinking W ater Well (ReO
Pipe inside the home 1.058 1.053 1.026
Pipe in to the property 1.129 1.087 1.093
Public Tab 1.439 • 1.410 -r 1.453 •
Surface 1.208 1.193 1.178
Type ofT oilet Facility:No Facility (Ref)
Bucket 0.763 + 0.756 - 0.763 r
Flush 0.541 • • • 0.517 • • • 0.530 • • •
Construction Baked Bricks: Unbaked Bricks (ReO 0.973 0.996 1.044
Mother's Age 15-29 (reO 0.811 + 0.822 1 -
Previous Birth Interval: 36 -4 7 (ReO
L ess than 18 m onths 1.385 * 1.439 •
1 8 - 3 5
1.241 * 1.297 •
4 8 and m ore
0.736 • 0.730 •
Sex o f the Child:Boy 1.075 l.092
Number o f Surviving Sibling: No Sibling (ReO
1-2 1.348 1.498
3-4 1.394 1.588
5 and over 1.680 + l.9 ll *
Sibling Death 1.145 l .157
Premature Birth 3.205 •
Baby-size at Birth:Average (ReO
S m all-size 2.360
B ig-size 1.136
Feeding with Nipple 0.778 •
Had Diarrhea During 15 days before survey 1.337 •
-2 Log Likelihood Ratio 6 5 2 . 3 7 6 7 2 . 3 4 7 0 3 . 1 6 7 4 4 . 2 2
Degrees o f Freedom 10 17 26 31
Score Test: p-value 0.42 0.66 0.52 0.48
• • • p < 0.001, • • p < 0.01, • p < 0.05. + p < 0.10
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.12: Odds Ratios Obtained from the Ordered Logistic Regression Analysis to Predict the
Probability o f Stunting and Severely Stunting Among Children 0-59 Months, PDHS, 1990-91
Independent Variables____________________________________________ Model-5______ Model-6
Intercepts
Intercept I -3.817 • • • -3.966 • • •
Intercept 2 -2.683 • • • -2.829 ***
Age o f the Child
Age o f the Child 1.371 1.378
Age-square 0.991 • • • 0.991 • • •
Age-cube 1.000 • • • 1.000 • • •
Mother's Education 0.949 •* 0.954 • •
Father's Education 0.972 • 0.973 •
Socioeconomic Status: Lower (Ref)
Higher 0.592 • • 0.599 • •
Medium 0.822 + 0.826 t
Place o f Residence: Urban 0.859 0.900
Province o f Residence:NWFP & Balochistan (ReO
Punjab 0.588 • • • 0.591 • • •
Sindh 1.605 • • 1.636 • •
Source o f Drinking W atenW ell (ReO
Pipe inside the home 1.066 1.051
Pipe in to the property 1.112 1.112
Public Tab 1.435 * 1.439 •
Surface 1.219 1.244
Type o f Toilet Facility: No Facility (ReO
Bucket 0.739 • 0.751 •
Flush 0.560 • • • 0.600 • • •
Construction Baked Bricks: Unbaked Bricks (ReO 1.066
Mother’s Age 15-29 (reO 0.813 f 0.802 -
Previous Birth Interval: 36 - 47 (ReO
Less than 18 months 1.441 • 1.497 • •
18-35 1.294 • 1.291 •
48 and more 0.745 • 0.750 *
Sex o f the Child: Boy t.tto 1.471 •
Number o f Surviving Sibling: No Sibling (ReO
1-2 1.529 1.642 -
3-4 1.586 1.713 *
5 and over 1.863 • 2.057 •
Sibling Death 1.214 + I.I96
Premature Birth 4.194 *• 4.080 • •
Baby-size at Birth:Avemge (ReO
Small-sizc 2.221 • • • 2.214 • • •
Big-size 1.163 1.274
Feeding with Nipple 0.803 • 0.793 *
Had Diarrhea During 15 days before survey 1.381 • • 0.865
Prenatal care visits 0.944 •* 0.947 • •
Place o f Delivery: At Home (ReO
At Government Hospital 0.728 0.790
At Private Hospital 0.406 • • • 0.434 • • •
Delivery Attended by Medical Personal 1.142
BCG Vaccine received 0.937
Contraceptive used 1.224
Interactions
Age o f the Child • Boy 0.988 •
Age o f the Child * Diarrhea l.022 *
-2 Log Likelihood Ratio 774.784 780.718
Degrees o f Freedom 37 35
Score Test: p-value 0.23 0.27
• • • p < 0.001, • • p < 0.01, • p < 0.05, + p < 0.10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, when the group o f the demographic variables is included in
the model along with the socioeconomic and environmental variables, the p-
value of score test shows that the proportion odds assumption o f the model
does not hold. To maintain the validity o f this assumption, several variables
were included/excluded from the analysis one by one. When the category of
maternal age 15-19 years was excluded from the model, the score test for the
proportionality assumption becomes non-significant (Model-3), therefore, only
one maternal age-group 30-49 left in the model leaving 15-29 as reference age-
group.
In model-3 (Score test: Chi-square 24.94 with p < 0.52), the effect of a
short preceding birth interval on stunting is significant (p < 0.05). Moreover,
preceding birth interval of 18-35 months also shows a marginally positive
significance on stunting (p < 0.10), whereas, the interval more than 48 months
shows a negative relationship on stunting (p < 0.05). Children of mothers thirty
and over years of age show a negative effect on stunting compared to the
children of mothers aged less than 30 years (p < 0.05). Moreover, it is
observed that children with more than 5 living siblings have a significantly
higher risk of stunting. However, sibling death does also appear to have a
marginal significant effect on stunting (p < 0.10).
When variables presenting the nutritional factors are included in
Model-4 (Score test: Chi-square 30.61 with p < 0.48), the effect of preceding
birth intervals and having more than 5 siblings become more pronounced
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
compared to the previous model with out nutritional factors included. It means
that children who bom with short birth intervals and have more than 5 living
siblings are more likely to be bom premature or small birth size. Mothers who
started giving bottle feeding with nipples to their children have significantly
better nourished children compared to mothers who did not bottle-fed their
children. Children who had episode of diarrhea two weeks before the survey
are significantly stunted than those children who did not have episode of
diarrhea.
Model-5 in table-5.12 (Score test: Chi-square 39.37 with p < 0.28)
includes all the explanatory variables; it shows that all the variables maintain
their significance as observed in the previous model. Among the health seeking
variables, mothers who delivered their babies in the private hospitals show a
significant negative effect on stunting (OR=0.4l, p < 0.001).
In the final model-6, two interactions are included to see whether there
are any sex differentials or if the effect of diarrhea varies as the child grows. It
is observed that both the interactions are statistically significant in the expected
direction. The addition of interactions in the model shows that boys are at a
significantly higher risk of stunting (OR=1.47, p < 0.05) compared to girls but
the difference reduces as the age o f the child increases (OR=0.99, p < 0.05).
The interaction between diarrhea and the age o f the child also shows
the effect of diarrhea on stunting increases (OR=1.02, p < 0.05) as the child
grows. As observed earlier that the prevalence o f diarrhea increase by age, so
2 0 5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the effects o f diarrhea on stunting. The incidence o f diarrhea increases by age
because o f the reduction o f exclusive breastfeeding along with increase in food
supplementation. The food can become contaminated more easily under poor
hygienic conditions so keeping the child on the breast is protective against
diseases (Ahiadeke, 2000).
T a b l e - 5 . 1 3 : P r e d i c t e d P r o b a b i l i t i e s o f S e v e r e l y S t u n t e d , M o d e r a t e l y S t u n t e d a n d N o r m a l
G r o w t h b a s e d o n t h e R e d u c e d O r d e r e d L o g i s t i c R e g r e s s i o n M o d e l ,
C h i l d r e n 0 - 5 9 M o n t h s , P D H S , 1 9 9 0 - 9 1
A g e o f
t h e c h i l d
S e v e r e l y
S t u n t e d
M o d e r a t e l y
S t u n t e d
N o r m a l
G r o w t h
A g e o f
t h e c h i l d
S e v e r e l y
S t u n t e d
M o d e r a t e l y
S t u n t e d
N o r m a l
G r o w t h
0 0 . 0 2 1 0 . 0 4 1 0 . 9 3 9 3 0 0 . 3 4 6 0 . 2 7 6 0 . 3 7 7
I 0 . 0 2 7 0 . 0 5 3 0 . 9 2 0 3 1 0 . 3 4 4 0 . 2 7 6 0 . 3 8 0
2
0 . 0 3 5 0 . 0 6 7 0 . 8 9 8 3 2 0 . 3 4 1 0 . 2 7 6 0 . 3 8 2
3 0 . 0 4 5 0 . 0 8 3 0 . 8 7 2 3 3 0 . 3 3 8 0 . 2 7 6 0 . 3 8 6
4 0 . 0 5 6 0 . 1 0 1 0 . 8 4 3 3 4 0 . 3 3 5 0 . 2 7 6 0 . 3 8 9
5 0 . 0 7 0 0 . 1 1 9 0 . 8 1 1 3 5 0 . 3 3 1 0 . 2 7 6 0 . 3 9 3
6 0 . 0 8 4 0 . 1 3 8 0 . 7 7 7 3 6 0 . 3 2 7 0 . 2 7 5 0 . 3 9 7
7 0 . 1 0 0 0 . 1 5 8 0 . 7 4 2 3 7 0 . 3 2 4 0 . 2 7 5 0 . 4 0 1
S 0 . 1 1 8 0 . 1 7 6 0 . 7 0 6 3 8 0 . 3 2 0 0 . 2 7 5 0 . 4 0 5
9 0 . 1 3 6 0 . 1 9 3 0 . 6 7 1 3 9 0 . 3 1 7 0 . 2 7 4 0 . 4 0 9
1 0 0 . 1 5 5 0 . 2 0 9 0 . 6 3 6 4 0 0 . 3 1 4 0 . 2 7 4 0 . 4 1 3
1 1 0 . 1 7 4 0 . 2 2 2 0 . 6 0 3 4 1 0 . 3 1 1 0 . 2 7 3 0 . 4 1 6
1 2 0 . 1 9 3 0 . 2 3 4 0 . 5 7 2 4 2 0 . 3 0 9 0 . 2 7 3 0 . 4 1 8
1 3 0 . 2 1 2 0 . 2 4 4 0 . 5 4 4 4 3 0 . 3 0 7 0 . 2 7 3 0 . 4 2 0
1 4 0 . 2 3 0 0 . 2 5 2 0 . 5 1 7 4 4 0 . 3 0 7 0 . 2 7 3 0 . 4 2 1
1 5 0 . 2 4 8 0 . 2 5 9 0 . 4 9 4 4 5 0 . 3 0 7 0 . 2 7 3 0 . 4 2 1
1 6 0 . 2 6 4 0 . 2 6 4 0 . 4 7 3 4 6 0 . 3 0 7 0 . 2 7 3 0 . 4 2 0
1 7 0 . 2 7 9 0 . 2 6 8 0 . 4 5 4 4 7 0 . 3 0 9 0 . 2 7 3 0 . 4 1 7
I S 0 . 2 9 2 0 . 2 7 0 0 . 4 3 8 4 8 0 . 3 1 2 0 . 2 7 4 0 . 4 1 4
1 9 0 . 3 0 4 0 . 2 7 2 0 . 4 2 4 4 9 0 . 3 1 7 0 . 2 7 4 0 . 4 0 9
2 0 0 . 3 1 4 0 . 2 7 4 0 . 4 1 2 5 0 0 . 3 2 2 0 . 2 7 5 0 . 4 0 3
2 1 0 . 3 2 3 0 . 2 7 5 0 . 4 0 2 5 1 0 . 3 3 0 0 . 2 7 5 0 . 3 9 5
2 2 0 . 3 3 1 0 . 2 7 6 0 . 3 9 4 5 2 0 . 3 3 9 0 . 2 7 6 0 . 3 8 5
2 3 0 . 3 3 7 0 . 2 7 6 0 . 3 8 7 5 3 0 . 3 4 9 0 . 2 7 7 0 . 3 7 4
2 4 0 . 3 4 2 0 . 2 7 6 0 . 3 8 2 5 4 0 . 3 6 2 0 . 2 7 7 0 . 3 6 1
2 5 0 . 3 4 5 0 . 2 7 6 0 . 3 7 9 5 5 0 . 3 7 7 0 . 2 7 6 0 . 3 4 6
2 6 0 . 3 4 7 0 . 2 7 6 0 . 3 7 6 5 6 0 . 3 9 5 0 . 2 7 5 0 . 3 3 0
2 7 0 . 3 4 8 0 . 2 7 6 0 . 3 7 5 5 7 0 . 4 1 5 0 . 2 7 4 0 . 3 1 2
2 8 0 . 3 4 9 0 . 2 7 6 0 . 3 7 5 5 8 0 . 4 3 7 0 . 2 7 0 0 . 2 9 2
2 9 0 . 3 4 8 0 . 2 7 6 0 . 3 7 6 5 9 0 . 4 3 7 0 . 2 7 0 0 . 2 9 2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The predicted probabilities for the children of severely stunted,
moderately stunted and with normal growth are calculated and presented to
highlight the effect of one or more predictors by controlling the effect of other
variables. Figure-5.1 depicts how the probability of severely stunted and
moderately stunted children increases and the probability o f the children with
normal growth decreases as the age o f the child increases. The probabilities of
being in each category of the dependent variable are calculated keeping the
average levels of the independent variables of the then age. Table-5.13 shows
that the probability of severely stunted children increases up to age 29 months
when it reaches about 34.9 percent, and then slightly decreases to 30.7 percent
at the age of 44 months and after that it again gets the momentum of higher
probability of severely stunting. On the other hand, the probability of
moderately stunting increases from 4 percent at age of 1 month to 27 percent at
the age of 23 months and remains stable for rest o f the months. However, the
probability of children for normal growth sharply dropped from 94 percent at
age 1 month to 80 percent at age 6 to 60 percent at age 12 and 37.5 percent at
age 30 months. After the age o f 30 months the probability o f normal growth
increases from 37.5 percent to 42 percent at age 47 months and then again
declines sharply to 29 percents at age 59 months. Figue-5.1 demonstrates these
differences more noticeably.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure-5.1: Predicted Probabilities of Stunting based on Final Ordered Logistic
Regression Model by Age of the Child, taking Mean Values of all the
Independent Variables, PDHS, 1990-91
N orm al G ro w th
□ N o rm a l G ro w th ■ S tu n te d O S ev erely S tunted
Figure-5.2: Predicted Probabilities of Stunting based on Final Ordered Logistic
Regression Model by Age of the Child, Lower SES Illeterate Parents living in Rural
with Wells water and No Toilet Facility, PDHS, 1990-91
N o rm a l G ro w th
Q N o rm a l G ro w th ( S t u n t e d ( S e v e r e ly S tu n te d
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure-5.2: Predicted Probabilities of Stunting based on Final Ordered Logistic
Regression Model by Age of the Child, Higher SES10 Years education Parents living
in Urban Punjab with Piped water and Flush System, PDHS, 1990-91
N orm al G row th
O N orm al G ro w th ■ S tun ted Q S everely S tu n ted
Figure-5.2 and figure-5.3 show the differences between the
advantaged1 and disadvantaged2 children as they grow. It is observed that, at
age zero, the proportion of stunting does not exist at birth for the advantaged
children, compared to 20 percent of stunted among the disadvantaged children
at birth. Moreover, the differences become widened as the children grow. It is
clear that at age 5, only 10 percent (figure-5.2) of the non-privileged children
have normal growth compared to more than 70 percent (figure-5.3) for the
privileged children.
Stunting is a phenomenon of early childhood and a direct result of poor
diets and infection (Martorell & Habicht,1986). According to them the intense
period of growth retardation is generally between 3 and 12 or 18 months.
1 Children with parents who have 10 years of education, live in urban Punjab and have piped water
connected to their houses and have flush toilet facilities.
2 Children with illiterate parents, living in Sindh rural with Wells water and have no toilet facility.
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, in some countries growth retardation continues into the third year or
longer but to a lesser extent. At the end o f this process, marked departure from
normality will have often occurred. There are multiple reasons why stunting
occurs in early childhood and not later. In childhood nutritional needs are
greater, in relation to weight, than at any time later. One o f the reasons that
nutritional requirements are high is that growth velocities are the highest they
will ever be (Martorell, et al., 1994). Thus, the opportunity for growth
retardation is great in early childhood, partly because more growth is taking
place. Moreover, infections limit growth in very young children because
episodes are more frequent and more severe, especially the malnourished.
Finally, young children are totally dependent on others for their care and are
hence most vulnerable to poor care taking (Martorell, et al., 1994).
5.3 Multivariate Logit Analysis: Mortality and stunting Together:
In this section, the dependent variables, mortality in chapter 4 and stunting in
the previous sections o f this chapter, are pooled together in a multivariate logit model
to see the relationship of independent variables and the combined dependent variable.
As explained in chapter 3 this dependent variable consists o f 4 categories namely
dead, severely stunted, moderately stunted, and normal growth. The SAS program
produces the results in the form of coefficient estimates of 3 equations out of 4
