The determinants of child growth retardation in India

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 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. T H E D E T E R M IN A N T S OF C H IL D G R O W T H R E T A R D A T IO N IN IN D IA b y Joy Inasu Cheenath A D issertation Presented to the FA C U L T Y O F T H E G R A D U A T E SC H O O L U N IV ER SIT Y OF SO U TH ER N C A L IFO R N IA In Partial Fulfillm ent o f the Requirem ents for the D egree D O C T O R OF PH IL O SO PH Y (Political E conom y and Public Policy) A ugust 2000 C opyright 2000 Joy Inasu C heenath Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3054854 ® UMI UMI Microform 3054854 Copyright 2002 by ProQuest Information and Learn; ing Company. All rights reserved. This microform edition is prolected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Cormpany 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 Joy Inasu Cheenath under the direction of hXs. Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies D a te August..7 . „.2Q Q Q DISSERTATION COMMITTEE Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. my parents and Rose, Alb in and Deepak Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A C K N O W L E D G E M E N T S I have been able to com plete this research w ith the substantial help o f m any agencies and individuals. The G overnm ent o f India generously permitted and enabled m e to undertake this study. The W orld B ank G raduate Scholarship Program financed the first tw o years o f m y study. D uring the follow ing tw o years I w as financed by the research assistantship provided by Prof. Richard A. Easterlin and the teaching assistantship provided by the E conom ics departm ent o f the U niversity o f Southern California. A D issertation Fellow ship from the Population C ouncil enabled m e to concentrate on and finance m y research, w ithout w hich I could not have produced a dissertation o f this quality. I am deeply grateful for all these. N ext, I thank the m em bers o f m y D issertation C om m ittee. Prof. Richard A. Easterlin introduced m e to research and helped m e to select an appropriate topic and w o rk on it. Prof. David M . H eer helped to design m y final year o f research and guided m e through the entire process till the very end, contributing substantially to it. Prof. Laurie Brand spent m any hours going through the draft and m ade valuable suggestions for im provem ents. Prof. Gail H arrison, despite her onerous responsibilities as the C hair o f the D epartm ent o f C om m unity Health Sciences, U C L A School o f Public H ealth, w ent through the draft and helped me to im prove m y w ork. Prof. John Elliott, D irector o f the Poltical E conom y and iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Public Policy program , has been a source o f inspiration and assistance in num erous w ays throughout the last five years. I thank Dr. Faride M otam edi for her help as the A ssociate D irector o f the Political E conom y and Public Policy program . I thank the Population Reference B ureau for the valuable training I was provided about population com m unications. I thank A rshad M ahm ood for the valuable help provided in SAS program ming. Finally, but not the least, I thank m y w ife R ose and children A lbin and D eepak for their im m ense help in typing, com paring, scanning and for bearing with m y research w ork w ith patience. W ithout their help I could not have accom plished this task. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Page D E D IC A T IO N jj A C K N O W L E D G E M E N T S iii L IST O F FIG U R ES ix L IST O F TA B LES xi XV A B S T R A C T C H A P T E R 1. IN T R O D U C T IO N 1 1.1 T he purpose o f the study 1 1.2 A dverse consequences o f grow th retardation 4 1.3 Effectiveness o f interventions 10 1.4 R elevance o f the study o f child grow th in India 16 1.5 T he organization o f this study 24 C H A PT E R 2. M E A SU R E M E N T O F G R O W T H 27 2.1 T hree approaches to m easuring undem utrition 28 2.2 A nthropom etric indices and indicators 30 2.2.1 W eight-for-age 31 2.2.2 H eight-for-age 32 2.2.3 W eight-for-height 33 2.2.4 W aterlow classification 35 2.3 C hoice o f anthropom etric indicators 35 2.4 A nthropom etric reference data 38 2.4.1 Justification for a com m on anthropom etric reference and creation o f reference data 38 2.4.2 Sources o f reference data for children 4 1 2.4.3 M aking com parisons using reference data 44 2.4.4 C u to ff points, sensitivity and specificity 46 2.5 C ontroversies in the assessm ent o f undem utrition 47 2.5.1 The ‘range o f hom eostasis’ controversy 50 2.5.2 The ‘sm all but healthy’ controversy 51 2.5.3 Sum m ing up the controversy 53 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3. GROWTH POTENTIAL AND GROWTH PATTERNS 56 3.1 Socio-econom ic status and grow th potential 56 3.2 A ge and regional patterns o f grow th retardation 63 3.3 C auses o f age patterns in grow th 72 3.3.1 Problem s related to w eaning 72 3.3.2 Enhanced dietary requirem ents o f children 73 3.3.3 P oor quality o f breastm ilk and w eaning diets 75 3.3.4 P o o r conditions for catch-up grow th 77 3.4 C oncluding rem arks 80 C H A P T E R 4. T H E B A SIC A N D U N D E R L Y IN G C A U SES OF C H ILD G R O W T H IN K ER A LA A N D U T T A R PR A D ESH 81 4.1 Schooling: Progress in K erala, and the poor situation in U ttar Pradesh 84 4.1.1 R egional disparities w ithin states 85 4.1.2 D isparities in gender and social class and the quality o f education 87 4.1.3 A field investigation in U ttar Pradesh 94 4.1.4 The historical progress o f schooling in K erala 101 4.1.4.1 Social m ovem ents for literacy and governm ent policy 101 4.1.4.2 T he lead taken by the C om m unist Party 108 4.1.4.3 A culture favoring fem ale literacy 109 4.1.5 Sum m ing up 111 4.2 H ealth services in K erala and U ttar Pradesh 111 4.2.1 H ealth services in U ttar Pradesh 112 4.2.2 H ealth services in K erala 117 4.2.3 Sum m ing up 123 4.3 P ublic distribution system 126 4.3.1 The creation o f the PDS in K erala 127 4.3.2 A field study in K erala 131 4.3.3 Sum m ing up 135 4.4 L and reform s 135 4.4.1 L and distribution in U ttar Pradesh and K erala before and after land reform s 136 4.4.2 The struggle for and the results o f land reform s in K erala 141 4.4.3 The failure o f land reform s in U ttar Pradesh 148 4.4.4 Sum m ing up 150 4.5 C onclusion 151 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5. DATA AND METHODS 155 5.1 Source o f data 155 5.2 Sam ple design and im plem entation 157 5.2.1 Q uestionnaires 157 5.2.2 Sam ple design 159 5.2.3 D ata quality control 160 5.2.4 The data 162 5.3 V ariables used in this study 163 5.3.1 D ependent variables 164 5.3.2 Independent variables 167 5.3.2.1 C hildem ’s age 168 5.3.2.2 Prenatal visits 169 5.3.2.3 Place o f birth 172 5.3.2.4 Size at birth 173 5.3.2.5 Prevalence o f fever 174 5.3.2.6 M easles im m unization 175 5.3.2.7 Schooling o f parents 177 5.3.2.8 Som e background characteristics o f N FH S respondents 179 5.3.2.9 A ge o f the m other at first birth 181 5.3.2.10 A gricultural land ow ned by households 182 5.3.2.1 1 Socio-econom ic status (SES) 184 5.3.2.12 O ccupational status o f parents 188 5.4 T he statistical m odel 189 5.4.1 O rdered logistic regression 189 5.4.2 The m odel 192 5.4.3 The BIC statistic 194 5.5 Sum m ing up 194 C H A PT E R 6 . R EG R ESSIO N RESU LTS 203 6 .1 T he results o f the country m odels 205 6.1.1 A ge o f the child 206 6 .1.2 Prenatal visits 211 6.1.3 Place o f birth 212 6.1.4 Size at birth 213 6 .1.5 H aving fever 215 6 .1.6 Schooling o f the m other 215 6.1.7 Schooling o f the father 218 6 .1.8 Low age o f the m other at first birth 218 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.1.9 A bsence o f toilet facility 219 6.1.10 Living in poor quality house 221 6.1.11 O w nership o f agricultural land 222 6.1.12 Interaction betw een prenatal visits and fathers’ schooling 224 6.1.13 Interaction betw een prenatal visits and m others’ schooling 226 6.1.14 R esults o f probability calculations in the country m odels 227 6.2 T he results o f the state m odels 232 6.2.1 A ge o f the child 232 6.2.2 Place o f birth 233 6.2.3 Size at birth 234 6.2.4 Schooling o f the father 235 6.2.5 Schooling o f the m other 235 6.2.6 L ow age o f m others at first birth 236 6.2.7 A bsence o f toilet facility 236 6.2.8 T he effect o f residence in K erala 237 6.2.9 Interaction between prenatal visits and fathers’ education 238 6.2.10 Interaction betw een prenatal visits and m others’ education 240 6.2.11 R esults o f probability calculations in the state m odels 241 6.3 C orrelation results 245 6.4 C onclusion 247 C H A P T E R 7. C O N C LU SIO N S A N D PO L IC Y IM PL IC A T IO N S 266 B ibliography 272 A ppendix 1 293 A ppendix 2 294 A ppendix 3 295 A ppendix 4 296 v iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure 1.1 C auses o f m alnutrition and poor grow th Figure 1.2 M ean differences (deficits) in length betw een 3-year-old G uatem alan children receiving varying levels o f protein-calorie supplem entation and the 50th percentile o f the N C H S Figure 1.3 C om parison o f the developm ental quotients (D Q ) on the G riffiths T est o f children w ith severe m alnutrition F igure 3.1 M ean heights o f Japanese boys in 1957 (•), 1967 (x) and 1977 (o) plotted on N C H S grow th curves Figure 3.2 M ean heights o f 7-year-old boys o f high (•) and low (o), socio- onom ic status in various developing countries Figure 3.3 M ean length o f 459 H onduran m ale infants relative to N CH S percentiles Figure 3.4 W eight-for-height and height-for-age z scores for children aged 0- 7 years in H onduras Figure 3.5 The age patterns o f w asting in H onduras, G uatem ala and Bangladesh Figure 3.6 A verage requirem ents for protein and for the sum o f eight essential am ino acids at 6 m onths, 10-12 years, and in adulthood Figure 3.7 (a) Effects o f different types o f input services on the m ean heights o f children from the N arangw al project in the first three years o f life (adjusted for sex, birth order, m other’s age, caste, year and season o f observation) (b) M ean heights o f the sam e children relative to the H arvard height standard (adjusted for caste and sex) Figure 6.1 Predicted probabilities o f severe underw eight, underw eight and norm al w eight-for-age o f children 1-4 years old: India, 1992-93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.2 Predicted probabilities o f severe stunting, stunting and norm al height-for-age o f children 1-4 years old: India, 1992-93 250 Figure 6.3 Predicted probabilities o f severe w asting, w asting and norm al w eight-for-height o f children 1-4 years old, India: 1992-93 251 Figure 6.4 Predicted probabilities o f severe underw eight, underw eight and norm al weight-for-age o f children 1-4 years old: K erala-U P, 1992-93 252 Figure 6.5 P redicted probabilities severe stunting, stunting and norm al height-for-age o f children 1-4 years old: K erala-U P, 1992-93 253 Figure 6.6 Predicted probabilities o f severe wasting, wasting and normal w eight-for-height o f children 1-4 years old: K erala-U P, 1992-93 254 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES T able 1.1 DHS-1 countries w ith the highest percentages o f underw eight children in 1986-89 and the percentages o f underw eight children in India in 1992-93. Table 1.2 DHS-1 countries w ith the highest percentages o f stunted children in 1986-89 and the percentages o f stunted children in India in 1992-93 Table 1.3 DHS-1 countries w ith the highest percentages o f w asted children in 1986-89 and the percentages o f w asted children in India in 1992-93 T able 1.4 D aily additional intake o f calories and protein by category o f A tole intake Table 1.5 D em ographic, Literacy and H ealth Indicators for India and States T able 1.6 Percentage o f population below P overty-line in India and states in 1977-78, 1983-84 and 1987-88 Table 1.7 Percentages o f children 0-4 years old undernourished in India and states in 1992-93 Table 2.1 W aterlow classification Table 3.1 A nthropom etry at 5.0 and 17.5 years o f age in rural Indian children Table 4.1 Percentages o f m ales and females literate in Kerala during 1951- 1991 T able 4.2 Proportion o f literate persons at age +7 am ong scheduled castes and scheduled tribes in K erala in 1961 and 1991 Table 4.3 Percentage o f SC and ST students in K erala and U ttar Pradesh in 1990-91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4 Percentage o f students enrolled in grades 1-5 and 6-8 in K erala, U ttar Pradesh and India in 1990-91 92 Table 4.5 R ates o f functional literacy in K erala, U ttar Pradesh and India, 1967-69 93 Table 4.6 G row th o f G overnm ent-recognized V ernacular Schools and Students in Travancore, 1873-1894 106 T able 4.7 The expenditure on education as a proportion o f developm ental and non-developm ental expenditure in K erala and other states 107 Table 4.8 Select indicators o f health services and public distribution system 114 in U ttar Pradesh and K erala (1980-90) Table 4.9 Percentage o f V illages with Selected Public A m enities, 1992-3 115 Table 4.10 The num ber o f beds in m edical institutions and death rates in the two regions o f K erala during 1951-1971 118 Table 4.11 Percentage distribution o f villages according to population in Indian states: 1981 120 T able 4.12 Percentage o f H ospitals and beds according to ow nership in K erala in 1991-92 122 Table 4.13 Percentage Shares o f land ow ned by bottom 40 and top ten percent o f households in U ttar Pradesh and K erala 137 Table 4.14 Percentage o f households ow ning no land in the rural areas o f U ttar Pradesh and K erala 138 Table 4.15 The percentage o f households not operating any land in U ttar Pradesh and K erala 140 T able 4.16 The percentage share o f the bottom 40 percent o f the households in the area operated in U ttar Pradesh and K erala 141 T able 4.17 Percentages o f children undernourished in U ttar Pradesh and K erala in 1991-92 154 xn Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.1 Percentages o f children undernourished by dem ographic characteristics o f m others and children in India, 1992-93 196 Table 5.2 Levels o f years o f schooling o f w om en and their husbands in the N FH S 1992-93 178 Table 5.3 B ackground characteristics o f N FH S respondents, India 1992-93 180 Table 5.4 Percentages o f households ow ning agricultural land: N FH S 1992- 93 183 Table 5.5 Percentages o f households having item s m easuring the SES 185 Table 5.6 The m eans o f variables used in the regression for India for height- for-age 198 T able 5.7 The m eans o f variables used in the regression for India for w eight-for-age and w eight-for-height 199 T able 5.8 The m eans o f variables used in the regression o f tw o states, K erala and U ttar Pradesh for w eight-for-age, height-for-age, and w eight-for-height 200 Table 5.9 The sam ple m eans o f variables for K erala 201 Table 5.10 The sam ple m eans o f variables for U ttar Pradesh 202 Table 6.1 Logistic regression coefficients, odds ratios, likelihood estim ate and BIC show ing the effects o f proxim ate, m aternal and household covariates on w eight-for-age in Indian children 1-4 years old in 1992-93 255 Table 6.2 Logistic regression coefficients, odds ratios, likelihood estim ate and BIC show ing the effects o f proxim ate, m aternal and household covariates on height-for-age in Indian children 1-4 years old in 1992-93 256 xiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.3 Logistic regression coefficients, odds ratios, likelihood estim ate and BIC show ing the effects o f proxim ate, m aternal and household covariates on w eight-for-height in Indian children 1-4 years old in 1992-93 Table 6.4 Logistic regression coefficients, odds ratios, likelihood estim ate and B IC show ing the effects o f proxim ate, m aternal and household covariates on w eight-for-age in U ttar Pradesh and K erala am ong children 1-4 years old in 1992-93 Table 6.5 Logistic regression coefficients, odds ratios, likelihood estim ate and BIC show ing the effects o f proxim ate, m aternal and household covariates on height-for-age in U ttar Pradesh and K erala am ong children 1-4 years old in 1992-93 Table 6.6 Logistic regression coefficients, odds ratios, likelihood estim ate and BIC Show ing the effects o f proxim ate, m aternal and household covariates on w eight-for-height in U ttar Pradesh and Kerala am ong children 1-4 years old in 1992-93 Table 6.7 Pearson Correlation coefficients between w eight-for-age, height- for-age and weight-for-height Table 6.8 Probabilities for the interaction betw een prenatal visits and fathers’ schooling keeping all other variables at their m eans in the w eight-for-age m odel for India, 1992-93 Table 6.9 Probabilities for the interaction betw een prenatal visits and father’s schooling keeping all other variables at their m eans in the height-for-age m odel for India, 1992-93 Table 6.10 Probabilities for the interaction betw een prenatal visits and father’s schooling keeping all other variables at their m eans in the w eight-for-age m odel for K erala and U ttar Pradesh, 1992-93 Table 6.11 Probabilities for the interaction betw een prenatal visits and father’s schooling keeping all other variables at their m eans in the height-for-age m odel for K erala and U ttar Pradesh, 1992-93 Table 6.12 Pearson Correlation coefficients between weight-for-age, height- for-age and weight-for-height Reproduced with permission of the copyright owner. Further reproduction prohibited without permission A b s tra c t T h e d e te rm in a n ts o f ch ild g ro w th re ta rd a tio n in In d ia The study analyzed the N ational Fam ily H ealth Survey o f India 1992-93 data fo r the follow ing purposes: (1) to estim ate the probabilities o f underw eight, stunting and w asting am ong children 1-4 years old, (2) to identify the m ajor determ inants o f child grow th retardation, and (3) to suggest policies that can im prove child grow th. Further, public schooling, public health, land reform s and public distribution o f essential com m odities in two states o f India, nam ely, K erala and U ttar Pradesh, w ere com pared to dem onstrate the effects o f underlying and basic causes on the w ell-being o f the population and child grow th. K erala and U ttar Pradesh represent states that have done relatively w ell and poorly, respectively in these areas. The data for these two states w ere also analyzed for purposes (1), (2) and (3) above. A nthropom etric m easurem ents, and proxim ate, m aternal and household characteristics o f children w ere analyzed using ordered logistic regression. The regression sam ples for the w hole o f India had 25550 children in the w eight-for-age m odel, and 19368 children in the height-for-age and w eight-for-height m odels. The state m odels had 4957 children. Prenatal visits and parental schooling had significant, independent and negative effects on child grow th retardation. Fathers’ schooling had significant xv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. effects on only on underw eight. M other’s schooling had significant effects on underw eight, stunting and w asting. H ow ever, the interactions betw een prenatal visits and parental schooling had greater effects. In addition, age o f the child, place o f birth, size at birth, having fever, m other being less than tw enty years at first birth, absence o f any toilet facility, poor-quality o f house and ow nership o f agricultural land had significant effects. Child age w as negatively associated with underw eight and w asting, but positively associated w ith stunting. Short term policies recom m ended include effective education o f less literate parents on m aternal and child health, and nutrition during prenatal visits, m ore effective use o f the m edia to im prove the health and nutritional know ledge o f less literate parents, and supplem entary feeding o f poor m others and young children. L ong term m easures include increasing schooling and aw areness, enabling w eaker sections to im prove their lot using the dem ocratic process. xvi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1 INTRODUCTION Child grow th has for long been used to assess the health and nutritional status o f children based on the prem ise that poor grow th, in m ost cases, reflects deviations from optim al environm ental conditions that support grow th. The m ain influence on the grow th o f children is the environm ent in w hich they live (M artorell and H abicht 1986). Poor diet and infections are the m ain environm ental conditions o f interest to biom edical researchers, m edical and health practitioners and health planners in developing countries (Haas and H abicht 1990). In the m odel at Figure 1.1, adapted from U n icef (1990), the im m ediate, underlying and basic causes o f m alnutrition and poor grow th are presented. These factors together m ake up the environm ent, w hich largely determ ines the nutritional status and grow th o f children. 1.1 The purpose of the study M y goals in this dissertation are to: (1) estim ate the probabilities o f grow th retardation am ong children betw een one and four years old in India, using three anthropom etric indicators, nam ely, weight-for- age, height-for-age and weight-for-height; 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2) identify the m ajor im m ediate, underlying and basic determ inants o f growth retardation am ong children in India (Figure 1.1); and (3) suggest policy to reduce the probability o f child grow th retardation in India. Figure 1.1 Causes of Malnutrition and Poor Growth Malnutrition and Poor Growth Inadequate Dietary Intake Disease Insufficient > 1 Health Services & Unhealthy Environment, Insufficient Household Food Security Inadequate Maternal & Child Care Formal and Non-Formal Institutions Political and Ideological Superstructure Manifestations Immediate C auses Underlying C auses \ I / Economic Structure Potential Resources Basic Causes Source: A dapted from U nicef 1990. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1.1 DHS-1 countries with the highest percentages o f underweight children in 1986-89 and the percentage o f underweight children in India in 1992-93. S tu n te d * S ev erely S tu n ted C o u n try (%) (%) India 5 3 .4 20.6 B u ru n d i 3 7 .5 10 Sri Lanka 3 7.3 8.3 G uatem ala 3 3 .2 8.5 M ali 3 0 .6 9.2 Table 1.2 DHS-1 countries with the highest percentages o f stunted children in 1986-89 and the percentage o f stunted children in India in 1992-93 U n d erw eig h t* S ev erely U nderw eight** C o u n try (%) (%) In d ia 5 2 2 8 .9 G uatem ala 5 7 .7 30.1 B u ru n d i 4 7 .4 18.8 U ganda 4 3 .8 17.3 B o liv ia 3 7 .7 14.5 Table 1.3 DHS-1 countries with the highest percentages o f wasted children in 1986-89 and the percentage o f wasted children in India in 1992-93. W asted * S ev erely W asted C o u n try (%) (%) In d ia 17.5 3.2 Sri L anka 11.6 0.6 M ali 11.5 1.3 G hana 8 0.7 O ndo State, N ig eria 6 .2 0.7 B u ru n d i 5 .7 0.8 T rinidad and T obago 4 .8 0.7 * W eight-for-age, h eight-for-age o r w eig h t-fo r-h eig h t b elo w -2 stan d ard d eviations from the international reference m edian ** W eight-for-age, h eight-for-age o r w eig h t-fo r-h eig h t b elo w -3 stan d ard deviations from the international reference m edian Sources for all three tables: A dapted from S om m erfelt and S tew art, 1994, for D H S-I countries; U PS 1995, fo r India. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A ccording to the U nited N ations H um an D evelopm ent R eport (1999), India had the highest percentage o f children w ho w ere underw eight, next only to Bangladesh, am ong the 174 countries listed in the Report. T ables 1.1, 1.2 and 1.3 list the countries with the highest levels o f grow th retardation in term s o f underw eight, stunting and w asting in children am ong the 19 developing countries o f A sia, Latin A m erica and A frica during 1986-89 (Som m erfelt and Stew art 1994) and in India during 1992-93 (UPS 1995). This com parison show ed that India had (1) the highest percentages o f children w ho w ere underw eight and w asted, and the second highest percentage o f stunted children; and (2) the second highest percentages o f children w ho are severely stunted, and the highest percentages o f children w ho are severely underw eight and severely w asted. G row th retardation has m any adverse consequences. H ow ever, studies that are discussed below have show n that these adverse consequences can be m itigated through appropriate interventions. 1.2 Adverse Consequences of growth retardation G row th retardation is a m anifestation o f po o r nutritional status. It has m any adverse consequences for psychom otor developm ent in childhood, and on w ork capacity and productivity in adulthood. Stunting in early childhood in G uatem ala w as found to have led to a w ide range o f adverse outcom es in adulthood such as reduced fat-free mass, short adult stature and reduced intellectual achievem ent 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (M artorell et al. 1992). It w as found that even tem porary food deprivation had adverse effects on cognitive abilities o f adequately nourished children (Pollitt et al. 1981; C onners and Blouin 1983; Pollitt et al. 1985), such effects being m ore severe in the case o f undernourished children (Sim eon and G rantham -M cG regor 1987). A fter review ing several studies conducted in com m unities w ith endem ic m alnutrition, G rantham -M cG regor (1990) concluded that children subject to severe m alnutrition w ere likely to suffer long-term effects on m ental developm ent. U ndernourishm ent reduces physical activity in young children and reduced physical activity affects their m otor developm ent and coordination (M alina 1984; C havez and M artinez 1979; Satyanarayana et al. 1979; R utishauser and W hitehead 1972; V iteri and Torun 1981), and this in turn, their cognitive and social developm ent (S am eroff and M cD onough 1984). M alnutrition in childhood affects physical w ork capacity and productivity in adulthood. B ecause o f less m echanization, m uch o f the w ork in developing countries is done by hum an labor. Hence physical size, strength and endurance are im portant for output and productivity. A fter review ing a num ber o f case studies done by him and others, Spurr (1990) cam e to the follow ing conclusions: chronic nutritional deficiency in childhood leads to reduced grow th and sm aller body size as adults in poor countries; reduced grow th and sm all adult size results in the reduced physical w ork capacity and the low er productivity o f those w ho engage in m oderate to heavy 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. physical w ork; such persons w ould be forced to w ork shorter hours or have increased periods o f rest, thus reducing their total w ork effort; this ultim ately underm ines the econom ic viability o f undernourished populations. Satyanarayana et al. (1977,1978) found that even in less strenuous factory w ork involving production o f detonator fuses, body w eight, height and lean body m ass w ere significantly and positively correlated w ith productivity. In developed countries it has been found that those w ho w ere o f sm aller stature w ent dow n in the social class scale whereas those w ho w ere larger and healthier m oved up the occupational scale (Eveleth and T anner 1990). Thus failure to grow norm ally has m any econom ic consequences. M alnutrition in childhood affects reproductive outcom es. Short m aternal size w as found to be associated with prolonged labor, cesarean section (A dadevoh et al. 1989; M ahm ood et al. 1988) and high infant m ortality (M artorell et al. 1981). K ram er (1987) cites low m aternal stature as one o f the factors leading to low birth w eight. L ow birth w eight was associated w ith poor psychom otor developm ent in childhood (Lasky et al. 1975). Latham (1997) has sum m arized docum ented evidence o f m alnutrition reducing the body’s resistance to infections, som e o f which are as follow s: Children w ith kw ashiorkor w ere found to be unable to form antibodies follow ing typhoid and diphtheria vaccine; this capacity w as restored after protein therapy. Children with protein m alnutrition have an im paired antibody response to yellow fever vaccine. 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M any studies found that the m alnourished body w as unable to defend itself against infection. P oor nutritional status im pairs cell-m ediated im m unity that is im portant in providing resistance against viral and bacterial diseases (Floud, W atcher and G regory, 1990). Infections in turn affect nutritional status by loss o f appetite, energy loss during fever as heat, loss o f nutrients in sweat, vom iting, diarrhea, reduced absorption o f nutrients, protein catabolism 1 , and reduced food intake resulting from cultural factors (Floud, G regory and W atcher 1990; Latham 1997). Serious infections like typhoid fever and m ilder infections like tonsillitis result in protein catabolism . Full recovery requires increased intake o f protein during the recovery period. In children w hose protein intake is inadequate, grow th w ill be retarded during and follow ing the infections. Children in developing countries w ho have successive infections and often, m ultiple infections, becom e prone to grow th retardation. Parasitic infections are w idely prevalent in developing countries and have an adverse effect on nutritional status. F or instance hookw orm s cause intestinal blood loss and loss o f iron. R oundw orm s are even m ore w idely prevalent2. R oundw orm s are large and require considerable am ount o f nutrition for them selves, w hich they absorb from the hosts. H igh parasitic densities are com m on in 1 Protein catabolism is the destructive metabolism of protein that constitutes lean body mass during infections or even psychological stress. Only part o f the protein is catabolized for energy use and the rest is excreted as urinary nitrogen (Tanner, 1978). 2 Hookworms are estimated to infect 800 million people in tropical and sub-tropical countries, and roundworms, about 1200 million people (Latham 1997). 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. environm ents w here sanitation is poor. A m oebas cause serious dysentery and can cause m alabsorption and abdom inal pain. D iarrhea often leads to serious protein- energy m alnutrition. M easles fatality w as found to be m uch higher for m alnourished children. M ortality rates from m easles in the early 1960s was 83 tim es higher in M exico and 268 tim es higher in G uatem ala, than in U SA (Scrim shaw , T aylor and G ordon 1968). D uring a m easles epidem ic in the U nited R epublic o f T anzania, it w as found that there w as considerable m ortality am ong children o f poorer fam ilies, but very few m ortalities am ong children o f families having m oderate incom e such as hospital w orkers (Latham 1997). The greater m ortality from m easles am ong children in developing countries w as not due to differences in virus virulence, but because o f the differences in the h o sts’ nutritional status. O ther com m on diseases like w hooping cough, diarrhea and upper respiratory infections also had m ore severe consequences in children w ho w ere m alnourished than in those w ho w ere well nourished. T uberculosis is another disease that has been associated w ith w eight loss and w asting. A n inter-A m erican investigation o f 35,000 cases o f m ortality o f children u n d er five in ten countries show ed that in fifty-seven percent o f the cases, m alnutrition was either the underlying or the associated cause o f death. The ‘synergistic’ relationship betw een nutritional status and infections is capable o f producing secondary infections, som etim es m ore dam aging than even the p rim ary infection (Cecil 1947). Som e studies have show n that undem utrition affects 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the severity rather than the incidence o f infections (M artorell, 1980; C handra, 1980; Chen and Scrim shaw 1983; K ielm ann, T aylor and De Sw eem er, 1983). Pelletier (1994) review ed the results o f 28 com m unity-based, prospective studies done in A sia and A frica in the 1980s and early 1990s that exam ined the relationship betw een anthropom etric indicators o f m alnutrition and child m ortality. O ne o f the im portant findings o f the study w as that m alnutrition had a potentiating or m ultiplicative effect on m ortality rather than an additive effect. A n increase or decrease in m alnutrition w as found to have a higher im pact on m ortality in populations w ith higher levels o f m ortality than in populations w ith low er levels o f m ortality. This im plied that the im pact o f m alnutrition on m ortality varied according to the burden o f m orbidity and the im pact o f m orbidity varied according to the prevalence o f m alnutrition. The study estim ated a 5.9% com pound rate o f increase in m ortality for each percent decline in w eight-for-age, with a standard error o f 0.8% (Pelletier et al. 1994b). The relative risk o f m ortality due to m alnutrition w as not found to vary across different form s o f m orbidity (Fauveau et al. 1990, V ella et al. 1994). These studies lead to the conclusion that the risk o f grow th retardation from infections is m uch higher in m alnourished populations than in w ell-nourished populations. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.3 Effectiveness of interventions W e have seen that undem utrition adversely affects psychom otor developm ent in children; physical w ork capacity and productivity; and reproductive outcom es in adults. M alnutrition interacts with infections to produce m ultiplicative adverse effects. The question that arises then is w hether interventions can reduce grow th retardation and its adverse consequences. Experim ental and other studies have show n that nutritional interventions can be effective in doing so. M artorell and H abicht (1986) cite tw o studies that show that children can recover from physical grow th retardation if there is tim ely intervention. T hey are the food supplem entation studies done by the Institute o f N utrition o f C entral A m erica and Panam a (IN C A P) study in four m alnourished G uatem alan villages (M artorell et al., 1982) and the study in Bogota, C olum bia (M ora et al., 1981). In the IN C A P study four groups o f children from four m alnourished villages w ere given a drink by the nam e 'Atole', high in energy and protein. The children w ere given ‘A to le’ tw ice daily from birth to the age o f three. The additional calories thus provided to the four groups o f children w ere as follows: less than 50 kcal/day to G roup 1,51 - 100 kcal/day to G roup 2, 101 - 200 kcal/day to G roup 3 and m ore than 200 kcal/day to G roup 4. A dditional protein was also provided to the four groups as given in Table 1.4. H om e diets rem ained the sam e. C hildren w ho w ere provided ‘A to le’ im proved as show n in Figure 1.2. On the left axis is the difference betw een the m ean length o f 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.4 Daily additional intake of calories and protein by category of 'Atole' intake C ategory A dditional kcal/day A dditional Sam ple size o f bovs Sam ple size o f girls protein/dav ( grams") 1 0-50 0-3.4 23 32 2 51-100 3.5-6.7 28 26 3 101-200 6.8-13.4 44 41 4 > 200 > 13.4 22 13 N ote: 'A tole' w as the nam e o f the energy-protein drink. 'A tole' w as given from birth to three years o f age. Source: A dapted from M artorell et al. 1982. Figure 1.2 Mean differences (deficits) in length between 3-year-old Guatemalan children receiving varying levels of protein-calorie supplementation and the 50th percentile of the NCHS. E u 1 4 I 2 10 8 6 i Baseline [ [ boys R^Igirls - I 4.0 3.5 3.0 2.5 2.0 1.5 1 .0 0.5 Q to 2 GROUP Source: M artorell 1985. 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1.3 Comparison of the developmental quotients (DO) on the Griffiths Test of children with severe malnutrition 100 O' Q Time after returning h om e (m onths) A: on admission to hospital and up to 60 months after returning home. Solid lines: mean (and two SDs) o f 16 children Dotted line: DQ o f child adopted 24 months after leaving hospital Source: G rantham -M cG regor 1990. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the groups and the N C H S 50th percentile3. On the right axis are the N C H S standard deviations. T he reference group was 12 cm shorter than the N C H S m edian height at 3 years. W hile Group 1 differed little from the reference group at age 3, all other groups got taller. G roup 4 im proved m ost and w as only 7.5 cm shorter than the N C H S m edian. T he reference group w as 3.3 standard deviations and G roup 4 was 2.0 standard deviations below the N C H S m edian. T he B ogota study also cam e to the sam e conclusions. Several studies have been done to determ ine the effects o f interventions on m ental developm ent o f m alnourished children. O ne study observed a severely m alnourished child in order to evaluate the effect o f enriched environm ent on mental developm ent (G rantham -M cG regor and B uchanan 1982). O f the seventeen children w ho w ere hospitalized for treatm ent, sixteen w ere returned to the im poverished conditions from w hich they cam e. A m iddle-class fam ily adopted one child tw o years after release from the hospital. T he adopted boy show ed dram atic im provem ent in physical and m ental developm ent after adoption as show n in Figure 1.3. T he other 16 children did not show any im provem ent after release from the hospital. In an intervention study designed to m easure the effect o f increased stim ulation on m ental developm ent, a group o f m alnourished children w ere given 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increased stim ulation at their hom es by giving them toys and teaching their m others to play w ith them regularly (G rantham -M cG regor et al. 1987). They w ere observed for over three years. A nother group o f m alnourished children and a group o f adequately nourished group w ere the control groups. T he group that w as given the stim ulation show ed rem arkable im provem ent in m ental developm ent and caught up w ith the norm al children. T he other m alnourished group did not show any im provem ent. Follow ing the end o f the experim ent the intervened group show ed a small decline in developm ent. T hese studies show that m alnutrition in childhood causes physical and m ental grow th retardation, w hich in turn affects physical w ork capacity and productivity in adulthood. F urther infections and m alnutrition interact to produce m ultiplicative effects on nutritional status and growth. Even sm all increases in diet have been show n to increase birth w eight o f children w hose m others suffer from protein-energy m alnutrition (Latham 1997). A ppropriate interventions during childhood can reverse physical and m ental grow th retardation and avoid the adverse consequences o f grow th retardation. H ow ever, it is difficult to estim ate the extent o f reversal o f linear grow th retardation in an individual child because the ch ild ’s genetic potential cannot be precisely estim ated. T here are also environm ental constraints that cannot be precisely estim ated for an individual child. H ow ever, this problem does not arise in population studies w here the com parison is done w ith a reference population (H aas 3 NCHS 50lh percentile indicates the median height achieved by children in the United States. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and H abicht 1990). It is also difficult to determ ine the exact age until which children can recover lost grow th in im proved environm ents. A fter review ing the results o f studies done in Jam aica and Peru W aterlow (1988) estim ated that catch-up grow th is possible at least until the age o f ten. FA O /W H O /U N U (1985:143) estim ated that capacity for catch-up grow th in height is retained until the end o f the adolescent grow th spurt. Steckel (1987) reported catch-up grow th am ong adolescent slave children w ho w ere extrem ely stunted when they w ere given better diet. 1.4 Relevance of the study of child growth in India In the context o f the high levels o f child grow th retardation in India, its adverse consequences and the effectiveness o f interventions in reversing growth retardation, it w ould be desirable to investigate the causes o f child grow th retardation and the interventions that are likely to be effective in reducing it in India. It is now generally accepted that the environm ent in w hich children live is the m ain influence on their physical growth and m ental developm ent (Figure 1.1). The N ational Fam ily H ealth Survey (N FH S) data gathered for the w hole o f India in 1992-93 has m ade it possible to detennine som e o f the m ajor causes o f m alnutrition and child grow th retardation in India. H ow ever factors such as household food security, education and health care that affect the nutritional status and the grow th o f children in India depend on public policy and the ability o f the people to force the state to adopt 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. policies and expand public services in their favor and to ensure that public services are efficiently delivered (Kannan 1988; Sen 1977; D reze and Sen 1989; D reze and Sen 1995; D reze 1995; Jos M ooij 1999; D reze and G azdar 1996; R am achandran 1996; Caldw ell 1986). M any o f the policies and m uch o f the policy im plem entation in education, public health, land reform s, m inim um w ages, public distribution o f essential com m odities and so on that affect the health o f the com m unity and the grow th o f children are the responsibility o f state governm ents in India. T herefore an evaluation o f public policy and peo p le’s involvem ent w ould require a study o f the social, econom ic and political situation in each state o f India. H ow ever such a detailed study is beyond scope o f this dissertation for w ant o f tim e and resources. T herefore I com pare two states o f India, nam ely, K erala and U ttar Pradesh, using published research findings and m y own analysis o f N FH S data o f these tw o states for children 1-4 years old. D ata presented in T able 1.5 show that these tw o states are representative o f states that have better and w orse indices o f quality o f life for both adults and children in India (D reze and Sen 1995). Included in this table are 16 states o f India that had populations exceeding 5 m illion according to the 1991 Census. The states are arranged in descending order o f fem ale literacy in the 7+ age group. The highest fem ale literacy rate was 86 percent in K erala and the lowest was 20 percent in Rajasthan. It w as 25 percent in U ttar Pradesh. Fem ale life expectancy varied from 74 years in K erala to 53 to 55 years in M adhya Pradesh, U ttar Pradesh 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and Orissa. Rural fem ale literacy rates in the ten to fourteen age group w ere 98 percent in K erala, and 39 percent in U ttar Pradesh; only tw o other states, nam ely, B ihar and Rajasthan w ere below U ttar Pradesh. Child death rate (age 0 to 4) w as 10 tim es higher in M adhya Pradesh and 8 tim es higher in U ttar Pradesh than in Kerala. The fem ale-m ale ratio (fem ales per 1000 m ales) w as 1036 in K erala, sim ilar to the ratio in developed countries. It w as the low est in H aryana (865), w hich w as low er than for any country in the w orld (D reze and Sen 1995). T his ratio is considered to be a good index o f fem ale status and autonom y in society. The fem ale-m ale ratio was 879 in U ttar Pradesh, low est in India, next only to H aryana. The num ber o f children bom per w om an (TFR ) w as 1.8 in K erala and 5.1 in U ttar Pradesh, w hich w as the highest in India. 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.5 Demographic, literacy and health indicators for India and states State Literacy rate in Rural literacy rate in Life Expectancy at birth. Death-rate for Total Female-male Per capita Populatioi 7+ aee group. 1991 Female Male 10-14 aee proup. 1987-8 Female Male 11990-921 Female Male 0-4 aee eroup. 1991 Fertility Rate. 1991 ratio. 1991 incom e 1991 fm illionl Kerala 8 6 .0 94.0 98.0 98.0 74.4 6 8 .8 4.3 1.8 1036.0 46 29.1 Himachal Pradesh 52.0 75.0 81.0 95.0 n/a n/a 19.3 3.1 976.0 53 5.2 Maharashtra 52.0 77.0 6 8 .0 8 6 .0 64.7 63.1 16.3 3.0 934.0 82 78.9 Tamil Nadu 51.0 74.0 71.0 85.0 63.2 61.0 16.1 2 .2 974.0 51 55.9 Punjab 50.0 6 6 .0 69.0 76.0 67.5 65.4 17.0 3.1 882.0 96 20.3 Gujarat 49.0 73.0 61.0 78.0 61.3 59.1 23.3 3.1 934.0 64 41.3 W est Bengal 47.0 6 8 .0 61.0 69.0 62.0 60.5 2 0 .6 3.2 917.0 54 68.1 Karnataka 44.0 67.0 56.0 74.0 63.6 60.0 23.6 3.1 960.0 55 45.0 Assam 43.0 62.0 78.0 83.0 n/a n/a 32.4 3.5 923.0 42 22.4 Haiyana 41.0 69.0 63.0 87.0 63.6 62.2 23.0 4.0 865.0 87 16.5 Orissa 35.0 63.0 51.0 70.0 54.8 55.9 39.0 3.3 971.0 41 31.7 Andhra Pradesh 33.0 55.0 42.0 6 6 .0 61.5 59.0 21.3 3.0 972.0 56 66.5 Madhya Pradesh 29.0 58.0 40.0 6 8 .0 53.5 54.1 44.5 4.6 931.0 41 6 6 .2 Uttar Pradesh 25.0 56.0 39.0 6 8 .0 54.6 56.8 35.6 5.1 879.0 40 139.1 Bihar 23.0 52.0 34.0 59.0 58.3 n/a 2 2 .8 4.4 911.0 29 86.4 Rajasthan 2 0 .0 55.0 22 .0 72.0 57.8 57.6 30.9 4.6 910.0 44 44.0 INDIA 39.0 64.0 52.0 73.0 59.4 59.0 *In 1992 yearly in Hundred rupees at current prices. Source: Dreze and Sen 1995 26.5 3.6 927.0 56 846.3 v O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.6 Percentages of population below poverty-line in India and states in 1987-88, 1983-84 and 1987-88 1977-78 1983-84 1987-88 1977-8 and 1987- State Percent Percent Percent Percent Andhra Pradesh 43.6 36.4 31.7 27.3 Assam 47.3 23.5 22.8 51.8 Bihar 56.3 49.5 40.8 27.5 Gujarat 38.9 24.3 18.4 52.7 Haryana 25.2 15.6 11.6 5 4.0 Himachal Pradesh 2 7 13.5 9.2 65.9 Jammu & Kashmir 18.4 16.3 13.9 24.5 Karnataka 50.8 35 32.1 36.8 Kerala 4 8.4 28.8 17 64.9 Madhya Pradesh 58.9 46.2 36.7 3 7.7 Maharashtra 50.6 34.9 29.2 42.3 Orissa 65.1 42.8 44.7 31.3 Punjab 16.4 13.8 7.2 56.1 Rajasthan 33.6 34.3 24.4 2 7.4 Tamil Nadu 52.8 39.6 32.8 37.9 Uttar Pradesh 49.7 45.3 35.1 29.4 West Bengal 52.2 39.2 27.6 47.1 India 48.3 37.4 29.9 38.1 *Not available. Source: NSS Reports and Economic Survey o f India 1994-95; from Agarwal, A.N. and H.O. Varma 1996. to o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.7 Percentages of children 0-4 years old undernourished in India and states in 1992-93 Below Underweight >-3 S.D Below Below Stunted >-3 S.D Below Below Wasted >-3 S.D Below -3 S.D. <-2 S.D. -2 S.D. -3 S.D. <-2 S.D. -2 S.D. -3 S.D. <-2 S.D. -2 S.D. India 20.6 32.8 53.4 28.9 23.1 52.0 3.2 14.3 17.5 Northern states Delhi 12.0 29.6 41.6 19.3 23.9 43.2 2.7 9.2 11.9 Haryana 9.0 28.9 37.9 19.3 27.4 46.7 0.6 5.3 5.9 Himachal Pradesh 12.9 34.1 47.0 u u u u u Jammu region o f J&K 13.8 30.7 44.5 18.6 22.2 40.8 3.4 11.4 14.8 Punjab 14.2 31.7 45.9 15.7 24.3 40.0 2.8 17.1 19.9 Rajasthan 19.2 22.4 41.6 26.6 16.5 43.1 5.2 14.3 19.5 Central states Madhya Pradesh 22.3 35.1 57.4 u u u u u u Uttar Pradesh 24.6 34.4 59.0 35.6 23.9 59.5 2.7 13.4 16.1 Eastern states Bihar 31.1 31.5 62.6 39.5 21.4 60.9 4.1 17.7 21.8 Orissa 22.7 30.6 53.3 25.2 23.7 48.9 3.6 17.7 21.3 W est Bengal 18.4 38.4 56.8 u u u u u u Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.7 continued Underweisht Stunted Wasted Below >-3 S.D Below Below >-3 S.D Below Below >-3 S.D Below -3 S.D. <-2 S.D. -2 S.D. -3 S.D. <-2 S.D. -2 S.D. -3 S.D. <-2 S.D. -2 S.D. Southern states Andhra Pradesh 15.6 33.5 49.1 U u U U U U Karnataka 19.4 34.9 54.3 22.7 24.9 47.6 2.6 14.8 17.4 Kerala 6.1 22.4 28.5 9.0 18.4 27.4 1.3 10.3 11.6 Tamil Nadu 13.3 29.5 42.8 u u U U U U Western states Goa 8.9 26.1 35.0 1 1 .0 21.5 32.5 2.4 12.9 15.3 Gujarat 17.6 32.5 50.1 25.3 22.9 48.2 3.7 15.2 18.9 Maharashtra 21.3 32.9 54.2 23.5 25.0 48.5 4.2 16.0 20.2 North-Eastern states Assam 18.7 31.7 50.4 26.3 25.9 52.2 1.7 9.1 1 0 .8 Manipur 7.2 22.9 30.1 16.0 17.6 33.6 1.2 7.6 8 . 8 Meghalaya 17.2 28.3 45.5 38.4 12.4 50.8 4.8 14.1 18.9 Mizoram 5.3 22.8 28.1 16.0 25.3 41.3 0.6 1.6 2.2 Nagaland 7.6 21.1 28.7 13.2 19.2 32.4 2.3 10.4 12.7 Tripura 18.6 30.2 48.8 21.3 24.7 46.0 0.7 16.8 17.5 Note: Figures are for children bom 1-47 months prior to the survey. Each index is expressed in standard deviation units (S.D) from the median o f the International Reference Population The number o f children for calculation o f height-for-age does not include children from Andhra Pradesh, Himachal Pradesh, Madhya Pradesh, Tamil Nadu and West Bengal. U: Data not available Source: Adapted from IIPS (1995) to to T able 1.6 presents the percentage o f the population below poverty line in Indian states. B oth K erala and U ttar Pradesh had sim ilar rates o f poverty in 1977- 78, but the poverty rate in K erala becam e less than h alf as that o f U ttar Pradesh by 1987-88. Poverty declined by 65 percent in K erala and 29 percent in U ttar Pradesh in 10 years. Table 1.7 gives the rates o f child undernourishm ent in Indian states. U nderw eight (below -2 S.D) in India ranged from 63 percent in B ihar and 59 percent in U ttar Pradesh to about 28 percent in M eghalaya, K erala and M izoram . The highest rates o f stunting (below -2 S.D) w ere in U ttar Pradesh and Bihar, and the low est w as in Kerala. U ttar Pradesh w as also am ong the states w ith higher levels of wasting in India. The com parison o f K erala and U ttar Pradesh can be useful in evaluating the effects o f public policy and p eople’s organized action because these quality o f life indices w ere achieved by the two states when they had sim ilar per capita incom es and poverty levels ( see Tables 1.5, 1.6 and 1.7). It is interesting to note that K erala, with a p er capita incom e w hich w as 18 percent less than that o f India could overall, achieve substantially m ore in term s o f life expectancy, literacy, TFR, fem ale-m ale ratio, child death rates and nutritional levels o f children than all other states o f India. M uch o f these have been achieved through public policies and people’s involvem ent in ensuring their own collective welfare. A m ong the public actions that the state can take are rem oval o f traditional inequalities, prom otion o f literacy, provision o f health care and social security (Dreze and Sen 1995). W hile 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. incom e is im portant for health and w ell-being, p eople’s collective action can achieve a lot that can w eaken the correlation betw een incom e and health. The study o f K erala and U ttar Pradesh should help to bring out the effects o f public policy and p eople’s action in shaping the basic and underlying factors affecting the grow th o f children in India. 1.5 The organization of this study C hapter 2 deals w ith the m easurem ent o f grow th. The three indices used in this study, nam ely, w eight-for-age, height-for-age and w eight-for-height, are introduced. Som e o f the m ain sources reference data, nam ely, the U nited States N ational C enter for H ealth Statistics (N C H S) reference data, the H arvard reference data, the reference data based on studies done in the U nited K ingdom , the C anadian reference data, and the reference data recom m ended by the W orld H ealth O rganization w hich is based on the N C H S reference data are very briefly described. T he different m ethods used for com paring the distribution o f anthropom etric indices w ith the reference data, nam ely, percentiles, percent o f m edian and standard deviation scores; and the choice o f cu to ff points are briefly gone into. Finally, the controversies that raged in the m easurem ent o f undem utrition based on the view s different from the m ainstream view represented by the W orld H ealth O rganization (W H O 1983) expressed, m ainly by Sukhatm e and Seckler, are discussed. 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C hapter 3 contains a discussion, based on studies done in developed and developing countries, about the relative roles o f genetic and environm ental factors in determ ining child grow th, the grow th potential o f children o f different races living in different parts o f the w orld and the age and regional patterns o f grow th retardation. Finally, the likely causes for these patterns in grow th are discussed. C hapter 4 contains a discussion o f the underlying causes affecting child grow th in U ttar Pradesh and K erala. This chapter is based on studies done by researchers about U ttar Pradesh, K erala and other Indian states. The failure o f schooling in U ttar Pradesh and the success o f schooling in K erala are discussed. The public health system ; the arrangem ents for public distribution o f food and other essential item s o f daily consum ption; fem ale autonom y and the attitude tow ards w om en; and the im plem entation o f land reform s in the tw o states are com pared. The effect o f public policies adopted by the state governm ents; the im pact o f the involvem ent o f the people through com m unity and political organizations in ensuring that the state governm ents adopt policies favorable to them and public services function effectively; and the outcom es in term s o f child grow th and the health o f the population are sought to be brought out in this chapter. C hapter 5 gives an idea o f the m ethods adopted by the N ational Fam ily H ealth Survey o f India (1992-93) for data collection. The m ain variables that w ere used are described in som e detail w ith a view to give the reader an idea o f the 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. background situation regarding these variables in India. Finally the statistical model and the m ethodology adopted in this study are discussed. C hapter 6 presents and discusses the regression results. A n effort is m ade to exam ine the policy am enability o f variables from the point o f view o f reducing the levels o f undem utrition and prom oting norm al growth in children. T his is done by taking into consideration the odds ratios and the different values o f probabilities obtained by putting in different values for the independent variables in the probability equations. The final chapter has the conclusions o f the study and suggestions for policy to reduce the high rates o f underw eight, stunting and w asting am ong children in India. