NOTE TO USERS This reproduction is the best copy available. UMI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The Impact of Parental Death and Socioeconomic Factors on Children in Sub-Saharan Africa by Jennifer C. Crain B.A. High Honours, Carleton University, 1998 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN INTERNATIONAL STUDIES THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA MARCH 2010 © Jennifer C. Crain, 2010 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Library and Archives Canada Bibliothdque et Archives Canada Published Heritage Branch Direction du Patrimoine de l'6dition 395 Wellington Street Ottawa ON K1A 0N4 Canada 395, rue Wellington Ottawa ON K1A 0N4 Canada Your file Votre reference ISBN: 978-0-494-61136-4 Our file Notre r6f6rence ISBN: 978-0-494-61136-4 NOTICE: AVIS: The author has granted a non­ exclusive license allowing Library and Archives Canada to reproduce, publish, archive, preserve, conserve, communicate to the public by telecommunication or on the Internet, loan, distribute and sell theses worldwide, for commercial or non­ commercial purposes, in microform, paper, electronic and/or any other formats. L'auteur a accorde une licence non exclusive permettant a la Biblioth&que et Archives Canada de reproduce, publier, archiver, sauvegarder, conserver, transmettre au public par telecommunication ou par I'lnternet, preter, distribuer et vendre des theses partout dans le monde, a des fins commerciales ou autres, sur support microforme, papier, electronique et/ou autres formats. The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. L'auteur conserve la propriety du droit d'auteur et des droits moraux qui prot&ge cette these. Ni la these ni des extraits substantiels de celle-ci ne doivent etre imprimes ou autrement reproduits sans son autorisation. In compliance with the Canadian Privacy Act some supporting forms may have been removed from this thesis. Conformement a la loi canadienne sur la protection de la vie privee, quelques formulaires secondares ont ete enleves de cette these. While these forms may be included in the document page count, their removal does not represent any loss of content from the thesis. Bien que ces formulaires aient inclus dans la pagination, il n'y aura aucun contenu manquant. M Canada Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT Mass orphanhood in sub-Saharan Africa (SSA) results from conflict, poverty, and disease. In the absence of resilient fostering systems, orphans may be more vulnerable to health hazards than non-orphans. Maternal and paternal orphans may also be vulnerable to differing degrees after losing one or both parents, depending on each parent's care roles. Using multiple logistic regression on national household survey data for four SSA countries, this thesis explores: 1) which socioeconomic and demographic factors are risk factors for child undernutrition; and 2) whether different types of orphans confer varied degrees of vulnerability to undernutrition. Child age, household wealth, and mother's education are significant and reliable predictors of undernutrition. The child's sex, and maternal and paternal orphanhood also have some impact, but the results are not as reliable. These findings add to limited scholarship about health outcomes of maternal versus paternal orphans. More investigations into individual countries' orphan crises are needed. ii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS PAGE Abstract ii List of Tables v List of Figures vi Acknowledgements viii Dedication ix Chapter One: Introduction 1 1.1 Background 1 1.2 Purposes and rationale of this research 4 Chapter Two: Literature Review 8 2.1 What is killing Africa's prime-aged adults? 8 2.1.1 HIV/AIDS 10 2.1.2 Tuberculosis 13 2.1.3 Obstetric complications 14 2.2 The roles of mothers vs. fathers for child development and care 17 2.2.1 Child psychosocial development 18 2.2.2 Direct care and access to basic needs: parents and the greater social safety net 21 2.3 Maternal and paternal death: consequences for children 30 2.3.1 Undernutrition 31 2.3.2 Increased mortality risk 34 2.3.3 Psychological distress 37 2.3.4 Reduced education -. 2.3.5 Increased labour force participation Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 44 PAGE Chapter Three: Methodology 48 3.1 Conceptual Model 48 3.2 Data source and units of analysis 51 3.3 Data preparation 56 3.4 Empirical approach 58 3.4.1 Independent (explanatory) variables 58 3.4.2 Dependent (outcome) variables 61 3.4.3 Statistical methods and fitting the models 61 Chapter Four: Results and Discussion 65 4.1 Sample characteristics 65 4.2 Descriptive results 66 4.3 Cross-tabulations 73 4.4 Spearman's rho correlations 77 4.5 Multiple logistic regression 79 Chapter Five: Conclusion 98 5.1 Greater context of findings: Social determinants of child undernutrition and factors of resilience 98 5.2 Limitations 101 5.3 Concluding Remarks 104 References 107 Appendix: Frequency distributions for anthropometry data 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES PAGE Table 1.1 Potential impacts of a mother's death on her children 3 Table 2.1 Estimates of leading causes of death in Africa's high child and very high adult mortality countries, women 15 to 59 years, 2000 9 Table 2.2 Estimates of leading causes of death in Africa's high child and very high adult mortality countries, men 15 to 59 years, 2000 9 Table 2.3 Fathers' involvement with infants, by world region 30 Table 2.4 Undernutrition (weight for age) among children 5 years of age and under, by survival of parents, for Burundi, 2000 33 Table 2.5 Percentage of children aged 7 to 13 who attend school, by survival of parents, for Burundi, 2000 42 Table 3.1 MICS3 questionnaire modules 52 Table 3.2 Comparison of economic, demographic, and health indictors for The Gambia, Cote d'lvoire, Sierra Leone, and Somalia, 2008 55 Table 4.1 MICS3 descriptive statistics, children under five, for select countries of sub-Saharan Africa 67 Table 4.2 Cross-tabulations of the presence of stunting, and socioeconomic and demographic characteristics, children under five, for select countries of sub-Saharan Africa 73 Table 4.3 Cross-tabulations of the presence of underweight, and socioeconomic and demographic characteristics, children under five, for select countries of sub-Saharan Africa 75 Table 4.4 Cross-tabulations of the presence of wasting, and socio-economic and demographic characteristics, for select countries of sub-Saharan Africa 76 Table 4.5 Correlation coefficients (Spearman's rho) between socioeconomic and demographic factors, and stunted, wasted, and underweight, for select countries of sub-Saharan Africa 78 Table 4.6 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of stunting, for select countries of sub-Saharan Africa (model 1) 80 Table 4.7 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of underweight, for select countries of sub-Saharan Africa (model 2) 88 Table 4.8 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of wasting, for select countries of sub-Saharan Africa (model 3) 94 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES PAGE Figure 2.1 Hypothesized pathways through which maternal social capital may affect child nutritional status 26 Figure 3.1 Schematic conceptual model of socioeconomic, demographic, and child care influences on child health outcomes 50 Figure 3.2 Map of West Africa 54 Figure 3.3 Map of Somalia and bordering states 55 Figure 4.1 Age distribution of children under five, for select countries of subSaharan Africa 68 Figure 4.2 Distribution of male and female children under five, for select countries of sub-Saharan Africa 68 Figure 4.3 Orphanhood distribution, children under five, for select countries of sub-Saharan Africa 69 Figure 4.4 Proportion of undernourished children under five years of age, for select countries of sub-Saharan Africa 70 Figure 4.5 Distribution of children under the age of five, by household wealth, for select countries of sub-Saharan Africa 71 Figure 4.6 Distribution of children under the age of five, by mother's education, for select countries of sub-Saharan Africa 72 Figure A1 Frequency distribution, height for age z-scores, children under five, for The Gambia, 2005/2006 123 Figure A2 Frequency distribution, height for age z-scores, children under five, for Cote d'lvoire, 2006 123 Figure A3 Frequency distribution, height for age z-scores, children under five, for Sierra Leone, 2005 124 Figure A4 Frequency distribution, height for age z-scores, children under five, for Somalia, 2005 124 Figure B1 Frequency distribution, weight for age z-scores, children under five, for The Gambia, 2005/2006 125 Figure B2 Frequency distribution, weight for age z-scores, children under five, for Cote d'lvoire, 2006 125 Figure B3 Frequency distribution, weight for age z-scores, children under five, for Sierra Leone, 2005 126 Figure B4 Frequency distribution, weight for age z-scores, children under five, for Somalia, 2005 126 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PAGE Figure CI Frequency distribution, weight for height z-scores, children under five, for The Gambia, 2005/2006 127 Figure C2 Frequency distribution, weight for height z-scores, children under five, for Cote d'lvoire, 2006 127 Figure C3 Frequency distribution, weight for height z-scores, children under five, for Sierra Leone, 2005 128 Figure C4 Frequency distribution, weight for height z-scores, children under five, for Somalia, 2005 128 Vll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS First and foremost, I would like to extend my sincerest thanks to my thesis supervisor Dr. Jalil Safaei who provided integral support during this project. His belief in me and my ideas, expert guidance in research and statistics, and his utmost patience and kindness allowed me to successfully produce a piece of work that at times seemed daunting. Guidance from my thesis committee members Dr. Neil Hanlon and Dr. Fiona MacPhail was also invaluable for the completion of this research. A special thank you to both of you, not only for your helpful insight and encouragement regarding this thesis, but also for enlightening research and coursework that I completed under your supervision. I would also like to express my gratitude to Dr. Anne George for offering her time and perspective in the external review and defence procedures. To the many other UNBC professors who I worked for and/or studied under, your teaching and mentorship not only helped me form the ideas and skills needed to complete this thesis, but also to financially support myself throughout my Master's. Sincerest thanks to Dr. Data Barata, Dr. Paul Bowles, Dr. Caroline Clarke, Dr. Tony Fang, Dr. Karima Fredj, Dr. Greg Halseth, Mr. Leslie Lax, Dr. Don Munton, and Dr. Heather Smith. It is no wonder that UNBC is "The Best in the West" among small universities. Sincerest thanks to my immediate family for their support, encouragement, and words of wisdom, most notably from my magnificent mother Donna Clark who always stressed the importance of getting a good education and being an independent, balanced, and resilient woman. Lucky for you Mom, this is one university paper that I did not ask you to proofread! A big thanks goes out to my cherished friends Honey, Tlell, Fraser, and Caroline, and office mates Erin, Sue, Kyla, and William, who provided lots of laughs, empathy, and encouragement throughout my studies. Also thank you to my friend and "Den Mother" Cynthia Coles who took such good care of me during my stay on Sullivan Crescent. Finally, I would like to acknowledge my appreciation towards Ngagne Diakhate of UNICEF for providing technical advice about working with the Multiple Indicator Cluster Survey (MICS) datasets. Vlll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION This thesis is dedicated to the women and children of Africa who shared intimate details of their lives through participating in the Multiple Indicator Cluster Survey (MICS). Without the time, energy, and information they volunteered, this thesis would not have been possible. It is my hope that their contributions were not in vain, and that the information generated by MICS and other UNICEF initiatives helps to brighten the future of generations to come. I am confident that it will. ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER ONE Introduction 1.1 Background Global scope of orphanhood. Widespread orphanhood1 in sub-Saharan Africa (herein SSA)2 is a tragic consequence of conflict, poverty, and disease epidemics. Incidence3 of orphanhood is expected to rise over the coming years as HIV positive parents succumb to AIDS-related illnesses (UNAIDS/UNICEF/USAID 2004). Estimates show that by the end of 2003, there were 143 million orphans (up to 17 years old) in 93 countries throughout SSA, Asia, Latin America and the Caribbean, with 90% of those being at least 6 years old (UNAIDS/UNICEF/USAID 2004). As a region, SSA bears the greatest proportion of orphaned children (12.3% of all children in SSA are orphans, compared with Asia at 7.3%, and 6.2% of all children in Latin America and the Caribbean), although the actual number of orphans in Asia is considerably higher (UNAIDS/UNICEF/USAID 2004). Maternal orphans outnumber paternal orphans in the hardest hit countries of southern Africa (in terms of HIV/AIDS), where 60% of all orphans are maternal orphans; without HIV/AIDS a child would be more likely to become a paternal orphan in this region (UNAIDS/UNICEF/USAID 2004). By contrast, in Asia, Latin America and the Caribbean, 40% of orphaned children are maternal orphans (UNAIDS/UNICEF/USAID 2004). 1 There are three types of orphans: 1) Double orphans are children up to 18 years of age whose mother and father have died; 2) Maternal orphans are children up to 18 years of age whose mother has died (children classified as double orphans are also included in counts of maternal orphans); and 3) Paternal orphans are children up to 18 years of age whose father has died (children classified as double orphans are also included in counts of paternal orphans) (UNAIDS/UNICEF/USAID 2004). 2 Sub-Saharan Africa refers to the region below the southern demarcation of the Sahara Desert. The non-subSaharan countries are Algeria, Egypt, Libya, Morocco, Tunisia, and Western Sahara. 3 The term incidence is defined as the number of newly diagnosed cases during a specific time period. It is different from prevalence which indicates the number of cases alive on a certain date. 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Impacts ofparental death. Gertler and colleagues (2003, 1) note that "the death of a parent is one of the most severe traumas that a child can suffer. The loss of a parent causes emotional distress and deprives the orphaned child of love, nurturing, values, information and discipline." Furthermore, differing gender roles of parents in virtually all societies imply differing consequences for the surviving child(ren) of a deceased parent. A Thai proverb poignantly contrasts the effects on a child of their father versus their mother dying prematurely (with the mother being the primary direct caregiver in many societies). It reads: "A boy without a father is a boat that sails without a rudder; without a mother, the situation is much worse: he is a castaway" (Calderon et al. 2007, 186). A mother's death means the loss of any social and economic investment made in her regard (UNFPA Country Support Team 2003), and has been linked to adverse consequences in surviving children, namely reduced education, increased participation in the labour force, decreased access to health care, and poorer health status. Correspondingly, research by Case, Lin, and McLanahan (2000) regarding South African households found that if the birth mother is not present in the household, the amount of money spent on food is likely to be less than what is spent on alcohol and tobacco; this suggests children's reduced consumption and increased adult (male) consumption. Furthermore, children not only lose their primary caregiver, but communities forego her social and economic contributions, including labour (UNFPA Country Support Team 2003). Women comprise a considerable proportion of the labour force in many African countries, most notably in the production of both sustenance and cash crops (UNFPA Country Support Team 2003). Table 1.1 summarizes some of the consequences for children that can result from a mother's death, and some of these are also discussed in greater detail in Chapter Two. 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1.1 Potential impacts of a mother's death on her children Domain Health Potential impacts on children Death Illness Malnutrition Poor hygiene Decreased access to health care Economic Increased labour force participation Poverty Social Reduced education Social isolation Reduced parental supervision and care Psychological Depression Anxiety Other psychological problems Source: Adapted from Reed, Koblinsky, and Mosley 2000. This is not to suggest, however, that only maternal orphans are at risk of these and other consequences. A father's death can also increase a child's vulnerability to health, economic, social, and psychological hazards. Accordingly, paternal orphans have also been linked to poorer socioeconomic status, reduced education, and stunted growth, although the evidence here is more equivocal (see Beegle, De Weerdt, and Dercon 2008; Case and Ardington 2007; and Evans and Miguel 2007).4 Moreover, depending on the availability and accessibility of social and other support systems, the risk of adverse health, economic, social, and psychological outcomes are presumably even greater in the wake of double orphanhood. Notwithstanding the generally accepted notion that orphans hold a potentially disadvantageous position in society, one theory of orphan care proposes that the informal and traditional childcare systems such as the extended family are well-equipped to sustain the large 4 Beegle, De Weerdt, and Dercon (2008) excluded all double orphans from their analysis therefore maternal orphans in their study are "maternal-only" orphans. The studies by Case and Ardington (2007) and Evans and Miguel (2007) differentiate between maternal-only and double orphans. 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. numbers of orphans under their care (Abebe and Asbjorn 2007). By contrast and more commonly maintained is that the traditional system of orphan care is being eroded by the HIV/AIDS epidemic to the point where it is collapsing (Abebe and Asbjorn 2007). Chapter Two documents these competing theories of orphan care in greater detail. 1.2 Purposes and rationale of this research The life course approach to understanding child heath outcomes warrants discussion because of its accordance with the underlying assumptions of this research. The life course approach regards health as "a consequence of multiple determinants operating in nested genetic, biological, behavioral, social, and economic contexts that change as a person develops" (Halfon and Hochstein 2002, 1). As such, life course epidemiology "studies how socially patterned exposures during childhood, adolescence, and early adult life influence adult disease risk and socioeconomic position, and hence may account for social inequalities in adult health and mortality" (Kuh et al. 2003, 778). For instance, undernutrition leads to delayed intellectual development (Brown and Pollitt 1996), and as such, better nutrition in early childhood can improve cognitive achievement (for instance, as measured through test scores) and affect successive schooling decisions (Behrman 1993). Malnutrition can also lead to stunted growth, which has been linked to lower incomes later in life. Specifically, research by Strauss and Thomas (1998) revealed that taller adults earn higher wages, even after controlling for education, and this is particularly the case in low-income countries. Accordingly, studying the potential impacts of a parent's death on their surviving children arguably has implications not only for increasing knowledge on the well-being of orphans, but also for drawing assumptions founded on the social determinants of health concerning the poorer health outcomes and hence reduced productivity of adult orphans. Improving the 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. understanding of how parental death impacts investments in children therefore may also have important policy implications for public health investments overall. The socioeconomic and health impacts of parental education have been given research attention (see Boyle et al. 2006; De and Bhattacharya 2002; Desai and Alva 1998; Frost, Forste, and Haas 2005; Gyimah 2003; Kiros and Hogan 2001; Sahn and Stifel 2002), although much less is known on an empirical level about the downstream effects of a parent's death on child health. The theoretical relationship between parental death and societal wellbeing is generally accepted and acknowledged on the basis of its face validity, in that the potential social and economic consequences of a parent's death on their children seem obvious. Notwithstanding these assumptions, there is a need for empirical research on the health of orphans versus non-orphans (see comments by Andrews, Skinner, and Zuma 2006, for instance). As such, this work intends to study the nutritional status of maternal, paternal, and double orphans versus other chidlren through a multivariate analysis of hypothesized socioeconomic and demographic influences on child anthropometric status. This study complements the descriptive statistics that individual countries generate from the information collected, and informs policy decisions (UNICEF 2000). As Carla AbouZahr, a leading researcher of reproductive health issues asserts "[s]ound information is the prerequisite for health action: without data on the dimensions, impact and significance of a health problem it is neither possible to create an advocacy case nor to establish strong programmes for addressing it" (2003, 1). This thesis has two main objectives. The first is to determine through exploratory literature research what is known about the factors, magnitude, and consequences for orphans, of prime-aged adults dying in developing countries. A review of literature also explores the key themes of this research, namely the mother's versus the father's roles for 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. emotional and physical nurturance of their children, theories of orphan care, and empirical evidence of the impacts of parental death on orphans' health. The second, and primary objective of this research, is to empirically assess these relationships by exploring the following research questions: i) which socioeconomic and demographic characteristics - including orphanhood - are risk factors for child undernutrition? More specifically, ii) do differing types of orphans confer differing degrees of vulnerability to undernutrition? The consideration of socioeconomic and demographic controls is important because HIV/AIDS (the primary cause of orphanhood in SSA) afflicts all strata of society, poor and non-poor alike, and as such, ".. .orphan status alone may not be a good correlate of poverty or adverse outcomes (Ainsworth and Filmer 2006, 1100). The theoretical framework of the flow of survival resources (food, shelter, and health care, among others) between parent and child is implied and acknowledged in the ensuing literature review. The conceptual model provided in Chapter Three (Figure 3.1) further illustrates additional possible influences on child health. These theoretical relationships will provide the basis of achieving the final objective of building and testing empirical models to assess the impacts of parental death on child health outcomes. The specific sampling frame, its rationale, and other methodological considerations are also described in Chapter Three. This research builds upon other studies using household survey data that measure indicators of child well-being in developing countries, including increased mortality risk, school enrollment, and labour force participation. For instance, it complements recent work by Ainsworth and Filmer (2006) who examine the relationship between orphan status, schooling, and other indicators of well-being (but not health) across 51 developing countries. Several other studies are also relevant to the context of this thesis (see Ainsworth and Filmer 2006; Deming et al. 2002; Kennedy et al. 2006; Miller et al. 2007; Monasch and Boerma 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2004; and Waters et al. 2004); however, they are distinct from this thesis across several dimensions. Certain studies are country-specific or draw data from a different sample of countries than this thesis, while other studies provide only descriptive statistics. Others examine different variables than those examined here. Also notable is that the data for this research were released for public use in February 2008, presumably making this research one of the earlier multivariate analysis of these data. Moreover, results based on the same data source for this thesis do not separate maternal from paternal orphans, and are derived from a smaller number of explanatory and outcome variables. All things considered therefore, this thesis contributes to filling a knowledge gap concerning the relationships between factors contributing to children's health outcomes and care in sub-Saharan Africa. Chapter Two reviews a selection of topics that are relevant to this research. While it is not intended to be a systematic, exhaustive review of the relevant literature, using key references it provides a thorough contextual and empirical reference for the statistical analyses that address the primary objective of this research. Chapter Three details the methodology, while Chapter Four presents results and discussion of the empirical analyses. The thesis comes to a close with concluding comments in Chapter Five. 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER TWO Literature Review: Determinants and Impacts of Death Among Prime-Aged Adults in sub-Saharan Africa The purpose of this chapter is to examine topics that are consistent with the theoretical underpinnings of the empirical analysis of this thesis. The discussion begins with estimates of the primary causes of death among Africa's prime-aged adults, followed by an examination of the magnitude of, and factors contributing to, mortality among women and men. Next, the gender-specific parenting roles of mothers versus fathers are discussed, followed by theories of orphan care. The chapter concludes with an analysis of the impacts of parental death on surviving children. 2.1 What is killing Africa's prime-aged adults? While the number one cause of death to both prime-aged women and men in Africa is AIDS, other main causes vary both in type and ranking between the sexes. For prime-aged women, obstetric causes and tuberculosis (TB) are the second and third main causes of death, respectively. For prime-aged men, with the exception of AIDS the majority of deaths are attributable to TB and war (in that order) (Rao, Lopez, and Hemed 2006). Tables 2.1 and 2.2 present estimates of the leading causes of death of women and men 15 to 59 years of age in Africa's high child, and very high adult mortality countries.5 Epidemiological research and demographic surveillance help determine causes of death, where deficiencies in the vital 5 High child mortality and very high adult mortality countries in Africa are designated by the World Health Organization as the APR E region, for epidemiological and demographic purposes. The "E" designation represents the highest mortality stratum amongst mortality strata "A" to "E" (WHO 2003). The AFR E region comprises: Botswana, Burundi, Central African Republic, Cote d'lvoire, Democratic Republic of Congo, Eritrea, Ethiopia, Kenya, Lesotho, Malawi, Mozambique, Namibia, the Republic of Congo, Rwanda, South Africa, Swaziland, Uganda, Tanzania, Zambia, and Zimbabwe (Rao, Lopez, and Hemed 2006) 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. registration systems of many developing countries place limitations on doing so6 (Rao, Lopez, and Hemed 2006). Table 2.1 Estimates of leading causes of death in Africa's high child and very high adult mortality countries, women 15 to 59 years, 2000 Cause of death in females (N=l ,228,010) 1. HIV/AIDS 2. Related to pregnancy, labour or birth 3. Tuberculosis 4. Cerebrovascular disease 5. Malaria 6. Ischemic heart disease Total deaths (%) 58.1 7.3 3.2 2.0 1.4 1.2 Source: Adapted from Rao, Lopez, and Hemed 2006. Notes: The original source table had 10 categories, 4 of which were obstetric complications, and therefore were combined into "Related to pregnancy, labour or birth" in Table 2.1 above. These were maternal hemorrhage (2.6%), maternal sepsis (1.9%), abortion (1.4%), and hypertensive disorders of pregnancy (1.4%). The age group 15 to 59 extends beyond women's reproductive years (15 to 44 years of age). Generating estimates for young adults (15 to 44 years of age) is particularly complicated due to the potential for multiple factors to be at play, namely HIV/AIDS, injury and violence (males in particular), as well as high rates of maternal mortality (Rao, Lopez, and Hemed 2006). Table 2.2 Estimates of leading causes of death in Africa's high child and very high adult mortality countries, men 15 to 59 years, 2000 Causes of death in males (N=l,308,048) 1. HIV/AIDS 2. Tuberculosis 3. War 4. Violence 5. Road traffic accidents 6. Lower respiratory infections 7. Ischemic heart disease 8. Cerebrovascular disease 9. Malaria 10. Syphilis Total deaths (%) 50.3 7.4 4.8 4.2 2.8 2.7 1.8 1.7 1.1 1.0 Source: Rao, Lopez, and Hemed 2006. 6 One hundred twenty-five developing countries were recently graded for adherence to key international statistical methods and standards of good practice. Sixty countries failed to reach the midpoint score. Except for Afghanistan, the lowest scoring countries were from SSA (World Bank 2002). 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The following sub-sections outline the magnitude and factors of leading causes of death to prime-aged adults in SSA, namely HIV/AIDS, tuberculosis, and obstetric complications for women. 2.1.1 HIV/AIDS7 Magnitude of the problem. According to estimates derived by the Joint United Nations Programme on HIV/AIDS (UNAIDS) and the World Health Organization (WHO) (2007) 33.2 million people worldwide are living with HIV, and of these, 15.4 million are women. In SSA, nearly 61% of adults living with HIV are women. By contrast, in the Caribbean, Asia, Eastern Europe and Central Asia, and Latin America, those figures are approximately 43%, 29%, 26%, and 26%, respectively. Moreover, women are arguably the "most susceptible to infection.. .get the most inadequate and inferior access to treatment, take most responsibility for caring for the sick and dying and have the shortest survival rate" (Gilbert and Walker 2002, 1094). The estimated proportion of the world's adult population living with HIV has been stable since 2001 (although trends vary by country), and a number of countries are even experiencing downward trends due to prevention efforts put forth since 2000 and 2001.8 Globally, adult AIDS deaths numbered approximately 1.7 million in 2007, and AIDS remains the leading cause of death in SSA, where 76% of all AIDS deaths occurred in 2007. Many of these adults leave behind their children, and there are an estimated 11.4 million AIDS orphans in this region. While there have been some declines in AIDS-related death partially attributable to increased availability of antiretroviral treatments, many countries (Cote D'lvoire and Nigeria for example) have not yet reached the peak of the AIDS epidemic. 