categories o f the dependent variable as well as the p-values of wald chi-square test.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.14: Odds Ratios obtained from the Multinomial Logit Model to Predict the Probability
of Death, Severely Stunted and Moderately Stunted children, PDHS, 1990-91
Independent Variables Dead Severely Stunted Stunted
Intercept -4.39I3*** -3.7805 *•• -2.8139***
Age o f the
child: Age 1.186+ 1.503 * * • 1.317***
Age-Square 0.990+ 0.989*** 0.992***
Age-Cube 1.000 1 .000* * * 1 .000* * *
M other Education 0.992 0.945+ 0.949+
Father Education 0.877 • • • 0.964* 0.967*
Household Possessions Index: Lower (Ref)
Higher 0.537 0.555 ** 0.781
Medium 0.694 0.833 1.040
Place o f Residence: Urban 0.794 0.864 0.865
Province o f Residence:NW FP & Balochistan (Ref)
Punjab 2.539+ 0.544 • • 0.570 **
Sindh 4.669 • • 1.688 ** 1.244
Source o f Drinking Water: Well (Ref)
Pipe in to the house 0.289 • • 1.189 1.133
Pipe on to the Property 1.096 1.099 1.132
Public Tap 0.990 1.620* 1.648*
Surface 0.322 1.430+ 1.554-
Latrine Facility: No Facility (Ref)
Bucket o r Pit 2.479 • 0.710+ 0.948
Flush 2.023 0.487 ** 0.871
Mother Age: 20-29 (Ref)
15-19 0.971 1.103 0.282 **
30-39 2 .4 1 0 " 0.786 0.858
Birth Order:3-5 (Ref)
2nd 0.864 0.857 1.477*
6th and over 0.490* 1.349+ 1.460*
Previous Birth Interval: 36 - 47 (Ref)
Less than 18 months 17.140*** 2.054 0.886
18-35 5.152*** 1.280 1.149
48 and more 0.286 * * 0.724+ 0.985
Sex o f the Child:Boy 0.593 1.975 •• 1.341
Previous Sibling Death 29.198*** 1.052 1.024
Premature births 9.801 *• 9.085 **• 3.584-
Baby-size at Birth:Average (Ref)
Small-size 1.248 2.786*** 1.149
Big-size 7.920* 1.202 0.656
Bottle Feeding with Nipple 0.625 0.786 0.702 **
Prenatal care visits 1.026 0.911 ** 0.987
Place o f Delivery: At Home (Ref)
A t Government Hospital 1.002 0.523 + 1.073
At Private Hospital 0.269+ 0.327 **• 0.673
Delivery Attendant 0.571 + 1.124 0.924
BCG Vaccine received 0.454** 0.858 1.132
Contraceptive Use 2.151 + 1.420 0.893
interactions: Birth interval < 18 & Prenatal care 0.073 **• 0.848 1.575
Birth interval 19-35 & Prenatal care 0 J5 6 + 1.215 0.586
Age o f the child & Boy 1.068+ 0.980* 0.983*
p < 0.001, • • p < 0.01, * p < 0.05, + p < 0.10
211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table-5.14 shows the, odds ratios and the significance level of each
independent variable for each equation.
First of all, the result of multinomial logit model shows that, among the
socioeconomic factors, the fathers’ education has a stronger effect on lowering, both,
the mortality and stunting among children aged 0-59 months than the mothers’
education. It confirms the earlier results of proportional hazard and ordered logistic
regression analyses. The multinomial logit analysis, however, shows that the effect of
father’s education on mortality is stronger (OR=0.88, p < 0.001) than severely stunting
(OR=0.96, p < 0.05) and moderately stunting (OR=0.97, p < 0.05). Mothers’
education does not show any significant effect on mortality whereas, it shows a
marginally significant effect on severely stunting (OR=0.95, p < 0.10) and moderately
stunting (OR=0.95, p < 0.10). Multinomial logit model also demonstrated that children
living in the province of Sindh have 4.7 times higher odds o f mortality than children
living in NWFP and Balochistan. Punjabi children have 45 percent lower odds of
severely stunting and Sindhi children are 69 percent higher odds of severely stunting
compared to children living in Balochistan and NWFP provinces.
Among the household environmental and hygiene factors, children o f mothers
living in households connected to piped water are enjoying significantly better
survival chances (OR=0.289, p < 0.01) than children of mothers living in households
depending on wells as a source of drinking water. Children living in households
depending on public tap water have significantly higher odds o f severely stunting
(OR=l.62, p < 0.05) and moderately stunting (OR=1.65, p < 0.05). Households using
212
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
bucket or pit as their toilet facility are more likely to have higher child mortality than
the households without any toilet facilities. It is also observed that the children living
in households equipped with flush system are significantly less likely to be severely
stunted (OR=0.49, p < 0.01) than the children living in households without any toilet
facilities. In the proportional hazard analysis, it was observed that toilet facilities do
not have any significant effect on post neonatal mortality.
The multinomial logit model maintains the importance of demographic
variables in predicting the childhood mortality as obtained from the proportional
hazard model analysis. Age of the mother at the time o f the birth, birth order, previous
birth interval, and sibling death all have statistically significant effect on mortality. For
severely stunted children, boys have significantly higher odds of stunting (OR=1.97, p
< 0.01) compared to girls, whereas, children of younger mothers have lower odds
(OR=0.28, p < 0.01) of moderately stunting and children of lower birth order
(OR=1.48, p < 0.05) and very higher birth order (OR=l.47, p < 0.05) have
significantly higher odds of moderately stunting.
Premature bom children have almost 10 times higher odds of mortality
(OR=9.8, p < 01) and, even if they survived, they still have 9 times higher odds of
severely stunting (OR=9.0, p < 0.001) or 3.5 times higher odds of moderately stunting
(OR=3.6, p < 0.10). It is clear that premature bom children have very little chances to
catch-up the normal growth compared to the full-term bom babies. Small-sized bom
children also have higher odds of severely stunting. Children who bottle-fed with
nipple have lower odds o f moderately stunted (OR=0.70, p < 0.01).
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Among the panel of health seeking behavior variables, the significant
advantage of utilization o f prenatal care for mothers with shorter birth intervals
demonstrated by the multinomial logit model. Moreover, babies delivered at private
hospitals have lower odds of mortality (OR=2.15, p < 0.10) compared to babies
delivered at home and babies who had BCG vaccination at birth also have the lower
odds of mortality (OR=0.45, p < 0.01). Babies delivered at hospitals have one-third
odds of severely stunting (OR=0.33, p < 0.001) compared to babies delivered at home.
The multinomial logit model confirms the results of the proportional hazards
model regarding the advantages of prenatal care for mothers with short birth intervals.
Mothers with shorter birth intervals have only 7 percent of the mortality if they availed
the prenatal care services during their pregnancies compared to the mothers with short
birth intervals who did not get these services.
It also confirms the results of logistic regression models that boys are more
likely to be stunted compared to the girls, but the risk of stunting reduces for boys as
they grow.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 6
CONCLUSION
In this analysis we have examined the mechanisms how the socioeconomic
factors operate through: environmental, demographic, nutritional and health care
factors that affect neonatal and post neonatal mortality as well as stunting of living
children in Pakistan. The findings of this analysis of PDHS data do not fully support
the hypothesis that socioeconomic factors affect child survival only through the
proximate determinants as proposed by Mosley and Chen (1984), rather also have
their independent and direct effect in improving the child survival in Pakistan. The
analysis shows that the proximate determinants have stronger influence on neonatal
and post neonatal mortality than the socioeconomic factors. The findings are
consistent with the findings of a previous study from Thailand (Zheng, 1993).
The possible explanation for this may be that, first, there are some restrictions
applied in the selections of the effective sample for this analysis. In order to avoid the
violation of the independence assumption of multivariate analysis, only the last births
were included in the analysis. The sample was also restricted to singleton births bom
1-59 months before the survey. To include the survival status of the older siblings,
only women who had at least two births were included. This selection xcluded all the
younger mothers who had only one child which normally has higher mortality in
developing countries. Second, in the absence of household income, the socioeconomic
variables selected may not truly represent the socioeconomic status of the families.
215
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to Mosley and Chen’s (1984) framework, socioeconomic factors
affect the outcome: in this case mortality and stunting, through the proximate
determinants. The empirical model developed for this analysis identifies four sets of
proximate determinants (domestic environmental, demographic, nutritional, and health
seeking behavior factors). To examine the importance of these factors in studying the
neonatal and post neonatal mortality in Pakistan, proportional hazards analyses are
employed. For the analysis of stunting, ordered logistic regression is used. Finally,
multinomial logit analysis is employed to observe the affect of each independent
variable on mortality and stunting together.
6.1 Socioeconomic Factors:
In, both, proportional hazards and multinomial logit analyses, the effect
o f the father’s education appears to be stronger on child survival than the
mother’s education. At a bivariate level, maternal education shows its
importance in improving neonatal mortality but the effect disappears after
including the fathers’ education, index of household possessions and place of
residence in the model. However, maternal education proves its importance on
better child survival in rural areas in the multivariate analysis. The
proportional hazards analysis also shows that mothers’ education improves
post neonatal mortality may be by providing hygienically cleaner bottle-
feeding with nipple to their children compared to their uneducated
counterparts.
216
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Both, at the bivariate and multivariate levels, the fathers’ education
shows significant negative effect on post neonatal mortality. Children of
parents with a lower socioeconomic index are also at significantly higher risk
of post neonatal mortality. The analysis shows that biological reasons, more
than socioeconomic factors, are responsible for higher neonatal mortality in
Pakistan. However, region of residence significantly improves the survival
status during neonatal and post neonatal periods. Children bom in the
provinces o f Punjab and Sindh have significantly higher mortality.
However, children living in the province of Punjab have significantly
lower odds of stunting and severely stunting and children living in the province
of Sindh have significantly higher odds of severely stunting compared to the
children living in the provinces of NWFP and Balochistan. These findings
support the earlier research that found higher mortality in the province of
Punjab and lower mortality in the province of Balochistan (UN, 1986). The
explanation given was complete enumeration of households in the Punjab
relative to other regions o f residence. Punjab is the most agricultural Province
of Pakistan where more than 50 percent o f the population lives.
The findings of this analysis confirm the results o f earlier research
conducted in developing countries that biological factors are the best predictors
of neonatal mortality than the socioeconomic factors. However, the
significance o f fathers’ education in predicting the post neonatal mortality is
much stronger than the mothers’ education.
217
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The results of the ordered logistic regression show that higher
socioeconomic status significantly improves the nutritional status of living
children. This pattern persists even when the differences in household
environment, demographic, nutritional and health care factors are taken into
account. These findings are also in accordance with the earlier results that say
stunting is caused by poverty in developing countries. As the socioeconomic
status o f parents increases, the prevalence o f stunting decreases. Higher
educated parents (both father and mother) may allocate more food to their
children.
6.2 Household Environment and Hygiene:
Among the environmental factors, the results of proportional hazards
model and multivariate logit analysis show that families who have piped water
connected to their houses have significantly lower post neonatal mortality than
those families who depend on wells for drinking water. Households who
depend on public tap water have significantly higher stunted children
compared to the children bom in households depending on wells for drinking
water. The proportional hazard analysis does not find evidence o f improved
neonatal (less than 1 month) and post neonatal mortality (1-23 months) in
households by type of toilet facilities. However, in the multivariate logit
analysis (0-59 months), it was observed that households using bucket or pit as
toilet have significantly higher mortality than those who do not have any. This
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
may be that, as the children grow, they have longer exposure to unhygienic
conditions which affect their survival status.
The highly statistically significant better nutritional status is observed
for children living in households equipped with flush toilet facilities. The
children living in households having at least pit or bucket types of toilet
facilities in their households also have better nutritional status compared to the
children living in households without any toilet facility. The results are
indicative of the importance of safe drinking water and better toilet facilities
for improving child survival in Pakistan.
6.3 Demographic Factors:
Among the demographic variables, preceding birth interval and the
survival status of the older siblings are the most important demographic
determinants of neonatal and post neonatal mortality in Pakistan. There are
many reasons that deaths tend to cluster in families, such as premature delivery
or intrauterine growth retardation. These are likely to be repeated in other
pregnancies as well. Also those children share risks associated with family