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 MEASUREMENT OF GROWTH A distinction is som etim es draw n betw een undem utrition and undernourishm ent (D reze and Sen 1989). The form er is seen in term s o f a shortage o f food intake and the latter as an unsatisfactory state o f health. T he form er im plies any nutrient deficiency, including food energy, protein and m icronutrients required by the body, the latter has to do w ith the health outcom e. M alnutrition is som etim es defined as undesirable physical or disease conditions that result from eating too little, too m uch or an unbalanced diet that does not contain all necessary nutrients required for good nutritional status. This m akes it sim ilar to undernourishm ent. A nother term used is protein-energy m alnutrition (PEM ), which has a m eaning sim ilar to that o f undernourishm ent, and m alnutrition. PEM im plies a condition produced by diet that is inadequate to supply the body w ith necessary energy, protein and m icronutrients. All these term s have sim ilar m eanings and results. For instance, undem utrition provides less than the m inim um requirem ent o f energy, protein and m icronutrients and the person is likely to end up m alnourished. In this dissertation w here the focus is on poor grow th in w eight and height, all three 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. term s are used to m ean a general deficiency in food intake, w hich is insufficient to m eet the basic requirem ents for body m aintenance, grow th and developm ent. G row th reflects nutritional status. P oor nutritional status leads to po o r grow th and is likely to be reflected in one or m ore o f the anthropom etric indicators considered in this study, nam ely, w eight-for-age, height-for-age and w eight-for-height. A distinction is som etim es draw n betw een stunting and stunted (W aterlow 1988). Stunting is considered to be the process o f becom ing stunted, a velocity term ; stunted is considered a distance term . In this dissertation, how ever, no such distinction is m ade. 2.1 Three approaches to measuring undernutrition There are three m ain approaches to m easuring undem utrition (Floud et al. 1990). The first is to m easure food intake and relate it to the requirem ents o f food. B ut the accurate m easurem ent o f intake is difficult because o f difficulties in quantifying the exact quantity and quality o f intake, recall problem s, and the large seasonal variations in the type and quantity o f food consum ed in developing countries (A cheson et al., 1980). N utritionists also em phasize the variations in requirem ents betw een individuals living and w orking in different conditions (Sukhatm e 1978; Sukhatm e (ed.) 1982; Sukhatm e and M argen 1982). C onsidering all 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. these aspects Payne (1987) concluded that direct m easurem ents o f intake to assess the nutritional situation w as very difficult in theory and practice. T he second m ethod o f m easuring undem utrition is by clinical examination. T he deficiency o f certain vitam ins and m inerals can be clinically identified. For instance, kw ashiorkor is an acute condition caused by very severe deficiency o f protein, com m only found am ong children o f ages tw o to four in developing countries. T his condition is identifiable by excess extracellular w ater, referred to as edema. M arasm us is another type o f severe condition o f m alnutrition caused by calorie deficiency. This results in em aciation. Such m easures o f severe undem utrition cannot how ever be used for com paring the nutritional statuses o f different sections o f a society (Floud, W atcher and G regory 1990). T he third m ethod o f m easuring undem utrition is by taking system atic m easurem ents o f the hum an body, know n as anthropom etry. A nthropom etric m easurem ents assess body size and com position and reflect the effects o f inadequate diet, insufficient exercise and disease (W H O 1995). A nthropom etry is the single m ost universally applicable m ethod that can be used for all ages, individuals and populations for assessm ent and for interventions designed to im prove health. It is useful for a w ide variety o f purposes including for prediction o f benefits from interventions, identification o f social and econom ic inequality and evaluation o f responses to interventions. In view o f the lim itations o f the first tw o m ethods, 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. anthropom etric indices are considered to be the best m easures for assessing the nutritional statuses o f individuals and populations (Floud et al. 1990; B uzina and U em era 1974). W eight, height, arm circum ference and skinfolds m easurem ents are recognized as the best anthropom etric m easures o f m alnutrition, both o f ovem utrition and o f undem utrition (C am eron, 1986). Sastry and Vijayaraghavan (1973) and the W orld Health O rganization (W aterlow et al. 1977) prefer the com parison o f height to age and w eight to height. O nly w eight and height were m easured in the N ational Fam ily H ealth Survey o f India. T his dissertation uses weight-for-age, height-for-age and w eight-for-height indices to assess undernourishm ent in India. 2.2 Anthropometric indices and indicators Since it is not possible to m easure nutritional status directly, anthropom etric indicators are used as an alternative m ethod for estim ating the true prevalence o f m alnutrition based on the assum ption o f a universally applicable standard (H abicht et al. 1974). Com parisons o f weight-for-age, height-for-age and w eight-for-height indices constructed out o f m easurem ents o f children, relative to a reference population, have becom e the m ost com m only used indicators o f undem utrition in children (W H O 1995). 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2.1 W eight-for-age W eight-for-age represents the sum o f protein, fat, w ater and bone m aterial. It reflects body m ass relative to chronological age and is the result o f linear grow th and w eight accum ulated both before and after birth, over the long term , as w ell in the short term (A CC-SCN , 1997). W eight-for-age in children is an index o f both chronic and acute m alnutrition and is w idely used to assess protein-energy m alnutrition and ovem utrition (G ibson 1990). H ow ever since underw eight represents the com posite index o f stunting and w asting it is difficult to interpret (W HO 1995). A s such it does not provide additional inform ation beyond that provided by stunting and wasting. H eight-for-age and w eight-for-age w ere found to be highly correlated. In surveys done in five developing countries, the correlation betw een height-for-age and w eight-for-age ranged between 0.6 and 0.87 in children under five years o f age (A nderson 1979). H ence, w hen the prevalence o f w asting in a population is low, w eight-for-age and height-for-age provide the sam e inform ation (W H O 1995). H ow ever w hen the prevalence o f w asting in a population is high, as is found in India, the prevalence o f underw eight does not necessarily parallel the prevalence o f stunting (W HO 1995). A m ajor deficiency o f w eight-for-age as an index o f protein-energy m alnutrition is that it does not account for height differences. C hildren w ith low w eight-for-age m ay not alw ays be w asted. T hey m ay be genetically short or stunted because o f disease or poor diet and m ay have a w eight appropriate to their short 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. height. In order to interpret w eight-for-age correctly it is necessary to know the exact age o f children, otherw ise accuracy o f this index is affected. W eight-for-age is a com m only used statistic in clinical settings to m onitor children on a longitudinal basis. C hildren w hose w eights are m ore than tw o standard deviations below the international reference m edian are considered underw eight and those m ore than three standard deviations below the reference m edian are considered severely underw eight. 2.2.2 Height-for-age Height-for-age is an index o f achieved linear grow th and a m easure o f linear grow th retardation. Stunting m ay be defined as the deficit in attained length or height com pared to international standards or any other appropriate reference (W aterlow 1988). Stunting is usually a chronic condition, w hich is the result o f extended periods o f reduced rates o f linear grow th caused by inadequate diet and high m orbidity. Therefore, T anner (1982) considered stunting as a proxy for the health o f populations. H eight-for-age is m ore stable than w eight-for-age and w eight-for- height. Illnesses and changes in diet affect height m uch less than w eight in the short run. A nthropom etric data in different com m unities around the w orld reveal a strong relationship betw een the heights o f children and adults in a com m unity, and the m easures o f socio-econom ic developm ent (W H O 1995). A s a country develops socially and econom ically, heights o f children and adults have been observed to 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increase. Eveleth and T anner (1990) described this as ‘the secular trend’. This occurred all over W estern Europe during the last one century and is still going on. In m any developing countries stunting is w idely prevalent as a result o f econom ic deprivation (M artorell and H abicht 1986; G opalan 1988; W H O 1995). W hile children o f higher socio-econom ic groups in developing countries have been attaining norm al heights, high levels o f stunting have been observed am ong children o f poor socio-econom ic groups (M artorell and H abicht 1986). The interpretation o f low height-for-age in a com m unity depends on its prevalence. I f prevalence is high, it m ay be safely assum ed th at m ost short children are stunted. I f prevalence is low, short stature m ay be the result o f genetic differences or factors particular to each child (W H O 1995). A deficit in height takes som e tim e to develop. Therefore a m easurem ent o f nutritional status taking only height-for-age into account m ay underestim ate undem utrition in infants (K eller et al. 1976). C hildren w hose heights are more than two standard deviations below the reference m edian are considered stunted and children m ore than three standard deviations below the reference m edian are considered severely stunted in this study. 2.2.3 W eight-for-height W eight-for-height is a sensitive index o f current nutritional status. W asting indicates a deficit in tissue and fat m ass (W HO 1986). W eight-for-height is relatively 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. independent o f age betw een one and ten years (G ibson 1990). It is particularly useful when the ages o f children are not available. W eight-for-height is also considered to be relatively independent o f ethnicity for children below 5 years (W aterlow et al. 1977). W asting can develop rapidly, but can be reversed ju st as rapidly with appropriate intervention. T he highest occurrence o f w asting has been observed during the w eaning period and to peak betw een one and tw o years o f age (W aterlow 1992; W H O 1995; W H O 1986). Percentage w asted indicates the prevalence o f acute m alnutrition in a population, w hich is generally caused by severe food shortages and high disease prevalence rates. South A sia and A frica have the highest rates o f w asting in the w orld (Som m erfelt and Stew art 1994). W eight-for-height is a useful index for evaluating the benefits o f nutritional interventions, as this is m ore sensitive to changes in nutritional status than height- for-age. O ne disadvantage o f this index is that it m ay classify stunted children as norm al. H ence it is preferable to com bine w eight-for-height w ith height-for-age as recom m ended by W aterlow et al. (1977). Prevalence o f w asting at 5% in a population is considered elevated by the W HO (1995). W asting is associated with im pairm ent o f m ental developm ent, w hich m ay be irreversible (G ibson 1990). C hildren w hose w eight-for-height are m ore than tw o standard deviations below the reference m edian are considered wasted and children m ore than three standard deviations below the reference m edian are considered severely wasted. 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2.4 W aterlow classification W aterlow (1972) m ade a cross-tabulation o f height-for-age and w eight-for- height called “W aterlow classification” (Table 2.1). F our broad categories o f m alnutrition w ere defined in this classification, nam ely, norm al, w asting, stunting, and stunting and w asting. This helps to m ake a distinction betw een children w ho are w asted, stunted and both w asted and stunted. Table 2.1 Waterlow Classification Height-for-age Degree o f Stunting W eight-for-height D egree o f W asting Percent (G rade) >90% 80-90% (0) (1) 70-80% 70% (2) (3) >90% (G rade = 0) 95-90% (G rade = 1) Norm al W asting 85-90% (G rade = 2) <85% (G rade = 3) Stunting Stunting and W asting Source: G ibson 1990 2.3 Choice of anthropometric indicators T he choice o f suitable anthropom etric indicators depends very m uch on the purpose for w hich they are used (H abicht and Pelletier 1990). F or instance low birth w eight m ay be used as an indicator o f m aternal m alnutrition in case o f m others who 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are thin. B ut low birth w eight is not an indicator o f m aternal m alnutrition in case o f overw eight m others. B ody m ass index (w eight divided by height squared) cannot be uniform ly applied to athletes and sedentary populations. I f an indicator is w rongly interpreted, it m ay lead to incorrect understanding o f conditions, w rong policies and m isdirected interventions. F or exam ple, if sm oking is the cause o f low birth weight, providing sm oking m others with supplem entary nutrition will be a w rong intervention. A nthropom etric indicators can be classified into different groups depending on the purpose for w hich they are to be used (W H O 1995). Som e o f the com m on purposes for w hich they are used are given below : Identification o f individuals or populations at risk: A n indicator m ay reflect present or future risk or both present and future risk. Severe underw eight m ay indicate present m alnutrition and risk o f death in the near future. Stunting that occurred in the past m ay not be a good predictor o f future risk. T he farther in the past the event that caused stunting, the less reliable the indicator becom es for prediction (B eaton et al. 1990). But w idespread prevalence o f stunting in a population m ay be a useful indicator for identifying areas for potential interventions. Som e indicators like obesity in children m ay not be good predictors o f future m ortality, but m ay be used in adults as indicators o f greater risk o f death. 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Selection o f individuals or populations for interventions: In this case, the indicator m ust predict the benefit to be derived from an intervention. For instance, low m aternal height m ay predict low birth w eight, but supplem entary feeding given to short pregnant w om en m ay not im prove birth w eight. H ence low m aternal height should not be used as an indicator to select m others for supplem entary feeding aim ed at increasing birth w eight. On the other hand, underw eight in pregnant w om en is likely to be a good indicator for selecting subjects for supplem entary feeding to reduce birth weight. A nthropom etry provides good indicators o f socio-econom ic status. Stunting am ong children is good indicator o f poor socio-econom ic backgrounds and a good indicator to identify populations for targeting developm ental program s. F or evaluation o f interventions: For this purpose, the indicators m ust reflect the response to past and present interventions. C hange in w eight-for-height can be a good indicator o f the effectiveness o f supplem entary feeding given to a m alnourished child. R eduction in levels o f stunting in a population is a good indicator o f socio econom ic developm ent. R eduction in low birth w eight w as found to be a good indicator o f a successful m alaria control program (B arbin 1991). A n aspect that should be kept in m ind w hile using anthropom etric indicators to evaluate responses to interventions, is the difference in response lags in individuals and populations for the sam e intervention. A w asted child m ay respond im m ediately to better feeding by 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. gaining w eight and quickly catching up in linear grow th. B ut such an increase in height m ay take decades to register in a population. Indicators that predict a risk m ay not be used to predict lack o f risk. For instance, indicators o f poor grow th w ere not found to be good predictors o f adequate grow th (Shekar et al. 1991). R elating anthropom etric indicators o f m alnutrition to the determ inants or consequences o f m alnutrition in the population requires careful distinction betw een causal and non- causal associations. For instance stunting in an elder sibling m ay be associated with stunting in a younger sibling, but there is no causal relationship betw een the two. T hese observations point to the need for caution in the use and interpretation o f anthropom etric indicators. 2.4 Anthropometric reference data 2.4.1 Justification for a com m on anthropom etric reference and creation o f reference data A grow th reference is a tool used for grouping and analyzing data (W H O 1995). It provides a com m on basis for com parisons but is different from a standard for growth. A standard is a norm or a desirable target and involves value judgm ent. Som etim es references are erroneously referred to and used as standards. H ow ever in m atters related to health and nutrition o f individuals and populations this is w idely 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. believed to be acceptable. This is based on the extensive evidence available that in populations, the effect o f ethnic differences in grow th o f children is sm all com pared w ith environm ental effects. M ean heights o f young children from m any affluent populations differ little across ethnic groups com pared w ith the socio-econom ic variability w ithin the sam e ethnic group (H abicht et al. 1974). H ence it w as considered appropriate to construct a com m on standard representing the grow th potential o f children all over in the w orld. T here are broad distributions found in attained heights and w eights am ong affluent populations, w hich appear to be o f genetic origin. B ut across m ost populations there are very sm all differences in m ean grow th in height or in its distribution across the m ean (W H O 1995). M any developing countries have not developed grow th curves o f their ow n. T hey use W H O (1983) reference data. The creation o f reference data involves detailed surveys, w hich require resources, time, and expertise, w hich are difficult for m any developing countries to m arshal. A nother difficulty is that m any developing countries have m ulti-ethnic societies and it is not practical to develop and use separate reference curves for each o f them . F or instance, G uha identified fifty-one ethnic groups in India (G uha 1944). A nother benefit from using a com m on reference standard to m easure grow th is that it enables the com parison o f hum an grow th in different countries. 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A reference has to be carefully chosen and has to be from a w ell-nourished population. To be considered a healthy population it has to m eet at least one o f the follow ing criteria or a com bination o f them (W H O 1995): T he population should live in a healthy environm ent. T he U nited States N C H S population m et this criterion. T he population should live in a healthy environm ent and should not as a rule contain sick individuals. T he population should live in a healthy environm ent, and longevity, or at least survival for som e years should prove their present good health, after the m easurem ents are taken. T he population should live in a healthy environm ent and consist o f only individuals w ho live according to certain prescriptions for healthy living. For instance, in the case o f infants and children, they should be breast-fed according to W H O recom m endations. F or reference data to be used as standards, certain critical conditions have been specified as desirable (W aterlow et al. 1977): T he sam ple should contain at least 200 individuals in each age and sex group. This is im portant for precision in calculating extrem e percentiles and Z-scores. In population-based studies this will allow differentiation betw een environm ental and genetic factors affecting growth. 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T he sam ple should be cross-sectional, since com parisons m ade w ould be cross-sectional. (H ow ever this is no longer necessary because longitudinal data can be presented cross-sectionally w ith m inor adjustm ents, even though, the reverse is not possible in the case o f grow th charts.) Sam pling procedures should be well defined and reproducible. T hose trained in anthropom etric techniques, using equipm ents w ith w ell- tested design and calibrated at frequent intervals should carefully record m easurem ents. The anthropom eric data should be available to anyone w ishing to use them , along with all relevant inform ation concerning instrum ent error, observer variability, m ethod o f im putation o f m issing data and data cleaning procedures used. A nthropom etric reference data m ay be collected locally from m easurem ents o f w ell-nourished children in good health. A lternatively reference data developed elsew here can be used. 2.4.2 Sources o f reference data for children The sources o f reference data for children based on w eight-for-age, height-for- age and weight-for-height are the following. 1. The U nited States N ational C enter for H ealth Statistics reference grow th data for children: These data w ere collected betw een 1960 and 1975. T he Fels R esearch Institute, O hio collected data for children from birth to 36 m onths by taking 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m easurem ents o f 720 w hite children attending the Institute. F or children 2 to 18 years o f age the data w ere collected through health and nutrition surveys o f all ethnic, geographic and incom e groups living in different regions o f the United States (Gibson 1990). Percentile curves, and m eans and standard deviations for w eight, stature and w eight for stature are available as charts and tables from N C H S. The C enter for D isease Control has m ade available com puter program s that use norm alized curves to calculate anthropom etric indices for children from w eight, height, sex and age data (D ibley et al. 1987). 2. The H arvard reference grow th data for children: These data w ere derived in the 1930s and early 1940s from a sm all cross-sectional sam ple o f w hite m iddle-class A m erican children. The data provide w eight-for-age, stature-for-age and w eight-for- stature percentiles for boys and girls, both separately and com bined from birth to 36 m onths o f age, also are provided sex-specific w eight-for-age and stature-for-age data for children aged tw o to eighteen years o f age. T he 50th percentile values o f this reference are very sim ilar to the 50th percentile o f the N C H S reference (Stephenson et al. 1983). Fifty percent o f the children in the reference distribution fall below the 50th percentile value and 50 percent fall above this value. 3. The U.K. reference data: T anner et al. (1966) com piled sm oothed percentile curves for children in the U nited K ingdom using both cross-sectional and longitudinal data. These w ere later revised in 1976. 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4. The Canadian reference data: T hese data w ere generated from cross-sectional data collected betw een 1970 and 1972 from a nationally representative sam ple. Percentile values for weight-for-age, height-for-age and weight-for-height are available for children from birth to 19 years o f age (G ibson 1990). 5. The W orld H ealth O rganization (W H O ) reference data: The W H O in 1966 adopted a com bined-sexes version o f the H arvard grow th curves as the international reference curve. A fter a decade this w as considered inadequate. A new W HO reference data for children w ere recalculated from the US N C H S data, intended for use as international reference by countries that do not have their ow n reference data (W H O 1978, 1983). T he second publication (W H O 1983) had m ore detailed percentile and m edian and standard deviation tables. Several percentile values for weight-for-age, height-for-age and w eight-for-height, the m edian and + 1 ,2 , and 3 standard deviations w ere available for prepubertal children. W H O recom m ends use o f these tables only for children 10 years or younger because o f differences in the age o f onset o f puberty in different populations. Stephenson et al. (1983) com pared the Harvard, NCH S and W H O references for w eight-for-age, stature-for-age and w eight- for-stature and found that the differences betw een these data for m ean and m edian w eights and statures w ere sm all. 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4.3 M aking com parisons using reference data To evaluate anthropom etric indices calculated for any population, an associated reference data for each index is required. The W orld H ealth Organization (W H O 1983) recom m ends the com parison o f distributions o f data derived from nutritional surveys o f populations w ith th at o f the N C H S distributions. C utoff points can be determ ined to identify and classify m alnourished individuals and to trigger interventions based on criteria such as the severity o f m alnutrition and resources available. In population studies, the distribution o f each anthropom etric index m ay be com pared using percentiles, standard deviation scores or percent o f m edian derived from the reference data. The distribution o f the population surveyed can be usefully presented as a table or a graph for highlighting its salient features. 1. Percentiles: Percentile is the rank position o f an individual on a given reference distribution stated in term s o f w hat percentage o f the group the individual equals or exceeds (D ibley et al. 1987). F or a G aussian distribution, the 50lh percentile represents both the m ean and the m edian. In a skew ed distribution the 50th percentile represents only the m edian. O ne disadvantage o f using percentiles derived from populations in developed countries fo r evaluating anthropom etric indices in developed countries is that m uch o f the study population m ay fall below the lowest 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reference percentile o f the reference population m aking it difficult to accurately classify a large num ber o f cases (W aterlow et al. 1977). 2. Percent o f m edian: Percent o f m edian is the ratio o f a m easured value in the individual, say weight, to the median value o f the reference data for the sam e age or height, expressed as a percentage. 3. Standard deviation scores: This m ethod calculates the deviation o f the m easurem ents from the reference m edian in term s o f standard deviations or Z scores. The standard deviation score calculated for each individual is the deviation o f the value for the individual from the m edian value o f the reference population divided by the standard deviation for the reference population. This Z score gives the individual’s value with respect to the distribution o f the reference population. W H O (1983) has published the standard deviation values for calculating the Z scores from the original N C H S data, the details o f w hich are given in D ibley et al. (1987). One advantage o f this m ethod is that Z scores can be defined beyond the lim its o f the reference data. This enables the accurate classification o f individuals beyond the extrem e percentiles o f the reference data. There are usually a large num ber o f cases in developing countries w ith indices below the low est percentile in the reference population. H ence this m ethod is suitable for use in developing countries. Standard deviation scores have the advantage o f being identical across age and sex, and for all m easurem ents o f underw eight, stunting and w asting (W aterlow et al. 1977). 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4.4 C u to ff points, sensitivity and specificity C utoff points are based on the relationship betw een nutritional assessm ent indices and functional im pairm ent or clinical signs o f deficiency or both. H owever, often, inform ation that can relate indices to deficiencies is not readily available. M oreover there is alw ays a m argin o f error and risk o f m isclassification because o f biological differences am ong individuals (Beaton 1986). T he cu to ff points influence the sensitivity and specificity o f indices. Sensitivity o f an index refers to the ability o f the index to identify and classify individuals in a population w ho are genuinely m alnourished. A n indicator with a 100% sensitivity w ill correctly identify all genuinely m alnourished individuals; no m alnourished individual will be classified as not m alnourished. In other w ords, there will be no false negatives. Specificity o f an index refers to the ability o f an index to identify and classify all genuinely well- nourished individuals. An indicator w ith a 100% specificity w ill correctly identify all well-nourished individuals; no w ell-nourished individual will be classified as m alnourished. T hat is, there will be no false positives. T he ideal index w ill have high specificity and sensitivity. If low er values o f an index are associated w ith m alnutrition, assigning a low er value to a cutoff point w ill increase specificity and decrease sensitivity. Predictive value o f an index o f m alnutrition is the ability o f an index to correctly predict the presence or absence o f m alnutrition. Positive 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. predictive value is the proportion o f positive (m alnourished) identifications that are true. N egative predictive value is the proportion o f negative (w ell-nourished) identifications that are true. Predictive value depends on sensitivity and specificity and the prevalence o f m alnutrition. Sensitivity and specificity rem aining the same, positive predictive value goes up with higher prevalence. 2.5 Controversies in the assessment of undernutrition T here w ere tw o controversies that arose about estim ating undem utrition. T hese concerned the determ ination o f standards for the level o f activity, body size and energy intake desired for a healthy hum an being. D ivergent view s existed am ong the traditional or establishm ent theorists on the one hand (W H O 1983; N utrition C anada 1980; W aterlow et al. 1977; Tanner 1978), and P.V. Sukhatm e and David Seckler and others, on the other (Sukhatm e 1978, 1982; Seckler 1980,1982; Pacey and Payne 1985). Sukhatm e and Seckler argued that traditional standards used for m easuring undem utrition resulted in exaggerated estim ates o f undem utrition. It is generally agreed that the hum an body requires energy for perform ing its internal and external functions and this energy is m ostly provided by food consum ed daily. B ut if at any tim e the energy provided by food consum ed is not sufficient to perform daily functions, the body utilizes reserves o f energy stored in the body. C onversely, w hen the needs o f energy are less than w hat is provided by food 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. consum ed, the body stores the excess. W hen there is an im balance betw een intake and expenditure, variations in energy stored occur, usually m anifested in changes in body w eight. This also results in changes in the com position o f the body m ass, such as the proportion o f fat and lean body mass. T he concept o f energy balance is linked to the level o f functions that are required to be supported. I f only survival w as at issue, the energy intake required would be m uch lower than if higher levels o f activity w ere desired. H ence there arose the need to set a standard for the desirable level o f activity. T he U N expert com m ittee on nutritional standards offered the follow ing definition (FA O /W H O /U N U 1985: 12): “The energy requirem ent o f an individual is that level o f energy intake from food which will balance energy expenditure w hen the individual has a body size and com position and level o f physical activity consistent with long-term good health and which will allow for the m aintenance o f econom ically necessary and socially desirable physical activity.” H ow ever there w as no single level o f physical activity and body size that w as considered appropriate. Even if these can be uniquely determ ined, will the energy requirem ent be the sam e for all individuals? T he traditional view was that all these can be uniquely determ ined, w hereas the heretical view expressed by Sukhatm e, Seckler and others w as that they cannot be so determ ined (Osm ani 1992). These two view s w ere also described as the ‘genetic potential’ view and the ‘individual adaptability’ view respectively (Pacey 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and Payne 1985). The establishm ent view w as that the efficiency o f m etabolism (conversion o f stored fat and protein into energy required for w ork) w as m ore or less fixed for all individuals w hen body w eight is held constant, as was done for the reference person. T hey also held the level o f activity constant for the reference person. W ith these assum ptions, the energy expenditure and requirem ent also remained fixed. The traditional approach recom m ended the use o f grow th curves derived from w ell-nourished children in the developed w orld as the standard for com paring observed m easurem ents o f children from all over the w orld (W H O 1983; W aterlow et al. 1977; T anner 1978). T he two im plicit assum ptions m ade in doing so are: (1) there is no difference in genetic potential o f different races in the w orld, and (2) any grow th below the genetic potential determ ined by the reference standard m ust be the result o f undem utrition (O sm ani 1992). The first assum ption w as supported by strong em pirical evidence, which has com e to be w idely accepted (H abicht et al. 1974, M artorell 1984, R oberts 1985; M artorell and H abicht 1986). B ut the second assum ption, w hich is the core o f the genetic potential theory, gave rise to the controversies (O sm ani 1992). Sukhatm e and Seckler m aintained that w here the establishm ent theorists w ent w rong was in understanding the biological response to the nutritional environm ent. 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5.1 T h e ‘R ange o f H om eostasis’ controversy Sukhatm e’s theory w as concerned w ith the intake-based m easure o f undem utrition and the fixed norm set by traditional theorists to com pare observed levels on intake. He contended that even if desired body w eight and activity level w ere taken as fixed, different individuals had different efficiencies o f m etabolism because o f genetic differences am ong individuals. Traditional theorists, who took an average rate o f m etabolic efficiency to get over this problem , conceded that such genetic differences existed. H ow ever, Sukhatm e w ent a step further and argued that the efficiency o f m etabolism also varied for the sam e individual over tim e. His hypothesis w as that costless variation in energy utilization perm itted a person to m aintain the sam e state o f health under a fairly w ide range o f intake. W ithin this range o f intra-individual variation in requirem ent the individual can rem ain in a state o f energy equilibrium and good health. Sukhatm e contented that if this w as true, the traditional standard exaggerated the m agnitude o f undem utrition. H ence he argued in favor o f a 'range o f hom eostasis' or a range o f variation in requirem ent o f intake for the sam e individual and suggested keeping the cu to ff point at the low er end o f this range to estim ate undem utrition. W hile the traditional view took the average requirem ent o f reference individuals who w ere persons in good health as the cutoff, S ukhatm e proposed tw o standard deviations below that average as th e cutoff. U sing this low er cu to ff Sukhatm e (1982) estim ated the level o f nutritional poverty in India 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to be betw een 15% and 20% . This was less than 50% o f the 46% estim ated by D andekar and Rath (1971) for India. The severely m alnourished w ere only 5 per cent am ong children according to this calculation. Sukhatm e’s view s generated a great deal o f controversy over the concept and m easurem ent o f undem utrition. Gopalan (1980) w as o f the view that only about 15 percent o f the children in India w ere w ell- nourished, im plying that about 85 per cent o f them w ere m alnourished. H e argued that perm anent adaptation to a low er level o f intake by those obliged to do so cannot be costless in term s o f functional im pairm ent. I f the body responds to a reduction in energy intake by burning body fat in excess, this cannot go on indefinitely without reducing the functional capabilities o f the individual, as claim ed by Sukhatm e (G opalan 1983). C ritics o f Sukhatm e w ere o f the view that there w as no evidence available to substantiate his views. 2.5.2 T he ‘Small B ut H ealthy’ controversy The controversy that was generated by Seckler concerned the use o f fixed anthropom etric standards by traditional theorists to m easure undem utrition (Seckler 1982, 1984a). A ccording to him the crucial response m echanism to changes in nutritional conditions w as the variation in the rate o f body grow th. H is hypothesis w as that different states o f grow th w ere consistent w ith a desirable state o f health. H ence the use o f a fixed anthropom etric standard for m easuring undem utrition was 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. inappropriate and it overestim ated the m agnitude o f undem utrition in developing countries. Seckler coined the term 'sm all but healthy' (SB H ) and argued that small people can also be healthy. Seckler argued that, w ithin limits, deviations from the genetic potential did not entail any functional im pairm ent (Pacey and Payne 1985). Sim ilar opinions that sm allness can be advantageous in conditions o f nutritional constraints w ere expressed earlier also (T anner 1978; G oldstein and T anner 1980). It is generally agreed that a sm aller body norm ally required lesser energy to survive and perform work. But Seckler w ent a step further and stated that a body that is constrained in grow th can still avoid all kinds o f functional disabilities. Initially he included all cases o f m ild to m oderate m alnutrition in SBH (Seckler 1980,1982), but later, responding to criticism , included only the moderately stunted and not wasted at all in SBH (Seckler 1984a, 1984b). S eckler’s hypothesis w as that when nutritional intake is inadequate for attaining genetic potential, stunting is an adaptive m echanism adopted by the body to adjust to reduced nutritional supply by reducing nutritional dem and (Seckler 1980). H e also asserted that m oderate stunting does not result in any functional im pairm ent or harm. But this assertion w as not backed up by evidence (O sm ani 1992). Critics disagreed w ith the view that good health and the sam e level o f functionings were possible with a small and a large body size. 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5.3 Sum m ing up the controversy D uring the course o f the controversy, proponents on both the sides traded som e unfair criticism s. It w as charged that S eckler’s view w as that sm allness was good in itself or desirable for its own sake. B ut Seckler did not hold such a view. He said that even the sm all could be as healthy as those larger than them selves. H eretics argued that traditional theorists im plied unlim ited feeding and this could lead to consequences such as obesity and shortened life span (Payne 1985; Payne and C utler 1984; Pacey and Payne 1985). This criticism , how ever, w as unfair. The traditional view advocated the achievem ent o f genetic potential in height, w eight and functionings, and not in accum ulation o f fat by unlim ited feeding (O sm ani 1992). Seckler’s argum ent that genetic potential theory was not falsifiable according to Popperian criterion and hence not scientific w as not valid; the FA O /W H O /U N U (1985) hypothesis that all functions are sim ultaneously m axim ized only w hen genetic potential is achieved is falsifiable (Osm ani, 1992). W aterlow (1988) stated that in societies w here earning capacity depends on height, as w as show n in a study done by S atyanarayana et al. (1986), poverty m ay the result o f shorter height than the other w ay around. B eaton (1989) noted the failure o f adaptationists establish their claim s about m etabolic adaptation. D reze and Sen (1989) noted som e o f the considerations that w ould m ake view ing small size as som ething that has no serious consequences: 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. But as know ledge stands at the m om ent, to dism iss sm allness as entirely costless w ould seem to be a dubious and prem ature position to take in the context o f nutritional deprivation across the w orld. Indeed, given (1) m edical inform ation on the observed relation betw een height, weight, m orbidity, and learning (especially am ong children, and particularly for cases o f severe stunting), (2) the econom ic inform ation on the relevance o f height for em ploym ent in som e occupations and strength in others, (3) the social inform ation on the relevance o f height in m oving up or dow n the social ladder, and (4) the cultural inform ation on peo p le’s ow n view o f their height, w eight, and strength, it w ould be difficult to view sm allness o f body size as being, in general, inconsequential and unproblem atic. F urther studies have established that even m ild to m oderate undem utrition m anifested as reduced grow th can have m easurable and im portant functional consequences. Protein-energy m alnutrition and even relatively m ild deficiencies o f m icronutrients w ere found to reduce resistance to infections and lead to a range o f related consequences adversely affecting child nutritional status and grow th (C handra 1997). A logistic regression study controlling for proxim ate and m aternal variables affecting child grow th found that anem ia w as associated w ith m ild to m oderate m ental retardation in early childhood (H urtado et al 1999). M any studies have show n that children in developing countries have the sam e grow th potential as children in developed countries; and the W H O reference standards have com e to be w idely used for assessing the grow th o f children in developing countries Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Latham 1997). The serious consequences o f grow th faltering in early childhood ju stify all possible public policy action to prevent it. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 GROWTH POTENTIAL AND GROWTH PATTERNS W H O recom m ended the use o f its reference for assessing the nutritional status o f children from birth to ten years o f age (W H O 1979, 1983). This recom m endation was based on the prem ise that differences in grow th potential am ong ethnic groups are m inor during childhood, and notable differences in height em erge after adolescence (M artorell and H abicht 1986). C lear patterns have been found in the grow th o f children based on age and in different regions in the world. 3.1 Socio-economic status and growth potential M any researchers have attem pted to assess the grow th potential o f children in different parts o f the world. They have done so by selecting children w ho were from b etter-off socio-econom ic groups w hose grow th w as likely to be unconstrained by environm ental factors. Farquharson (1976) studied N epalese children attending private clinics, private schools and children o f health w orkers. T he children’s grow th potential w as m uch higher than that o f their poorer counterparts. Johnston et al. (1976) studied G uatem alan children studying in an elite private school. Their grow th w as close to N C H S norm s. O ther studies o f Latin A m erican pre-adolescent children 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in w ell-to-do fam ilies found sim ilar results (Johnston et al. 1973; Villarijos et al. 1971). Studies o f black children living in A frica (Janes 1974; G raciter and G entry 1981), in the U nited States (H am ill et al. 1970; G am and C lark 1976) and elsew here in N orth and South A m erica (K ing et al. 1963; A shcroft et al. 1966) found their height potential equal to or slightly above children o f E uropean descent. W ell- nourished children in India (Vijaya Raghavan et al. 1971; D atta Banik 1982) were found to grow in height and w eight equal to European children. Eveleth and Tanner (1976) considered that differences in child grow th in India w ere m ainly the result o f socio-econom ic differences rather than due to ethnic variations. O ther studies that have brought out the effect o f socio-econom ic differences in child grow th in India are Udani (1963), M ukerji and Kaul (1970), Sikri (1972), Sw am inathan (1971), Agarwal et al. (1974) and D atta et Al. 1970). M any studies had found children o f F ar East A sian origin to be shorter than children o f other racial groups (G reulich 1957; Tanner et al. 1982; C hang et al. 1965; A shcroft et al. 1966; W heeler and Tan 1983; Kim 1982), near the 25lh percentile o f N C H S grow th curves (M artorell and H abicht 1986). B ut w ith better environm ental conditions differences in heights have narrow ed. H eight differences between Japanese children brought up in the U nited States and Japanese children grow ing up in Japan disappeared by m id- 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. STATURE I N CM Figure 3.1 M ean heights of Japanese boys in 1957 ('•'). 1967 (x) and 1977 (o) plotted on NCHS growth curves. 190 95 90 75 50 25 180 170 160 150 140 150 120 0 100 90 80 j . x i J___L 1 AGE IN YEA R S Source: M artorell and Habicht 1986. 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.2 Mean heights of 7-vear-old bovs of high (V) and low (o^. socio economic status in various developing countries. 123 N C O m C O a O C C O + - > Vi o O JO C O E 0) 4 - > C O o CD C O X C O o 'co E to JO "C Q ) C O 05 -0 z £ 05 c o 05 c o X c / 5 C D C O X C C J < u z E a? a. 121 —50th 119 E o -25th £ 117 .05 ‘< 1 5 sz § 115 05 - -10th 113 —5th 1 1 1 109 Source: M artorell 1985. 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1970s as a result o f im provem ents in nutritional conditions in Japan G reulich (1976). Figure 3.1 based on data from Tanner et al. (1982) show s the secular trend in heights o f Japanese boys. H eights o f six-year old Japanese boys in 1957 w ere betw een the N C H S 5th and 10th percentiles, but in 1977 they w ere betw een the 25th and 50th percentiles (T anner et al. 1982). Eveleth and T anner (1976, 1990) have sum m arized m ost o f the grow th studies conducted around the w orld. T hey cam e to the conclusion that a considerable proportion o f the m ean differences in body size am ong populations they exam ined was due to the effects o f environm ental conditions. W hile difference betw een the w ell-to-do and the less w ell-to-do in developed countries w as relatively sm all, the difference betw een the poor and the rich in developing countries w as m uch greater. This w as also the case in Europe earlier in the 19th century. In the w ords o f Floud, W atcher and G regory (1990), “It was only m ildly flippant to assert that, in the early nineteenth century, upper classes o f France and B ritain could literally look down on the low er classes.” The differences in size according to class reduced after the basic needs o f alm ost all households w ere taken care o f in European countries. The differences in height betw een socio-econom ic and ethnic groups are brought out clearly in Figure 3.2. The figure illustrates the influence o f genetic factors and the environm ent on stature. The solid dots represent high and the em pty 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. dots low socio-econom ic groups. On the right hand side percentiles from NCH S survey are given. T he height differentials betw een the dark circles show that differences in height betw een children o f different ethnic groups, controlling for socio-econom ic status, are relatively small. T he distance betw een the dark and em pty circles show that height differentials am ong children o f high and low socio econom ic groups are indeed very large, com ing to about two and a h a lf times the height differentials due to genetic differences. M ost o f the low er class children hover around the N C H S 10th percentile. Indian and G uatem alan children are below the range o f variation o f the N C H S charts. A recent m ultivariate analysis found that household econom ic status had significant effects on stunting o f pre-school children in India (M ishra and R etherford 2000). T his show s that environm ental factors far outw eigh genetic effects in determ ining height. The results o f m any studies support the relationship betw een socio-econom ic status and height. O w nership o f land other assets for food production (V alverde et al. 1977), quantity o f food produced by the fam ily (M artorell et al. 1984), w ealth m easured in term s o f the quality o f house construction and fam ily incom e (Levinson 1974; D evadas et al. 1980; V alverde et al. 1981) w ere found to be positively associated w ith height. M artorell et al. (1983) found that in urban and rural H onduras, the am ount o f w ealth m easured in term s o f the quality o f house construction and the num ber o f room s in the house w ere positively related to the height o f children. Fifty-eight percent o f the children living 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the poorest quality o f housing w ere stunted w hile only 19% o f children living in relatively better hom es w ere stunted. T he consensus that has em erged is that children o f different ethnic groups at prepubertal ages exhibit very sim ilar grow th potential and grow th patterns all over the w orld (M artorell and H abicht 1986). Therefore the use o f an international standard for evaluating the adequacy o f children’s grow th was justified. A t the sam e tim e, it is advisable to investigate the relevance o f a grow th standard to populations under study. This is because the use o f grow th curves developed from a population with a greater grow th potential to assess nutritional status o f children w ith a low er growth potential can exaggerate the extent o f undem utrition. In India, a scientific report from the N utrition Foundation o f India found that the W H O standard was applicable to Indian children (Agarwal et al. 1991). In all, 1429 boys and 1206 girls below age 6 living in 7 cities in 5 regions o f India and not subject to any obvious socio-econom ic and environm ental constraints on grow th w ere selected. T he grow th perform ance in tw o cities, nam ely, Ludhiana and D elhi nearly corresponded to the grow th perform ance represented by the W H O /N CH S international reference population. In other regions the grow th observed w as lesser. H ow ever, considering that the secular trend in grow th w as still under w ay the study expected that the gap to be gradually elim inated betw een the regions. The study attributed the differences in grow th betw een regions to environm ental rather than ethnic or genetic factors. 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 Age and regional patterns of growth retardation Children require adequate nutrition to undertake physical activity, grow , and com bat disease and other form s o f stress. W hen nutrition available is insufficient for these, the children’s grow th slow s down. U sually w eight is affected first, and then height and other indices o f growth. W hen favorable conditions return, a spurt in grow th occurs as the body attem pts to return to the grow th path it w as follow ing previously. This is called catch-up grow th (Tanner 1978; Floud, W atcher and G regory 1990). It is believed that the capacity for catch-up grow th exists till the end o f adolescent grow th spurt (FA O /W H O /U N U 1985). W aterlow (1993) and G olden (1988) review ed several studies that reported catch-up grow th w hen nutritional supplem entation w as given to m alnourished children or they w ere adopted by w ell- to-do families. C ertain age patterns in grow th retardation have been observed in children living in poor environm ents in developing countries. T hey w ere found to grow norm ally in height during the first tw o to three m onths o f life. A fter that, a decline in their rate o f grow th in height and w eight w as observed, w ith the decline becom ing sharp at about six m onths (M artorell and H abicht 1986). Figure 3.3 illustrates the age pattern in linear growth. This figure is based on cross-sectional data from a sam ple o f boys living in slum s in Tegucigalpa, H onduras. The m ean lengths o f 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. children for the first three m onths o f life w ere equal to the N C H S m edian represented by the 50th percentile curve. D uring the next three m onths it w as slightly below the N C H S m edian. A fter the sixth m onth, length fell increasingly and precipitously, and w as below the N C H S tenth percentile by the end o f the first year o f life. The pattern seen in F igure 3.3 did not vary by sex. M any studies have reported the absence o f significant grow th differences based on sex in Latin A m erica (M artorell and H abicht 1986). Figure 3.4 illustrates the effect o f age in years on height and w eight-for- height. This is based on cross-sectional data o f children in rural H onduras for the first seven years o f life. H eight drops from NCH S m edian at a rapid rate during the first year o f life, at a gradual rate during the second year o f life and returns to the norm al N C H S rate o f grow th by the third year o f life. M any longitudinal studies have found this pattern in G uatem ala, Jam aica, G am bia and U ganda (M artorell and H abicht 1986). Figure 3.4 also shows the absence o f sex based differentials in stunting and w asting. The children w ere som ew hat “chubbier” in infancy and a bit leaner during the second year o f life com pared to N C H S data. N o w asting w as found except during the second year o f life. This w as in agreem ent w ith th e general pattern observed in Latin A m erica w here children had norm al w eight for height values along w ith a m oderate level o f stunting 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.3 Mean length of 459 Honduran male infants relative to NCHS percentiles 80 n 90th 50th 75- 10th 70- £ 60- 55- 50- 45- 0 1 2 3 4 5 6 7 8 9 10 11 12 Age (months) Source: M artorell and H abicht 1986. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.4 Weight-for-height and height-for-age z scores for children aged 0-7 years in Honduras -0 Girls (n = 4 8 1 ) or 0.0 W EIGHT FOR HEIGHT a > L . 0 u I/I 1 N H EIG HT - 2.0 -3 .0 0 3 2 4 5 6 7 AGE (Years) Source: M artorell and H abicht 1986. 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Y arbrough et al., 1975; M ora et al., 1981; D elgado et al., 1983 quoted in M artorell and H abicht 1986). M uch higher levels o f stunting and w asting w ere found in B angladesh (B row n et al. 1982), N epal (M artorell et al. 1984) and India (Rao et al. 1959; UPS 1995). M any studies have supported the hypothesis that m uch o f the growth retardation occurs during the first tw o to three years o f life. A t age three or four norm al growth is resumed. Calloway et al. (1992) found evidence for this in data from Egypt, K enya and M exico from the N utrition C ollaborative Research Program (CRSP). Further, it has been found that failure to grow in height in early childhood is rarely m ade up later as adolescents. Table 3.1 presents the results o f a study in rural Hyderabad, India (Satyanarayana et al. 1980). T he sam e boys w ere m easured at the ages o f five and seventeen and a half. They all made sim ilar gains in height during these ages, irrespective o f the level of deficits observed at age five. But the boys who were shorter at age five did not catch up with those who were taller at age five. The average increase in height between five and seventeen and a half years o f age was 62 cm , which was very close to the NCHS norm s (67 cm). H ow ever, those who had greater weights at age five gained greater w eight between ages five and seventeen and a half. Also, unlike in the case o f heights, the average w eight gains were (betw een 31.6 kg and 24.3 kg) far below the NCHS average (49 kg). 