7 Unless otherwise cited, estimates in this section were obtained from UNAIDS and WHO (2007). 8 Cote D'lvoire, Kenya, Zimbabwe, Cambodia, Myanmar and Thailand all show declines in HIV prevalence (UNAIDS and WHO 2007). 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Consequently, according to the United States Agency for International Development (USAID), by 2010 there will be an estimated 35 million AIDS orphans (Deininger, Garcia, and Subbarao 2003). Factors contributing to the spread of HIV/AIDS. Although biological factors9 can impact an individual's susceptibility to HIV infection, upstream social factors exert the most powerful influence on vulnerability. Upstream factors comprise the social determinants of health, in contrast to the immediate medical determinants of health such as receipt of preventive or curative medical treatment. Commonly recognized social determinants of health include occupation, education, age, geography, gender, cultural norms, and socioeconomic status. Inequalities within these domains can also influence one's risk of contracting HIV. Gilbert and Walker even go so far as to argue that "social inequality is the greatest transmitter of HIV/AIDS" (2002, 1094). Occupation influences HIV transmission. Sex workers are a prime example of a vulnerable occupational group. People employed in mobile workforces such as construction, mining, trucking, agriculture, industry, forestry, fishing, and shipping (mostly men) also have an increased risk of contracting HIV because of their propensity to engage in commercial sex (Bates et al. 2004). Education can also influence risk of infection and progression to AIDS. For example, in many developing countries women's lack of education impacts their awareness of their right to access protection for sexually transmitted infections (like condoms, for example), as well as health care for HIV/AIDS symptoms (Silberschmidt and Rasch 2001). Lack of 9 Certain biological conditions mean that some people can be exposed to HIV multiple times but do not become infected, while others become infected but do not appear to progress to AIDS. For example, the presence of protective cytotoxic T-cell and T-helper-cell immune responses appear to affect progression to AIDS (Beattie, Rowland-Jones, and Kaul 2002; Hogan and Hammer 2001) 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. knowledge about HIV transmission is also linked to higher rates of HIV (Bloom 2001; Parker 1999). In addition to occupation and education, gender, age and geography can also contribute to HIV transmission. Women in South Africa and many other SSA countries, tend to be poorer and less educated than their male counterparts. There are also more women than men in rural areas (rural areas tend to be the poorest areas, and many men migrate to urban areas for work), and female-headed households are also more likely to be poor than maleheaded households (Baden, Hassim, and Meintjies 1999; Gilbert and Walker 2002). Poverty implies less access to preventive health care measures such as condoms, for example, and a greater risk for engaging in prostitution. Accordingly, it is not surprising given these social and gender inequalities combined with geographical influences, that the burden of HIV/AIDS is overwhelmingly shouldered by young women in African countries. In South Africa for instance, rates of HIV are higher among women than men (Gilbert and Walker 2002). Sexual-cultural norms such as men's and women's acceptance of multiple sexual partners, and a man having a right to force himself on a woman, also increase women's vulnerability (Leclerc-Madlala 2000). Moreover, given women's limited employment opportunities in many cultural contexts, they may turn to sex work as a last resort for income (especially in desperate times), thereby increasing their risk of HIV infection (Smith 2002). Young girls are especially vulnerable because of violence, having sexual experiences with older men, and being less informed about HIV/AIDS (Smith 2002). These and other social inequalities put women at increased risk of HIV/AIDS and other health-related vulnerabilities. 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.1.2 Tuberculosis Magnitude of the problem. Tuberculosis (TB) is curable, yet it remains the leading cause of death from a curable infectious disease (Dye 2006). In 2006, more than 9 million new TB cases and 1.7 million TB deaths were documented. Of those, 0.7 million cases and 0.2 million deaths were in HIV-positive people (WHO 2008). Africa has the highest estimated incidence rate at 363 cases per 100,000 population each year (WHO 2008), but the majority of people with tuberculosis are found in (ranked from first to fifth in terms of absolute numbers) India, China, Indonesia, South Africa and Nigeria (WHO 2008). Approximately 80% of annual incidence of TB arises in the 22 most populated countries (Dye 2006). Overall, the world has endured a gradual increase of TB incidence (in terms of infections per person) for the last twenty years or so, with the exception of dramatic increases in Eastern Europe and SSA since 1990 and the mid 1980s, respectively, as well as declines in some parts of South East Asia, the Western Pacific, central Europe, Latin America and the Eastern Mediterranean (Dye et al. 2005). Factors contributing to the spread ofTB. Weakened economies, inadequate containment measures, and deficient health care systems have played major roles in the spread of TB across Eastern Europe over the last two decades (Dye 2006). By contrast, much of the spread of TB in Africa in the same time period can be explained by the proliferation of HIV because TB affects those with weakened immune systems (Corbett et al. 2003; Dye et al. 2005; WHO 2006a). Although TB is more common among males, in African populations with high rates of HIV a high proportion of TB patients are women 15 to 24 years old (Corbett et al. 2003; Dye 2006; WHO 2006a). Risk factors garnering increased consideration include diabetes, undernutrition, and tobacco- and pollution-related respiratory conditions (Dye 2006). 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.1.3 Obstetric complications Magnitude of the problem. Complications related to pregnancy (including labour and childbirth) are the leading cause of death,10 disease and disability among women in their childbearing years in the developing world (WHO 2005), amounting to at least 18%u of the burden of disease among women of reproductive age in developing countries (United Nations Population Fund [UNFPA] Country Support Team 2003). In 2000, it is estimated that 529,000 women died from complications during pregnancy and childbirth (WHO 2006b). For every woman who dies, it is estimated that another 20 suffer significant injury or disability — between 8 million and 20 million each year (UNFPA 2005). Moreover, women who become pregnant more than once face a cumulative, lifetime risk of maternal death as high as 1 in 7 the ratio for Ethiopia (WHO 2001). Compare this to Canada where the lifetime risk of maternal death is estimated at 1 in 8,700 (WHO 2001). Regionally, lifetime risk of maternal death is highest in SSA12 (1 in 16), followed by South-central Asia (1 in 46), Oceania (1 in 83) and Latin America and the Caribbean (1 in 160) (WHO 2006b). Together, SSA and South Asia account for 74% of global maternal conditions (Disease Control Priorities Project [DCPP] 2007). Little progress has been made throughout the world in reducing maternal mortality over the last two decades (DCPP 2007). 10 Given the focus in this sub-section on obstetric complications as a cause of death, maternal death/mortality refers specifically to the death of a woman from a pregnancy-related cause. However, in other sections throughout this thesis not focusing specifically on pregnancy-related causes of death, maternal death/mortality refers to the death of a mother from any cause. The Tenth Revision of the International Classification of Diseases (ICD-10) (2007) defines a maternal death (from unspecified cause) as "death from unspecified cause occurring during pregnancy, labour and delivery, or the puerperium" (WHO 2007, Chapter XV). " The burden of disease comprises not only causes of death, but also illness and disability. In the context of maternal health, gestational diabetes (for example) would be included in burden of diseases estimates for women of childbearing age. As such, the 18% value differs from the 7.3% value in table 2.1, because the latter measures deaths only. 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The breadth of this health crisis also raises questions concerning the potential societal impact of maternal ill-health,13 as well the potential for investments in this area to offset the societal burden. In 2001, maternal and perinatal conditions comprised close to 6% of total Disability-Adjusted Life Years (DALYs)14 globally, representing the single largest proportion of the global burden of disease (DCPP 2007). In contrast to the growing consideration given to the incidence and determinants of maternal ill-health however, there has been relatively little empirical research devoted to its impacts on society. This is partially due to difficulties in accurately measuring both maternal mortality and morbidity, as well as the methodological and theoretical challenges associated with tracking the socioeconomic and health sequelae, and establishing causality given the vicious circles associated with disease and poverty. Indirect and direct determinants of maternal death. Adverse socioeconomic conditions are key underlying upstream factors contributing to the prevalence of maternal death and disability. Where poverty, undernutrition, and lacking health care are rampant, women who are biologically too young or old for childbearing, or who have closely spaced pregnancies face increased risks of complications. The WHO identifies five major complications as responsible for over 70% of all maternal deaths, namely severe bleeding (haemorrhage) (25%), sepsis (infection) (15%), complications from unsafe abortion (13%), hypertensive disorders (eclampsia—seizures occurring with pregnancy-associated high blood pressure— 13 When a woman dies from a pregnancy- or childbirth-related complication it is termed maternal mortality, whereas maternal morbidity involves illness or disability of the woman that is attributable to or aggravated by pregnancy, labour, childbirth, or the postpartum period, and can be manifested as any number of several conditions. Unless otherwise specified, maternal mortality and morbidity herein are referred to in a generalized manner as maternal ill-health. 14 The DALY (disability-adjusted life years) is a health gap measure. DALYs for a disease are the sum of life years lost due to premature mortality in the population and the years lost due to disability for incident cases of the health condition (See http://www.who.int/healthinfo/boddaly/en/) 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and pre-eclampsia) (12%), and obstructed labour (8%) (2005). Further, even if care is accessed, the quality and/or provider of care may be unacceptable (for example a male health care practitioner in some cultural contexts is inappropriate), and/or the travel distance required to access care may be too great. Once at the health care facility, human resource and emergency care shortages are likely to be encountered. Health systems in many African countries are not fully functional due to human resource and financial shortages, and no other region is as desperate to improve its health systems (WHO 2006c). Seeking maternity care is also financially debilitating for impoverished women and their households (see Kowalewski, Mujinja, and Jahn 2002; Nahar and Costello 1998), and as such, is not an option for many. A woman's lower social status in the household and community may also deny her the power to make choices that could greatly impact her life (Lule et al. 2005), like the decision to get a formal education, for instance. This may imply that she is not sufficiently educated about the importance and/or availability of prenatal care (Lule et al. 2005). An analysis by Shen and Williamson (1999) revealed evidence that a woman's status (indicated by the ratio of female over male secondary school enrollment; births attended by trained personnel [or health attendants]; contraceptive prevalence; average age of first marriage; and total fertility) has significant effects on maternal mortality, even after controlling for per capita gross domestic product (GDP) and economic growth. Clearly numerous factors contribute to maternal ill-health. Those addressed here represent merely a summarized account of select factors, and the complexity of the interplay between them. The following sub-section presents literature that is relevant to the underlying themes of this research, namely the importance of mothers and fathers for child development, theories of orphan care, and the consequences parental death for children across several 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. domains. Note that the discussion is disproportionately centered on mothers, given the relative dearth of literature about fathers' familial roles, and consequences of paternal orphanhood. 2.2 The roles of mothers versus fathers for child development and care Few people would contest a mother's fundamental role in fulfilling her child's basic physical and emotional needs such as food, shelter, and feeling loved and secure. By contrast, fathers are often seen as being the protectors, providers, and authority figures. Are these mere stereotypes, or does evidence throughout the literature suggest there is some truth to these labels? While not intended to be an exhaustive review of child development discourse, this section addresses selected literature that illuminates the fundamental parent-child linkages that influence a child's health and well-being. As such, these works inherently shed light on the theoretical underpinnings of the emotional and physical sustenance that an orphan may be lacking. This is not to suggest that other kin or caregivers cannot provide an emotionally and physically nurturing environment to the child that ultimately engenders a secure, confident, healthy and well-adjusted adult — especially in societies where the extended family and community play an active role in raising children (as observed in one theory of orphan care discussed in subsection 2.2.2). Rather this discussion centres on the concept of parent-child attachment, with attachment as defined by Mary D.S. Ainsworth as "an affectional tie that one person forms to another specific person, binding them together in space and enduring over time. Attachment is discriminating and specific" (1973, 1). As such, the distinct elements of the parent-child relationship that are entrenched in biological and gender-based 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. processes that perhaps cannot be entirely replicated by other care figures are acknowledged in the ensuing discussion. It is also important to acknowledge that in terms of psychosocial development, childparent attachment theory may be more prognostic of familial social norms in Western societies than societies where child-rearing is more communal (Schneider, Atkinson, and Tardif 2001), due to a dearth of cross-cultural research into parent-child attachment especially regarding father-child relationships. This is especially true in relation to fatherchild attachment because "paternal roles are more culturally variable than maternal roles" (Paquette 2004 194). Nonetheless, in spite of its shortcomings for cultural applicability in terms of Western versus African contexts, published knowledge on fathers' roles and child development provides a useful starting point for addressing issues of fatherhood. Child psychosocial development is first examined, followed by direct care and access to basic needs. 2.2.1 Child psychosocial development [T]he prolonged deprivation of a young child of maternal care may have grave andfar reaching effects on his character and so on the whole of his future life. It is a proposition exactly similar in form to those regarding the evil after-effects of German measles before birth or deprivation of vitamin D in infancy. — John Bowlby (1907 - 1991)15 The notion that a parent holds a special place in a child's life is based on the concept of attachment theory. Derived from the study of animal behaviour, information processing, developmental psychology, and psychoanalysis (Bretherton 1992), attachment theory postulates that children can be divested "not by the economic, nutritional, medical or housing conditions in which they [live], but simply by separation from their mothers" 15 Bowlby 1953, p. 53 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Stroebe 2002, 129). Early attachment theory was founded by John Bowlby, although Bowlby's close colleague Mary Ainsworth is also credited with making significant and creative contributions to attachment theory (Bretherton 1992). The theory argues that it is "the mother with whom the child forms the basis for the development of subsequent relationships in life" (129). As such, a child's separation from their mother may lead to emotional problems and diminished success parenting their own children later in life (Stroebe 2002). Specifically, those who study attachment contend that children's first relationships especially with their parents or other primary caregiver - address two fundamental needs (Ainsworth 1973; Belsky and Cassidy 1994; Bowlby 1969; Cassidy 1999; Sroufe 1996). First, the caregiver's presence allows the child to approach unfamiliar or challenging circumstances with more confidence, and to better deal with stress (Ainsworth 1967; Emde 1980; Emde and Easterbrooks 1985; Gunnar 2000; Gunnar et al. 1996). Second, a young child's sense of efficacy and competence in their influence on the world is strengthened through attachment relationships (Carson and Parke 1996; Denham et al. 1997; Hooven, Katz, and Gottman 1994). While research on attachments does not denounce the importance of children developing close relationships with other caregivers including relatives, child care providers, and friends, some researchers argue that the security of attachment between a mother and her child is more influential on early psychosocial growth than are the relationships a child has with other caregivers at home or outside the home (Easterbrooks and Goldberg 1990; NICHD Early Child Care Research Network 1997 and 1998). In contrast to mother-child attachment, much less research attention has been devoted to father-child dynamics of child psychosocial development. The work of several researchers, however, is relevant to this discussion and warrants attention. Anthropologist Barry S. 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hewlett has done some important cross-cultural research on father-child relationships, and has found that fathers across many cultural contexts16 give limited or no direct care to children (2000). Work by Paquette however, emphasizes, that notwithstanding the seemingly distant relationship fathers have with their children, in many cultures fathers still have responsibilities for nurturing their children's social skills and adaptation to physical surroundings, and perform an "important provider role" (2004, 193). Moreover, other research shows that infants do form attachments to their fathers, and that these attachments are important especially for the development of fewer behavioural problems (Verschueren and Marcoen 1999), greater sociability (Lamb et al. 1982; Sagi, Lamb, and Gardner 1986), and higher quality peer interactions (Parke 2002). Research by Kromelow and colleagues (1990) also revealed that fathers tend to encourage risk-taking, often through boisterous play. Such encouragement of children to exhibit initiative in unfamiliar circumstances, face risks head-on, and have greater self-assurance around strangers (among other traits of development), are critical for children's early introduction to the social skills they will need to face the world around them (Le Camus 1995; Paquette 2004). These cross-cultural differences, especially regarding how much direct care and/or provision of household resources a father is responsible for, have important implications for the well-being of paternal orphans across differing cultures. For instance, a paternal orphan from a cultural context whereby the father provides a considerable amount of direct physical and emotional nurturance, could conceivably experience more grief in the wake of his/her father's death in contrast to a paternal orphan whose deceased father fulfilled mainly a financial provider role and spent little time in the household. The latter child, however, could conceivably be at a 16 While some of Hewlett's work concludes that fathers across many cultural contexts give limited or no direct care to children, other research he has conducted presents evidence that fathers in some African cultures are more involved with the direct care of children than others. Discussion of this aspect of Hewlett's work is forthcoming in this thesis. 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. greater disadvantage regarding his/her standard of living given the loss of the father's income. 2.2.2 Direct care and access to basic needs: parents and the greater social safety net " ...[W] omen are said to be 'homemakers', they are 'led', are 'submissive', are 'workers', are 'dependants', and play the role of mothers of the nation."I? "That is the thankless position of the father in the family-the provider for all, and the enemy of all."1S In many societies, there is significant variation in residence, occupation, intrahousehold resource allocation, marriage patterns, and living arrangements (Caldwell 1976; McNicoll 1980; Ryder 1984), child bearing and child-rearing in SSA tend to take place within a larger social unit including the extended family and even friends (Madhavan 2001). The death of an adult in the household, therefore, does not necessarily imply that a child has lost his or her parent, and may be regarded as a temporary shock to the production of child health (Ainsworth and Semali 2000). The death of a parent, however, may have an everlasting impact because household members may not be appropriate substitutes for a mother (for example, as in the case of the death of a breastfeeding mother) (Ainsworth and Semali 2000). As such, parents are likely to "be more efficient at transforming health and food inputs into child health" (Ainsworth and Semali 2000, 5) than another caregiver who may not know the child as thoroughly as the parent. Household adults can therefore be substituted, but the loss of a parent and their unique contributions to their child's development is likely to have an impact well into the future (Ainsworth and Semali 2000). Gertler et al. (2003, 7) further explain, "If the primary promoter of good health and educational norms and values dies, and 17 (Ochwada 1997, 18 126) As quoted by Swedish dramatist, novelist, and poet J. August Strindberg. Obtained from http://www.wisdomquotes.com/002304.html 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the surviving parent does not adequately fill this role, surviving children may lose the motivation to attend school and implement good health and nutritional practices." Moreover, Africa is often misguidedly regarded as homogeneous across several dimensions including strategies of orphan care and their efficacy in the wake of premature parental death. In actuality, the two dominant theories of orphan care in the literature are divided in their assessment of the social safety net's integrity. One theory posits that there exists a looming, complete disintegration of the social safety net in AIDS afflicted communities (see Ayieko 2000; Foster 2000; George, Oudenhoven, and Wazir 2003; Okeyo 1995; and UNICEF 2003). This phenomenon is revealed in the mass reduction of prime-aged adults, leaving the elderly generation to care for surviving children because the aunts and uncles also often succumb to AIDS-related illnesses. In their research pertaining to poverty and ageing in Africa, Kakwani and Subbarao (2005, 2) call such arrangements "skipped generation" households. As such, it is likely that in contexts where social capital is weakened due to a compromised social safety net, becoming an orphan could have detrimental effects on the child's health and well being. Evidence also points to the extended family safety nets as being undermined even more in countries with greater urbanization (Foster and Williamson 2000). By contrast, another theory proposes that the social rupture thesis as described above is too simplistic (Chirwa 2002) and that informal and customary childcare systems such as the extended family are more resilient for sustaining the large numbers of orphans under their care than once perceived (Abebe and Asbjorn 2007; Chirwa 2002). In reference to his research on orphan care in Malawi, Chirwa (2002, 94) argues that "alternative forms of social organization and new social relationship patterns, with increasingly broad adaptive capacities, are emerging as a result of the HIV/AIDS crisis.. .There is a process of social reconfiguration 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in which some strategies are discarded while others are adopted." Furthermore, communitybased strategies are also benefiting orphans and foster families in Malawi according to Ghosh and Kalipeni (2004). Malawi's National Orphan Care Task Force does its work through a network of community-based childcare centres which aim to improve care and learning opportunities for children. Moreover, many NGOs and community-based groups in Malawi aim to reunite and reintegrate orphans with their surviving parents, families, and communities. Community support is also portrayed in Guest's (2004) story of Molatela, a 17 year old girl whose parents died of AIDS. With little or no support from other family members, Molatela had to assume running the household including caring for her siblings, one of whom is HIV positive. Fortunately, Molatela and her siblings were able to receive food parcels and other support from a local community group which has made a significant contribution to their ability to cope. Moreover, in her literature review, Strebel provides numerous examples of community interventions for the care of orphaned and vulnerable children in several subSaharan African countries including Cote d'lvoire (one of the countries studied in the empirical component of this thesis). In the Ivorian city of Abidjan "...two NGOs - one of which is a network of people with AIDS (PWAs) - provide material (including payment of school fees), medical, and skills-training support for poor families with AIDS orphans. Results indicate that orphans benefit from staying within the extended family, families have become more independent through income generating activities.. .and that the PWAs experience ownership of the project, which is due to be replicated elsewhere" (2004, 3). Further attention to social conditions as they relate to theories of orphan care is provided in 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter Four, Results and Discussion. The following section explores women's and mother's roles in ensuring child well-being. Women's and mother's roles in ensuring child well-being. Socio-cultural norms regarding the household division of labour mean that child care, domestic labour, and food preparation remain essentially women's roles in SSA. Women in SSA undertake as much as 95% of all household chores, supply up to 30% of labour in ploughing, 50% in planting, 60% in weeding, and 85% in preparing food (Meena 1994). Overall, women's contribution to labour in rural Africa is thought to be three times greater than men's (Meena 1994). As such, the death of a woman has far-reaching implications for her immediate and extended family not only in terms of grief and psychological distress, but also in terms of her inputs into the functioning of the household and care of children. In addition to the direct labour and care practices in which women partake, a mother's social connectedness or social capital is also hypothesized to influence children's well-being. Putnam (1993) describes social capital as the norms, networks and associations that facilitate cooperative action. In earlier work defining social capital, Coleman (1988, 4) contends, "Social capital is defined by its function. It is a not a single entity but a variety of different entities, with two elements in common: they all consist of some aspect of social structures, and they facilitate certain actions of actors within the structure. Like other forms of capital, social capital is productive, making possible the achievement of certain ends that in its absence would not be possible." 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moreover, De Silva and Harpham (2007) theorize that in contrast to social isolation, social capital may enable mothers to know'9 more via knowledge dissemination among one another (for example, about food sources), do things differently (for example, breastfeeding duration), think differently (for example, attitudes towards hygiene), and feel different (for example, abundant emotional support is linked to good maternal mental health, and improved child growth) (Patel et al. 2004; Harpham et al. 2005) than if they were lacking such social assets. Figure 2.1 illustrates the hypothesized pathways through which maternal social capital may affect child nutrition status. Note that structural social capital refers to objective measurable behaviours that build cohesion, such as being a member of a group, whereas cognitive social capital embodies people's subjective perceptions of cohesion, such as having a sense of trust and mutual cooperation (Harpham, Grant, and Thomas 2002). An examination of social capital in Peru carried out by De Silva and Harpham (2007) reveals some of the pathways illustrated in Figure 2.1. Mothers' advocacy for obtaining title deeds to their land, and to obtain water, electricity and other key utilities for their impoverished communities, not only improves citizens' everyday quality of life, but also curbs communicable disease and improves child nutritional status. Moreover, some community women's and mothers' groups in Peru are directly linked to child feeding initiatives such as the state-subsidized Glass of Milk Program which provides free milk and cereal to babies of mothers who are members of the associated community groups. Also relevant to the discussion, Madhavan (2001) examines possible ways in which female relationships in SSA can affect demographic outcomes, namely women's fertility and child survival. She hypothesizes that the nature of the female relationships within this social context can impact the degree to which resources such as food, medicine, and attention are 19 In the original article by De Silva and Harpham (2007), the italicized words were also emphasized (but using upper case). 