behavior and childcare practice, such as infant feeding, use o f health services
and general standards of domestic environment and hygiene.
The analysis highlights the advantage o f prenatal care to prevent these
neonatal and post neonatal deaths if those mothers avail the prenatal care
services when they got pregnant after shorter birth intervals. The analysis
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
shows that the survival of children improves significantly, both, at neonatal
and post neonatal periods when the mothers with shorter birth intervals
attended prenatal care services compared to the mothers with longer birth
intervals. The analysis also finds the importance of prenatal care to prevent the
neonatal mortality for mothers living in rural areas compared to the mothers
living in urban areas. There is a need to improve the knowledge and use of
family planning and health care services.
The ordered logistic regression analysis shows that children bom at
least 3 years after the preceding birth interval are significantly taller in
Pakistan. Children who have more than 5 living siblings are significantly
shorter than children who do not have any living siblings. The analysis also
shows that children having 1-2 or 2-4 living siblings are also more likely to be
shorter than children who do not have any living siblings. This may be due to
the food competition between siblings in the family.
Our results clearly indicate that the children in families with shorter
birth intervals and higher number of living siblings are at a higher risk of, both,
mortality and stunting. Longer intervals between births will allow more time
for the allocation of sufficient family resources for the provision of food for
additional needs of these children. In addition, the nutritional problems are
also compounded when another child is bom with shorter birth interval.
Therefore, information, education, and communication efforts should
encourage mothers not only 3-4 years intervals between births to reduce the
220
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mortality and stunting, but also nutritional vulnerabilities as well as the
attention needed for these children. Moreover, use of family planning to
increase birth intervals and reduced family size can result in significant
reductions in childhood mortality and stunting.
6.4 Nutritional Factors:
Breastfeeding proved its importance during the first two years o f the
child’s life. The proportional Hazards analysis shows a significant negative
effect of continued breastfeeding on post neonatal mortality. The age of
children at which mothers start giving supplementary foods is also very
important in preventing the post neonatal mortality. This analysis shows that
the delay in start giving supplementary foods significantly increase the risk of
post neonatal mortality. In Pakistan, very few children receive supplementary
foods during the first 2-3 months, the period during which this appears to be
harmful to long-term growth. The major challenge is to protect the practice of
breastfeeding rather than to promote change in feeding practice.
Premature births and big baby sizes at birth are at significantly higher
risk o f mortality during neonatal period. Research has identified that maternal
factors affect not only birth weight and birth length, but also growth afterward.
Mothers who suffer serious under-nutrition during pregnancy tend to produce
babies who carry a life-long disadvantage in terms of, both, physical and
mental health. These mothers also produce lower birth-weight babies, who
221
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tend to die sooner than babies of normal birth-weight. On the other hand,
diabetic mothers are more likely to deliver very big size babies which have
also higher mortality during neonatal period.
6.5 Health Seeking Behavior:
Regular prenatal care is needed to help detect and manage some
pregnancy-related complications and to educate women about danger signs,
potential complications, and where to seek help. Prenatal care beginning early
in the first trimester of pregnancy and continuing on a regular basis is
important to the health of both mother and infant. Early prenatal care provides
an opportunity to offer preventive care that will benefit the infant as well as the
mother such as, counseling on hygiene, breastfeeding, nutrition, family
planning, immunization and iron supplementation. Prenatal care also benefits
treatment o f existing diseases that may be aggravated by pregnancy. Prenatal
care helps to prevent complications during pregnancy and labor. Our analysis
confirms these early findings and shows the importance of attending prenatal
services care in preventing both mortality and stunting. However, sick bom
babies are not given any vaccination including BCG. These sick bom babies
have higher chances o f mortality during neonatal period.
Moreover, babies delivered at private hospitals have lower odds of
stunting compared to babies delivered at home. Babies who had BCG
vaccination at birth also have the lower odds of mortality. The BCG
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vaccination not only shows the prevention of tuberculosis, it also shows the
parental enthusiasm of utilizing the preventive measures for their children.
This vaccine is given to the children at birth, therefore, children who received
this vaccine are much more likely to get full immunization.
The analysis shows a curvilinear relationship between stunting and child age.
It is because in early childhood nutritional needs are greater in relation to weight than
at any time later in the life. Therefore, the opportunity for growth retardation is greater
in early childhood, partly because more growth is taking place during this period.
Moreover, infections limit growth in very young children because episodes are more
frequent and more severe. The findings of this analysis show that diarrhea has a
greater negative impact on child’s nutritional status as the child grows. Diarrheal
diseases are widespread in Pakistan. This analysis finds that the diarrheal disease is
strongly associated with lower heights in children. In Pakistan, the ORS packets are
distributed through the government service delivery outlets for free. In many cases
people are taught how to use ORS, but do not use it because it does not conform to
their understanding o f the character of diarrhea. It is a common practice in Pakistan
that mothers reduce the feeding or even stop feeding the child during diarrhea because
they believe that giving food will increase the stool output and thus it make the
diarrhea worse. To prevent growth faltering, good nutrition must be maintained both
during and after an episode o f diarrhea. This can be achieved by continuing to give
considerable amounts of nutritious foods throughout the episode and during the period
of recovery.
2 2 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, our findings aiso show that boys are at a higher risk o f stunting
compared to the girls; but as boys grow, the odds o f stunting decreases compared to
girls. A study in Bangladesh found that when boys experienced diarrhea, parents are
more likely to take them for medical treatment at earlier stage compared to their
sisters. In contrast, the PDHS data shows that 54 percent of girls were taken to health
care provider compared to only 43 percent o f boys (NIPS, 1992).
The data presented in this analysis documents a disturbing picture o f mortality
and under-nutrition among children less than five years of age in underprivileged
subgroups. The findings confirm the great magnitude of under-nutrition which
continues to hamper the physical growth and mental development o f more than a half
of the Pakistani children. Indeed, it is a major threat to their very survival.
The analysis shows that the causes o f growth retardation are deeply rooted in
poverty, unhygienic household environment, non-utilization of the health services and
lack of education. In order to eliminate the existing imbalances and disparities
between urban rural female education, mixed schools (co-education) may be opened in
rural areas. In addition, the recruitment age o f female teachers may be relaxed to
increase their availability. In order to retain girls in rural schools, free textbooks,
stipends and nutritional food may also be provided in disadvantaged and far-flung
areas. This will result in an increase in enrolment and a reduction in the female drop
out rate.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The nutritional well being of people is a precondition for the development of
societies. The tragic consequences o f malnutrition include death, disability, and
stunted physical and mental growth and, as a result, retarded national socioeconomic
development (WHO, 2000).
In order to continue to allow underprivileged environments to affect child
development not only perpetuates the vicious cycle o f poverty, but also leads to an
enormous waste o f human potential. The Pakistani government may not be successful
in their efforts to accelerate economic development in any significant long-term sense
until optimal child growth and development are ensured for the majority.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bibliography
Acheson, K.J., I.T. Campbell, O.G. Edholm, D.S. Miller, and M.J. Stock, 1980,
“The measurement o f food and energy intake in man - An evaluation of some
techniques”, American Journal of Clinical Nutrition 35:1147-54
Afzal, M., Raja, T. A., Mohammad, Ali, 1988, “Some differentials in infant and
child mortality risks in Pakistan 1962-1986”, Pakistan Development Review
27(4):635-43
Aksit, B. and B. Aksit, 1989, “Sociocultural determinants o f infant and child
mortality in Turkey” Social Science and Medicine 28(6):571 -576.
Al-Kabir, A., 1984, “Effects of community factors on infant and child mortality in
rural Bangladesh, WFS Scientific Report NO. 56, ISI, Voorburg.
Alam, A. and J. Cleland, 1984, “Some differentials in infant and child mortality
risks in Pakistan”, in Fertility in Pakistan: a review o f findings from the
Pakistan Fertility Survey, edited by Alam, Aqbal and Betzy Dinesen, pp. 187-
212. International Statistical Institute, Voorbrg.
Anderson, P., D. Pelletier, H. Alderman, 1995, “Beyong child survival:An
overview of the issues”, In Child Growth and Nutrition in Developing
Countries, edited by P.Pinstrup-Anderson, D. Pelletier, H. Alderman, Ithaca,
NY:Comell University Press
Armstrong, J.R. and H. Campbell, 1991, “Indoor air pollution exposure and lower
respiratory infections in young Gambian children”, International Journal o f
Epidemiology 20(2):424-429.
Annan, A, 1985, ‘T he socioeconmic determinants o f nutrition among preschiil
children in Ghana”, Unpublished Ph.D dissertation, Cornell University,
Ithacam, NY
Akther. H. H., Chowdhury, M. E. and Sen, A., 1996, A cross-sectional study on
maternal morbidity in Bangladesh, Bangladesh Institute of Research for
Promotion of Essential and Reproductive Health and Technologies, Dhaka.
Al-Kabir, A., 1987, “Effects of community factors on infant and child mortality in
rural Bangladesh”, Studies in Family Planning 18(2): „
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bairagi, R., 1985, “Why mortality-discriminating power o f anthropometric
indicators differs among populations”, Journal o f Tropical Pediatics 31:63-64
Bairagi, R, 1981, “On validity of some anthropometric indicators as predictos of
mortality”, American Journal o f Clinical Nutrition 34:2592-2594
Balk, D., 1997, “Defying gender norms in rural Bangladesh: a socio demographic
analysis”, Population Studies 5 1:153-172
Bang, A., O. Bang, P. Tale, R. Sontakke, J. Solanki, R. Wargantiwar and P.
Kelzarkar, 1990, “Reduction in pneumonia mortality and total child mortality
by means of community-based intervention trial in Gadchiroli, India”, Lancet,
336:201-206.
Barrera, A., 1990, “The role o f mother’s schooling and its interaction with public
health programs in child health production”, Journal o f Development
Economics 32:69-91
Basu, A.M. 1989, “Is discrimination in food really necessary for explaining sex
differentials in childhood mortality?”, Population Studies 43(2): 193-210.
Bateman, O.M., Smith, S., and Roark, P., 1993, “Acomparison of the health effects
o f water supply and sanitation in urban and rural areas o f five African
countries, (WASH Field Report No. 398), Water and Sanitation for Health
Project, Arlington, VA
Beaton, G.H., 1986, “E.V.McCoIlum International Lectureship in nutrition, toward
harmonization o f dietary, biochemical and clinical assessments:the meanings
o f nutritional status and requirements”, Nutritional Review 44:349-358
Bender, W. and M.Smith, 1997, “Population, food, and nutrition”, Population
Bulletin 51(4), A publication o f the Population Reference Bureau, Inc.
Bernhardt, E. 1992, “Crowding and child survival in Stockholm 1895-1920”
Paper presented at IUSSP Seminar on Child and Infant Mortality in the Past,
Universite de Montreal, Montreal, 7-9 October.
Bicego, George T. and J. Ties Boerma 1993, "Maternal education and child
survival: a comparative study o f survey data from 17 countries", Social
Science and Medicine 36(9): 1207-1227.
2 2 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bhatia, S. 1983, Traditional practices affecting female health and survival:
evidence from countries in South Asia, in Sex Differentials in Mortality:
Trends, Determinants and Consequences, eds., A.D. Lopez and L.T. Ruzicka,
The Australian National University, Canberra, pp. 165-177.
Bhuiya, Abbas and K. Streatfield, 1991, “Mothers’ education and survival of
female children in a rural area of Bangladesh”, Population Studies 45:253-264
Bhuiya, Abbas, 1989, “Factors affecting child survival in Matlab, Bangladesh”,
Ph.D Thesis, Department of Demography, Australian National University,
Canberra
Black, R.E., G. Lopez de Romana, K.H. Brown, N. Bravo, O.G. Bazalar, and H.C.
Kanashiro, 1989, “Incidence and etiology of infantile diarrhea and major
routes of transmission in Huascar, Redu”, American Journal o f Epidemiology:
129:785-799
Black R.E., Brown K.H. Becker S, 1984, “Malnutrition is a determining factor in
diseeheal duration, but not incidence, among young children in a longitudinal
study in rural Bangladesh”, American Journal o f Clinical Nutrition 39:87-94
Blossfeld, H.-P. and G. Rohwer, 1995, Techniques o f Event History Modeling.
New Approaches to Causal Analysis, Mahwah NJ: Lawrence Erlbaum
Associates.