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1 Height and weight of children at 5 years and 17.5 years in rural India G roups N um ber In 1965 A t age 5 H eight W eight In 1978 At age 17.5 Height W eight Growth between 5 and 17.5 years (cm) (kg) I 23 5.04 (cm) 105.0 (kg) 15.0 17.43 (cm) 164.5 (cm) 46.1 60.5 31.6 II 25 5.22 99.8 13.8 17.45 160.1 42.4 61.6 29.3 III 30 5.14 95.6 12.8 17.41 157.0 42.0 62.6 29.8 IV 14 5.16 88.5 11.1 17.39 149.0 34.8 62.2 24.3 Pooled 0.32 3.21 1.27 0.28 6.40 5.30 6.23 5.25 SD F ratio 1.16 85.45 30.32 0.30 16.12 11.84 0.47 5.14 P-value 0.05 0.001 0.001 0.05 0.001 0.001 0.05 0.01 Note: The groups were defined on the basis o f height at 5 years o f age and in reference to the Boston median and standard deviation as follows: Group I, values within the range o f 0 to - 2 SD; Group II, -2 to - 3 SD; Group III, -3 to — 4 SD; Group IV, less than — 4 SD. Mean Z scores (based on NCHS reference data) were -1 .8 8 , -2.25, -3.03, and -4 .2 6 , respectively, at age 5 and -1 .0 7 , -2.20, -3.11 and -4 .6 5 , respectively, at 17.5 years o f age. Source: Satyanarayana et al. 1980 as oo M any studies have found that failure to grow in early childhood is rarely m ade up later in developing countries. A longitudinal study in W est B engal, India, m easured m iddle-class boys and girls for fourteen years (H auspie et al.1980). The study concluded that differences in adult height w ere m ostly the product o f prepubertal growth. B illew icz and M cG regor (1982) in G am bia and (M artorell et al. 1990) in G uatem ala obtained sim ilar results. T hese findings support the conclusion that height and w eight deficits occurring in early life determ ine adult size. M ost researchers believe that the m ain reason for this is that children continue to live in the sam e poor environm ents that caused the initial stunting (M artorell 1995; O sm ani 1992; W aterlow 1993; M artorell and H abicht 1986). Figure 3.5 presents a com parative picture o f such w asting patterns in Latin A m erica and South Asia. The data pertain to children living in poor households in H onduras, G uatem ala and Bangladesh. The w eight-for-height data are show n as percent w eight-for-height (i.e., actual w eight as a percentage o f the expected NCHS w eight m edian for children o f the sam e height and sex). All three data w ere reported for both sexes together. In the Central A m erican countries, w eight-for-height started o ff at higher levels than in B angladesh, but in both the regions there w as a sharp decline in w eight-for-height in infancy. W eight-for-height rem ained low er than the N C H S reference all along in Bangladesh. B y age three, children in Honduras and G uatem ala w ere very near the N C H S m edian, w hereas in B angladesh, they were 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. WEIGHT FOR HEIGHT (%) Figure 3.5 The age patterns of wasting in Honduras. Guatemala and Bangladesh •—•—• Honduros 0 -0 --O Guoiemolo Banglodesh 105 - 0 100 ■ o r 90 0 1 2 3 4 5 6 AGE (Years) Source: M artorell and Habicht 1986. 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. still below ninety percent o f m edian w eight-for-height. In India and neighboring countries children o f poorer households w ere found to rapidly lose w eight as shown in Figure 3.5, w hereas children o f richer households rem ained close to the NCHS m edian longer, but by one and a h alf years even they w ere at about ninety-six percent o f the NCHS m edian w eight-for-height or 0.5 S.D below the N C H S m edian (M artorell and H abicht 1986 based on data from R ao et al. 1959; D atta B anik et al. 1970; Rao et al. 1976). The phenom enon o f rapid decline in w eight in infancy w as reported in developing countries all over the w orld w ith the low est figures for w eight-for-height during the second year o f life as show n in Figure 3.5 (M artorell and H abicht 1986). A fter analyzing DHS (1986-89) data for nineteen developing countries o f A sia, A frica and Latin A m erica Som m erfelt and Stew art (1994) identified three m ain patterns in stunting, w asting and underw eight across these regions o f the world. C ountries in Latin A m erica, the C arribean and N orth A frica (also U ganda and Z im babw e in Sub-Saharan A frica) had levels o f stunting that w ere m uch higher than levels o f underw eight. B ut the proportion classified as w asted w as sim ilar to that seen in the International R eference Population. In som e countries w here w asting is com m on, like Sri Lanka and M ali, there w as both stunting and underw eight. Levels o f stunting and underw eight in India w ere higher than in either Sri L anka or Mali. 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The third pattern seen m ostly in Sub-Saharan A frican countries w as an interm ediate one, w ith substantial levels o f stunting and w asting. The prevalence o f stunting and underw eight w ere very sim ilar there with the exception o f U ganda and Zim babw e. 3.3 Causes for age patterns in growth The exact ways in w hich undem utrition affects grow th and functional capabilities o f the m ind and body are not precisely understood (Floud, W atcher and G regory 1990; W aterlow J.C. 1988). Hence it is not possible to determ ine the exact contributions o f nutrition and infection, and their interactions to retarded growth. H ow ever, there is enough evidence to suggest that the patterns o f w eight and height retardation during the first tw o years o f life in developing countries is due to the inadequate diet and the poor environm ent that increase the risk o f infections. Som e o f the m ain reasons for the age patterns observed are as follows: 3.3.1 Problem s related to w eaning U ndem utrition usually sets in w hen breast m ilk is supplem ented by other foods. T his is the tim e w hen children need m ore food for grow th, frequent feeding is required, and they are dependent on others to feed them (Pacey and Payne 1985; G urney 1979). This is also the tim e when toddlers begin to m ove around and are m ore exposed to infections. Prior to about six m onths, breast m ilk alone is adequate 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for norm al grow th and children have little other food. D uring this period they are protected by the anti-infective properties o f breast m ilk. A fter six m onths breast m ilk is not adequate, and com plem entary foods are given to the child (Scrim shaw and U nderw ood 1980). T hese foods becom e a m ajor source o f infections (Floud, G regory and W atcher 1990). U nclean cooking vessels, lack o f hygiene on the part o f those preparing food, contam inated food and w ater, incom plete cooking that leave pathogens alive, exposure o f food to flies, and proliferation o f bacteria w hen food is allow ed to stand are som e o f the w ays in w hich children get infected (B lack et al. 1982; B arrell and R ow land 1979). 3.3.2 Enhanced dietary requirem ents o f children C hildren require higher quantity and quality o f diet relative to their size because o f the heavy dem ands on nutrition for growth and for fighting infections (FA O /W H O /U N U 1985). This is brought out by Figure 3.6, w hich com pares the average requirem ents o f protein and essential am ino acids in infancy, at 10 to 12 years and in adulthood. Infants at 6 m onths need a larger portion o f their o f diet in the form o f protein, about tw o to three tim es as much as 10 to 12 year olds and about five tim es as m uch as adults. T he proportion o f am ino acids required in total protein ingested is 43 percent for infants as against 36 percent for 10 to 12 year olds and nineteen percent for adults p er kilogram o f body w eight. T herefore the 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.6 Average requirements for protein and for the sum of eight essential amino acids at 6 months. 10-12 years, and in adulthood □ Nonessentio E ssen tial C P E C P 0.5 0.0 Infants I0 - I2yrs Adults Source: M unro 1985 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. quantity and nutritional content o f w eaning diets are im portant. P oor socio econom ic conditions prevailing in developing countries affect the ability o f households to buy food and obtain adequate and tim ely m edical care required for the norm al grow th o f children. Poor feeding practices adopted by less literate parents and the poor environm ent add to infectious diseases. T he interaction betw een poor nutritional status and infections further w orsen the nutritional status and grow th o f children. T hese factors constitute the environm ent in w hich children live determ ine the availability o f nutrients at the cellular level and affect the grow th o f children. 3.3.3 P oor quantity o f breast m ilk and w eaning diets It has been found that the breast m ilk produced by poorly nourished w om en is generally low er in quantity and quality. W hile w ell-nourished w om en in developed countries produce 600-800 m l o f breastm ilk daily, poorly nourished w om en produce only 400-700 ml (Jelliffe and Jelliffe 1978; Lauber and R einhardt 1979). B reast milk produced by poorly nourished m others provides few er vitam ins and less energy; m aternal diet was found to be a determ inant o f this (Jelliffe and Jelliffe 1978). H ow ever, the difference in quality is m inor, w hereas, the quantity produced is considerably less w hen a m other is undernourished (Latham 1997). M any studies have concluded that breastfed children are healthier than non-breastfed children both in developed and developing countries (French 1967; K anaaneh 1972; C handra 1979; 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cunningham 1979; Ferguson et al. 1981; G arrett 1981). In developing countries w here environm ental conditions are poor, the protection given by the anti-infective effect o f breast m ilk is m uch higher (B utz et al. 1984). Form ula feeds are often given to children in developing countries. This has been found to be harm ful in m any ways (M artorell and H abicht 1986). B ottle-fed infants m ay not breastfeed as w ell as non bottle-fed infants because they are not as hungry and thirsty to suckle vigorously. T his w ould reduce the anti-infective effect o f breast m ilk. If the bottle-feed is less nutritious than breast m ilk it w ould the nutritional intake. The high cost o f form ula feeds and the inability o f m others in developing countries to read and follow mixing instructions m ay lead to their giving children thin, diluted feeds. The failure o f m any m others to sterilize nipples and bottles for w ant o f know ledge or resources is another source o f infections. The quality o f locally-m ade w eaning diets provided to children in m any developing countries are insufficient to provide enough nutrition for norm al grow th (Payne 1975). In G am bia, the energy content o f m illet gruel w as only h alf as that o f breast m ilk (R ow land et al. 1978). Such diets w ould require children to be fed m any m ore tim es for them to get enough energy for norm al grow th than if they w ere fed energy-rich diets. 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3.4 P oor conditions for catch-up growth The poor environm ents in w hich children live m ay probably explain the significant relationships betw een illnesses and physical grow th in developing countries as w ell as the absence o f such relationships in developed countries (M artorell and Yarbrough 1983). M oreover, the frequent recurrence o f infections and the non-availability o f enhanced diet during convalescence prevent catch-up grow th from taking place in poor countries (M artorell and H abicht 1986). R epeated infections w ithout enough tim e in betw een for catch-up has been found to cause stunting (W hitehead 1977; FA O /W H O /U N U 1985). G row th is affected during infections by a variety o f factors such as cultural practices that reduce intake during infections, reduced appetite, and diarrhea and vom iting that cause reduced absorption (M ata 1978; B isel 1977; M artorell et al. 1980; Hoyle et al. 1980; M olla et al. 1983). C hildren w ho suffered from greater incidence o f diarrhea in Guatem ala w ere found to be shorter by 3.5 cm at age seven than those relatively free o f diarrhea (M artorell et al. 1975). Such deficits in height w ere reported by studies done in B ritain in the 1940s and 1950s in children betw een one and five years old, w ho suffered from severe infections (Tanner 1962). R ea (1971) analyzed data on children in Lagos w ho w ere two to five years old and had suffered severe grow th reverses in infancy. It w as found that children o f higher socio-econom ic groups w ere able to catch-up to above the 60th percentile o f the British standard o f height by age five, whereas 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. children o f the low est socio-econom ic group rem ained below the 10th percentile. The availability o f good diet is likely to have enabled children o f richer fam ilies to catch up. An experim ent done in N arangw al, Punjab, India, w as designed to test the effects o f nutritional inputs, health care and a com bination o f the tw o on grow th in height (K ielm ann, Taylor and D eSw eem er 1983). The results are show n in Figure 3.7. T he thick broken lines in both parts o f the figure show s that health care was m ore significant than nutritional intake in the first year o f life; the solid lines show that nutritional inputs had the greatest effect on height betw een the ages o f eighteen m onths to three years; and the thin broken lines show that the com bination o f nutritional intake and health care had an effect that w as in betw een that o f health care and nutritional intake. The results o f this study appeared to indicate that health care m ay be m ore im portant during the first year o f life and nutritional intake m ore im portant thereafter. This conclusion is in line w ith the earlier evidence about the high rates o f infection causing deterioration in nutritional status during the w eaning period. 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.7 (a) Effects of different types of input services on the mean heights of children from the Narangwal project in the first three years of life (adjusted for sex, birth order, mother’s age, caste, year and season of observation! (bl Mean heights of the same children relative to the Harvard height standard (adjusted for caste and sex! (a) 3.00-1 — 2.00 & 1.00 Input services Nutrition care (NUT). Nutrition and health care (NUTHC)V Health care (HC) s Zero baseline Control (CONT-N) \ \ \ \ v V Age (m onths) = 0.00 15 17 19 21 - 1.0 0 - (b) Input services Nutrition care (NUT). Nutrition and health care (NUTHC). Health care (HC). Control (CONT-N). 100 98- 96- 92- 90- Age (months) Source: Kielm ann et al. 1983. 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Concluding remarks Im m ediate, underlying and basic factors presented in Figure 1.1 affect the nutritional status and grow th o f children. T he physical environm ents, the socio econom ic status, diet, literacy o f parents and infections, largely determ ine their grow th. The first tw o to three years o f grow th are very critical for children in developing countries and largely determ ine their adult stature, w ork capacity and productivity. This is because m ost o f the children continue to live in the sam e poor environm ents that preclude catch-up grow th. G enetic factors have a m uch sm aller role because m ost o f them do not realize their genetic potential. These findings have im plications for policy. Interventions during pregnancy and the first tw o to three years o f life are likely to be effective in lowering the high rates o f undem utrition in India. 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4 THE BASIC AND UNDERLYING CAUSES AFFECTING CHILD GROWTH IN KERALA AND UTTAR PRADESH In this chapter, a com parison is m ade o f the tw o states o f India, nam ely U ttar Pradesh and K erala, to bring out the effects o f governm ent policy and the organized action by the people, on the quality o f life o f the population. The grow th o f children, especially in height, is considered to be a reflection o f the quality o f life o f a population (W aterlow 1988). This is because grow th in height is the product o f the cum ulative effect o f the environm ent in w hich children live on their health. A s studies discussed in the first three chapters have show n, m uch o f the grow th retardation occurs in the first three years o f life. F urther, the deficit in height developed in childhood is rarely regained because children continue to live in the sam e poor environm ents that initially caused the deficit. It w as seen in Section 3.3.2 that children o f low er ages require m uch higher quantity and quality o f diets relative to their w eight for norm al grow th. M any poor households in India cannot ensure this for their children. U ndem utrition affects the physical and m ental grow th and developm ent o f children and this affects th eir productivity and earning capacity as adults (Section 1.2). This in turn is likely to affect the 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. earnings and nutritional status o f future generations. I f these consequences are to be avoided, the quality o f life o f the entire population needs to be raised. The term quality o f life is used to m ean the capabilities to be healthy and well nourished, free from diseases caused by m alnutrition, to be able to grow norm ally in w eight and height. The discussion in this chapter largely concerns the basic and underlying causes o f child grow th retardation. C om pared to the rest o f India, the quality o f life in K erala is better (Table 1.5), the poverty rate has declined (Table 1.6) and the rate o f child grow th retardation is overall low relative to other states o f India (Table 1.7). In U ttar Pradesh, on the other hand, the quality o f life is p o o r relative to other Indian states and the poorest countries in the w orld, and the level o f child undem utrition is one o f the highest in India and the w orld (see Tables 1.1, 1.2, 1.3, 1.5 and 1.7). T hese tw o states do not stand alone am ong Indian states. U ttar Pradesh, Bihar, R ajasthan and M adhya Pradesh are very sim ilar in m atters related to literacy, health and the position o f w om en (Table 1.5). Fem ale literacy rate in the 7+ age group w as tw enty-five percent in U ttar Pradesh, tw enty percent in R ajasthan, tw enty-three p ercent in B ihar and tw enty-nine percent in M adhya Pradesh, life expectancy w as below the national average and w om en on average had betw een 4.4 and 5.1 children. These four states together had a population o f 335 m illion, w hich accounted for forty percent o f India's population in 1991. A m ong the states that have done relatively w ell, in 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. addition to Kerala, are Tamil N adu and H im achal Pradesh (Table 1.5). The female literacy rate in the 7+ age group w as eighty-six percent in K erala, fifty-tw o percent in H im achal Pradesh and fifty-one percent in Tam il N adu. The total fertility rate (T F R ) w as 1.8 in Kerala, 2.2 in Tam il N adu and 3.1 in H im achal Pradesh. M ale life expectancy at birth w as sixty-nine in K erala and sixty-five in Punjab. This study treats U ttar Pradesh and K erala as states representing the better and w orse perform ers in term s o f quality o f life and child grow th in India. T he purpose is to see w hat w orked in K erala and w hat kept m ajor sections o f the society deprived in U ttar Pradesh. T w o m ajor factors that affect the well being o f a population are people’s organized action and public policy. P eople’s organized action through com m unity and political organizations has been found to be effective in reducing traditional inequalities, im proving the position o f w om en, and ensuring that favorable public policies are adopted and public services function efficiently. Public policy that effectively im proves the lives o f w eaker sections o f society is, m ore often than not, the result o f sustained and organized action by these sections for their own betterm ent. O rganized action o f the people and public policy affect the underlying and basic causes that affect the nutritional status and grow th o f children (Figure 1.1). A m ong the m ajor factors that have affected quality o f life in India are the availability and quality o f schooling and health services, regular 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. availability o f food and other essential com m odities at affordable prices, and the reduction o f rural inequality through land reform s. In this chapter I com pare the situation in these four respects in K erala and U ttar Pradesh. I have based m y study on the research and field investigations done by other researchers, m y own fam iliarity w ith conditions in India and the findings o f the N ational Fam ily Health Survey o f India (1992-93). 4.1 Schooling: Progress in Kerala, and the poor situation in Uttar Pradesh Robin Jeffrey, w riting in the context o f K erala, called literacy the basic personal skill that underlies the entire m odernization sequence (quoted by R am achandran 1996). Literacy w as the basis o f econom ic grow th and social progress the w orld over. K erala had achieved near-universal literacy for both m ales and fem ales by 1991 (Table 1.5). M any inform ed observers consider the high literacy rate o f w om en in K erala to be the foundation o f its high quality-of- life indices (Jeffrey 1987; Caldwell 1986; Zacharia 1984). A notable feature of K erala’s achievem ent in literacy is the absence o f m ajor urban-rural, gender, caste or regional disparities. In contrast, in U ttar Pradesh, not only w ere m ale and fem ale literacy rates m uch lower than the national average, but the fem ale literacy rates w ere less than h a lf as that o f the m ales (Table 1.5). T hree out o f four w om en there w ere illiterate. There w ere also m ajor urban-rural, gender, caste or 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. regional disparities. There are very few studies available on the historical developm ent o f literacy in U ttar Pradesh, perhaps because o f the absence o f any m ajor initiative there to im prove m ass literacy. The history o f the progress o f literacy in K erala is presented below in order to highlight the role o f public policy and com m unity action in prom oting literacy. O ther aspects related to this basic requirem ent o f social and econom ic progress are also discussed. K erala w as form ed on a linguistic basis in 1956. Prior to that K erala consisted o f three regions, nam ely, M alabar, Travancore and Cochin. M alabar w as ruled directly by the British. T ravancore and Cochin w ere princely states. M alabar had thirty-four percent, T ravancore had fifty-three percent and C ochin had thirteen percent o f the population o f K erala in 1940. U ttar Pradesh w as under B ritish rule before India becam e independent in 1947. First, the literacy rates in the two states in the post-1950 period are discussed. T his includes an account o f the reduction o f disparities in literacy based on gender, region and social class in K erala and the low rates and persisting disparities in U ttar Pradesh. Then the historical progress o f literacy in K erala is discussed. 4.1.1 R egional disparities w ithin states T here w ere considerable regional disparities in literacy in the 1950s betw een M alabar and the rest o f Kerala. In 1950-51, the literacy rate in M alabar 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w as thirty-one percent w hile it w as about forty-five percent in the rest o f K erala (N air 1981; Jeffery 1987). In the 1950s and 1960s literacy increased at a rapid pace in M alabar and the region caught up w ith the rest o f the state by 1971 (N air 1981). Regional disparities in U ttar Pradesh, on the other hand, have rem ained high. In 1991, in the 7+ age group, both the m ale (seventy-six percent) and the fem ale literacy rates (forty-three percent) w ere m uch higher in the H im alayan region than the m ale (fifty-five percent) and fem ale literacy rates (tw enty-one percent) in the eastern region o f U ttar Pradesh. T he west, central and southern regions lay in betw een (D reze and G azdar 1996). There w ere also substantial disparities betw een urban and rural areas o f U ttar Pradesh in 1991. The m ale literacy rate w as seventy percent in urban and fifty-tw o percent in rural areas. A m ong fem ales, the urban-rural gap w as even w ider. The literacy rates for w om en w ere fifty percent in urban and nineteen percent in rural areas. In 1981, the crude fem ale literacy rate for scheduled castes1 w as below 1.5 percent in eighteen o f the fifty-six districts o f U ttar Pradesh, and below 2.5 percent in m ajority o f the districts (N una 1990). 1 Scheduled castes were untouchables, who were below the four-tier caste system. They still are one o f the most backward sections o f Indian society and large sections o f them are illiterate still live in poverty. The scheduled tribes were the original inhabitants o f India. For thousands o f years they lived undisturbed in the hills and densely wooded regions, which others considered unattractive. They are still educationally and econom ically one o f the most backward groups in India. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1.2 D isparities in gender and social class and the quality o f education In the 1950s, there w ere considerable gender and class-based disparities in literacy in K erala, sim ilar to the disparities found in m any o f the states o f India in 1991 (Table 1.5). In 1951, fifty percent o f the m ales and thirty-one percent o f the fem ales w ere literate. B ut these disparities w ere rapidly reduced and nearly elim inated by 1991 (Table 4.1). In 1987-88, the rural literacy rate for both boys and girls ten to fourteen years old w as equal at 98 percent (Table 1.5). Table 4.1 Percentages of males and females literate in Kerala during 1951-1991 Year M ale Female T otal 1951 50 31 41 1961 55 39 47 1971 67 54 60 1981 75 65 69 1991 94 86 90 Source: A dapted from Jeffrey 1987; Prakash 1994; N air 1981. In contrast, in U ttar Pradesh, w ide gender disparities have persisted (D reze and G azdar 1996). In 1991 the fem ale literacy rate was less than h a lf the 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m ale literacy rate. The gender gap in literacy am ong the Indian states w as the highest in U ttar Pradesh and the low est in K erala (Table 1.5). Table 4.2 illustrates the reduction o f disparities in literacy based on social class in Kerala. Scheduled castes and scheduled tribes are considered, socially and econom ically, the m ost backw ard classes o f Indian society. The literacy rates o f scheduled castes and scheduled tribes w ere m uch behind the rest o f the population in literacy prior to the 1960s, but thereafter, the literacy rates am ong them increased at m uch faster rates than the literacy rate o f the total population. The female literacy rates am ong scheduled caste and scheduled tribe w om en increased by m ore than four times, w hile the literacy rate o f all w om en doubled betw een 1961 and 1991. Further, the gender gap in literacy betw een m ales and fem ales in these tw o groups narrow ed substantially as is evident from Table 4.2. The scheduled caste population in K erala had literacy rates sim ilar to the rest o f the state population in 1991. In 1991, scheduled caste fem ales in K erala w ere m ore literate (74 percent) than the overall m ale population o f India (64 percent). This is im pressive, considering that only tw enty-three percent o f scheduled caste w om en in India w ere literate in 1991 (D reze and G azdar 1996, R am achandran 1996). Fifty-one percent o f scheduled tribe w om en in K erala w ere literate as against 18 percent in India (Tyagi 1993). Scheduled tribes still have considerable catching up to do with 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the rest o f the population in Kerala. A nother feature aggravating the failure in U ttar Pradesh w as the com bination o f both gender and class differences. In 1987- 88, only eight percent o f the scheduled caste w om en in rural U ttar Pradesh w ere Table 4.2 Proportion of literate persons at age +7 among scheduled castes and scheduled tribes in Kerala in 1961 and 1991 Year Scheduled Castes Scheduled Tribes Total Population M ale Female M ale Female M ale Female 1961 32 17 23 12 55 39 1991 85 74 63 51 94 86 Source: A dapted from Ram achandran 1996 literate as against thirty-nine percent scheduled caste m en, and nineteen percent o f the state rural fem ale population (D reze and G azdar 1996). Sixty-eight percent o f the girls 12 to 14 years old in rural U ttar Pradesh w ere never enrolled in a school in 1986-87 as against tw enty-seven percent o f the boys in the sam e age group. T able 4.3 com pares the enrollm ent ratios o f students belonging to the scheduled castes (SC) and the scheduled tribes (ST) in K erala, U ttar Pradesh and India. O verall, the enrollm ent and the dropout rates w ere higher in U ttar Pradesh. 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.3 Enrollment of SC and ST students in Kerala and Uttar Pradesh and India in 1990-91 State Scheduled Castes Scheduled Tribes Grade 1-5 G rade 6-8 Grade 1-5 Grade 6-8 B ovs G irls B ovs Girls B ovs Girls B ovs Girls Kerala 120 117 117 112 103 128 88 87 U ttar Pradesh 92 39 32 9 114 74 51 18 India 123 81 61 33 127 79 51 28 Source: A dapted from A .N . A garw al et al. 1993 Secondly, the enrollm ent ratios o f girls belonging to scheduled castes and scheduled tribes at both the grade levels w ere far low er than boys. Except at Grade 1-5 in the case o f scheduled tribe boys, enrollm ent ratio o f girls was only about one-third as that o f boys. O nly nine percent o f scheduled caste girls and 18 percent o f scheduled tribe girls w ere enrolled in grades 6 to 8 in U ttar Pradesh. The drop out rates for children belonging to these backw ard castes in U ttar Pradesh, w ere very high. R oughly tw o-thirds o f the boys and three-fourths o f the girls belonging to scheduled castes in U ttar Pradesh dropped out before reaching Grade six. T he dropout rates am ong scheduled tribes betw een the tw o levels o f schooling w ere about fifty percent for boys and seventy-five percent for girls. In 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. contrast, the enrollm ent rates in K erala w ere m ore than a hundred percent at both the grade levels for boys and girls belonging to scheduled castes. A m ong scheduled tribes enrollm ent w as m ore than a hundred percent at the G rades 1-5 for boys and girls. Further, the enrollm ent o f scheduled tribe girls w as higher than boys at both the grade levels, show ing the absence o f gender bias against fem ales even in this m ost backw ard social group in Kerala. N early ninety percent o f the scheduled tribe children continued in grades 6 to 8 in K erala, as against only fifty-one and eighteen percent o f boys and girls respectively in U ttar Pradesh. D ata in T ables 1.5 and 4.3 dem onstrate that disparities in literacy levels and enrollm ent based on gender and social class w ere high in U ttar Pradesh in 1991. T ables 4 .1 ,4 .2 and 4.3 dem onstrate that the disparities in literacy based on gender and social class w ere high in K erala in 1951, but practically disappeared by 1991. T able 4.4 show s that the patterns seen in enrollm ent and dropout for scheduled castes and scheduled tribes in Table 4.3 also apply to the general population. T here w ere high drop out rates in U ttar Pradesh for both boys and girls, and the drop out rates for girls w ere far higher than for boys. 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4 Enrollment of students in grades 1-5 and 6-8 in Kerala, Uttar Pradesh and India in 1990-91 State G rades 1-5 G rades 6-8 B o v s Girls B o v s Girls Kerala 104 101 107 104 U ttar Pradesh 99 62 65 28 India 115 86 73 46 Source: A dapted from A.N. A garw al et al. 1993 T he quality o f education is as im portant as, if not m ore im portant than, the num ber o f years o f schooling (B ehrm an 1995). H igh drop-out rates are partly the result o f the p o o r quality o f education. A nother m easure o f the quality o f education is the functional literacy rate which is defined in the table below as the num ber o f students successfully com pleting three years o f school education out o f an initial cohort o f 100 enrolled in grade one (U nited N ations 1975). The functional literacy rate o f girls in K erala w as m ore than three tim es higher than girls in U ttar Pradesh and double that o f India in 1967-69 (Table-4.5). The average educational attainm ent o f children leaving school w ould have been low er in U ttar Pradesh than in K erala and India. A recent field investigation show ed that 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the quality o f public education in U ttar Pradesh w as low in the 1990s (D reze and G azdar 1996). The findings o f this investigation are discussed shortly. Table 4.5 Rates of functional literacy in Kerala, Uttar State B ovs Kerala 81 U ttar Pradesh 37 India 47 Source: U nited N ations 1975. It is also notew orthy that the functional literacy rate for girls in K erala was higher than boys, w hereas for India and U ttar Pradesh, girls had low er functional literacy rates than boys. O ne reason for the lower educational attainm ents o f girls in U ttar Pradesh could be the low er preference parents give to fem ale education (D reze and G azdar 1996). This aspect is discussed later. The higher functional literacy o f girls in K erala w as considered to be the result o f the absence o f gender-based preferences am ong parents in K erala (Jeffrey 1987). W e have seen that the attainm ents in literacy in K erala are im pressive in m any respects and problem atic in m any respects in U ttar Pradesh. It is im portant 93 Pradesh and India, 1967-69 Girls 89 28 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to exam ine the reasons behind the poor progress in U ttar P radesh and the m ore im pressive progress in K erala. In order to do this, first, I describe the findings o f a field investigation o f the public school system in U ttar Pradesh. This investigation revealed that the social, cultural and political setting in U ttar Pradesh w as hindering progress not only in literacy, but also in health and other aspects affecting the decline in poverty and the im provem ent in the quality o f life o f the m ajority o f the population in the state. It will becom e evident that im provem ent in the various indices o f quality the o f life and child grow th are closely connected and affected by the basic and underlying causes described in Figure 1.1. These aspects cannot be seen in isolation. The grow th o f children is a good indicator o f im provem ent in the quality o f life o f a population. 4.1.3 A field investigation in U ttar Pradesh In order to find out the reasons for the poor perform ance o f schooling and the considerable disparities that existed based on gender, social class and region in U ttar Pradesh, a field investigation w as done in February-M arch 1994 by D reze and G azdar (1996). The authors visited prim ary schools and a few m iddle and high schools in 16 villages in four districts o f U ttar Pradesh, covering the different regions o f the state. The authors spoke to teachers, parents, governm ent officials, local leaders and com m unity activists. They found the school buildings in 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. dilapidated and m ostly unfit conditions for instruction. The classes w ere m ostly held in the verandas o f school buildings or under trees, w here children o f all grades sat together. M ost o f the furniture w as m issing, taken aw ay for use in the residences o f village leaders, teachers or headm asters. U nauthorized absences o f teachers from schools w ere com m on, som etim es for several days in a w eek. W hen they did attend, they rarely stayed after the lunch break at 12:30. W hile at school, teachers w ere rarely found to be teaching, but busy talking w ith each other or visitors, preparing rolls for elections, playing cards and the like. A bsenteeism and shirking o f duties w ere found m ore com m only in sm aller schools, w here there w ere only one or tw o teachers and supervision w as lacking. H ow ever, 41 percent o f the prim ary schools in U ttar Pradesh had few er than three teachers, m aking shirking a m ajor problem underm ining the school system . A s a result, on m any days schools w ere often closed w ithout prior notice. T he absence o f fem ale teachers was another feature o f rural schools in U ttar Pradesh. T here w as not one fem ale teacher em ployed in any o f the 16 schools visited. The declared policy in the state sought to increase the percentage o f w om en teachers. H ow ever, contrary to policy, the percentage o f w om en teachers had declined from tw enty-one percent in 1986 to eighteen percent in 1992-3. U ttar Pradesh had the low est percentage o f w om en teachers am ong all the Indian states. In com parison, sixty-six percent o f the teachers in K erala and 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. thirty-nine percent o f the teachers in the four southern states o f India w ere w om en (N C ER T, 1992 and G overnm ent o f India 1994a). In the cultural and social setting o f rural U ttar Pradesh, m any parents w ere reluctant to have even girls seven to eight years old taught by m ale teachers (G upta et al. 1993). M ost o f the prim ary schools in rural areas, particularly in rem ote areas, had exclusively m ale teachers. H ence, the absence o f female teachers is likely to dam pen the progress o f fem ale literacy in the rural areas o f U ttar Pradesh (D reze and G azdar 1996). D uring the field visit, only about one-third o f the children attending schools w ere found to be girls. B ecause o f the poor quality o f public schools, m any parents preferred to send th eir children to private schools. H ow ever, only about three percent o f the private prim ary schools w ere given recognition by the governm ent (NCERT, 1992). Students attended unrecognized private schools and transferred to governm ent schools in grade five to obtain official prim ary school passing certificates. Private schools, both recognized and unrecognized w ere well m anaged and w ell attended. T his show ed that lack o f dem and for education w as not the reason for the slow progress o f literacy in U ttar Pradesh. T he teachers in private schools w ere, on average, paid m uch less than teachers in governm ent schools, but they attended schools regularly and stayed on for the full tim e, teaching well. A nother aspect that had im plications for gender equality in U ttar Pradesh w as 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. that m any m ore boys than girls w ere attending private schools. This was attributed to tw o factors: the reluctance o f parents to send girls out o f the village to attend private schools for cultural reasons and their reluctance to pay fees for the girls in private schools because o f the low priority they attached to female education. A nother notew orthy aspect w as the failure o f the state governm ent to utilize the readiness o f the private sector to im prove literacy. By recognizing private schools m any m ore private schools w ould have been started, and the quality and spread o f education w ould have risen. W hile the w hole population suffered, the only beneficiaries seem ed to be governm ent school teachers and politicians. The form er had a captive pool o f students to protect their jo b s and postings. T he latter w ere able to dispense patronage and advance their personal and partisan goals by appointing teachers o f their choice and in turn getting their help during elections, census enum erations, etc. This nexus betw een politicians and teachers provided protection to teachers and rendered the school inspection system ineffective in U ttar Pradesh. O ne o f the m ajor hurdles in the w ay o f rapid progress in literacy appears to be governm ent policy in U ttar Pradesh. In m any respects, the prim ary school system appeared to be getting w orse in U ttar Pradesh. For instance, the num ber o f prim ary school teachers per 1000 pupils declined from 26 in 1981-2 to 17 in 1992-3 and public expenditure per capita on education in U ttar Pradesh reduced Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. by alm ost 20 percent betw een 1991-2 and 1993-4 (Tyagi 1993). Large grants given for prim ary education by the G overnm ent o f India and international agencies w ere lost because o f non-utilization by the state governm ent (D reze and G azdar 1996). There was also considerable scope for im proving the availability o f schools in rural U ttar Pradesh. O nly 66 percent o f the villages in U ttar Pradesh had a prim ary school and only 23 percent o f the villages had a m iddle school; every village in K erala had a prim ary and m iddle school (UPS 1995). W ith parents reluctant to send girls out o f the village for schooling, this is likely to have depressed fem ale literacy. Overall, the governm ent in U ttar Pradesh seem ed to be giving low priority to education. A nother im portant aspect that w as noted in U ttar Pradesh was the lack o f public action to force the governm ent to im prove schooling. P oor governm ent policy and non-im plem entation o f projects had hardly drawn protests from the press, opposition parties, interest groups or the general public. The field investigation encountered m any parents w ho w ere highly critical o f the school system , but they w ere not even considering m aking organized dem ands for im proving it. The lack o f literacy itself m ay be one o f the m ain reasons w hy the disadvantaged groups have not been able to m ake effective dem ands for better education. 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Poverty did not appear to be the cause o f low literacy rates in U ttar Pradesh. M any studies had concluded that poverty had very little explanatory pow er in explaining illiteracy after controlling for adult literacy, schooling facilities, caste and gender (M aharatna 1995; Sinha 1995; D reze and G azdar 1996). Social divisions, how ever, appear to have played a role in depressing literacy rates am ong low er social classes, w ho w ere far less educated than others in U ttar Pradesh. Instances w ere found w here influential higher caste m en prevented schools from being built in scheduled caste inhabited villages; and low er caste students w ere ignored, ill-treated and discrim inated against in schools by teachers and higher-caste students (W adley and D err 1989; B anerjee 1994; D reze and G azdar 1996). Social divisions cause conflicts o f interest and prevent or w eaken organized action. In m any parts o f India, landlords w ere not the m artial castes. In U ttar Pradesh landlords belonged to the m artial castes (know n as R ajputs, Thakurs or K shatriyas). This m ade it m uch harder to challenge the prevailing rural inequalities (Dreze and G azdar 1996). Social attitudes am ong different social groups also appeared to have depressed schooling in U ttar Pradesh. In a survey o f P alam pur village in U ttar Pradesh it w as found that both m ales and fem ales belonging to certain higher castes like Kayasths (a higher caste long engaged in clerical occupations) w ere highly literate (D reze and G azdar 1996). H ow ever 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. certain higher castes like Thakurs and low er castes like scheduled castes had very low rates o f literacy. The K ayasths considered it socially acceptable to send their girls out o f the village to study. N one o f the other castes sent th eir girls to study outside the village, as they considered it socially unacceptable. T here has been no credible initiative to alter these attitudes either from com m unity organizations or the government. M any literacy program s that have been successful in other states have failed to m ake an im pact in U ttar Pradesh. The Total Literacy C am paign (TLC) is one such case. T he aim o f TLC w as to identify every person w ho w as not literate and to teach them to read w ith the help o f volunteers. It w as a great success in K erala, w ith m ore than 350,000 volunteers participating in the cam paign (R am achandran 1996). B ut in U ttar Pradesh the state governm ent took little interest in this. T his also happened w ith the TLC in other states w here literacy rates w ere low (G hosh et al. 1994). L iteracy is a basic requirem ent for dem ocratic participation and for disadvantaged groups to organize for their rights (D reze and Sen 1995). The political leadership in the state has been traditionally from privileged classes. E xperience in the last five decades has show n that they are not likely to take initiatives that w ill better the lot o f w eaker sections. This is because the existing system goes in their favor. U nless political and social m ovem ents organize w eaker 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sections to dem and a better deal for them selves, the present situation is likely to persist in U ttar Pradesh. T he experience o f K erala in raising literacy, discussed below , ju stifies this conclusion. The problem s in the prim ary schooling system prevailing in U ttar Pradesh w ere also characteristic o f other areas like health services, public distribution system and the im plem entation o f land reform s. T hese aspects in the tw o states w ill be com pared in subsequent sections. 4.1.4 The historical progress o f schooling in K erala The historical progress o f schooling in K erala reveals the role o f com m unity organizations in dem anding social reform and in im proving literacy at public expense, as w ell as the constructive role o f public policy in literacy and social transform ation. In 1875 only about ten percent o f the m ales and less than h a lf a percent o f the fem ales w ere literate in the three areas that w ere m erged to create K erala in 1956. In the last quarter o f the 19th century, the princely governm ents o f Travancore and Cochin took m any initiatives to prom ote literacy. H ow ever these policies did not com e about without p eo p le’s agitating for change. 4.1.4.1 Social m ovem ents for literacy and governm ent policy Literacy rates in K erala w ere low in the 19th century. T here w ere m any disparities and inequalities in social, political and econom ic status am ong the 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. people o f Kerala. H igher castes discrim inated against low er castes and m ade their lives difficult. In the early 19lh century, C hristian m issionaries encouraged depressed classes to organize and fight for their social and econom ic rights (Tharakan 1984). T hey organized resistance to discrim inatory practices against low er castes. O ne m ethod o f discrim ination w as to deny the low er classes access to m odem education. T he oppressed com m unities saw education, especially o f w om en, as an im portant m eans for their econom ic and social advancem ent. T he leaders o f tw o m ajor H indu com m unities, the N airs and Izhavas, dem anded that the governm ent expand the public educational system (R am achandran 1996). O ne reason for this dem and w as the decision o f the Travancore governm ent in 1844 to give preference to those w ho had an English education in recruitm ent to senior positions in governm ent. Further, literacy and arithm etic w ere m ade com pulsory for all governm ent jo b s (Caldw ell 1986). This m ade literacy desirable for getting the m uch-coveted governm ent jo b s. A s early as in 1817, the governm ent o f Travancore had declared that the rem oval o f illiteracy, paid for by the state, w as the responsibility o f the governm ent. T his had set the tone for governm ent support o f education, even though it w as only in 1865 that the first m ajor initiative to prom ote vernacular education cam e (Jeffrey 1987). T he expansion o f education was one o f the m any econom ic, social and other changes dem anded by the low er classes. The agitations becam e w idespread 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. by the 1850s. B ut higher castes, that w ere the beneficiaries o f the old system opposed reform s. The T ravancore governm ent w as caught betw een the agitators and the higher castes and felt that the only w ay to set a clim ate suitable for social and econom ic reform s w as by the spread o f literacy (Jeffrey 1987, 1976b). This resulted in policies that expanded literacy in Travancore. T he leadership for the m ajor m ovem ents dem anding changes in governm ent policy in areas such as literacy, econom ic and social reform s cam e from w ithin the different com m unities them selves. In the early 20th century Izhava leaders (Izhavas w ere a backw ard com m unity, socially at the bottom o f the caste structure, but above the scheduled castes and scheduled tribes) organized the com m unity for all-round reform in T ravancore and other regions o f present-day K erala. The Izhavas w ere discrim inated against in m any ways. T hey w ere seventeen percent o f the population in T ravancore but had less than four p ercent o f governm ent jobs. By the 1920s and the 1930s, this initiative developed into a large-scale m ovem ent. N airs, w ho w ere also seventeen percent o f the population, had fifty-four percent o f governm ent jobs. Izhava leaders dem anded higher representation in governm ent jo b s and the rem oval o f social restrictions im posed on them such as the prohibition on the use o f certain public roads. The leaders o f the m ovem ent w orked for a rapid increase in enrollm ents, and organized scholarships and fee concessions for Izhava children. In dem anding reform , they found a com m on 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cause w ith C hristians and M uslim s dem anding sim ilar reform s in Travancore. A t about the sam e tim e, the untouchables o f K erala under their ow n leaders successfully organized m ovem ents for econom ic and social ju stice in m atters such as the rem oval o f prohibitions on use o f public roads used by up p er castes, adm issions to governm ent schools and so on. A m ong N airs who w ere a relatively forward caste, too, there were organized m ovem ents for social reform . N airs follow ed a m atrilineal system o f inheritance. N air w om en took the initiative for reform ing outdated social practices and increasing literacy in the com m unity in the late 19th and early 2 0 th centuries (R am achandran 1996). This enabled Nairs to enter professions and the bureaucracy. A m ong B rahm ins w ho w ere at the top o f the caste structure also, there w ere m ovem ents to reform and m odernize the com m unity and to get rid o f the outdated social practices like the seclusion o f w om en. T he C hristians who w ere in the forefront o f reform m ovem ents also encouraged fem ale literacy. These m ovem ents initiated in Travancore spread to Cochin and M alabar because o f the com m on language, com m on social custom s and practices, and day-to-day contacts am ong the people living in all the three areas. Thus all m ajor sections o f society organized them selves for reform and m odernization. The goal o f m ass literacy form ed the core o f all these reform m ovem ents and helped to spread literacy in every section o f society in Kerala. 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A large num ber o f governm ent schools w ere started in Travancore. Follow ing the announcem ent o f a liberal policy o f grants-in-aid in 1886, many private organizations and individuals came forward and started schools with governm ent aid. For instance the Catholic C hurch in K erala had 10 schools under its m anagem ent in 1881-82; by 1887 it had m ore than a 1000 schools (Jeffrey 1987). D uring the tw enty years from 1873 to 1894 there w as a substantial increase in the num ber o f schools and students enrolled as show n in Table 4.6. The num ber o f governm ent schools nearly doubled, the num ber o f governm ent- aided private schools increased seventy tim es and the num ber o f students w ent up by six times. A long w ith providing grants-in-aid, m any other im portant initiatives w ere taken to prom ote literacy (Jeffrey 1987; R am achandran 1996). Fees for girls in prim ary schools w ere abolished in 1896 and backw ard castes w ere given incentives to attend schools. Ten years later, prim ary education w as m ade free for all castes. C hildren w ere provided m id-day m eals starting from the early 1940s. A sim ilar course o f events happened in C ochin too, w hich had an even more orthodox society than Travancore. The B ritish adm inistration in M alabar attached low priority to education. In 1921 the literacy rate in M alabar was only h a lf that o f Travancore. H ow ever, the India A ct o f 1919 passed by the British Parliam ent caused substantial 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allocations to be m ade for education in M alabar. The people o f M alabar w ho had w itnessed the spread o f literacy in T ravancore and C ochin took advantage o f the Table 4.6 Growth of government-recognized vernacular schools and students in Travancore, 1873-1894. Year G overnm ent G overnm ent- Students in governm ent schools aided private and private-aided schools schools N um ber Index N um ber Index N um ber Index 1873-74 117 100 20 100 9637 100 1882-83 223 191 440 2200 35588* 369 1894 225 192 1388 6940 57314** 594 *in 1881, **in 1893 Source: A dapted from N air 1981. opportunity. B y 1936 the M alabar D istrict B oard m anaged m ore than 1200 schools. O f these, private m anagem ents ran about 70 percent o f the schools with grants-in-aid from the governm ent. From about ten percent in 1919, the literacy level in M alabar increased to thirty-one percent in 1951 (Jeffrey 1987). D espite the m ass expansion o f schooling, there w as no m ass literacy by the end o f the 19th 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. century. O nly about tw elve percent o f the population in the three regions w as literate in 1901. A fter the form ation o f K erala state in 1956 successive governm ents attached high priority to education. T he proportion o f governm ent expenditure on education in K erala has been higher than in other states o f India. T able 4.7 gives the share o f governm ent expenditure on education in K erala and all other states o f India. T he public school system in K erala w as found to function regularly and Table 4.7 The expenditure on education as a proportion of developmental and non- developmental expenditure in Kerala and other states Year Kerala O ther s 1960-61 35 20 1971-72 36 22 1981-82 32 21 1991-92 25 17 1992-93 27 17 Source: Ram achandran 1996 efficiently during a field investigation by the sam e authors (D reze and G azdar 1996). D uring the thirty years beginning in 1961, the percentage o f literate 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. persons in K erala’s population increased from forty-seven to ninety-one (N air 1981; Agarwal and Varm a 1996). A part from the m ajor initiatives that cam e from w ithin com m unities for all-round reform and governm ent policy, there w ere tw o other significant factors that helped in the rapid expansion o f literacy in K erala. O ne w as the role o f the C om m unist Party and the other w as the culture o f K erala that favored fem ale literacy. 