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. invested in children within the extended social unit. For example, women who have good rapport with their counterparts (such as co-wives in a polygamous marriage, or and/or sistersin-law) may feel more competent at maintaining their children's health than those who live alone or in a household where competition among members is high (Madhavan 2001). The extended family system has generally functioned to foster orphans, but much remains to be Figure 2.1 Hypothesized pathways through which maternal social capital may affect child nutritional status High cognitive social capital High structural social capital Citizenship Group membership Social support Psychosocial well-being Economic development Access to food resources Food security Access to knowledge networks Improved living conditions r n Reduced childhood physical illness Ability to cope Access to health services H n Increased ability to care for child T Good child nutritional status Source: De Silva and Harpham 2007 learned about their lives following the shock of losing one or both parents (Masmas et al. 2004). There is some evidence, for instance, that fostering arrangements can actually adversely affect child survival (Bledsoe 1989; Bledsoe, Ewbank, and Isiugo-Abanihe 1988; Foster 1997), depending on the social context. For example, linking to the work of 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Madhavan (2001) women may have little reason to care for one another's children in a competitive household. Jones (1999) examined woman-headed households in the Eastern Cape of Southern Africa. Women in this region believe that men are an economic hindrance, in that men are known to hide, or deny the household a share of their earnings, and insist on indulgences which are well beyond the household budget. As such, many women choose singlehood and collaborative housekeeping and child-rearing relationships with other women. The benefits consist of having more social and economic support for themselves and their children; a greater ability to meet basic needs and to manage household consumption and expenses; diversification of unemployment risk and income generation; and having better control over their fertility. As demonstrated in the featured literature, mothers are a critical link in a child's health and well-being. A child's psychosocial development in terms of basic virtues such as confidence, trust, and self-worth largely hinge upon a child's relationship with their mother, especially in their early years. Mothers are also the primary caretakers of children in many societies, and are thus responsible for food production, medical care, and other basic needs. Moreover, mothers have been shown to collaborate on household management and childrearing, thereby reducing the risk of household shocks such as unemployment and illness; but what about men's and fathers' roles? Men's and fathers' roles in the household division of labour. The critical roles that mothers fulfill notwithstanding, Nsamenang contends that..fathers cannot... be assumed to be peripheral or insignificant figures in the lives of their children" (1992, 326). This comment stems from the nearly absent research attention paid to African fathers' parenting roles, likely 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. partially due to the assumption that they simply are not involved as parents, but rather play a largely reproductive role. By contrast, as with mothers in all parts of the world, the vast majority of fathers everywhere are trying to do the best they can within their particular cultural, ecological, economic, and demographic contexts (Hewlett 2004). Granted some evidence does support the claim that in general some African fathers are not as involved in the direct care of their children, it is dangerous to assume such is the case throughout all of Africa given evidence of variations in gender roles for child-rearing (as discussed in the following paragraph). In many cases, it is also inaccurate to assume that the father's absence from the household is associated with a lack of support, and vice versa - common conjectures of social science research (Madhavan, Townsend, and Garey 2008). As such, it is important to consider the varied implications for child support of a father's location in relation to his children, as well for child-rearing in terms of varied (and sometimes unexpected) gender roles across different communities within a society. For instance, as demonstrated by Barry S. Hewlett's research (2004) fathers within the !Kung20 and Aka foragers of Central Africa are very engaged in child-rearing. In contrast to fathers in urban-industrial societies, he found many Aka fathers to be in close proximity or to be holding their infants almost half of the day. He also observed Aka fathers to be more tender and loving towards their infants while holding them than were mothers. Some Aka fathers, however, were much less involved in the care of their infants, which is symptomatic of the intracultural variability of fathers' level of care. Research by Fouts, Hewlett, and Lamb (2005) also revealed that fathers within Bofi foraging communities of Central Africa are known to frequently carry/hold their children, typically during periods of weaning. Mothers and fathers also take turns transporting infants and toddlers using side-slings. 20 The unusual, but correct placement of the exclamation point in MKung' indicates a sound that is made in the non-English enunciation of the word. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Additional evidence of the household division of labour (including child care) as deviating from 'traditional' gender roles is found in the work of Smit (2006). Her study of the evolving role of South African husbands/fathers in dual-income families, revealed that while "remnants of the "the husband/father as head of the family" still exist, some men have moved toward greater equality in the marriage and more involvement in the family and domestic sphere" (412). Statistically significant (p = 0.000) positive correlations were found between greater responsibility in performing domestic tasks and child-rearing and care. On fathers' presence versus absence and support of the household, a study by Barbarin and Richter (2001) warrants attention. They compared a large cohort of South African children born in 1990, to Ugandan and African-American children. Findings revealed that notwithstanding a biological father's frequent absence from the household, 61% of mothers surveyed reported that the fathers are engaged in financially supporting their children. Furthermore, research on rural South African fathers by Madhavan, Townsend, and Garey (2008) implies it is imprudent to assume an absent father implies their lack of financial support for their children, as well as the opposite scenario. Children are also just as likely to get a share of earnings from fathers of different households, as they are from fathers with whom they reside. Madhavan, Townsend, and Garey (2008) also found that paternal financial support provided at any point in a child's upbringing is likely to continue throughout their lives. In terms of rural subsistence African communities, Hewlett's (2004) findings diverge from this trend, and reveal that in most settings fathers provide almost no direct care to infants and young children. Moreover, a study by Barry and Paxson (1971) found that fathers in SSA were the least involved in infant and child development out of nine world regions studied. Table 2.3 presents these findings. 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.3 Fathers' involvement with infants, by world region Region Sub-Saharan Africa Middle Old World North Eurasia and Circumpolar Southeast Asia and Pacific Islands Australia, New Guinea, Melanesia Northwest Coast Northern and Western North America Eastern Americas (includes N. and S. America) Mesoamerica, Central America, and Andes Number of cultures 22 12 12 22 12 7 7 16 9 Mean score* 2.40 2.87 3.17 3.60 3.42 3.29 2.71 3.05 3.22 *A 1-3 score means the father is never, seldom, or occasionally near the infant, and a 4-5 denotes the father is in regular or frequent proximity. Source: Adapted from Hewlett (2004). Based on coding of father involvement by Barry and Paxson (1971). Regional groupings based on the analysis of Burton et al. (1996). The next section explores the consequences for children of maternal and paternal death. 2.3 Maternal and paternal death: consequences for children As previously illustrated, the extended family and social network serve as a fundamental fostering system of support for child-rearing and other domestic responsibilities. The quality of care from a new caregiver is one of the most important factors in a child's adjustment to loss (Bowlby 1973). Sadly however, some foster caregivers may take advantage of children's labour and neglect their health and education (Bedri, Kebede, and Negassa 1995; Foster et al. 1995). Accordingly, empirical evidence points to the importance of the said linkages between parents and their children's health and well-being. The following section demonstrates the consequences of a mother's versus a father's death for their children's health, namely their nutritional status, survival, and psychological health, and their education and labour force participation. 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.1 Undernutrition Stunting (based on height for age), underweight (based on weight for age), and wasting (based on weight for height) are physical indicators of undernutrition. Stunted growth, an indicator of chronic nutritional deficiency, is regarded as a well-established measure of a population's capacity to meet basic needs such as food, water, housing and health care (deOnis, Frongillo, and Blossner 1997). Wasting is a measure of an acute period of nutritional deficiency, (Martorell and Habicht 1986), whereas underweight is a composite measure of both acute and chronic conditions. In their analysis of data from the Indonesia Family Life Survey, Gertler et al. (2003) found positive associations between a mother's death and increased probability of her surviving children being undernourished. Paternal death was found to have no effect on child anthropometric measurements (coefficients are positive but not significant). Conversely, maternal death had a large and significant impact on children's short term health, as shown by coefficients on the 25 weight-related measurements. For example, maternal deaths reduced a child's weight for age z-score by 0.7 standard deviations, weight for height z-score by 0.9 standard deviations, and body mass index (BMI) by 0.75 points (Gertler et al. 2003). Ainsworth and Semali's (2000) research on the consequences of parental death also warrants discussion. Their longitudinal study (1991-1994) in Kagera, Tanzania (a region with high rates of HIV/AIDS) revealed very high levels of stunting among children under five years old. Regression models revealed that children who lost their mothers were much more likely to be stunted than children who lost their fathers, or than non-orphans. Holding the values of all other variables at their means, maternal orphans in poor households have considerably lower height for age (1.18 SD lower) than children in the same type of household with living mothers, while paternal orphans also show large negative effects (0.4 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SD lower). Predicted stunting among poor maternal orphans and paternal orphans is 59.3% and 58.3%, respectively, compared to 39.6% among poor non-orphans.21 Beegle, De Weerdt, and Dercon (2006) also examined the long-run impact of orphanhood on children in the Kagera region. Using baseline data from the Kagera Health and Development Survey (1991 to 1994) and another round of data collected in 2004, they examined the impact of orphanhood on children's height, controlling for a broad range of household and child characteristics before orphanhood, and for community fixed effects. An association between maternal orphanhood and height deficiencies among children 11 to 18 years old in 2004 was revealed, although it could not be concluded whether those effects were permanent. Paternal orphanhood was not found to be associated with child height. Recent data collected by UNICEF on over 10,000 children in Burundi also show that a higher percentage of double orphans and maternal orphans are undernourished, compared with non-orphans (Ministere de la Planification du Developpement et de la Reconstruction 2001). Moreover, the data also reveal a marked inequality between male versus female undernutrition among double orphans, although no interpretation of this gap is provided by the authors of the referenced source (Table 2.4). Findings about orphanhood and undernutrition from the Multiple Indicator Cluster Survey 3 (MICS3). According to data collected in 2006 via the third cycle of the Multiple Indicator Cluster Survey (MICS3) (also the data source for this thesis), orphans in The Gambia ranging from 0 to 59 months of age were also more likely to be stunted (25.9%) or underweight (22%) than children who were not orphaned or considered vulnerable (22.4% 21 In their results, Ainsworth and Semali (2000) make no distinction between single maternal orphans (only the child's mother is dead) and double orphans (both the child's mother and father are dead). It is therefore assumed that their results pertaining to maternal orphans apply to both single maternal orphans and double orphans. 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.4 Undernutrition (weight for age) among children 5 years of age and under, by survival of parents, for Burundi, 2000 Survey conditions Parental status > 2 standard deviations (moderately underweight) > 3 standard deviations (severely underweight) Both parents alive Mother dead Father dead Both parents dead Both parents alive Mother dead Father dead Both parents dead Number of children surveyed 687 14 50 1 647 17 47 5 Percentage of children undernourished Males Females 30.6 32.9 61.5 23.1 40.0 21.3 40.0 25.0 12.4 13.0 23.1 38.5 12.9 20.0 20.0 0 Source: Adapted by Subbarao, Mattimore, and Plangemann (2001), using data from Enquete Nationale d'evaluation des Conditions de vie de I 'enfant de la Femme au Burundi, Institut de Statistiques et d'Etudes Economiques du Burundi et UNICEF, 2000. stunted, and 20.2% underweight) (Gambia Bureau of Statistics, UNICEF, and The World Bank 2007). In terms of other possible socioeconomic factors of undernutrition measured by MICS3, children in The Gambia's rural areas are also more likely to be underweight, stunted, or wasted as urban children. Children whose mothers have primary or higher education are least likely to be underweight and stunted than children of mothers with no education. The MICS3 initiative also revealed that children from rural areas of Cote d'lvoire are more likely to be moderately or severely stunted, wasted, or underweight that urban children. Children whose mothers have no formal education are also more likely to be undernourished (INS Cote d'lvoire 2007). Similarly, children in Sierra Leone whose mothers have secondary or higher education are less likely to be undernourished than children of mothers with no or only primary education. Boys appear to be slightly more likely to be underweight, stunted, and 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wasted as girls. Orphans were also more likely to be moderately or severely underweight, stunted or wasted as other children (Statistics Sierra Leone and UNICEF-Sierra Leone 2007). Children living in Somalia's rural areas are almost twice as likely to be moderately underweight than those living in urban areas. Somali children whose mothers have primary or secondary education are less likely to be underweight and stunted compared to children of mothers with no education or Koranic education. Orphans were also more likely to be moderately or severely, stunted, underweight, or wasted than non-orphans (UNICEF Somalia 2006). Malnutrition may also lead to increased mortality among orphans. The following section presents empirical evidence of the increased mortality risk that orphans experience. 2.3.2 Increased mortality risk Numerous studies support the assertion that surviving children of a mother who died are significantly more likely to die than children who live with both parents, and are less likely to receive adequate nutrition throughout their lives. Anderson et al. (2007) studied the consequences of maternal mortality (in terms of an obstetric cause) on infant and child survival in rural Haiti. They examined deaths among women of reproductive age between 1997 and 1999, and classified them as maternal (related to pregnancy) or non-maternal (not related to pregnancy). It was found that the odds of the family experiencing the death of a child were 1.01 (95% CI 0.81-1.26) when a non-maternal death occurred, and were higher when a maternal death occurred (odds ratio [OR] 1.55, 95% CI 1.12-2.15). These figures translate into a 55% increased probability of experiencing the loss of a child (younger than 12 years) when a maternal death occurs, as opposed to a non-maternal death of a woman of reproductive age. While the authors were not able to determine whether the maternal death 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. was causal of a child's death or if health care factors contributed to both, they hypothesize that care factors (such as breastfeeding, nutrition, and health maintenance performed by the mothers) are related to deaths of infants and older children. Since the 1960s, Matlab, Bangladesh has been the focus of numerous demographic surveillance activities carried out by the International Centre for Diarrhoeal Disease Research, Bangladesh (herein ICDDR,B; formerly the Cholera Research Laboratory). Several studies generated by those surveillance activities have shown that adult death is associated with significantly higher mortality risk of children during the five years following the death of a parent, and these risks are much higher when an adult female dies (Chen et al. 1974; Roy et al. 2000). For example, one study (study period from 1983 to 1987) monitored children for two years whose households had experienced an adult death from any cause. Logistic regression analysis revealed that 70% of children under one month old whose mother had died, and half as many children 1 to 11 months old, were likely to die, controlling for health service area, SES measures (dimension of dwelling), and the year the observation began. Also, children whose mothers had died were significantly more likely to die than those whose fathers had died. After the first month of life, female children were more likely to die than male children (Strong 1992). The level of detail in Strong's empirical results does not allow for the calculation of child mortality rates. Moreover, Gertler et al. (2003) used data from the first and second waves of the Indonesian Family Life Survey (IFLS), and six waves of the Mexican evaluation surveys for the Progresa health, education, and nutrition program (ENCASEH [Encuesta De Caracteristicas Socioeconomicas de los Hogares] and ENCEL [Encuesta Evaluation de los Hogares]22), to empirically examine two key ways through which parental death could impact 22 The expanded versions of these acronyms were obtained from http://www.iipri.org/data/mexico01.asp 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. children: (1) a decline in household economic assets, and (2) "removal of parental presence that causes grief and removes love and guidance" (p. 2). They found that for both countries, a mother's death increases the probability of child death. In their examination of survival among maternal orphans in Guinea-Bissau, Masmas et al. (2004) find that most people interviewed for the study (88%23) felt that a maternal orphan's life was considerably inferior to that of a child whose mother was alive. Controlling for numerous socioeconomic, biological, and demographic factors, considerably higher mortality was measured among maternal orphans than among controls in both urban and rural settings. Almost all of the excess mortality was among children whose mother died when they were less than 2 years old, and few maternal orphans had access to substitute breastfeeding (Masmas et al. 2004). A study of orphans living in Spain between 1870 and 1950 revealed similar findings in that the loss of a mother lead to significant excess mortality of young children, especially up to the age of 2, while a father's death had a comparatively limited adverse influence (Reher and Gonzalez-Quinones 2003). Likewise, in Blantyre, Malawi, Taha et al. (1996) found that, controlling for household socioeconomic characteristics, the child's gender, birthweight, first-born status, and the age and HIV status of the mother, young children whose mothers died were 3.3 times more likely to die themselves, compared to the children of mothers who did not die. Maternal orphans may also be vulnerable to early mortality because of their heightened risk of HIV/AIDS infection and other health hazards. In their study of eastern Zimbabwe, Gregson et al. (2005) found that among a sample of girls 15 to 18 years old, high proportions of HIV infections, other sexually transmitted infections, and pregnancies can be 23 The relatives of 128 motherless children from a rural cohort and 192 motherless children from an urban cohort. 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. linked to maternal orphanhood. Moreover, a study of maternal AIDS orphans in Zaire revealed higher rates of missing scheduled clinic visits, early weaning and poor adult supervision compared with children whose parents were alive (with serostatus either HIV positive or negative) (Kamenga et al. 1990). Also, while not directly attributable to increased mortality but still a serious public health issue, psychological distress among orphans has also been given research attention, and is discussed in the following section. 2.3.3 Psychological distress A parent's death is arguably the most emotionally devastating event for a child, and the psychological response to the shock of losing one's parent(s) can persist for months or even years (Goodman 2001). An AIDS-related death is also often considered dishonourable given the perceived linkages to shameful behaviours such as prostitution and drug use. Moreover, even changes in the nature of work performed by youth are perceived to be linked to the spread of HIV. With specific reference to Tanzanian youth, the shift from doing "proper work" in farming (that causes fatigue), to working in the informal business sector which "doesn't tire them enough," is blamed for their "wandering around" and "causing AIDS" (Setel 199924). Consequently, adding to the severe stress of losing a parent is taunting, teasing, and other forms of social stigmatization and exclusion that many AIDS orphans experience (FHI2003; Nyamukapa, Foster, and Gregson 2003). Moreover, children in many African societies are often ill-prepared for bereavement because of cultural norms dictating the avoidance of discussing, preparing for, or even acknowledging death (Horizons 2003; Posel 2004). These approaches have the potential to limit a child's understanding of the 24 Lily M. speaking to Community Development Committee, Mbokomu, January 30, 1992, as quoted in Setel's book (1999). 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. finality of death, thereby hindering the grieving process which is necessary for recuperating from the loss (Brodzinsky, Gormly, and Ambron 1986). Potential outcomes of unresolved emotions include anger and depression (Sengendo and Nambi 1997). Moreover, the death of a parent may involve moving to a home of lower socioeconomic status, changes in educational opportunities including switching schools or dropping out, separation from siblings, or even forced self-sufficiency (Sengendo and Nambi 1997). Adapting to such shocks can be extremely difficult for even the most physically and emotionally healthy child, let alone a child who is severely stressed from living in extreme poverty, and having to cope with the social stigmatization associated with AIDS. Atwine, Cantor-Graaea, and Bajunirweb (2005) used the Beck Youth Inventories of Emotional and Social Impairment (BYI) to measure psychological distress among AIDS orphans in rural Uganda. They found orphans to be at greater risk for higher anxiety levels (OR = 6.4), depression (OR = 6.6), and anger (OR = 5.1). Orphan status was the only significant outcome predictor, when other multiple factors with possible relevance for BYI outcome were considered. Likewise, Makame, Ani, and Grantham-McGregor (2002) found considerably more internalizing problems (including anxiety, a sense of failure, depression, pessimism and others) and less positive affect (measured using questions that ask about experiencing positive feelings and moods) among orphans than non-orphans in their study of the psychological well-being of orphans in Dar El Salaam, Tanzania. In their study of teenage orphans' experiences of parental AIDS-related illness and bereavement in Zimbabwe, Wood, Chase, and Aggleton (2006) revealed that many orphaned teenagers want to discuss parental illness and death with adults. The adults, however — whether an ill parent or other caregivers — are often ill-prepared to cope with children's distress in a constructive manner. Furthermore, a particularly relevant finding demonstrated 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the loss associated with a mother's death was more severe than a fathers' because of the "superior quality of the mother-child relationship, which was said to be characterised by trust and good-quality communication." By contrast, fathers were accused of being "careless" and abusive of alcohol, while mothers were said to be better caregivers, more approachable, and that they showed "love with greater ease" (1927). Orphans in the Rakai district of Uganda were asked whether they were angry about their parents' death and whether they blamed their parents for their death (Sengendo and Nambi 1997). Many orphans (52%) were angry about their parents' death, with children living with relatives the most likely to be angry (65%), followed by those living with grandparents (57%). Children who lost their mother were actually less likely to be angry (43%) than children whose father had died (48%). Most children who reported anger attributed it to times when they were faced with problems or fearful situations. In terms of depression, all orphans were classified as depressed, and were positive about the future compared to non-orphans (82% versus 100%). Children 10 to14 years old whose mother had died were much more likely to be depressed than those whose father had died (mean depression scores 22.9 and 17.3, p= .044, respectively). Children 15 to 19 years old who lost their mother were also much more likely to be depressed than those whose father had died (mean depression scores 17.8, and 14.2, p= .001, respectively). Sengendo and Nambi (1997) contend that the difference in depression scores between maternal and paternal orphans can be attributed to the generally poor capacity of fathers to offer love, care, and domestic assistance to their children and the household, as compared to the mother. The discussion now turns to reduced education as an outcome of orphanhood. 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.4 Reduced education The loss of a parent has the potential to reduce a child's chances of starting, continuing or completing school (Ainsworth, Beegle, and Koda 2002). Schooling costs may be considered prohibitive for some households, children's labour may be needed at home, and/or foster caregivers may not be as willing to finance the education of those who are not their biological kin (Ainsworth, Beegle, and Koda. 2002). Orphans therefore, are at a disadvantage in terms of schooling and other activities that require a significant financial investment by foster caregivers, as is the dominant theme throughout the literature; notwithstanding, some literature reveals otherwise. In their study of orphans in Kagera, Tanzania, Beegle, De Weerdt, and Dercon (2006) measured strong effects of orphanhood on education. Children who lost their mother between the ages of 7 and 15 lost upwards of a year of schooling on average, contrasted with children whose father died (and who were orphaned at a younger age than the maternal orphans described above) that lost on average 0.4 years of schooling. Moreover, the data revealed that children not enrolled at the time of the parent's death lose considerably more schooling in comparison to non-orphans or those already in school at the time of the parent's death. Some evidence also pointed to orphans from wealthier households as being less vulnerable to schooling losses. Ainsworth, Beegle, and Koda. (2002) also examined the effect of parental death on children's schooling in Kagera, Tanzania. They found that enrollment rates for children aged 7 to 10 years in households that had experienced a female adult death (whether or not she was a parent) in the previous 12 months were less than half the enrollment rate than that among children from households with no female deaths (18% versus 39%, respectively; statistically significant). The death of a prime-aged male had no effect on enrollment. 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In their study of Indonesia's and Mexico's orphans using the Indonesia Family Life Survey (IFLS) and the Progresa welfare program, respectively, Gertler et al. (2003) found that for Indonesia, a father's death doubles the school dropout rate in relation to non-bereaved children (14% and 7%, respectively), while a mother's death lowers the probability of school enrollment more so than a father's death. They also found that in Mexico, there is a higher probability of school dropout for children whose mother had died, as compared to whose father had died. Recent data collected by UNICEF on over 10,000 children in Burundi also confirm that maternal orphans are at a significantly higher risk of not being in school than paternal orphans (Ministere de la Planification du Developpement et de la Reconstruction 2001) (Table 2.5). Likewise, using data from the Demographic Health Surveys program, for several countries in East and West Africa, Bicego, Rutstein, and Johnson (2003) found that the probability of being at the proper education level was lower for maternal orphans in both the younger (ages 6-10) and older age groups (ages 11-14) than for children who had lost only their father, or who were non-orphans. Deininger, Garcia, and Subbarao (2003) used panel household data to assess the impact of AIDS-induced orphanhood in Uganda, and found that foster children faced a clear disadvantage.25 However, this disadvantage became less apparent over time, parallel to the increase in enrollment via the implementation of Universal Primary Education. 