Blum, D. and Feachem, R.G. 1983, “Measuring the impact o f water supply and
sanitation investments on diarrheal: Problems of methodology”, International
Journal o f Epidemiology 12:357-365
Boerma, J.T., Sommerfelt, A.E. and Rutrtein S.O, 1991, Childhood Mortality and
Treatment Patterns, Demographic and Health Surveys Comparative Studies
No. 4, Macro Intemation, Inc. Maryland.
Boerma J.T., and George T. Bicego, 1992, “Preceding birth interval and child
survivahsearching for pathways o f influence”, Studies in Family Planning
23(4):243-256
Booth, H. and I.H.Shah,1984 “The data and their quality” in Fertility in Pakistan:
a review offindings from the Pakistan Fertility Survey, edited by Alam, Aqbal
and Betzy Dinesen, International Statistical Institute, Voorbrg.
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Brakohiapa LA, Yartey J, Bille A, Harrison E, Quansah E, Armar MA, 1988,
“Does prolonged breastfeeding adversely affect a child's nutritional
status!"Lancet, ii:416-8.
Briend A, Wojtyniak B, Rowland MGM., 1988, “Breast feeding, nutritional state,
and child survival in rural Bangladesh”, British Medical Journal 296:879-82.
Brown KH, Black RE, de Romana GL, de Kanashiro HC., 1989, “Infant-feeding
practices and their relationship with diarrheal and other diseases in Huascar
(Lima), Peru”, Pediatrics 83:31-40.
Brown K.H. and Begin F, 1993, “Malnutrition among weaning of developing
countries: Still a problem begging for solutions”, Jounral o f Pediatric
Gastroenterol Nutrition 17:132-138
Butz, W.P., Habecht, J. P. and DaVanzo, J. 1984, “Environmental factors in the
relationship between breastfeeding and infant mortality: The role o f sanitation
and water in Malaysia”, American Journal o f Epidemiology 119:516-525.
Cabigon, J.V., 1997, “The effect of birthspacing and breastfeeding on childhood
mortality in the Philippines ”, Journal o f Population 3 (1): 1-8
Caldwell, J., 1979 “Education as a factor in mortality decline: An examination of
Nigerian data”, Population Studies 33(2):395-413.
Caldwell J.C., 1980, “Mass education as a determinant of the timing of fertility
decline”, Population and Developing Review 6(2):225-255
Caldwell, J. and P. McDonald, 1981, “Influence of maternal education on infant
and child mortality: levels and causes”, International Population Conference,
Vol.2, Manila: International Union for the Scientific study o f Population, pp.
79-96.
Caldwell, J.C. and McDonald P., 1982, “Imfluence o f matema; education on infant
and child mortality: Levels and causes”, Health Policy and Education 2:251-
267
Caldwell. J, P.H. Reddy and P. Caldwell, 1983, “The social components of
mortality: An investigation in South India employing alternative
methodologies”, Population Studies 37(2)185-205.
2 2 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Caldwell, J., 1986, “Routes to Low Mortality in Poor Countries”, Population and
Development Review 12(2): 171 -220.
Caldwell, John C., I. Gajanayake, Pat Caldwell and I. Pieris, 1989a, “Sensitization
to illness and the risk o f death: an explanation for Sri Lanka’s approach to
good health for all”, Social Science and Medicine 28(4):365-379
Caldwell, John C., 1989, “Mass education as a determinant of mortality decline”
Pp. 101-111 in Selected Readings in the Cultural, Social and Behavioural
Determinants o f Health, ed. J. C. Caldwell and G. Santow. Canberra: Health
Transition Centre, The Australian National University.
Caldwell, John C., 1990, “Cultural and social factors influencing mortality levels
in developing countries”, Annals o f the American Academy o f Political and
Social Science 510:44-59.
Caldwell, J.C., P.Caldwell, L.Gajanayake, I.OrubuIoye, I.Pieris and P.H.Reddy,
1990, “Cultural, social and behavioural determinants of health and their
mechanisms: a report on related health programs”, Pp. 534-541 in What We
Know about Health Transition: The Cultural, Social and Behavioural
Determinants o f Health, edited by John C. Caldwell, S.Findley, P.Caldwell,
G.Santow, W.Cosford, J.Braid and D.Broers-Freeman.Canberra: Health
Transition Centre, The Australian National University.
Caldwell, J.C., 1991, “Major new evidence on health transition nd its
interpretation, Health Transitional Review l(2):22l-229
Caldwell, J.C. 1994a. “How is greater maternal education translated into lower
child mortality?” Health Transition Review 4(2):224-229.
Caldwell, J.C. 1994b, “Health transition: The cultural, social, and behavioural
determinanats of health in the Third World,” Social Science and Medicine
36(2): 125-135
Calloway, D.H., S. Murphy, J. Balderstone, O.Receveur, D.Lein, and M.Hudes,
1992, Village Nutrition in Egypt, Kenya, and Maxico:Looking across the
CRSP Project, University of California, Berkeley, Mimeographed
Carlaw, R.W. and Vaidya K.., 1983, “Birth intervals and the survival od children to
age five:some data from Napal”, Journal o f Tropical Pediatrics 29(1):31-34
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Carlson B.A. and Tessa M. Wardlaw, 1990, “A global, regional and country
assessment of child malnutrition”, UNICEF Staff Working Paper No.7.
UNICEF, New York,
Chen, L.C., Huq, E. and D’Souza, S. 1981, “Sex bias in the family allocation of
food and health care in rural Bangladesh”, Population and Development
Review
Chen, L.C., Chowdhury, A.K.M.A., and Huffman, S.L., 1980, “Anthropometric
assessment o f enrtgy-protine malnutrition and subsequent risk o f mortality
among preschool aged children”, American Journal o f Clinical Nutrition
33:1836-1845
Cleland, J., 1990, “Maternal education and child survival: Future evidence and
explanation”, in John Caldwell, S. Findley, P.caldwell, W. Cosford, J. Braid
and D.B. Freeman (eds), What we know about Health Transition: the cultural,
social and behavioral determinants o f health, Health Transition Series No. 2,
Vol. ii, pp.400-419.
Cleland, J., Bicegom G. and Fegan, G., 1992, “Socioeconomic inequalities in
childhood mortality:the 1970s to 1980s”, Health Transition Review 2(1): 1-18
Cleland, J., and Van Ginneken, J. K., 1988, “Maternal education and child survival
in developing counties: The search for pathways of influence”, Social Science
and Medicine 27:1357-1368
Cleland, J, and Sathar Z.A., 1984, “The effect o f birth spacing on childhood
mortality in Pakistan”, Population Studies 38(3)401-418
Clemens J.D, Stanton B, Stoll B, Shahid NS, Banu H, Chowdhury AKMA., 1986,
“Breast feeding as a determinant of severity in shigellosis. Evidence for
protection throughout the first three years of life in Bangladeshi children”,
American Journal o f Epidemiology 123:710-20.
Cochrane, S. H., 1980, “The socio-economic determinants of mortality: the cross
national evidence”, World Development Report 1980
Cogill, B, 1982, Ranking anthropometric indicators using mortality in rural
Bangaladesh xhildren, Ithaca, New York:Comell University, Master of
Science Thesis.
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cohen, R.J., K.H. Brown, J. Canahuait, L.L. Rivera, and K.G. Dewey, 1994,
“Effect of age of introduction of complementary foods on infant breastmilk
intake, total energy intake, and growth: A rendomized intervention study in
Honduras”, Lancet 334(8918):288-293
Colombo M, I.de Andraca, I.Lopez, 1988, “Mental development and stunting”, In
Linear growth retardation in less developed countries, edited by J.C.
Waterlow, Nestle Nutrition Workshop Series, Vol. 14, Nestle Ltd.
Vevy/Raven Press, Ltd. New York
Curtis, S.L., I Diammond, and J.W., McDonald, 1993, “Birth interval and family
effects on postneonatal mortality in Brazil”, Demography 30(l):33-43
Curtis, S.L., and Fred Arnold, 1994, An evaluation of the Pakistan DHS Survey
based on the reinterview survey, DHS Occasional Paper No. I, Calverton,
MartlandrMacro International
Curtis, S.L., 1995, Assessment o f the quality of data used for direct estimation of
infant and child mortality in DHS-II surveys”, Demographic and Health
Survey, Occasion Paper No. 3, Calverton, Martland: Macro International
Cunningham AS, Jelliffe DB, Jelliffe EFP., 1991, “Breastfeeding and health in the
1980s: a global epidemiological review”, Journal o f pediatrics 118: 659-66.
Das Gupta, M., 1987, “Selective discrimination against female children in rural
Punjab, India”, Population and Development Review 13(1):77-100.
Das Gupta, M., 1990, “Death clustering, mother’s education and the determinants
of child mortality in rural Punjab, India”, Population Studies 4493):489-505
Da Vanzo, Ellen H. Starbird, 1991, “Correlates of Short Inter-birth Intervals in
Peninsular Malaysia: Their Pathways of Influence Through Breast-feeding and
Contraceptive Use (in Reports)”, Studies in Family Planning 22(4):241-254.
DaVanzo, J., Butz, W.P. and Habicht, J.P., 1983, “How biological and behavioural
influences on mortality in Malaysia vary during the first year of life”,
Population Studies 37(3):381-402.
DaVanzo, Julie, 1988, ‘Infant mortality and socioeconomic development: evidence
from Malaysian household data’, Demography 25(4):581-95
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DaVanzo, J. and J.P. Habicht, 1986, "Infant mortality decline in Malaysia, 1946-
1975: the roles of changes in variables and changes in the structure of
relationships", Demography 23(2): 143-160.
Defo, B.K., 1997, “Determinants of infant and early childhood mortality in
Cameroon:The role o f socioeconomic factors, housing characteristics, and
immunization status”, Social Biology 41(3-4): 181-211
Defo, B. K., 1992, “Mortality and attrition processes in longitudinal studies in
Africa:An appraisal of the iford surveys”, Population Studies 46:327-348
Desai, S., 1992, “Children at risk:the role of family structure in Latin America and
West Africa”, Population Development Review 18:689-717
Desai, S., 1993, ‘The impact od family size on children’s nutritional
status:Insights from a comparative perspective”, In Fertility, Family Size, and
Structure:Consequences for families and children, edicted by C.B.Llovd, The
Population Council, New York
Dettwyler K., 1987, “ Breastfeeding and weaning in Mali: cultural context and
hard data”, Social Science and Medicine 24:633-44.
Dewey K.G., 1998, “Cross-cultural patterns of growth and nutritional status of
breast-fed infants, American Journal o f Clinical Nutrition 67:10-17
D’Souza, S., and Chen L.C., 1980, “Sex differencials in mortality in rural
Bangladesh”, Population and Development Review, 6(2):257-270
D’Souza. S and Abbas Bhuiya, 1982, “Socioeconomic mortality differentials in a
rural area o f Bangladesh”, Population and Development Review 8(4):753-769.
D’Souza, R., 1997, “Housing and environmental factors and their effects on the
health of children in the slums of Karachi, Pakistan”, Journal o f Biosocial
Science, 29:271-281
Dibley, M.J., J.B. Goldsby, N.W. Staehling, and F.L. Trowbridge, 1987a,
“Development of normalized curves for the international growth reference:
Historical and technical considertaions”, American Journal o f Clinical
Nutrition 46(5):736-748
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dibley, M.J., M.W. Staehling, P. Neiburg, and F.L. Trowbridge, 1987,
“Interpretation of Z-score anthropometric indicators derived from the
international growth reference”, American Journal o f Clinical Nutrition
46(5):749-762
Dickin K.L., Brown K.H., Fagbule D., Adedoyin M., Gottelsoln J., Esrey S.A. and
Oni G.A., 1990, “Effect of dusrrhea on dietray intake o f infants and young
children in rural villages in Kwara State, Nigeria, Eurapian Journal o f Clinical
Nutrition 44:307-317
Dyson, T. and Moore, M., 1983, “On kinship structure, female autonomy, and
demographic behavior in India, Population and Development Review 9:35-60.
Elegbe, I.A., O.O. Ojoteitimi, I. Elegbe, and M.O.. Akinola, 1982, “Pathogenic
bacteria isolated from infant feeding teats. Contamination of teats used by
illiterate and educated nursing mothers in Ile-Ife, Nigeria”, American Journal
o f Diseases Child 136:672-674
EIo, I.T., and J.E.Miller, 1991, “Ever-breasrfed status and health at birth”, Paper
presented at the IUSSP Seminar on measurement o f metemal and child
mortality, morbidity, and health care: Interdisciplinary approaches
Esrey, S. A. and Habicht, J. P., and G.Casella, 1992, “The complementary effcet of
latrines and increased water usage on the growth of infants in rural Lesotho”,
American Journal o f Epidemiology 135:659-666
Esrey, S. A. and Habicht, J. P., 1988, “Maternal literacy modifies the effect of
toilet and piped water on infant survival in Malaysia”, American Journal of
Epidemiology 127(5): 1079-1087
Esrey, S. A. and Habicht, J. P., 1986, “Epidemiologic evidence for health benefits
from improved water and sanitation in developing countries”, Epidemiological
Review 8:117-128
Ewbank, D.C., and J.N. Gribble, 1993, Effects o f health programs on child
mortality in Sub-saharan Africa, Commission on behavioral and social
seciences education, National Research Council, Washington, D.C: National
Academy Press
Ewbank, D.C., 1994, “Maternal education and theories of behaviour, A cautionary
note”, Health Transition Review 4:215-223
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Farah, A.A. and Preston S.H., 1982, “Child mortality differentials in Sudan”,
Population Development Review 8(2):365-383
Feachem RG, Hoblinsky MA., 1984, "Interventions for the control of diarrhoeal
diseases among young children: promotion o f breast-feeding”, Bulliten o f
World Health Organization 62:271-91
Federal Bureau o f Statistics, 1995, Women and Men in Pakistan, Statistics
Division, Govememnt of Pakistan, Islamabad
Fauveau, V., M.A. Koeing and B. Wojtyniak, 1991, “Excess female deaths among
rural Bangladeshi children: an examination o f cause-specific mortality and
morbidity”, International Journal o f Epidemiology 20(3):729-735
Floud, R., K.Wachter and A. Gregory, 1990, “Height, health and history,
Nutritional status in the United Kingdom, 1750-1980", Cambridge University
Press
Forste, Renata, 1994, “The Effects of Breast-feeding and Birth Spacing on Infant
and Child Mortality in Bolivia”, Population Studies 48:497-511.
Freed, Ruth S.; Freed, Stanley A., 1989, “Beliefs and practices resulting in female
deaths and fewer females than males in India”, Population and Environment
10(3): 144-61
Gaspari, K.C. and A.Woolf, 1985, “Income, public works, and mortality in early
twentieth-century American cities”, Journal o f Economic History 45:355-361.
Gibson, R.S., 1990, Principles if nutritional assessment, New York:Oxford
Goldberg, H.I., W. Rodrigues, A.M.T. Thome, BJanowitz, and L. Morris, 1984,