4.1.4.2 T he lead taken by the C om m unist Party B eginning in the 1930s, the leftist m ovem ent led by the C om m unist Party in K erala played a very im portant role in organizing and radically changing the social, econom ic and political outlook o f the people and in giving a pro-poor direction and purpose to the policies o f successive governm ents in K erala. C om m unist Party activists w ere at the forefront o f m ovem ents and agitations for tenancy reform , raising the m inim um wages o f workers, rem oval o f social discrim inations like untouchability, political freedom from the B ritish, dem ocracy, integration o f the state on a linguistic basis and so on (R am achandran 1996). School teachers m ade up a disproportionate num ber o f C om m unist Party activists in rural areas in the pre-1950 period (Caldw ell 1986). C om m unist Party volunteers organized village schools, adult education and study classes, and 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. libraries in villages all over the three regions o f the state. The C om m unist Party m ade literacy a requirem ent for its m em bership and laid stress on m aking the rural population literate. T hey organized w om en in several occupations and gave them leadership positions. W hen the com m unists cam e to pow er in 1957 in the first elections held to the state legislative assem bly, they doubled the expenditure on education. They also initiated and provided leadership in other areas o f reform like land reform s and the public distribution system (T he PDS distributes food and other essential articles o f daily consum ption such as sugar, edible oil, fuel for lighting and cooking, cloth etc. at affordable prices through governm ent-licensed outlets). 4.1.4.3 A culture favoring fem ale literacy A m ajor factor in the spread o f literacy in K erala w as the absence o f opposition to fem ale literacy in the society. The only exceptions to this w ere the Brahm ins, who w ere the highest caste, and the M uslim s, both o f w hom practiced the seclusion o f w om en in their hom es as a m ark o f social status or social advancem ent. M alabar had a substantial M uslim population (30 percent in 1940). T hey attached low priority to w om en’s education. T he culture in every other region and am ong other social groups favored female literacy (Jeffrey 1987). 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There w as no know n organized social opposition to fem ale literacy in K erala (M ateer 1871 cited in R am achandran 1996). T here w ere also substantial sections o f society that practiced m atrilineal system o f inheritance. T hey included N airs, som e sections o f Izhavas, and a significant section o f M uslim s in M alabar, who w ere converts from H induism to Islam . B eing holders and inheritors o f property w om en in these m ajor com m unities played a crucial role in decision-m aking w ithin their households. The birth o f a fem ale in a N air fam ily w as an occasion for rejoicing rather than a disaster as in m any other parts o f India (R am achandran 1996). In the royal fam ilies o f Travancore and Cochin too w om en had a prom inent role. W hen a M aharaja (king) died, his brother succeeded him, if he had no brother, his sister’s son succeeded him. I f the sister’s son w as elder than the brother o f the M aharaja, the form er becam e the next M aharaja (Ram achandran 1996). T hese social and political arrangem ents resulted in w om en’s enjoying positions o f relative prestige and equality to m en in K erala com pared to the rest o f India. T his cultural background also enabled w om en to acquire literacy w ithout social hindrances. The advance o f fem ale literacy is thought to have played a m ajor role in raising the overall literacy levels in K erala (Jeffery 1987, 1992). 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1.5 Summing up The detailed com parison o f schooling in U ttar Pradesh and K erala shows that the com m unity initiatives, the leadership o f the C om m unist P arty, the favorable cultural attitudes and governm ent policy that helped to spread m ass literacy in K erala w ere absent in U ttar Pradesh. This absence resulted in the low literacy rates, and the considerable disparities based on region, gender and social class in U ttar Pradesh. Literacy is basic requirem ent for aw areness and collective action by the people to prom ote and safeguard their interests. Lack o f literacy handicaps the less advantaged sections in effectively dem anding public policy favorable to them and in ensuring that they get w hat is granted to them by law. 4.2 Health services in Kerala and Uttar Pradesh A ccessibility and affordability o f health services has an im portant and direct im pact on the health and nutritional status o f a population, in particular, o f children. Prenatal care o f pregnant w om en, assistance o f trained personnel during delivery, im m unization against preventable diseases and the treatm ent o f maternal and childhood diseases are all factors that affect the grow th o f children. Il l Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2.1 H ealth services in U ttar Pradesh The indicators o f m ortality and longevity in U ttar Pradesh w ere one o f the poorest in India (Table 1.5). M ale and fem ale life expectancy at birth w as one o f the low est and child m ortality rates (the probability o f a ch ild ’s dying before the age o f two m ultiplied by 1000) in U ttar Pradesh were one o f the highest in the w orld; m aternal m ortality rates in 1982-86 w as estim ated to be 931 per 100000 births, the average for the four southern states o f India w as 365 (D reze and G azdar 1996). A new bom girl in K erala could expect to live 20 years longer than in U ttar Pradesh (G overnm ent o f India 1993a). Table 4.8 gives a com parative picture o f factors affecting m aternal and child health in K erala and U ttar Pradesh. T he contrast is striking in each o f these indicators that have bearings on the health and nutritional status o f adults and children. O nly 44 percent o f the m others received tetanus toxoid vaccines and only 30 percent had prenatal checkups in U ttar Pradesh. In K erala nearly 100 percent o f the m others received these during pregnancy. O nly 4 percent o f deliveries took place in m edical institutions in U ttar Pradesh as against 92 percent in Kerala. The availability o f hospital beds in U ttar Pradesh w as only one-seventh o f that o f K erala. The public distribution system (PD S) in U ttar Pradesh contributed very little. In Kerala, eighty-eight percent o f the rural population received food grains from the PDS and the per capita quantity supplied was tw enty tim es as high as in U ttar Pradesh. T he people in U ttar 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Pradesh appeared to be benefiting very little from these m edical and public services (D reze and G azdar 1996). The health and other facilities available to the people in U ttar Pradesh w ere highly inadequate. M ore than three quarters o f the villages in U ttar Pradesh did not have any health facility. T w o-thirds o f the villages did not have a trained birth attendant. Only 4 percent o f w om en had deliveries in m edical institutions in U ttar Pradesh (Table 4.8), even though tw enty percent o f the villages had a health center or a sub-center (Table 4.9). T he non-utilization o f these public m edical facilities could be either because services are not provided or because o f decisions m ade to have deliveries at hom e. A nganw adis are centers that provide advice related to pregnancy and child care, nutrition to pregnant m others and young children, and other related services. T hey w ere available only in 19 percent o f the villages in U ttar Pradesh. O nly 38 percent o f the villages had a PDS shop. In K erala on the other hand, all these facilities w ere available to nearly all villages. T able 4.9 lists the am enities available in the villages o f U ttar Pradesh and Kerala. T he situation in U ttar Pradesh w as actually m uch w orse than w hat is reflected in Tables 4.8 and 4.9. M any o f these facilities existing in governm ent records did not actually function, as w as found by the field investigation (D reze and G azdar 1996). A s in the case o f schooling, there w ere also considerable disparities in health indices. The literacy rates w ere the highest, and child 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.8 Select indicators of health services and public distribution system in Uttar Pradesh and Kerala (1980-90) U ttar Pradesh Kerala P ercentaee o f births preceded bv Tetanus vaccine (1992-93) 44 94 Prenatal checkup (1992-93) 30 97 Percentage o f births taking place in m edical institutions (1991) 4 92 Percentage o f children aged 12 -23 m onths w ho received som e vaccination (1992-93) 57 89 Percentage o f villages with medical facilities (1981) 10 96 N um ber o f hospitals beds per m illion persons (1991) 340 2418 Percentage o f the rural population receiving subsidized cereals from the public distribution system (1986-7) 2 88 Per-capita supply o f food grains through the public distribution system (1986-7) (kg/year) Source: D reze and G azdar 1996. 3 60 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.9 Percentages of villages with selected public amenities, 1992-93 U ttar Pradesh Kerala H ealth Facilities Any health facility 23 98 Prim ary health center or sub-center 20 96 Trained birth attendant 33 46 M obile health unit 0.4 27 O ther institutions A nganw adi (M other and C hild 19 99.5 Centers) F air price shop (PDS) 38 97 C ooperative society 14 87 M ahila mandal 5 89 Youth club 14 96 Source: D reze and G azdar 1996 m ortality and poverty rates w ere the low est in the H im alayan region o f the state. O ther regions w ere quite behind. For instance, the fem ale child m ortality rate in 1981 w as 106 in the H im alayan region and 166 in the southern region; the m ale 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. literacy rate in the 7+ age group in 1991 w as seventy-six percent in the H im alayan region and the next highest rate was fifty-eight in the Southern region, other regions being betw een fifty-five and fifty-six; the rural poverty rate in 1987-88 w as eight in the H im alayan region and fifty in the southern region (D reze and G azdar 1996). T he poor functioning o f governm ent health facilities and other institutions has been docum ented by m any studies (K han et al. 1986; Sinha, A 1995; M urthi et al. 1996). T hese are likely to affect w om en, pregnant m others and young children m ore than m en in settings like U ttar Pradesh. Few er w om en in U ttar Pradesh w ere literate, fem ale seclusion w as com m only practiced, and w om en rarely traveled unaccom panied by m en, in particular in rural areas (D reze and G azdar 1996). The entire governm ent health system in U ttar P radesh w as found to be preoccupied w ith m eeting fam ily planning targets rather than m eeting the health needs o f the people (D reze and G azdar 1996). The public health system in U ttar Pradesh w as as poor as the public school system . The reasons for the non-functioning o f the health system w ere very sim ilar to those found in the case o f the school system . They included absenteeism and shirking by governm ent m edical personnel, absence o f public action to ensure that public facilities functioned efficiently, the low priority attached by the governm ent to m eeting the 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. health needs o f the people and the obsession o f the governm ent m edical system and the adm inistration w ith m eeting fam ily planning targets. 4.2.2 H ealth services in K erala Table 4.8 show ed that the utilization o f prenatal, delivery and im m unization services was nearly 100 percent in K erala. M edical facilities were available in alm ost all the villages and a high percentage o f the rural population obtained substantial am ounts o f food grains from PD S shops. T able 4.9 show ed that m ost o f the villages in K erala had the basic facilities required by the people. H ow ever, m ost o f these facilities did not exist in the 1950s. T here w ere also disparities in m edical facilities and dem ographic indicators betw een M alabar and the rest o f K erala as show n in Table 4.10. H ow ever, these disparities w ere reduced in a relatively short period through public policy as show n in T able 4.10. The num ber o f hospital beds in M alabar increased 2.3 tim es betw een 1951 and 1971 w hile the increase w as 1.7 tim es in the rest o f the state. T he death rate went dow n by 55 percent in M alabar and by 34 percent in the rest o f the state. K erala m ade further progress and doubled the num ber o f hospital beds in rural areas per 100,000 persons betw een 1979-80 and 1991-92 (C entral B ureau o f H ealth Intelligence 1981; Z acharia 1984; H ealth M onitor 1994). D uring the sam e period 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the corresponding increase in India w as tw elve to tw enty beds. T h e p er capita availability o f hospital beds in K erala w as ten tim es higher than India by 1991. Table 4.10 The number of beds in medical institutions and death rates in the two regions of Kerala during 1951-1971 Region N um ber o f beds per 100000 D eath rate people in rural areas 1951-52 1960-61 1970-71 1956-57 1970-71 M alabar 33 49 76 23 10.4 T rav an co re- 67 94 117 12 7.9 Cochin Source: U nited N ations 1975 A unique advantage o f K erala am ong Indian states w as its high density o f population and the large population size o f its villages. L arge populations m ake it viable to provide services w ithin villages. In 1991 the num ber o f persons per square m ile w as 747 in Kerala, 471 in U ttar Pradesh and 267 in India as a whole. H ow ever the size o f the village populations w ere disproportionately larger in Kerala. The favorable position o f K erala can be seen from Table 4.11. Seventy- four percent o f the villages in Kerala had populations exceeding 10,000 and 92 percent o f the villages had populations in excess o f 5000. N early 50 percent o f the villages in U ttar Pradesh had less than 500 people, seventy-four percent o f its 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. villages had less than 1000 people and ninety-tw o percent o f its villages had less than 2000 people. There w as no other state in India that was close to K erala in this respect. Partly due to its favorable distribution o f population, m ost o f the villages in K erala had all the basic am enities. T he N ational Fam ily H ealth Survey o f India (1992-93) found that ninety-six percent o f the villages in K erala had a governm ent Prim ary H ealth C enter or a Sub-C enter com pared to only tw enty percent in U ttar Pradesh. T he distance to the nearest tow n or transportation facility w as m uch closer in K erala: tw enty-seven percent o f the villages w ere w ithin 5 km and another thirty-three percent w ere w ithin 5 to 9 km . In U ttar Pradesh only 2.5 percent o f the villages w ere w ithin 5 km of the nearest tow n or transport facility; tw enty-six percent o f the villages w ere within 5 to 9 km ; and the rem aining 71.5 percent villages w ere farther away. H ow ever, even m ore im portant than distances and physical infrastructure w ere the actual utilization rate o f health and other facilities. This w as very high in K erala (D reze and Sen 1996, R am achandran 1996). T he people w ere keenly aw are o f their rights to health care and dem anded that public health services work efficiently. M encher (1980) w rote: In K erala if a PH C 2 w ere unm anned for a few days, there w ould be a dem onstration at the nearest 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.11 Percentage distribution of villages according to population in Indian states: 1981 State Population o f villages Less 500 1000 2000 5000 M o re Total than to to to to than 500 999 1999 4999 9999 10000 % % % % % % % A ndhra Pradesh 32.4 19.1 16.3 20.5 3.9 0.6 100 Bihar 50.1 23.4 16.0 8.6 1.6 0.3 100 G oa 30.3 14.6 24.8 21.8 8.3 0.2 100 G ujarat 26.8 27.4 28.2 15.0 2.3 0.3 100 H aryana 22.2 26.3 27.9 20.0 3.3 0.3 100 Him achal Pradesh 89.9 7.4 2.2 0.5 0.0 0.0 100 Jam m u & K ashm ir 51.0 26.5 16.1 5.8 0.3 0.0 100 K arnataka 41.8 27.2 19.4 10.1 1.5 0.1 100 Kerala 0.4 0.2 1.1 5.9 18.2 74.2 100 M adhya Pradesh 58.5 27.0 11.4 2.9 0.2 0.0 100 M aharashtra 35.0 31.3 22.8 9.2 1.5 0.3 100 M an ip u r 73.4 13.0 8.6 4.4 0.6 0.0 100 M eghalaya 91.6 6.2 1.7 0.5 0.0 0.0 100 2 Primary Health Center (o f the government) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.11 continued State Population o f villages Less 500 1000 2000 5000 M ore Total than to to to to than 500 999 1999 4999 9999 10000 % ° A ° A % % % % Nagaland 61.2 21.7 12.9 4.2 0.1 0.0 100 O rissa 66.4 21.1 9.8 2.7 0.1 0.0 100 Punjab 35.9 30.1 22.8 10.2 1.0 0.0 100 R ajasthan 52.2 24.4 15.6 6.7 0.9 0.1 100 Sikkim 53.4 31.1 13.6 1.8 0.0 0.0 100 Tam il N adu 16.0 20.0 28.9 27.7 6.3 1.1 100 T ripura 14.6 47.9 29.6 32.0 6.5 0.9 100 U ttar Pradesh 47.3 26.7 18.1 7.2 0.7 0.0 100 W est Bengal 41.9 24.7 19.8 11.5 1.8 0.2 100 India* 48.6 24.4 17.0 8.4 1.3 0.3 100 *Excludes A ssam Source: A dapted from A garw al, A.N. and H.O. V arm a, 1996. 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. collectorate3 led by local leftists, w ho w ould dem and to be given w hat they knew they w ere entitled to. .. .[T]he availability o f doctors at a prim ary health facility, and public know ledge that som ething w ill be done at any tim e o f day or night if it is an em ergency, has gone a long w ay in lowering child death. N ag (1989) also found that public aw areness contributed to the efficient provision o f health services in Kerala. T he partnership o f the governm ent and the private sector that existed in the case o f schooling in Kerala, also existed in the area o f health care as shown in Table 4.12 Table 4.12 Percentages of hospitals and beds according to ownership in Kerala in 1991-92 G overnm ent organizations T otal Source: H ealth M onitor 1994 H ospitals Beds D ispensaries Beds 6.8 35.5 2.7 100 93.3 64.5 97.3 0 100 100 100 100 3 District administration office 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In 1991-92 the governm ent m anaged 6.8 percent o f the hospitals and 35.5 percent o f the hospital beds w hereas the private sector m anaged 93.3 percent o f the hospitals and 64.5 percent o f the beds. In India the private sector m anaged only 57.4 percent o f the hospitals and 32.3 percent o f hospital beds (H ealth M onitor 1994). In hospitals run by the governm ent, churches, and com m unity organizations, treatm ent w as provided free o f cost or at a nom inal charge. The private hospitals in m ost states o f India w ere sm all w ith only basic facilities. In K erala, the hospitals run by com m unity organizations w ere large institutions w ith m any beds, specialist doctors and m odem m edical equipm ent (H outart and Lem ercinier 1979). The easy accessibility o f these institutions by public transport put high quality health services within the reach o f the com m on people in K erala. In addition, the traditional ayurveda system o f treatm ent, w as also popular. T here w ere thirty-seven hospitals and two hundred and seven ayurveda dispensaries in K erala w here 2.4 m illion patients were treated in 1967-68 (U nited N ations 1975). A yurveda is still very popular in Kerala. 4.2.3 Sum m ing up The health system in K erala w as functioning well and providing high- quality services to the people. The utilization o f health facilities w as high. M ore than ninety percent o f the deliveries took place in m edical institutions and 89 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. percent o f the children tw elve to tw enty three m onths received som e vaccination (Table 4.8). Both the public and the private sector contributed to the creation o f health facilities and provision o f services. G overnm ent hospitals in K erala had 35.5 percent o f the beds (T able 4.12), and thirty-nine percent o f the deliveries took place in public facilities (UPS 1995). Forty-nine percent o f the deliveries took place in private institutions (UPS 1995). T he role o f the private sector (which included m any m odem facilities m anaged by com m unity organizations) in health care w as m uch higher in K erala than in India as a w hole. The com m on people o f the state effectively dem anded and obtained efficient and tim ely services from public institutions through political and other organizations (D reze and G azdar 1996; M encher 1980). This helped to m aintain the efficiency o f public services. T he high literacy rate and aw areness o f their rights helped em pow er the people to do so. Even though tw enty p ercent o f the villages in U ttar Pradesh had a prim ary health center or a sub-center (T able 4.9), only seven percent o f the deliveries took place in public facilities in U ttar Pradesh (IIPS 1995). T herefore the utilization o f existing facilities in U ttar Pradesh can be considered to be low. T he p o o r quality o f services and the w rong priorities o f the public health system , the social and cultural restrictions on w om en, and the low fem ale and m ale literacy rates, appear 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to be the m ain causes o f this. In rural K arnataka4, C aldw ell, R eddy and Caldwell (1983) found that som e education was necessary for a person to identify with m odem m edicine, as against traditional cures. In Kerala, N ag (1985) found that wom en, w ho w ere m ore literate, were less fearful o f m edical exam ination. Fifty- five percent o f the w om en in U ttar Pradesh did not have prenatal care in Uttar Pradesh; o f this, sixty percent cited “not necessary” as the reason for not obtaining prenatal care and nine percent cited “not perm itted to g o ” as the reason (IIPS 1995). O nly four percent o f the deliveries took place in private hospitals; only fifty-seven percent o f the children tw elve to tw enty-three m onths old had any im m unization; and there w ere only 340 hospital beds p er m illion people. Thus the contribution o f the public and the private sector in public health was highly inadequate in U ttar Pradesh. The people, despite being dissatisfied appeared to have done nothing to im prove m atters by m aking dem ands for better services (D reze and G azdar 1996). The w idespread illiteracy, social division based on caste, the culture o f seclusion o f w om en in their hom es and lack o f com m unity and political leadership to advocate the basic needs o f the people appear to be the m ain reasons for the poor w orking o f the health system in U ttar Pradesh. 4 A south Indian state that has a common border with Kerala 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3 Public distribution system The m ajority o f people in developing countries spend a large portion o f their incom e on food. Y et their incom e m ay not be enough to ensure the regular supply o f food required by the household year round. W hen food prices go up or other unexpected expenses arise, they are forced to reduce food consum ption. Thirty-six percent o f the households in rural India did not ow n any land and 2.5 percent ow ned less than one acre o f land (IIPS 1995). Yet, the share o f land in the assets o f rural households was sixty-eight percent in K erala and sixty-five percent in U ttar Pradesh (Sharm a 1994). In these conditions, the availability o f food and other basic consum ption item s at regulated prices can greatly enhance the health and w ell-being o f the people, in particular children. A w ell-functioning public distribution system (PD S) can do this. The creation and m anagem ent o f the PDS in India is left to the state governm ents. A s w e saw in tables 4.9 and 4.8, the PDS in K erala covered ninety- seven percent o f the villages and supplied 60 kg o f food grains p er head per year. Since only eighty-eight percent o f the households obtained the food grains, the per capita purchase o f those w ho bought the food grains w ould still be higher. There w ere PDS outlets only in tw o percent o f the villages in U ttar Pradesh and the per capita supply w as ju s t 3 kg p er year. T he PDS in U ttar Pradesh supplied the 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. low est per capita quantity o f food grains am ong m ajor Indian states (D reze and G azdar 1996). Even this m eager quantity distributed, how ever, did not fully benefit the poor because PDS outlet ow ners sold o ff a large portion o f the supplies in the black m arket at a profit (D reze and G azdar 1996). Thus the people in U ttar Pradesh, unlike in K erala, hardly benefited from the PDS. 4.3.1 The creation o f the PDS in Kerala The precarious food situation that existed in K erala and how security o f food was im proved in K erala through the PDS, is discussed below . The PD S in K erala’s form er regions was established during W orld W ar II w hen there w ere several cases o f fam ine deaths and m alnutrition. T here w ere near-fam ine conditions prevailing in M alabar and Travancore and m any died o f starvation (R am achandran 1996). The people staged dem onstrations dem anding that the governm ent supply food to them through PDS shops at reasonable prices. A s a result the governm ents procured food grains from farm ers and started distributing them through PD S outlets in all the three form er regions o f K erala. H ow ever, the establishm ent o f an effective PDS took place after 1957. T he C om m unist M inistry in K erala negotiated and obtained an assurance from the governm ent o f India to supply food grains to K erala for distribution through the PDS. This led to the establishm ent o f the PDS in K erala. 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. D andekar and Rath (1971) estim ated that in 1961, the highest levels o f poverty in India w ere in K erala w here ninety percent o f the population w as below the poverty line, taking 2250 calories as the per capita daily required consum ption. K erala had roughly thirty percent o f its cultivated land under food crops since the 1960s and produced about thirty percent o f its requirem ent o f rice, w hich w as the staple diet, and farm ers used m ost o f the rice they produced for self-consum ption (M ooij 1999). M ost o f the rice sold in the m arket was im ported into the state from other states o f India. The price o f rice in K erala was very sensitive and fluctuated w ith, am ong other factors, changes in railw ay freight, m arket prices in other states supplying rice to K erala, and the food released for public distribution in K erala by the Food C orporation o f India (an agency that holds stocks o f food grains on b eh alf o f the governm ent o f India). The low levels o f incom e in K erala m ade fluctuations in food prices a politically sensitive issue (M ooij 1999). In 1964, the G overnm ent o f India discontinued the existing arrangem ents for food supply when there w as a general shortage o f food in the w hole country and there was no popularly elected governm ent in K erala. This led to a sharp rise in food prices and popular protests broke out all over the state led by the C om m unist Party. T he protests could not be controlled, and it w as form ally decided by the G overnm ent o f India to introduce public distribution o f food grains 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in K erala from N ovem ber 1, 1964. The G overnm ent o f India undertook to supply food grains to the PDS in K erala regularly after procuring food grains from other states (Ram achandran 1996). Food had becom e a m ajor issue in the elections and this w as considered as the m ain reason for the decision o f the governm ent o f India to supply food grains to K erala (M ooij 1999). Thereafter, there w as a m ajor expansion o f the PDS in K erala, and it w as further expanded in the 1980s. K erala was the only state in India that had a PD S covering the entire population. A bout 97 percent o f the fam ilies had PDS cards in 1987 (K oshy et al. 1989 cited by R am achandran 1996). In January 1992 there w ere 13028 authorized retail outlets, one for every 389 households in K erala (G overnm ent o f K erala 1992). This ensured that PDS food m aterial was easily available to the people. It w as estim ated that the PDS supplied about one-third o f the total requirem ent o f food grains in K erala (M ooij 1999). This com prised fifty percent o f the total supply o f calories consum ed by households in K erala (M ooij 1999). A survey in 1987 found that the poor used the system m ore than the rich and obtained m ost o f their requirem ents o f food grains from PDS outlets. Low -incom e beneficiaries w ere m ostly satisfied w ith the quality, price and the service they obtained in PDS outlets (R am achandran 1996). D reze and Sen (1989) considered that the PDS in K erala had contributed substantially to food price stabilization and protection o f the entitlem ents o f the people. By increasing the supply o f 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. food grains and other essential com m odities during difficult periods such droughts and floods, the PD S has been able to stabilize food prices in K erala (Balakrishnan and R am asw am y 1992). A com parison o f the prices o f rice and w heat distributed through the PDS and the open m arket prices o f rice and w heat show ed that PDS prices w ere roughly one-third low er from 1980 to 1994 (M ooij 1999). A ccordingly it was estim ated that the average per capita financial transfer through the PD S was Rs.15 per head and Rs.75 p er fam ily per m onth in 1991. This cam e to R s.20 p er w eek per fam ily and m ay be considered as a substantial assistance to poor fam ilies in K erala w here the annual per capita incom e w as Rs.4600 in 1992 (T able 1.5). In addition, the PDS outlets sold other essentials like w hole w heat flour, sugar, kerosene for cooking and lighting, pulses, spices, condim ents and cloth at prices low er or considerably low er than prevailing m arket prices. The K erala State Civil Supplies C orporation, a governm ent agency established in 1974, operated its own stores selling rice, w heat, pulses, vegetables, and a large num ber o f other essential item s procured by it at w holesale prices to consum ers. The state governm ent has been able to recover the cost o f the PDS w ithout subsidizing its operating costs (G eorge 1979). The only subsidy provided w as by the governm ent o f India, to cover difference betw een the issue price and the total cost o f procurem ent (R am achandran 1996). The assessm ent o f G eorge (1979) has been that the gains 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to producers and consum ers from the PDS in K erala exceeded the cost o f governm ent subsidy borne by the G overnm ent o f India. 4.3.2 A field study in Kerala In his recent study o f the PDS system s in the states o f K erala and K arnataka, M ooij (1999) found that the PDS in K erala w as largely free o f the corruption and m alpractices that characterized the PD S in the neighboring state o f K arnataka. Instances o f paying bribes to get PDS card s5 w ere not at all found in K erala, w hereas bribes w ere com m only paid to get PD S cards in the neighboring state o f K arnataka. A lm ost all village households interview ed in K erala had PDS cards, w hereas five to tw enty percent o f households in K arnataka did not have them . The near 100 percent coverage o f the rural households through PD S cards w ithout giving bribes w as attributed by M ooij (1999) to the w idespread literacy and w ell-developed sense o f rights and entitlem ents in K erala, w hich m ade the people m ore assertive and apt to resist unlaw ful actions restricting access to public program s. T he study found that the public adm inistration in K erala w orked effectively and in favor o f the people. O n the other hand, vigilance com m ittees appointed in K arnataka by the governm ent to supervise the w orking o f PDS 5 PDS cards are issued to households specifying the maximum amount o f food grains and other commodities they can get per week or month, usually based on the size o f the household. 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. outlets w ere packed w ith the supporters o f the ruling party, w ho often m isused th eir position to obtain bribes and favors for them selves from the PDS outlet ow ners. T he outlet ow ners in turn w ere allow ed to com m it m alpractices and w ere given protection by the com m ittee m em bers. In K erala, there w ere no such com m ittees. T he people directly, or through local political activists or other organizations, m ade com plaints to responsible officials. Im m ediate action was taken, and even on oral or anonym ous com plaints licenses o f the PD S outlets were suspended pending a quick investigation. The licensee was cleared, or punished and rem oved, and the supply was restored forthw ith by appointing a new licensee. Prom pt action w as taken by officials at low er levels to avoid com plaints about their inaction or delays. Such action by the people kept the licensed ow ners o f PD S outlets on their toes. The study also found that politicians in K erala advocated the interests o f the people. T his w as also partly due to the aw areness o f the people and the fear o f exposure. Therefore, political functionaries aided rather than hindered the proper functioning o f the PDS. Food security and ability o f the people to protect their entitlem ents and dem and their rights are interlinked w ith other aspects o f public policy. Land reform s in K erala, for instance, had m any beneficial effects on the poor. A lm ost all households o f agricultural w orkers gained ow nership o f about one-tenth o f an acre o f land around their house as a result o f the land reform s. T his gave the 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w eakest sections o f the society secure housing sites from w hich they could not be evicted, strengthening their position for dem anding other rights. For instance, take the case o f a person having his living on land ow ned by a PD S licensee as a tenant, w ho could be evicted from the hom estead at short notice. This person cannot com plain against the irregular functioning o f the PDS outlet, w ithout facing the risk o f eviction from his hom estead. This w ould reduce the p erso n ’s ability to dem and his rights. Even small additional assets help people to increase their w ell being. For instance, it was a com m on practice am ong the poor in India to raise loans for em ergencies by m ortgaging their PDS cards (Gulati 1977). H ow ever when they did so, they lost the subsidy in the PD S food, for the period o f the m ortgage. In Kerala, the poor w ere able to pledge a few coconut trees on their hom estead lands for raising loans instead o f the PDS cards (M ooij 1999). Thus even sm all assets provided to the poor through public policy can im prove their financial, nutritional and political position. I f regular supply o f food is uncertain, people are often forced to agree to unfair tenancy arrangem ents and w orking conditions, and m ay be forced to take loans at very high interest rates to obtain food. O nce food supply is ensured at an affordable price, w eaker sections are able to dem and other rights, including the efficient w orking o f the PD S. The proper im plem entation o f food policy and other public policies also depends on other policies that enhance the capabilities o f the people, such as education. 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The state and federal governm ents in India are both developm ental and rent seeking (M ooij 1999). Those in pow er aim to m axim ize their own w elfare unless they are, through dem ocratic m eans, forced to enhance people’s welfare. For instance, the politicians in pow er, who have the financial ability to obtain all their supplies o f food grains from the open m arket w ould not norm ally be concerned about the PDS. O nly if the poor dem and it, and vote for those w ho prom ise to set it up, w ill those in pow er w ould set up a PD S. T he actions that politicians take depend, to a large extent, on the dem ands that people m ake, their vigilance and the fear o f exposure by the opposition. In K arnataka, the PDS outlet ow ners w ere found to be im posing restrictions on the poor so that they w ould not able to buy the food grains they need. For instance, they m ade it a condition that households buy their entire supply o f both rice and w heat together. M any households did not consum e w heat regularly and did not w ant to buy w heat. Further, m any daily w age earners did not have the m oney to pay for the full w eek ’s allocation o f grains at the sam e time. T his prevented them from buying from the PDS. The PDS licensees sold excess food grains in the m arket at higher prices and m ade an irregular profit. Had the people been vocal, as in K erala, the PDS licensees could not have im posed such restrictions. In Kerala, the legitim acy o f the ruling party w as closely linked to food prices and the PD S. T his m ade 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. politicians to cham pion the cause o f PDS and act as w atchdogs. Thus the m any aspects o f public policy are interrelated. 4.3.3 Sum m ing up It has been argued that the rem arkable achievem ents o f K erala in raising the indices o f the quality o f life w ould have been difficult w ithout a well functioning PDS. On the other hand, the PDS did not apparently m ake any significant positive im pact in U ttar Pradesh. D ata in Table 4.8 show that PDS w as not a public policy priority, nor w ere the people able to force the governm ent to expand and ensure the effectiveness o f the system . The w idespread illiteracy and lack o f aw areness in rural U ttar Pradesh, and the social divisions based on caste preventing organized dem ands by the disadvantaged, appear to be behind this failure, too. 4.4 Land reforms A s in the case o f the PDS, land reform s are another area w here K erala and U ttar Pradesh are cases o f relative success and failure respectively. Land is an im portant asset in rural India and land reform law s passed by the Indian Parliam ent provided an opportunity for the poor to im prove their w ell-being. The im plem entation o f land reform law s w as left to the states. 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.1 Land distribution in Uttar Pradesh and Kerala before and after land reforms The Indian Parliam ent abolished the Zamindari6 system o f land tenure all over India soon after independence in 1947. It w as left to the Indian states to enact law s laying dow n appropriate land ceilings and for the distribution o f surplus land. The enactm ent o f basic laws took the better part o f a decade in U ttar Pradesh. U ttar Pradesh w as prim arily agricultural, w ith over 80 percent o f the population living in rural areas and engaged in agriculture in 1991. Agriculture w as the m ain source o f livelihood o f the rural population and determ ined their standard o f living. H ow ever, land reform s w ere not effectively im plem ented in U ttar Pradesh. Inequalities in landholdings did not decline and the share o f the land ow ned by the bottom 40 percent did not increase (T able 4.13). The m axim um ceilings fixed for landholdings in U ttar Pradesh w ere higher than in any other state in India, w ith num erous loopholes in the law and in its application that enabled the landlords to retain m uch o f the land (Singh and M isra 1964; H aque and Sirohi 1986; K ohli 1987; H asan 1989). The pow erful rural elite also m anaged to grab m uch o f the land that w as allotted for com m on use o f the village (R am achandran 1996). Land reform s in U ttar Pradesh m ade no change in the 6 Zamindars were landlords who were given legal rights to the land and revenue collection by the British government. Zamindars sublet the land to tenants, collected revenue and paid the British, after retaining a part o f it. 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. shares o f land held the bottom 40 and top 10 percent o f the landow ners as show n in T able 4.13. Table 4.13 Percentage shares of land owned by the bottom forty and top ten percent of households in Uttar Pradesh and Kerala State Share o f the bottom 40 Share o f the top 10 percent landholders percent landholders 1953-54 1982 1953-54 1982 U ttar Pradesh 2.5 2.6 45.8 45.5 Kerala 3.5 8.3 67.0 55.4 Source: A dapted from Sharm a 1994 T he im provem ent in production techniques, popularly know n as the G reen R evolution, increased productivity and private incom es, but there w as very slow im provem ent in the lives o f the m ajority o f the people in U ttar Pradesh as they did not own m uch land (D reze and G azdar 1996). A s such the inequalities in the rural areas w ere not reduced and a great opportunity for progress w as lost. In K erala, on the other hand, land reform s had a significant effect. T he share o f the bottom forty percent o f households w ent up by nearly 2.5 tim es betw een 1953- 54 and 1982 (Table 4.13). The share o f the top ten percent declined by seventeen percent. T he land reform s helped to reduce rural inequalities. T he data for the 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. top one percent o f the households bring out the difference in the im plem entation o f land reform s in the two states. The share o f the top one percent in U ttar Pradesh declined only m arginally from 11.2 percent to 10.5 percent; in Kerala, there w as near fifty percent reduction, from 26.7 percent to fourteen percent (Sharm a 1994). Table 4.14 show s the percentage o f households ow ning no land in 1953-54 and 1982. There w as a 64 percent reduction in the percentage o f landless households in K erala, w hereas in U ttar Pradesh the reduction w as less than fifty percent. The data in tables 4.13 and 4.14 show that a larger percentage o f households in K erala benefited from land reform s and the percentage o f total land redistributed w as higher than in U ttar Pradesh. Table 4.14 Percentages of households owning no land in the rural areas of Uttar Pradesh and Kerala State 1953-54 1982 U ttar Pradesh 9.36 4.85 Kerala 36.3 12.8 Source: A dapted from Sharm a 1994 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.15 show s the percentage o f households that did not operate7 any land in U ttar Pradesh and Kerala. W hile the percentage o f such households quadrupled in U ttar Pradesh betw een 1953-54 and 1982, there was a reduction o f tw enty-seven percent in households that did not operate any land in K erala. W hen w e com pare the figures in Table 4. H a n d T able 4.15 the paradoxical position o f U ttar Pradesh com es to light. W hile only 4.85 percent o f the households w ere landless, tw enty percent w ere not operating any land. This confirm s reports that lands allotted to the poor in U ttar Pradesh w ere actually not handed over to them , and w ere under the illegal cultivation o f pow erful persons in the village (Shankar 1991). The latter group o f persons for the m ost part continued to cultivate their lands that w ere declared surplus, and in addition, often cultivated com m on village lands that w ere either allotted to the w eaker sections or not allotted to anyone. In either case the poor w ere too w eak to challenge them (Shankar 1991). In K erala, on the other hand, there w as hardly any difference in the percentages o f households not ow ning and not operating land, show ing that persons w ho w ere allotted lands actually possessed and cultivated those lands (Tables 4.14 and 4.15). The above conclusions, based on data in the last tw o tables are supported by the data in Table 4.16. B etw een 1953-54 and 1982, the share o f the bottom 7 Operating land is different from owning it. A person may own land according to official records, but may not be cultivating the land because someone else is, by force, cultivating it. 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. forty percent o f the households in the area operated declined from 3.8 percent o f the land operated to 2.2 percent o f the land, a reduction o f forty-tw o percent. In K erala, during the sam e period, there w as an increase o f 117 percent. The rural rich in U ttar Pradesh appear to have not only m anaged not to give up surplus lands, but also w ere able to take possession o f lands that should have gone to the disadvantaged sections according to the laws enacted. T his did not happen in K erala and the disadvantaged sections w ere able to get their share o f land according to the land reform laws enacted in the state. H ow these different results occurred in K erala and U ttar Pradesh is discussed below. Table 4.15 The percentages of households not operating any land in Uttar Pradesh and Kerala State 1953-54 1982 U ttar P radesh 5 20 Kerala 18 13 Source: A dapted from S hanna 1994 Also, the government may allot land to a person, but until a specific piece o f land is handed over, the person cannot start cultivation. In such cases the person owns the land, but does not operate it. 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.16 The percentage share of the bottom forty percent of the households in the area of land operated in Uttar Pradesh and Kerala State 1953-54 1982 U ttar P radesh 3.8 2.2 K erala 1.8 3.9 Source: A dapted from Sharm a 1994 4.4.2 The struggle for and the results o f land reform s in K erala The history o f land reform s in K erala, as in the case o f other public policy m easures, w as one o f achievem ents gained by sustained struggles o f the people, in w hich the C om m unist Party played a m ajor role. Table 4.13 show s that the distribution o f land w as m ore unequal in K erala than in U ttar Pradesh in 1953-54. T here w as an oppressive system o f land tenure in the M alabar region o f Kerala, w hich had been under the British rule since 1792, w ith very large landlords, and im poverished and exploited tenants (R am achandran 1996). In 1920-21, thirty- tw o landlords in M alabar held 628,921 acres (51 percent) o f the total o f 1,229,217 acres o f the cultivated land in M alabar (Panikkar 1989). O ne landlord (Sam uthiri o f K ozhikode) ow ned land in 520 villages. All the agricultural, forest and cultivable w astelands in M alabar w ere ow ned by the landlords. B elow them were 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the superior tenants. T hey w ere the interm ediaries w ho hired the land from the landlords and sublet it to the tenants. B elow the tenants w ere the landless w orkers. The landlords had absolute right to all the lands under the law. They could evict tenants at will, including from their hom esteads, m aking them hom eless. T he interm ediaries had a longer tenure o f tw elve years, but they had to pay very heavy rents to the landlord when the leases w ere granted and renewed. T he B ritish, the landlords and the interm ediaries tried to get as m uch out o f the tenants as possible. The levies w ere so oppressive that rents varied betw een 75 and 85 percent o f the produce. In addition, there w ere about fifty different types o f levies im posed on the people by the B ritish: on cattle, houses, shops, loom s, ferries, fishing nets etc., w hich left virtually nothing ow ned or used by the tenant untaxed (Panikkar 1989). Further, there w ere other exactions on tenants like free labor associated w ith festivals, life cycle events and rituals in the households o f the landlords. W ith all these, the peasants w ere alw ays heavily in debt, because o f w hich, m any o f them becam e bonded laborers and w ere som etim es sold to other landlords, even after slavery w as abolished in 1843. In Travancore, sixty to seventy percent o f all cultivable land belonged to the Padm anabha tem ple situated in the capital city. The rest o f the land belonged to the royal fam ily, the landlords and fam ilies that w ere directly given land by the governm ent. In 1865 the king conferred full rights o f ow nership and transfer o f all 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tem ple lands to the tenants for a sm all paym ent. C ultivable w astelands w ere offered to cultivators along w ith financial incentives to bring them under cultivation. T hese steps m ade agriculture dynam ic, and m any agricultural entrepreneurs took advantage and raised plantations (R am achandran 1996). The agrarian situation in C ochin was som ew here in betw een that o f M alabar and Travancore. Landlords ow ned sixty percent o f the land and the governm ent ow ned forty percent o f the land. Levies w ere high on tenants but not as oppressive as in M alabar. O w nership rights to governm ent lands w ere conferred on tenants in 1905 by the royal adm inistration. Further, cultivable w aste lands did not belong to the landlords, and w ere assigned to cultivators. T herefore the type o f stranglehold the landlords had in M alabar w as avoided and m ade agriculture relatively dynam ic (Ram achandran 1996). W hile it was possible to ow n and cultivate land in T ravancore and Cochin by the latter h a lf o f 19lh and early 20th century respectively, the m ajority o f the people in M alabar w ere oppressed by the land tenure system enforced by the British. T hey lagged behind the rest o f K erala in health, education and econom ic developm ent. The rural society there w as restless. R evolts, robberies, thefts and social banditry occurred regularly as the poor and the oppressed tenants and landless w orkers revolted against exploitation. U prisings o f tenants and agricultural laborers against the landlords and the governm ent started as early as 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1834 and culm inated in the violent rebellion o f 1921 called Mopilla Rebellion, w hich lasted for eight m onths and cost thousands o f lives (R am achandran 1996). T he colonial governm ent ultim ately suppressed the rebellion and conceded few o f the dem ands for econom ic justice. B y 1930s, thousands o f tenants w ho w ere indebted w ere evicted in M alabar. The C ongress Party initially and later the C om m unist Party, m ost o f the leaders o f w hom w ere persons w ho had left the C ongress Party, gave leadership to the discontented and im poverished peasants and agricultural w orkers. In order to m ake the peasants aw are o f the prevailing injustices and to organize them to fight for econom ic and social justice, the C om m unists m ade a com prehensive action plan. T heir aim w as to set up a unit o f the Party, a reading room and a night school in each village. The Peasants U nion established in 1933 by the C om m unist Party becam e the rallying point o f m ost o f the protest m ovem ents. The U nion opposed the landlords by conducting num erous protest m arches and m eetings all over M alabar. The Com m unist Party also organized Peasant U nions in T ravancore and C ochin and organized protest m ovem ents against big landlords there. The m ain dem and o f the protesters w as the grant o f ow nership to the actual cultivators. H ow ever until independence there w as no notable change in land distribution in the three regions o f present-day Kerala. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The com m unist party cam e to pow er in 1957 in the first general election held to the K erala State A ssem bly. The C om m unists had a w ell-planned agenda for the state. W ithin a w eek o f the form ation o f the m inistry the governm ent issued an ordinance bringing in sw eeping land refonns. The m ain features o f the reform s w ere the follow ing (R am achandran 1996): 1. Evictions o f tenants w ere prohibited and the tenants evicted during the previous 10 years w ere restored their land. 2. All rights on land leased out to tenants by landlords and interm ediaries were taken over by governm ent. All arrears o f rent w ere cancelled. 3. Tenants w ere given the ow nership and transfer rights at an affordable price to be paid in easy installm ents. 4. A ceiling on agricultural land holdings o f individual households w as decided. The ceiling depended on the num ber o f persons in the household, the m axim um being 25 acres per household. Land in excess w as taken over by governm ent. 5. Surplus land taken over w as given to tenants and landless w orkers. Taxes on land were drastically reduced. 6. The hom esteads o f the landless w ere given to them w ith ow nership and transfer rights for a sm all paym ent. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T hese reform s evinced strong protests by landow ners and forw ard com m unities in the state w ho stood to lose land. The C ongress Party in K erala supported those w ho opposed land reform s. The G overnm ent o f India, controlled by the C ongress Party, dism issed the C om m unist governm ent in 1959. In the next elections, the C ongress Party cam e to pow er. T hey brought in new legislation, w eakening reform s enacted by the C om m unist Party. B ut even this was challenged and struck down in the courts o f law. Later, the K erala A grarian R elations A ct w as passed in 1964 and placed in the N inth Schedule o f the C onstitution o f India, taking it out o f the purview o f courts. W hen the C om m unist party w as elected to pow er for the second tim e in 1967, they enacted a m ore effective K erala Land R eform s A ct, which becam e the basis for the land reform s carried out in the 1970s. The C om m unist governm ent, how ever, did not rem ain in pow er long enough to im plem ent the provisions o f the A ct. In order to press for the im plem entation o f reform s, peasants and w o rk ers’ organizations staged dem onstrations and cam paigns. T hey had to face the opposition o f the landlords and the state police. D uring June to A ugust 1972 alone 160,000 agricultural w orkers and peasants w ere arrested and 10,000 jailed (N ayanar 1982 cited in R am achandran 1996). H ow ever these m ass struggles m ade the im plem entation o f land reform s unavoidable. 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Tw o sets o f provisions o f the land reform law s that had to do w ith tenancy and hom estead land w ere relatively w ell-im plem ented. A ccording to official data 1,970,000 acres o f agricultural land w as transferred to 1,270,000 households and 20,000 acres o f hom estead land w as transferred to 270,000 households. H ow ever the identification o f surplus land above ceiling w as not successful. O nly 50,000 acres o f surplus land w as identified and transferred to 90,000 households (R adhakrishnan 1989). This w as because a large num ber o f bogus transfers o f land were m ade by landlords in the m eantim e to avoid losing land. M any o f these bogus transfers w ere regularized by the K erala Land Reform s A m endm ent A ct passed by the C ongress governm ent in 1979, thereby underm ining the provisions o f the 1967 A ct. D espite these setbacks, land reform s ended the period o f extrem e concentration and inequalities in rural areas and gave the poor a considerable am ount o f econom ic security (Franke 1993; U nited N ations 1975). There was a great deal o f opposition in K erala to the im plem entation o f land reform s and the C ongress party delayed and w eakened the reform s w henever they cam e to pow er. H ow ever the strong peasants and w orkers m ovem ents led by the C om m unist Party kept up the pressure on governm ents for the im plem entation o f land reform laws. This forced the C ongress Party also to support land reform s to som e extent. The relative effectiveness o f land reform s 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. com pared to other states o f India, along w ith other public policy actions, had a significant role in reducing inequality and poverty in Kerala. 