25 Deinenger et al. (2003) used panel data which focused on foster children (that is children who are not physically related to the household head). As such, it was not possible to confirm whether they were true orphans. 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.5 Percentage of children aged 7 to 13 who attend school, by survival of parents, for Burundi, 2000 Survival status of parents Both parents alive One/both parents dead Mother dead Father dead Both parents dead All children Sex Male 53.3 42.4 33.6 47.6 33.0 50.5 Female 45.8 37.3 30.2 40.4 35.4 43.7 7-9 years 40.4 29.2 20.8 32.7 24.6 37.8 Age 10-13 years 57.0 47.1 38.6 52.5 39.0 54.3 Total 49.5 39.8 31.7 44.0 34.1 47.0 Source: Ministere de la Planification du Developpement et de la Reconstruction 2001. Nyamukapa and Gregson (2005) explored primary education among orphans and nonorphans in rural Zimbabwe. They found that primary school completion rates were lowest among maternal orphans (boys 45.9% and girls 59%), and girls whose fathers had died actually had higher completion rates than non-orphaned girls (76.3% versus 67.2%, respectively). This can be explained by their greater tendency to live in female-headed households with the surviving parent or other relatives. The study also revealed that women succeeded in raising education fees through their own work as well as other sources including relatives and social programs (Nyamukapa and Gregson 2005). By contrast, there is also evidence suggesting that some orphans are not as disadvantaged in terms of schooling as is generally perceived. A study in the Rakai district of Uganda found that maternal orphans living with their father were the least affected group in terms of lost school time following a parent's death (Sengendo and Nambi 1997). The other categories were children living with their widowed mother, grandparents, or other relatives, or living in child-headed households. Moreover, results from an orphan enumeration survey completed by foster caregivers in Zimbabwe (including surviving parents) show little discrimination between measures of education between orphaned and non-orphaned children living in the same family, and only 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. limited differences between orphan and non-orphan families in the same community. These results, however, contradict testimony given by community members. This implies that foster caregivers may be underreporting discriminatory behaviour due to the indignity associated with favouring one child over another (Foster et al. 1995). Further, Lloyd and Blanc (1996) found that a child's biological parents appear to play a less critical role in children's education outcomes when comparing orphans to non-orphans. Conversely, it is namely the household-head's education and the household standard of living which better explain differences among children's education. Notwithstanding, Lloyd and Blanc (1996) also point out that after controlling for the household's socioeconomic status and other variables, children living in female-headed households are consistently more likely to be enrolled in school and to have completed grade four (a critical marker of school progress according to UNICEF26) than children living in male-headed households. This is indicative of women's tendency to invest more time, money, and emotional support in ensuring children in their household have access to education (Lloyd and Blanc 1996). Research by Kobiane, Calves, and Marcoux (2005) revealed mixed results regarding the impact of orphanhood on children's schooling in Burkina Faso. They found that being orphaned is unfavourable for entering school, most notably for double orphans in rural areas. Moreover, paternal orphans were found to be more likely to enter school than maternal orphans. Kobiane Calves, and Marcoux (2005) assert that this is likely due to the largely patrilineal nature of Burkinabe society whereby orphan care is largely assumed by the paternal extended family. Surprisingly, however, the research also revealed that double orphans in urban settings are twice as likely to enter school as non-orphans. One explanation 26 See UNICEF 1993. 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is that some of these double orphans were orphaned in rural areas and then were sent to urban areas for schooling (Kobiane, Calves, and Marcoux (2005). 2.3.5 Increased labour force participation Children who have lost one or both parents may also have to work outside the home in order to supplement the household income and pay for their schooling. Examples of occupations include agricultural and domestic labourers, porters, street hawkers, and most tragically, commercial sex trade workers (Foster and Williamson 2000; Omokhodion, Omokhodion, and Odusote 2006). Not only does working lead to less time for education, but many jobs also put children at risk of adverse health and safety outcomes including (but not limited to) respiratory infections, skin diseases, drug abuse, and road traffic accidents (Ayaya and Esamai 2000; Omokhodion, Omokhodion, and Odusote 2006). Although empirical evidence on child labour as an outcome of orphanhood is not as well-documented as other outcomes such as reduced education, increased mortality and decreased nutrition, a few studies warrant attention. While the reviewed studies in general suggest that orphans' participation in the labour force is more pronounced than non-orphans, some results are inconclusive, and one even suggests that male maternal orphans are more likely to assume a greater burden of household chores than their female counterparts. Using data for 20 SSA countries, a UNICEF study revealed that in all countries studied, orphaned children (single and double orphans) aged 5 to14 were less likely to be in school and working in excess of 40 hours per week (Guarcello et al. 2004). Studies by Monasch and Boerma (2004) and Suliman (2003) also found that orphans were more involved in labour than non-orphans. Likewise, surveys of working children carried out by the International Labor Organization have also shown that orphans are much more likely than 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. non-orphans to be working in commercial agriculture, domestic service, the sex trade, and as street vendors (Guarcello et al. 2004). Mushingeh et al. (2002) found that in Zambia, HIV/AIDS among parents and guardians increased the child labour force by an estimated 23% to 30%. Moreover, data from Zambia's 1999 Child Labour Survey confirms the increased likelihood of orphaned children to be working, in contrast to non-orphaned children (Republic of Zambia Central Statistical Office, Lusaka, and International Labour Organization/International Programme on the Elimination of Child Labour 1999). By contrast, using data for Senegal to create a composite index of child labour27 combining economic activity and household chores, no clear association between child labour and orphanhood was revealed (Guarcello, Lyon, and Rosati 2004c). The results stratified by sex showed that for boys, rates of both economic activity and chores are lower for maternal and paternal orphans than for non-orphans, while child labour rates for double orphans are relatively the same as for non-orphans. For girls, maternal and paternal orphans experienced higher child labour rates than non-orphans, but rates for double orphans were found to be lower than for non-orphans. Explaining such an unexpected result is a complicated endeavour given the numerous individual and household influences dictating both orphans' and nonorphans' use of time within the household (Guarcello, Lyon, and Rosati 2004c). One can speculate, however, that double orphans are at increased risk of having a poorer health status than non-orphans, and as such, are potentially less productive members of the household, in contrast to healthier non-orphans. Moreover, the children examined in the study were living in a formal household, and therefore were less at risk of entering into the labour force, in contrast to children living outside of any formal household (who were not examined in the 27 Regarding the composite index of child labour, Guarcello, Lyon, and Rosati (2004b) define children engaged in child labour as all economically active children aged 5-14, excluding children aged 12-14 involved in light work, (<14 hours per week), in addition to all children aged < 14 involved in household chores > 28 hours per week. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. study which uses data gathered via a household survey) (Guarcello, Lyon, and Rosati 2004c). When involvement in household chores is solely considered, female maternal and paternal single orphans were found to be much more likely to spend a minimum of 28 hours per week doing chores than other children. This may be explained by gender roles in SSA, whereby females shoulder more of the burden of running the household. Household survey data collected in the year 2000 for Burundi show that maternal orphans and foster children28 are more involved in economic activity than non-orphans, but that the situation for paternal and double orphans is similar to that of non-orphans (Guarcello, Lyon, and Rosati 2004a). This may be explained in part by patrilineage customs whereby children whose father has died stays with paternal kin, whereas a maternal orphan is more likely to be separated from their father and become what is called a de facto double orphan (Guarcello, Lyon, and Rosati 2004a). As such, there is a greater chance that the child will be living with biological kin if the father dies than if the mother dies. Biological kin may be more willing to invest an orphan's well being in contrast to non-kin foster caregivers, especially if they have biological children of their own. In accordance with the results for Senegal, Guarcello Lyon, and Rosati (2004a) also found for Burundi a strong association between maternal orphanhood and involvement in household chores. Interestingly however, it is the male maternal orphans who were found to be more likely to spend at least 28 hours per week doing chores than non-orphans and orphans from other categories. This chapter provided context for the subsequent empirical examination of orphanhood as it relates to outcomes of child well-being. The HIV/AIDS epidemic, tuberculosis, maternal causes, war, and other hazards have left millions of children orphaned in SSA. There is mixed evidence that children who have lost their mothers versus their 28 Children who are not orphaned but live in a separate household from their parents (Guarcello, Lyon, and Rosati 2004) 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fathers are at a greater disadvantage in terms of their lifespan, nutrition, educational attainment, labour provision, and other indicators of well-being, given the critical roles that each parent play as gatekeepers for their children's access to resources. As such, the impact of paternal orphanhood versus maternal orphanhood will vary across countries given differences in traditional gender roles of parents. The next chapter presents the methodological approach of this thesis. 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER THREE Methodology This chapter describes five key aspects of the methodology of this thesis and is organized as follows. The conceptual model is introduced followed by a description of the data source, unit of analysis, and the data screening/preparation strategy. Finally, the empirical approach is discussed. 3.1 Conceptual model As noted in Chapter One, at an empirical level, comparatively little research attention has been given to the socioeconomic and demographic influences of a parent's death on child health, in contrast to the influence of parental education on child health, for instance. Accordingly, this research intends to study the nutritional status of maternal, paternal, and double orphans versus other children through a multivariate analysis of socioeconomic and demographic factors, including parental death. To reiterate, the specific research questions are: i) what socioeconomic and demographic characteristics - including orphanhood - are risk factors for child undernutrition? More specifically, ii) do differing types of orphans confer differing degrees of vulnerability to undernutrition? While many health determinants are not quantifiable, previous research has shown (as described in Chapter Two) that certain socioeconomic, demographic and health care factors can have measurable effects on child health. Even before an individual is born, the likelihood that they will become seriously ill, be exposed to pollutants and other health hazards, or receive adequate health care can be significantly influenced by the community or household they are born into. Such socioeconomic characteristics are often referred to as upstream influences because they have 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a more indirect impact on one's health, than a more direct input such as medical care. Household wealth, for example, plays a part in determining parents'/caretakers' capacity to provide nutrition, education, and health care to their children. It may also tell a story regarding the wealth of a caretaker's own household when they were a child. In turn, their own children may be more vulnerable to disease or other health hazards. Consider for example, a high-risk pregnancy due to the malnourishment of the mother. The infant is consequently also more likely to be underweight and susceptible to illness as a result. Parents' education can also influence child health in a number of ways. A more educated parent is more likely to earn a higher wage, thereby affording them better access to nutrition and health care for themselves and their family. Further, knowledge acquired by future mothers may be gained through education, and in turn, enhanced literacy and numeracy increases a parent's capacity and receptivity to diagnose and treat child health problems through medical care (Glewwe 1999). Demographic influences can also influence a child's health. As described throughout Chapter Two, a mother's death likely means the loss of the child's primary caregiver, as well as a potentially productive member of the household, while a father's death could mean the loss of the main source of income. Moreover, children fostered by extended family or other social ties may be lower on the list of children to receive expensive health and/or education inputs. As suggested in the literature, orphanhood has been associated with children's increased vulnerability to premature death and other health threats. A child's age may also influence their health outcomes, since younger children, for example, may be less immune to certain illnesses than older children, and/or may be higher on the priority list for certain health care investments. Moreover, older children are more likely to have stunted growth than younger children, because stunting is a measure of chronic undernutrition over time. By 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. contrast, underweight can occur as a result of a short term nutrition shock or crisis (Kennedy et al. 2006), but also accounts for longer term nutritional deficiencies as well. Wasting also indicates acute undernutrition that could, for example, be attributed to "significant seasonal shifts associated with changes in the availability of food or disease prevalence" (Gambia Bureau of Statistics, UNICEF, and The World Bank 2007, 23). Figure 3.1 Schematic conceptual model of socioeconomic, demographic, and child care influences on child health outcomes. SOCIOECONOMIC INFLUENCES Household wealth Mother's education (FURTHEST "UPSTREAM") Mother dead DEMOGRAPHIC INFLUENCES Father dead Child's sex Child's age CHILD NUTRITIONAL STATUS Stunted Wasted Underweight Source: The nested ellipses design was adapted from a figure by the National Committee on Vital and Health Statistics (2002). A child's sex may also have implications for their health status, linked to differing gender status of male versus female children. Such gender differences could influence a child's access to health care and other inputs necessary for their well-being. For instance, in some cultures male children have higher status than females, therefore boys may receive preferential treatment when it comes to health care and nutrition. 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Beginning with a description of the Multiple Indicator Cluster Survey, the following section examines the data source, units of analysis, and data preparation steps for this thesis. 3.2 Data source and unit of analysis The Multiple Indicator Cluster Survey (MICS) is a household survey program designed by the United Nations Children's Fund (UNICEF) to help countries monitor the situation of women and children.29 Originally developed to help countries monitor progress towards indicators for the goals of World Summit for Children,30 the first round of MICS was conducted in 1995 in over 60 developing countries (UNICEF 2006). Since then, there have been second and third cycles of MICS, and fieldwork for MICS4 is currently underway in several countries (UNICEF 2010). The most recent completed round of MICS (MICS3) gathered data in approximately 65 countries around 2005 (in that there is some variation between countries in the year(s) data were collected), and covers a wider range of topics than those contained in previous rounds of the survey. In doing so, the information collected helps in the monitoring of the goals of: 1) the Millennium Declaration, 2) the World Fit for Children Declaration and Plan of Action, 3) the United Nations General Assembly Special Session on HIV/AIDS and 4) of the African Summit on Malaria (UNICEF 2006). Table 3.1 illustrates the general topics covered in MICS3. The model questionnaire can be customized to each country's needs by adding questions/question modules, and comprises a household questionnaire, a questionnaire for women 15 to 49 years of age, and a questionnaire for children under five years old (aged 4 29 The model MICS3 household and child questionnaires are publicly available online at http://www.childinfo.org/mics3_questionnaire.html. Also see the following UNICEF sources in the reference list of this thesis for bibliographic information on the specific datasets used in the empirical analyses: UNICEF 2008a, 2008b, 2008c, and 2008d. 30 Representing the largest gathering of world leaders in history, 71 heads of State and Government and 88 other senior officials level assembled at the United Nations in September 1990 to attend the World Summit for Children. The Summit adopted a Declaration on the Survival, Protection and Development of Children and a Plan of Action for implementing the Declaration in the 1990s (UNICEF n.d.). 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. completed years or younger, that is 0 to 59 months). The child questionnaire is addressed to the mother or primary caretaker of the child (UNICEF 2006). Table 3.1 MICS3 questionnaire modules Household modules Modules for children under five Household Information Panel Household Listing or Extended Household Listing Education Water and Sanitation Household Characteristics ITN (Insecticide-treated Nets) Children Orphaned and Made Vulnerable by HIV/AIDS Child Labour Salt Iodization Under five Child Information Panel Birth Registration and Early Learning Vitamin A Breastfeeding Care of Illness Malaria Immunization Anthropometry Modules for women Optional modules Women's Information Panel Child Mortality Tetanus Toxoid Maternal and Newborn Health and Malaria Prevention Marriage/Union and Polygyny Contraception Female Genital Mutilation/Cutting Sexual Behaviour HIV/AIDS Additional Household Characteristics Security of Tenure and Durability of Housing Child Discipline Source and Cost of Supplies for InsecticideTreated Nets, ORS Packets, Antibiotics, and Antimalarials Contraception and Unmet Need Attitudes Towards Domestic Violence Child Development Disability Maternal Mortality Italics indicate additional modules to be included in all Multiple Indicator Cluster Surveys, where they apply. Source: UNICEF 2006. The primary unit of analysis of this thesis is children 0 to 59 months of age, and as such, the child questionnaire is the primary data source for child health variables. Although information on older children is also collected by MICS3 (for example, age, orphanhood status) data on children's anthropometric status are only gathered in the child questionnaire 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (children 0 to 59 months). As such, data on the health status of older children are not captured by MICS3. Women aged 15 to 49 years are the main survey respondents, including for those questions pertaining to children under five. Given the cluster sample design, sample survey weights are not used in the empirical analysis of this thesis. Permission to access the data was granted upon providing UNICEF with a brief description of the intended use of the data during the request process.31 Because UNICEF seems quite liberal in granting access to its data and does not require a detailed research proposal from individuals requesting data access, the number of individuals who can access and analyze the data is increased, thereby fostering a win-win situation for UNICEF and researchers alike. Recall from earlier discussion that on a regional level SSA bears the highest percentage of children who are orphans (12.3%; compared with Asia at 7.3%, and 6.2% of children in Latin America and the Caribbean) (UNAIDS/UNICEF/USAID 2004). As such, it seemed fitting to focus on that region, in addition to it being on the forefront of international development discourse. The countries from which data for this thesis were generated comprise The Gambia, Cote d'lvoire, Sierra Leone, and Somalia, which are all West African countries with the exception of Somalia which is located in East Africa (Figure 3.2). These are the countries of choice because they were the only SSA countries for which MICS3 data were released at the time the empirical work of this thesis began. Each country was analyzed separately (as opposed to pooling the data into one dataset) for comparison purposes. 31 Requests to access MICS3 data are made through an online form available at http://www.childinfo.org /mailforml .php 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.2 Map of West Africa National Boundaries Major Routes SENEGAL Mopti MALI National Capital Scale (KM) 0 200 400 San Kita Bamako Kedoi BURKINA FASO o ^Bougouni Labe GUINEA Kinlk .Wa Tamale Makeni ' COTE D'lVOIRE jcKreetown 3|^RRA H OGC _ LEONE GHANA Man ^8 Voita LIBERIA Kumasi^ Monrovia ATLANTIC OCEAN Buchanan Abidjan* Accra 199 3 DeL omie M apping donated by Response.N et Source: ReliefWeb 1996. Regarding the countries' overall conditions (Table 3.2), Cote d'lvoire was in a state of civil war until 2007, and Sierra Leone and Somalia are considered failed states in that they "... can no longer perform [their] basic security, and development functions and [have] no effective control over [their] territory and borders" (Crisis States Research Centre 2004, 1). Moreover, all four countries rank poorly on the Human Development Index (HDI). The HDI is "...a composite measure of three dimensions of human development: living a long and healthy life (measured by life expectancy), being educated (measured by adult literacy and enrolment at the primary, secondary and tertiary level) and having a decent standard of living (measured by purchasing power parity, PPP, income)" (UNDP 2009a, 1). 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.3 Map of Somalia and bordering states "" , *8erbera SOMAULAND Hargeysa Garoowe ETHIOPIA Baidoa I NDI A N OCEAN MOGADISHU y KENYA Merca too 200 km Source: USAID n.d. Table 3.2 illustrates how the economic, demographic, and health contexts vary between countries. Table 3.2 Comparison of economic, demographic, and health indictors for The Gambia, Cote d'lvoire, Sierra Leone, and Somalia, 2008 Indicator **Human Development Index (HDI) ranking out of 206 states GDP (USD) *GDP per capita (PPP USD) Population Population growth rate Life expectancy Under-5 mortality rate1 Prevalence of HIV, per 100,000 population >15 years*** (2005) The Gambia 168 Cote d'lvoire Sierra Leone Somalia 163 180 204 $781 million $1,300 1.7 million 3% 56 109 2091 $23.4 billion $1,700 20.6 million 2% 58 126 6442 $1.9 billion $700 5.6 million 2% 48 262 1361 $2.6 billion $600 9.0 million 3% 46 142 870 Sources: *Central Intelligence Agency 2008; **UNDP 2009b; ***WHO 2010. World Bank 2009 (2008 and f2007 data). 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Poverty, conflict, and other difficult social conditions have direct implications for children's long- and short-term nutritional status. As mentioned in Chapter Two for instance, long-term abject poverty and consequent chronic undernutrition from insufficient food intake and/or a diet lacking in essential nutrients can lead to stunted growth over time (as indicated by height for age measurements). Moreover, inadequate sanitation and an unsafe/insufficient water supply can cause diarreah, which can result in both acute and chronic undernutrition due to weight loss and malabsorption of essential nutrients (WHO 2009a). In turn, El Taguri et al. (2009) also found stunting to be a risk factor for other health problems later in life, including obesity, "developmental delay, impaired immune function, reduced cognitive function, metabolic disturbances leading to accumulation of body fat, loss of lean mass and increased risk of hypertension" (558). They explain, "Stunted children have impaired regulation of food intake and have higher susceptibility to the effects of high-fat diets.. .Underprivileged groups may have the knowledge, the attitude and the capacity to correct stunting, but unfortunately not always in the most appropriate approach" (559). Further discussion about the vulnerability of children in the wake of conflict and other adverse social conditions is provided in Chapter Four. 3.3 Data preparation MICS3 data are subject to a comprehensive data editing process by UNICEF staff prior to being released for public use. Data are entered twice and any differences between the two data entries are captured and addressed during a basic consistency check. Secondary editing is also performed in order to investigate more complex inconsistencies. Data are also 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. scrutinized in terms of ensuring the number of questionnaires in the data file is consistent with the number of questionnaires received from field interviewers (UNICEF 2006).32 In addition to the said precautions taken by the data provider, for each of the four country data sets in the sample, various practical considerations were addressed by the author during the data screening and preparation stage. Descriptive statistics were generated to serve as an assessment of overall data coverage, and variables found to have sizeable amounts of missing data were not included in the analysis (for instance, father's education level). Moreover, although on a regional basis SSA bears the greatest percentage of children who are orphans (12.3%), this thesis only looks at children 0 to 59 months of age. As such, a significant proportion of orphans in the study population were not represented in the analyses (for reasons discussed earlier regarding the capture of anthropomorphic data only among children up to 59 months of age). Some independent variables were recoded into dichotomous variables, namely sex, and maternal, paternal, and double orphanhood, where a value of 1 denotes the presence of a condition and a 0 denotes the absence. Moreover, each category within child's age (categories 1 to 5), mother's education (categories 1 to 3), and household wealth (categories 1 to 5) were treated as a dichotomous variable by the SPSS software, using the "categorical" stipulation with the binary logistic regression analysis option. Final models are streamlined to represent the most theoretically plausible factors of child undernutrition, and for reasons of theoretical significance and model manageability (such as preserving degrees of freedom and limiting collinearity). 32 The structure checking process is technically complex, and does not lend itself to a succinct explanation here. Additional details on the structure checking process can be found in UNICEF. 2006. Multiple Indicator Cluster Survey Manual 2005: Monitoring the Situation of Children and Women. New York: UNICEF. http://www.childinfo.org/mics3_manual.html 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The following section describes the empirical approach used to estimate whether certain socioeconomic, demographic, and orphanhood factors influence child health, as measured by anthropometric status. 3.4 Empirical approach 3.4.1 Independent (explanatory) variables Demographic measures Child's sex: The child's sex is measured using the dichotomous variable "male" where a 1 denotes a male child and a 0 denotes a female. Child's age: For descriptive statistics, the child's age is measured in terms of age categories, namely "agel" (first year of life, child is 0 to 11 months old), "age2" (second year of life, 12 to 23 months old), "age3" (third year of life, 24 to 35 months old), "age4" (fourth year of life, 36 to 47 months old), and "age5" (fifth year of life, 48 to 59 months old). For logistic regression, each category is treated as a dichotomous variable, with the first (youngest) age category as a reference for odds ratios (2:1, 3:1, 4:1, 5:1). Orphanhood status: Orphanhood status is measured in terms of maternal orphanhood (variable label "morphan"), paternal orphanhood ("porphan"), and double orphanhood ("dorphan"). For the purpose of generating descriptive statistics, a fourth orphanhood variable measuring non-orphanhood was also created ("norphan"). Orphanhood is measured using dichotomous variables where a 1 denotes the presence of orphanhood, and a 0 the absence. 