“Infant mortality and breastfeeding in North-Eastern Brazil, Population
Studies 38:105-115
Goodbum E., GJ.Ebrahim, and Sushir Senapati, 1990, “Strategies educated
mothers use to ensure the health o f their children,” Journal od Tropical
Pediatrics 36:235-239
Gopalan C, 1992, “Undemutrition: measurement and implications”, in Nutrition
and Poverty (ed) Osmani SR, Oxford:Clarendon press, pp. 17-48
235
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gopalan C., 1987, “Growth monitoring: some basic issues”, Bulletin o f the
Nutrition Foundation o f India 8(2): 1-4
Gopalan C., 1986, “Stunting: Significance and implications for public health
policy”, in Linear growth retardation in less developed countries, edited by
John C. Waterlow, Nestle Nutrition Workshop Series, Vol. 14, Nestle Ltd.
Vevy/Raven Press, Ltd. New York
Gopalan, C., 1983, “Measurement o f undemutrition: biological considerations”,
Bulletin o f the Nutrition Foundation o f India 4(oct):4-8
Gorstein, J., Sullivan, K., Yip, R., de Onis, M., Trowbridge, F., Fajans, P., and
Clugston, G., 1994, “Issues in the assessment o f nutritional status using
anthropometry, Bulletin o f the World Health Organization 72:273-283
Government of Pakistan, 2000, “Economic Survey o f Pakistan, 1999-2000,
Planning Commission, Ministry o f Planning and Development, Islamabad
Government o f Pakistan, 1999, Population and development: Pakistan country
report ICPD+5, Ministry of Population Welfare, Islamabad
Gracey M, 1991, “Nutrition and physical growth, In Anthrometric assessment o f
nutritional status ” , edited by Himes J.H. New York:Wiley-Liss, pp.29-50
Graitcer,P.L., and E.M.Gentry, 1981, “Measuring children:One reference for all”,
Lancet 2:297-299
Grantham-Mcgregor, S., 1990, “Morbidity, nutritional deficiencies and child
development in developind countries”, In Diet and Disease, edited by G.A.
Harrison and J.C. Waterlow, New York:Cambridge University Press
Grant, J.P., 1994, “The State o f the World's Children - 1994”, New York: Oxford
University Press.
Grosse, Robert N., and Christopher Auffrey, 1989, “Literacy and Health Status in
Developing Countries”, Annual Review of Public Health 10:281-297
Gubhaju, B., 1984, “Demographic and social correlates o f infant and child
mortality in Nepal”, Ph.D. thesis, The Australian National University.
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gubhaju, B., 1985, “The effect of previous child death on infant and child
mortality in rural Nepal”, Journal o f Biosocial Science 16:231-236
Gubhaju, Bhakta B., 1986a, “Level and trends of infant and child mortality in
Nepal”, Demography India 15(l):46-60, Delhi, India.
Gubhaju, Bhakta B., 1986b, ‘Effect o f birth spacing on infant and child mortality
in rural Nepal’, Journal o f Biosocial Science\%(4):
Gubhaju, Bhakta B., Streatfleld Kim, Majumder, A. K., 1991, “Socioeconomic
demographic and enverionmental determinants of infant and child mortality in
Nepal”, Journal o f Biosocial Science 23(4)4:524-35
Gubhaju, Bhakta B., Choe, Minja Kim, Retherford, R.D., Thapa Shyam, 1987,
“Infant mortality trends and differentials in Nepal”, Studies in Family Planning
18(1):22-31
Guo, Guang, 1993, “Use o f sibling data to estimate family mortality effects in
Guatemala”, Demography 30(1): 15-32
Habtemariam, G. 1994, “Determinants of child health and survival in Central
Ethiopia”, Ph.D Thesis, London Schoold of Hygiene and Tropical Medicine.
Habicht J.P, DaVanzo J, Butz W.P., 1986, “ Does breastfeeding really save lives,
or are apparent benefits due to biases?” American Journal o f Epidemiology
123: 279-90.
Haggerty P.A., and S.O. Rutsteinm, 1999, “Breastfeeding and complementary
infant feeding, and the postpartum effects of breastfeeding”, DHS Comparative
studies No. 30.
Haines, M.R., and Avery, R.C., 1982, “Differential infany and child mortality in
Costa Rica, 1968-1973”, Population Studies 36(1):31-43
Hakim, Rubina, J.Thomas, S. Badruddin, 1998, “Urban-rural differences in the
nutrient intake o f Pakistani children”, Journal o f Pediatric Gastroenterology &
Nutrition 27(2):259
Halli, S.S. and K.V. Rao (1992), Advanced Techniques o f Population Analysis
(New York and London, Plenum Press).
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hobcraft, JN., 1993, “Women’s education, child welfare and childsurvivaha
review o f the evidence”, Health Transition Review, 3(2): 159-175
Hobcraft, JN., 1991, “Family formation patterns and child survival: A comparative
analysis”, Paper presented at the annual meeting o f the Population Association
o f America, 21-23 March, Washington, DC.
Hobcraft, JN., J. W. Mcdonald and S. O. Rutstein, 1985, “Demographic
determinants of infant and early child mortality: a comparative analysis”,
Population Studies 39:363-385
Hoberaft, J., J. W. Mcdonald and S. O. Rutstein, 1984 ‘Socio-economic factors in
infant and child mortality: A cross-national comparison’, Population Studies
38(2): 193-223.
Hobcraft, J., J.W. McDonald and S.Rutstein, 1983, “Child-spacing effcets on
infant and early child mortality”, Population Indx 49:585-618
Horton, S, 1988, “Birth order and child nutritional status:evidence from the
Philippines”, Economic Development and Cultural Change 36:341-354
Heywood, P., 1982, “The functional significance of malnutrition: Growth and
prospective risk of death in the highlands of Papua New Guinea”, Journal o f
Food & Nutrition 39(1): 13-19
Huffman, S.L. and Steel, A., 1995, “Do child survival interventions reduce
malnutrition? The dark side o f child survival”, in Child growth and nutrition in
developing countries: Priorities fo r action, edited by P.Pinstrup-Anderson, D.
Pelletier, and H. Alderman, Ithaca, New York:ComeIl University Press
Huffman, S.L., and C. Combest, 1990, “Role o f breastfeeing in the prevention and
treatment of diarrhea”, Journal o f Diarrheal Disease Resaerch 3:68-81
Huttly SRA, Blum D, Kirkwood BR, Emeh RN, Feachem RG,1987, “The
epidemiology of acute diarrhoea in a rural community in Imo state, Nigeria”,
Trans R Soc Trop Med Hyg 81:865-70.
Jeffery, R. and P. Jeffery, 1993, “Jats and Sikhs in Bijnor: education, women’s
autonomy and fertility outcomes”, Paper presented at workshop on Female
Education, Autonomy and Fertility Change in South Asia, Delhi
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Jelliffe D.B., Jelliffe E.FP.,1978, “Human milk in the modem world.Psychosocial,
Nutritional and Economic Significance”, Oxford University Press
Jelliffe E.P., 1976, “Maternal nutrition and lactation”, In Breastfeeding and the
mother, Ciba Foundation Symposium, Elsevier 45:119-145
Jelliffe D.B., 1969, “Child Nutrition in developing countries:A handbook for
fieldworkers”, Office of War on Hunger, Agency for International
Development, U.S. Department of State, Washington, D.C
Jelliffe D.B., 1966, The assessment o f the nutritional status o f a community, WHO
Monograph Series 53
Jelliffe D.B., 1962, “Culture, socialchange and infant feeding”, American Journal
o f Clinical Nutrition 10.
Jelliffe D.B., 1955, “Infant nutrition in the subtropics and tropics, WHO
monograph, No. 29, Geneva
Johansson, S.R.,1990, “Cultural software, institutional hardware, and health
information processing in social systems”, in What We Know about Health
Transition: The Cultural, Social and Behavioural Determinants o f Health,
edited by John C. Caldwell, Sally Findley, Pat Caldwell, Gigi Santow, Wendy
Cosford, Jennifer Braid and Daphne Broers-Freeman. Canberra: Health
Transition Centre, The Australian National University.
Johnson, G. Z., 1964, Health conditions in rural and urban areas of developing
countries”, Population Studies 17:293-309
Jason JM, Nieburg P, Marks JS., 1984, “Mortality and infectious disease
associated with infant-feeding practices in developing countries”, Pediatrics
74: 702-27.
Kasongo Project Team, 1983, “Anthropometric assessment o f young children’s
nutritional status as an indicator o f subsequent risk of dying”, Journal o f
Tropical Pediatrics 29:69-75
Kasongo Project Team, 1981, “Influence o f measles vaccination in survival pattern
o f 7-35 month-old children in Kasongo, Zaire”, Lancet, 4(l):764-767
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Karki, A. and R. Levine, 1993, “Maternal schooling and child health: preliminary
analysis on the intervening mechanisms from a community - level study in
rural Nepal” Unpublished manuscript.
Katz, J., and others, 1989, “The importance of age in evaluating anthropometric
indices for predicting mortality”, American Journal o f Epidimiology
130(6): 1219-1226
Keilmann, A.A. and McCord, C., 1978, “Weight-for-age as an index of risk of
death in children”, Lancet 1:1247-1250
Keller, W., 1991, “Stature and weight as indicators of undemutrition”, in
Anthropometric assessment of nutritional status, edited by J.Himes:l 13-122,
Wiley, New York
Kent, G., 1991, The politics o f children's survival, Praeger:New York, New York.
Kielmann, A.A., and C.McCord, 1978, “Weight for Age as an index o f risk of
death in children”, Lancet 1:1247-1250
Kimm, Sue, and G. Gonoso,1979, “Control of protein-calorie malnutrition”, In
Food and nutrition policy in a changing world, (eds.) Mayer, J., and Johanna
T. Dwyer, Oxford University Press, New York.
Klebanoff MA., 1988, “Short interpregnancy interval and the risk o f low
birthweight”, American Journal o f Public Health 78:667-670.
Koeing M.A., J.F.Phillips, O.M.Campbell, and S.D’Souza, 1990, “Birth intervals
and childhood mortality in Bangladesh”, Demography 27:251-266
Koeing M.A. and D’Souza S., 1986, “Sex differences in mortality in rural
Bangladesh”, Social science and Medicine 22(2): 15-22
Kundi, M.Z., M. Anjum, D. Mull and J. Mull, 1993, “Maternal perceptions of
pneumonia and pneumonia signs in Pakistani children”, Social Science and
Medicine 37(5):649-660.
Kusin J.A., Kardjati S., van Steenberger W.M. and Renqvist U.H., 1992, “Energy
supplementation during pregnancy and postnatal growth”, Lancet 340:623-
626.
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lantz, P., M. Partin and A. Palloni, 1992, ‘Using Retrospective surveys for
estimating the effects of breast-feeding and child spacing on infant and child
mortality”, Population Studies 46(1):121-139
Leowski, J. 1986. Mortality from acute respiratory infections in children under
five years. World Health Statistics Quarterly 39:138-144.
Leslie J, 1991, “Women’s nutritionrthe key to improving family health in
developing countries”, Health Policy Planning 6:1-19
LeVine, R., Emily Dexter, Patricia Velasco, Sara LeVine, Arun R. Joshi, Kathleen
W. Stuebing, and F. Medardo Tapia-Uribe, 1994, “ Maternal literacy and
health care in three countries: a preliminary report” Health Transition Review
4:186-191
Lindskog, U., P. Lindskog and S. Wall, 1987, “Water supply, sanitation and health
education programmes in developing countries: problems of evaluation”,
Scandinavian Journal of Social Medicine 15:123-130.
Lindskog, U., Lindskog, P., Carstensen, J., Larsson, Y., and Gebre-Medhin, M.,
1988, “Childhood mortality in relation to nutritional status and water supply: A
prospective study from rural malawi”, Acta Paediatrica Scandinavica 77:260-
268
Maclean V., 1974, “Magical medicine: a Nigerian case study”, London, Penguin
Mahmood, Muhammad A., 1992, “Socio-economic factors, personal illness control
and infant and child mortality in Pakistan”, Pakistan Population Review
3(l):67-86, Islamabad, Pakistan.
Majumder, K., and S.M. Islam and S. M. Shafiqul, 1993, “Socioeconomic and
Environmental determinants o f child survival in Bangladesh,” Journal o f
Biosocial Science, 25(3):311-318
Majumder, K., 1991, “Beast-feeding, birth interval and child mortality in
Bangladesh”, Journal o f Biosocial Sciences 23:297-312
Majumder, K., 1990, “Child survival and its effects on mortality of siblings in
Bangladesh”, Journal of Biosocial Science 22:333-347
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Majumder, K., 1989, “Factors affecting child survival in Bangladesh”,
unpublished Ph.D. dissertation, Australian Nationall University, Canberra
Makinson, Carolyn, 1986, “Sex differentials in infant and child mortality in
Egypt” University Microfilms International: Ann Arbor, Michigan
Malik, I, 1997, “State and civil society in Pakistan:Politics o f authority, ideology
and ethnicity", St. Martin’s Press, Inc
Martin, L., J. Trussell, F. R. Salvail and N.M.Shah, 1983, “Covariates of child
mortality in the Philippines, Indonesia, and Pakistan: An analysis based on
Hazard Model”, Population Studies 37(4):417-432.
Martorell, R., 1995, “Promoting health growth:Rationale and benefits” in Child
Growth and Nutrition in Developing Countries, edited by P. Pinstrup-
Andersen, D.Pelletier and H. Alderman, Ithaca:Comell University Press
Martorell R., Kettle Khan L., and Schroeduer D.G., 1994, “Reversibility of
stunting:epidemiological findimgs in children from developing countries”,
Eurpian Journal o f Clinical Nutrition 48:S45-S57
Martorell, R.,J.Rivera, H.Kaplowitz, and E.Pollitt, 1992, “Long-tem consequences
of growth retardation during early childhood”, in Human Growth:Basic and
Clinical Aspects, edited by M. hemanadez and J. Argente, New York:Eksevier
Science Publishers B.V.
Martorell, R.,J.Rivera, H.Kaplowitz, 1990, “Consequences o f stunting in early
childhood for adult body size in rural Guatemala”, Annales Nestle 48:85-92
Martorell, R. and J.P. Habicht, 1986, “Growth in early childhood in developing
countries”, in Human Growth: A comparative treatise, voluntary 3:241-262
edited by F. Falkner and J.N.Tanner, New Yoyk:Plenum Press
Martorell, R., 1985, “Child grwoth retardations discussion o f its causes and its
relationship to health”, In Ntritional Adaptation in Man, edited by K.L. Blazter
and J.C. Waterlow, London:John Libbey
Martorell, R., and T J ilo , 1984, “Malnutrition, Morbidity, and Mortality”, In
Child Survival:Strategies for Resaerch, edited by H.Mosley amd L.Chen,
Population and Development Review 10 (Supplement):49-68
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Martorell R., Yarbrough C., Yarbourgh S., and Klein R.E., 1981, “Maternal
stature, fertility and infant mortality”, Human Biology 53:303-312
Mata L.J., Urrutia, Albertazzi E, Pellecer O & Aerllano E, 1972, “Influence of
recurrent infections on nutrition and growth of children in Guatemala”,
American Journal o f Clinical Nutrition 25:1267-1275
Mata, L.J., R.A. Kromal, JJ. Urrutia, and B. Garcia, 1977, “Effect o f infection on
food intake and the nutritional state:Persoectives as viewed from the village”,