4.4.3 T he failure o f land reform s in U ttar Pradesh L and reform s in U ttar Pradesh w ere poorly im plem ented, largely because o f the inability o f the poor to organize and dem and their rights and the absence o f effective support for land refonns from any source. T here w as no effective support given to the im plem entation o f land reform s b y any political party. L andlords dom inated the C ongress Party that w as in pow er in U ttar Pradesh. The Z am indari A bolition C om m ittee o f U ttar Pradesh, dom inated by C ongress Party leaders, even w hile recognizing the extrem e concentration o f land, recom m ended against any ceiling on land held for cultivation. The reason given for this w as that landlords w ere experiencing hardships because o f the loss o f their position follow ing the end o f the B ritish rule, and taking aw ay land from them w ould increase their hardships further (Shankar 1991). The C om m ittee considered, strangely, that giving land to the tenants, agricultural w orkers and other w eaker sections w ould not benefit them m uch. A ccordingly law s enacted in U ttar Pradesh to im plem ent land reform s laid down the highest ceilings am ong all Indian states. A s a result, the share o f land ow ned by the bottom forty percent and the top ten percent o f the rural households did not change (T able 4.13). On the 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. contrary, the percentage o f households not operating any land increased from five percen t to tw enty percent during 1953-54 and 1982 (T able 4.15). The landlords in U ttar Pradesh sold aw ay large areas o f land w hile the land reform law w as under litigation in courts o f law. M oreover the laws had num erous exem ptions and w ere full o f loopholes in law and im plem entation (Shankar 1991). For instance, groves w ere exem pted, and this allow ed landlords to plant trees on the land and claim exem ption from the ceiling. Lands could also be donated to trusts, cooperatives, educational institutions etc. M any pow erful individuals in villages used these provisions to m ake bogus transfers and continued w ith the cultivation o f those lands. The land ow ned by the villages cam e to 900,000 acres. T hese lands w ere to be distributed to persons ow ning less than 3.12 acres o f land. T he village headm an and the accountant w ere the sole authorities w ho w ere authorized to distribute land. They w ere rarely in favor o f the poor. They transferred a lot o f land to the children o f the rich farm ers, w ho w ere technically landless as they had no land in their names. O nly in 1968 w as the approval o f a sub-divisional governm ent official m ade necessary for transferring village land. B y that tim e m ost o f the surplus lands w ere already transferred in the nam es o f various persons/organizations to evade the land reform laws. 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Further, large areas o f village lands w ere encroached upon by leaders and pow erful persons in the village. C om plaints for eviction had to be m ade by the village headm an in the courts. H e had to produce w itnesses in the court to prove the case. B ut w itnesses w ere often not forthcom ing for fear o f reprisals. There w as also a provision in the law that m ade any encroachm ent o f village land m ore than three years old autom atically legal. A s a result m any o f the encroachm ents becam e legal, and in the rest illegal possession and cultivation continued (Shankar 1991). Even in cases where land w as allotted to landless persons, m any w ere unable to cultivate the land as it w as not dem arcated and handed over to them . U ltim ately, the p o o r in U ttar Pradesh could not benefit from land reform s. 4.4.4 Sum m ing up Land reform s in Kerala, though not entirely successful because o f opposition from those w ho ow ned land and the support they had o f certain political parties, w ere successful in reducing rural inequalities, providing hom esteads, reducing landlessness and abolishing the crushing debts o f the tenants and other w eaker sections in K erala. The security provided by land reform s have enabled w eaker sections to dem and and secure m any other entitlem ents such as better m inim um w ages and subsidized food grains through the PDS. Land reform s w ere one o f the m any policy initiatives taken by the governm ent th at have raised 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the living standards o f the people o f K erala. The C om m unist Party played a leading role, both w hile in and w hile out o f power, in organizing and leading the people. The gains that the poor in K erala had through land reform s w ere not easily achieved. It w as the culm ination o f the long struggles by the oppressed tenants and landless w orkers that started in 1834 in M alabar. In U ttar Pradesh the land reform s were a failure, sim ilar to the other failures areas such as education, public health and the PDS. The disadvantaged sections in U ttar Pradesh could not organize and fight for even the lim ited rights granted to them by under the land reform law o f U ttar Pradesh. The w idespread illiteracy, lack o f aw areness, lack o f leadership for the w eaker sections and social divisions perm itted large landow ners to keep their surplus lands and grab village lands, that was granted to the landless and marginal land holders by the law. 4.5 Conclusion T he states o f India have m uch in com m on by w ay o f physical resources, adm inistrative structure, and opportunities in key areas such as education, public health, public distribution and land reform s. The im pact o f governance on the quality o f lives o f the people depends on public policy and the quality o f the public services in each state. The ability o f the w eaker sections to dem and their rights, organize and struggle for these rights in a dem ocratic 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fram ew ork appears to be an im portant determ inant o f their w ell-being. In some states like K erala m uch was achieved through public policy and p eo p le’s organized action and continued vigilance. U ttar Pradesh is an exam ple where w eaker sections w ere not able to organize and act to im prove their lot. There w ere m any factors that helped in K erala, som e o f w hich are listed below. Initiatives from w ithin the different com m unities for reform and m odernization in the late 19th and the early 20th centuries. The positive response o f the princely states to the dem ands for social, educational and econom ic reform. T he em phasis laid on education o f m ales and fem ales in all the com m unity reform m ovem ents. A culture that encouraged fem ale literacy, and the absence o f fem ale seclusion in m ost o f the com m unities in the state. T he leadership o f the C om m unist Party that laid stress on literacy, raised the aw areness o f the w eaker sections and led them in their struggles for a better life. T he large sizes o f village populations, that m ade it viable to provide the basic services within m ost o f the villages. 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T he high level o f participation o f private and voluntary agencies in the creation and m aintenance o f educational and health facilities. T he rem oval o f m arked social divisions in society that often prevent organized action by w eaker sections. T he determ ined and sustained efforts o f the people to im prove their lot and public policy interventions that usually follow such organized dem ands in dem ocracies, appear to be effective in im proving the quality o f people’s lives. In the absence o f literacy, aw areness, and organized and persistent actions on the part o f the disadvantaged, traditional inequalities and social injustices are likely to persist and public resources are likely to be m isused for private gain. In such conditions the w eaker sections are likely to rem ain deprived. In this chapter, four m ajor areas that affected the lives o f com m on people in U ttar Pradesh and K erala w ere discussed. Sim ilar results w ere found in other areas o f public policy that had to do w ith the w elfare o f the poor and for im proving the nutritional status o f m others and children (D reze and G azdar 1996; D reze and Sen 1995). Som e states in India w ere sim ilar to K erala in their achievem ents and others w ere sim ilar to U ttar Pradesh, as is evident from the data in Table 1.5. The quality o f life o f the population and the nutritional status o f children are intim ately connected. Table 4.17 presents the level o f undernourishm ent am ong children in U ttar Pradesh and K erala. U ndernourishm ent w as m uch higher in 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U ttar Pradesh than in Kerala. Children o f deprived parents cannot be expected to grow norm ally. In order to bring down the rates o f child grow th retardation, the quality o f life o f the whole population has to im prove. This is possible only when the disadvantaged groups becom e literate, aw are and are able to safeguard their ow n interest. L iteracy appears to be a basic requirem ent for this to happen. The com m unity reform m ovem ents in K erala laid stress on literacy and literacy appears to b e the basis for the progress achieved in K erala. Even though raising literacy rates is a long-term process, it appears to be necessary condition for im provem ents in quality o f life. Table 4.17 Percentages of children undernourished in Uttar Pradesh and Kerala in 1991-92 State U ttar P radesh U nderw eight B elow -3 B elow -2 S.D S.D 2 4 .6 5 9 .0 Stunted B elow -3 B elow -2 S.D S.D 3 5 .6 5 9 .5 W asted B elow -3 B elow -2 S D S D 2 .7 16.1 Kerala 6.1 2 8 .5 9 .0 Source and N otes: Sam e as in Table 1.7 2 7 .4 1.3 11.6 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5 DATA AND METHODS 5.1 Source of data T he data used in this dissertation w as produced by the N ational Fam ily H ealth Survey o f India (N FH S) conducted in 1992-93. The survey was initiated by the M inistry o f H ealth and Fam ily W elfare (M O H FW ) o f the G overnm ent o f India w ith financial support from the U nited States A gency for International D evelopm ent (U SA ID ). T he N FH S covered 99 percent o f the population o f India in 24 states and the N ational Capital Territory o f D elhi. T he data w ere collected from April 1992 to Septem ber 1993 in three phases. M aking available national and state level data on fertility and fam ily planning, m aternal and child health, and child nutrition w ere som e o f the objectives o f the N FH S. T he m ain agencies that w ere involved in the exercise w ere the International Institute o f Population Sciences, B om bay (IIPS 1995), eighteen Population R esearch Centers located in Indian states, several consulting organizations, the E ast-W est C enter (H aw aii) and M acrointem ational, Princeton, USA. T he E ast-W est C enter and M acrointem ational, U SA provided technical assistance. The descriptions o f the N FH S survey design and im plem entation and data tabulations w ere published by 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the International Institute for Population Sciences, B om bay in N ational Fam ily H ealth Survey 1992-93. India CUPS 1995). M acrointem ational m ade additional docum entation available at their w eb site http://m acroint.com . T he N FH S data w ere available for dow nloading from the above site. W om en w ere the respondents in the survey. The data for children were created from the w om en-based data using the child package provided by M acrointem ational. The N FH S 1992-93 was very sim ilar to the D em ographic and Health Surveys (D H S) conducted in m any developing countries in survey design and im plem entation, b u t w as not considered a D H S because o f certain differences betw een D H S and N FH S. T h e D HS w ere started in the m id-1980s to assist developing countries in the collection and use o f data to m onitor and evaluate population, health, and nutrition program s. D em ographic and health surveys provide inform ation on fam ily planning, maternal and child health, child survival, H IV /A ID S/STIs (sexually transm itted infections), and reproductive health. In addition, DHS offers assistance for im proving on going data collection efforts, such as health inform ation system s, to efficiently m eet inform ation needs in a cost-effective m anner. D H S I w as carried out during 1984-89, D H S II during 1988-89 and DHS III during 1992-99. B ecause the N FHS w as very sim ilar to the DHS, the data can b e used for m aking interstate com parisons in India and for com parisons w ith D H S data o f other developing countries. 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Sample design and implementation 5.2.1 Q uestionnaires H ousehold, individual and village questionnaires w ere used to gather data. These are reproduced in A ppendix F o f IIPS (1995). The questionnaires were based on the D H S M odel B Q uestionnaire, designed prim arily for countries with low contraceptive prevalence. M odifications w ere m ade to the questionnaires to suit conditions in India after detailed considerations. A few state-specific questions w ere also incorporated, on the recom m endation o f agencies involved in the state surveys. F or instance, in U ttar Pradesh a question w as asked about sex preference for children and in K erala a question w as asked about international m igration. T he questionnaires w ere pre-tested and m odified suitably before they w ere finalized. T he questionnaires were then translated into state languages and again pre-tested before finalization. T he household questionnaire collected inform ation on the household such as the type o f house, source o f w ater supply, type o f toilet facility and fuel used for lighting and cooking, ow nership o f land, livestock and durable goods, religion, caste, tribe and other characteristics o f all those w ho w ere present, including visitors, at the tim e o f the survey. T he w om an’s questionnaire collected inform ation on the respondent (respondents w ere ever-m arried w om en aged 13 to 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49) such as age, education, m arital status, age at m arriage, profession, reproduction, contraception, child health, partner’s characteristics, and the height and the weight o f children. The height/length and w eight o f children from birth up to four years o f age w ere m easured. In five states that w ere included in the first phase o f the survey, nam ely, A ndhra Pradesh, H im achal Pradesh, M adhya Pradesh, Tam il N adu and W est Bengal, height was not m easured, as instrum ents for m easuring heights w ere not ready at the tim e o f the survey. H ence in these states only w eight-for-age o f children are available, not height-for-age and w eight- for-height. H ow ever w eight-for-age in these states clusters closely around the national estim ate o f the children’s w eight-for-age (IIPS 1995). Those who m easured heights and weights w ere given training for taking m easurem ents accurately. For m easuring w eight standard spring balance w eighing m achines (Salter scales) w ere used. Children below the age o f tw o were m easured lying down, and older children w ere m easured standing. The height/length o f the children w ere m easured using adjustable boards o f adequate specifications suitable for such m easurem ent. W eight and height w ere m easured according to instructions contained in the U nited N ations m anual, “H ow to W eigh and M easure C hildren” (U nited Nations 1986). The village questionnaire w as used to collect inform ation about am enities like electricity, w ater supply, transportation, and educational and health facilities available in villages covered by NFHS. 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2.2 Sample design T he uniform sam ple design adopted for all the N FH S states w as a system atic and stratified sam ple o f households, w ith tw o stages in rural areas and three stages in urban areas. The sam ple size for each state w as decided considering aspects such as the size o f the state, tim e and resources available for the survey, need for separate estim ates for rural and urban areas, and for backw ard districts. A tw o-stage stratified sam pling design was adopted in rural areas. The 1981 Census o f India list o f villages served as the sam pling fram e in all the states except Assam , D elhi and Punjab. In these three states the 1991 C ensus data w ere used. First the villages in each district w ere divided into regions according to their geographic characteristics. Then each region was further stratified on the basis o f variables such as fem ale literacy, percentage o f scheduled caste/tribe population, distance to the nearest tow n, and proportion o f non-agricultural w orkers. The variables used w ere not all the sam e for every state. Fem ale literacy w as often used for stratification by listing villages in the order o f fem ale literacy rates. The prim ary sam pling units (PSU ) w ere selected system atically w ith probability proportional to size (PPS). O n an average, 30 households w ere selected for interview ing from each PSU . In PSU s w ith few er than 500 households, all the households w ere listed. In PSU s w ith m ore than 500 or m ore households, 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. segm entation w as done on the basis o f existing w ards in the PSU and two segm ents selected using either system atic sam pling or PPS. A ll the households in the selected segm ents w ere listed. T hen the households to be interview ed w ere selected from the household lists. The households, once selected, w ere never substituted by other households for any reason. In urban areas a three-stage sam ple design w as follow ed. C ities/tow ns were selected based on the 1991 Census, and then urban blocks and finally households w ere selected. Self-selecting cities (cities with population above a m inim um ), tow ns that w ere district headquarters and other tow ns w ere arranged in the sam e w ay that geographic stratification as w as done in rural areas. In large cities a two- stage sam ple design was used: urban blocks w ere selected first, and from them the households. In the case o f district headquarters and other tow ns a three-stage sam ple design w as used. T he tow ns w ere selected w ith PPS, tw o census blocks w ere selected per tow n and then 20 households each w ere selected from each block as w as done in rural areas. 5.2.3 D ata quality control In order to m aintain uniform ity in survey procedures across states, four detailed m anuals w ere prepared: an Interview er’s M anual, a m anual for Field Editors and Supervisors, a H ousehold Listing m anual, and Training G uidelines. 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. R epresentatives from population research Centers, consulting organizations, supervisors and surveyors w ere trained in all necessary aspects o f the survey. T he details o f training and how the survey w as carried out are described in detail in IIPS (1995). Each interview er was allow ed to conduct a m axim um o f three interview s a day so that a m inim um o f three callbacks w as m ade if no suitable person w as available for interview . The com pleted questionnaires w ere checked by field editors for com pleteness, consistency and other details, and corrections w ere m ade, if necessary, by revisiting the household. Field supervisors collected village inform ation using the village questionnaires and further checked on the accuracy o f data gathered by interview ers. One R esearch O fficer was assigned to each state by IIPS to ensure that correct procedures w ere follow ed and high quality o f data w as m aintained. Sim ultaneously data w ere entered into com puters and checked to identify problem areas and m ake im provem ents. Q uestionnaires com pleted and checked in the field w ere again checked centrally by the consulting organization in each state for consistency o f answ ers. C onsistency regarding age and birth history w as given special attention. The data w ere processed using the data entry and editing softw are called Integrated System for Survey A nalysis (ISSA). C om puter checks w ere done to clean the data and rem ove inconsistencies and im pute ages w here missing. 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2.4 The data The N FH S gathered inform ation from 88,562 households, tw o-thirds w hom w ere rural. T he response rate w as 96 percent in rural areas and 94 percent in urban areas. T he low est household response rate w as 88 percent in A runachal Pradesh. T he households had 500,492 residents. Interview s w ere com pleted with 89,777 ever-m arried w om en aged 13 to 49 w ho slept in these households the night before the interview . The w om en’s response rate w as 96 percent in both rural and urban areas. T he low est individual response rate w as 91 percent in A runachal Pradesh. The N FH S m easured living children who w ere 47 m onths or o f low er age at the tim e o f the survey. Section 4 o f the w om an’s questionnaire w as used to gather inform ation about the health o f children bom in the year o f the survey and in the four years prior to that on m atters including prenatal care, breastfeeding, vaccinations, illnesses and treatm ent o f children. Section 7 o f the questionnaire collected inform ation about the height and w eight o f children under age four. A bout 16 percent o f the N FH S sam ple o f children o f age 0 to 47 m onths w ere not m easured, m ainly because the children w ere not at hom e or the m other refused to allow the m easurem ents to be taken for cultural reasons. NFHS w as the first national survey to gather dem ographic, health and anthropom etric inform ation for the w hole o f India sim ultaneously. In all, there w ere 49,369 live births during the 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 m onths prior to the survey, for w hom tabulations w ere presented in IIPS (1995). T he data gathered on w eights and heights o f children w ere by the organizers o f the survey used to calculate three indices o f nutritional status, nam ely, w eight-for-age, height-for-age, and w eight-for-height. All the three indices have been expressed in standard deviation units (z-scores) from the m edian o f international reference population, recom m ended b y the W orld H ealth O rganization (D ibley et al. 1987a, 1987). 5.3 Variables used in this study T his study analyzed data pertaining to children w ho w ere 12 to 47 m onths old. T he variables used in the reduced m odels and som e o f the variables that show ed notable associations to the dependent variables are described below. In this chapter I also describe the position prevailing in India w ith respect to each variable, based on the tabulations contained in IIPS (1995). In a few cases where data are not available in IIPS (1995) I have presented data that I have derived from N FH S data. 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.1 Dependent variables The three trichotom ous dependent variables used in ordered logistic regressions are weight-for-age, height-for-age and w eight-for-height. The one-tim e m easurem ents concerning the grow th o f Indian children in relation to their age and height w ere com pared to the international reference population recom m ended by the W orld H ealth O rganization (D ibley et al. 1987a, 1987). In the case o f w eight- for-age, children w ere divided into three categories, nam ely, norm al weight, underw eight and severely underweight. In the case o f height-for-age the categories w ere norm al height, stunted and severely stunted. In the case o f w eight-for-height the categories w ere norm al weight-for-height, w asted and severely w asted. T he categories are m ade on the basis o f the deviations o f the m easurem ents in standard deviations from the international reference m edian. C hildren were classified as norm al if their weight-for-age, height-for-age or w eight-for-height w ere not below -2 standard deviations. C hildren w ere classified as underw eight, stunted or w asted if their weight-for-age, height-for-age or w eight-for-height, respectively, w ere below - 2 but not less than - 3 standard deviations. Children w ere classified as severely underw eight, severely stunted or severely w asted if their w eight-for-age, height-for-age or w eight-for-height, respectively, w ere below - 3 standard deviations. In the reference population, 2.3 percent o f the children 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. have a z-score below - 2 standard deviations. O nly 0.1 percent o f the children have z-scores below -3 standard deviations. Three cum ulative logit m odels w ere used to estim ate the probabilities o f underw eight, stunting and w asting for children 0 to 47 m onths old in the whole o f India with weight-for-age, height-for- age and w eight-for-height as dependent variables. Three m ore m odels w ith the sam e dependent variables w ere used for analyzing data for the tw o states o f U ttar Pradesh and Kerala. In selecting the variables that w ere significant in affecting the dependent variables, an iterative process w as follow ed. The BIC statistic (described in section 5.4.3) w as used to select the best m odels. The results o f the data analysis are presented in C hapter 6. Table 1.7 presents the levels o f undem utrition in India and states presented in IIPS (1995). Tables 1.1, 1.2 and 1.3 give an international com parison. T hese tables present the high levels o f undem utrition in India and its states. There were large differences in the levels o f undem utrition am ong Indian states. F or instance, the percentages underw eight w ere 29.5 percent in K erala and 62 percent in Bihar. Such interstate differences prevailed w ith respect to stunting and w asting as well. The percentages undernourished differed greatly by background characteristics (Table 5.1). The percentage o f children less than six m onths old w ho w ere undernourished for w eight (below -2 S.D) was about four tim es low er than children m ore than one year old. T his pattern w as even m ore 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. pronounced in the case o f stunting. W hile percentages underw eight peaked in the age group o f 12 to 23 m onths and then fell slightly, the rate o f stunting kept rising with age, show ing that stunting becam e w orse w ith age. W asting peaked sharply at 12 to 23 m onths and then dropped as fast as it rose during 24 to 35 m onths and less sharply during 36 to 47 m onths. The pattern o f sudden w orsening o f the nutritional status during w eaning has been discussed earlier. There w as no gender difference in percentages o f the undernourished. H igher birth orders had greater percentages o f undernourished. There w ere substantial differences in undem utrition by m others’ education. The percent underw eight and severely underw eight am ong children o f m others with no education was tw ice as high as o f m others who had a high school or higher education. The situation w as the sam e in the case o f stunting also. In the case o f w asting the differences w ere com paratively lesser in m agnitude, but still substantial. The low er rates o f w asting appear to be the result o f the very high rates o f stunting at all ages beyond six m onths. B oth M uslim s and H indus w ho constituted 94 percentage o f the total N FH S sam ple, had sim ilar levels o f undem utrition. The rates o f undem utrition am ong children belonging to scheduled castes and scheduled tribes w ere higher than those o f other children. 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.2 Independent variables Independent variables w ere divided into proxim ate determ inants and underlying m aternal and household determ inants. The proxim ate determ inants in the reduced m odels that w ere found to be significant in affecting child nutritional status w ere age o f the children, prenatal visits, place o f delivery, size at birth, and suffering from fever in the two w eeks p rior to the survey. The significant underlying m aternal and household determ inants in the reduced m odels w ere the schooling o f the m other, schooling o f the father, age o f the m other at first birth, the occupational status o f the father and the am ount o f agricultural land ow ned by the household. Interactions betw een prenatal visits and father’s education, and prenatal visits and m other’s education w ere also found to be significant. In the state regressions (using data from U ttar Pradesh and K erala) the state o f residence (K erala) residence in K erala was used as a variable in order to bring out the effects o f the underlying and basic causes found in Figure 1.1. Som e o f the other variables that were found to be significant in the analysis, but w ere not in the reduced m odels, w ere socio-econom ic status and m easles vaccination. The independent variables are described below in order to dem onstrate the prevailing situation regarding each variable in India. The descriptions are based on tabulations in IIPS (1995) and are for the N FH S sam ple and not restricted to the observations included in the regressions in this study. 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Proximate determinants 5.3.2.1 C hildren’s age Children w ho had com pleted one, tw o and three years o f age w ere included in the analysis. A ge w as used both as a continuous variable in the height-for-age and w eight-for-height country m odels. In the w eight-for-age country model age three (children w ho had com pleted age three) was used as a dum m y variable, as it w as yielding a better m odel based on a low er BIC statistic. The low er the BIC statistic, the better the m odel is considered to be capturing the m ain features o f the data. The BIC statistic is described in section 5.4.3. In the N FH S sam ple households, there w ere 25,021 boys and 23,497 girls betw een one and four years old w ho w ere located at the tim e o f the survey. They m ade up 9.8 percent o f the total household population. A s given in T able 5.1 underw eight and w asting peaked in the 12-23 m onth period, w hereas the rate o f stunting continued to rise w ith age. This is in line w ith the patterns observed in Figures 3.5 and 3.6. A ge appears to be an im portant factor that should be considered w hile deciding on interventions aim ed at preventing undem utrition. 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.2.2 Prenatal visits This variable w as used as a continuous variable that indicates the num ber o f prenatal visits the m others had at their hom es or outside. V isits beyond tw elve w ere considered as tw elve based on the recom m endation that prenatal visits should be m ade once every four w eeks during the first seven m onths o f pregnancy, then once every tw o w eeks until the last m onth and every w eek thereafter (M cD onald and Pritchard 1980). I f tw elve visits w ere not possible for socio-econom ic or other reasons, a m inim um o f four visits w ere recom m ended during the third, sixth, eighth and ninth m onths o f pregnancy (Park and Park 1989). In 1992, the M inistry o f H ealth and Fam ily W elfare o f the G overnm ent o f India set a m ore m odest target o f a m inim um o f at least three prenatal visits for all pregnant w om en (M O H FW 1992b). The m edian num ber o f prenatal visits in the N FH S sam ple w as 4.6 in urban areas and 3.5 in rural areas. H ow ever, eighteen percent o f the respondents in urban areas and forty-tw o percent o f the m others in rural areas did not have any prenatal visit either at hom e or elsew here (total 37 percent). Fifty-one percent o f all m others had one to three visits. Tw enty-seven percent o f them had four or m ore visits. T he m edian m onths o f pregnancy at the tim e o f the first visit w ere 5.3. A llopathic doctors (doctors practicing the W estern system o f m edicine) provided prenatal care to 40 percent o f the m others; nurses, ayurvedic doctors 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (doctors practicing the traditional Indian system o f m edicine) and other care givers provided care to 9 percent. The care provided by allopaths m ay be considered superior to the care given by other doctors. N on-allopaths in India, usually do not provide surgical support and institutional facilities for delivery. In addition, thirteen percent o f the m others received prenatal care at hom e. U rban w om en (71 percent) w ere m ore likely than rural w om en (31 percent) to have received prenatal care from an allopath. G reater percentages o f younger w om en received prenatal care than older w om en: 64 percent o f the w om en below 20 years and 42 percent o f w om en above 34 years received prenatal care. W hile 52 percent o f first order births received prenatal care from allopaths, only 20 percent o f 6th or higher order births received prenatal care from allopaths. Prenatal care increased w ith the education o f the m other. W hile only 50 percent o f the m others w ith no schooling had prenatal care, 90 percent o f the m others w ho com pleted m iddle school had prenatal care. A m ong the m others w ho did not have any prenatal care, 59 percent cited 'not necessary' as the reason for not having prenatal care, 7 percent cited 'financial cost', 6 percent cited 'not custom ary' as the reason and 5 percent said they w ere 'not allow ed to go'. O ne o f the im portant benefits o f prenatal care to m others in India is that it enables them to get tetanus toxoid vaccinations. Tetanus is m ajor cause o f death am ong new born children in India (IIPS 1995). N eonatal tetanus is com m on when 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. deliveries take place in unhygienic conditions and unsterilized instrum ents are used to cut um bilical chords. Tetanus typically develops in the first or the second w eek o f life and is fatal in 70 to 90 percent o f the cases (Foster 1984). Expert medical help is often not available in rural areas w here m ost tetanus infections occur. Tw o doses o f tetanus toxoid vaccines given in the early m onths o f pregnancy can effectively prevent tetanus am ong new borns (IIPS 1995). Seventy four percent o f the m others in urban areas received tetanus toxoid vaccinations as against 48 percent in rural areas. W hile 84 percent o f the m others w ho had com pleted m iddle school received tw o doses o f tetanus toxoid vaccinations, only forty percent o f m others w ho had no schooling received them . Prenatal visits also enable m others to obtain iron and folic acid tablets. The iron and folate status o f the m other during pregnancy affects the health o f the m other and child, birth w eight, and the m other’s ability to adequately breastfeed the child (W H O 1994). The M inistry o f H ealth and Fam ily W elfare o f the G overnm ent o f India had recom m ended that a pregnant w om an consum e a hundred iron and folic acid tablets. H ow ever only 51 percent o f the m others obtained iron and folic acid tablets. 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53.23 Place o f birth Two dum m ies w ere created for delivery at hom e and delivery at a governm ent m edical facility to capture the effect o f the place o f delivery. D elivery at private m edical facilities w as the om itted group. H om es included hom es o f the m others, m others’ paren t’s hom es, other hom es and locations other than medical facilities. Governm ent m edical facilities included governm ent hospitals and governm ent health posts. Private m edical facilities included private hospitals and clinics. The W orld Health O rganization (1994) considered safe and clean delivery as essential requirem ents for child survival and m other’s health. For this, it is essential that delivery be conducted under proper hygienic conditions w ith the assistance o f a trained person. Lack o f trained assistance can cause maternal deaths and other com plications during delivery. A bout 74 percent o f the births in the N FH S sam ple took place at the hom es o f the m others (62 percent) or o f their parents (12 percent). A bout fifteen percent o f the deliveries took place in governm ent m edical facilities. The position w as better in urban areas w ith 58 percent o f the births taking place in m edical institutions (private or public) as against only sixteen percent in medical institutions in rural areas. Thirteen percent o f the m others w ho w ere m ore than 35 years old had institutional deliveries, w hile 24 percent o f m others less than tw enty years old had 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. institutional deliveries. H igher order births had low er percentages o f institutional deliveries. W hile tw elve percent o f the w om en without any schooling had institutional deliveries, 75 percent o f the w om en w ith at least a high school education had institutional deliveries. Institutional deliveries w ere 9 percent for Scheduled tribe wom en, sixteen percent for scheduled caste w om en, 29 percent for H indus and 24 for M uslim s. W hile 62 percent o f the m others had prenatal care only 26 percent had institutional deliveries. Those w ho had prenatal care w ere m ore likely to have institutional deliveries. Trained birth attendants did not attend tw o-thirds o f all deliveries. In rural areas five out o f six births took place at hom es and three- quarters o f all births had no m edical assistance. 5.3.2.4 Size at birth Sizes o f children at birth w ere classified as “sm all”, “average” or “large” in the N FH S. D um m ies w ere created for the first two and “large” w as the om itted group in this study. Size at birth w as recorded during N FH S based on the subjective report o f the respondent. T his w as because seventy nine percent (50 percent in urban and 88 percent in rural areas) o f the children surveyed w ere not w eighed at birth. This is not surprising considering the fact that 74 percent o f the births took place at hom e. Even when children w ere w eighed at birth, the records 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w ere not available and m any m others did not rem em ber the exact w eights o f the children. H ow ever m others’ m em ory o f the size o f children at birth appeared good. Previous experience show ed that m others give useful inform ation about the size o f their new borns (IIPS 1995). Slightly m ore than one-fifth o f the children w ere reported to be sm all by their m others. Tw enty six percent o f the babies w hose w eights w ere available w ere underw eight (less than 2500 gm ). A n estim ate based on past studies had placed the prevalence o f low birth w eight in India at 30 percent (W H O -U N IC E F 1992). 5.3.2.5 Prevalence o f fever A dum m y w as created to capture the effect o f having fever during the tw o w eeks prior to the survey. C hildren w ho did not have fever w ere the reference group. O ther infections w ere not found to be statistically significant in m ultivariate regression. N FH S asked m others about three conditions prevailing during the tw o w eeks prior to the survey: fever, cough accom panied by fast breathing, and diarrhea. In the regression analyses only fever was found to be significant. Fever w as the m ost com m on ailm ent am ong children less than four years old. T w enty percent o f the children had fever during the tw o w eeks p rio r to the survey, 6.5 percent had a cough accom panied by fast breathing and 10 percent had diarrhea. 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Children w ho w ere 6 to 23 m onths had the highest prevalence rates o f fever, and children less than six m onths and m ore than three years or older had the lowest prevalence rates. This tim ing coincides with the age patterns in grow th faltering observed in Figures 3.5 and 3.6. There were no m ajor differences betw een m others o f different levels o f education, religion or social class in the prevalence rates o f fever am ong children. There w ere considerable differences am ong states in the prevalence rates o f fever, ranging from 36 percent in T ripura to 11 percent in Rajasthan. In all the states except Jam m u and K ashm ir, fever w as the m ost com m on o f the three conditions surveyed. 5.3.2.6 M easles Im m unization If a child had a m easles vaccination according to the im m unization card or as reported by the m other, the child w as considered to have received a m easles vaccination. A dum m y w as created for those who had a m easles vaccination. Im m unizations for other diseases w ere not found to be statistically significant, individually or w hen taken together. A m ong im m unizable diseases, m easles and pertussis w ere found to cause the greatest num ber o f deaths, w ith m easles alone accounting for m ore than 880,000 deaths annually (U nicef 1993). M ata (1978) found that m easles resulted in w eight loss, w ith over one-fourth o f the children studied losing over 5 percent 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o f body w eight. M easles w as associated w ith increased diarrhea, acute respiratory infection, prolonged anorexia and reduced food intake am ong young children and has been correlated w ith increased rates o f m alnutrition and poor grow th (H uffm an and Steel 1995). Hence im m unization against m easles can help prevent grow th faltering. Im m unization o f children against six preventable diseases (tuberculosis, diphtheria, pertussis, poliom yelitis, tetanus and m easles) has been an im portant part o f the child health care system in India. In order to m ore fully im m unize Indian children the G overnm ent o f India started the U niversal Im m unization Program (U IP) in 1986-87. The U IP aim ed at vaccinating at least 85 percent o f the children against the six preventable diseases and achieving self-sufficiency in the production o f vaccines and refrigerated equipm ent required for transporting the vaccines by 1990 (M O H FW 1991). International guidelines prescribe that children should be fully im m unized by age one (IIPS 1995). H ence only children m ore than one year old w ere selected for analysis in this study. T he N FH S found that only 35 percent o f the children 12 to 23 m onths old w ere fully im m unized. O f the 35 percent o f the children w ho were fully im m unized, 25 percent w ere fully im m unized by 12 m onths o f age. A nother 35 percent received som e vaccinations and 30 percent received none. O nly 42 percent o f the children w ere vaccinated against m easles. M ale and female children w ere alm ost equally 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. im m unized for m easles. Thirty-one percent o f the children o f illiterate m others and 77 percent o f the children o f m others w ho had a high school or higher education w ere im m unized for m easles. U nderlying m aternal and household determ inants 5.3.2.7 Schooling o f parents T he num ber o f years o f schooling o f the m others and th e fathers had w ere used as continuous variables. T hese variables w ere constructed by taking into account the levels o f schooling com pleted (prim ary, secondary and higher) and the num ber o f years at the highest level. The m inim um num ber o f years o f schooling w as zero and the m axim um w as 20 years for m others and 22 years for fathers. T he num ber o f years o f schooling o f the m others and the fathers are im portant variables that affect m any factors that determ ine child health, such as, utilization o f available health inform ation, w om en’s autonom y, higher expectations for children, fam ily size, hygiene, feeding practices and treatm ent o f diseases. T he table below gives the educational attainm ents o f the N FH S sam ple o f 89777 ever-m arried w om en aged 13 to 49 and their husbands. W om en w ere the respondents in the survey and the nutritional status o f th eir children w ho w ere 12 to 47 m onths old are studied in this dissertation. S ixty-three percent o f the 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w om en surveyed in the N FHS had no schooling at all. O f the rem aining 37 percent, 19 had not reached m iddle school and 18 percent had com pleted m iddle school or gone higher. T heir husbands w ere better educated, and only 35 percent w ere w ithout any schooling. O f the 65 percent w ho w ere literate, 39 percent had com pleted m iddle school or gone higher. There w ere m ajor interstate differences in the level o f education o f m others, sim ilar to w hat was seen in Table 1.5, which contains data from the 1991 Census. W om en in M izoram (92 percent) and K erala (84 percent) had the highest literacy rates. The highest percentages o f w om en w ithout any schooling w ere in Table 5.2 Levels schooling of women and their husbands, NFHS 1992-93 Educational Level W omen ('Percentage') H usbands ( ’ Percentage') N o education Prim ary incom plete Prim ary com plete M iddle school com plete High school com plete H igher than high school Source: A dapted from IIPS 1995. 63 7 12 7 8 3.4 35 10 18 12 16 9 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. R ajasthan (82 percent), B ihar (78 percent), U ttar Pradesh (76 percent) and M adhya Pradesh (74 percent). H ow ever in these four states the percentage o f the husbands who w ere literate was tw ice as high as the wom en. The percentage o f wom en w ho w ere literate in urban areas (72 percent) was tw ice as high as the percentage o f w om en w ho were literate in rural areas. Only 11 percent or rural wom en had com pleted m iddle school in rural areas com pared to 41 percent in urban areas. H igher percentages o f respondents who were aged 13 to 14 (61 percent) and 15 to 19 (44 percent) had no schooling at all, than in the higher age groups (20 to 44 years). T his show ed the tendency o f less educated w om en to m arry earlier. Eighty-one percent o f scheduled caste and 84 percent o f scheduled tribe w om en had no schooling as against 58 percent for other wom en. 5.3.2.8 Som e background characteristics o f N FH S respondents Table 5.3 gives som e o f the other background characteristics o f respondents in N FH S w ho w ere ever-m arried w om en 13 to 49 years old. H indus and M uslim s together m ade up 94 percent o f the sam ple, sim ilar to their share in the population according to the 1991 census. Scheduled castes and scheduled tribes m ade up 21 percent o f the sam ple. N inety four percent o f ever-m arried w om en rem ained currently m arried. M ost o f the w om en did not w ork outside the 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.3 Background characteristics of NFHS respondents, India 1992-93 R espondents Percentage Rural residents 74.0 Hindu 82.0 M uslim 12.0 Christian 2.2 Sikh 1.9 Scheduled caste 12.2 Scheduled tribe 8.8 C urrently M arried 94.3 W idowed 3.8 D ivorced/Separated 1.9 N ot w orking 68.5 W orking in fam ily farm /business 12.2 E m ployed by som eone else 16.2 Self-em ployed 3.1 Regularly exposed to the m edia 53.0 Source: IIPS 1995. hom e. O verall, forty seven percent o f N FH S respondents w ere not regularly exposed to any kind o f m ass m edia. O nly 39 percent ow ned radio sets and 21 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. percent ow ned T V sets. These, considered along w ith the percentage o f respondents w ho are not educated (63 percent), have im plications for im parting inform ation through the m ass m edia. The exposure w as m uch higher in urban areas w here 81 percent o f the respondents w ere exposed to the m edia com pared to only 43 percent in rural areas. W om en w ith no schooling w ere the least exposed to any m edia; only 36 percent, as com pared to 94 percent for those w ith at least a high school education. 5.3.2.9 A ge o f the m other at first birth M others w ho w ere 11 to 19 years old at first birth w ere classified as low- age m others and those older as high-age m others. A dum m y was created to capture the effect o f low -age on nutritional status, w ith m others w ho w ere older than 19 years being the reference group. T he N FH S found that about 50 percent o f the w om en w ho w ere 20 to 49 years old had their first births w hen they w ere 15 to 19 years old; and 6 percent o f them had their first births at less than 15 years o f age. H ow ever, the percent o f w om en w ho had their first births before age 20 had substantially declined; in 1992-93, only 19 percent o f w om en aged 15 to 19 years had had th eir first births. N inety percent o f the w om en in urban areas and 78 percent in rural areas had not yet had any birth at the tim e o f the N FH S. T here w ere m ajor differences betw een 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w om en w ith different levels o f education in the m ean ages at first birth. W om en w ho had no schooling had their first births four to five years earlier than w om en w ho had at least a high school education. The m edian ages at first birth varied considerably am ong Indian states. F or w om en 25 to 49 years old, the highest m edian age w as in 23.7 years in Goa. The lowest m edian age w as 17.9 years in A ndhra Pradesh. A m ong teenage w om en in the households surveyed, only 17 percent had children or w ere pregnant. A m ong ever-m arried teenage respondents, 58 percent had begun childbearing. The onset o f childbearing is an im portant dem ographic indicator. R ise in age at first birth is associated w ith a rise in age at first m arriage and is usually accom panied by fertility decline. Births to teenage wom en have been found to be detrim ental to the health o f the m other and the child. 5.3.2.10 A gricultural land ow ned by households This variable w as constructed by adding together the irrigated and non irrigated agricultural land owned by each household living in urban and rural areas and creating a scale as follows: H ouseholds with no agricultural land=0 H ouseholds with >0 to 5 acres o f agricultural land =1 H ouseholds w ith 6 to 10 acres o f agricultural land =2 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. H ouseholds w ith m ore than 10 acres o f agricultural land =3. In the N FH S sam ple 48 percent o f the households did not ow n any land. The follow ing table gives the percentages o f households that ow ned land (adapted form UPS 1995). N ineteen percent o f the households in the N FH S sam ple ow ned irrigated land, 22 percent ow ned unirrigated land and 11 percent ow ned both irrigated and unirrigated land. Thirteen percent o f the households ow ned less than an acre o f land, 28 percent betw een one and five acres and 11 percent m ore than five acres. Land ow nership is a source o f incom e and security for rural households. L and is used to raise loans for social, econom ic, m edical and other purposes and em ergencies. L and ownership can indicate the econom ic and social status o f the household in rural areas w here land is the m ain asset. Table 5.4 Percentages of households owning agricultural land: NFHS 1992-93 A cres O n ly O n ly Both irrigated and Total irrigated unirrigated unirrigated <1 6 5 2 13 1 to 5 10 13 5 28 > 6 3 4 4 11 T otal 19 22 11 52 Source: A dapted from IIPS 1995. 183 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.2.11 Socio-economic status ('SES1 The (SES) variable was created by constructing an index o f the possession o f the follow ing six item s: any type o f transportation (car, bicycle, m otorcycle or tractor), radio, flush toilet, som e toilet facility, pucca house and sem i-pucca house. In India housing quality is described as kachha, pucca or sem i-pucca. K achha houses are m ade from m ud, thatch or other low quality m aterials. Pucca houses are m ade entirely from high-quality m aterials, including roofs, floor and walls. Sem i-pucca houses are m ade partly from low -quality and partly from high-quality m aterials. The index w as created as follows: High S E S : Possession o f four or m ore item s = 3 M edium S E S : Possession o f 2 or 3 item s = 2 Low S E S : Possession o f less than 2 item s = 1. Sim ilar indices have been created by researchers to evaluate SES (Knodel and W ongsith 1991; Speizer 1995; see M ontgom ery et. al 1999 for a review o f indices created). H ow ever, a tractor is not usually included as a m ode o f transport. T ractor w as included as a m ode o f transportation in this study because in m any parts o f India tractors are used by farm ers to travel to the m arket, on social occasions, to carry the sick to hospitals, w om en for delivery and so on. 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The table below gives the percentages o f households in the N F H S sam ple that possessed the item s taken into account to create the SES index. Table 5.5 Percentages of households having items measuring SES Item U rban Rural T otal B icycle 48 38 42 M otorcycle/scooter 19 4 8 T ractor 2.10* Car 3.2 0.3 1.1 Flush toilet 60 7 22 Som e toilet facility 16 6 9 N o toilet facility 24 87 70 Pucca house 57 11 24 Sem i-pucca house 26 28 28 K achha house 17 60 49 * Percentage ow nership in the households with children 0 to 47m onths old Source: A dapted from IIPS 1995. M eans o f transport: The possession o f som e form o f transportation allow s faster travel and easier access to m arkets, shops, doctors, hospitals and other facilities. Forty eight percent o f the households in urban areas and 38 percent in rural areas (total 42 percent) ow ned a bicycle. N ineteen percent o f the households in urban areas and four percent in rural areas (total 8 percent) ow ned a 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m otorcycle or scooter. Three percent o f the households in urban areas and 0.3 percent in rural areas (total 1.1 percent) ow ned a car. R adio: The possession o f a radio m akes it possible to get inform ation on a variety o f subjects including health and fam ily planning. Fifty-nine percent o f the households in urban areas and 32 percent in rural areas (total 39 percent) ow ned radio sets. H ouseholds that have T V sets are very likely to have radio sets. Hence T V sets have not been included in calculating SES to avoid duplication. In all 21 percent o f the households in India ow ned TV sets. T oilet facility: H ouseholds w ere considered to have flush toilet facility if they had their ow n flush toilets or shared flush toilets w ith others. H ouseholds w ere considered to have som e toilet facility if they had access to public flush toilets, traditional or im proved pit latrines or public pits/latrines. I f the households had none o f the facilities m entioned above, they w ere considered to have no toilet facility. Sixty percent o f the households in urban areas and seven percent in rural areas had access to flush toilets (total 22 percent). In addition, sixteen percent o f the households in urban areas and six percent in rural areas had som e toilet facility (total nine percent). T w enty four percent o f the households in urban areas and 87 percent in rural areas had no type o f toilet facility (total seventy percent). T oilet facilities w ere inadequate in all the states o f India. There w ere m ajor interstate differences in toilet facilities available to the households. 