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Socioeconomic measures Wealth index: The wealth index variable is an asset-based measure derived using principle components analysis (PCA) by MICS technical staff. In PCA, several variables are aggregated into a more manageable number of components based on like relationships between the variables and their respective weight as a measure (Rutstein and Johnson 2004). Regarding the household wealth index specifically, the assigned weight of each variable relates to its relative validity as a measure of wealth.33 Household assets include (but are not limited to) access to a radio; household ownership of a bicycle; the dwelling's flooring, walls, and roofing materials; the number of rooms in the dwelling; the main source of household drinking water; and the type of toilet facility (UNICEF 2000). To further explain how the wealth index is derived, [e]ach household is.. .weighted by the [total] number of household members [including adults and children], and the household population is divided into five groups of equal size, from the poorest quintile to the richest quintile, based on the wealth scores of households they [are] living in. The wealth index is assumed to capture the underlying long-term wealth through information on the household assets, and is intended to produce a ranking of households by wealth, from poorest to richest. The wealth index does not provide information on absolute poverty, current income or expenditure levels, and the wealth scores calculated are applicable for only the particular data set they are based on (Statistics Sierra Leone and UNICEF-Sierra Leone 2007, 9). For descriptive statistics, the wealth index variable is measured by the quintile categories "wealthl" through "wealth5." For logistic regression, each category is treated as a dichotomous variable, with the last (richest) wealth category as a reference for odds ratios (1:5, 2:5, 3:5, 4:5). Either the poorest or richest quintile would be a suitable reference category for this analysis because they represent the extremes on either end of household 33 For a detailed account of how the wealth index is derived see DHS Comparative Reports 6: The DHS Wealth Index (Rutstein and Johnson 2004). http://www.childinfo.org/files/ DHS_Wealth_Index_ (DHS_Comparative_Reports).pdf 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wealth status, and as such, allow for intuitive interpretation of odds ratios as they relate to undernutrition. The richest quintile was chosen as the reference category. The wealth index is an important measure of a household's standard of living, and therefore has direct implications for interpreting the resources available for health care expenditures, education, proper nutrition, and other basic determinants of health. Mother's education'. The mother's (or primary female caretaker's)34 education is measured by the "melevel" variable. The categories comprise "melevell," "melevel2," and "meleveB," indicating i) no formal schooling, non-standard curriculum, or religious studies,35 herein referred to throughout this thesis as 'no or limited schooling' ii) primary level schooling, and iii) secondary or higher education, respectively. For logistic regression, each category is treated as a dichotomous variable, with the first (least educated) education category as a reference for odds ratios (2:1, 3:1).36 While the father's education is also documented in MICS3, it could not be included in the analyses due to considerable amounts of missing data. Education is an important indication of socioeconomic status, which has implications for the ability of a caretaker and/or the household as a whole to provide basic needs for their children. 34 In the case of maternal or double orphans, mother's education measures the education of the child's primary female caretaker (15-49 years of age), who was the respondent of the MICS survey. 35 The author aggregated these categories due to low counts, and a lack of available information in MICS3 documentation on what these categories measure in terms of literacy, numeracy, and other learning indicators; also see UNICEF 2005a. 36 The author chose the first education category as the reference category because in some instances, there were zero cases of mothers with secondary+ education (third category); as such, when the secondary+ education was used as the reference category, it resulted in hyper-inflated odds ratios. 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4.2 Dependent (outcome) variables Child health (anthropometry) measures Stunting: Stunting is measured using a dichotomous variable derived from the mean zscores for height for age (the "haz" variable). The cut-point for anthropometric status is two standard deviation units (< -2SD) below the mean z-score as estimated by the National Centre for Health Statistics/WHO childhood growth curves.37 Accordingly, using the cut point of < -2SD below the mean, a dichotomous variable called "stunted" was created. Underweight: Underweight is measured using a dichotomous variable derived from the mean z-scores for weight for age (the "waz" variable), based on the recommended cut point of < -2SD below the mean. Wasting: Wasting is assessed using a dichotomous variable derived from the mean z-scores for weight for height (the "whz" variable), based on the recommended cut point of < -2SD below the mean. 3.4.3 Statistical methods and fitting the models To reiterate, the empirical analyses of this thesis serve to explore whether socioeconomic and demographic factors, including orphanhood, have an empirical influence on child nutritional status. Three main types of quantitative analyses were performed in this thesis. Firstly, for each of the four countries (The Gambia, Cote d'lvoire, Sierra Leone, and Somalia) simple cross-tabulations between explanatory and outcome variables were performed in order to 37 Ngagne Diakhate (of UNICEF) in an email to the author of this thesis, April 7, 2008. Also see in References Measure DHS (n.d.) 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. assess bivariate frequencies between variables - an early indication of any potential associations between variables. Secondly, a more advanced measure of correlation, the Spearman's rank order correlation coefficient (or Spearman's rho), was applied to reveal the strength of any correlations between independent and dependent variables, and is an appropriate test of correlation given the ordinal nature of the variables in the empirical analysis. Spearman's rho functions on the assumption that ordinal variables contain a minimum of three ranked categories. In order to meet this assumption, variables were recoded as follows: • Stunted, wasted, and underweight: 1 = the absence of undernutrition, 2 = the presence of moderate undernutrition, and 3 = the presence of severe undernutrition (dichotomous versions of the variables used for logistic regression, whereby a 0 denotes the absence of undernutrition, and a 1 denotes presence of undernutrition).38 • Orphanhood status: 1 = both parents are known to be alive, 2 = one parent is known to be alive, and 3 = both parents are known to be dead. • Age: 1 = 0 to 19 months, 2 = 20 to 39 months, 3 = 40 to 59 months. • Household wealth: 1 = first or second quintile (poorer), 2 = third quintile, 3 = fourth or fifth quintile (richer). • Mother's education: 1 = no/limited education, 2 = primary education, 3 = secondary or higher education. 38 A child is considered moderately stunted, underweight, or wasted when the height for age, weight for age, and weight for height z-scores, respectively, fall between -2.0 and 2.99 standard deviations (SD) below the mean. A child is classified as severely stunted or wasted when the associated z-scores are less than -3.0 SD below the mean (based on a pre-determined international standard). Severely underweight children have weight for age z-scores that are less than -3.0 SD below the mean (WHO 2009b). 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, multiple logistic regression (LR) (logistic regression involving multiple independent variables) was performed, since cross-tabulations and correlations do not control for multivariate influences on child health outcomes. For a number of reasons, multiple logistic regression was determined to be the most appropriate strategy for addressing whether empirical associations exist between socioeconomic factors, demographic factors, and child undernutrition. Logistic regression is considered appropriate where the goal is to predict the presence or absence of a discrete outcome (dependent outcome variables) according to a selection of covariates (explanatory variables). Moreover, the objective of LR is to model the odds of the outcome as a function of the independent variable(s), and to then express the results in terms of odds ratios to estimate the magnitude of differentials between the different outcomes (Hosmer and Lemeshow 2000). Unlike linear regression involving continuous data, LR is a nonparametric test in that it does not have distributional requirements for predictor variables, does not assume linearity of relationship between the predictor and outcome variables, and can involve a mix of continuous, discrete, and dichotomous predictors (Garson 2008; Tabachnick and Fidell 2007). The multiple logistic regression model can be described a follows.39 Consider a grouping of p independent variables indicated by the vector x' = (xi, X2, .. .xp). The logistic regression model is then specified as: gg(*) 7l(x) = ' whereby the conditional probability that the outcome is present is indicated by P( F=llx) = jt(x). As such, the logit of the MLR model is given by the equation g(x) = /i + Pix, + fi2x2 +... + PpXp 39 Source: Hosmer and Lemeshow 2000. 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Each of the three empirical models was fitted using manual logistic regression methods (that is, as opposed to automated stepwise methods). Manual entering of variables is performed on the basis of the posited associations between the independent and dependent variables. The overall fit of the model is measured using the Nagelkerke r-squared statistic, which is a generalized r-squared statistic, which also measures the maximum likelihood estimation of the model in terms of proportion (that is, a value of 1 indicates the model predicts the variance perfectly). Three models were analyzed, whereby each anthropometric variable, namely stunting (model 1), underweight (model 2), and wasting (model 3), was regressed on all of the independent variables comprising: child's age, child's sex, maternal orphan, paternal orphan, double orphan, household wealth status, and mother's education.40 Descriptive statistics as well as the results of the Spearman correlations and multivariate analyses are presented in Chapter Four, Results and Discussion. 40 During the model fitting process, three other models were also tested whereby each anthropometric variable was regressed only on the three orphanhood variables comprising maternal orphanhood (morphan), paternal orphanhood (porphan), and double orphanhood (dorphan). Due to low explanatory value, those models are not reported in this thesis. 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER FOUR Results and Discussion This chapter provides a variety of empirical results. Firstly, sample characteristics for each country are presented. Secondly, descriptive statistics are presented for each country, followed by cross tabulations and bivariate correlations. Finally, results and discussion of multiple logistic regressions are presented for each country. 4.1 Sample characteristics Of the 6,175 households selected for The Gambia's full sample, 6,071 were interviewed (98.4% response rate). Within these households, 9,982 women 15 to 49 years old were interviewed (97.4% response rate). In addition, of the 6,641 children under five years old listed in the household questionnaire, 6,543 child questionnaires were completed (98.5% response rate) (Gambia Bureau of Statistics, UNICEF, and The World Bank 2007). All 7,600 Ivorian households in the full sample were interviewed (100% response rate). Within these households, 12,888 women aged 15 to 49 were interviewed (99% response rate), and under five child questionnaires were completed for 8,604 children (99% response rate) (INS Cote d'lvoire 2007). In Sierra Leone, 7,078 households were interviewed (99.3% response rate), and within those households, 7,654 women aged 15 to 49 were interviewed (82.7% response rate). Completed under five child questionnaires numbered 5,246 (88.9% response rate) (Statistics Sierra Leone and UNICEF-Sierra Leone 2007). Of the Somali households selected for the full sample 5,969 were interviewed (99.5% response rate). Of these, 6764 were women interviewed (aged 15 to 49) (93% response rate), 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and under five child questionnaires were completed for 6,305 children (98.9% response rate) (UNICEF Somalia 2006). 4.2 Descriptive results Table 4.1 and figures 4.1 to 4.6 present descriptive statistics for socioeconomic, demographic, and anthropometric status among children 0 to 59 months of age for all four countries. Histograms of the raw z-scores of children's anthropometric data can be found in the Appendix (Figures A1 to C4). As shown in Table 4.1, there is relatively little variation in age distribution between countries, although contrasting results are emphasized as follows. Sierra Leone has the smallest proportion of children 0 to 11 months of age (18.7%) in contrast to The Gambia, where 23.7% of children are in their first year of life. Somalia has the smallest proportion of children 12 to 23 months of age (17.3%), whereas Gambian children in their second year of life comprise 22.6% of the sample. The least amount of variation between countries is revealed in the 24 to 35 months age category, whereby Ivorian children in their third year of life comprise the smallest proportion of all four countries' samples at 19.7%, in contrast to Gambian children of the same age group comprising the highest proportion at 21.0% of the sample. Gambian children in their fourth year of life (36 to 47 months) comprise 19.0% of sample, in contrast to Sierra Leone at 23.4%, which is also the most populated age category for Sierra Leone. The Gambia has the smallest proportion of children 48 to 59 months of age (13.7%) which is in stark contrast to Somalia where 20.6% of children belong to that age group (Table 4.1 and Figure 4.1). 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.1. MICS3 descriptive statistics, children under five, for select countries of subSaharan Africa Variable and category The Gambia (2005/2006) n=6641 # of % children 23.7 1551 1481 22.6 1373 21.0 1242 19.0 896 13.7 Child's age (mos.) Oto 11 12 to 23 24 to 35 36 to 47 48 to 59 Child's sex Male 3343 Female 2300 Orphanhood status Maternal 37 Paternal 115 Double 8 Non-orphan 6497 Child's anthropometric status Stunted 1465 Underweight 1341 Wasted 412 Household wealth (t juintile) First 1575 Second 1342 Third 1351 Fourth 1249 1026 Fifth Mother's education level None/limited 4975 Primary level 690 Secondary+ 878 Cote d'lvoire (2006) n=8650 # of % children 2009 23.3 1757 20.4 1693 19.7 19.1 1642 17.5 1503 Sierra Leone (2005) n=5904 # of % children 1105 18.7 1140 19.3 1170 19.8 1379 23.4 1110 18.8 Somalia (2006) n=6373 # of % children 1344 21.3 1093 17.3 1257 19.9 1311 20.8 1300 20.6 51.1 48.9 4413 4191 51.3 48.7 2936 2968 49.7 50.3 3307 3066 51.9 48.1 0.6 1.8 0.1 97.8 45 172 8 8441 0.5 2.0 0.1 97.6 105 211 36 5624 1.8 3.6 0.6 95.3 69 198 21 6127 1.1 3.1 0.3 96.1 22.9 20.9 6.5 2969 1813 616 35.6 21.4 7.5 1752 1436 373 40.5 32.2 8.7 1990 2003 673 36.0 35.4 12.2 24.1 20.5 20.6 19.1 15.7 2136 1916 1825 1661 1112 24.7 22.2 21.1 19.2 12.9 1263 1360 1306 1159 816 21.4 23.0 22.1 19.6 13.8 1167 1251 1322 1423 1210 18.3 19.6 20.7 22.3 19.0 76.0 10.5 13.4 5698 2193 757 65.9 25.4 8.8 4747 628 527 80.4 10.6 8.9 5528 918 206 82.3 14.5 3.2 Regarding the distribution of male versus female children, Somalia has the highest proportion of males at 51.9%, whereas Sierra Leone has the highest proportion of females at 50.3% (Table 4.1 and Figure 4.2). 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.1 Age distribution of children under five, for select countries of sub-Saharan Africa Age (months) • 0-11 § 15.0 - H 12-23 o 10.0 • 24-35 • 36-47 • 48-59 Gambia (2005/2006) Cote d'lvoire (2006) Sierra Leone (2005) Somalia (2006) Figure 4.2 Distribution of male and female children under five, for select countries of sub-Saharan Africa 53.0 -| 52.0 51.0 Sex 50.0 • Male 49.0 0 Female 48.0 47.0 46.0 I I 1 Gambia Cote d'lvoire Sierra Leone (2005/2006) (2006) (2005) Somalia (2006) 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.3 Orphanhood distribution, children under five, for select countries of sub-Saharan Africa 4 i 3.5 Orphanhood status 3 - • Paternal orphans H Maternal orphans • Double orphans 0 Gambia Cote d'lvoire Sierra Leone (2005/2006) (2006) (2005) Somalia (2006) In terms of orphanhood status, paternal orphans are more common among all samples than other types of orphans, and proportions are comparable across the four countries studied (The Gambia 1.8%; Cote d'lvoire 2.0%; Sierra Leone 3.6%; and Somalia 3.1%) (Table 4.1 and Figure 4.3). This could be partially attributed to children's fathers often being significantly older than their mothers due to cultural norms (and therefore dying sooner), as well as conflict - common in all four countries - being a factor of premature death among prime-aged males. Sierra Leone shows the highest proportion of all types of orphans (maternal at 1.8%, paternal at 3.6%, and double orphans at 0.6%). Cote d'lvoire has the lowest proportion of maternal orphans at 0.5%, whereas The Gambia has the lowest proportion of paternal orphans at 1.8%. The Gambia and Cote d'lvoire both share the lowest proportion of double orphans at 0.1%. 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.4 Proportion of undernourished children under five years of age, for select countries of sub-Saharan Africa 45.0 40.0 Anthropometric status • Stunted H Underweight • Wasted Gambia Cote d'lvoire Sierra Leone (2005/2006) (2006) (2005) Somalia (2006) Undernutrition is widespread in all four countries examined, with stunting being the most common. However, there is notable variation in the percentages of the samples afflicted by stunting, with the lowest percentage being 22.9% in The Gambia, and the highest being 40.5% of Sierra Leone's sample (Cote d'lvoire 35.6%, and Somalia 36.0%). Regarding underweight children, Somalia's sample reveals the highest proportion at 35.4%, in contrast to The Gambia at 20.9%. The highest percentage of wasted children is found in Somalia at 12.2% which is nearly double that of Gambia children who are wasted (6.5%) (Table 4.1 and Figure 4.4). 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.5 Distribution of children under the age of five, by household wealth, for select countries of sub-Saharan Africa 30.0 Wealth quintile 25.0 • First 20.0 - <* 15.0 ^ 10.0 - Hi Second • Third E3 Fourth • Fifth Gambia Cote d'lvoire Sierra Leone (2005/2006) (2006) (2005) Somalia (2006) The share of children belonging to the five household wealth status categories across all four countries is relatively consistent, although there are certain observations worth noting. Overall, results for The Gambia and Sierra Leone appear to fall between Somalia and Cote d'lvoire. Also, Somalia shows the most equitable distribution of children under five among the five wealth status categories, with the exception of the fourth wealth quintile which accounts for a slightly larger percentage of children at 22.3% (the other quintiles measure closer to the 20% mark); the distribution of Ivorian children across the wealth categories appears to be the least equitable, with the exception of the fifth (richest) quintile, which only 12.9% of the children belong to, representing the smallest percentage of children from wealthy households across the four countries. Not surprisingly, children from the richest quintile account for the smallest proportion all four samples, with the exception of Somalia. Conversely, 24.7% of Ivorian children belong to the poorest wealth quintile, representing the highest such percentage among the four countries (Table 4.1 and Figure 4.5). 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.6 Distribution of children under the age of five, by mother's education, for select countries of sub-Saharan Africa 90.0 80.0 70.0 Mother's education aj 60.0 On 50.0 • None/limited H Primary o 40.0 £ 30.0 • Secondary+ Gambia Cote d'lvoire Sierra Leone (2005/2006) (2006) (2005) Somalia (2006) Regarding maternal education, the vast majority of mothers of children in the countries studied have no or limited formal schooling, with Somalia bearing the greatest percentage at 82.3%, in contrast to Cote d'lvoire at 65.9%. Cote d'lvoire bears the greatest percentage of children whose mothers have primary level education (25.4%), while The Gambia shares the lowest percentage with Sierra Leone at 10.5% (Sierra Leone is 10.6%). However, Gambian mothers who have secondary or higher education comprise 13.4% of the sample (the highest out of the four countries studied), while Somali children have the smallest share of mothers with secondary or a higher level of education at 3.2% (Table 4.1 and Figure 4.6). 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3 Cross-tabulations The full results for stunting, underweight, and wasting each cross-tabulated with the independent variables are presented in tables 4.2 to 4.4. Select results are described below. Table 4.2 Cross-tabulations of the presence of stunting, and socioeconomic and demographic characteristics, children under five, for select countries of sub-Saharan Africa Variable and category The Gambia (2005/2006) n==6641 # Child age (mos.) 174 Oto 11 12 to 23 435 24 to 35 352 36 to 47 311 48 to 59 193 Child's sex 752 Male Female 713 Orphanhood status Maternal 12 29 Paternal Double 3 Non-orphan 1427 Household wealth (quintile) 472 First (poorest) Second 361 Third 291 Fourth 223 Fifth (richest) 118 Mother's education level None/limited 1209 Primary level 129 Secondary+ 127 % Cote d'lvoire (2006) n==8650 Sierra Leone (2005) n=5904 Somalia (2006) n==6373 # % # % # % stunted within category 11.5 30.3 26.5 25.5 22.0 337 703 634 691 604 stunted within category 17.3 41.2 38.4 43.6 41.6 189 407 415 468 273 stunted within category 21.0 45.0 46.7 49.6 39.7 191 355 500 505 439 stunted within category 16.1 36.6 44.3 44.7 39.4 22.9 22.9 1618 1351 37.7 33.4 897 855 41.8 39.2 1046 944 36.5 35.5 35.3 25.2 37.5 22.8 23 63 2 2885 52.3 38.9 25.0 35.4 27 63 8 1670 39.7 42.0 33.3 40.4 20 59 5 1916 34.5 36.6 27.8 36.0 30.7 27.7 22.1 18.2 11.6 861 735 644 455 274 42.1 39.7 36.9 28.3 25.2 379 424 384 370 195 44.1 44.6 42.6 41.6 26.9 472 510 451 331 226 46.6 46.0 40.4 27.5 20.8 25.0 19.0 14.6 2024 731 214 37.0 34.4 28.8 1453 156 142 42.8 33.4 30.7 1751 196 34 38.6 24.4 19.4 Stunting. Overall, Sierra Leonean children are shown to be the most vulnerable to stunting, while Gambian children appear to be the least at risk (Table 4.2). In terms of patterns in relation to age, in The Gambia and Cote d'lvoire, there is a stark increase in vulnerability to stunting in the second year of life, although the risk for Gambian children then decreases 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. thereafter as the child ages. In contrast, for Ivorian children, there are fluctuations in risk as the child ages, and the risk of stunting is actually highest in children 36 to 47 months of age. As with The Gambia and Cote d'lvoire, the cross tabulations for Sierra Leone and Somalia also demonstrate a spike in vulnerability to stunting in children 12 to 23 months of age. However, unlike the fluctuating results for Cote d'lvoire, the jump in vulnerability to stunting in Sierra Leone and Somalia is followed by a steady increase in risk until the onset of the fifth year of life at which time the risk is lower than the previous year. Furthermore, with the exception of The Gambia, boys appear to be more vulnerable to stunting than girls, and the interactions between orphanhood and stunting reveal no clear apparent trends. Certain conspicuous results, however, warrant noting. For instance, Ivorian maternal orphans face the greatest of risk stunting in comparison to other children in the four countries studied, followed by paternal orphans in Sierra Leone; Gambian non-orphans are the least vulnerable. Finally, as expected, in all four countries studied the risk of being stunted decreases as household wealth and mother's education increase. Underweight. Overall, Somali children are shown to be the most vulnerable to being underweight, while Gambian children appear to be the least at risk (Table 4.3). Resembling the results for stunting, the risk of being underweight jumps considerably in the second year of life across all four countries. There is also a steady decline in risk with each successive age category across all four countries. It is also worth nothing that Somalia is the only country for which the risk of being underweight in higher in the fifth year of life than the second; for all other countries it is considerably lower. As with stunting, boys appear to be more vulnerable to being underweight than girls in all four countries studied, and the associations between orphanhood and underweight are 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mixed with no clear patterns. It is important to acknowledge, however, that Gambian double orphans face the greatest risk of being underweight in comparison to other children in the four countries studied. Finally, also as with stunting, in all four countries studied the risk of being underweight decreases as household wealth and mother's education increase. Table 4.3 Cross-tabulations of the presence of underweight, and socioeconomic and demographic characteristics, for select countries of sub-Saharan Africa Variable and category The Gambia (2005/2006) n=6641 # Child age (mos.) Oto 11 168 12 to 23 469 24 to 35 336 36 to 47 216 48 to 59 152 Child's sex Male 695 Female 646 Orphanhood status Maternal 9 Paternal 28 Double 4 Non-orphan 1308 Household wealth (t iuintile) First (poorest) 413 Second 305 290 Third 198 Fourth Fifth (richest) 135 Mother's education level 1095 None/limited Primary level 137 Secondary+ 109 % Cote d'lvoire (2006) n=8650 Sierra Leone (2005) n=5904 Somalia (2006) n=6373 # % # % # % under­ weight within category 11.1 32.2 25.0 17.7 17.3 265 523 424 337 264 under­ weight within category 13.3 30.1 25.5 21.0 18.0 200 442 343 285 166 under­ weight within category 21.5 46.9 37.6 29.3 23.4 160 378 513 494 458 under­ weight within category 13.2 37.8 44.9 42.8 40.0 21.1 20.6 987 826 22.7 20.1 744 692 33.6 30.7 1074 929 36.6 34.1 26.5 24.3 50.0 20.8 9 24 1 1781 20.5 14.6 12.5 21.6 22 49 8 1373 31.0 32.0 29.6 32.2 17 64 2 1924 29.3 39.5 11.1 35.3 26.6 23.3 22.0 16.1 13.3 564 462 399 257 131 27.2 24.6 22.5 15.7 11.9 337 326 311 297 165 37.9 33.5 33.4 32.1 22.1 499 531 463 336 174 48.0 46.8 40.1 27.3 15.8 22.5 20.1 12.5 1282 436 95 23.0 20.3 12.7 1175 142 119 33.6 29.3 24.8 1785 183 27 38.4 22.6 15.2 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4 Cross-tabulations of the presence of wasting, and socioeconomic and demographic characteristics, for select countries of sub-Saharan Africa The Gambia (2005/2006) n==6641 Sierra Leone (2005) n==5904 Somalia (2006) n=6373 % wasted # within category # % # % wasted within category 6.5 12.1 4.3 3.4 4.6 212 233 82 55 34 wasted within category 11.4 13.7 5.0 3.5 2.3 81 136 71 47 38 wasted within category 9.4 15.1 8.1 4.9 5.5 127 160 129 131 126 10.9 16.7 11.5 11.5 11.3 6.7 6.1 344 272 8.1 6.8 211 162 9.9 7.5 375 298 13.1 11.3 5.9 3.5 12.5 6.5 2 5 0 609 4.5 3.1 0 7.6 6 14 1 354 8.7 9.7 4.0 8.6 6 30 0 637 10.5 18.5 0 12.0 7.1 7.4 7.6 4.3 5.4 166 140 145 100 65 8.2 7.6 8.4 6.3 6.1 71 83 88 69 62 8.5 8.9 9.8 7.8 8.5 152 160 165 136 60 15.0 14.5 14.8 11.4 5.6 6.5 6.9 5.7 426 146 44 7.9 7.0 6.0 287 51 35 8.6 10.9 7.6 587 72 14 13.0 9.2 % # Child age (mos.) Oto 11 98 12 to 23 175 24 to 35 57 36 to 47 42 48 to 59 40 Child's sex Male 221 Female 191 Orphanhood status Maternal 2 Paternal 4 Double 1 Non-orphan 407 Household wealth (quintile) First (poorest) 109 Second 96 Third 100 Fourth 52 Fifth (richest) 55 Mother's education level No/limited schooling 315 Primary level 47 Secondary+ 50 C6te d'lvoire (2006) n==8650 OO o Variable and category Wasting. As with underweight, Somali children are shown to be the most vulnerable to being wasted, while Gambian children appear to be the least at risk (Table 4.4). For all four countries studied, as with other forms of undernutrition, the risk of wasting jumps in the second year of life, although the discrepancy is not as pronounced as it is for stunting and underweight. Also, with the exception of The Gambia which shows some minor fluctuation, the risk of wasting then steadily decreases after 12 to 23 months of age. It is also important to acknowledge that Somalia is the only country for which the risk of wasting is higher in the fifth year of life, than in the first. 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As with other forms of undernutrition, boys in all four countries are more vulnerable to wasting than girls, and results for orphanhood are mixed. Out of the four countries studied, Somali orphans appear to be the most vulnerable to wasting overall, especially paternal orphans (Table 4.4). Results for interactions between wealth and wasting are also varied, showing fluctuations in percentages within three of the four countries. This could be indicative of the relatively limited responsiveness of wasting to wealth given that it is a measure of acute undernutrition, which could result from acute illness and/or other circumstances that can afflict wealthy and poor households alike. Regarding mother's education, overall as education increases wasting risk decreases, although there are some fluctuations in the percentages. Results of the cross-tabulations are an early indication of associations that can be foreseen throughout additional empirical results of this thesis, beginning with Spearman's rho correlations. 4.4 Spearman's rho Correlations Table 4.5 presents Spearman's rho correlations between stunted, underweight, and wasted, and socioeconomic and demographic factors of undernutrition for all four countries. The results indicate that child age, orphanhood status, wealth status, and mother's education are significantly correlated with undernutrition. Results for all four countries show that age is positively associated with stunting, while wealth and education are negatively associated, as expected. Similarly, wealth and education are negatively associated with underweight in all four countries. Results for age and underweight are mixed, however, in that while both Sierra Leone and Somalia exhibit statistically significant coefficients for age, Sierra Leone's result is negative but Somalia's is 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.5 Correlation coefficients (Spearman's rho) between socioeconomic and demographic factors, and stunted, wasted, and underweight, for select countries of sub-Saharan Africa Variable Age GAMBIA COTE D'lVOIRE SIERRA LEONE (2005) (2005/2006 (2006) Stunted Under­ Wasted Stunted Under­ Wasted Stunted Under­ Wasted weight weight weight .087" .006 -.083" .159** .013 -.155" .166" -.036* -.122 SOMALIA (2006) Stunted Under­ Wasted weight .174" .169" -.022 -.017 .027* -.013 -.024* .012 .002 .008 .002 .008 .035" Household Wealth Mother's Education -.149" -.111" -.042" -.125" -.122 -.025* -.091" -.071" -.008 -.216" -.246*" -.090" -.077" -.006 -.051" -.063" -.022* -.098" -.058" .008 -.127" -.140" -.046** » .019 " .022 fov"o o * Orphanhood 1 **P < 0.01 *P < 0.05 (2-tailed) 78 positive. Given that underweight measures both chronic and acute undernutrition, it is not surprising that results are varied, since a child can experience acute undernutrition at any age, in contrast to stunting which is indicative of chronic undernutrition over longer periods of time. Coefficients for wasting are also varied. The Gambia, Cote d'lvoire, and Sierra Leone exhibit significant negative correlations between wasting and age, while The Gambia, Cote d'lvoire, and Somalia all show significant negative coefficients for household wealth. In addition, mother's education is significantly negatively correlated with wasting only in Cote d'lvoire and Somalia. Oddly, orphanhood is shown to be significantly negatively associated with wasting in Cote d'lvoire, but positively associated with wasting in Somalia. These results are an early indication of significant associations in the regression results. The following sub-section presents multiple logistic regression results for each country, followed by critical analyses (discussion) of the results as they relate to the literature presented in Chapter Two. 4.5 Multiple Logistic Regression Three models per country were analysed using multiple logistic regression. Models 1, 2, and 3 estimate the relative influence of socioeconomic and demographic variables, along with orphanhood, on stunting, underweight, and wasting, respectively. Table 4.6 presents multiple logistic regression results for model 1, followed by descriptions and analysis of the regression results. Results are described at the 1% level of significance unless otherwise specified. 