American Journal o f Clinical Nutrition 30(8): 1215-1227
Mata, L.J., 1978, The children o f Santa Maria Cauque, Cambridge, MA:MIT Press
Mata, L.J., 1980, “Child malnutrition and deprivation-observations in Guatemala
and Costa Rica”, Food Nutrition 6:7-14
McCance, David R.,Pettitt, David J., Hanson, Robert L., Jacobsson, Lennart T. H.,
Knowler, William C., Bennett, Peter H., 1994, “Birth Weight and Non-Insulin
Dependent Diabetes: Thrifty Genotype, Thrifty Phenotype, or Surviving Small
Baby Genotype”? The British Medical Journal 308(6934): 942-945
McDevitt, T. M., Adlakha, A., Fowler, T. B. and Vera Bourne, H. 1996, Trends in
adolescent fertility and contraceptive use in the developing world, Bureau of
Census, US Department o f Commerce.
Merrick, T.W, 1985, “The effect of piped water on early childhood mortality in
urban Brazil, 1970 to 1976”, Demography 22:1-23.
Mensch, B., H. Lentzner and S. Preston, 1985, “Socio-Economic Differences in
Child Mortality in Developing Countries " , New York: United Nations.
Michael R. Haines, Roger C. Avery, 1982, “Differential Infant and Child
Mortality in Costa Rica: 1968-1973”, Population Studies 36(1):31-43.
Michaelsen KF., 1988, “Value o f prolonged breastfeeding”, Lancet ii:788-779.
Miller, B.D. 1981, “The Endangered Sex”, Cornell University Press, Ithaca.
Miller JE., 1991, “Birth intervals and perinatal health: an investigation of three
hypotheses ”, Family Planning Perspectives 23:62-70
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Miller, J.,James Trussell, Anne R. Pebley, Barbara Vaughan, 1992, “Birth Spacing
and Child Mortality in Bangladesh and the Philippines (in Mortality)”,
Demography 29(2):305-318.
Mitra, S.N., A. Al-Sabir, A.R. Cross and K. Jamil, 1997, “Bangladesh
Demographic and Health Survey 1996-1997 Main Report” (Dhaka: National
Institute of Population Research and Training (NIPORT),Mitra and Associates,
Dhaka, and Macro International, Inc., Maryland, United States).
Mosley, W.and Lincoln C. Chen, 1984, “nalytical framework for the study of child
survival in developing countries”, Population Development Review, A
supplement to Vol. 10:25-45.
Mosley, W. and P.Cowley, 1991, “The Challenge of World Health”, Population
Bulletin 46(4), A publication of the Population Refrerence Bureau, Inc
Molbak, K, A. Gottschau, P. Abely, N. Hojlyng, L. Ingsholt, and A.P. J. Silva,
1994, “Prolonged breastfeeding, diarrhoea! disease, and survival of children in
Guinea-Bissau”, British Medical Journal 308:1403-1406
Molbak K., Aaby P, Ingholt, Hojlyng N., Gottschau A., Andersen H., 1992,
“Persistent and acute diarrhoea as the leading causes o f child mortality in
urban Guinea Bissau”, Trans R Soc Trop Med Hyg 86:216-20.
Molbak K, Hojlyng N, Ingholt L, da Silva APJ, Jepsen S, Aaby P., 1990, “An
epidemic outbreak of cryptosporidiosis: a prospective community study from
Guinea Bissau”, Pediatric Infectious Disease Journal 9:566-70.
Moy R.J.D., Marshall T.F., Choto R.G.A., Mcneish A.S. and Booth J.W., 1994,
“Diarrhea and growth falteming in rural Zambabwe”, Eurpian Journal of
Clinical Nutrition 48:810-821
Mozumder, A.B.M.K., Barkat-e-Khuda, T.T.Hane A. Levin and S. Ahmad, 2000,
“The effect o f birth interval on malnutrition in Bangladesh infants and young
childre”, Journal o f Biosocial Science 32:289-300
Muhuri, P, 1995, “Health Programs, Maternal Education, and Differential Child
Mortality in Matab, Bangladesh”, Population and Development Review
2 1(4):813-834
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mull, J.Dennis and Dorothy S. Mull, 1988, “Mothers’ concepts of childhood
diarrhea in rural PakistanzWhat ORT program planners should know”, Social
Science and Medicine 27(l):53-67
Murray CJL, Lopez AD., 1996, “The global burden of disease: a comprehensive
assessment o f mortality and disability from diseases, injuries and risk factors
in 1990 and projected to 2020”, Global Burden o f Disease and Injury Series,
vol I. Cambridge: Harvard School of Public Health, 1996.
National Academy Press, 1992, “Nutritional issues in developing countries,
Subcommitte on nutrition and diarrheal diseases control, Subcommitte on diet,
physical activity. And pregnancy outcome, subcommitte on international
nutrition programs food and nutrition board, Institute of Medicine,
Washington D.C.
National Institute of Population Studies (NIPS), 1992, "Pakistan demographic and
health Survey 1990-91", Institute of Research Development (IRD)/ Macro
International, Columbia, MD.
Neumann, C.G. and Harrison, G.G., 1994, “Onset and evolution of stunting in
infants and childrenzexamples from the human nutrition collaborative research
support program, Kenya and Egypt studies”, European Journal o f Clinical
Nutrition 48(Suppl 1):S90-S 102
Orubuloye, I.O. and John C. Caldwell, 1975, “The impact o f public health services
on mortality: a study of mortality differentials in a rural area in Nigeria”,
Population Studies 29(2):259-272.
Osmani, S.R., 1999, “Famine, demography, and endemic poverty”, In Population
and poverty in the developing countries, (eds.) Livi-Bacci, M. and Gustava DE
Santa, Clarendon Press, Oxford
Osmani, S.R, 1992, “On some controversies in the measurement of
undemutrition”, in Nutrition and Poverty, (ed.) Osmani SR, Oxford:Clarendon
press pp. 121.
O’Toole, J. and Robert E. Wright, 1991, “Parental education and child mortality in
Brundi”, Journal o f Biosocial Sciences 23(3):255-262
Palloni, A. and Tienda, M, 1986, “The effects of breastfeeding and pace of child
bearing on mortality at early ages”, Demography 23:31-52
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Palloni, A., G. Kephart, 1989, “The Effects o f Breast-feeding and Contraception
on the Natural Rate o f Increase: Are there Compensating Effects”? Population
Studies 43(3):455-478.
Pakistan Medical Research Council, 1998, National health survey of
Pakistan:Health profile of the people of Pakistan, Islamabad
Pandey, M. R., J.S. Boleij, K.R. Smith and E.M. Wafula, 1989, “Indoor air
pollution in developing countries and acute respiratory infection in children”,
Lancet 1:427-428.
Patel, M. 1980, “Effects o f the health and environmental factors on infant
mortality: The case of Sri Lanka”, Journal o f Epidemiology and Enviormenatl
Health 34:76-82
Payne, P. 1987, “Malnutrition and human capitakProblems o f theory and
practice”, In Poverty, Development and Food, edited by E. Clay and J. Shaw,
London: Macmillan
Pebley, A.R, and S.Millman, 1986,“Birthspacing and child survival”, International
Family Planning Perspectives 12(3): 71-9
Pelletier, D.L., 1999, “Nutritional status:The master key to child survival”, In
Scaling up, Scaling down:Overcoming malnutrition in developing countries,
(eds.) Marchione T.J.,Gordon and Breach Publishers
Pelletier, D.L.,1998, “Malnutrition, morbidity and child mortality in developing
countries”, Too young to die: Genes or gender? Economic and Social Affairs
United Nations
Pan American Health Organization, 1991, “Measles elimination in the Americas”,
Bulleetin o f the Pan-American Health Organization 26(3):271-274
Pelletier, D.L., and L.A. Msukwa, 1991, ‘T he use of national sample surveys for
nutritional surveiIlance:Lessons from Malawi’s National Sample Survey of
Agriculture”, Social Science and Medicine 32(8):887-898
Pelletier, D.L., E.A. Frongillo and J-P. Habicht, 1993, “Epidemiologic evidence
for a potentiating effect of malnutrition on child mortality”, American Journal
o f Public Health 83(8): 1130-1133
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Pelletier, D.L., 1994a, “The relatioship between child anthropometry and mortality
in developing countries: implications for policy, programs and future
research", American Journal o f Nutrition Supplement 124(10S):2047S-2081S.
Pelletier, D.L., 1994b, “A methodology for estimating the contribution of
malnutrition to child mortality in developing countries”, Journal o f Nutrition
Supplement 124( 10S):2082S-2105S.
Pelletier, D.L., 1994c, “Child anthropometry and mortality in Malawi:testing for
effect modification by age and length of follow-up and confounding by
socioeconomic factors”, Journal o f Nutrition Supplement 124( 1 0S):2106S-
2122S.
Pelletier, D.L., 1998, “Malnutrition, marbidity and child mortality in developing
countries”, In Too young to die:genes or gender? Department of Economic &
Social Affairs, Population Division, United Nations
Pelletier, D.L., 1999, ‘The master key to child survival”, in Scaling up, scaling
down:overcomming malnutrition in developing countries, edited by
TJ.Marchione, Gordon and Beech Pubkishers
Perez-Escamilla, R., S. Seguro-Millan, J. Canahuati, and H. Allen, 1996,
“Prelacteal feeds are nagatively associated with breastfeeding outcomes in
Honduras”, Journal o f Nutrition 126(11):2765-73
Phillips, David I. W. PhD, FRCP, 1998, “Birth Weight and the Future
Development o f Diabetes: A review of the evidence”, Diabetes Care 21(2S)
Supplement
Phillips, L., S.K. Lwanga, W. Lore and D. Wassawa, 1969, “Methods and hygiene
o f infant feeding in an urban area o f Uganda”, Journal o f Tropical Pediatrics,
15:167-171
Piwoz, E.G., Lopez de Romana, G., Creed de Kanashiro, K., B;ack, R.E., and
Brown, K..H., 1994, “Indicators for monitoring the growth of Reruvian
infants:weight and length gain vs attained weight and length”, American
Journal o f Public Health 84:1132-1138
Pollitt E., 1992, “Malnutrition and infection in the classroom”, Belgium:UNISCO
247
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Poppel, F. V. and Heijden, C. V. D., 1997, “The effects o f water supply on infant
and childhood martality: A review o f historical evidence”, Health Transitions
Review 7:1-113
Population Reference Bureau, 1997, “Population Food, and Nutrition”, Population
Bulletin 51(4), A publication of the Population Reference Bureau, Inc
Population Reference Bureau, 1999, “ 1999 World Data Sheet”, Washington, D.C.
Population Reference Bureau, 2000, “2000 World Data Sheet”, Washington, D.C
Potter, J.E., 1988, “Birth spacing and child survival: A cautionary note regarding
the evidence from the WFS”, Pupulation Studies 42:443-450
Preston, S.H., 1985, Resources. Knowledge and child mortality: A comparison of
the U.S. in the late ninteenth century and developing countries today, In
International Population Conference, Florence 1985, Vo;.4, Intervational
Union for the Scientific Study o f Population, Liege, 373-388
Preston, Samual and Alberto Palloni, 1977, “Fine-tuning Brass-type mortality
estimates with data on ages o f surviving children”, Population Bulletin o f the
United Nations, No. 10, pp.72-87,(Sales No. E.XIII.6),
Preston, S. H. and Van De Walle, E, 1978, “Urban French mortality in the
nineteenth century”, Population Studies 32:275-297
Puffer, R.C., and C.V. Serrano, 1973, “Patterns of Mortality in childhood”,
Scientific Publication, No. 262, Washington, D.C.:Pan American Health
Organization.
Rahman, Mizanur (1998). "The effect of child mortality on fertility regulation in
rural Bangladesh" Studies in Family Planning 29(3).
Rahman, M., Rahaman M., Wojtyniak, B., Aziz K.M., 1985, “Impact of
environmental sanitation and crowding on infant mortality in rural
Bangladesh”, Lancet 11:28-31
Retherford, Robert D. and Minja Kim Choe, 1993, “Statistical Medels fo r Causal
Analysis", New York:John Wiley & Sons.Inc.
248
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Retherford, Roberr D., Minja Kim Choe, Shyam Thapa, and Bhakta B. Gubhaju.
1989, ‘T o what extend does breastfeeding explain birth-inmterval effects of
early childhood mortality?” Demography 26:439-50
Rees, Jane M; Lederman, Sally A; Kiely, John L , 1999, “Neonatal Mortality
among High Birth Weight Infants Bom to African American Adolescent
Mothers”, Pediatric Research 45(4)
Rivera J. and Ruel M.T., 1997, “Growth retardation in the first three months of life
among rural Guatemalan children”, Eurpian Journal o f Clinical Nutrition 51
Rona R.J., S.Chinn, 1995, “Genetic and environmental influences on growth”,
Journal o f Medical Screening 2:133-139
Rowland, M.G.M., S,G.J.G.Rpw!and, and .J.Cole, 1988, “Impact o f infection on
the growth o f children from 0 to 2 years in an urban West African
community”, American Journal o f Clinical Nutrition 47(1):134-138
Rowland, M.G.M., Goh, S.G.J., Tulloch, S., Dunn, T.,and Hayes, R.J., 1985,
“Growth and weaning in urban Gambian children”, In ChildHealth in the
Tropics, edited by R.E.Eeckels, O. Ransome-Kuti, and C.C. Kroonenberg, The
Hague: Martinus Nijhoff Publishers
Rowland, M.G.M., R.Barrel, and R.Whitehead, 1978, “Bacterial contamination in
traditional Gambian weanig foods”, Lancet 1:136-138
Rowland, M.G.M., TJ.Cole, Whitehead R.G., 1977, “A quantitative study into the
role of infection in determining nutritional status in Gambian village children”,
British Journal o f Nutrition 37:441-450
Rutstein, Shea Oscar, 1984, Infant and Child Mortality: levels trends and
demographic differentials, Revised edition. WFS comparative studies No. 43.
Voorburg, Netherland: International Statistical Institute.
Rutstein, Shea Oscar, 1983, Infant and Child Mortality: levels trends and
demographic differentials, Revised edition. WFS comparative studies No. 24.
Voorburg, Netherland: International Statistical Institute.
Ruzicka, L. T., 1989, “Problems and issues in the study of mortality differentials”,
in Differentials Mortality: Methodological issues and biosocial factors, edited
by Ruzicka, G. Wunsch and P.Kane, Oxford: Clarendon Press, pp.3-17
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SAS Institute Inc. (1990). SAS/STAT User’ s Guide, Version 6, Cary, NC, United
States
Sachdev, H.P.S., J. Krishna, R.K. Puri, L. Satyanarayana, and S.Kumar, 1991,
“Water suplementation in exclusively breastfed infants during summer in the
tropics, Lancet, 337(8747):929-933
Sandiford P., Cassel J., Montenegro M. and Sanchez G., 1995, “The impact of
women’s literacy on child health and its interaction with access to health
services”, Population Studies 49(l):5-17
Sathar, Z. A., 1985, “Infant and Child Mortality in Pakistan, some trends and
differentials”, Journal of Biosocial Science 17(3):351-359
Sathar, Z.A., 1987a, “Seeking explanations for high levels of infant mortality in
Pakistan,”, Pakistan Development Review 26(1), Islamabad.
Sathar, Z.A., 1987b, “Sex differentials in mortality: a corollary of son
preference?” Pakistan Development Review 26(4):555-68
Sathar, Z.A, 1988, “Birth Spacing in Pakistan, Journal o f Biosocial Science
20(2): 175-94
Satyanarayana, K., A.N. Naidu, M.C. Swaminathan, and B.S.N. Rao, 1980,