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M ore than seventy percentage o f the households in Delhi, states o f the N orth-east (except A ssam and M eghalaya) and K erala had som e toilet facility. In nine states less than 25 percent o f the households had any type o f toilet facility. T oilet facilities are very im portant for the health and hygiene. They have a direct potential to reduce exposure to pathogens, m orbidity and poor nutritional status o f children and other m em bers o f the household. Absence o f toilet facilities m ay result in contam ination o f drinking w ater and the spread o f intestinal parasites. Provision o f clean drinking w ater m ay not im prove health and nutritional status unless proper arrangem ents for disposal o f feces and hands are w ashed w ith soap to prevent transfer o f contam ination through . Q uality o f house construction: The quality o f house construction w as found to be associated w ith the health and grow th o f children in rural H onduras (M artorell et al. 1983). M any studies have used housing in m easuring living standards (M ontgom ery et al. 1999). In the N FH S sam ple forty nine percent o f the houses w ere kachha, 28 percent w ere sem i-pucca and 24 percent w ere pucca. The quality o f housing in urban areas w as better, w ith 57 percent pucca houses. Crow ded conditions affect health by increasing the spread o f infections, especially respiratory infections. T he num ber o f persons to a room is a sim ple m easure o f crowding. On an average there w ere 2.8 persons to a room in India. Sixty percent o f households had less than three persons to a room , sixteen percent had five or 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m ore persons and five percent o f the households had seven or m ore persons to a room . G ujarat, M aharashtra, U ttar Pradesh and R ajasthan had m ore than three persons on average to a room . K erala had the low est num ber o f persons (1.4) to a room . 5.3.2.12 O ccupational status o f parents T he occupations o f fathers w ere divided into tw o groups. O ne group was in relatively better occupations that included high-level professional technical work, low-level professional technical w ork, adm inistrative, executive or m anagerial w ork, and clerical w ork. A ll other types o f w ork, starting from sales w orkers dow n, m ade up the second group. A dum m y was related to capture the effect o f the father having an occupation o f low er status. T he hig h er status occupations w ere the reference group. M others w ere also sim ilarly classified, but m o th ers’ occupation w as not statistically significant. Sixty-eight percent o f the m others in the N FH S sam ple w ere not working, and tw elve percent w orked in the fam ily farm or business. In the sam ple children 1 to 4 years old, approxim ately 91 percent o f the fathers w ere in occupations o f low er status. T he occupation o f the father w as therefore im portant for household incom e. This affects the diet and m edical care that can be afforded by the fam ily and the nutritional status o f children. 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.4 The statistical model 5.4.1 O rdered logistic regression O rdered logistic regression is used in this study to analyze the data. Logistic regression has in recent years becom e the analytical technique o f choice for the m ultivariate m odeling o f categorical dependent variables (D em aris 1995). Independent variables m ay be categorical or continuous in logistic regression. The m easurem ent scale o f each dependent variable consists o f a set o f categories. Such variables com m only occur in social and biom edical sciences, public health, econom ics and other behavioral sciences (Agresti 1990). Categorical variables that have ordered levels are called ordinal categorical variables. For instance, upper, m iddle and low er social class. A interval variable is one that has num erical distances betw een any tw o levels o f the scale. Exam ples w ould be incom e and age. N om inal categorical variables do not have a natural ordering. Religion and race are exam ples o f nom inal categorical variables. A variable can be m easured as nom inal, ordinal or as an interval variable (Agresti 1990). For instance education is nom inal w hen the type o f schooling is considered (public or private school), ordinal when levels o f schooling are considered (prim ary, m iddle and secondary) and interval w hen m easured by num ber o f years o f schooling, as is done in this study. V ariables can be classified as continuous or discrete according to the 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. num ber o f values they take. A ctual m easurem ents o f all variables occur in a discrete m anner. In practice, those variables that can take a lot o f values are considered continuous and those that can take relatively few values are considered discrete (Agresti 1990). N om inal variables are qualitative, whereas interval variables are quantitative. Ordinal variables fall in between. Each level o f an ordinal variable signifies a certain m agnitude o f the characteristic. H ence ordinal variables often com e close to being continuous. The categorization will need to be done taking into consideration practical know ledge and experience available from research in the field. L ogit analysis provides a global test for the significance o f a given predictor controlling for all other predictors in the m odel as w ell as a test for the significance o f a set o f predictors controlling for other effects. The im pact o f a given predictor on the dependent variable, adjusted for other effects in the m odel, is m anifested in the odds ratios (D em aris 1992). The odds indicate the relative probability o f falling into one o f the categories o f the dependent variable. H ow ever in m ultivariate analysis, conditional odds are o f greater interest. For instance the odds o f a child becom ing underw eight given certain characteristics such as SES and m other’s education. Log odds have a theoretical range o f m inus infinity to plus infinity. 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The m odel used in this analysis is the cum ulative or ordinal logit m odel developed by M cC ullagh (1980). The SAS package w as used to fit the cum ulative logit m odel. W e obtain tw o intercepts in the case o f a trichotom ous dependent variable. B ut intercepts are not o f substantive interest (A llison 1999). H ow ever, it m ay be m entioned that the first intercept gives the predicted log odds o f being in category one rather than in categories tw o or three when all the explanatory variables have a value o f zero. The second intercept predicts the log odds o f being in categories one or tw o rather than in category three when all the explanatory variables have a value o f zero. In this study, categories one, tw o and three consist o f children w ho are severely undernourished, undernourished and having norm al grow th respectively. The cum ulative logit m odel calculates the coefficients for each category o f the dependent variable and then estim ates a w eighted average o f the corresponding coefficients. Hence, irrespective o f the num ber o f categories in the dependent variable, there is only one coefficient. The “Score T est for the Proportional O dds A ssum ption” in the SAS output indicates w hether the ordinal restrictions are valid, and high p -values are desirable (Allison 1999). The score test is a test o f the null hypothesis th at the corresponding coefficients calculated for the categories m atch. T he SA S/STA T U ser’s Guide (SAS Institute Inc. 1989) w arns that the test m ay tend to reject the null hypothesis m ore often than is w arranted. A llison (1999) reports that w hen there are m any independent 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. variables and w hen the sam ple size is large, the score test for proportional odds assum ption will usually pro d u cep -v alu es below 0.05. A llison (1999) considered that this was not a sufficient reason to eject the m odels. H ow ever, to m ake sure that coefficients are not variant across dichotom izations, separate m odels can be estim ated for different dichotom izations. 5.4.2 The m odel In this m odel, pij is the probability that an individual falls into category j o f the dependent variable. T he categories are ordered in the sequence j= 1 to 3. T he cum ulative probabilities are given by ^• = S av„ m=I where j-1,2 Fij gives the probability that an individual falls into a category or a low er one. Each Fij corresponds to a different dichotom ization o f the dependent variable. The m odel for J-l equations is given by log ' J L ' — C(j + + @2^2 + +.... + w here /3,,/?>... are regression coefficients a n dXt,X2... are the independent variables. 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A lthough there is a single set o f coefficients there is a different intercept for each o f the equations. In this study, the explanatory variables predict the probability o f being in a low er rather than in a higher category. H ence the odds ratios can be interpreted as the effect o f the variable on the odds o f being in the low er categories, nam ely severely undernourished or undernourished, than in the higher category, nam ely, having norm al growth. The m axim um likelihood method is used to estim ate the m odel. The probabilities o f being norm al (growth in weight, height or w eight-for- height), undernourished (underw eight, stunted or w asted), and severely undernourished (severely underw eight, severely stunted or severely w asted) are estim ated by using the follow ing procedure (SAS Institute 1995). P(Y, = 2) = 1/[1 + e x p ( -a , - P(Yt = I) = 1/[1 + e x p ( - a 2 - X P A ) ] - 1/[1 + e x p ( -a , - £ & * « ) ] P(Y, = 0) = 1 - {1/[1 + e x p ( - a 2 - T hese three probabilities sum up to one. 7=2 gives the probability o f being norm al, 7=1 gives the probability o f being underw eight, stunted o r wasted and 7=0 gives the probability o f being severely underw eight, severely stunted or severely w asted. 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.4.3 The BIC statistic Tw o logistic regression m odels can be com pared by taking the difference o f their BIC values, w ith the m odel having the sm aller BIC value being the preferred one (Raftery 1995). T he BIC for logistic regression is given by the follow ing formula: BIC=Z2 + p In N W here, X~ is the likelihood ratio test statistic (-2 log I in SAS) for com paring the null m odel w ith no covariates w ith the m odel o f interest, p is the num ber o f independent variables in the m odel o f interest (not counting the intercepts), and N is the sam ple size. T he low er the BIC, the better the model captures the m ain features o f the data relative to other models. BIC can be used as a guide for an iterative m odel selection process (R aftery 1995). T he iterative process w as follow ed in this study to select the best m odel. T he BIC can give accurate estim ates even in sam ples as small as 40 observations (R aftery 1995). 5.5 Summing up T he N FH S o f India 1992-93 was the first fam ily health survey that gathered dem ographic, maternal and child health inform ation for the w hole o f India. It w as for the first tim e that a sam ple o f children from all over India w as system atically m easured. T his study sought to use N FH S data fo r estim ating the 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extent and severity o f undem utrition and the im m ediate and underlying factors associated undem utrition am ong children 12 to 47 m onths old. In addition, basic factors such as the social, econom ic, cultural and political factors determ ine the environm ent in w hich children grow (Figure 1.1). T herefore an assessm ent of such factors associated w ith the undem utrition o f children in India w as attem pted, based on published findings o f other researchers and the research er’s own fam iliarity w ith conditions in India (C hapter 4). In the study, data for children 12 m onths to 47 m onths w ere analyzed. The m eans o f the m ajor variables analyzed are given in Tables 5.6, 5.7 and 5.8. The num ber o f observations (N) in Table 5.6 (w eight-for-age m odel) is 25,550. The N in T able 5.7 (height-for-age and weight- for-height m odels) is 19,368. The reason for this difference w as that height was not m easured in five states o f India. The N in Table 5.8 pertaining to data for U ttar Pradesh and K erala w as 4957. In both these states w eights and heights were m easured. T he m eans o f variables for K erala and U ttar Pradesh separately are given in T ables 5.9 and 5.10. T h e N o f K erala w as 978 and the N for U ttar Pradesh w as 3979. R egression results are discussed in the next chapter. 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 5.1 PERCENTAGES OF CHILDREN UNDERNOURISHED BY DEMOGRAPHIC CHARACTERISTICS OF MOTHERS AND CHILDREN IN INDIA, 1992-93 Demographic Characteristic Und Belov/ -3 S.I). erweight (Percentage! Child's age <6 months 2.8 6-11 months 14.1 12-23 months 26.3 24-35 months 25.9 36-47 months 21.8 Sex Male 20.2 Female 21 Mother's education Illiterate 24.7 Lit.,<m iddle 16.7 complete Middle school 12.4 complete High school and 7.8 above >-3 S.D Below No. o f <-2 S.D. -2 S.D. children 12.8 15.6 4406 29.2 43.3 4792 37.1 63.4 9560 36.3 62.2 8406 36.7 58.5 8643 33.1 53.3 18208 32.4 53.4 17599 34.5 59.2 22946 33.7 50.4 6251 31.1 43.5 2765 22.5 30.3 3844 Stunted (Percentage! Below >-3 S.D Below -3 S.D. <-2 S.D. -2 S.D. 5.7 10 15.7 14.3 20 34.3 30.7 25.9 56.6 34.6 25.6 60.2 40.7 26 66.7 28.4 23.9 52.3 29.4 22.3 51.7 34.5 24 58.5 22.6 23.8 46.4 17.9 21.4 39.3 12.2 17.8 30 Wasted (Percentage1 ) Below >-3 S.D Below -3 S.D. <-2 S.D. -2 S.D. 2 7.5 9.5 2.9 12.8 15.7 5.6 22.4 28 2.5 14.1 16.6 1.8 9.8 11.6 3.7 15.1 18.8 2.6 13.5 16.1 3.4 15.4 18.8 3 13.8 16.8 2.7 12 14.7 2.3 10 12.3 N o. o f children 3225 3176 6945 6033 6204 13040 12543 16639 4260 1905 2780 V O O V Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 5.1 continued Demographic Underweight (Percentage! Stunted ('Percentage') Wasted (Percentage! Characteristic Below >-3 S.D Below N o. o f Below >-3 S.D Below Below >-3 S.D Below N o. o f Previous birth intervel -3 S.D. <-2 S.D. -2 S.D. children -3 S.D. <-2 S.D. -2 S.D. -3 S.D. <-2 S.D. -2 S.D. children First birth 17.5 32 49.5 9762 24.8 23.3 48.1 3 13.5 16.5 6664 <24 months 23.3 33.6 56.9 6106 33.1 23.8 56.9 3.7 12.6 16.3 4549 24 to 47 months 21.5 33.7 55.2 14713 30.4 23.5 53.9 2.9 15.1 18 10677 48 or more months 20.7 30.8 51.5 5227 26.4 20.9 47.3 3.6 15.6 19.2 3694 Birth order 1 17.4 32 49.4 9719 24.8 23.3 48.1 3 13.5 16.5 6630 2 to 3 19.5 32.7 52.2 15209 27.3 22.5 49.8 3.2 14.2 17.4 10634 4 to 5 23.7 34 57.7 6848 32.6 24 56.6 3.6 15.5 19.1 5125 6 and above 26.8 33 59.8 4031 36.6 23.4 60 2.9 14.5 17.4 3194 Total 20.6 32.8 53.4 35807 28.9 23.1 52 3.2 14.3 17.5 25584 Note: Figures are for children bom 1-47 months prior to the survey. Each o f index is expressed in standard deviation Units (S.D ) from the median o f the international Reference Population. The number o f children for calculation o f height-for-age does not include children from Andhra Pradesh, Himachal Pradesh, Madhya Pradesh, Tamil Nadu and W est Bengal In the case o f first-born twins, both twins are counted as first births because neither has a previous birth intervel. Source: Adapted from UPS (1995) -J Table 5.6 The means of variables used in the regression for weight-for-age, India W eighted N=25550 N am e o f v ariab le P ro x im ate d e te rm in a n ts M ean M inimum M aximum Age o f the child 1.96 1 3 Prenatal visits (num ber) 2.71 0 12 Delivery a t hom e (proportion o f cases) 0.73 0 1 Delivery at governm ent m edical facility (proportion o f cases) 0.15 0 1 Delivery at private medical facility (proportion o f cases) 0.11 0 1 Sm all size at birth (proportion o f cases) 0.19 0 1 M edium size at birth (proportion o f cases) 0.66 0 1 Large size at birth (proportion o f cases) 0.15 0 1 Im m unized for m easles (proportion o f cases) 0.45 0 1 Had fever during tw o w eeks before the survey (proportion o f cases) 0.21 0 1 U ndcrlvins m a tern al an d household d e te rm in an ts Education o f m other (years o f schooling) 2.81 0 20 Education o f father (years o f schooling) 5.54 0 22 M other's age less than 20 years (proportion o f cases) 0.65 0 1 M other's age 20 or more years (proportion o f cases) 0.35 0 1 Owned agricultural land (scale 0-1) 0.58 0 3 Socio-econom ic status (scale 1 to 3) 1.48 1 3 Owned radio (proportion o f cases) 0.40 0 1 Owned bicycle (proportion o f cases) 0.46 0 1 O w ned m otorcycle (proportion o f cases) 0.08 0 1 O wned car (proportion o f cases) 0.01 0 1 O wned tractor (proportion o f cases) 0.02 0 1 Had access to flush toilet (proportion o f cases) 0.16 0 1 Had som e toilet facility (proportion o f cases) 0.11 0 1 Had no toilet facility (proportion o f cases) 0.73 0 1 Had pucca house (proportion o f cases) 0.20 0 1 Had kuchha house (proportion o f cases) 0.52 0 1 Had sem i-p u c ca house (proportion o f cases) 0.28 0 1 Father in professional occupation (proportion o f cases) 0.09 0 1 Father in low er occupations (proportion o f cases) 0.91 0 1 In te ra c tio n s Prenatal visits*M other's education 13.59 27.99 228 Prenatal visits’ Father's education 19.62 31.76 240 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.7 The means of variables used in the regression for height-for-age and weight-for-height, India W eighted N =I9368 N am e o f v ariab le P ro x im a te d e te rm in a n ts M ean M inimum A ge o f the child 1.96 1 Prenatal visits (num ber) 2.46 0 Delivery at hom e (proportion o f cases) 0.76 0 Delivery at governm ent m edical facility (proportion o f cases) 0.13 0 Delivery at private medical facility (proportion o f cases) 0 .1 1 0 Small size at birth (proportion o f cases) 0.19 0 M edium size at birth (proportion o f cases) 0.68 0 Large size at birth (proportion o f cases) 0.14 0 Im m unized for m easles (proportion o f cases) 0.43 0 Had fever during tw o w eeks before the survey (proportion o f cases) 0.21 0 Maximum 3 12 U nderlying m a tern al and household d eterm in an ts Education o f m other (years o f schooling) Education o f father (years schooling) M other's age less than 20 years (proportion o f cases) M other's age 20 or m ore years (proportion o f cases) O wned agricultural land (acres) Socio-econom ic status (scale 1 to 3) O wned radio (proportion o f cases) Owned bicycle (proportion o f cases) Owned m otorcycle (proportion o f cases) Owned car (proportion o f cases) Owned tractor (proportion o f cases) Had access to flush toilet (proportion o f cases) Had som e toilet facility (proportion o f cases) Had no toilet facility (proportion o f cases) Had pucca house (proportion o f cases) Had kuchha house (proportion o f cases) Had sem i-p u c ca house (proportion o f cases) Father in occupation o f high status (proportion o f cases) Father in occupation o f low status (proportion o f cases) 2.79 5.76 0.63 0.37 0.61 1.50 0.39 0.47 0.08 0.01 0.02 0.16 0.11 0.73 0.21 0.49 0.30 0.10 0.90 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 20 22 1 I 3 3 1 1 1 I 1 I I 1 In te ra c tio n s Prenatal visits*M other's education Prenatal visits*Father's education 13.13 18.87 26.90 29.99 228 240 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.8 The means of variables used in the regression of two states, Kerala and Uttar Pradesh, for weight-for-age, height-for-age, and weight-for-height W eighted N=4957 N am e o f variab le P ro x im a te d e te rm in a n ts M ean Minimum Maximum A ge o f the child 1.93 1 3 K erala (proportion o f observations) 0.12 0 0 U ttar Pradesh (proportion o f observations) 0.88 0 1 Prenatal visits (num ber) 2.12 0 12 Delivery at home (proportion o f cases) 0.79 0 1 Delivery at governm ent m edical facility (proportion o f cases) 0.11 0 I Delivery at private m edical facility (proportion o f cases) 0.10 0 1 Small size at birth (proportion o f cases) 0.16 0 1 M edium size at birth (proportion o f cases) 0.75 0 1 Large size at birth (proportion o f cases) 0.09 0 1 Im m unized for m easles (proportion o f cases) 0.34 0 1 Had fever during tw o w eeks before the survey (proportion o f cases) 0.22 0 1 U nderlying m a tern al an d household dete rm in an ts Education o f m other (years o f schooling) 2.56 0 18 Education o f father (years o f schooling) 6.15 0 21 M other's age less than 20 years (proportion o f cases) 0.61 0 1 M other's age 20 or m ore years (proportion o f cases) 0.39 0 1 O wned agricultural land (scale 0 to 3) 0.51 0 3 Socio-econom ic status (scale 1 to 3) 1.54 1 3 O wned radio (proportion o f cases) 0.39 0 I O wned bicycle (proportion o f cases) 0.60 0 I O wned m otorcycle (proportion o f cases) 0.07 0 1 O wned car (proportion o f cases) 0 .0 1 0 1 O w ned tractor (proportion o f cases) 0.03 0 1 Had access to flush toilet (proportion o f cases) 0.16 0 1 Had som e toilet facility (proportion o f cases) 0.09 0 I Had no toilet facility (proportion o f cases) 0.74 0 1 Had pucca house (proportion o f cases) 0.17 0 1 Had kuchha house (proportion o f cases) 0.51 0 1 Had sem i-p u cca house (proportion o f cases) 0.32 0 1 Father in professional occupation (proportion o f cases) 0.09 0 1 Father in lower occupations (proportion o f cases) 0.91 0 1 In te ra c tio n s Prenatal visits’ M other's education 12.46 31.30 192 Prenatal visits*Fathcr's education 17.22 32.42 240 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.9 The sample means of variables for Kerala W eighted N = 978 Name o f variable Proximate determ inants Mean M inimum Maximum Age o f the child 1.96 I 3 Kerala (proportion o f observations) 1.00 1 1 Uttar Pradesh (proportion o f observations) 0.00 0 0 Prenatal visits (number) 7.50 0 12 Delivery at home (proportion o f cases) 0.11 0 1 Delivery at government medical facility (proportion o f cases) 0.39 0 1 Delivery at private medical facility (proportion o f cases) 0.50 0 1 Small size at birth (proportion o f cases) 0.21 0 1 Medium size at birth (proportion o f cases) 0.65 0 1 Large size at birth (proportion o f cases) 0.14 0 1 Immunized for measles (proportion o f cases) 0.67 0 1 Had fever during two weeks before the survey (proportion of 0.36 0 1 cases) Underlvine maternal and household determ inants Education o f mother (years o f schooling) 7.61 0 18 Education o f father (years o f schooling) 7.32 0 20 Mother's age less than 20 years (proportion o f cases) 0.38 0 1 Mother's age 20 or more years (proportion o f cases) 0.62 0 1 Owned agricultural land (scale 0 to 3) 0.10 0 3 Socio-economic status (scale 1 to 3) 1.40 1 3 Owned radio (proportion o f cases) 0.61 0 I Owned bicycle (proportion o f cases) 0.19 0 1 Owned motorcycle (proportion o f cases) 0.07 0 1 Owned car (proportion o f cases) 0.01 0 1 Owned tractor (proportion o f cases) 0.00 0 0 Had access to flush toilet (proportion o f cases) 0.62 0 1 Had no toilet facility (proportion o f cases) 0.28 0 1 Had pucca house (proportion o f cases) 0.17 0 1 Had kuchha house (proportion o f cases) 0.20 0 1 Had semi-pucca house (proportion o f cases) 0.63 0 1 Father in high-status occupation (proportion o f cases) 0.10 0 1 Father in low-status occupations (proportion o f cases) 0.90 0 1 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.10 The sample means of variables for Uttar Pradesh Weighted N = 3979 Name o f variable P ro x im a te d e te rm in a n ts M ean M in im u m M axim um A ge o f the child 1.93 1.93 3 K erala (p ro p o rtio n o f observ atio n s) 1.00 1.00 1 U ttar P rad esh (pro p o rtio n o f o b serv atio n s) 1.00 1.00 1 Prenatal v isits (num ber) 1.42 1.42 12 D eliv ery at hom e (proportion o f cases) 0 .8 8 0 .8 8 1 D elivery at govern m en t m ed ical facility (prop o rtio n o f cases) 0 .0 7 0 .0 7 1 D eliv ery at p rivate m edical facility (proportion o f cases) 0 .0 5 0 .0 5 1 Sm all siz e a t birth (proportion o f cases) 0 .1 6 0 .1 6 1 M edium size at birth (p ro p o rtio n o f cases) 0 .7 6 0 .7 6 1 L arge siz e a t birth (proportion o f cases) 0 .0 8 0 .0 8 1 Im m unized fo r m easles (pro p o rtio n o f cases) 0 .3 0 0 .3 0 1 H ad fev er d u rin g tw o w eeks b efo re th e survey (proportion o f 0 .2 0 0 .2 0 cases) U n d e rly in g m a te rn a l a n d h o u se h o ld d e te rm in a n ts E ducation o f m o th er (years o f schooling) 1.90 0 17 E ducation o f father (years o f schooling) 6 .0 0 0 21 M other's ag e less than 20 y ears (proportion o f cases) 0 .6 4 0 1 M other's age 20 o r m ore years (proportion o f cases) 0 .3 6 0 1 O w ned agricultural land (scale 0 to 3) 0 .5 6 0 3 S ocio-eco n o m ic status (scale 1 to 3) 1.55 1 3 O w ned radio (proportion o f cases) 0 .3 6 0 1 O w ned bicy cle (proportion o f cases) 0 .6 6 0 1 O w ned m o to rcy cle (pro p o rtio n o f cases) 0 .0 7 0 1 O w ned c a r (proportion o f cases) 0.01 0 1 O w ned tracto r (proportion o f cases) 0 .0 3 0 1 H ad access to flush to ile t (pro p o rtio n o f cases) 0 .1 0 0 1 H ad no to ile t facility (p ro p o rtio n o f cases) 0 .8 0 0 1 Had p u cca h ouse (proportion o f cases) 0 .1 7 0 1 H ad k u ch h a h o use (proportion o f cases) 0 .5 5 0 1 H ad se m i-p u cca house (p ro p o rtio n o f cases) 0 .2 8 0 1 F ath er in hig h -statu s o ccu p atio n (p ro p o rtio n o f cases) 0 .0 9 0 1 F ath er in low -status o ccu p atio n s (p ro p o rtio n o f cases) 0.91 0 1 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 6 REGRESSION RESULTS The regression results o f reduced m odels w ith the best BIC statistics for the w hole o f India (referred to as country m odels) are presented in Tables 6.1, 6.2 and 6.3. The regression results for reduced m odels w ith the best BIC statistics for K erala and U ttar Pradesh (referred to as state m odels) are presented in Tables 6.4, 6.5 and 6.6. All the six m odels had the low est B IC statistic am ong the m odels tried. H ow ever, the country and state m odels for w eight-for-age and height-for- age had p values low er than 0.05 for the “ Score T est for the Proportional O dds A ssum ption”. Taking into account the statem ents in the SA S/STA T U ser’s Guide (SAS Institute Inc. 1989) regarding the possibility o f the score test rejecting the null hypothesis m ore often than is w arranted, and o f A llison (1999) about the high likelihood o f getting low er p values for the score test (see Section 5.4.1) when there are m any independent variables and when the sam ple size is large, it was decided to calculate separate m odels, dichotom izing the dependent variables, to see the effect on the coefficients across the dichotom izations. T he results are in A ppendices 1 to 4. The coefficients o f all the variables that had statistically significant effects on the dependent variables in Tables 6.1, 6.2, 6.4 and 6.5 have 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the sam e signs in the dichotom ized m odels presented in the A ppendices. H ence the six m odels w ere considered to be the best, from am ong the m any m odels tried, in capturing the m ain features o f the data for explaining underw eight, stunting and w asting am ong Indian children tw elve to forty-seven m onths old. T able 6.7 gives the m ean probabilities o f severe undernourishm ent, undernourishm ent and norm al grow th in the country and state m odels. The country m odels are discussed first, and then, the state m odels. L ogit coefficients indicate log odds, w hich are rather difficult to interpret. It is easier to interpret the effects o f variables in terms o f predicated odds. The odds ratios, which are the exponents o f the coefficients, indicate the predicted odds. O dds ratios are also called adjusted odds ratios because they control for other variables in the m odel. T hey indicate the effects o f the respective variables on the odds o f being in the low er categories than being in the higher category (A llison 1999). In this study, the low er categories are severe undernourishm ent and undernourishm ent, and the higher category is normal growth. H ence w hile explaining the odds ratios, the low er categories are referred to com m only as underw eight, stunted or wasted. F or instance, odds o f being underw eight includes odds o f being both severe underw eight and underweight. H ow ever, probability equations give separate probabilities for all the three categories. H ence the three 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. categories are m entioned separately while explaining the results o f calculations using probability equations. 6.1 The results of the country models The five proxim ate determ inants that w ere significant in affecting the dependent variables in the reduced country m odels w ere the age o f the child, the num ber o f prenatal visits, the place o f birth, size at birth and having fever in the tw o w eeks before the survey. T he six m aternal and household determ inants that w ere significant w ere the schooling o f the m other, the schooling o f the father, the age o f the m other at first birth, the type o f toilet facility, living in a kachha (poor quality) house and the ow nership o f agricultural land by the household. The interaction betw een fath ers’ schooling and prenatal visits w ere statistically significant in the w eight-for-age m odel, and the interaction betw een m others’ schooling and prenatal visits w ere significant in height-for-age m odel. T here was no significant interaction in the w eight-for-height m odel. Gender, caste, religion, urban residence and source o f drinking w ater were som e o f the variables that did not have statistically significant effects in any o f the three models. There w ere other variables that w ere significant in m odels that had 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. slightly low er B IC statistics. O ne such variable w as socio-econom ic status (ses) that had a significant negative effect on stunting. H ow ever the dum m ies for absence o f access to any toilet facility and residence in a kachha or low quality house (both these variables w ere part o f the index m aking up the ses variable) w ere part o f the best-B IC m odels. A nother variable that had a significant positive effect on underw eight w as the low er occupational status o f the father. I shall explain below the results o f the best-B IC country m odels. Proxim ate determ inants: 6.1.1 A ge o f the child Children w ho had com pleted ages one, two and three years w ere included in this study. The m axim um age was forty-seven m onths. T hirty-six percent o f the children in the regression m odels had com pleted one year o f age, and thirty- tw o percent each had com pleted tw o and three years. T he average age o f the children in the regression sam ples analyzed was 1.96 years. C hildren w ho were less than one year old w ere om itted in order to capture the effect o f im m unizations uniform ly. International guidelines prescribe that children should be fully im m unized by age one (UPS 1995). M any studies in developing countries had found that pre-school children’s grow th rate w as influenced by their age 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. because o f w eaning from breast m ilk, increased exposure to infections w hen toddlers start m oving around, and inadequate diet com pared to th eir high requirem ents o f quantity and quality o f diet needed for grow th and to fight infections. T hese aspects w ere discussed in detail in Sections 3.2 and 3.3. In the w eight-for-age m odel presented in Table 6.1, a dum m y variable for com pleted age three w as constructed to capture the effect o f age, w ith children w ho com pleted ages one and tw o as the om itted category. In the height-for-age and weight-for- height m odels presented in Tables 6.2 and 6.3, age w as used as a continuous variable. In the w eight-for-age and w eight-for-height m odels, age had highly significant1 negative effects on underweight. In the height-for-age m odel, age had a highly significant, but positive effect on stunting. The evidence presented in C hapter 3 show ed that m ost o f the grow th retardation occurred during the first tw o years o f life. The pattern th at em erged in the w eight-for-age m odel (Table 6.1) w as very sim ilar to the pattern observed in Figure 3.5. T he odds ratio o f 0.79 indicated that, controlling for other variables, children w ho had com pleted three years o f age had only seventy-nine percent o f the odds o f being underw eight than children w ho w ere younger. T he use o f age as a continuous variable instead o f age three in the sam e m odel yielded a negative coefficient w ith an odds ratio o f 0.89, w hich show ed that each additional year o f 1 p-value below 0.01 (highly significant effect) denoted by ** in tables 6.1 to 6.6. 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. age was associated w ith an eleven percent decrease in the odds o f being underweight. T he use o f tw o dum m ies for com pleted ages tw o and three in the sam e m odel (children w ho had com pleted age one as the reference) also yielded negative coefficients. The odds o f underw eight w ere m arginally low er for children w ho had com pleted tw o years than for those w ho had com pleted only one year o f life, but the effect w as not statistically significant2. H ow ever, com pleted age three had a highly significant effect on underw eight and the odds o f underw eight w ere tw enty- tw o percent low er than for children w ho had com pleted only one year o f life. A dum m y for com pleted age one w as also tried in the sam e m odel w ith com pleted ages tw o and three being the om itted category. The dum m y show ed a highly significant and positive effect on underw eight. T he odds o f being underw eight for children less than tw o years old was 1.14 tim es or fourteen percent higher than for children who had com pleted tw o and three years. T he overall results o f the w eight-for-age m odel show ed that for children one to four years old, the highest predicted odds o f underw eight w ere before they com pleted two years o f age and the risk o f being underw eight declined as they becam e older. Table 6.2 presents the best-B IC m odel for height-for-age. A ge as a continuous variable had a positive coefficient and w as highly significant. The 2 /?-value higher than 0.05 (no significant effect) 208 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. odds ratio o f 1.24 show ed that at higher ages, the odds o f stunting w ere higher. Each y ear’s increase in age increased the odds o f stunting by nearly tw enty-four percent. R egressing age three com pleted as a dum m y instead o f age as a continuous variable in the sam e m odel show ed that the odds o f stunting for children w ho had com pleted three years w ere higher by forty-one percent than for younger children. Regressing tw o dum m ies for com pleted ages tw o and three show ed that the odds o f stunting for older children w ere higher than for children one year old. T he odds ratio for com pleted age tw o w as 1.2 and for com pleted age three w as 1.5 w hen com pared to children w ho had com pleted only one year o f age. T his show ed that the relative odds o f stunting at com pleted age three w as higher than at com pleted age two. A ge tw o w as statistically significant in the case o f stunting, w hereas it w as not so in the case o f underw eight. Several studies in developing countries had found sim ilar relationships betw een age and stunting. A cross-sectional study in rural H onduras described in Section 3.2 show ed that standard deviation values for height-for-age continued to decline until the third year o f life (Figure 3.4). A study in G am bia show ed th at substantial deficits in height and w eight developed early in life, and that the height lost w as rarely regained (B illew icz and M cG regor 1982). A longitudinal study in Philippines found th at ninety-six percent o f the children w ho w ere stunted at six m onths w ere 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. also stunted at tw o years o f age. This study also found that children rarely recovered from stunting (A dair and G uilkey 1997). Table 6.3 presents the results o f the w eight-for-height m odel. A ge had a highly significant negative effect on wasting. Each additional year o f age w as associated w ith a forty-tw o percent decline in the odds o f w asting. R egressing tw o dum m ies for com pleted ages tw o and three in the sam e m odel show ed that the odds o f w asting w ere forty-eight percent and sixty-five percent low er respectively than for children w ho had com pleted age one. D um m ies for com pleted ages one and tw o w ith age three as the om itted category show ed that the low er ages were significantly and positively associated with w asting. T he odds o f w asting for children w ho had com pleted one year o f age were nearly three tim es as high as that o f children w ho had com pleted three years o f age; and for children w ho had com pleted tw o years, it w as 1.5 tim es higher than those w ho had com pleted three years. T he regression results show ed that w asting w as the highest at com pleted age one and low er at ages tw o and three. This pattern w as sim ilar to w hat was found in Section 3.2 (Figure 3.5) for children in H onduras, G uatem ala and B angladesh. W eight-for-height sharply declined in all the three countries after the first three to six m onths o f life. This effect bottom ed out by the m iddle o f the second year o f life and thereafter w eight-for-height im proved. B y age three there was no evidence o f w asting am ong children in H onduras and G uatem ala. 21 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A ge w as negatively associated w ith underw eight and w asting, w hereas it was positively associated w ith stunting. M ore than 90 percent o f the children in this study sam ple w ere children o f fathers in occupations o f low er status. U nderw eight and w asting are usually acute conditions and stunting usually a chronic condition (V ictora 1991). C hildren in India appeared to have caught-up in w eight after the surviving the w orst initial years. H ow ever, conditions did not im prove enough for catch-up in height to occur. Poor living environm ents increase the risk o f infections and inadequate diets m ake it difficult to fight infections. T hese tw o reasons possibly prevented catch-up growth in height from taking place. 6.1.2 Prenatal visits Prenatal visits enable m others to obtain inform ation about m aternal and child health, health check-ups, vitam in and m inerals tablets, and tetanus toxoid injections. T hese help to im prove the health o f the m others, and the grow th o f the children before and after birth. Prenatal visits are also an occasion for both the parents to get educated about child-care practices that can help ensure the norm al grow th o f children in the crucial years after birth. M ore educated parents are usually in a better position to understand and follow the advice given during prenatal visits. G reater schooling was also associated with greater num ber o f 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. prenatal visits. T he m eans for prenatal visits in the country sam ples w ere about 2.5. Prenatal visits w ere used as a continuous variable w ith a m axim um value o f tw elve. Thirty-five percent o f the m others had no prenatal visits and forty-three percent had betw een one and four prenatal visits in the regression sam ples. Prenatal visits had highly significant and negative effects on both underw eight and stunting. Each additional prenatal visit reduced the odds o f children being underw eight by tw o percent (Table 6.1). Each additional prenatal visit reduced the odds o f stunting by four percent (Table 6.2). Prenatal visits had a significant3 negative effect on w asting (Table 6.3). Each additional prenatal visit reduced the odds o f w asting by tw o percent. Thus, the greater the num ber o f prenatal visits the m others had, the low er w ere the odds o f their children being underw eight, stunted and wasted. 6.1.3 Place o f birth The percentage o f births taking place at hom e can indicate socio-econom ic conditions such as the educational attainm ents o f parents and other m em bers o f the household, and the availability and utilization o f prenatal and delivery services in the com m unity. A bout seventy-five percent o f the births took place at hom e, fifteen percent in governm ent health facilities, and ten percent in private health facilities. Clean and safe delivery is im portant for the health o f the m other and the 3 p-value o f 0.05 or less (significant effect) denoted by * in tables 6.1 to 6.6. 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. child (The W orld H ealth O rganization 1994). M any o f the m aternal deaths and chronic m orbidity resulting from child birth in India w ere attributed to the failure to get tim ely help for com plications at birth (UPS 1995). Birth at hom e and birth at a governm ent facility w ere the tw o dum m ies created to capture the effect o f place o f birth on underw eight, w ith birth at a private hospital or clinic being the om itted category. B irths taking p lace at hom e had highly significant effects on increasing the odds o f both underw eight and stunting (Tables 6.1 and 6.2). B irths taking place at governm ent health facilities had highly significant effects in increasing the odds underw eight, but not the odds o f stunting. The predicted odds o f underw eight for children bom at hom e w ere thirty-tw o percent higher, and for those bom at a governm ent health facility w ere nineteen percent higher than for children w ho w ere b om at a private hospital or clinic (T able 6.1). The predicted odds o f stunting w ere thirty-seven percent higher for children bom at hom e than for those bom at a governm ent facility or a private hospital or clinic. Place o f birth did not have a significant effect on the predicted odds o f w asting (Table 6.3). 6.1.4 Size at b irth : Tw o dum m ies for sm all size at birth and average size at birth w ere created to capture the effect o f birth size, large size being the om itted category. Low birth 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w eight has been associated w ith greater physical ill health and p o o r m ental developm ent leading to failure in school (Gribble 1993). Poor m aternal nutrition has been associated w ith low birth weight. Poor m aternal nutrition affects pregnancy outcom es and the ability o f the m other to breastfeed the child (Khan and K abir 1997). Thus size at birth is a proxy for m any factors that affect child grow th both before and after birth. Both sm all and average sizes a t birth had positive and highly significant effects on underw eight, stunting and wasting. N ineteen percent o f the children in the w eight-for-age sam ple w ere sm all and sixty-six percent w ere average and fifteen percent w ere large at birth. The odds o f underw eight for children w ho w ere sm all at birth w ere m ore than tw ice the odds for children w ho w ere large at birth (Table 6.1). For children o f average size at birth, the odds o f underw eight w ere nearly one and a h a lf tim es the odds for children who w ere large. The odds o f stunting for children w ho w ere sm all at birth w ere nearly 1.5 tim es higher than for children large at birth. Children w ho w ere o f average size at birth had 1.3 tim es higher odds o f stunting than those large at birth (Table 6.2). T he odds o f w asting for children o f sm all size at birth w ere 1.75 tim es higher than for children w ho w ere large at birth (Table 6.3). The odds o f w asting for children o f average size w ere 1.3 tim es higher than for children w ho w ere large at birth. 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thus, sm all and average sizes at birth increased the odds o f underw eight, stunting and w asting highly significantly o f pre-school children in India. 6.1.5 H aving fever Fever had a highly significant and positive effect on w eight-for-age and w eight-for-height (Tables 6.1 and 6.3). T w enty-one percent o f the children had fever during the tw o w eeks p rior to the survey. The odds for children w ho had fever being underw eight w ere tw enty-four percent higher than the odds for children who did not have fever. The odds o f w asting for children w ho had fever w ere tw enty-tw o percent higher than for those w ho had no fever. Fever had no significant effect on stunting. 6.1.6 Schooling o f the m other The m ean num ber o f years o f schooling o f the m others in the regression sam ples w as 2.8 years. Sixty-three percent o f the m others and thirty-five percent o f the fathers w ere illiterate. T w enty five percent o f the m others h ad five years or m ore o f schooling. Fifty percent o f the fathers had com pleted five years or m ore o f schooling. T he years o f schooling o f both the parents w ere regressed on the dependent variables so that the effect o f m aternal schooling w as not exaggerated. Ignoring father’s education w as found to upw ardly bias the effect o f m others’ 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. schooling by 40 to 70 percent in Thailand (B ehrm an and S ussangkam 1989). Fathers’ schooling in T hailand w as found to b e at least as im portant as m others’ schooling in determ ining child nutritional outcom es. There is likely to be a sim ilar relationship in m ost states o f India, w here m en m ake im portant decisions regarding children’s health. M any researchers and agencies have advocated m others' schooling for im proving child health and nutrition in developing countries (W orld B ank 1993; K ing 1990; K ing and Hill 1993; Schultz 1993a, b). Several studies observed the positive effects o f m others' schooling on the grow th o f children. A recent study found that the level o f schooling o f the m other w as negatively associated w ith stunting in India (M ishra and R etherford 2000). Som e studies found that the m ain benefit to child health from w om en’s education w as the ability o f educated w om en to process and use available inform ation to im prove children’s health (D a V anzo and Levy 1989; T hom as Straus and H enriques 1991). Schooling w as also found to increase parents’ abilities and m otivation to im prove the child health and nutritional status (Z acharia 1984; R osenw eig and Schultz 1982). M any studies indicate that better educated m others w ere able to utilize public health facilities to a greater extent for im proving the nutritional status o f children (Sen 1996; Caldw ell 1986; D reze and G azdar 1996). Some researchers found that the effect o f m others’ education was a proxy for other background characteristics like socio- 216 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. econom ic status, m others’ nutrition in her youth and household endow m ents (Desai and A lva 1998; Thom as, Strauss and H enriques (1990); W olfe and B ehrm an 1987). It has also been pointed out that variations in the quality o f schooling, rather than only the quantity, should be taken into account. It w as also found that higher quantity o f schooling produced greater quality o f schooling (Behrm an 1995). N FH S survey gathered inform ation on w hether w om en could read and w rite based on the ability o f wom en to w rite their nam e and read som e literature given. This variable, however, was found to be less effective in capturing the effect o f schooling on undem utrition than the num ber o f years o f schooling. The results o f the country regressions in this study show ed that m aternal schooling w as an im portant determ inant o f children’s health and nutritional status in India and had highly significant and negative effects on underw eight, stunting and wasting. The w eight-for-age m odel presented in T able 6.1 show ed that each y ear’s increase in m others' schooling was associated w ith a four percent reduction in the predicted odds o f the children being underw eight. In the height-for-age model (Table 6.2), maternal schooling had a higher negative effect. Each year’s increase in m aternal schooling was associated w ith a five percent decline in the odds o f stunting. In the w eight-for-height m odel (Table 6.3), each year o f 217 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m aternal schooling w as associated w ith a three percent decline in the odds o f wasting. 6.1.7 Schooling o f the father The m ean num ber o f years o f fathers' schooling w as about 5.5 years. Fathers' schooling had a significant and negative effect on underw eight, but did not figure in the other tw o reduced models. In the w eight-for-age m odel, each year’s increase in fathers' schooling was associated with a percentage reduction in the odds o f underw eight. Fathers' education w as an im portant variable in the w eight- for-age m odel because it w as part o f the interaction variable. T he im portance o f father’s education in determ ining the nutritional status o f children is discussed in Section 6.1.12. 6.1.8 Low age o f the m other at first birth M others w ho w ere less than tw enty years old at their first birth w ere grouped together as low -age m others. N early sixty-five percent o f the m others in the regression m odels had their first births before the age o f tw enty. The dum m y variable for low -age first births had highly significant and positive effects on underw eight and stunting; but w as not part o f the reduced w eight-for-height m odel. The odds o f underw eight for children o f low-age m others w ere tw enty- 218 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. three percent higher than for children o f m others older at their first birth. Low age at first birth w as associated w ith an eighteen percent increase in the odds o f stunting. Thus low age at first birth increased the odds o f both underw eight and stunting. Low age at first birth was associated w ith low er parental literacy in the N FH S data. Teenage pregnancies w ere found to affect fetal grow th because grow ing m others appeared to reserve fat reserves for their ow n grow th even w hen there was adequate w eight gain during pregnancy (Scholl et al. 1994). 6.1.9 A bsence o f o r poor toilet facility D iet and infections are the tw o im portant determ inants o f the nutritional status o f children. Poor toilet facilities have been found be associated w ith greater incidence o f infections and poor nutritional status o f children. In the w eight-for- age model, the dum m y for having no toilet facility was highly significant and increased the odds o f underw eight o f children by thirty percent as com pared to children living in households with som e toilet facility or a flush toilet (Table 6.1). In a slightly low er-B IC m odel, the two dum m ies for having som e toilet facility and no toilet facility w ere significant. H aving som e toilet facility increased the odds o f underw eight by sixteen percent and having no toilet facility increased the odds o f underw eight by forty percent, as com pared to having a flush toilet. This show ed 219 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. that better toilet facilities resulted in better nutritional status and grow th o f children. In the height-for-age m odel presented in Table 6.2, both the dum m ies for having som e toilet facility and no facility w ere highly significant and increased the odds o f stunting by tw enty-three and tw enty-seven p ercent respectively, as com pared to having a flush toilet. In the w eight-for-height m odel, having no toilet facility w as highly significant and increased the odds o f w asting by tw enty-five percent as com pared to having som e toilet facility or flush toilet facility (Table 6.3). The dum m y for having som e toilet facility w as not significant. Sim ilar results w ere found by other studies. B etter w ater and sanitation w ere associated w ith less diarrheal diseases and less m orbidity and im proved nutritional status in children (B urger and E srey 1995). T he m agnitude o f im provem ent in health outcom es w as show n to increase from no sanitation, to latrines and to toilets (A nker and K now les 1980; H aines and A very 1982; Esrey 1993). P roper disposal o f feces w as found to be likely to reduce the num ber o f pathogens and transm itted through routes such as food and other m edia that is ingested by young children (Esrey and Feachem 1989; Im ong et al. 1989). Further, im proved sanitation w as found to com pensate for poverty. Y ee (1984) found that the m ean height-for-age w as significantly higher for children in low- incom e households that had flush toilets than for children in low -incom e 220 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. households that had pit latrines or no toilets. This study also found that differences in height-for-age in children o f high-incom e households with or without flush toilets w ere m uch less. B etter sanitation w as found to have the largest im pact in contam inated environm ents w hen incom e w as low and breastfeeding w as absent, and appeared to com pensate for these tw o factors (B urger and Esrey 1995). Som m erfelt and Stewart (1994) analyzed D H S data o f nineteen countries and found that the m edian levels o f underw eight for children living in households w ithout any to ilet facilities, with latrines and w ith flush toilets w ere tw enty-six, tw enty-tw o and nine percent respectively. T he corresponding figures for stunting w ere thirty-five, tw enty-eight and seventeen percent respectively. A sim ilar pattern w as observed in the case o f w asting also, though not uniform ly in all the countries. 6.1.10 L iving in poor quality house T he quality o f house w as found to be associated w ith the grow th o f children in m any studies. In rural Honduras, the quality o f the house in term s o f the num ber o f room s and quality o f construction o f the floor w ere found to be associated w ith the heights o f children (A nderson 1979; M artorell et al. 1984). The better the quality o f the house, the taller w ere the children. Poor housing has been associated with acute low er respiratory infections (A LR I) that are a m ajor 221 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cause o f grow th faltering and child m ortality in developing countries (Huffm an and Steel 1995; Ballard and Neum ann 1995). In the w eight-for-age m odel presented in Table 6.1, residence in a kachha house (poor quality house defined in Section 5.3.2.11) w as highly significant and increased the odds o f being underw eight by tw enty-six percent as com pared to those residing in a sem i-pucca (m edium quality) or pucca (high-quality) house. In the height-for-age m odel, residence in a kachha house increased the odds o f stunting by 10 percent (Table 6.2). The quality o f the house did not significantly affect wasting. 