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.6 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of stunting for select countries of sub-Saharan Africa (model 1) Variable Child's age 2:1 Child's age 3:1 Child's age 4:1 Child's age 5:1 Sex (:male) Maternal orphan Paternal orphan Double orphan Household wealth 1:5 Household wealth 2:5 Household wealth 3:5 Household wealth 4:5 Mother's education 2:1 Mother's education 3:1 Constant Nagelkerke R square THE GAMBIA (2005/2006) OR CI **3.386 2.782—4.121 **2.768 2.261-3.388 **2.705 2.200-3.326 **2.305 1.835-2.896 .969 .859-1.093 1.520 .662-3.491 1.014 .642-1.601 1.511 .261-8.765 **3.119 2.464-3.948 **2.770 2.180-3.520 **2.011 1.579-2.563 **1.619 1.263-2.075 .819 .664-1.009 .759 .612-942 **.238 .084 COTE D'lVOIRE (2006) CI OR **3.410 2.923-3.977 **3.019 2.584-3.527 **3.782 3.236-4.419 **3.422 2.918—4.013 **1.211 1.103-1.329 *2.792 1.407-5.540 1.081 .773-1.512 .178 .030-1.057 **2.184 1.834-2.600 **1.986 1.667-2.367 **1.762 1.477-2.103 1.155 .963-1.385 1.001 .897-1.118 .942 .785-1.130 **.454 .094 SIERRA LEONE (2005) CI OR **3.216 2.607-3.969 **3.544 2.869-4.378 **3.921 3.183-4.829 **2.663 2.127-3.335 *1.160 1.022-1.316 .995 .530-1.868 1.050 .725-1.520 .265-2.647 .838 **2.063 1.629-2.614 **2.076 1.645-2.619 **1.994 1.579-2.517 **1.915 1.525-2.405 **.676 .547-836 .823 .646-1.047 **.516 .094 SOMALIA (2006) CI OR **3.258 2.644-4.013 **4.327 3.543-5.284 **4.538 3.714-5.545 **3.520 2.876-4.308 1.045 .930-1.174 .956 .490-1.863 .880 .616-1.256 .648 .178-2.360 **3.075 2.496-3.787 **3.058 2.493-3.749 **2.523 2.061-3.088 **1.429 1.166-1.751 **.724 .600-.874 *.630 .423-940 **.422 .140 **/><0.01 *P < 0.05 Data are expressed as odds ratios (OR) (95% confidence interval [CI]). 80 4.5.1 Model 1 Results: Socioeconomic and demographic risk factors of stunting Table 4.6 presents logistic regression results for The Gambia, Cote d'lvoire, Sierra Leone, and Somalia regarding socioeconomic and demographic risk factors, including orphanhood (estimated by odds ratios [OR]), of stunting in children. The Gambia. Child age and household wealth are found to be statistically significant risk factors of early childhood stunting in The Gambia (Table 4.6). For instance, children in the second youngest age group (12 to 23 months) demonstrate a three-fold increased risk of stunting in contrast to the youngest children (0 to 11 months) as indicated by an odds ratio (OR) of 3.39 (95% CI 2.78^.12), followed by a gradual decline in risk as the child ages (although the risk is still higher than the reference category). Children in the poorest households are three times more likely to be stunted than children in the wealthiest households (OR 3.12, 95% CI 2.46-3.95), and as expected, the risk of stunting decreases with each successive wealthier quintile. Cote d'lvoire. Child age, sex, maternal orphanhood, and household wealth are found to be statistically significant risk factors of early childhood stunting in Ivorian children (Table 4.6). Risk of stunting is greater for older children. For example, Ivorian children in the second oldest age group (36 to 47 months) face much higher odds of stunting compared with the youngest children (0 to 11 months) as indicated by an odds ratio of 3.78 (95% CI 3.24-4.42); these children face the highest risk of stunting than any other age group of Ivorian children studied. Moreover, male children are found to be slightly more vulnerable to stunting than female children, as indicated by an odds ratio of 1.21 (95% CI 1.10-1.33). 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Maternal orphans are nearly three times more vulnerable to stunting than other children in the sample, as indicated by an odds ratio of 2.79 (5% level of significance; 95% CI 1.41-5.54). In terms of household wealth status, Ivorian children in the poorest households are more than twice as likely to be stunted as children in the wealthiest households (OR 2.18, 95% CI 1.83-2.60), and as expected, the risk of stunting decreases with each successive wealthier quintile. Sierra Leone. Child age, sex, household wealth, and mother's education are found to be statistically significant risk factors of early childhood stunting in Sierra Leonean children (Table 4.6). Children in the second oldest age group (36 to 47 months) face much higher odds of stunting compared with the youngest children (0 to 11 months) as indicated by an odds ratio of 3.92 (95% CI 3.18-4.83); these children face the highest risk of stunting than any other age group in Sierra Leone. Moreover, male children are found to be 16% more vulnerable to stunting than female children, as indicated by an odds ratio of 1.16 (5% level of significance; 95% CI 1.02-1.32). As expected, the risk of stunting decreases with successive wealthier quintiles. Vulnerability to stunting is highly comparable across all wealth quintiles, ranging from the lowest OR of 1.92 (95% CI 1.53-2.41) for the fourth wealth quintile, to the highest OR of 2.08 (95% CI 1.65-2.62) for the second wealth quintile. Regarding maternal education, the children of mothers who have primary education face the least amount of risk of stunting in contrast to children whose mothers have no or limited education (OR .676, 95% CI .547.836). 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Somalia. Child age, household wealth, and mother's education are found to be statistically significant risk factors of early childhood stunting in Somalia (Table 4.6). For instance, children in the second oldest age group (36 to 47 months) are more than four and a half times more vulnerable to stunting in contrast to the youngest children (0 to 11 months), as indicated by an odds ratio 4.54 (95% CI 3.71-5.55). Results are varied across age groups. Children in the poorest households are three times more likely to be stunted than children in the wealthiest households (OR 3.08, 95% CI 2.50-3.79), and as expected, the risk of stunting decreases with each successive wealthier quintile. Furthermore, the children of mothers who have the highest level of education face the least risk of stunting in contrast to children whose mother have no or limited education (5% level of significance; OR .630, 95% CI .423-940). Children of mothers who have primary education are also less likely to be stunted than children whose mother have no or limited education (OR .724, 95% CI .600.874). The following section presents the author's analysis of the model 1 results in terms of expected versus unexpected results, linkages with key research questions and literature presented throughout this thesis, and any variation across the countries. 4.5.2 Model 1 Discussion: Socioeconomic and demographic risk factors of stunting The author's interpretation of each model considers the core objective of this thesis, namely to explore which socioeconomic and demographic characteristics, including different degrees of orphanhood, are risk factors of child undernutrition. This structure will be repeated for models 2 and 3, although greater detail is provided in discussion for models 1 and 2 so as to limit redundancies in the interpretation of associations between variables. 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Overall, maternal orphanhood is unreliable as a risk factor for stunting, given that it is only Ivorian maternal orphans who are shown to be more vulnerable than other children. Notwithstanding, the finding that maternal orphans are more vulnerable to undernutrition is consistent with research by Gertler et al. (2003), who found positive associations between a mother's death and increased probability of her surviving children being undernourished. Moreover, Ainsworth and Semali's (2000) longitudinal research (1991 to 1994) on the consequences of parental death in Kagera, Tanzania revealed that children who lost their mothers were much more likely to be stunted than children who lost their fathers. Beegle, De Weerdt, and Dercon (2006) also discovered an association between maternal orphanhood and height deficiencies among older children (11 to 18 years old) using data from the same study. A study by Case, Lin, and McLanahan (2000) also warrants attention, due to its relevance as evidence of the importance of mothers for child nutrition. Their study of South African households revealed that".. .the presence of a child's biological mother increases expenditure, in particular, on healthy foods. The benefits are limited to households in which mothers have control over food expenditure, and they are limited to children in the youngest age groups. In this way, biological mothers protect their offspring during the children's most vulnerable years" (797). In addition, recall the theory of orphan care which posits a disintegration of social safety nets in the wake of high rates of premature adult deaths in sub-Saharan Africa. Bear in mind Evans' (2005) explanation whereby the community suffers a productivity loss related to an adult's death, although the surviving child(ren) can often rely on the surviving parent for their care. Many orphans must also rely on the community for support; however if the community or other social safety nets like the extended family have been weakened by 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. poverty or another stressor such as conflict, orphaned children will likely be more vulnerable to both short- and long-term undernutrition and other health hazards. Child age is a reliable risk factor for stunting, and odds ratios are relatively consistent across all four countries studied, with children in the 12 to 23 month age category being three times more vulnerable to stunting than infants (0 to 11 months). The sharp increase in vulnerability associated with the 12 to 23 months age category across all four countries may be associated with weaning - commonly practiced in the second year of life - and the child's subsequent increased reliance on solid foods which may be in short supply and/or lacking essential micronutrients (World Bank 2005). Moreover, the prevalence in stunting across the older age categories may be explained by chronic undernutrition associated with long term poverty, as opposed to measures of shortterm nutrition shocks such as wasting, and to a lesser degree underweight (which captures both chronic and acute undernutrition). Accordingly, stunting is a well-established indicator of chronic undernutrition related to socioeconomic and environmental conditions (deOnis, Frongillo, and Blossner 1997). This finding is supported by the literature. It is also worth noting, that the risk of stunting across all four countries was highest in Somalia, and in particular, among Somali children in the 36 to 47 months age category. Somalia is among the poorest of the four countries studied and ranks 204th out of 206 countries on the Human Development Index (UNDP 2009). Male children in Cote d'lvoire and Sierra Leone, but not The Gambia or Somalia, are revealed to be slightly more vulnerable to stunting. Such variation in stunting risk between the sexes across the four countries studied could be attributable to numerous factors, one being differences in cultural values associated with the relative status attributed to differing genders. As such, in some countries, it is possible that male children are favoured in terms of 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. nutrition investments, while in others females are favoured, depending on social hierarchy and familial roles and expectations. For example, Boserup's theory contends that agricultural practices determine differential treatment of males and females, depending on their roles in food production (Svedberg 1990). The vulnerability of boys versus girls to undernutrition has been debated in the literature since around the early 1990s, although the evidence pertaining to SSA largely suggests boys are more vulnerable. In his thorough empirical analysis of anthropometric status of male versus female children in SSA, Svedberg (1990) found that females' status was equivalent and sometimes better than males. More recently, Sahn and Stifel (2002) also found that with respect to African children "gender plays an important role in determining nutritional outcomes, with girls achieving better outcomes than boys" (42). This is also corroborated by Christiaensen and Alderman (2004) in their study of child malnutrition in Ethiopia. Household wealth is a reliable predictor of stunting. This result is expected given the relationship between stunting and chronic poverty, and is consistent with the literature. For instance, Kennedy et al. (2006) found that being poor increased the risk of stunting in Angola by more than three times, by three times in Senegal, and by one and a half times in Central African Republic. They also found that poor children living in an urban area did not confer any advantages for an increased height for age. Also, armed conflict, a hazard especially in Sierra Leone and Somalia at the time the data for this thesis were collected, has adverse implications for food security, namely the disruption of food production and supplies, as well as interference with breastfeeding (Machel 2000). Machel explains, "During conflicts, breastfeeding is often interrupted by separation, maternal trauma and exhaustion" (21). As such, the greater risk of stunting among Somali children in older age groups in particular may 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. also suggest that weaning is purposefully delayed in order to extend nutritional security during times when other food sources are lacking due to conflict and/or other adverse conditions. Mother's education is negatively associated with stunting. This is an expected result for several reasons. Recall from comments presented earlier in Chapter Three that a more educated parent is more likely to earn a higher wage, thereby affording them better access to nutrition and health care for themselves and their family. Further, enhanced literacy and numeracy increases a parent's capacity and receptivity to diagnose and treat child health problems through modern medicine Glewwe (1999). 4.5.3 Model 2 Results: Socioeconomic and demographic risk factors of underweight Recall that underweight is a measure of both acute and chronic undernutrition. Table 4.7 presents logistic regression results for the four countries of this study regarding socioeconomic and demographic risk factors including orphanhood, of underweight in children. The Gambia. Child age, household wealth, and mother's education are found to be statistically significant risk factors of underweight among Gambian children (Table 4.7). For instance, Gambian children in the second youngest age group (12 to 23 months) demonstrate a four-fold increased risk of being underweight in contrast to the youngest children (0 to 11 months) as indicated by an odds ratio of 3.91 (95% CI 3.21-4.76), followed by a gradual decline in risk as the child ages (although the risk is still higher compared to infancy). Regarding wealth status, children in the poorest households are twice as likely to be underweight as children in the wealthiest households (OR 2.08, 95% CI 1.65-2.61), and as 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.7 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of underweight for select countries of sub-Saharan Africa (model 2) Variable Child's age 2:1 Child's age 3:1 Child's age 4:1 Child's age 5:1 Sex (:male) Maternal orphan Paternal orphan Double orphan Household wealth 1:5 Household wealth 2:5 Household wealth 3:5 Household wealth 4:5 Mother's education 2:1 Mother's education 3:1 Constant Nagelkerke R square THE GAMBIA COTE D'lVOIRE SIERRA LEONE (2005/2006) (2006) 2005) OR **3.911 **2.640 **1.749 **1.721 1.008 .780 1.107 5.623 **2.075 **1.779 **1.622 1.153 .951 **.626 **.204 CI 3.213—4.760 2.151-3.239 1.404-2.178 1.354-2.189 .891-1.142 .288-2.116 .692-1.770 .918-34.436 1.648-2.614 1.405-2.254 1.281-2.053 .901-1.476 .773-1.169 .498-787 OR **2.842 **2.220 **1.745 **1.410 *1.170 1.163 *.633 .799 **2.537 **2.228 **2.048 *1.294 .923 **.665 ** {99 .082 **/><0.01 *P < 0.05 Data are expressed as odds ratios (OR) (95% confidence interval [CI]). CI 2.404-3.358 1.869-2.637 1.460-2.085 1.168-1.701 1.052-1.302 .522-2.593 .403-996 .079-8.047 2.043-3.152 1.791-2.772 1.642-2.555 1.027-1.631 .813-1.047 .525-842 .066 OR **3.362 **2.342 **1.551 1.166 *1.165 1.072 .999 1.148 **2.200 **1.745 **1.767 **1.664 .863 .888 **.380 CI 2.739-4.126 1.900-2.886 1.256-1.916 .921-1.477 1.023-1.326 .557-2.064 .676-1.476 .367-3.592 1.721-2.814 1.367-2.228 1.386-2.255 1.311-2.113 .696-1.071 .691-1.142 .076 SOMALIA (2006) OR **4.561 **5.806 **5.539 **4.741 1.120 .954 1.101 *.158 **4.598 **4.422 **3.472 **2.020 **.728 *.520 **.356 CI 3.670-5.668 4.705-7.165 4.486-6.839 3.837-5.857 .995-1.260 .487-1.869 .775-1.565 .030-.836 3.692-5.726 3.565-5.485 2.804^4.298 1.630-2.504 .600-.884 .334-810 .187 expected, the risk of being underweight decreases in wealthier quintiles, with no greater risk in the fourth quintile. Finally, in terms of maternal education, children of mothers who have the highest level of education face the least amount of risk of being underweight in contrast to children whose mothers have no or limited education (OR .626, 95% CI .498-787). Primary education does not seem to lower the risk compared to no/limited education. Cote d'lvoire. Child age, sex, paternal orphanhood, household wealth, and mother's secondary or higher education are found to be statistically significant risk factors of underweight in Ivorian children (Table 4.7). To explain, children in the second youngest age group (12 to 23 months) are nearly three times more vulnerable to being underweight than the youngest children (0 to 11 months), as indicated by an odds ratio of 2.84 (95% CI 2.40-3.36); these children face the highest risk of being underweight than any other age group of Ivorian children studied. Moreover, as age increases in the sample, the risk of being underweight declines. Regarding sex, male Ivorian children are found to be 17% more vulnerable to being underweight than female children, as indicated by an odds ratio of 1.17 (5% level of significance; 95% CI 1.05-1.30), a similar result to that of stunting. Strangely, paternal orphans are found to be less vulnerable to being underweight than other children, as indicated by an odds ratio of .633 (5% level of significance; 95% CI .403-996). Ivorian children in the poorest households are more than twice as likely to be underweight as children in the wealthiest households (OR 2.54, 95% CI 2.04-3.15), and consistent with other results pertaining to wealth, the risk of being underweight decreases with each successive wealthier quintile. Regarding mother's education as a risk factor for 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. being underweight, the children of mothers who have the highest level of education are the least vulnerable in contrast to children whose mothers have no or limited education (OR .665, 95% CI .525-842). Sierra Leone. Child age, sex, and household wealth are found to be statistically significant risk factors of early childhood underweight in Sierra Leonean children (Table 4.7). Again, children in the second youngest age group (12 to 23 months) face much higher odds of being underweight compared with the youngest children (0 to 11 months) as indicated by an odds ratio of 3.36 (95% CI 2.74-4.33); here too, the risk decreases for older children. Moreover, male children are found to be more vulnerable to being underweight than female children, as indicated by an odds ratio of 1.17 (5% level of significance; 95% CI 1.02-1.33), a similar result to that for stunting. As expected, the risk of being underweight decreases as wealth increases. Children in the poorest households are more than twice as vulnerable to being underweight than children in the wealthiest households (OR 2.20, 95% CI 1.72-2.81), while the risk is 67% higher for children in the fourth wealth quintile (OR 1.67, 95% CI 1.31-2.11). Somalia. Child age, double orphanhood, household wealth, and mother's education are found to be statistically significant risk factors of being underweight in Somalia (Table 4.7) To illustrate, children in the 24 to 36 months age group are nearly six times more vulnerable to being underweight in contrast to the youngest children (0 to 11 months), as indicated by an odds ratio 5.81 (95% CI 4.71-7.17). Results are varied across age groups, although the risk of being underweight is at least four and a half times as high in all age 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. groups. Though not expected, double orphans are considerably less likely to be underweight than other Somali children (5% level of significance; OR .158, 95% CI .030-836). Wealth poverty plays a more pronounced role in Somali children being underweight. Children in the poorest households are nearly five times more vulnerable to being underweight than children in the wealthiest households (OR 4.6, 95% CI 3.69-5.73), and as expected, the risk of being underweight decreases with each successive wealthier quintile, although it remains twice as high for the fourth quintile. The children of mothers who have the highest level of education are half as vulnerable to being underweight in contrast to children whose mothers have no or limited education (5% level of significance; OR .520, 95% CI .334-.810). Children of mothers who have primary education are also less likely to be underweight than children whose mothers have no or limited education (OR .728, 95% CI .600-.884). 4.5.4 Model 2 Discussion: Socioeconomic and demographic risk factors of underweight The results for sex as a predictor for underweight are consistent with those of stunting, with Ivorian and Sierra Leonean boys facing a higher risk of being underweight than girls. Recall from the discussion of model 1 that boys are expected to be slightly more vulnerable to undernutrition than girls; considering past scholarly debate surrounding anti-male versus anti female biases of intrahousehold allocation of resources, however, some mixed results for sex as a predictor of undernutrition are expected. Underweight is also responsive to age, although overall slightly lower odds ratios in the older age categories (in contrast to model 1, with the exception of Somalia) suggest that the risk of being underweight as children age is lower in contrast to stunting. The slight decreased responsiveness of underweight in relation to age is indicative of the variable's dual 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. capture of both chronic undernutrition and acute nutritional shocks that are not always dependent upon age; for example, weaning is dependent on age, while famine can afflict individuals young and old alike. Notwithstanding, several statistically significant associations are revealed, and as such, age is still considered a reliable predictor of being underweight in all four countries studied, with vulnerability decreasing overall with age (but still remaining higher than during infancy). The results for Somalia, however, diverge from this trend and the degree of risk is considerably higher than in the other countries studied, with odds ratios ranging between 4.56 (rounded) (12 to 23 months age category) to as high as 5.81 (rounded) (24 to 35 months) (Table 4.7). As with the increased risk of stunting in the second year of life, the sharp increase in vulnerability to being underweight associated with the 12 to 23 months age category across all four countries can be associated with weaning and increased reliance on solid foods (World Bank 2005). Note however, that the greatest risk of being underweight to Somali children occurs in the third year of life (24 to 35 months), as noted earlier. This could be related to the aforementioned conflict and overall destitute conditions of the country. Regarding results for orphanhood, in general there is no consistent, significant support for the impacts of orphanhood on underweight. This supports the orphan care theory that families and communities are coping with the influx of orphans in the wake of HIV/AIDS (and other health hazards) better than once perceived. This could indicate a "patching-up" of the social safety nets. The underweight variable is responsive to household wealth, although there is some variation between countries. This is consistent with the work of Kennedy et al. (2006) whose findings pertaining to disparities in nutritional status across Angola, Central African Republic and Senegal revealed an inconsistent relationship between wealth and weight for age (from 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. which the underweight variable is derived). This can be explained by underweight capturing chronic undernutrition, as well as acute nutritional deficiencies. Although in relation to stunting acute undernutrition is not as closely linked to chronic poverty, an individual whose immune system is already compromised - due to the stress of living in poverty or on its brink - is less resilient to nutritional shocks stemming from personal illness, household income shocks, and other acute source of undernutrition. Consistent with, and for the same explanations given for model 1, Somali children in the poorest households are the most at risk of being underweight, with the level of risk decreasing with each successive quintile, as expected. In contrast, Ivorian children are the least vulnerable to being underweight (Table 4.7). Recall from Table 3.2 that Cote d'lvoire has the highest GDP, life expectancy, and HDI ranking of the four countries studied. Regarding mother's education as a risk factor for underweight, with the exception of Sierra Leonean children (because the results were insignificant), children whose mothers who had at least primary education were less vulnerable to being underweight than children whose mothers had no or limited education. This is an expected result, given the association of education with wealth, as well as the implications of education for greater knowledge about child nutrition and care practices (Glewwe 1999), as explained in reference to model 1. 4.5.5 Model 3 Results: Socioeconomic and demographic risk factors of wasting Table 4.8 presents logistic regression results for the four countries of this study, regarding the socioeconomic and demographic risk factors of wasting among children. Recall that wasting is a measure of acute undernutrition. 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.8 Logistic regression results of socioeconomic and demographic risk factors (including orphanhood) of wasting for select countries of sub-Saharan Africa (model 3) Variable Child's age 2:1 Child's age 3:1 Child's age 4:1 Child's age 5:1 Sex (:male) Maternal orphan Paternal orphan Double orphan Household wealth 1:5 Household wealth 2:5 Household wealth 3:5 Household wealth 4:5 Mother's education 2:1 Mother's education 3:1 Constant Nagelkerke R square THE GAMBIA (2005/2006) CI OR **1.990 1.534-2.582 *.650 .464-910 **.522 .360-756 .709 .485-1.037 1.115 .911-1.366 .551 .073—4.145 .467 .147-1.490 9.796 .413-232.460 1.298 .904-1.863 .951-1.968 1.368 1.365 .955-1.952 .770 .516-1.148 1.088 .786-1.506 .939 .672-1.311 **.055 .050 COTE D'lVOIRE (2006) OR CI *1.242 1.017-1.517 **.405 .311—.527 **.278 .205-378 **.185 .128-268 *1.218 1.029-1.441 .899 .210-3.839 .196-1.197 .484 1.337 1.240 *1.421 1.014 .852 .788 **.051 .975-1.833 .901-1.706 1.035-1.951 .727-1.415 .697-1.041 .561-1.106 .080 SIERRA LEONE 2005) OR CI **1.729 1.287-2.321 .855 .611-1.196 ** 494 .339-719 *.562 .376-839 1.339 1.078-1.664 1.591 .614-4.124 1.426 .786-2.588 .250 .025-2.494 .675-1.490 1.003 1.026 .699-1.506 .755-1.602 1.100 .863 .588-1.266 1.295 .935-1.792 .888 .584-1.349 **.075 .044 SOMALIA (2006) OR CI **1.685 1.307-2.172 .794-1.346 1.034 1.071 .824-1.393 1.005 .770-1.312 *1.197 1.015-1.412 1.518 .626-3.686 *1.808 1.192-2.742 **3.030 **2.866 **3.000 **2.218 .913 .960 ** J14 2.168^.237 2.058-3.991 2.165—4.156 1.599-3.078 .694-1.200 .529-1.743 .040 **P<0.01 *P< 0.05 Data are expressed as odds ratios (OR) (95% confidence interval [CI]). .. .Odds ratios and confidence intervals are not computed because there are zero cases of wasted children belonging to this category. 