“Adolscent growth among rural indians goys in relation to their nutritional
status in early childhood”, Annals o f Human Biology 7(4):359-366
Schultz, P.T., 1984, “Studying the impactof household economic and community
variables on child mortality”, In Child Survival: Strategies for research,
Population Development Review, Supplement to volume 10, edited by W.
Henry Mosley and L.c. Chen
Sen, A. and Sengupta, S., 1983, “Malnutrition o f rural children and the sex bias”,
Economic and Political Weekly 18:855-864.
Shah, M A., M.A.Yousafzai, M.H.Rahber, 1998, “Prevalence and determinants of
diarrhoea among rural Pakistani children o f less than three”, Journal o f
Pediatric Gastroenterology & Nutrition. 27(2):246
250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Shah, N.M., 1986, Pakistani Women:A socioeconomic and demographic profile,
Pakistan Institute of Development Economics, Islamabad and East-West
Population Institute, Honolulu
Sigman M., Neumann C., Baksh M., Bwibo N. and McDonald M.A., 1989,
“Relationship between nutrition and development in Kenyan toddlers”,
Journal Pediaritrics 115:357-364
Smedman, L., Sterky, G., Mellander, L., and Wall, S., 1987, “Anthropometry and
sunsequent mortality in Guinea-Bissau”, American Journal o f Clinical
Nutrition 46:369-373
Smith, David P., 1980, “Life Table Analysis”, World Fertility Survey, Technical
Bulletins, No.6 / Tech. 136, International Statistical Institute, Voorburg,
Netherlands.
Sommer, A. and Loewenstein, M.S., 1975, “Nutritional status and mortality: A
prospective validation o f the QUAC stick”, American Journal o f Clinical
Nutrition 28:287-292
Sommerfelt, A.E., and J.T. Boerma, 1993, ' " ‘ Anthromopetric status o f young
children in DHS surveys: An assessment o f data quality, In am assessment of
the Quality o f the Health data in DHS-l Surveys, DHS Methodological
Reports, No. 2., Calverton, Matyland: Macro Intemationl
Sommerfelt, A.E., and Steward, M.K., 1994, “Children’ s Nutritional Status,
Demographic and Health Surveys Comparative Studies No. 12, Macro
Intemation, Inc. Maryland.
Sommerfelt, A.E., and F. Arnold, 1998, “Sex differencials in the nutritional status
of young children”, In TooYoung to Die.Genes or Gender? Edited by
Economic and Social Affairs, United Nations:New York
Spurr, G.B., 1990, ‘The impact of chronic undemutrition on physical work
capacity and daily energy intake”, In Diet and Disease, edited by G.A.
Harrison and J.C. Waterlow, New York:Cambridge University Press
Spurr, G.B., M.G. Maksud, and Barac-Nieto, 1977, “Energy expediture,
productivily, and physical work capacity o f sugar cane loaders”, American
Journal o f Clinical Nutrition 30:1740-1746
251
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Strauss, J, 1990, “Households, communities and preschoold children’s nutritional
status:evidence from rural Cote d’Ivoire”, Economic Development and
Cultural Change 38:231-262
Sullivan, J.M., 1972, “Models for the estimation o f the probability o f dying
between birth and exact ages of early childhood”, Population Studies 26(1):79-
97
Sullivan, Jeremiah M., George Bicego, and Shea Oscar Rutstein, 1990,
“Assessment of the Quality o f Data Used for the Direct Estimation o f Infant
and Child Mortality in the Demographic and Health Surveys An Assessment o f
DHS-l data Quality, DHS Methodological Reports No. I.Columbia, Maryland:
Sukhatme, P.V. and S. Margen, 1982, “Autoregulatory homeostatic nature of
energy balance”, American Journal o f Clinical Nutrition 35:355-65
Tanner, J.M., 1982, “The potential of auxological data measuring economic and
social well-being”, Social Science History 6:571-581
Tawiah, E.O., 1989, “Child mortality differentials in Ghana: A priliminary report,”
Journal o f Biosocial Science 21:349-355
Timaeus, I. M. and Lish, L. 1995, “Intra-urban differentials in child health”,
Health Transitions Review 5:163-190
Tinger, Anne G., 1998, Improving Women Health in Pakistan, Human
Development Network,: Health, Nutrition, and Population Series, The World
Bank, Washington D. C.
The Cebu Study Team, 1992, “Undrelying and proximate determinants of child
health: The Cebu longitudinal health and nutritional study” American Journal
o f Epidemiology 133(2): 185-201
Togoe, E., 1995, “Maternal education and infant and child morbidity in Ghana:
The case o f diarrhea. Evidence from the Ghana DHS”, In: Women's Position
and demographic change in Sub-Saharan Africa, pp. 169-200. Edited by
Makinwa and A. M. Jensen.
Tomkins A, 1988, “The risk of morbidity in a sttunted child”, In Linear growth
retardation in less developed countries, edited by J.C. Waterlow, Nestle
Nutrition Workshop Series, Vol. 14, Nestle Ltd. Vevy/Raven Press, Ltd. New
York
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Trussell T.J.,1975,”A re-estimation o f the multiplying factors for the Brass
technique for determining childhood survivorship rates”, Population Studies
29(1):97-108
Trussell, J.,Anne R. Pebley, 1984, ‘The Potential Impact of Changes in Fertility on
Infant, Child and Maternal Mortality’, Studies in Family Planning 15(6):267-
280.
Trussell. J. and S.H. Preston, 1982, “Estimating the covariates of childhood
mortality from retrospective reports o f mothers”, Health Policy and Education
3:1-36
Trussell, J., L. Martin, R. Feldman, J. Palmore, M. Cenception and D. Abu Bakar,
1985, “Determinants o f birth-interval length in the Philippines”, Malaysia, and
Indonesia: A hazard model analysis”, Demography 22(2): 145-168.
UNESCO, 1999, “The Education For All (EFO) 2000 assesment Pakistan country
report”, http://w w w 2.unesco o rg/efa/w ef/eountryrepnrts/pnlci<;fan/contents html
UNFPA, 1998, “The state o f the World’ s children", New York:Oxford University
Press
UNFPA, 1997, “The state o f the World’ s children", New York:Oxford University
Press
UNICEF, 1997, “Water, Environment and Sanitation (WES): Guidelines for
implementing WES strategies.an Overview", Water, Environment and
Sanitation Section of UNICEF.
UNICEF, 1996, “The State o f the World's Children", New York: Oxford
University Press
UNICEF, 1997, “The State o f the World’ s Children", New York: Oxford
University Press
UNICEF, 1998, “The State o f the World's Children", New York: Oxford
University Press
UNICEF, 1990, “Strategy for improved nutrition o f children and women in
developing countries”, A UNICEF Policy Review, New York
253
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
United Nations, 1992, Second Report on the World Nutrition Situation: Global and
regional results
United Nations, 1985, “Socio-economic differencials in child mortality in
developing countries”, Department of International Economic and Social
Affairs, New York.
United Nations, 1986, “How to Weigh and Measure Children: Assessing the
Nutritional Status o f Young Children in Household Surveys, Sales
NO.E.88.IV.2
Van Den Broeck, J., Eeckels, R., and Uylsteke, J., 1993, “Influence of nutritional
status on child mortality in rural Z aif’, Lancet 341:1491-1495
Van Norren B. and Van Vianen H.A.W., 1986, “The Malnutrition-infection
syndrome and its demographic outcome in developing countries: A new model
and its application”, Programmimg committee for demographic research.
Publication No. 4, The Hague, Netherlands.
Van Norren B., Boerma and Sempebwa, 1989, “Simplifying the evaluation of
primary health care programmes”, Social Science and Medicine 28:1091-1097
Van Ginneken, J. K., 1990, “Behavioural factors affecting transmission and
treatment o f acute respiratory infections”, in What We Know about Health
Transition: The Cultural, Social and Behavioural Determinants o f Health, ed.
John C. Caldwell, Sally Findley, Pat Caldwell, Gigi Santow, Wendy Cosford,
Jennifer Braid and Daphne Broers-Freeman. Canberra: Health Transition
Centre, The Australian National University.
Vella, V., Tomkins A., Borghesi A., Migliori G.B., Ndiku J. and Adriko B.C.,
1993, “Anthropometry amd childhood mortality in northwest and southwest
Uganda, American Journal o f Public Health 83:1616-1618
Vella,V., Tomkins A., Ndiku, J., Marshal T., and Cortinovis, I., 1994,
“Anthropometry as a predictor for mortality among Ugendan children,
allowing for socio-economic variables”, European Journal o f Clinical
Nutrition 48(3): 189
254
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Victoria CG., 1996, “Infection and disease: the impact of early weaning”, Food
Nutrition Bulliton 17:390-96.
Victora, C.G., J., P.. Vaughan, C.Lombardi, S.M.Fuchs, L.P. Gigante, P.G. Smith,
L.C. Nobre, A.M. Teixeira, L.B. Moreira, and F.C. Barros, 1987, “Evidence
for protection by breastfeeding against infant deaths from infectious diseases
in Brazil”, Lancet 2(8554):319-322
Victoria CG, Vaughan JP, Martines JC, Barcelos LB., 1984, “Is prolonged
breastfeeding associated with malnutrition”, American Journal o f Clinical
Nutrition 39:307-14.
Villa,V. and others, 1992, “Determinants of child mortality in south-west
Uganda”, Journal o f Biosocial Science 24:103-112
Visaria L., J. Simons, 1997, “Mother’s education and child survival: An
overview”, in Maternal Education and Child Survival:Pathways and Evidence,
ed. L.Visaria, J.Simons and P.Berman. Ahmedabad:Gujrat Institute of
Development Research.
Waldron, Ingrid, 1986, “What do we know about causes of sex differences in
mortality? A review o f the literature”, Population Bulletin o f the United
Nations 18:59-76
Ware, Halen, 1984, “Effects of maternal education, women’s roles and child care
on child mortality”, Population and Development Review, A supplement to
Vol. 10, pp. 191-214.
Waterlow J.C., 1972,“Classification and definition of protein-calorie malnutrition”,
British Medical Journal 3:566-569
Waterlow, J.C., R.Buzina, W.Keller, J.M.Lane, M.Z.Nichaman, and J.M.Tanner,
1977, ‘T he Presentataion and use o f height and weight data for comparing the
national status o f groups of children under the age o f 10 years, Bullettin o f the
World Health Organization 55(4):489-498
Waterlow J.C, 1988a, “Linear growth retardation in less developed countries”,
Nestle Nutrition Workshop Series no. 14. Vevey: Nestle Nutrition/New York:
Raven Press.
255
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Waterlow J.C, 1988b, “Observations on the natural history o f stunting”, in Linear
growth retardation in less developed countries”, ed. JC Waterlow, pp. 1-16.
Vevey: Nestle Nutrition/New York: Raven Press.
Waterlow J.C., 1992, Nutrition and Growth”, in Protein energy malnutrition with
contributions edited by A.M. Tomkins and S.M. Grantham-McGregor, Edward
Arnold, a division of Hodder and Stoughton, London
Waterlow, J.S, 1993, “Reflections on Stunting”, In recent trebds in Nutrition”,
edited by C.Gopalan, Delhi‘ .Oxford University Press
Wells J.C.K., Davies P.S.W. and Lee P.C., 1993, ’’Patrens of infants weight gain in
developing countries”, Journal o f o f Tropical Pediatrics 39:214-218
Winikoff, B., 1982, “Weaning:nutrition, morbidity and mortality consequences”,
in Biological and SocialAspects of Mortality and the Length of Life, pp. 113-
150, edited by S.Oreston, Ordina Editions, Liege
Winikoff, Beverly, 1983, “The effects of birth spacing on child and maternal
health”, Studies in Family Planning 14(10):231-245.
Winikoff, Beverly, Castle, M.A., 1987, “The maternal depletion
syndrome:Clinical diagnosis or eco-demographic condition?”, Paper presented
at the International Conference for better health for women and children
through family planning, Nairobi, Kenya
Winkvist A, Rasmussen KM, Habicht J-P., 1992, “A new definition of maternal
depletion syndrome”, American Journal o f Public Health 82:691-694.
Woldemicael, G.,2000, “The effects of water supply and sanitation on childhood
mortality in urban Eritrea”, Journal of Biosocial Science, 32:207-227
World Bank, 2000, "World Development Indicators 2000" Stock no. 14553 (ISBN
0-8213-4553-2)
World Bank, 1998, “Pakistan towards a health sector strategy”, Washington, D.C
(Report No. 16695-PAK)
WHO, 1995, '"'Physical status:the use and interpretation o f anthropometry”,
World Health Organization.
256
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WHO Working Group, 1986, “Use and interpretation o f anthropometric indicators
o f nutritional status”, Bulletin o f the World Health Organization 64(6) :929-
941.
WHO/UNICEF, 1990, “Meeting on breastfeeding in the 1990s: A global initiative,
Innocentive dectaration on the protection, promotion, and support of
breastfeeding, Florence, Italy, August 1, New York
WHO, 1997, “Tuberclusis prevention and control:Country profile:Pakistan”
Annual Report 1997.
WHO, 1991, Indicators for assessing breast-feeding practices, Report of an
informal meeting, 11-12 June, 1991, Geneva:WHO
WHO, 1983, Measuring change in nutritional status. Annex 3. Reference data for
the weight and height of children. Geneva: WHO.
WHO, 1987, Global surveillance through anthropometric measurements. WHO
Weekly Epidemiological Record, March 1987, 1-20.
Wray, J.D., 1978, “Direct nutrition intervention and the control of diarrheal
diseases in pre-school children”, American Journal o f Clinical Nutrition
31:2073-2082
Yamaguchi, K., 1991, Event History Analysis, Newbury Park: Sage.
Yambi, O., 1988, Nutritional status and the risk o f death: a prospective study o f
children six to thirty months old in Iringa region, United Republic o f Tanzania,
Ithaca, New York:ComelI University, Doctoral Thesis.
Yerushaimy et al., 1965, “Births and gestation as indices of Immaturity”,
American Journal o f Children Diseases 109:43-57
Yoon, Paula W., R.E.Black, L.H. Moulton, and S. Becker, 1996, “Effect of not
breastfeeding on th erisk of diarrhea and respiratory mortality in children under
2 years of age in Metro Cebu, The Philippines,” American Journal o f
Epidemiology 143(11): 1142-1148
257
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Zeitlin, M., H. Ghassemi and M. Mansour, 1990, Positive deviance in child
nutrition, with emphasis on psychosociol and behavioral aspects and
implications for development, Tokyo:United Nations University
Zeitlyn, S. and F. Islam. 1993, “Mother’s education, autonomy and innovation”,,
in Maternal Education and Child Survival .Pathways and Evidence, ed. L.
Visaria, J.Simons and P.Berman. Ahmedabad:Gujrat Institute of Development
Research.
Zheng, Jianbin, 1993, “Covariates of infant and early childhood mortality in
Thailand: A hazards model analysis o f the DHS data with policy implecations
(Infant mortality)”, unpublished Ph.D dissertation, University of Oregon.
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