6.1.11 O w nership o f agricultural land A gricultural land is an im portant asset in rural areas in India. Land accounted for m ore than fifty percent o f the total assets o f rural households and m ore than eighty percent o f the rural households depended on land for their livelihood in 1981-82 (Sharm a 1994). Yet, sixty percent o f the households in the regression sam ples did not own any land. A nother tw enty-six p ercent o f the households ow ned five or less acres o f land. O w nership o f agricultural land had highly significant negative effects on underw eight and stunting, but did not figure in the best-BIC w eight-for-height m odel. In creating the variable, a scale was created for different levels o f land ow nership (no land, 5 or less acres, 6 to 10 222 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. acres and m ore than 10 acres). For each increase in the level o f agricultural land held, the odds o f being underw eight declined by nine percent and the odds o f stunting declined by eleven percent. H ow ever, these odds ratios can at best only indicate a rough approxim ation o f the effect o f agricultural land for the follow ing reasons. B oth irrigated and unirrigated lands w ere added up w ithout giving a higher w eight to irrigated land even though irrigated land is usually m ore productive. T he groups created w ere also broad. F o r instance, in the category o f greater than zero to five acres, about fifty percent o f the households ow ned less than tw o acres o f land. In the tw o highest groups, the frequencies w ere low: less than seven percent each. Even so, the results give an indication o f the effect o f ow ning agricultural land in India. R edistribution o f agricultural land on large scale is difficult at this juncture and there is lim ited surplus land available according to the existing land ceiling laws. H ow ever there are m any loopholes that can be closed in the existing ceiling law s in m any states, that can benefit the w eaker segm ents o f the population (Sharm a 1994). F or instance, m any tenants still cultivate leased land. T hey can be given ow nership o f the land at a price laid dow n by the law, or, the resum ption o f the land for self-cultivation by the landlord prohibited and the m axim um rents that landlords can charge can be laid down. Such steps w ould m ake tenancy m ore secure and enable tenants to invest in the land and im prove its productivity. A s 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w e saw in Sections 4.3 and 4.4, even sm all assets enhance the econom ic security and the ability o f the poor to secure their rights and im prove their entitlem ents. A n anthropom etric study o f 802 children betw een one and three years o f age in B razil found that children o f laborers w ere m ore likely to be underw eight, stunted and w asted than those o f landow ners; and children o f sharecroppers fell in betw een the form er two categories (V ictora et al 1985). The study also found that there w ere no differences between children based on the area o f land ow ned by households. H ow ever, m alnutrition w as found to be associated w ith land tenure patterns: children in districts with very large ranches w ere m ore likely to present underw eight, stunting and w asting than districts w ith sm all farm s. This study indicates beneficial effects from greater socioeconom ic equality for child growth. 6.1.12 Interaction betw een prenatal visits and fathers’ schooling T he interaction betw een prenatal visits and fathers’ schooling had a highly significant and negative effect on the odds o f underw eight (Table 6.1). Prenatal visits and the num ber o f years o f schooling o f fathers also had highly significant independent effects on underw eight. T he odds ratio o f the interaction indicated that w hen prenatal visits and fathers’ education increased by a unit, the odds o f being underw eight declined by 0.4 percent. 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In interpreting the interaction it is relevant to consider the role o f the fathers in decision-m aking about child health and household earnings. N inety percent o f the fathers w ere em ployed in occupations lower in status than clerks. Sixty-six percent o f the fathers w orked in farm s or as fisherm en. W hen the dum m y for low er w ork status o f the fathers w as added to the best-B IC m odel, it had highly significant and positive effects on underw eight and yielded a m arginally low er-B IC . This m eant that low er w ork status o f fathers increased the odds o f underw eight. Since eighty percent o f the w om en in the N FH S sam ple w ere not w orking for an income, fathers w ere the m ain earners. Children require diets o f higher quality and quantity, relative to their size, for norm al grow th (Section 1.3, Figure 1.2). Low er earnings m ake better-quality diets for children less affordable. H igher schooling is usually associated w ith higher socio-econom ic status and higher earnings. This m akes fathers’ schooling an im portant determ inant o f earning capacity, household diet and child grow th. Further, educated fathers are able to be better inform ed and m ake better decisions about child health. Paternal education assum es even greater im portance in the context o f the low literacy rates and the low autonom y o f w om en in m ost states o f India, and the social restrictions on w om en’s travel unaccom panied by m ales in the household. T hese factors lim it the role o f m others in child health. An educated husband accom panying a pregnant w om an can better appreciate the m edical advice given 225 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and use it to produce better child health. Therefore fathers schooling is an im portant determ inant o f child grow th in India. The other term in the interaction is the num ber o f prenatal visits. Prenatal visits can play an im portant role in the health o f the m other and the norm al grow th o f children, as w as discussed earlier (Section 6.1.2). W hen the parents are illiterate or less educated, prenatal visits can provide them w ith inform ation about pregnancy and childcare, provided prenatal-care providers m ake the effort and succeed in conveying the know ledge to the parents. T he interaction show ed that such efforts w ere inadequate and children o f less literate parents benefited very little from prenatal visits. These reasons m ay explain the highly significant interaction effects o f prenatal visits and fathers’ schooling in the w eight-for-age model. 6.1.13 Interaction betw een prenatal visits and m others’ schooling In the height-for-age m odel presented in T able 6.2, the interaction betw een prenatal visits and m others’ education w as highly significant and negatively associated w ith stunting. The interaction reduced the odds o f stunting by 0.4 percent for each unit increase in the interaction. The beneficial effects o f m others’ schooling on the nutritional status o f children w ere discussed in Section 6.1.6 and 226 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o f prenatal visits in Sections 5.3.2.1 and 6.1.2. B oth these variables also had independent highly significant and negative effects on stunting. 6.1.14 R esults o f probability calculations in the country m odels Probability equations used in this study are given in S ection 5.4.2. These equations provide separate probabilities for each o f the three categories o f the dependent variable (Table 6.7). In the w eight-for-age m odel, the m ean probability o f severe underw eight was 0.18, o f underw eight was 0.36 and o f norm al grow th w as 0.46. In term s o f percentages, the m ean probability o f an average child being severely underw eight w as 18 percent, o f being underw eight w as 36 percent and o f being norm al in w eight-for-age was 46 percent. A ssigning different values in place o f m ean values allow us to see the effect o f changes in these variables on the different probabilities. T he m ean age was 1.96 years. T he probability o f severe underw eight was 0.22 at age one, 0.18 at age tw o and 0.15 at age three. The probability o f underw eight w as 0.38 at age one, 0.36 at age tw o and 0.33 at age three. Thus w ith higher age the probability o f undem utrition for w eight-for-age declined. The m ean num ber o f years o f m others’ schooling w as about three. Increasing m others’ schooling to six years, keeping all other variables at their m eans, reduced the probabilities o f severe underw eight and underw eight by tw o percentage points 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. each. For each y ear’s increase in m aternal schooling, the probability o f severe underw eight and underw eight in children reduced m ore or less at a constant rate. K eeping all other m eans constant, and m others’ education at eight and tw elve years, the probabilities o f severe underw eight w ere fifteen and thirteen percent; o f underw eight thirty-three and thirty-one percent; and o f norm al w eight-for-age fifty-tw o and fifty-five percent respectively. This show ed th at increasing m others’ education w as an effective m eans to reduce underw eight am ong Indian children. B y assigning different values for prenatal visits and fathers’ education, the effects o f changes in these variables on the nutritional status o f children w ere estim ated. The results are given in Table 6.8. K eeping all other variables at their m eans and increasing prenatal visits from their m ean o f about three to six, reduced the probability o f severe underw eight and underw eight by tw o percentage points each, show ing the effectiveness o f prenatal visits in reducing the levels o f grow th retardation. Increasing prenatal visits to eight and tw elve, reduced the m ean probabilities o f severe underw eight to fifteen and thirteen percent, and o f underw eight to thirty-three and thirty-one percent, respectively. Increasing fathers’ schooling to nine years reduced the probability o f severe underw eight and underw eight by one percentage point each. 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Increasing both prenatal visits and fathers' schooling from their m eans to eight and nine respectively, decreased severe underw eight by five percentage points and underw eight by four percentage points. T he interaction o f prenatal visits and fathers' education show ed that sim ultaneous increase in them w as m ore effective in reducing severe underweight and underw eight than increasing them one at a time. B efore deciding on interventions to im prove child grow th, the relative cost effectiveness and the tim e required in effecting changes in these variables will have to be considered. C hanges in tw o other variables w hose effects w ere exam ined from the point o f view o f policy w ere the age o f the m other at first birth and the place o f delivery. Increasing age at first birth and reducing births at hom e depend on m any socio-econom ic and cultural conditions, and m edical services available in the com m unity. T herefore im provem ents in them take a longer tim e and involve considerable resources. C alculations using probability equations show ed that reducing the percentage o f low -age births and births at hom e alone w ere not viable policy options for reducing underw eight. A ge at birth and institutional deliveries are usually positively associated with literacy rates, prenatal visits, better health awareness and availability o f m edical facilities. In the height-for-age m odel, the m ean probability o f severe stunting was thirty-four percent and the m ean probability o f stunting w as tw enty-eight 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. percent. It m ay be noted that w hile the probabilities o f severe underw eight w as m uch low er than underw eight, and o f severe w asting m uch low er than w asting, the probability o f severe stunting w as considerably higher than that o f stunting (Table 6.7). This indicates the severity o f the problem o f stunting and o f chronic m alnutrition in India. As seen in earlier, age w as positively associated with stunting, w hereas it w as negatively associated with underw eight and wasting. Substituting values for age show ed that the probability o f severe stunting was tw enty-nine percent at age one, thirty-four percent at age tw o, and thirty-nine percent at age three. The probability o f stunting varied very little w ith age, it increased ju st by a percentage point betw een ages one and three. Therefore age has im portant im plications for policy. T his w ill be discussed in the next chapter. The m eans for prenatal visits and m aternal schooling w ere 2.5 and 2.8 respectively. Increasing m aternal schooling and prenatal visits one at a tim e had independent negative effects on stunting. H ow ever, increasing prenatal visits to eight and m aternal schooling six years at the sam e tim e had a greater negative effect on severe stunting (-11) than the additive effect (-10) o f increasing these variables one at a time. The results using the probability equations, presented in Table 6.9, show ed that increasing m aternal education and prenatal care at the sam e time produced better results than increasing each individually. 230 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Changes m ade in the m ean values for place o f birth, size at birth, ow nership o f agricultural land and age at first birth in probability equations did not yield viable policy options. In the w eight-for-height m odel, the m ean probabilities o f severe w asting and w asting w ere three percent and fifteen percent respectively (Table 6.7). T he probability equations show ed that the probabilities o f severe w asting at ages one, tw o and three w ere five percent, three percent and tw o percent respectively. The probabilities o f w asting w ere tw enty-tw o percent, fourteen percent and nine percent at ages one, two and three respectively. Thus age had a significant negative effect on severe w asting and wasting. Policy im plications o f age are discussed in the next chapter. Increasing m aternal schooling from its m ean o f three to six years did not have any effect on severe w asting and had a m arginal effect on w asting. Increasing m aternal schooling from three to ten years reduced the probability o f severe w asting from three percent to tw o percent. C hanges in the values o f other variables in the m odel, including prenatal visits, toilet facilities and fever, did not change the probabilities o f w asting and severe w asting. The prevalence o f w asting at the rate o f five percent in a population w as considered to be elevated by the W H O (1995). A ccording to this, the prevalence o f w asting in India was extrem ely high. 231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2 The results of the state models In the state regressions nine variables and two interactions had significant effects on undernourishm ent (Tables 6.4 to 6.6). T he additional variable that was included in the state m odels was the state o f residence. A dum m y w as created to capture the effect o f residence in K erala, w ith residence in U ttar Pradesh as the reference. T he effects o f the variables in the state m odels are given below . Dum m ies for measles vaccination and fem ale gender had significant negative and positive effects respectively on w asting w hen these variables w ere added to the best-B IC m odels. H ow ever they w ere not significant in the other tw o state models. 6.2.1 A ge o f the child In the three state m odels, age w as used as a continuous variable taking values o f one, tw o and three years o f com pleted ages. T he age patterns in the three state m odels w ere sim ilar to the patterns observed in the country m odels. The predicted odds o f underw eight and w asting declined by nine percent and forty-eight percent, respectively, w ith each y ear’s increase in age. The m agnitudes o f the reduction in the odds w ere sim ilar to those found in the country m odels. In the height-for-age m odel presented in Table 6.5, the predicted odds o f 232 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stunting w ere positively associated w ith age. Each y ear’s increase in age increased the predicted odds o f stunting by tw enty-nine percent. The corresponding the increase in the country m odel w as tw enty-four percent. T he higher odds o f stunting in the state m odel is likely to be the result o f the higher prevalence o f stunting in U ttar Pradesh than for India as a whole. T hus age had a significant effect on undem utrition in the state m odels as well. 6.2.2 Place o f birth Births at hom e had a highly significant effect on the odds o f underw eight and stunting. Seventy-nine percent o f the births w ere at hom e, ten percent w ere in governm ent m edical facilities, and the rest in private hospitals or clinics. The predicted odds o f underw eight for births at hom e w ere forty-eight percent higher than for births in a governm ent or a private m edical facility. T he odds o f stunting for births at hom e w ere forty-eight percent higher than for births in a governm ent or a private m edical facility. Both these odds w ere higher than the odds in the respective country m odels. Place o f birth w as not part o f the best B IC m odel for w eight-for-height. This w as also the case in the country m odel (T able 6.3). 233 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2.3 Size at birth Sixteen percent o f the children in the state sam ple w ere o f sm all size and seventy-five percent w ere o f average size at birth. Sm all size at birth had highly significant and positive effects on the predicted odds o f underw eight, stunting and w asting. T he odds o f underw eight for children who w ere sm all at birth w ere 130 percent higher than the odds for large children (Table 6.4). The po o r environm ent that produced sm all birth size appeared to have persisted, negatively affecting w eight-for-age in early childhood. The odds o f stunting for children sm all at birth w ere tw enty-four percent higher than the odds for those w ho w ere o f average or large size at birth. T he odds o f w asting for children sm all at birth w ere forty-four percent higher than the odds for those o f average or large size at birth. A verage size at birth had a highly significant effect on increasing the odds o f underw eight, but had no significant effect on stunting and w asting in the best- BIC m odels. The odds o f underw eight for children w ho w ere o f average size w ere fifty-three p er higher than the odds for children large at birth. In the country m odels, both sm all and average sizes had significant effects on underw eight, stunting and w asting. Small and average sizes at birth appeared to be good indicators predicting the probabilities o f grow th faltering in early childhood years. 234 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2.4 Schooling o f the father The m ean num ber o f years o f schooling o f the fathers w as 6 years. Schooling o f the father had a highly significant effect in reducing the odds o f underw eight, but had no effects on th e odds o f stunting or w asting. Each y ear’s increase in the schooling o f the father reduced the odds o f underw eight by two percent. T he reasons for the likely im portance o f father’s schooling and earning capacity on the nutritional status o f children w ere discussed in Section 6.1.12. 6.2.5 Schooling o f the m other The m ean num ber o f years o f schooling o f the m others w as 2.5 years. The w ider gap betw een m aternal and paternal education in the state m odels than in the country m odels w as because o f the very low fem ale literacy rates in U ttar Pradesh (m ean 1.9 years). M others’ schooling had highly significant and negative effects on stunting and w asting, but not on underw eight in the reduced state m odels (Tables 6.5 and 6.6). Each y ear’s increase in m others’ schooling resulted in the predicted odds o f stunting and w asting declining by four percentage points each. Thus, m aternal education had a m ajor effect on stunting and w asting in K erala and U ttar Pradesh. In the country m odels, m aternal education had significant negative effects on all three types o f undem utrition. 235 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2.6 Low age o f m others at first birth Sixty-one percent o f the m others in the state m odels had their first births w hen they w ere less than tw enty years old. The age at first birth had positive and highly significant effects on the odds o f underw eight and w asting. This variable was not part o f the height-for-age m odel. The odds o f underw eight for children w hose m others had their first births before the age o f tw enty w ere tw enty-four percent higher than the odds for children w hose m others w ere older at their first birth. The odds o f w asting was tw enty-six percent higher for children o f m others less than tw enty at first birth than for children o f older m others. The pattern in the state m odels differed from that o f the country m odels, w here the dum m ies for low age at first birth had highly significant effects on underw eight and stunting, but not on w asting. 6.2.7 A bsence o f toilet facility A bsence o f toilet facilities had significant and positive effects on the odds o f underw eight and w asting. The odds o f underw eight for children living in households w ithout access to toilet facilities w ere tw enty-eight percent higher than the odds o f underw eight for children having access to flush toilets or som e type o f toilet facility (Table 6.4). The type toilet facility the household had 236 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. access to did not have significant effects in the height-for-age and w eight-for- height m odels. 6.2.8 T he effect o f residence in Kerala This variable w as used to capture the differences in the environm ent o f child grow th in K erala and U ttar Pradesh. T he environm ent includes the im m ediate, underlying and basic factors affecting child grow th that are depicted in Figure 1.1. In the weight-for-age and the height-for-age m odels presented in Tables 6.4 and 6.5 respectively, the dum m ies for residence in K erala had highly significant and negative effects on underw eight and stunting. Residence in Kerala reduced the odds o f underw eight by forty-nine percent and the odds o f stunting by fifty-tw o percent. T he negative effect o f residence in K erala on underw eight and stunting w ere substantial even after controlling for the proxim ate, and m aternal and household variables that affected undernourishm ent. This show ed that the environm ent o f child grow th w as m uch better in K erala than in U ttar Pradesh. T he discussion in C hapter 4 show ed that w eaker segm ents o f the population in K erala w ere better educated and better provided for through m any public policy m easures than in U ttar Pradesh. B etw een 1977-78 and 1987-88 K erala had the highest percentage o f reduction in the population living below the poverty line am ong all the m ajor states o f India (T able 1.6). K erala also had better 237 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. indices reflecting quality o f life (Table 1.5) and low er rates o f undem utrition am ong children in India (Table 1.7). H ence these results w ere not unexpected. H ow ever, residence in K erala had no significant effect on wasting. 6.2.9 Interaction betw een prenatal visits and fath ers’ education Prenatal visits had no independent significant effects in the w eight-for-age and height-for-age m odels (Tables 6.4 and 6.5), and w ere not part o f the w eight- for-height m odel (Table 6.6). H ow ever, prenatal visits w ere part o f the interactions that had highly significant and significant negative effects in the w eight-for-age and w eight-for-height m odels respectively. Each unit increase in the interaction betw een prenatal visits and fathers’ schooling reduced underw eight by 0.8 percent. The corresponding reduction in the county m odel w as only 0.4 percent. The m ean num ber o f prenatal visits was tw o and the m ean num ber o f years o f fathers’ schooling w as six. T he prenatal visits variable had a maxim um value o f twelve and m axim um v a lu le for fathers’ schooling w as twenty-one. Hence the interaction could have a large negative effect on underw eight if the values o f these variables w ere large. T he interaction show ed that in the absence o f fathers’ education, prenatal visits did not reduce the odds o f underw eight. O n the other hand, fathers’ schooling had a highly significant and negative effect on underw eight, even in the absence o f prenatal visits. This m ade 238 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fathers’ education im portant for child grow th. C ross-tabulations show ed that sixty-seven percent o f the w ives o f illiterate m en had no prenatal visits and the num ber o f prenatal visits increased w ith the num ber o f years o f schooling o f the fathers. T hus fathers’ schooling was associated w ith greater num ber o f prenatal visits and low er underw eight o f children. A s w e saw in Section 6.1.12, fathers’ education affects child grow th through factors such as earnings and better diets for children, and decisions about medical care. In Section 6.1.2, the beneficial effects o f prenatal visits w ere also discussed. In sum , the sim ultaneous increase in prenatal visits and fathers’ schooling had significant effects on reducing odds o f underweight. Paternal schooling is all the m ore im portant w hen m others are illiterate or have low levels o f education, and have low autonom y in decisions related to child health. T his w as the case in U ttar Pradesh. In order to test this hypothesis, regressions w ere run for data o f the tw o states o f K erala and U ttar Pradesh separately. T he num ber o f observations (N ) in K erala w as 978 and the N in U ttar Pradesh w as 3979. The m eans for m aternal schooling w ere 7.6 years in Kerala and 1.9 years in U ttar Pradesh; the m eans for paternal schooling w ere 7.3 years in K erala and six years in U ttar Pradesh. T he interaction betw een prenatal visits and fathers’ schooling in K erala w as not statistically significant, w hereas it w as highly significant in U ttar Pradesh. This difference is probably the resu lt o f the 239 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relatively less im portant role o f fathers in decisions regarding child health in Kerala than in U ttar Pradesh. 6.2.10 Interaction betw een prenatal visits and m others’ education In the height-for-age m odel, the interaction betw een prenatal visits and m others’ schooling had a significant and negative effect on the odds o f stunting. W ith each u n it increase in the interaction, the odds o f stunting declined by 0.6 percent. T he corresponding increase in the country m odel for stunting w as 0.4 percent. C onsidering the m axim um values for the prenatal visits (tw elve years) and the m aternal schooling variables (eighteen years), the interaction could have a large negative effect on stunting if the values o f these variables w ere large. The benefits o f prenatal visits and m aternal education for the health and grow th o f children w ere discussed in Sections 6.1.6 and hence, is not discussed here. The absence o f significant effects o f prenatal visits on w eight-for-age and height-for-age o f children and the significant effect o f prenatal visits in the presence o f m aternal and paternal schooling has im plications for policy. It is likely that in the absence o f schooling parents are not able to understand w hatever advice is given about child health and im prove their children’s health. Therefore doctors and other prenatal care providers should advise illiterate and less educated parents about child health in a m anner that they understand the advice and are able 240 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to follow it. D octors and others m ay need to be specially sensitized and trained for this and provided w ith charts and other pictures that they can use to instruct these parents. T his also applies to the use o f m ass m edia to im part know ledge about nutrition. Program s should be specifically designed keeping in m ind the illiterate and less educated segm ents o f the population. T hese conclusions are supported by a recent study w hich found that exposure to the electronic m edia did not significantly affect child m alnutrition in India (M ishra and R etherford 2000). 6.2.11 R esults o f probability calculations in the state m odels T he m ean probabilities o f severe underw eight, underw eight and norm al grow th in the state m odel was tw enty-five percent, forty percent and thirty-five percent respectively (Table 6.7). In the state m odels, the m ean o f prenatal visits was tw o and the m ean num ber o f years o f paternal schooling w as six years. Increasing prenatal visits to eight, keeping other variables at their m eans, decreased severe underw eight by four percent and underw eight by one percent; increasing prenatal visits to tw elve decreased severe undeiw eight by six percent and underw eight by tw o percent (Table 6.10). Increasing paternal schooling to nine years, keeping other variables at their m eans, decreased severe underw eight by tw o percent, with no change in percentage underw eight; increasing paternal 241 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. education to tw elve years decreased severe underw eight by four percent and underw eight by one percent. Increasing prenatal visits to eight and fathers’ schooling to nine years sim ultaneously reduced the probability o f severe underw eight by from tw enty-five percent to eighteen percent and underw eight from forty percent to thirty-seven percent. A sim ultaneous increase in prenatal visits to tw elve and fathers’ schooling to tw elve years reduced the probability o f severe underw eight to eleven percent and o f underw eight to thirty percent, respectively. The effect o f increasing prenatal visits and fathers’ schooling one at a tim e, w as lesser than the effect o f increasing both sim ultaneously. This show ed that sim ultaneously increasing prenatal visits and paternal schooling can decrease the probability o f severe underw eight and underw eight m ore than by increasing them by the sam e am ount one at a tim e. Substituting zero in place o f the m ean o f seventy-nine percent for births at hom e, keeping other variables at their m eans, decreased severe underw eight from tw enty-five percent to nineteen percent, and underw eight from forty to thirty-eight percent. Substituting zero in place o f its m ean o f sixteen percent for sm all size at birth decreased severe underw eight by one percent, with no change in percentage underw eight. H ow ever, increasing institutional births are a long-term process involving changes in cultural and social attitudes, aw areness, and availability o f m edical facilities. Increasing size at birth depends on im provem ents in m aternal nutrition, socioeconom ic conditions and the 242 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. greater utilization o f prenatal services. Hence changes in both these variables involve tim e and substantial financial outlays. Probability calculations in the height-for-age m odel show ed that increasing prenatal visits and m aternal education, one at a time, w ere m ore effective in reducing severe stunting and stunting than increasing both o f them together (Table 6.11). Increasing prenatal visits to eight, keeping other variables at their m eans, decreased severe stunting by five percent, with no change in percentage stunted. Increasing m aternal schooling to six years, keeping other variables at their means, decreased severe stunting by four percent, with no change in percentage stunted. Increasing both prenatal visits and m aternal schooling to eight and six respectively reduced severe stunting by eight percent, which w as one percent less than the additive effect o f increasing these tw o variables one at a tim e. T his w as the only interaction that produced low er effects than the additive effect o f increasing the variables separately. R educing births at hom e and sm all size at birth did not appear to be viable policy options as discussed in the previous section. A ge affected the probability o f severe stunting. The probability o f severe stunting at ages one, tw o and three w ere thirty-three percent, thirty-nine percent and forty-five percent respectively. A ge had no significant effect on the probability o f stunting. 243 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The probabilities o f severe w asting and w asting declined w ith age. A t com pleted ages one, tw o and three the probabilities o f severe w asting w ere four percent, tw o percent and one percent respectively. The probabilities o f w asting at ages one, tw o and three w ere tw enty-tw o percent, fourteen p ercen t and nine percent respectively. T hese results w ere sim ilar to the results found in other countries (Section 3.2, Figure 3.6). It was also found that increasing m aternal schooling from 2.5 years to six years did not result in a decline in severe w asting at all, and only a one percent decline in wasting. H ow ever, increasing m aternal schooling to tw elve years decreased the probability o f severe w asting to tw o percent and o f w asting to fourteen percent. N one o f the other variables yielded policy relevant results in the w eight-for-height model. Figures 6.1 to 6.6 show the predicted probabilities o f severe undernourishm ent, undernourishm ent and norm al grow th for children w ith different values for the independent variables. ‘A verage’ indicates the probabilities for children w ith the m ean values for the independent variables. ‘P olicy’ indicates the probabilities for children with greater n um ber o f prenatal visits, higher schooling, institutional delivery etc. as show n below each figure. ‘W orst’ indicates the probabilities for children living in environm ents that are highly unfavorable for norm al grow th w ith no prenatal visits, delivery at hom e, illiterate parents and so on. “B est’ indicates the probabilities for children living in 244 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. conditions th at are m ore or less ideal for norm al growth. T he figures show that environm ent has a strong effect on grow th. Prenatal care, parental education, size at birth, sanitation and housing are som e o f the variables that have strong effects on child growth. 6.3 Correlation results Table 6.12 show s the P earson’s correlation coefficients betw een the three indices for undernourishm ent in the country m odels (the results for the state m odels w ere also alm ost the sam e). T he N for correlations w as 19368. All the correlation coefficients w ere significant at .0001 levels. U nderw eight and stunting w ere highly and positively correlated. This indicated that w hen height increased w eight also increased. U nderw eight and w asting were also positively correlated, but the correlation w as not as high as the correlation betw een underw eight and stunting. Stunting and w asting w ere negatively correlated show ing that w hen height increased, w eight-for-height reduced. B ut the correlation cannot be considered substantively significant, though statistically highly significant because o f the large sam ple size. 245 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V ictora (1991) analyzed the correlation betw een w asting and stunting am ong children 12-23 m onths old in Africa, Asia, Latin A m erica and Eastern M editerranean countries using data from 175 previous studies. F our o f the thirty- tw o studies in A sia w ere from India. He found a strong correlation betw een w asting and stunting in A sia. The results obtained for India from the analysis o f N FH S data do not show such a correlation. In studies done in five developing countries A nderson (1979) found that w eight-for-age and height-for-age w ere highly correlated, as is the case in Table 6.12. These results are a clear indication that in India grow th is largely determ ined by environm ental, rather than genetic factors. W ith im provem ents in living conditions, both the height and the w eight o f children are likely to go up sim ultaneously. T he study by A garw al et al. (1991) also arrived at the sam e conclusion. Pearson Correlation coefficients between weight-for-age, height-for-age and weight-for-height W eight-for-age Height-for-age W eight-for-height Table 6.12 W eight-for-age 0.597 0.384 Height-for-age 0.597 -0.065 W eight-for-height 0.384 -0.065 246 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.4 Conclusion T he regression and correlation analyses show ed that the poor environm ents in which they lived w ere largely responsible for the high levels o f undernourishm ent am ong children in India. The num ber o f prenatal visits and the num ber o f years o f parental schooling w ere the tw o ch ief variables that were found to be affecting m alnutrition and poor growth. The interaction betw een parental schooling and prenatal visits w ere found to have strong effects on child grow th and increasing both these sim ultaneously was m ore effective than increasing them one at a time. O ther variables that affected child grow th w ere size at birth, sanitation and housing. A ge had a strong effect on the nutritional status o f children. H igher percentages o f older children w ere undernourished for weight- for-age and w eight-for-height than younger children, w hereas a higher percentage o f older children w ere stunted than w ere younger children. This show ed that chronic undem utrition w as a serious problem in India. Im provem ent in protein- energy intake and low ering the incidence o f infections can im prove nutritional status and reduce chronic m alnutrition. C ollective efforts o f w eaker segm ents o f the population to influence public policy priorities in their ow n favor, and their being vigilant to ensure the high quality o f public services, can im prove living conditions for the w hole population and lead to lesser child grow th retardation in 247 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. India. The com parative study o f K erala and U ttar Pradesh in C hapter 4 supports this view . 248 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.1 Predicted probabilities of severe underweight, underweight and normal weight-for-age of children 1-4 years old: India, 1992-93 1.00 0 .9 0 0 .8 0 0 .7 0 0 .6 0 0 .5 0 0 .4 0 0 .3 0 0.20 0.10 0.00 0 .8 2 0 .6 7 0 .4 6 0 .3 6 0 .2 4 0 .2 5 0 .1 8 0 .1 4 0 .0 8 Average P o lic y W orst N Severely underweight ti Underweight “ Normal Best Average: mean sample values Policy: 8 prenatal visits, birth at a government facility, medium size at birth, 6 years o f maternal schooling, 9 years o f fathers’ schooling, has flush toilet or som e toilet facility, lives in better quality house Worst: N o prenatal visits, birth at home, sm all size at birth, both parents illiterate, mother o f low age at first birth, does not own agricultural land, has no toilet facility, lives in poor quality house Best: 12 prenatal visits, birth at a government facility, medium size at birth, had no fever, 12 years each o f maternal and paternal schooling, mother not o f low age at first birth, has flush toilet or som e toilet facility t o -i^ vo 0552 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.2 Predicted probabilities of severe stunting, stunting and normal height-for age of children 1-4 years old: India, 1992-93 1.00 0 .9 0 0 .8 0 0 .7 0 0 .6 0 0 .5 0 0 .4 0 0 .3 0 0.20 0.10 0.00 0 .8 1 0.6 0 0 .5 1 0 .3 8 0 .3 4 0 .2 8 0.22 0 .1 7 0.12 0 .0 7 A verage P olicy W orst B est * Severely Stunted o Stunted ~ Normal Average: mean sample values Policy: 8 prenatal visits, institutional delivery, medium size at birth, 6 years o f maternal schooling, has flush toilet or som e toilet facility, lives in better quality house Worst: N o prenatal visits, birth at home, small size at birth, mother illiterate, mother o f low age at first birth, does not own agricultural land Best: 12 prenatal visits, institutional delivery, medium size at birth, 12 years o f maternal schooling, mother not o f low age at first birth, has flush toilet or som e toilet facility, lives in better quality house to la o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.3 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Predicted probabilities of severe wasting, wasting and normal weight-for-height of children 1-4 years old: India, 1992-93 0.82 0.87 ~07W 0.75 0.15: 0.03 0.11 0.20 0.02 . W W W 0.05: 0.09: 0.02 Average P olicy W o rs t Best ® Severely wasted i Wasted - Normal Average: mean sample values Policy: 8 prenatal visits, medium size at birth, 6 years o f maternal schooling, has flush toilet or som e toilet facility Worst: N o prenatal visits, small size at birth, mother illiterate, has no toilet facility Best: 12 prenatal visits, medium size at birth, had no fever, 12 years o f maternal schooling, has flush toilet or som e toilet facility t o 499999999 9999991 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.4 Predicted probabilities of severe underweight,underweight and normal weight-for-age of children 1-4 years old: Kerala-Uttar Pradesh, 1992-93 1.00 0 .9 0 0 .8 0 0 .7 0 0 .6 0 0 .5 0 0 .4 0 0 .3 0 0.20 0 .8 4 0 .6 1 0 .4 4 0 .4 0 0 .3 7 0 .3 5 0 .2 9 0 .2 5 0 .1 8 0 .1 3 0 .0 3 m s s 0.00 Average Policy Severely underweight + Underweight - Normal Worst Best Average: mean sample values Policy: 8 prenatal visits, no births at home, medium size at birth, 9 years o f fathers’ schooling, mother not o f low age at first birth Worst: N o prenatal visits, birth at home, small size at birth, father illiterate, mother o f low age at first birth Best: residence at Kerala, 12 prenatal visits, medium size at birth, had no fever, 12 years o f fathers’ schooling, mother not o f low age at first birth to cn t o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.5 Predicted probabilities of severe stunting, stunting and normal height-for-age of children 1-4 years old: Kerala-Uttar Pradesh, 1992-93 1.00 0 .9 0 0.86 0 .8 0 0 .7 0 0 .5 6 0 .6 0 0 .5 0 0 .4 0 0 .3 0 0.20 0 .5 2 0 .3 9 0 .3 3 0 2 5 0 .2 3 0 .2 4 0.20 0 .0 9 0 .0 5 0.10 0.00 Average Policy Worst Best ai Severely Stunted E Stunted - Normal Average: mean sample values Policy: 8 prenatal visits, no deliveries at home, not o f small size at birth, 5 years o f maternal schooling Worst: N o prenatal visits, birth at home, small size at birth, mother illiterate Best: residence at Kerala, 12 prenatal visits, no deliveries at home, not o f small size at birth, 12 years o f maternal schooling to on C O 148544422955 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 6.6 Predictedprobabilities of severe wasting, wasting and normalweight-for-height of children 1-4 years old: Kerala-Uttar Pradesh, 1992-93 1.00 0.90 0.80 i 0.70 0.60 0.50 0.40 H 0 3 0 020 0.10 0.00 0.85 0.89 0.92 0.12 0.02 0.09 0.02 0.78 0.19 O.Ot IM S W t 0.01 0.07 J E E E B Average Policy Worst Best ^ Severely w asted & W asted - Normal Average: mean sample values Policy: not o f small size at birth, 8 years o f maternal schooling, mother not o f low age at first birth Worst: small size at birth, mother illiterate, mother o f low age at first birth Best: residence at Kerala, not o f small size at birth, 12 years o f maternal schooling, mother not o f low age at first birth to Ol 2769887144 Table 6.1 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on weight-for-age in Indian children 1-4 years old in 1992-93 W eighted N =25550 Variables C oefficients Standard E rror O dds Ratios Intercept 1 -1.8327** 0.0675 Intercept2 -0.1675** 0.0666 Proxim ate D eterm inants A ge o f the child (3 years com plete) -0.2380** 0.0253 0.79 N um ber o f prenatal visits -0.0221** 0.0072 0.98 Delivery at home 0 .2809** 0.0467 1.32 Delivery at governm ent facility 0 .1710** 0.0489 1.19 Small size at birth 0 .7432** 0.0411 2.10 Average size at birth 0 .3710** 0.0344 1.45 H ad fever during 2 weeks before the 0 .2173** survey U nderlying m aternal and household determ inants 0.0290 1.24 N um ber o f years o f m others’ schooling -0.0391** 0.0043 0.96 N um ber o f years o f fathers’ schooling -0.0077* 0.0036 0.99 M other less than 20 years at first birth 0 .2037** 0.0258 1.23 Ow nership o f agricultural land -0.0995** 0.0138 0.91 H ousehold has no toilet facility 0 .2 5 6 8 * * 0.0329 1.29 H ouse is kachha 0 .2 2 7 6 * * 0.0264 1.26 Fathers' schooling*Prenatal visits -0.0 0 3 6 * * 0.0009 0.996 Score Test p- value 0.0001 -2 log likelihood estim ate -2646.6 D F 14 B IC -2504.5 255 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.2 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on height-for-age in Indian children 1-4 years old in 1992-93 W eighted N =19368 Variables Coefficients Standard O dds Error Ratios Intercept 1 -1 .5768** 0.0822 Intercept2 -0.4336** 0.0814 Proxim ate Determ inants Age o f the child 0.21 3 5 * * 0.0170 1.24 N um ber o f prenatal visits -0 .0398** 0.0078 0.96 Delivery at home 0.31 0 6 * * 0.0418 1.36 Small size at birth 0.3659** 0.0502 1.44 Average size at birth 0.2567** 0.0422 1.29 U nderlvine m aternal and household D eterm inants N um ber o f years o f m others’ schooling -0 .0519** 0.0062 0.95 M other less than 20 years at first birth 0.1673** 0.0302 1.18 O w nership o f agricultural land -0.1157** 0.0159 0.89 H ousehold has som e toilet facility 0.20 8 0 * * 0.0584 1.23 Household has no toilet facility 0.2386** 0.0501 1.27 H ouse is kachha 0.0972** 0.0309 1.10 M others' schooling* Prenatal visits -0.0035** 0.0012 0.996 Score Test />value 0.0001 -2 log likelihood estim ate -1827.7 D F 12 BIC -1709.2 256 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.3 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on weight-for-height in Indian children 1-4 years old in 1992-93 W eighted N =19368 Variables Coefficients Standard O dds E rror Ratios Intercept 1 -2.7716** 0.0976 Intercept2 -0.8245** 0.0909 Proxim ate Determ inants A ge o f the Child -0.5473** 0.0245 0.58 N um ber of prenatal visits -0.0175* 0.0082 0.98 Small size at birth 0.5518** 0.0706 1.74 Average size at birth 0.2457** 0.0624 1.28 Had fever during 2 w eeks before the 0 .2017** 0.0451 1.22 survey U nderlying maternal and household determ inants N um ber o f years o f m others’ schooling -0.0285** 0.0061 0.97 H ousehold has no toilet facility 0 .2190** 0.0525 1.25 Score Test p-value 0.5739 -2 log likelihood estim ate -768.8 D F 7 BIC -699.7 257 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.4 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on weight-for-age in Uttar Pradesh and Kerala in children 1-4 years old in 1992-93 W eighted N =4957 Variables C oefficients Standard O dds Error Ratios Intercept 1 -1 .7226** 0.1520 Intercept2 0.0108 0.1503 Proxim ate Determ inants Residence at Kerala -0 .7065** 0.1191 0.49 A ge o f the child -0 .0949** 0.0294 0.91 Num ber o f prenatal visits 0.0141 0.0215 1.01 Delivery at hom e 0 .3 938** 0.0870 1.48 Small size at birth 0 .83 2 2 * * 0.1045 2.30 Average size at birth 0 .42 7 7 * * 0.0897 1.53 Underlying m aternal and household determ inants N um ber of years of fathers’ schooling -0 .0210** 0.0062 0.98 Household has no toilet facility 0.24 6 3 * * 0.0517 1.28 M other less than 20 years old at first birth 0 .2 115** 0.0647 1.24 Fathers' schooling*Prenatal visits -0 .0077** 0.0022 0.99 Score Test p-value 0.0001 -2 log likelihood estimate -660.2 D F 10 B IC -575.1 258 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.5 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on height-for-age in Uttar Pradesh and Kerala in children 1-4 years old in 1992-93 W eighted N =4957 Variables Coefficients Standard O dds Error Ratios Intercept 1 -1.011** 0.110 Intercept2 0.138 0.111 Proxim ate Determinants R esidence at Kerala -0.657** 0.116 0.52 A ge o f the Child 0 .2 5 5 * * 0.030 1.29 N um ber o f prenatal visits -0.030 0.017 0.97 Delivery at home 0.390** 0.087 1.48 Small size at birth 0.215** 0.067 1.24 U nderlying maternal and household determ inants N um ber o f years o f m others’ schooling -0.046 0.010 0.96 M others' schooling*Prenatal visits -0.006* 0.002 0.99 Score T est p-value 0.0047 -2 log likelihood estimate -760.32 D F 7 B IC -701.28 259 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.6 Logistic regression coefficients, odds ratios, likelihood estimate and BIC showing the effects of proximate, maternal and household covariates on weight-for-height in Uttar Pradesh and Kerala in children 1-4 years old in 1992-93 W eighted N =4957 Variables Coefficients Standard O dds E rror Ratios Interceptl -2.63** 0.125 Intercept2 -0.59** 0.103 Proxim ate Determ inants Residence at Kerala -0.01 0.136 0.99 A ge of the child -0.65** 0.047 0.52 Sm all size at birth 0.36** 0.089 1.44 U nderlying m aternal and household determ inants N um ber of years o f m others’ schooling -0.05** 0.011 0.96 M other less than 20 years at first birth 0.23* 0.077 1.26 Score Test p-value .2626 -2 log likelihood estim ate -270.19 D F 5 BIC -227.64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.7 Probabilities of severe undernutrition, undernutrition and normal growth in weight-for-age, height-for-age and weight-for-height in India, and in the states of Kerala and Uttar Pradesh Probabilities fin percentages') India K erala and U ttar Pradesh Severe underweight 18 25 Underweight 36 40 Normal growth in w eight-for-age 46 35 Severe stunting 34 39 Stunting 28 28 Normal growth in height-for-age 38 33 Severe wasting 3 2 W asting 15 12 Normal growth in w eight-for-age 82 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.8 Probabilities for the interaction between prenatal visits and fathers’ schooling keeping all other variables at their means in the weight-for-age model for India, 1992-93 N um ber o f N um ber o f years Probability (percentage') o f Prenatal visits F ath er’s schooling Severe U nderw eight Normal underweight growth ean (2.7) M ean (5.5) 18 36 46 8 M ean 15 33 52 M ean 9 17 35 48 8 9 13 32 55 12 M ean 13 31 56 M ean 12 16 34 49 12 12 10 26 64 262 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.9 Probabilities for the interaction between prenatal visits and father’s schooling keeping all other variables at their means in the height-for-age model for India, 1992-93 N um ber o f N um ber o f years o f Probability ('percentage') Prenatal visits M others’ schooling Severe stunting Stunting Norm al growth M ean (2.7) M ean (2.8) 34 28 38 8 M ean 28 27 45 M ean 6 30 27 43 8 6 23 25 51 12 M ean 24 26 50 M ean 12 23 25 52 12 12 12 18 70 263 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.10 Probabilities for the interaction between prenatal visits and father’s schooling keeping all other variables at their means in the weight-for-age model for Kerala and Uttar Pradesh, 1992-93 N um ber o f N um ber o f years Probability (percentage) of Prenatal visits F ath er’s Severe U nderw eight Normal schooling underweight growth M ean M ean 25 40 35 8 M ean 21 39 40 M ean 9 23 40 37 8 9 18 37 45 12 M ean 19 38 43 M ean 12 21 39 40 12 12 11 30 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.11 Probabilities for the interaction between prenatal visits and father’s schooling keeping aii other variables at their means in the height-for-age model for Kerala and Uttar Pradesh, 1992-93 N um ber o f N um ber o f years o f Prenatal visits M others’ schooling M ean M ean 8 M ean M ean 6 8 6 12 M ean M ean 12 12 12 Probability ('percentage) Severe stunting Stunting Normal erowth 39 28 33 34 28 38 35 28 37 31 28 41 32 28 40 29 27 44 23 26 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 7 CONCLUSIONS OF THE STUDY AND POLICY IMPLICATIONS The five proxim ate determ inants that had significant effects on child grow th retardation in India were: children’s age, num ber o f prenatal visits, place o f birth, size at birth and having fever. T he six m aternal and household determ inants that had significant effects w ere: num ber o f years o f m others’ schooling, num ber o f years o f fathers’ schooling, age o f m others at first birth, toilet facilities, quality o f the house and area o f agricultural land ow ned by the family. In addition, residence in K erala had significant effects on child growth. The tw o significant interactions w ere: prenatal visits and fathers’ schooling, and prenatal visits and m others’ schooling. The follow ing conclusions em erge from this study: Prenatal visits had significant negative effects on underw eight, stunting and w asting independently, and in interaction w ith father’s schooling in the w eight-for-age, and in interaction w ith m others' schooling in the height-for-age country m odels. Both the interactions had greater negative effects on grow th 266 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. retardation than the additive effects o f prenatal visits and schooling. This is evident from Tables 6.8 and 6.9. It appears that prenatal visits w ere m uch m ore beneficial to m ore literate parents than to less literate parents. Parents w ith greater schooling are m ore likely to understand and put into practice, the know ledge prenatal visits provide about m aternal and child health. N early fifty percent o f the illiterate m others did not have prenatal visits. M ore than sixty percent o f the m others and thirty-five percent o f the fathers w ere illiterate. O verall, the probabilities o f grow th retardation w ere fifty-four and sixty-tw o percent respectively for w eight-for-age and height-for-age. The probabilities o f grow th retardation w ere higher for the children o f less educated parents. H ence, to im prove the nutritional status and grow th o f children in India, it is im portant to target less literate parents. The inform ation that less literate parents should receive during prenatal visits should be clearly laid down. It should be ensured that parents understand and follow the instructions about m aternal and child health. For this, prenatal caregivers should spend m ore tim e with less literate parents and repeat the inform ation as often as required, and check that the advice given has been understood and follow ed correctly. Prenatal care providers should be m ade aw are o f these aspects and trained to effectively to com m unicate with illiterate and less-educated parents. Specially designed pictures, charts etc. are likely to help in better com m unication. T he m ass m edia should be used to air 267 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. radio and television program s specifically designed to enhance the nutritional know ledge o f the less educated. Further, parents should be m ade aw are o f the adverse consequences o f grow th retardation on physical and m ental developm ent, w ork capacity, and future earnings o f their children. This is likely to alert and encourage them to pay greater attention to the nutritional status and the grow th o f their children. U nless pregnant and lactating m others and children obtain adequate nutrition, child grow th is likely to retarded. Im provem ents in child grow th usually follow im provem ents in the quality o f life o f the entire population. G row th retardation in childhood affects physical and m ental w ork capacity, productivity and the econom ic well being o f the nation. T herefore public expenditure for providing adequate nutrition to w eaker segm ents o f the population should be considered as necessary developm ent expenditure. In addition to providing nutrition to poor m others and children through program s like the Integrated C hild D evelopm ent Schem e (ICD S), program s like the public distribution system (PD S) should be expanded and w ell im plem ented, as w as done in K erala. Even though the scope for reducing rural inequalities through land reform s is limited, it is possible to im plem ent the existing law effectively and close loopholes, to enhance the food security and bargaining pow er o f w eaker sections in rural areas. For instance, tenants could be given security o f tenure by preventing eviction by 268 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. landlords for self-cultivation, and m axim um rents that m ay be charged could be laid dow n. Som e o f the other m easures that can help the poor are distribution o f hom esteads, assistance for housing and higher and m ore effective public spending on schooling, public health and public sanitation. T hese steps are likely to raise the nutritional levels o f the poorer segm ents o f the population and decrease the rates o f child grow th retardation. C hildren’s age w as found to be an im portant determ inant o f their nutritional status. U nderw eight and w asting declined w ith age. H ow ever, stunting increased w ith age, indicating that those who w ere stunted rarely recovered lost height. Infections and inadequate diet had the greatest adverse effects on grow th in the first couple o f years o f life. Parents should be m ade aw are o f the crucial im portance o f tim ely m edical care and proper feeding during these early years. In the long run, there appears to be no alternative to the spread o f literacy for achieving sustained im provem ents in the quality o f peo p le’s lives and child grow th. A part from enabling parents to take the right decisions about child health, literacy is a basic requirem ent for effectively participating in the dem ocratic process, dem and favorable public policies and to ensure that public services function effectively. A w areness about health and literacy also encourages private and com m unity organizations to com e forw ard to provide schooling and health services to the people. T hese organizations can supplem ent governm ent’s efforts 269 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and im prove the quality o f services available to the com m unity. G lobalization and liberalization tend to result in deceased public spending on social services and social security in developing countries. This does not augur w ell for productivity and econom ic growth in the long run. G iven the high rate o f m alnutrition in India, public spending to raise nutritional standards and child grow th needs to be enhanced rather than be cut down. G row th potential, grow th patterns and the basic causes o f child grow th retardation are largely the sam e throughout the world. The differences in child grow th are, by and large, the result o f the different environm ents in w hich children live. Therefore, m any o f the findings and conclusions o f this study are likely to be relevant in other developing countries. C hild grow th is influenced by im m ediate, underlying and basic causes. In order to reduce the high levels o f child grow th retardation in India the follow ing short-term and long-term policies are suggested: Short-term policies: (1) Im proving m aternal and child health by increasing the num ber o f prenatal visits and effectively im parting health inform ation to less literate parents. Parents should be m ade aw are o f the causes, age patterns and the consequences o f child grow th retardation. T he 270 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. enhanced aw areness is likely to result in their paying greater attention to child health. (2) Expanding and effectively im plem enting food supplem entation program s for poor m others and children, PD S, land reform s and other social security m easures like distribution o f hom esteads to benefit vulnerable segm ents o f the population. Long-term policies: (1) O nly im provem ents in the underlying and basic factors that affect child grow th can result in better child grow th in India. The m ost effective w ay to m ake this happen w ould be to em pow er w eaker sections o f the society to collectively w ork for their well being through the dem ocratic process. For this, w idespread literacy is a basic requirem ent. Therefore priority should be given to expand and raise the quality o f schooling. 271 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Bibliography ACC-SCN. 1997. The Third Report on the World Nutrition Situation. United Nations. Geneva: Switzerland. 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 o f some techniques.” American Journal of Clinical Nutrition 35:1147-54. Adadevoh, S.W., C. Hobbs, and T.E. Elkins. 1989. “The relation o f the True Conjugate to Maternal Height and Obstetric Performance in Ghanians.” International Journal of Gynecology and Obstetrics (Ireland) 28 (3): 243-251. Adair, L.S., and K.D. Guilkey. 1997. “Age-specific determinants of stunting in Filipino children.” Journal of Nutrition 127 (2): 314-320. Agarwal, A.N., and H.O. Varma. 1996. India: Economic Information Yearbook 1996. New Delhi: National Publishing House. Agarwal, A.N., H.O, Varma, and R.C. Gupta. 1993. India: Economic Information Yearbook 1992-93. New Delhi: National Publishing House. Agarwal, K.N., A.M. Tripathi, and G.P. Katiyar. 1974. “Physical growth at adolescence.” Indian Pediatrics 11, 93-97. Agarwal, K.N., D.K. Agarwal, D.G. Benakappa, S.M. Gupta, P.C. Khanduja, S.P. Khatua, K. Ramachandran, P.M. Udani and C. Gopalan. 1991. Growth Performance of Affluent Indian Children (Under-jives): Growth Standard for Indian Children. New Delhi: Nutrition Foundation of India. Agresti, A. 1990. Categorical Data Analysis. New York: John Wiley and Sons. Allison, P. D. 1999. Logistic regression: Using the SAS system, theory and application. Cary: SAS Institute Inc. Anderson, M.A. 1979. “Comparison of anthropometric measures o f nutritional status in preschool children in five developing countries.” American Journal o f Clinical Nutrition 32:2339-2345. Anker, R., and J.C. Knowles. 1980. “An empirical analysis o f morbidity differentials in Kenya at the macro and micro levels.” Economic Development and Cultural Change 29: 165-185. 272 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Balakrishnan, P., and B. Ramaswami. 1992. Public storage and private speculation. The competition for supplies in the Indian wheat market. Paper prepared for presentation at the Far East and South Asia Econometrics Society M eeting in Bombay, December 1992. Ballard, T.J., and C.G. Neumann. 1995. “The effects o f malnutrition, parental literacy and household crowding on acute lower respiratory infections in young Kenyan children.” Journal of Tropical Pediatrics 41: 8-13. Banerjee, S. 1994. “Obstacles to change.” Economic and Political Weekly 26 March. Barbin, B. 1991. “An assessment of low birthweight risk in primiparae as an indicator o f malaria control in pregnancy.” International Journal of Epidemiology 20: 276- 283. Beaton G.H. 1986. “E.V. McCollum International Lectureship in nutrition. Toward harmonization of dietary, biochemical and clinical assessments: the meanings of nutritional status and requirements.” Nutrition Reviews 44: 349-358. Beaton G.H., A. Kelly, J. Kevany, R. Martorell, and J. Mason. 1990. Appropriate uses of anthropometric indices in children. ACC/SCN Nutrition Policy Discussion Paper No. 7. Geneva: World Health Organization. Beaton, G.H. 1989. “Small but healthy? Are we asking the right question?” Human Organization 48 (I): 127-134. Behrman, J.E., and C. Sussangkam. 1989. “Parental Schooling and Child Outcomes: Mother versus Father, Schooling Quality, and Interactions.” University o f Pennsylvania, Mimeographed. Behrman, J, R. 1995. Household behavior, Pre-school child health and nutrition and the role o f information. In Child Growth and Nutrition in Developing countries, edited by P. Pinstrup-Anderson, D. Pelletier, H. Alderman. Ithaca, NY: Cornell University Press. Beisel, W. R. 1977. “Resume of the discussin concerning the nutritional consequences of infection.” American Journal of Clinical Nutrition 30:1294-1300. Bhat, R. 1993. “The private health care sector in India.” In Paying for India’ s health care, edited by P. Berman and M.E. Khan. New Delhi: Sage Publications. Billewicz, W.Z., and I. A. McGregor. 1982. “A Birth-to-Maturity Longitudinal study o f Heights and Weights in Two West African (Gambian) Villages, 1951-1975. ” Annals of Human Biology 9 (4): 309-320. Black, R. E., K.H. Brown, S. Becker, A.R.M. Abdue Alim, and M.H. Merson. 1982. “Contamination o f weaning foods and transmission of enteropathogenic Escherichia coli diarrhea in children in rural Bangladesh.” Transactions of the Royal Society of Tropical Medicine and Hygiene 72:259-264. 273 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Brown, K.H., R.E. Black, S. Becker, S. Nahar, J. Sawyer. 1982. “Consumption of foods nutrients by weanlings in rural Bangladesh.” American Journal of Clinical Nutrition 36:878-889. Burger, S.E., and S.A. Esrey. 1995. “Water and sanitation: Health and nutritional benefits to children.” In Child Growth and Nutrition in Developing countries, edited by P. Pinstrup-Andersen, D. Pelletier and H. Alderman. Ithaca: Cornell University Press. Butz, W.P., J.P. Habicht and J. Da Vanzo. 1984. “Environmental factors in the relationship between breastfeeding and infant mortality: The role of sanitation and water in Malaysia.” American Journal of Epidemiology 119:516-525. Buzina, R., and K. Umera. 1974. “Selection of the minimal anthropometric characteristics to assess nutritional status.” Adv. Exp. Med. Biol. 49:271. Caldwell, J.C. 1986. “Routes to low mortality in poor countries.” Population and Development Review 12. Caldwell, J.C., P.H. Reddy, and P. Caldwell. 1983. “The social component of mortality decline: An investigation in South India employing alternative methodologies.” Populations Studies 37, no. 2 (July): 185-205. Calloway, D.H., S. Murphy, J. Balderstone, 0 . Receveur, D. Lein, and M. Hudes. 1992. Village Nutrition in Egypt, Kenya, and Mexico: Looking across the CRSP Projects. University of California, Berkeley. Mimeographed. Cameron, N. 1986. “The methods o f auxological anthropometry.” In Human Growth: A Comprehensive Treatise, Vol. 3, edited by F. Falkner and J.M. Tanner. New York: Plenum Press. Cecil, R.L. 1947. A textbook of medicine. 7th edition. Philadelphia: W.B. Saunder. Central Bureau o f Health Intelligence. 1981. Health Information of India-1981. New Delhi: CBHI, Directorate General o f Health Services, Ministry o f Health and Family Welfare. Central Bureau o f Health Intelligence (CBHI). 1991. Health Information of India- 1991. New Delhi: CBHI, Directorate General o f Health Services, Ministry o f Health and Family Welfare. Chandra, R.K. 1979. “Prospective studies o f the effect o f breast feeding on incidence of infection and allergy.” Acta Paediatria Scand 68:691-694. Chandra, R.K. 1980. Immunology of Nutritional Disorders. London: Edward Arnold. Chandra, R.K. 1997. “Nutrition and the immune system: an introduction.” American Journal of Clinical Nutrition 66:460S-463S. 274 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chang, K.S.F., M.N.C. Lee, and W.D. Low. 1965. “Standards o f heights and weights of Southern Chinese children.” Far Eastern Medical Journal 1:101-109. Chavez, A., and C. Martinez. 1979. “Consequences o f insufficient nutrition on child character and behavior.” In Malnutrition, Environment and Behavior, edited by D.A. Levitzky. Ithaca: Cornell University Press. Chen, C., and N.S. Scrimshaw. 1983. Diarrhea and Malnutrition. Interactions, Mechanisms, and Interventions. New York: Plenum Press. Conners, K.C., and A.G. Blouin. 1983. “Nutritional effects on behavior of children.” Journal of Psychiatric Research 17, 193-201. Cunningham, A. S. 1979. “Morbidity in breast-fed and artificially fed infants. II” Journal of Pediatrics 95:685-689 Da Vanzo, J., and D.S. Levy. 1989. Influences on Breastfeeding in Peninsular Malaysia. Rand Corporation, Santa Monica, California, Mimeographed. Dandekar, V., and N. Rath. 1971. “Poverty in India.” Economic and Political Weekly, 2 January. Datta Banik, N. D. 1982. “Semilongitudinal growth evaluation o f children from birth to 14 years in different socioeconomic groups.” Indian Paediatrics 19:353-359 Datta, B.N.D., K.R. Datta, S.I.S Mane, L. Raj, and A.D. Taskar. 1970. “A longitudinal study o f physical growth o f children from birth up to five years o f age in Delhi”. Indian Journal of Medical Research 58: 135-142. Delgado, H., V. Valverde, and R.E. Klien. 1983. “Analisis critico de la validez del perimetro del brazo como indicador del estado nutricional proteinico-energetico en ninos preescolares.” Arch. Lat. Nutr. 33:170-188. Demaris, A. 1992. Logit Modeling: Practical Applications. London: Sage Publications. Demaris, A. 1995. “A tutorial in logistic regression.” Journal of Marriage and the Family 57:956-968. Devadas, R. P., R. Rajalakshmi, and R. Kaviri. 1980. “Influence of family income and parent’s education on the nutritional status of pre-school children.” Indian Journal of Nutrition and Diet. 17:237-244. Dibley, M.J., J.B. Goldsby, N.W. Staehling, and F.L. Trowbridge. 1987a. “Development o f normalized curves for the international growth reference: Historical and technical considerations.” American Journal of Clinical Nutrition 46(5):736-748. 275 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dibley, M.J., N. Staehling, P. Nieburg, and F.L. Trowbridge. 1987. “Interpretation of Z-score anthropometric indicators derived from the international growth reference.” American Journal of Clinical Nutrition 46: 749-762. Dreze, J. 1995. “Famine prevention in India.” In The Political Economy of Hunger edited by J. Dreze, A. Sen and A. Hussain. Oxford: Caldreon Press. Dreze, J., and A. Sen. 1989. Hunger and Public Action. Oxford: Caldreon Press. Dreze, J., and A. Sen. 1995. India: Economic Development and Social Opportunity. Delhi: Oxford University Press. Dreze, J., and H. Gazdar. 1996. “Uttar Pradesh; The burden o f inertia.” In Indian Development, Select Regional Perspectives, edited by J. Dreze and A. Sen. Delhi: Oxford University Press. Esrey, S.A. 1993. “Multi-country study to examine relationships between the health and children and the level of water and sanitation service, distance to water, and type o f water used.” Report prepared for CIDA, W ater Sector, Ottawa. Esrey, S.A., and R.G. Feachem. 1989. Interventions for the control of diarrheal diseases among young children: Promotion of food hygiene. Geneva: World Health Organization. Eveleth, P. B., and T.M. Tanner. 1976. Worldwide variation in Human Growth. New York: Cambridge University Press. Eveleth, P.B., and J.M. Tanner. 1990. Worldwide Variation in Human Growth. New York: Cambridge University Press. FAO/WHO/UNU. 1985. Energy and Protein Requirements. Report of a Joint FAO/ WHO/UNU Expert Consultation. WHO Technical Report Series 724. Geneva: WHO. Farquharson, S.M. 1976. “Growth patterns and nutrition in Nepali children.” Archives of Diseases in Childhood 51:3-12. Ferguson, D. M., L J. Horwood, F.T. Shannon, and B. Taylor. 1981. “Breast-feeding, gastrointestinal and lower respiratory illness in the first two years.” Australian Paediatric Journal 17:191-195. Floud, R., K. Watcher and A. Gregory. 1990. Height, Health and History. New York: Cambridge University Press. Foster, S. 1984. Immunizable and respiratory diseases and child mortality. In Child survival: Strategies for research, edited by W. H. Mosley and L. C. Chen. Population and Development Review Supplement to volume 10. 276 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Francois, H., and L. Genevieve. 1979. Church and development in Kerala. Bangalore: Theological Publications o f India. Franke, R.W. 1993. Life is a little better: Redistribution as a development strategy in Nadur village, Kerala. Conflict and Social Change Series. Oxford: Westview Press. French, J.G. 1967. “Relationship o f morbidity to the feeding [patterns o f Navajo children from birth through twenty-four months.” American Journal of Clinical Nutrition 20: 375-384. Gam, S.M., and D.C. Clark. 1976. “Problems in the nutritional assessment o f Black individuals.” American Journal of Public Health 66:262-267. George, P.S. 1979. Public distribution of foodgrains in Kerala. IFPRI. Ghosh, A., U.R. Ananthamurthy, A. Beteille, S.M. Kansal, V. Mazumdar, and A. Vanaik. 1994. Evaluation of literacy campaigns in India. Report of an independent Expert Group appointed by the Ministry o f Human Resource Development, New Delhi: Ministry o f Human Resource Development. Gibson, R. S. 1990. Principles of nutritional assessment. New York: Oxford University Press. Golden, M.H.N. 1988. “The role of individual nutrient deficiencies in growth retardation o f children as exemplified by zinc and protein.” in Linear Growth Retardation in Less Developed Countries. New York: Raven Press. Goldstein, H., and J.M. Tanner. 1980. “Ecological considerations in the creation and use o f child growth standards.” Lancet 1:582. Gopalan, C. 1980. “ ‘Small is healthy?’ For the poor, not the rich.” Bulletin of the Nutrition Foundation of India 4(Oct): 4-8. Gopalan, C. 1983. “Measurement o f undemutrition: biological considerations.” Nutrition Foundation of India Bulletin 4. Gopalan, C. 1988. “Stunting: significance and implications for public health policy.” in Linear Growth Retardation in Less Developed Countries, edited by J.C. Waterlow. New York: Raven Press. Government o f India. 1993a. Sample Registration System: Fertility and mortality indicators 1991. New Delhi: Office o f Registrar General. Government o f India. 1994a. Annual Report 1993-94 (Part I) of the Department of Education. New Delhi: Ministry o f Human Resource Development. Government o f Kerala. 1992. “Economic Review, 1991.” Thiruvananthapuram: State Planning Board. 277 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Government o f Kerala. 1992. State Planning Board, Economic Review, 1991, Thiruvananthapuram . 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 developing countries.” In Diet and Disease edited by G.A. Harrison and J.C. Waterlow. New York: Cambridge University Press. Grantham-McGregor, S.M., and E. Buchanan. 1982. “The development o f an adopted child recovering from severe malnutrition-a case report.” Human Nutrition'. Clinical Nutrition 36C, 251-6. Grantham-McGregor, S.M., W. Schofield, and C. Powell. 1987. “Development of severely malnourished children who received psychosocial stimulation: six year follow-up.” Pediatrics 79, 247-54. Greulich, W.W. 1957. “A comparison of the physical growth and development of American-born and native Japanese children.” American Journal of Physical Anthropology 15:489-516. Greulich, W.W. 1976. “Some secular changes in the growth o f American-born and native Japanese children.” American Journal of Physical Anthropology 45:553- 568. Gribble, J.N. 1993. “Birth intervals, gestational age and low birth weight: Are the relationships confounded?” Population Studies 47: 133-146. Guha, B.S. 1944. Racial affinities of the people of India. London: Oxford University Press. Gulati, L. 1977. “Rationing in a peri-urban community. Case study o f a squatter habitat.” Economic and Political Weekly March 19, 501-6. Gurney, J.M. 1979. “The young child: protein-energy-malnutrition.” In Human Nutrition: a comprehensive treatise, Vol. 2. Nutrition and Growth. New York: Plenum Press. Gupta, D.B., A. Basu, and A. Asthana. 1993. Population change, women’ s role and status and development in India: A review. Mimeo, Institute o f Economic Growth, Delhi University. Haas, J.D., and J.P. Habicht. 1990. “Growth and Growth Charts in the Assessment of pre-school nutritional status.” In Diet and Disease, edited by G.A. Harrison and J.C. Waterlow. New York: Cambridge University Press. Habicht, J.P., and D.L. Pelletier. 1990. “The importance of context in choosing nutritional indicators.” Journal of Nutrition 35: 1241-1254. 278 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Habicht, J.P., R. Martorell, C. Yarbrough, R.M. Malina, and R.E. Klein. 1974. “Height and weight standards for preschool children. How relevant are ethnic differences in growth potential?” Lancet I: 611-615. Haines, M.R., and R.C. Avery. 1982. “Differential infant and child mortality in Costa Rica: 1968-1973.” Population Studies 36: 31-44. Hamill, P.V.V., F.E. Johnston, and W. Grams. 1970. Height and Weight of Children: United States, Vital Health Statistics Series 11, No. 104, Washington. D.C: U.S. Government Printing Office. Haque, T., and A.S. Sirohi. 1986. Agrarian reforms and institutional changes in India. New Delhi: Concept. Harriss, B. 1990. “Food distribution, death and disease in South Asia.” In Diet and Disease edited by G.A. Harrison and J.C. Waterlow. New York: Cambridge University Press. Hasan, Z. 1989. “Power and mobilization: Patterns o f resilience and change in Uttar Pradesh politics.” In Dominance and state power in modern India: Decline of a social order, by Frankel and Rao, 2 volumes. Delhi: Oxford University Press. Hauspie, R.C., S. R. Das, M.A. Preece, and J.M. Tanner. 1980. “A Longitudinal Study of the Growth in Height o f Boys and Girls o f West Bengal (India) Aged Six Months to 20 Years.” Annals of Human Biology 7 (5): 429-441. Health Monitor. 1994. New Delhi: Foundation for research in health systems. Hoyle, B., M. Yunus and L.C. Chen. 1980. “Breastfeeding and food intake among children with acute diarrheal disease.” American Journal of Clinical Nutrition 33:2365-2371. Huffman, S.L., and A. Steel. 1995. Do child survival interventions reduce malnutrition? The dark side of child survival. In Child growth and nutrition in developing countries, edited by P. Pinstrup-Andersen, D. Pelletier and H. Alderman. Ithaca: Cornell University Press. Hurtado, E.K., A.H. Claussen and K.G. Scott. 1999. “Early childhood anemia and mild or moderate mental retardation.” American Journal of Clinical Nutrition 69: 115-119. IIPS (International Institute o f Population Sciences). 1995. National Family Health Survey (MCH and Family Planning), India 1992-93. Bombay: IIPS. Imong, S.M., K. Rungruengthanakit, C. Ruangyuttikam, L. Wongsawasdi, D.A. Jackson, and R.F. Drewett. 1989. “The bacterial content o f infant weaning foods and water in rural northern Thailand.” Journal of Tropical Pediatrics 35: 14-18. 279 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Janes, M. D. 1974. “Physical growth o f Nigerian Yoruba children”. Tropical Geographic Medicine 26:389. Jeffrey, R. 1976a. “Temple entry movement in Travancore, 1860-1940.” Social Scientist March. Jeffrey, R. 1976b. The decline ofNayar dominance: society and politics in Travancore, 1847-1908. Delhi: Vikas Publishing House. Jeffrey, R. 1987. “Culture and governments: how women made Kerala literate.” Pacific Affairs 60, 4, Fall. Jeffrey, R. 1992. Politics, women and well-being: How Kerala became a model. London: Macmillan. Jelliffe, D.B., and E.F.P. Jelliffe. 1978. “The volume and composition o f human milk in poorly nourished communities: A review.” American Journal o f Clinical Nutrition 31:492-515. Jhonston, F. E., M. Borden, and R.B. MacVean. 1973. “Height, weight, and growth velocities in Guatemalan private school children o f high socioeconomic class.” Human Biology 45:627-641. Johnson, F.E., H. Wainer, D. Thissen, and R. MacVean. 1976. “Hereditary and environmental determinants o f growth in height in a longitudinal sample of children and youth of Guatemalan and European ancestry.” American Journal Physical Anthropology 44:469-475. Kananeeh, H. 1972. “The relationship o f bottle feeding to malnutrition and gastroenteritis in a pre-industrial setting.” Journal of Tropical Pediatrics 18:302- 306. Kannan, K.P. 1988. Of rural proletarian struggles. Delhi: Oxford University Press. Kannan, K.P. 1993. Public intervention and poverty alleviation: A study of the declining incidence of poverty in Kerala. Thiruvananthapuram: Centre for Development Studies. Keller, W., G. Donoso, and E.M. De Maeyer. 1976. “Anthropometry in nutritional surveillance: a review based on results of the WHO collaborative stud on nutritional anthropometry.” Nutrition Abstracts and Reviews 46: 591-609. Khan, M.E., S.K. Ghosh, Dastidar, and R. Singh. 1986. “Nutrition and health practices among the rural women: A case study o f Uttar Pradesh.” Journal of Family Welfare 33 (2). Khan, M.M.H., and M. Kabir. 1997. “Differentials in infant mortality in Bangladesh by selected background characteristics: A life table approach.” The Journal of Family Welfare 43: 62-70. 280 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kielmann, A.A., C.E. Taylor, and C. De Sweemer. 1983. Child and Maternal Health Services in Rural India: The Narangwal Experiment. Vol. 1: Integrated Nutrition and Health Care. Baltimore: Johns Hopkins University Press. Kim, Y.S. 1982. “Growth status of Korean schoolchildren in Japan” . Annals of Human Biology 9:453-458. King, E.M. 1990. Educating girls and women: Investing in Development. Washington D.C: World Bank. King, E.M., and M.A. Hill. 1993. Women’ s education in developing countries: Barriers, benefits and policies. Baltimore: Johns Hopkins University Press. King, K. W., J. Foucauld, W. Fougera, and E.L. Severinghaus. 1963. “Height and weight of Haitian children.” American Journal of Clinical Nutrition 13:106-109. Knodel, J., and M. Wongsith. 1991. “Family size and children’s education in Thailand: Evidence from a national sample.” Demography 28( 1): 119-131. Kohli, A. 1987. The state and poverty in India: The politics of reform. Cambridge: Cambridge University Press. Koshy, A., A.A. Gopalakrishnan, V. Vijayachandran, and N.K. Jayakumar. 1989. Report of the study on the evaluation of the public distribution system in Kerala. Department o f Civil Supplies, Ministry o f Food and Civil Supplies, Government of India, Centre for Management Development, Thiruvananthapuram. Kramer, M.S. 1987. “Determinants o f low birth weight: Methodological assessment and meta-analysis.” Bulletin of the World Health Organization 65 (5): 663-737. Lasky, R.E., A. Lechtig, H. Delgado, R.E. Klein, P. Engle, C. Yarbrough, and R. Martorell. 1975. “Birth weight and psychomotor performance in rural Guatemala”. American Journal of Diseases in Children, 129, 566-70. Latham, C.L. 1997. Human Nutrition in the Developing World. Rome: FAO. Lauber, E., and M. Reinhardt. 1979. “Studies on the quality o f breast milk during 23 months of lactation in a rural community o f Ivory Coast.” American Journal of Clinical Nutrition 32:1159-1173. Levinson, F.J. 1974. Morinda: An economic analysis of malnutrition among young children in rural India. Cambridge: Cornell-MIT International Nutrition Policy Series. MacDonald, P.C., and J.A. Pritchard. 1980. Williams Obstetrics. Sixteenth Edition. New York: Appleton-Century-Crofts. 281 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Maharatna, A. 1995. “Children’s work, activities, surplus labor and fertility: a case study of peasant households in Birbhum, West Bengal.” Paper presented at the second workshop on Applied Development Economics, 6-10 January 1996, Centre for Development Economics at the Delhi School o f Economics. Mahmood, T.A., D.M. Campbell and A.W. Wilson. 1988. “Maternal height, show size, and outcome of labor in white primigravidas: A prospective anthropometric study.” British Medical Journal 297 (6647, August 20-27):515-517. Malina, R. 1984. “Physical activity and m otor development/performance in populations nutritionally at risk”. In Energy Intake and Activity, pp. 285-302, edited by E. Pollitt and P. Amante. New York: A.R. Liss. Martorell R. 1985. “Child growth retardation: a discussion o f its causes and its relationship to health.” In Nutritional Adaptation in Man, edited by K.L. Blaxter and J.C. Waterlow. London: John Libbey. Martorell, R. 1980. “Interrelationships between diet, infectious disease and nutritional status.” In Social and biological predictors of nutritional status, physical growth and nutritional development, edited by L.S. Greene and F. Johnston. New York: Academic Press. Martorell, R. 1984. “Genetics, Environment and Growth: Issues in the Assessment of Nutritional Status.” In Genetic Factors in Nutrition, edited by L.S. Greene and F. Johnston. New York: Academic Press. Martorell, R. 1995. “Promoting Healthy Growth: Rationale and Benefits.” In Child Growth and Nutrition in Developing countries, edited by P. Pinstrup-Andersen, D. Pelletier and H. Alderman. Ithaca: Cornell University Press. Martorell, R., and J.P. Habicht. 1986. “Growth in early childhood in developing countries.” in Human Growth: A Comprehensive Treatise, Voluntary. 3, pp 241- 262, F. Falkner and J.N. Tanner. New York: Plenum Press. Martorell, R., and C. Yarbrough. 1983. “The energy cost o f diarrheal diseases and other common illnesses in children.” In Diarrhea and Malnutrition: Interactions, Mechanisms and Interventions, pp. 125-141, edited by L.C. Chen and N. S. Scrimshaw. New York: Plenum Press. Martorell, R., C. Kendall, and D.R. Foote. 1983. “Growth and poverty in Honduras.” Paper presented at the Fifty-Second Annual Meeting of the American Association of Physical Anthropologists, Indianapolis, Indiana. Martorell, R., C. Yarbrough, S. Yarbrough, and R.E. Klein. 1980. “The impact of ordinary illnesses on the dietary intakes o f malnourished children.” American Journal of Clinical Nutrition 33:345-350. Martorell, R., H.L. Delgado. V. Valverde, and R.E. Klein. 1981. “Maternal Stature, Fertility and Infant Mortality.” Human Biology 53. 282 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Martorell, R., J. Leslie, and P.R. Moock. 1984. “Characteristics and determinants o f child nutritional status in Nepal.” American Journal of Clinical Nutrition 39:74- 8 6 . Martorell, R., J. Rivera, and H. Kaplowitz. 1990. “Consequences o f Stunting in Early Childhood for Adult Body Size in Rural Guatemala.” Annales Nestle’ 48: 85-92. M artorell, R., J. Rivera, H. Kaplowitz, and E. Pollitt. 1992. “Long-term consequences of growth retardation during early childhood.” In Human Growth: Basic and Clinical Aspects, edited by M. Hernandez and J. Argente. New York: Elsevier Science Publishers B.V. Martorell, R., J.P. Habicht, C. Yarbrough, A. Lechtig, R.E. Klein, and D. Western. 1975. “Acute morbidity and physical growth in rural Guatemalan children.” American Journal of Diseases of Children 129: 1296-1301. M artorell, R., J.P. Habicht, R.E. Klein. 1982. “Anthropometric indicators o f changes in nutritional status in malnourished populations.” In Proceedings Methodologies for Human Population Studies in Nutrition Related to Health, pp.96-110, NIH Publ. No. 82-2462B, edited by Underwood. Washington, D.C: U.S. Government Printing Office. Mata, L. 1978. The children of Santa Maria Cauque. A Prospective Field Study of Health and Growth. Cambridge: MIT Press. Mateer, S. 1871. The land of charity: A descriptive account of Travancore and its people, with especial reference to missionary labour. London: John Snow and Co. McCullagh, P. 1980. “Regression models for ordinal data.” Journal of the Royal Statistical Society, Series B, 42: 109-142. Mencher, J. 1980. “The Lessons and Non-Lessons o f Kerala: Agricultural Labourers and Poverty.” Economic and Political Weekly, Special Number, XV, 41-3. Ministry of Health and Family Welfare (MOHFW). 1991. Family welfare programme in India , Year book 1989-90. New Delhi: Department o f Family Welfare, MOHFW. Ministry of Health and Family W elfare (MOHFW). 1992B. Module for health workers. New Delhi: National child survival and safe motherhood program, MOHFW. Mishra, V.K, and R.D. Retherford. 2000. “W omen’s education can improve child nutrition in India.” NFHS Bulletin Number 15, February. 283 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Molla, A.M., A. Molla, S.A. Sarkar, and M. Mujibur Rahaman. 1983. “Food intake during and after recovery from diarrhea in children.” In Diarrhea and Malnutrition. Interactions, Mechanisms and Interventions, edited by L. C. Chen and N.S. Scrimshaw. New York: Plenum Press. Montgomery, M.R., M. Gragnolati, K. Burke, and E. Paredes. 1999. Measuring living standards with proxy variables. New York: Population Council, No. 129 Mooij, J. 1999. Food policy and the Indian state. Delhi: Oxford University Press. Mora, J.O., S.G. Sellers, J. Suescun, and M.G. Herrera. 1981. “The impact of supplementary feeding and home education on physical growth o f disadvantaged children.” Nutrition Reviews 1:213-225. Morley, D. 1976. “Measles and whooping cough.” In Epidemiology and Community Health in Warm Climate Countries edited by R. Cruikshank, K.L. Standard and H.B.L. Russell. Mukerji. B., and K.K. Kaul. 1970. “Anthropometric observations on urban primary school children.” Indian Journal of Medical Research 58, 1257-1271. Munro, H. N. 1985. “Historical perspective on protein requirements: objectives for the future.” In Nutritional Adaptation in Man edited by K.L. Blaxter and J.C. Waterlow. London: John Libbey. Munro, H.N., and M. Crim. 1980. The proteins and amino acids, in Modern Nutrition in Health Care in Developing Countries. New York: Oxford University Press. Murthi, M., A.C. Guio, and J. Dreze. 1996. “Mortality, fertility and gender bias in India: A district-level analysis.” In Indian Development, Select Regional Perspectives, edited by J. Dreze and A. Sen. Delhi: Oxford University Press. Nag, M. 1985. “The impact of social and economic development on mortality: Comparative study o f Kerala and West Bengal.” in "Good health at low cost ” edited by S.B. Halstead, J.A. Walsh, K.S. Warren. Proceedings o f a conference held at the Bellagio Conference Center, Bellagio, Italy. New York: Rockefeller Foundation. Nag, M. 1989. “Political awareness as a factor in accessibility o f health services: A case study o f rural Kerala and West Bengal.” Economic and Political Weekly 24, 8. Nair, P.R.G. 1981. Primary Education, Population Growth and Socio-Economic Change. New Delhi: Allied Publishers Private Limited. Nayanar, E.K. 1982. My struggles: An autobiography. New Delhi: Vikas. NCERT (National Council o f Educational Research and Training). 1992. Fifth All- India Educational Survey, 2 volumes. New Delhi: NCERT. 284 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Nuna, S.C. 1990. Women and Development. New Delhi: National Institute of Educational Planning and Administration. Nutrition Canada. 1980. Anthropometry report: height, weight, and body dimensions. Bureau of Nutritional Sciences, Health Protection Branch, Health and Welfare, Ottawa. Osmani, S.R. 1992. “On some controversies in the measurement of undemutrition.” In Nutrition and Poverty, edited by S.. Osmani. Oxford: Caldreon Press. Pacey, A., and P. Payne, (eds.) 1985. Agricultural Development and Nutrition. By arrangement with FAO and UNICEF. London: Hutchinson. Panikkar, K.N. 1989. Against lord and stat: Religion and peasant uprisings in Malabar, 1836-1921. New Delhi: Oxford University Press. Park, J.E., and K. Park. 1989. Textbook of Preventive and Social Medicine. Twelfth Edition. Jabalpur: M/S Banarsidas Bhanot Publishers. Payne, P. 1987. “Malnutrition and human capital: Problems o f theory and practice.” In Poverty, Development and Food, edited by E. Clay and J. Shaw. London: Macmillan. Payne, P. R. 1975. “Safe protein-calorie ratios in diets. The relative importance of protein and energy intake as causal factors in malnutrition.” American Journal of Clinical Nutrition 28:281-286. Payne, P.R. 1985. “Nutritional Adaptation in Man: Social Adjustments and their Nutritional Implications.” In Nutritional Adaptation in Man edited by K.L. Blaxter and J.C. Waterlow. London: John Libbey. Payne, P.R., and P. Cutler. 1984. “Measuring malnutrition: technical problems and ideological perspectives.” Economic and Political Weekly , 25 August. Pelletier, D.L. 1994. “The Relationship Between Child Anthropometry and Mortality in Developing Countries: Implications for Policy, Programs, and Future Research.” American Journal of Clinical Nutrition 124: 2047S-2081S. Pelletier, D.L., E.A. Frongillo Jr., D.G. Shroeder, and J.P. Habicht. 1994b. “A methodology for estimating the contribution o f malnutrition to child mortality in developing countries.” Journal of Nutrition 2106s-2122s. Pollitt, E., A.G. Soemantri, F. Yunis, and N.S. Scrimshaw. 1985. “Cognitive effects of iron-deficiency anemia.” Lancet I, 158. Pollitt, E., R. Leibel, and D. Greenfield. 1981. “B rief fasting, stress and cognition in children.” American Journal of Clinical Nutrition 34, 1526-33. 285 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Prakash, B.A. 1994. “Kerala’s economy: An overview.” In Kerala’ s Economy, edited by B.A. Prakash. New Delhi: Sage Publications. Radhakrishnan, P. 1989. Peasant struggles, land reforms and social change: Malabar, 1936-82. New Delhi: Sage Publications. Raftery, A.E. 1995. Bayesian Model Selection in Social Research. Sociological Methodology. Cambridge: Blackwells. Ramachandran, V.K. 1996. “On Kerala’s Development Achievements.” In Indian Development, Select Regional Perspectives, edited by J. Dreze and A. Sen. Delhi: Oxford University Press. Rao, D. H., K. Satyanarayana, and J.G. Sastry. 1976. “Growth pattern o f well-to-do Hyderabad pre-school children.” Indian Journal of Medical Research 64:629-638. Rao, K. S., C.C. Swaminathan, S. Swarup, and V.N. Patwardham. 1959. Protein malnutrition in South India, Bulletin of the World Health Organization 20:603- 639. Rea, J. N. 1971. “Social and economic influences on the growth o f pre-school children in Lagos.” Human Biology 43:46-63. Roberts, D.F. 1985. “Genetics and Nutritional Adaptation.” In Nutritional Adaptation in Man, edited by K.L. Blaxter and J.C. Waterlow. London: John Libbey. Rosenweig, M.R., and T.P. Schultz. 1882. “Child Mortality and Fertility in Columbia: Individual and Community Effects.” Health Policy and Education, 2 (3-4): 305- 348. Rowland, M.G.M., R.A.E. Barrel], and R.G. Whitehead. 1978. “Bacterial contamination in traditional Gambian weaning foods.” Lancet 1:136-138. Rutishauser, I.H.E., and R.G. Whitehead. 1972. “Energy intake and expenditure in 1- 3 year old Ugandan children living in a rural environment.” British Journal of Nutrition 28, 145-52. Sameroff, A.J., and S.C. McDonough. 1984. “The role o f motor activity in human cognitive and social development.” In Energy Intake and Activity, pp. 331-53, edited by E. Pollitt and P. Amante. New York: A.R. Liss. SAS Institute Inc. 1995. Logistic Regression Examples Using the SAS System. Cary: SAS Institute Inc. SAS Institute Inc. 1989. SAS/STAT User’ s Guide, Version 6, Fourth Edition, Volumes 1 and 2. Cary: SAS Institute Inc. 286 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sastry, J.G., and K. Vijayaraghavan. 1973. “Use o f anthropometry in grading malnutrition in children.” Indian Journal of Medical Research 61:1225. Satyanarayana, K., A.N. Naidu, and B.S.N. Rao. 1978. “Nutrition, physical work capacity and work output.” Indian Journal of Medical Research 68 (suppl.): 88- 93. Satyanarayana, K., A.N. Naidu, and B.S.N. Rao. 1979. “Nutritional Deprivation in Childhood and the Body Size, Activity and Physical Work Capacity o f Young Boys.” American Journal of Clinical Nutrition 32. Satyanarayana, K., A.N. Naidu, B. Chatterjee and B.S.N. Rao. 1977. “Body Size and Work Output.” American Journal of Clinical Nutrition 30. Satyanarayana, K., A.N. Naidu, M.C. Swaminathan, and B.S. Narasinga Rao. 1980. “Adolescent Growth Spurt among Rural Indian Boys in Relation to Their Nutritional Status in Early Childhood.” Annals of Human Biology 1 (4): 359-366. Satyanarayana, K., T. P. Krishna, and B.S.N. Rao. 1986. “Effect o f early childhood undemutrition and child labour on growth and adult nutritional status o f rural Indian boys around Hyderabad.” Human Nutrition: Clinical Nutrition 40C: 131-40. Scholl, T.O., M.L. Hediger, J.I. Schall, C.S. Khoo and R.L. Fischer. 1994. “Maternal growth during pregnancy and the competition for nutrients.” American Journal of Clinical Nutrition 60: 183-188. Schultz, T.P. 1993a. Investments in Schooling and Health o f Women and Men: Quantities and Returns. New Haven: Yale University. Mimeographed. Schultz, T.P. 1993b. “Returns to W omen’s education.” In Women’ s education in developing countries: Barriers, benefits and policies. Baltimore: Johns Hopkins University Press. Scrimshaw, N.S., and B.A. Underwood. 1980. “Timely and appropriate complementary feeding of the breastfed infant: An overview.” Food and Nutrition 2:19-22. Scrimshaw, N.S., C.E. Taylor, and J.E. Gordon. 1968. “Interaction o f Nutrition and Infection.” World Health Organization, Monograph Series 57, Geneva. Seckler, D. 1980. “Malnutrition: an intellectual odyssey.” Western Journal of Agricultural Economics: 5. Reprinted in Newer concepts in nutrition and their implications for policy edited by P.V. Sukhatme 1982. Pune: Maharashtra Association for Cultivation o f Science. Seckler, D. 1982. “Small but Healthy: A Basic Hypothesis in the Theory, Measurement and Policy o f M alnutrition.” In Newer concepts in nutrition and their implications for policy edited by P.V. Sukhatme. Pune: Maharashtra Association for Cultivation o f Science. 287 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Seckler, D. 1984a. “The ‘small but healthy?’ hypothesis: a reply to critics.” Economic and Political Weekly 3. Seckler, D. 1984b. “The ‘ small but healthy?' hypothesis: an invitation to a refutation." In Impact of P. V . Sukhatme on Agricultural Statistics and Nutrition, edited by P. Narain. New Delhi: Indian Society of Agricultural Statistics. Sen, A. 1977. “Starvation and exchange entitlements: A general approach and its application to the great Bengal famine.” Cambridge Journal of Economics 1. Sen, A. 1996. “Radical needs and moderate reforms.” In Indian Development, Select Regional Perspectives, edited by J. Dreze and A. Sen. Delhi: Oxford University Press. Shankar, K. 1991a. “Politics o f land distribution in Uttar Pradesh.” Economic and Political Weekly 27 April. Sharma, H.R. 1994. “Distribution o f Landholdings in Rural India, 1953-54 to 1981- 82: Implications for Land Reform.” Economic and Political Weekly 24 September. Shekar, M., J.P. Habicht, and M.C. Latham. 1991. “Is positive deviance in growth the converse o f negative deviance?” Food and Nutrition Bulletin 13: 7-11. Sikri, S.D. 1972. “A comparative study o f height and weight of government and public school children o f Punjabi population.” Indian Journal of Medical Research 60, 491-500. Simeon, D., and S.M. Grantham-McGregor. 1987. “Cognitive function, undemutrition and missed breakfast.” Lancet II, 737-8. Singh, B., and S. Misra. 1964. A Study of Land Reforms in Uttar Pradesh. Honolulu: East-West Center. Sinha, A. 1995. Village Visit Reports, compendium of selected field reports of the 60fh Foundation Course. Mussoorie: Lai Bahudur Shastri National Academy o f Administration. Sommerfelt, A.E., and M.K. Stewart. 1994. Children’ s Nutritional Status. DHS Comparative Studies No. 12. Calverton: Macro International. Sonalde, D., and A. Soumya. 1998. “Maternal Education and child Health: Is there a strong causal relationship?” Demography 35(1):71-81. 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. 288 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Spurr, G.B., M.G. Maksud, and M. Barac-Nieto. 1977a. “Energy expenditure, productivity, and physical work capacity o f sugar cane loaders.” American Journal of Clinical Nutrition 30, 1740-6. Steckel, R. H. 1987. “Growth depression and recovery: the remarkable case of American slaves.” Annals of Human Biology 14:111-32 Stephenson, L.S., M.C. Latham, and A. Jansen. 1983. “A comparison of growth standards: similarities between NCHS, Harvard\ Denver and privileged African children and differences with Kenyan rural children.” Cornell International Nutrition Monograph Series Number 12. Sukhatme, P. V. (ed.) 1982d. Newer concepts in nutrition and their implications for policy. Pune: Maharashtra Association for Cultivation o f Science. Sukhatme, P. V. 1978. “Assessment of adequacy o f diets at different income levels.” Economic and Political Weekly Special number, August. Sukhatme, P. V., and S. Margen. 1982. “Autoregulatory homeostatic nature o f energy balance.” American Journal of Clinical Nutrition 35: 355-65; reprinted in Sukhatme (1982d). Swaminathan, M.C. 1971. “Observation on growth and development of undernourished children o f Andhra Pradesh.” Proceedings of the Nutrition Society of India: Nutrition, Growth and Development N o.10:149-155. Tanner, J. M. 1962. Growth at Adolescence. 2n d edition. Oxford: Blackwell. Tanner, J.M. 1966. “Galtonean eugenics and the study o f growth.” Eugenics Review 58:122-35. Tanner, J.M. 1978. “Foetus into man: physical growth from conception to maturity.” Cambridge: Harvard University Press. Tanner, J.M. 1982. “The potential o f auxological data for measuring economic and social well-being.” Social Science History 6: 571-81. Tanner, J.M., T. Hayashi, M.A. Preece, and N. Cameron. 1982. “Increase in length of leg relative to trunk in Japanese children and adults from 1957 to 1977: Comparisons with British and Japanese Americans.” Annals of Human Biology 9:411-423. Thomas, D., J. Strauss, and M.H. Henriques. 1990. “Child Survival, eight for Age, and Household Characteristics in Brazil.” Journal of Development Economics 33 (2): 197-234. Thomas, D., J. Strauss, and M.H. Henriques. 1991. “How Does M other’s Education Affect Child Height?” Journal of Human Resources 26 (2): 183-211. 289 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Tyagi, P.N. 1993. Education for all: A Graphic Presentation. 2n d edition. New Delhi: National Institute o f Educational Planning and Administration. Udani, P.M. 1963. “Physical growth o f children in different socio-economic groups in Bombay.” Indian Journal of Child Health, 12, 593-611. Unicef. 1990. Strategy for Improved Nutrition of Women and Children in Developing Countries. New York: United Nations Children’s Fund. Unicef. 1993. State of the World’ s Children. New York: United Nations Children’s Fund. United Nations Development Program Staff. 1999. Human Development Report. New York: Oxford University Press. United Nations. 1975. Poverty, unemployment, and development policy. New York: United Nations. United Nations. 1986. How to weigh and measure children: Assessing the nutritional status of young children in household surveys. Department of technical Co operation for development o f statistical office, United Nations. New York: UN. Valverde, V., R. Martorell, V. Mejia-Pivaral, H. Delgado, A. Zechtig, C. Teller, and R.E. Klein. 1977. “Relationship between family land availability and nutritional status.” Ecology of Food and Nutrition 6:1-7. Valverde, V., V. Mejia-Pivaral, H. Delgado, J. Belizan, R.E. Klein, and R. Martorell. 1981. Income and growth retardation in poor families with similar living conditions in rural Guatemala. Ecology of Food and Nutrition 10:241-248. Victora, C.G., J.P. Vaughan, B. Kirkwood, J.C. Martines and L.B. Barcelos. 1985. “Child malnutrition and land ownership in southern Brazil.” Ecology of Food and Nutrition 18: 265-275. Victora, G. 1991. The association between wasting and stunting: An international perspective. American Institute o f Nutrition. Vijay Raghavan, K.D., D. Singh, and M.C. Swaminathan. 1971. “Heights and weights of well-nourished Indian school children.” Indian Journal of Medical Research 59:648-654. Villarejos, V.M., J.A. Osborne, F.J. Payne, and G.J.A. Arguedas. 1971. “Heights and weights o f children in urban and rural Costa Rica.” Journal of Tropical Pediatric Environment and Child Health 17:31-43. Viteri, F.E., and B. Torun. 1981. “Nutrition, Physical activity, and growth.” In The Biology of Normal Human Growth, pp. 265-73, edited by M. Ritzen, A. Aperia, K. Hall, A. Larsson, and A. Zetterstorm. New York: Raven Press. 290 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wadley, S.S., and B.W. Derr. 1989. “Karimpur 1925-1984: Understanding rural India through restudies.” In Conversations Between Anthropologists and Economists, edited by P. Bardhan. Delhi: Oxford University Press. Waterlow J.C. 1972. “Classification and definition o f protein calorie malnutrition.” Annals of Human Biology 3: 229-233. Waterlow, J.C. 1976. “Classification and definition o f protein-calorie malnutrition.” in Nutrition in preventive medicine.” British Medical Journal 3:566-569. Waterlow, J.C. 1988. “Observations on the natural history o f stunting.” In Linear Growth Retardation in Less Developed Countries, edited by J.C. Waterlow. New York: Raven Press. Waterlow, J.C. 1992. “Nutrition and growth.” In Protein Energy Malnutrition with contributions by A.M. Tomkins and S.M. Grantham-McGregor. New York: Edward Arnold, a division of Hodder and Stoughton. Waterlow, J.C., R. Buzina, W. Keller, J.M. Lane, M.Z. Nichaman, and J.M. Tanner, 1977. “The Presentation and Use o f Height and Weight Data for Comparing the Nutritional Status of Groups o f Children under the Age o f 10 years.” Bulletin of the World Health Organization 55(4): 489-498. Waterlow, J.S. 1993. “Reflections on Stunting.” In Recent Trends in Nutrition, edited by C. Gopalan. Delhi: Oxford University Press. Wheeler, E., and S.P. Tan. 1983. “Trends in growth o f ethnic Chinese children living in London.” Annals of Human Biology 10:441-446. Whitehead, R.G. 1977. “Protein and energy requirements o f young children living in the developing countries to allow for catch-up growth after infections.” American Journal of Clinical Nutrition 30: 1545-7. WHO (World Health Organization) Working Group. 1986. “Use and interpretation o f anthropometric indicators of nutritional status.” Bulletin of the World Health Organization 64: 929-941. WHO. 1978. A growth chart for international user in maternal and child health care. Geneva: World Health Organization. WHO. 1995. Physical status: the use and interpretation of anthropometry. World Health Organization. Wolfe, B.L., and J.R. Behrman. 1987. “W omen’s schooling and children’s health: Are the effects robust with adult sibling control for women’s childhood background?” Journal of Health Economics 6 (3):239-254. World Bank. 1993. World Development Report: Investing in Health. W ashington D.C.: World Bank. 291 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. World Health Organization. 1979. Measurement of Nutritional Impact, pp.85. Geneva: WHO. World Health Organization. 1983. Measuring Change in Nutritional Status. Geneva: WHO. World Health Organization. 1994. Care of the mother and the baby at the health center: a practical guide. Geneva: Maternal Health and Safe Motherhood program, Division o f Family Health. World Health Organization-UNICEF. 1992. Low Birth Weight: A Tabulation of Available Information. Geneva: World Health Organization. Yarbrough, C., J.P. Habicht, R.M. Malina, A. Lechtic, and R.E. Klein. 1975. “Length and Weight in rural Guatemalan Ladino children: Birth to seven years o f age.” American Journal of Physical Anthropology 42:439-447. Yee, V. S-Y. 1984. “Household level o f correlates of child nutritional status in Fiji.” M.P.S. Thesis, Cornell University. Zacharia, K.C. 1984. The Anomaly of the Fertility Decline in India’ s Kerala State: A Field Investigation. Staff Working Paper No. 700. Washington, D.C: The World Bank. 292 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. A ppendix 1 D ichotom ized logistic regression coefficients and sta n d a rd e rro rs show ing the effect of proxim ate and m atern al and household covariates on w eight-for-age in In d ian children 1-4 years old, 1992-93 S everely U nderw eight -O th e rs N o rm a l-O th e rs Param eter Standard P Param eter Standard P Variable Estim ate Error value Estim ate Error value INTERCPT -1.7474 0.0917 0.0001 -0.2617 0.0726 0.0003 Proximate Determinants Age o f the child -0.2788 0.0322 0.0001 -0.2238 0.0285 0.0001 Number o f prenatal visits -0.0471 0.0092 0.0001 -0.0023 0.0083 0.7824 Delivery at home 0.2221 0.0669 0.0009 0.3205 0.0500 0.0001 Delivery at government hospital 0.1486 0.0715 0.0377 0.1858 0.0520 0.0004 Small size at birth 0.7889 0.0534 0.0001 0.7288 0.0467 0.0001 Medium size at birth 0.4313 0.0470 0.0001 0.3422 0.0375 0.0001 Had fever during 2 weeks before the survey 0.1980 0.0354 0.0001 0.2416 0.0339 0.0001 Underlving Maternal and Household Determinants Number o f years mother's schooling -0.0479 0.0058 0.0001 -0.0361 0.0047 0.0001 Number o f years father’s schooling -0.0121 0.0045 0.0066 -0.0012 0.0042 0.7708 M other’s lowage at first birth 0.1012 0.0327 0.0020 0.2654 0.0289 0.0001 Household has no toilet facility -0.0762 0.0177 0.0001 -0.1197 0.0155 0.0001 House is kachha 0.2960 0.0440 0.0001 0.2503 0.0363 0.0001 Ownership o f agricultural land 0.2660 0.0328 0.0001 0.1941 0.0301 0.0001 Prenatal care * father's education -0.0018 0.0013 0.1605 -0.0053 0.0009 0.0001 N > *o u> Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. A ppendix 2 D ichotom ized logistic regression coefficien ts and standard errors show ing the effect o f proxim ate and m aternal and household covariates on height-for-age in Indian children 1-4 years old, 1992-93 S ev erely S tu n ted -O th ers N o r m a l-O th e r s Param eter Standard P Param eter Standard P Variable Estim ate E rror value Estimate Error value INTERCPT -1.0583 0.0817 0.0001 -0.0722 0.0746 0.3329 Proximate Determinants Age o f the child 0.3337 0.0339 0.0001 0.3479 0.0342 0.0001 Number o f prenatal visits -0.0419 0.0103 0.0001 -0.0117 0.0096 0.2229 Delivery at home 0.3141 0.0506 0.0001 0.3204 0.0451 0.0001 Small size at birth 0.3170 0.0588 0.0001 0.4104 0.0560 0.0001 Medium size at birth 0.2521 0.0500 0.0001 0.2511 0.0463 0.0001 Underlying Maternal and Household Determinants Number o f years mother’s schooling -0.0579 0.0060 0.0001 -0.0560 0.0053 0.0001 Mother’s lowage at first birth 0.0979 0.0349 0.0050 0.2166 0.0336 0.0001 Ownership of agricultural land -0.1092 0.0187 0.0001 -0.1031 0.0176 0.0001 Household has no toilet facility 0.1318 0.0453 0.0037 0.1011 0.0423 0.0169 House is kachha 0.1083 0.0349 0.0019 0.0944 0.0348 0.0067 Prenatal care * m other’s education -0.0052 0.0012 0.0001 -0.0046 0.0010 0.0001 to VO Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. A ppendix 3 D ichotom ized logistic regression coefficien ts and standard errors show ing the effect o f proxim ate and m aternal and household covariates on w eigh t-for-age in K erala-U P children 1-4 years old, 1992-93 S everely U n d erw eig h t-O th ers N o r m a l-o th e r s Param eter Standard P Param eter Standard P Variable Estimate Error value Estimate Error value INTERCEPT -1.5158 0.1995 0.0001 -0.0820 0.1669 0.6233 Proximate Determinants Residence at Kerala -0.9768 0.1840 0.0001 -0.6723 0.1278 0.0001 Age o f the child -0.0589 0.0363 0.1048 -0.1260 0.0341 0.0002 Number o f prenatal visits -0.0463 0.0288 0.1080 0.0506 0.0245 0.0391 Delivery at home 0.1249 0.1154 0.2791 0.5060 0.0946 0.0001 Small size at birth 0.8743 0.1390 0.0001 0.8103 0.1187 0.0001 Medium size at birth 0.4803 0.1251 0.0001 0.3987 0.0978 0.0001 Underlying Maternal and Household Determinants Number o f years mother's schooling -0.0359 0.0076 0.0001 -0.0089 0.0072 0.2204 M other’s lowage at first birth 0.1365 0.0647 0.0347 0.3205 0.0589 0.0001 Household has no toilet facility 0.3233 0.0850 0.0001 0.1526 0.0729 0.0363 Prenatal care * mother's education -0.0004 0.0031 0.8880 -0.0115 0.0024 0.0001 to vO efl Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. A ppendix 4 D ichotom ized logistic regression coefficients an d sta n d a rd e rro rs show ing the effect of proxim ate and m atern al and household covariates on height-for-age in K erala-U P children 1-4 years old, 1992-93 S everely S tu n te d -O th e rs N o rm a l-O th e rs Param eter Standard P Param eter Standard P Variable Estimate Error value Estim ate Error value INTERCPT -0.9049 0.1265 0.0001 0.0417 0.1224 0.7333 Proximate Determinants Residence at Kerala -0.8416 0.1572 0.0001 -0.6338 0.1220 0.0001 Age o f the child 0.2593 0.0332 0.0001 0.2625 0.0348 0.0001 Number of prenatal visits -0.0441 0.0196 0.0243 -0.0163 0.0195 0.4026 Delivery at home 0.2898 0.1029 0.0049 0.4515 0.0960 0.0001 Small size at birth 0.1739 0.0744 0.0194 0.2739 0.0797 0.0006 Underlvine Maternal and Household Determinants Number o f years mother's schooling -0.0306 0.0119 0.0103 -0.0529 0.0112 0.0001 Prenatal care * mother's education -0.0090 0.0031 0.0037 -0.0052 0.0025 0.0337 V O O N

Asset Metadata

Creator Cheenath, Joy Inasu (author)

Core Title The determinants of child growth retardation in India

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Political Economy and Public Policy

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

Tag Economics, General,health sciences, nutrition,health sciences, public health,OAI-PMH Harvest

Language English

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-171359

Unique identifier UC11329534

Identifier 3054854.pdf (filename),usctheses-c16-171359 (legacy record id)

Legacy Identifier 3054854.pdf

Dmrecord 171359

Document Type Dissertation

Rights Cheenath, Joy Inasu

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