94 The Gambia. Only child age appears to be statistically significant risk factor of wasting among Gambian children (Table 4.8). Children 12 to 23 months are twice as vulnerable to wasting as infants, as indicated by an odds ratio of 1.99 (95% CI 1.53-2.58). Older children, on the other hand, are less likely to be wasted compared to the youngest ones. This pattern may be indicative of weaning, followed by increased food security and resilience to acute illness. Cote d'lvoire. For this country, child age, sex, and the third wealth quintile are found to be statistically significant risk factors of wasting in Ivorian children (Table 4.8). As with Gambian children, Ivorian children in the second youngest age group (12 to 23 months) are more vulnerable to wasting that the youngest children (0 to 11 months) as indicated by an odds ratio of 1.24 (5% level of significance; 95% CI 1.01-1.52). The risk of wasting reverses for children who are 24 months and older (0 to 11 months). Moreover, male children are shown to be 22% more vulnerable to being wasted than female children, as indicated by an odds ratio of 1.22 (5% level of significance; 95% CI 1.03-1.44). Even though a statistically significant result is observed for the third wealth quintile, wealth does not generally affect the risk of wasting to the degree it does for stunting. Sierra Leone. Child age is the only statistically significant risk factor of early childhood wasting in Sierra Leonean children (Table 4.8). To illustrate, children in the second youngest age group (12 to 23 months) are nearly twice as vulnerable to wasting than the youngest children (0 to 11 months) as indicated by an odds ratio of 1.73 (95% CI 1.29-2.32). Also, as with the other countries observed thus far, the risk of wasting reverses as age increases, to the degree where children who belong to the two oldest age categories are half as vulnerable to 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wasting than infants (0 to 11 months) (OR .494, 95% CI .339-719, and OR .562, 95% CI .376-839, 5% level of significance, respectively). Somalia. In Somalia, sex, paternal orphanhood, and household wealth are found to be statistically significant risk factors of wasting among Somali children (Table 4.8). As well, belonging to the second age group (12 to 23 months) appears to be a significant risk factor for wasting. Children in this age group are considerably more likely to be wasted than the youngest children (0 to 11 months) as indicated by an odds ratio of 1.69 (95% CI 1.31-2.17). Male Somali children are found to be 20% more vulnerable to wasting than female children, as indicated by an odds ratio of 1.20 (5% level of significance; 95% CI 1.02-1.41), and paternal orphans are nearly twice as vulnerable to wasting as other children (5% level of significance; OR 1.81, 95% CI 1.19-2.74). Unlike in other countries, in Somalia, wealth poverty is found to be a significant risk factor for wasting. Children in the poorest households are three times more vulnerable to being wasted than children in the wealthiest households (OR 3.03, 95% CI 2.17-4.24). The degree of risk is also similar for the second and third wealth quintiles (OR 2.87, 95% CI 2.06-3.99, and OR 3.00, 95% CI 2.17-4.16, respectively), but then decreases to a two-fold increased risk of wasting within the fourth household wealth quintile (OR 2.22, 95% CI 1.60— 3.08). 4.5.6 Model 3 Discussion: Socioeconomic and demographic risk factors of wasting The results in model 3 show that, as with stunting and underweight, wasting is responsive to age. However, unlike the other undernutrition variables, the only children in all four countries who are more vulnerable compared to children 0 to 11months are those 12 to 23 months of 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. age (Table 4.8). As with stunting and underweight, the spike in vulnerability that appears among the 12 to 23 month age group could be related to weaning, and the consequent difficulties faced by the child in accessing and/or adjusting to solid foods. However, wasting risks for older children drop substantially below that for infants. The results for sex as a predictor for underweight in Ivorian children are relatively consistent with those for stunting and underweight, in that Ivorian boys are found to be more vulnerable to undernutrition than girls. Recall from the discussions of models 1 and 2 that boys were found to be slightly more vulnerable to undernutrition than girls. Household wealth also influences vulnerability to wasting, although to a lesser degree in relation to stunting and underweight (Table 4.8), and only for Somalia among the four countries studied. As with models 1 and 2, poor Somali children face the greatest risk of undernutrition, compared to the other countries. To reiterate, wasting is a measure of acute undernutrition stemming from short term nutrition shocks and/or illness, in contrast to stunting and to a certain extent, underweight. As such, wasting is not as valid or reliable an indication of long term poverty as stunting and underweight are. Moreover, wasting is not as prevalent in the samples studied, as other forms of undernutrition. This might help to explain the differing results and decreased predictability offered by model 3, in comparison to models 1 and 2. Chapter Five concludes this thesis with a discussion about implications of the empirical findings for the greater context of social conditions in SSA including social and civil strife, theories of orphan care, and factors of resilience to undernutrition. Concluding remarks then follow commentary on the limitations of the research. 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER FIVE Conclusion Chapter Five begins with a brief exploration of the greater context of the findings of this thesis. In doing so, the author acknowledges how the empirical results relate to the greater context of Africa's social landscape and the orphan crisis. The limitations of this thesis are then discussed, followed by concluding remarks. 5.1 The greater context of findings and limitations: Social determinants of child undernutrition and factors of resilience This discussion considers the implications of the empirical findings, for the greater context of social conditions in SSA including social and civil strife, theories of orphan care, and factors of resilience to undernutrition. The findings that age, sex, household wealth, and maternal education are reliable predictors of chronic undernutrition, and to a lesser degree, acute undernutrition, are consistent with the literature (see deOnis, Frongillo, and Blossner 1997; Gyimah 2003; Kennedy et al. 2006; Sahn and Stifel 2002; Svedberg 1990; and WHO 2005). Moreover, new evidence supporting the theory that some orphans are at greater risk of undernutrition is revealed in this thesis, though the overall reliability of the three orphanhood variables as predictors of childhood undernutrition is low across the countries studied. It is important to acknowledge that mixed results regarding orphanhood as a factor of undernutrition may therefore be indicative of manifestations of varied social conditions inherent to competing theories of orphan care. As such, this suggests that orphans are for the most part coping as well as other children in the four countries studied, although not without 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. some exceptions. Recall that one orphan care theory postulates that social support networks such as the extended family and community are more resilient to income shocks including the loss of one or both parents than once assumed, and that social support mechanisms such as community safety nets do not necessarily fail when compromised (Abebe and Asbjorn 2007; Chirwa 2002; Foster 2007; Guest 2004; and Rugalema 1999). Several studies have revealed that an increased burden on the safety net may necessitate the rationing and diversification of support, but the added stress does not always lead to total collapse. For instance, Foster (2007) points to both formal and informal community safety nets that may include collective burial societies to help offset funeral costs of bereaved families; rotating savings and credit associations which are used to save for purchasing capital; co-operatively owned and controlled businesses; loan providers; philanthropic groups such as schools and faith-based organizations; and friends and family. As such, households facing hardship may be able to draw on a number of sources of support. The importance of the extended family for child-rearing in the wake of orphanhood is also demonstrated in a longitudinal study of Gambian orphans by Sear, Mace, and McGregor (2000). Their research revealed ".. .higher survival probabilities for children with living maternal grandmothers," and specifically that. .orphanhood after the age of two years does not increase the probability of death for children" (1647). This could help to explain the lack of statistically significant results for Gambian orphans in relation to their vulnerability to undernutrition. In contrast, another theory posits that the social safety net in AIDS afflicted communities is completely disintegrating, thereby leaving the elderly generation to care for surviving children because the aunts and uncles also often succumb to AIDS-related illnesses (see Ayieko 2000; Foster 2000; George, Oudenhoven, and Wazir 2003; Okeyo 1995; and 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNICEF 2003). Moreover, abject social conditions stemming from civil conflict, poverty, and other stressors - all of which are prevalent in the countries studied in this thesis - have implications for the formal and informal social safety net. As Foster (2007) explains, "Community safety nets may be stronger where social connectedness is higher and in rural areas where traditional customs are maintained, and weaker in countries affected by civil strife, social dislocation and urbanization" (S56). Furthermore, formal government expenditures to strengthen the social safety net may not always be effectively directed, as Maclean (2002) found with regard to Cote d'lvoire's social policy. She explains, ... [T]he government ... has tried to replace ... informal social networks with the centralized arm of the bureaucracy rather than through a more decentralized social policy .... [Ljittle has been done ... to recognize or address the needs of vulnerable groups. Despite a greater awareness of the options available, no targeting policy exists. There are only a few specific services that are provided free, for example, immunizations for children and childbirth kits for pregnant women. Instead, most social welfare policy focuses on formal sector workers such as teachers or civil servants, who make up at most 10 percent of the total population (64 and 82). In addition, regarding Somalia, civil war at the time of data collection could also partially explain the results given the likelihood that some parents - most notably fathers lost their lives in armed conflict. In turn, conflict increases the potential for populations to be displaced and face hardships inherent with such upheaval including impediments to food access and production. Also recall from the discussion of model 1 that Somalia ranks the lowest on the HDI out of the four countries studied. As such, children, their households, and extended families within severely impoverished states are less likely to be resilient to income shocks such as conflict and/or the loss of a prime-aged adult and other potential incomeearners, than those in states that are better off. Insight into how social landscapes can help or hinder kin- and/or community-based care for orphans, complements existing knowledge on the prevention of undernutrition 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. through targeted interventions. According to the World Health Organization (2000, 16), key areas of national level strategic action should: • • • • • • • • • mainstream nutrition goals into development policies and programmes, improve household food and nutrition security, protect consumers through improved food quality and safety, prevent and manage infectious diseases, promote breastfeeding, care for the socioeconomically deprived and nutritionally vulnerable, prevent and control specific micronutrient deficiencies, promote appropriate diets and healthy lifestyles, and assess, analyse and monitor nutrition situations. This commentary considered the relevance of the empirical findings of this thesis, for the greater context of social conditions in SSA including social and civil strife, theories of orphan care, and factors of resilience to undernutrition. The following section acknowledges the limitations of this thesis. 5.2 Limitations This research faces various limitations. Cross-sectional data provide information for a single point in time, and therefore do not capture long term changes. Although successive cycles of cross-sectional surveys can capture changes to populations over time (such as with the 4 cycles of MICS), they do not monitor the same individuals each cycle as a panel study would. Such studies are costly and complex to execute, however, and are not always feasible at the national level. The richness of longitudinal research and the associations it can reveal notwithstanding, as Kendall and Stuart underscore, "[a] statistical relationship, however strong and however suggestive, can never establish causal connection: our ideas of causation must come from outside statistics, ultimately from some theory or another" (1961, 279). 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Residual heterogeneity - where omitted variables may help to explain the relationships being analyzed - is also an issue. This could help to explain the weak goodnessof-fit of the three empirical models of this thesis. Neither MICS3 nor any national survey (longitudinal or otherwise) could conceivably collect complete data on the myriad biological, social, and medical determinants of health. Omitted variables that were tested in early phases of the model fitting process include (but are not limited to) measures for immunization status, the presence/absence of malaria, pneumonia, and diarreah, and whether the child has access to a health care professional. MICS3 datasets also contain information on the total number of household members, the number of children in the household, and other data that could be used in estimating associations between familial support and child health outcomes. Notwithstanding the rich source of variables in MICS3 that perceivably could have been included in the empirical component of this thesis, for the sake of preserving model manageability and degrees of freedom, the author chose a select few that are traditionally accepted as valid socioeconomic and demographic measures. There are also the issues of endogenity, where some omitted variables could be correlated with both orphan status and children's health, in addition to the time order of events. At the time of the survey, it is known whether the child is an orphan and what their nutritional status is, but it is not when they became an orphan, nor at what exact point they became undernourished. As such, it is conceivable that some children were undernourished or vulnerable to undernutrition prior to becoming orphans. Moreover, the cross sectional data may include very recently orphaned individuals, whom have not yet experienced the potential effects of orphanhood. Further, only surviving orphans could be studied in this thesis, therefore results are biased in that the socioeconomic and demographic circumstances of the most vulnerable orphans - those who died - remain unknown. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moreover, national surveys commonly collect cross-sectional data, especially in countries facing human resource constraints and other impediments to longitudinal demographic and health surveillance such as conflict and other manifestations of social strife, as with the countries studied in this thesis. As such, the quality of the data may also be limited by the conditions under which they are collected, which in turn can impact the robustness of the empirical models and the nature of results. Notwithstanding these limitations, cross-sectional health and demographic studies are a good source of information on the extent of orphanhood, provide results that are generalizable across national populations (Bicego, Rutstein, and Johnson 2003), and make possible comparisons between countries which can assist in revealing regional patterns (Kobiane, Calves, andMarcoux 1996). Finally, the author of this thesis found the literature to be dominated by discourse about SSA and orphan care in general, with the exception of studies specific to a handful of countries (see Heyman et al. 2007; Howard et al. 2006; Kakooza and Kimuna 2005; Kidman, Petrow, and Heyman 2007; Nyamukapa and Gregson 2003 and 2005; and Strebel 2004). As such, the discussion about the social safety nets throughout this thesis is largely limited to generalizations about SSA given that little is published about the specific countries studied. This dearth of national case studies presents an opportunity for future research. Gaining more detailed insight into what support is available to vulnerable children and their families, either through in-depth qualitative research and/or by adding questions to national household surveys such as MICS could help to fill these knowledge gaps. For instance, possible research questions could focus on the timing and cause of parental death, availability/accessibility of social-services, family and community coping mechanisms including the nature/degree of family extendedness, and the household dependency ratio in 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relation to orphans' health outcomes. Such information could extend the scope of the conceptual framework for studying associations between orphanhood and child health. Also, MICS does not currently collect anthropometric or other health status data for children 5 years of age and older. This presents an opportunity for developing health outcome variables for this population, and in turn, creates possibility for conducting future research which examines the impacts of orphanhood and other socioeconomic and demographic factors on the health of older children. 5.3 Concluding remarks Poverty, conflict, and undernutrition are widespread throughout sub-Saharan Africa. Moreover, widespread orphanhood due to the HIV/AIDS epidemic and other common health hazards such as TB and obstetric complications have exacerbated the already arduous social and economic conditions of this region, including the four countries studied in this thesis. This research responded to recommendations put forth by UNICEF to examine health outcomes of orphans versus other children, and multivariate determinants of child health. Accordingly, the author first set out to explore in the literature what is known about the factors, magnitude, and consequences for orphans of prime-aged adults dying in sub-Saharan Africa (SSA). Key themes in the literature review included the mother's versus the father's roles for emotional and physical nurturance of their children, theories of orphan care, and empirical evidence of the impacts of parental death on orphans' health. Some evidence suggests that orphans are more at risk of undernutrition, but the literature is mixed regarding the vulnerability of differing types of orphans, as well as the resilience of social safety nets in the wake of income shocks such as premature parental death. 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The second and primary purpose of this thesis was to empirically assess these relationships in samples of four SSA countries by exploring the following research questions: i) what socioeconomic and demographic characteristics - including orphanhood - are risk factors for child undernutrition? More specifically, ii) do differing types of orphans confer differing degrees of vulnerability to undernutrition? To recap, empirical models 1, 2, and 3 assessed risk factors of stunting (chronic undernutrition), underweight (a composite measure of chronic and acute undernutrition), and wasting (acute undernutrition) among children 0 to 59 months old, respectively. Broadly speaking, results indicate that child age, household wealth, and mother's education are reliable predictors of undernutrition; sex and orphanhood were also revealed as risk factors, but their influence was not as consistent across all models and countries. In particular, maternal orphanhood was a statistically significant predictor of stunting (model 1, Table 4.6) among Ivorian children. Somali double orphans and Ivorian paternal orphans were revealed as less vulnerable to underweight (model 2, Table 4.7), while Somali paternal orphans were found to be more vulnerable to wasting (model 3, Table 4.8). In relation to these inconsistencies, the author of this thesis advises readers to interpret the results with caution, but to also consider that these finding may be indicative of the competing theories of orphan care as described in Chapter Two. This research was limited by a number of factors including limited country-specific case studies and empirical work looking at social structures of the household and extended family and community. In spite of such limitations, the findings of this thesis contribute to relatively limited empirical scholarship on the impacts of maternal versus paternal death on child health. As such, this new information may assist decision makers in better addressing the needs of orphans when formulating policies and calling for development assistance. 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The findings also complement the descriptive country reports produced from MICS3 data, and as such, may help guide the design of future MICS cycles and reports. For instance, new variables could be created to capture other potential health impacts of orphanhood such as psychological stress and delayed development, and data analysts may also consider reporting results by type of orphanhood. Furthermore, more longitudinal research is needed in order to monitor the lives of orphans as they mature; this would eliminate many of the limitations associated with cross-sectional research. In the wake of HIV/AIDS, abject poverty, conflict, and myriad other health hazards, the orphan crisis in Africa at times give the impression of a "runaway train." However, this thesis demonstrates that although orphans are found to be more vulnerable to undernutrition at times, they are not always worse off than other children; this is consistent with mixed evidence pertaining to theories of orphan care. This gives hope to the notion that social networks such as extended families and other coping structures are able to provide orphans with basic sustenance. Notwithstanding, so as to better appreciate the differing needs and vulnerabilities of maternal versus paternal orphans as they relate to specific cultural norms surrounding gender-based parental roles, more in-depth investigation into individual countries is recognized and called for. 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Abebe, T., and A. Asbjorn. 2007. Children, AIDS and the politics of orphan care in Ethiopia: the extended family revisited. Social Science and Medicine 64(10):2058-2069. AbouZahr C. "Global burden of maternal death and disability." In Reducing maternal death and disability in pregnancy, edited by C. Rodeck, 1-11. Oxford: Oxford University Press, 2003. Ainsworth, M.D.S. 1967. Infancy in Uganda: Infant care and the grounds of love. Baltimore, MD: Johns Hopkins University Press. Ainsworth, M.D.S. "The development of infant-mother attachment." In Review of child development research, Volume 3, edited by B.M. Caldwell and H.N. Ricciuti, 1-94. Chicago: University of Chicago Press, 1973. Ainsworth, M., and D. Filmer. 2006. Inequalities in children's schooling: AIDS, orphanhood, poverty, and gender. World Development 34(6): 1099-1128. Ainsworth, M., and I. Semali. 2000. The impact of adult deaths on child health in Northwestern Tanzania. World Bank Policy Research Paper 226. http://wwwwds.worldbank.org/external/ default/ WDSContentServer/IW3P/IB/2000/02/ 09/000094946_00012505525064/additional /116516322_20041117153034.pdf Ainsworth, M., K. Beegle, and G. Koda. 2002. The impact of adult mortality on primary school enrollment in Northwestern Tanzania. Africa Region Human Development Working Paper Series. http://siteresources.worldbank.org/AFRICAEXT/Resources/ tz_adult_mortality.pdf Washington, DC: World Bank. Anderson, F.W.J., S.U. Morton, S. Naik, and B. Gebrian. 2007. Maternal mortality and the consequences on infant and child survival in rural Haiti. Maternal and Child Health Journal 11:395-401. Andrews, G., D. Skinner, and K. Zuma. 2006. Epidemiology of health and vulnerability among children orphaned and made vulnerable by HIV/AIDS in sub-Saharan Africa. AIDS Care 18(3):269-276. Atwine, B., E. Cantor-Graaea, and F. Bajunirweb. 2005. Psychological distress among AIDS orphans in rural Uganda. Social Science and Medicine 61:555-564. Ayaya, S.O., and F.O. Esamai. 2001. Health problems of street children in Eldoret, Kenya. East African Medical Journal 78(12):624-629. Ayieko, M.A. 2000. From single parents to child-headed households: The case of children orphaned by AIDS in Kisumu and Siaya Districts. New York: UNDP HIV & Development Program. 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Baden, S., S. Hassim, and S. Meintjies. 1999. Country gender profile: South Africa. http://www.womensnet.org.za/links/genderpr.htm Barbarin, Oscar, and Linda Richter. 2001. Mandela's Children: Child Development in PostApartheid South Africa. New York: Routledge. Barry, H., and L.M. Paxson. 1971. Infancy and early childhood: Cross-cultural codes 2. Ethnology 10:466-508 Bates, I., C. Fenton, J. Gruber, D. Lalloo, A. Medina Lara, S. Bertel Aquire, S. Theobald, R. Thomson, and R. Tolhurst. 2004. Vulnerability to malaria, tuberculosis, and HIV/AIDS infection and disease. Part 1: determinants operating at individual and household level. The Lancet Infectious Diseases 4(5):267-277. Beattie, T., S. Rowland-Jones, and R. Kaul. 2002. HIV-1 and AIDS: what are protective immune responses? Journal of HIV Therapy 7:35-39. Bedri, A., S. Kebede, and H. Negassa. 1995. Sociodemographic profile of children affected by AIDS in Addis Ababa. Ethiopia Medical Journal 33(4):227-234. Beegle, K., J. De Weerdt, and S. Dercon. 2006. Orphanhood and the long-run impact on children. American Journal of Agricultural Economics 88(5):1266-77. Beegle, Kathleen., Joachim De Weerdt, and Stefan Dercon. 2008. The intergenerational impact of the African orphans crisis: a cohort study from an HIV/AIDS affected area. International Journal of Epidemiology 1-8 doi:10.1093/ije/dynl97. Behrman, Jere R. 1993. The economic rationale for investing in nutrition in developing countries. World Development 21(11):1749-1771. Belsky, J., and J. Cassidy. "Attachment: Theory and Evidence." In Development Through Life, edited by M. Rutter and D. Hay, 373-402. Oxford, UK: Blackwell, 1994. Bicego, G., S. Rutstein, and K. Johnson. 2003. Dimensions of the emerging orphan crisis in sub-Saharan Africa. Social Science and Medicine 56:1235-1247. Bledsoe C., and A. Brandon. "Le placement des enfants et son influence sur la mortalite" In Mortalite et societe en Afrique, edited by G. Pison, E. van de Walle, and M. SalaDiakanda, 271-293. Paris: Presses Universitaires de France, 1989. Bledsoe C.H., D.C. Ewbank, and U.C. Isiugo-Abanihe. 1988. The effect of child fostering on feeding practices and access to health services in rural Sierra Leone. Social Science and Medicine 27:627-636. Bloom, D. 2001. Health, wealth, AIDS and poverty: the case of Cambodia. Geneva: UNAIDS. 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Bowlby, J. 1969. Attachment and Loss, Volume 1 (Attachment). New York. Basic Books. Bowlby, J. 1973. Attachment and Loss: Vol. II. Separation: Anxiety and anger. New York: Basic Books. Bowlby, J. 1953. Child care and the growth of love. Harmondsworth: Penguin Books. Boyle, M.H., Y. Racine, K. Georgiades, D. Snelling, S. Hong, W. Omariba, P. Hirley, and P. Rao-Melcini. 2006. The influence of economic development level, household wealth and maternal education on child health in the developing world. Social Science and Medicine 63(8):2242-2254. Bretherton, I. 1992. The Origins of Attachment Theory: John Bowlby and Mary Ainsworth. Developmental Psychology 28(5):759-775. Brodzinsky. D.M., A.V. Gormly, and S.R. Ambron. 1986. Lifespan Human Development, 3rd ed. New York: CBS College Publishing. Brown, J.L., and E. Pollitt. 1996. Malnutrition, poverty and intellectual development. Scientific American 274: 38—43. Burton, M.L., C.C. Moore, J.W.M. Whiting, and A.K. Romney. 1996. Regions based on social structure. Current Anthropology 37:87-123. Calderon, M.A.K, P.B. Montes, M.C. Gonzalez, and A.C. Pliego. 2007. Social aspects of maternal mortality: A case study of the state of Mexico. Social Medicine 2(4):186-191. Caldwell, John C. 1976. Toward a restatement of demographic transition theory. Population and Development Review 2(3/4):321-366. Carson, J., and R.D. Parke. 1996. Reciprocal negative affect in parent-child interactions and children's peer competency. Child Development 67:2217-2226. Case, Anne, and Cally Ardington. 2007. The Impact of Parental Death on School Outcomes: Longitundinal Evidence from South Africa. Demography 43(3):401-420. Case, A. Lin, I.F. and McLanahan, S. 2000. How Hungry is the Selfish Gene? The Economic Journa, 110(466):781-804. Cassidy, J. "The nature of the child's ties." In Handbook of Attachment Theory, Research, and Clinical Applications, edited by J. Cassidy and P.R. Shaver, 3-20. New York: The Guildford Press, 1999. Central Intelligence Agency. 2008. The World Factbook. https://www.cia.gov/ library/publications/the-world-factbook/geos/SO.html Chen, L.C., M.C. Gesche, S. Ahmed, A.I. Chowdhury, and W.H. Mosley. 1974. Maternal Mortality in Rural Bangladesh. Studies in Family Planning 5(11):334-341. 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chirwa, W.C. 2002. Social Exclusion and Inclusion: Challenges to Orphan Care in Malawi. Nordic Journal of African Studies 11(1):93—113. Coleman, J.S. 1988. Social capital in the creation of human capital. The American Journal of Sociology Suppl. No. 94:S95-S120. Corbett, E.L., C.J. Watt, N. Walker, D. Maher, B.G. Williams, M.C. Raviglione, and C. Dye. 2003. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine 163:1009-21. Crisis State Research Centre. 2006. Crisis, Fragile and Failed States: Definitions used by the CSRC. London: London School of Economics and Political Science http://www.crisisstates.com/download/drc/FailedState.pdf De, P., and B.N. Bhattacharya. 2002. Determinants of child immunization in four lessdeveloped states of North India. Journal of Child Health Care 6(1):34—50. Deininger, K., M. Garcia, and K. Subbarao. 2003. AIDS-Induced orphanhood as a systemic shock: magnitude, impact, and program interventions in Africa. World Development 31(7): 1201-1220. Deming, M.S., J. Roungou, M. Kristiansen, I. Heron, A. Yango, A. Guenengafo, and R. Ndamobissi. 2002. Tetanus toxoid coverage as an indicator of serological protection against neonatal tetanus. Bulletin of the World Health Organization 80(9):696-703. Denham, S.A, J. Mitchell-Copeland, K. Strandberg, S. Aurerbach, and K. Blair. 1997. Parental contributions to preschoolers' emotional competence: Direct and indirect effects. Motivation and Emotion 27:65-86. deOnis, M., E. Frongillo, and M. Blossner. 1997. Is undernutrition declining? An analysis of changes in levels of child undernutrition since 1980. Bulletin of the World Health Organization 78(10):1222-33. Desai, S., and S. Alva. 1998. Maternal education and child health. Is there a strong causal relationship? Demography 35(1):71-81. De Silva, M.J., and T. Harpham. 2007. Maternal social capital and child nutritional status in four developing countries. Health & Place 13:341-355. Disease Control Priorities Project. 2007. Maternal Deaths: An unacceptable lack of progress. (March fact sheet), http://www.dcp2.org/file/69/DCPP-MaternalDeaths.pdf Dye, Christopher. 2006. Global epidemiology of tuberculosis. Lancet 367:938-940. Dye, C., C.J. Watt, D.M. Bleed, S.M. Hosseini, and M.C. Raviglione. 2005. Evolution of tuberculosis control and prospects for reducing tuberculosis incidence, prevalence, and deaths globally. Journal of the American Medical Association 293:2767-2775. 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Easterbrooks, M.A., and W.A. Goldberg. "Security of toddler-parent attachment: Relation to children's sociopersonality functioning during kindergarten." In Attachment in the Preschool Years, edited by M.T. Greenberg, D. Cicchetti, and E.M. Cummings, 221-244. Chicago: University of Chicago Press, 1990. El Taguri, A., F. Besmar, A. Abdel Monem, I. Betilmal, C. Ricour, and M-F. RollandCachera. 2009. Stunting is a major risk factor for overweight: results from national surveys in 5 Arab countries. Eastern Mediterranean Health Journal 15(3):549-562. Emde, R.N. "Emotional Availability: A reciprocal reward system for infants and parents with implications for prevention of psychosocial disorders," in Parent-Infant Relationships. Ed. P.M. Taylor (Orlando: Grune and Stratton, 1980) 87-115. Emde, R.N. and M.A. Easterbrooks. "Assessing emotional availability in early development." In Early Identification of Children at Risk: An International Perspective, edited by W.K. Frankenburg, R.N. Emde, and J.W. Sullivan, 79-101. New York: Plenum, 1985. Evans, David. 2005. The Spillover Impacts of Africa's Orphan Crisis. http://129.67.96.112/conferences/2006-EOI-RPI/papers/csae/Evans_spillovers.pdf Evans, David K., and Eduard Miguel. 2007.0rphans and Schooling in Africa: A Longitudinal Analysis. Demography 44(l):35-57. Family Health International (FHI). 2003. Voices from the communities: The impact of HIV/AIDS on the lives of orphaned children and their guardians, http://www.fhi.org /en/index.htm Foster, G. 2000. The capacity of the extended family safety net for orphans in Africa. Psychology, Health and Medicine 5(1):55—62. Foster, G. 2007. Under the radar: Community safety nets for AIDS-affected households in sub-Saharan Africa. AIDS Care 19(Supplement 1):S54-S63 Foster, G., R. Shakespeare, F. Chinemana, H. Jackson, S. Gregson, C. Marange, and S. Mashumba. 1995. Orphan prevalence and extended family care in a peri-urban community in Zimbabwe. AIDS Care 7(1):3—18. Foster, G., and J. Williamson. 2000. A review of current literature of the impact of HIV/AIDS on children in sub-Saharan Africa. AIDS 14 (Suppl. No. 3): S275-S284. Fouts, Hillary N., Barry S. Hewlett, and Michael E. Lamb. 2005. Parent-Offspring Weaning Conflicts among the Bofi Farmers and Foragers of Central Africa. Current Anthropology 46 (l):29-50. Frost, M.B., R. Forste, and D.W. Haas. 2005. Maternal education and child nutritional status in Bolivia: finding the links. Social Science and Medicine 60(2):395-^I07. Ill Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gambia Bureau of Statistics (GboS), UNICEF, and The World Bank. 2007. The Gambia Multiple Indicator Cluster Survey Report 2005/2006. Cape Point: UNICEF. Garson, David G. 2008. Logistic regression, http://www2.chass.ncsu.edu/garson/pa765/ logistic.htm. George, S., Oudenhoven, N., and Wazir, R. 2003. Foster care beyond the crossroads: Lessons from an international comparative analysis. Childhood 10(3):343-361. Gertler, P.J., S. Martinez, D. Levine, and S. Bertozzi. 2003. Losing the presence and presents of parents: how parental death affects children. Berkeley: Haas School of Business. Gilbert, L., and L. Walker. 2002. Treading the path of least resistance: HIV/AIDS and social inequalities South African case study. Social Science and Medicine 54:1093-1110. Ghosh, Jayati, and Ezekiel Kalipeni. "Rising Tide of AIDS Orphans in Southern Africa." In HIV and AIDS in Africa: beyond epidemiology, edited by Ezekeil Kalipeni, Susan Craddock, Joseph R. Oppong, and Jayati Ghosh, 304. USA: Blackwell Publishing, 2004. Glewwe, Paul. 1999. Why Does Mother's Schooling Raise Child Health in Developing Countries? Evidence from Morocco. Journal of Human Resources 34(1): 124-59. Goodman, Robin F. 2001. Children and grief: what they know, how they feel, how to help. New York: New York University School of Medicine, http://www.aboutourkids.org/ articles/children_grief_what_they_know_how_they_feel_how_help Gregson, S., C.A. Nyamukapa, G.P. Garnetti, M. Wambe, J.J.C. Lewis, P.R. Mason, S.K. Chandiwana, and R.M. Anderson. 2005. HIV infection and reproductive health in teenage women orphaned and made vulnerable by AIDS in Zimbabwe. AIDS Care 17(7):785794. Guarcello, L., S. Lyon, and F. Rosati. 2004a. Orphanhood and child vulnerability: Burundi. Understanding Children's Work Working Paper, http://www.ucw-project.org/pdf/ publications/Burundiorphanscountryreports.pdf Guarcello, L., S. Lyon, and F. Rosati. 2004b. Orphanhood and child vulnerability: Cote d'lvoire. Understanding Children's Work Working Paper. Guarcello, L., S. Lyon, and F. Rosati. 2004c. Orphanhood and child vulnerability: Senegal. Understanding Children's Work Working Paper, http://www.ucw-project.org/ pdf/publications/standard_senegal_orphans_countryreport_21sept2004.pdf Guarcello, L., S. Lyon, F. Rosati, and C.A. Valdivia. 2004. The influence of orphanhood on children's schooling and labour: evidence from sub-Saharan Africa. Understanding Children's Work Working Paper, http://www.ucw-project.org/pdf/publications/ standard_orphans_and_CL_Oct2006.pdf 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Guest, Emma. 2004. "Children of AIDS: Africa's Orphan Crisis: A Mother to Her Brothers: A Child-headed Household's Story, Johannesburg, South Africa." In HIV and AIDS in Africa: Beyond Epidemiology, edited by Ezekiel Kalipeni, Susan Craddock, Joseph R. Oppong, and Jayati Ghosh, 316-322. USA: Blackwell Publishing, 2004. Gunnar, M.R. "Early adversity and the development of stress reactivity and regulation." In The Effects of Adversity on Neurobehavioral Development: The Minnesota Symposia on Child Psychology, Volume 31, edited by Charles A. Nelson, 163-200. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 2000. Gunnar, M.R., L. Brodersen, M. Nachmias, K. Buss, and J. Rigatuso. 1996. Stress reactivity and attachment security. Developmental Psychobiology 29(3):191-204. Gyimah, S.O. 2003. Interaction Effects of Maternal Education and Household Facilities on Childhood Diarrhea in sub-Saharan Africa: The Case of Ghana. The Journal of Health and Population in Developing Countries http://jhpdc.unc.edu Halfon, N., and N. Hochstein. 2002. Life Course Health Development: An integrated framework for developing health, policy, and research. The Milbank Quarterly 80(3):1— 41. Harpham, T., E. Grant, and E. Thomas. 2002. Measuring social capital within health surveys: key issues. Health Policy and Planning 17(1):106—111. Harpham, T., S. Huttly, M.J. De Silva, and T. Abramsky. 2005. Maternal Mental Health and Child Nutritional Status in Four Developing Countries. Journal of Epidemiology and Community Health 59:1060-1064. Hewlett, Barry S. "Fathers in Forager, Farmer, and Pastoral Cultures." In The role of the Father in Child Development: Fourth Edition, edited by Michael E. Lamb, 182-195. Hoboken, N.J.: John Wiley and Sons, Inc., 2004. Hewlett, B.S. 2000. Culture, history, and sex: Anthropological contributions to conceptualizing father involvement. Marriage and Family Review 29(2-3):59-73. Heyman J., A. Earle, A., D. Rajaraman, D., C. Miller, C., and K. Bogen, K. 2007. Extended family caring for children orphaned by AIDS: balancing essential work and caregiving in a high HIV prevalence nations. AIDS Care (19)3:337-345. Hogan, C.M., and S.M. Hammer. 2001. Host determinants in HIV infection and disease: part 1, cellular and humoral immune responses. Annals of Internal Medicine 134:761-76. Hooven C., L. Katz, and J. Gottman. 1994. The family as a meta-cognition culture. Cognition and Emotion 9:229-264. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Horizons Project. 2003. Succession planning in Uganda: Early outreach for AIDS affected children and their families. Washington: Population Council, http://www.popcouncil.org /pdfs/horizons/orphanssum.pdf Hosmer, David W., and Stanley Lemeshow. 2000. Applied Logistic Regression: Second Edition. New York: John Wiley and Sons Inc. Howard, Brian H., Carl V. Phillips, Nelia Matinhure, Karen Goodman, Sheryl A. McCurdy, and Cary A. Johnson. 2006. Barriers and incentives to orphan care in a time of AIDS and economic crisis: a cross-sectional survey of caregivers in rural Zimbabwe. BMC Public Health 6:27-36. Institut National de la Statistique (INS) [Cote d'lvoire], 2007. Enquete a indicateurs multiples, Cote d'lvoire 2006, Rapport final, Abidjan, Cote d'lvoire: Institut National de la Statistique. Jones, S. 1999. Singlehood for security: towards a review of the relative economic status of women and children in woman-lead households. Society in Transition 30(1):13-27. Kakooza, James, and Sitawa R. Kimuna. 2005. HIV/AIDS Orphans' Education in Uganda: The Changing Role of Older People. Journal of Intergenerational Relationships (3)4:6381. Kakwani, Nanak, and Kalanidhi Subbarao. 2005. Aging and Poverty in Africa and the Role of Social Pensions. Washington DC: World Bank. Kamenga, M., M. DaSilva, K. Muniaka, B. Matela, V. Batter, and R. Ryder. 1990. AIDS orphans in Kinshasa, Zaire . International Conference on AIDS. June 20-23 1990 (abstract number Th.D.127). Kendall, M.G., and A. Stuart. 1961. The Advanced Theory of Statistics. New York: Charles Griffin Publishers. Kennedy, Gina, Guy Nantel, Inge D. Brouwer, and Frans J. Kok. 2006. Does living in an urban environment confer advantages for child nutritional status? Analysis of disparities in nutritional status by wealth and residence in Angola, Central African Republic and Senegal. Public Health Nutrition 9(2):187—193. Kidman, S.E., S. Petrow, and J. Heyman. 2007. Africa's orphan crisis: Two communitybased models of care. AIDS Care 19(3):326-329. Kiros, Gebre-Egziabher, and Dennis P. Hogan. 2001. War, famine and excess child mortality in Africa: The role of parental education. International Journal of Epidemiology 30:447455. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kobiane, Jean-Francois, Anne-Emmanuele Calves, and Richard Marcoux. 2005. Parental death and children's schooling in Burkina Faso. Comparative Education Review 49(4):468-489. Kobiane, J., A. Calves, and R. Marcoux. 1996. What's wrong with the extended family system? Paper presented at the 2005 Population Association of America Annual Meetings, Philadelphia, PA. Koenig, Michael, Vincent Fauveau, A. I. Chowdhury, J. Chakraborty, M. A. Khan. 1988. Maternal Mortality in Matlab, Bangladesh: 1976-85. Studies in Family Planning 19(2):69-80. Kowalewski, M., P. Mujinja, and A. Jahn. 2002. Can Mothers Afford Maternal Health Care Costs? User Costs of Maternity Services in Rural Tanzania. African Journal of Reproductive Health 6 (l):65-73. Kromelow, S., C. Harding, and M. Touris. 1990. The role of the father in the development of stranger sociability during the second year. American Journal of Orthopsychiatry 60:521530. Kuh, D., Y. Ben-Schlomo, J. Lynch, J. Hallqvist, and C. Power. 2003. Life Course Epidemiology. Journal of Epidemiology and Community Health 57:778-783. Lamb, M.E., C.P. Hwang, A.M. Frodi, and M. Frodi. 1982. Security of mother- and fatherinfant preferences for mothers and fathers. Infant Behavior and Development 5:355-367. Leclerc-Madlala, S. 2000. The silences that nourish AIDS in Africa. Mail and Guardian, Johannesburg, 28-29 August:11-17. Le Camus, J. 1995. Le dialogue phasique: Nouvelles perspectives dans l'etude des interactions pere-bebe. Neuropsychiatrie de I'Enfance 43 (1—2):53—65. Lloyd, C.B., and A.K. Blanc. 1996. Children's schooling in sub-Saharan Africa: the role of fathers, mothers, and others. Population and Development Review 22(2):265-298. Lule, E., G.N.V. Ramana, N. Ooman, J. Epp, D. Huntington, and J.E. Rosen. 2005. Achieving the Millennium Development Goal of improving maternal health: determinants, interventions, and challenges. Health, Nutrition and Population Discussion Paper. Geneva: World Bank. Machel, Graga. 2000. The Impact of Armed Conflict on Children: A critical review of progress made and obstacles encountered in increasing protection for war-affected children. "International Conference on War Affected Children," Winnipeg, Canada, September 2000 http://saiv.net/SourceBook/Storage/documents/doc_armedconflict_ children.pdf 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Maclean, Lauren Morris. 2002. Constructing a Social Safety Net in Africa: An Institutionalist Analysis of Colonial Rule and State Social Policies in Ghana and Cote d'lvoire. Studies in Comparative International Development 37(3): 64-90. Madhavan, S. 2001. Female relationships and demographic outcomes in sub-Saharan Africa. Sociological Forum 16(3):503—527. Madhavan, Sangeetha, Nicholas W. Townsend, and Anita I. Garey. 2008. Absent breadwinners: Father-Child Connections and Paternal Support in Rural South Africa. Journal of Southern African Studies 34 (3):647-663. Makame, V., C. Ani, and S. Grantham-McGregor. 2002. Psychological well-being of orphans in Dar El Salaam, Tanzania. Acta Paediatrica 91(4):459-465. Martorell, R., and J.P. Habicht. "Growth in early childhood in developing countries." In Human Growth: A comprehensive treatise, Vol. 3: Methodology, Genetic, and Nutritional Effects on Growth, Second edition, edited by F. Falkner and J.M. Tanner. New York: Plenum Press, 1986. Masmas, T.N., H. Jensen, D. da Silva, L. Hej, A. Sandstrom, and P. Aaby. 2004. Survival among motherless children in rural and urban areas in Guinea-Bissau. Acta Paediatrica 93:99-105. McNicoll, G. 1980. Institutional determinants of fertility change. Population and Development Review 6(2):441—462. Meena, R. 1994. Women and sustainable development. Voices from Africa. http://www.unsystem.org/ngls/documents/publications.en/voices.africa/ number5/vfa5.07.htm Miller, C.M., S. Gruskin, S.V. Subramanian, and J. Heyman. 2007. Emerging health disparities in Botswana: Examining the situation of orphans during the AIDS epidemic. Social Science and Medicine 64:2476-2486. Ministere de la Planification du Developpement et de la Reconstruction. 2001. Enquete Nationale d'evaluation des Conditions de vie de l'enfant de la Femme au Burundi, Institut de Statistiques et d'Etudes Economiques du Burundi et UNICEF, 2000. Monasch, R., and J. T. Boerma. 2004. Orphanhood and childcare patterns in sub-Saharan Africa: an analysis of national surveys from 40 countries. AIDS 18(2):S55-S65. Mushingeh, A., A. Mkandawire, M. Nkula, R. Chinkupula, R. Kalomo. 2002. HIV/AIDS and child labouring Zambia: a rapid assessment on the case of Lusaka, Copperbelt and Eastern Provinces. Paper No. 5, International Labour Organization, International Programme on the Elimination of Child Labour, Geneva/Lusaka, August, http://www.comminit.com/ en/node/213370/306 (only the executive summary is available). 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Nahar, S., and A. Costello. 1998. The hidden cost of 'free' maternity care in Dhaka, Bangladesh. Health Policy and Planning 13(4):417—422. National Committee on Vial and Health Statistics. 2002. Shaping a Health Statistics Vision for the 21st Century. Washington DC: Department of Health and Human Services Data Council, Centers for Disease Control and Prevention, National Center for Health Statistics. http://ncvhs.hhs.g0v/hsvisi0n/21st%20final%20report.pdf Nsamenang, Bame A. "Perceptions of Parenting among the Nso of Cameroon." In Fatherchild relations: Cultural and biosocial contexts, edited by Barry S. Hewlett, 321-344. Piscataway, NJ: Aldine Transaction Publishers, 1992. NICHD Early Child Care Research Network. 1997. The effects of infant child care on infantmother attachment security: Results of the NICHD Study of Early Child Care. Child Development 68:860-879. NICHD Early Child Care Research Network. 1998. Relations between family predictors and child outcomes: Are they weaker for children in child care? Developmental Psychology 34(5):1119-1128. Nyamukapa, C., and S. Gregson. 2005. Extended family's and women's roles in safeguarding orphans' education in AIDS-afflicted rural Zimbabwe. Social Science and Medicine 60: 2155-2167. Nyamukapa, C., Foster, G., and Gregson, S. 2003. Orphans' household circumstances and access to education in a maturing HIV epidemic in eastern Zimbabwe. Journal of Social Development in Africa 18(2):7-32. Ochwada, H. 1997. Politics and gender relations in Kenya: A historical perspective. Africa Development 22:121-139. Okeyo, T.M. 1995. AIDS Orphans - a New Epidemic with Disaster Implications. Discovery and Innovation 7(1):1-2. Omokhodion, F.O., S.I. Omokhodion, and T.O. Odusote. 2006. Perceptions of child labour among working children in Ibadan, Nigeria. Child: Health Care & Development 32(3):281-286. Paquette, Daniel. 2004. Theorizing the Father-Child Relationship: Mechanisms and Developmental Outcomes. Human Development 47:193-219. Parke, R.D. "Fathers and families." In Handbook of Parenting Vol. 3, Second edition, edited by M. Bornstein, 27-73. Hillsdale, NJ: Erlbaum, 2002. Parker, R.G. 1999. Historic overview of Brazil's AIDS programme and review of the World Bank ADS project. Family International/AIDSCAP. Rio de Janeiro: State University of Rio de Janeiro. 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Patel V., Rahman A, Jacob KS, and Hughes M. 2004. Effect of maternal mental health on infant growth in low income countries: new evidence from South Asia. British Medical Journal 328(7443):820-823. Posel, D. 2004. Sex, death and embodiment: Reflections on the stigma of AIDS in Agincourt, South Africa. "Life and death in a time of AIDS: The Southern African experience," a symposium held at Johannesburg, October 14-16 2004. Putnam, 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Rao, C., A.D. Lopez, and Y. Hemed. "Causes of Death." In Disease and Mortality in SubSaharan Africa: Second Edition, edited by Dean T Jamison, Richard G. Feachem, Malegapuru W. Makgoba, Eduard R. Bos, Florence K. Baingana, Karen J. Hofman, and Khama O. Rogo, 43. Washington DC: The World Bank, 2006. Reed, Holly E., Majorie A. Koblinsky, and W. Henry Mosely (eds.). 2000. The Consequences of Maternal Morbidity and Mortality: Report of a Workshop, http://www.nap.edu/catalog/ 9800.html Washington, DC: The National Academy Press. Reher, D.S., and F.O Gonzalez-Quinones. 2003. Do parents really matter? Child health and development in Spain during the demographic transition. Population Studies 57(l):63-75. ReliefWeb. 1996. West Africa, http://www.reliefweb.int Republic of Zambia Central Statistical Office, Lusaka, and International Labour Organization/International Programme on the Elimination of Child Labour. 1999. Zambia 1999 Child Labour Survey Report, http://www-ilo-mirror.cornell.edu/public/english/ standards/ipec/simpoc/zambia/report/zambia.pdf Roy, N.C., T.T. Kane, Barkat-e-Khuda, I. Haque. 2000. Socioeconomic and health implications of adult deaths in families of rural Bangladesh. ICDDBR, B Working Paper No. 132. Mohakhali, Bangladesh, ICDDR, B: Centre for Health and Population Research. Rugalema, G. "It is not only the loss of labor: HIV/AIDS, loss of household assets and household livelihood in Bukoba district, Tanzania." In AIDS and African Smallholder Agriculture, edited by G. Mutangadura, H. Jackson, and D. Mukurazita, 41-52. Harare: SAfAIDS, 1999. Rutstein, S.O. and Johnson, K., 2004. The DHS Wealth Index. DHS Comparative Reports No. 6. Calverton, Maryland: ORC Macro. http://www.childinfo.org/files/DHS_Wealth_ Index_(DHS_Comparative_Reports).pdf Ryder, N. 1984. Fertility and family structure. Population Bulletin of the United Nations. 15: 15-34. New York: United Nations, http://www.popline.org/docs/015145 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sagi, A., M.E. Lamb, and W. Gardner. 1986. Relations between Strange Situation behavior and stranger sociability among infants on Israeli kibbutzim. Infant Behavior and Development 9:271-282. Sahn, David E., and David C. Stifel. 2002. Parental Preferences for Nutrition of Boys and Girls: Evidence from Africa. The Journal of Development Studies 39(1):21—45. Schneider, Barry H., Leslie Atkinson, and Christine Tardif. 2001. Child-Parent Attachment and Children's Peer Relations: A Quantitative Review. Developmental Psychology 37(1):86—100. Sear, R., R. Mace, and I.A. McGregor. 2000. Maternal grandmothers improve nutritional status and survival of children in rural Gambia. Proceedings of the Royal Society of London. Series B. Biological Sciences 267:1641—1647. Sengendo, J., and J. Nambi. 1997. The psychological effect of orphanhood: a study of orphans in Rakai district. Health Transition Review 7:105-124. Setel, P. 1999. A plague of paradoxes: AIDS, culture, and demography in northern Tanzania. Chicago: Chicago University Press. Shen, C., and J.B. Williamson. 1999. Maternal mortality, women's status, and economic dependency in less developed countries: a cross-national analysis. Social Science and Medicine 49:197-214. Silberschmidt, M., and V. Rasch. 2001. Adolescent girls, illegal abortions and "sugar daddies" in Dar es Salaam: vulnerable victims and active social agents. Social Science and Medicine 52 (12):1815-1826. Smit, Ria. 2006. The Changing Role of the Husband/Father in the Dual-Earner Family in South Africa. Journal of Comparative Family Studies 33: 401—415. Smith, M.K. 2002. Gender, poverty, and intergenerational vulnerability to HIV/AIDS. Gender and Development 10(3):63-70. Sroufe, L.A. 1996. Emotional development. New York: Cambridge University Press. Statistics Sierra Leone and UNICEF-Sierra Leone 2007. Sierra Leone Multiple Indicator Cluster Survey 2005, Final Report. Freetown, Sierra Leone: Statistics Sierra Leone and UNICEF-Sierra Leone. Strauss, J., and D. Thomas. 1998. Health, nutrition, and economic development. Journal of Economic Literature 36:766-817. Strebel, Anna. 2004. The Development, Implementation and Evaluation of Interventions for the Care of Orphans and Vulnerable Children in Botswana, South Africa and Zimbabwe: 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A Literature Review of Evidence-based Interventions for Home-based Child-centred Development. Cape Town: Human Sciences Research Council. Stroebe, M.S. 2002. Paving the way: from early attachment theory to contemporary bereavement research. Mortality 7(2):127-138. Strong, M.A. 1992. The health of adults in the developing world: A view from Bangladesh. Health Transition Review 2(2):215-224. Subbarao, K., A. Mattimore, and K. Plangemann. 2001. Social Protection of Africa's Orphans and Other Vulnerable Children: Issues and Good Practice Program Options. Washington D.C.: The World Bank. http://siteresources.worldbank.org/AFRICAEXT/ Resources/African_Orphans.pdf Suliman, E.D. 2003. HIV/AIDS effects on AIDS orphans in Tanzania. Working paper. Baltimore, Maryland: Johns Hopkins University. Svedberg, P. 1990. Undernutrition in sub-Saharan Africa: Is there a Gender Bias? The Journal of Development Studies 26(3):469-486. Tabachnick, Barbara G., and Linda S. Fidell. 2007. Using Multivariate Statistics: Fifth Edition. Boston: Pearson Education Inc. Taha, T.B.T., P. Miotti, G. Liomba, G. Dallabetta, and J. Chiphangwi. 1996. HIV, maternal death, and child survival in Africa. AIDS 10(1):111-112. UNAIDS//UNICEF/USAID. 2004. Children on the Brink: A Joint Report of New Orphan Estimates and a Framework for Action. Washington D.C.: USAID. http://data.unaids.org/ Publications/External-Documents/unicefchildrenonthebrink 2004_en.pdf UNAIDS and WHO. 2007. AIDS Epidemic Update. Geneva: UNAIDS and WHO. UNICEF. N.d. A promise to children, http://www.unicef.org/wsc/ UNICEF. 2003. Africa's orphaned generations. New York: UNICEF. UNICEF. 2005a. Field work manual for MICS3. New York: UNICEF http://www.childinfo.org/mics/mics3/archives/gambia/survey0/data/Fieldwork /MANUALMICS3FIELDWORK.pdf UNICEF. 2008b. MICS3 national household dataset for Cote d'lvoire, http://www.childinfo.org/mics3_surveys.html (first accessed January 31, 2008). UNICEF. 2008a. MICS3 national household dataset for The Gambia, http://www.childinfo.org/mics3_surveys.html (first accessed May 14, 2008). UNICEF. 2008c. MICS3 national household dataset for Sierra Leone, http://www.childinfo.org/mics3_surveys.html (first accessed January 28, 2008). 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNICEF. 2008d. MICS3 national household dataset for Somalia, http://www.childinfo.org/mics3_surveys.html (first accessed March 13, 2008). UNICEF. 2000. Monitoring the Progress Towards the Goals of the World Summit for Children: End Decade Multiple Indicator Survey Manual New York: United Nations Children's Fund. UNICEF. 2006. Multiple Indicator Cluster Survey Manual 2005: Monitoring the Situation of Children and Women. New York: UNICEF. http://www.childinfo.org/mics3_manual.html UNICEF. 2010. Surveys: list of confirmed surveys, http://www.childinfo.org/mics4_ surveys.html UNICEF Somalia. 2006. Multiple Indicator Cluster Survey 2006: Monitoring the situation of women and children. Mogadiscio: UNICEF United Nations Development Programme (UNDP). 2009b. HDR 2009 Statistical Tables. http://hdr.undp.org/en/reports/global/hdr2009/. New York: United Nations Development Programme. United Nations Development Programme (UNDP). 2009a. Human Development Reports. http://hdrstats.undp.org/en/countries/country_fact_sheets/cty_fs_LIE.html United Nations Population Fund (UNFPA) Country Support Team, Addis Ababa, Ethiopia. 2003. Safe Motherhood In Africa: Social and Economic Impact of Maternal Deaths. Paper Presented at the "7th Conference Of African Ministers of Health," Tripoli, Libya, April 26-30 2003 http://cst.addisababa.unfpa.org/docs/ad-safemotherhood.doc United Nations Population Fund (UNFPA). 2005. The State of the World Population 2005, The Promise of Equality: Gender Equity, Reproductive Health and the MDGs. New York: United Nations Population Fund. USAID. N.d. Map of Somalia, http://www.usaid.gov/our_work/cross-cutting_programs/ transitioninitiatives/country/somalia/somap.jpg Verschueren, K., and A. Marcoen. 1999. Representation of self and socioeconomic competence in kindergarteners: Differential and combined effects of attachment to mother and father. Child Development 70:183-201. Waters, H.R., L. Dougherty, S.P. Tegang, N. Tran, C.S. Wiysonge, K. Long, N.D. Wolfe, and D. S. Burke. 2004. Coverage and costs of childhood immunizations in Cameroon. Bulletin of the World Health Organization 82(9):668-675. Wood, K., E. Chase, and P. Aggleton. 2006. Telling the truth is the best thing: Teenage orphans' experiences of parental AIDS-related illness and bereavement in Zimbabwe. Social Science and Medicine 63:1923-1933. 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. World Bank. 2002. Building Statistical Capacity to Monitor Development Progress. http://siteresources.worldbank.org/SCBINTRANET /Resources/239410 to 1113334813340Zboard-paper-feb4.pdf World Bank. 2005. Sri Lanka: Attaining the Millennium Development Goals in Sri Lanka How Likely and What Will it Take to Reduce Poverty, Child Mortality and Malnutrition, and to Increase School Enrollment and Completion? Washington D.C.: World Bank. World Health Organization (WHO). 2001. Advancing Safe Motherhood through Human Rights. Geneva: World Health Organization. World Health Organization (WHO). 2007. International Statistical Classification of Diseases and Related Health Problems 10th Revision, Version for 2007. http://apps.who.int/ classifications/apps/icd/icd1Oonline/. World Health Organization (WHO). 2007. International Statistical Classification of Diseases and Related Health Problems 10th Revision Version for 2007. Geneva: World Health Organization. http://www.who.int/classifications/apps/icd/icdlOonline/ World Health Organization (WHO). 2006a. Global tuberculosis control: surveillance, planning, financing. Geneva: World Health Organization. World Health Organization (WHO). 2006b. Making a difference in countries: Strategic Approach to Improving Maternal and Newborn Survival and Health. Geneva: World Health Organization. World Health Organization (WHO). 2006c. The Health of the People: The African Regional Health Report. Geneva: The World Health Organization. World Health Organization (WHO). 2005. The world health report: 2005: make every mother and Child count. Geneva: World Health Organization. World Health Organization (WHO). 2003. The world health report: 2003: shaping the Future. Geneva: World Health Organization. World Health Organization (WHO). 2009a. Water-related diseases. Geneva: World Health Organization, http://www.who.int/water_sanitation_health/diseases/undernutrition/en/ print.html World Health Organization (WHO). 2009b. WHO child growth standards and the identification of severe acute malnutrition in infants and children. Geneva: World Health Organization, http://www.who.int/nutrition/publications/severemalnutrition/ 9789241598163/en/index.html World Health Organization (WHO). 2010. WHOSIS. WHO Statistical Information System. http://www.who.int/whosis/en/index.html. 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX Frequency distributions for anthropometry data Figure Al. Frequency distribution, height for age z-scores, Figure A2. Frequency distribution, height for age z-scores, children under five, for The Gambia, 2005/2006 children under five, for Cote d'lvoire, 2006 Mean = 1.03 SD= 1.47 n = 6388 -rriTfll' Mean =1.41 SD= 1.74 n= 8341 i-r -5.00 I I -2.50 0.00 2.50 Height for age z-score I 5.00 -5.00 -2.50 0.00 2.50 Height for age z-score 5.00 Figure A3. Frequency distribution, height for age z-scores, Figure A4. Frequency distribution, height for age z-scores, children under five, for Sierra Leone, 2005 children under five, for Somalia, 2006 400- 400Mean =1.51 SD= 1.89 n = 4326 300- Mean= 1.33 SD= 1.94 n = 5526 300- 100- 100- 0J 0 -5.00 -2.50 0.00 2.50 Height for age z-score 5.00 -5.00 -2.50 0.00 2.50 5.00 Height for age z-score 124 Figure Bl. Frequency distribution, weight for age zscores, children under five, for The Gambia, 2005/2006 Figure B2. Frequency distribution, weight for age z-scores, children under five, for Cote d'lvoire, 2006 800" Mean = 0.99 SD= 1.34 n = 8463 Mean = -1.02 SD= 1.23 n = 6422 600- I §"400£ 200- o-5.00 -2.50 0.00 2.50 Weight for age z-score 5.00 -5.00 -2.50 0.00 2.50 5.00 Weight for age z-score 125 Figure B3. Frequency distribution, weight for age zscores, children under five, for Sierra Leone, 2005 500" Figure B4. Frequency distribution, weight for age z-scores, children under five, for Somalia, 2006 600- Mean = 1.29 SD= 1.50 n = 4465 400- Mean = 1.36 SD = 1.53 n = 5657 500400- g 300- 200200- 100100- -5.00 -2.50 0.00 2.50 Weight for age z-score 5.00 -5.00 -2.50 0.00 2.50 5.00 Weight for age z-score 126 Figure CI. Frequency distribution, weight for height zscores, children under five, for The Gambia, 2005/2006 Figure C2. Frequency distribution, weight for height zscores, children under five, for Cote d'lvoire, 2006 800- Mean = 0.47 SD= 1.11 n = 6386 600- Mean = 0.13 SD= 1.39 n = 8245 600- g I 400& 400- 200200- -4.00 -2.00 0.00 2.00 Weight for height z-score 4.00 6.00 0 -4.00 -2.00 0.00 2.00 Weight for height z-score 4.00 Figure C3. Frequency distribution, weight for height zscores, children under five, for Sierra Leone, 2005 500- Figure C4. Frequency distribution, weight for height zscores, children under five, for Somalia, 2006 600- Mean = 0.32 SD = 1.34 n = 4285 400- Mean = 0.64 SD = 1.33 n = 5497 500- 400g 300- 200200100— 100Mb o-1 -4.00 -2.00 0.00 2.00 Weight for height z-score 4.00 6.00 0 -4.00 -2.00 0.00 2.00 Weight for height z-score 4.00 6.00