HIV/AIDS preventive behavior in Botswana: Trends and determinants at the turn of the 21st century

PDF

"Relative strangers": An ethnography of grandmothers raising grandchildren in the Los Angeles County welfare system

PDF

Gambling with public health: How government officials brought Los Angeles County to the brink of disaster

PDF

An exploration of nonresponse with multiple imputation in the Television, School, and Family Project

PDF

Changing patterns of interracial marriages of Asian Indians in California

PDF

An analysis of nonresponse in a sample of Americans 70 years of age and older in the longitudinal study on aging 1984-1990

PDF

Interbirth interval analysis of fertility: The case of Kenya in 1993

PDF

Impact of housing arrangements on social support and health status among Chinese American elderly

PDF

Assessment of fatigue as a late effect of therapy among survivors of childhood leukemia

PDF

Heart disease among middle -aged and elderly persons in the United States: Trends and a multistate model

PDF

Effect of genetic factors in the development of childhood lymphocytic leukemia (ALL)

PDF

Influence of body weight and weight change on carotid IMT and IMT progression: A pooled analysis using subject-level data from four clinical trials

PDF

Cross -cultural examination of mental health measures in dementia caregivers: Assessment of the Center for Epidemiologic Studies -Depression Scale and the Zarit Burden Inventory

PDF

Familiality and environmental risk factors of peptic ulcer: A twin study

PDF

Gender-biased education investment intention among Chinese rural parents: A glimpse into the intra-household resource allocation process

PDF

Immigrants and housing attainment: Different forms of residential assimilation

PDF

Comparative research of groundwater issues in Mexico City and Los Angeles: Some possible solutions

PDF

Dementia caregiving and ethnicity: African American caregivers and the sociocultural stress and coping model

PDF

Health interventions, health status, and labor market outcomes in Indonesia: Three essays

PDF

In this nature: Culture, place, and attitudes toward marine animals

Asset Metadata

Creator Mahmood, Muhammad Arshad (author)

Core Title Factors affecting child survival in Pakistan

School Graduate School

Degree Doctor of Philosophy

Degree Program Sociology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag health sciences, nutrition,health sciences, public health,OAI-PMH Harvest,sociology, demography,sociology, public and social welfare

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Heer, David (committee chair), James, Angela (committee member), Mack, Wendy (committee member), Silverstein, Merril (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-131235

Unique identifier UC11329226

Identifier 3041495.pdf (filename),usctheses-c16-131235 (legacy record id)

Legacy Identifier 3041495-0.pdf

Dmrecord 131235

Document Type Dissertation

Rights Mahmood, Muhammad Arshad

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA