The Prevalence and Context of Adult Female Overweight and Obesity in Sub-Saharan Africa

Home , Eleme people, Gashaka, Ilaje, Mafa people, Obolo people

East Tennessee State University Digital Commons @ East Tennessee State University

Electronic Theses and Dissertations Student Works

5-2019

Ifeoma Ozodiegwu East Tennessee State University

Follow this and additional works at: https://dc.etsu.edu/etd

Part of the Epidemiology Commons, International Public Health Commons, and the Nutritional and Metabolic Diseases Commons

Recommended Citation Ozodiegwu, Ifeoma, "The Prevalence and Context of Adult Female Overweight and Obesity in Sub-Saharan Africa" (2019). Electronic Theses and Dissertations. Paper 3566. https://dc.etsu.edu/etd/3566

This Dissertation - unrestricted is brought to you for free and open access by the Student Works at Digital Commons @ East Tennessee State University. It has been accepted for inclusion in Electronic Theses and Dissertations by an authorized administrator of Digital Commons @ East Tennessee State University. For more information, please contact [email protected].

The Prevalence and Context of Adult Female Overweight and Obesity in Sub-Saharan Africa

A dissertation

presented to

the faculty of the Department of Biostatistics and Epidemiology

East Tennessee State University

In partial fulfillment

of the requirements for the degree

Doctor of Public Health in Epidemiology

Ifeoma Doreen Ozodiegwu

May 2019

Dr. Megan Quinn, Chair

Dr. Hadii Mamudu

Dr. Mary Ann Littleton

Dr. Henry Doctor

Dr. Laina Mercer

Keywords: Overweight, Obesity, Sub-Saharan Africa, Adult Women

ABSTRACT

The Prevalence and Context of Adult Female Overweight and Obesity in Sub-Saharan Africa

Ifeoma Doreen Ozodiegwu

Adult women bear a disproportionate burden of overweight and obesity in Sub-Saharan Africa

(SSA). Precise information to understand disease distribution and assess determinants is lacking.

Therefore, this dissertation aimed to: (i) analyze the prevalence of adult female overweight and obesity combined in lower-level administrative units; (ii) analyze the effect modification of educational attainment and age on the association between household wealth and adult female overweight and obesity; (iii) synthesize qualitative research evidence to describe contextual factors contributing to female overweight and obesity at different life stages. Bayesian and logistic regression models were constructed with Demographic and Health Survey (DHS) data to respectively estimate the prevalence of overweight and obesity and assess the interaction of education on the association between household wealth and overweight. The synthesis of qualitative research studies was conducted in accordance with PRISMA guidelines and findings were grouped by themes. Prevalence estimates revealed heterogeneity at second-level administrative units in the seven SSA countries examined, which was not visible in first-level administrative units. The combined prevalence of overweight and obesity ranged from 7.5 –

42.0% in Benin, 1.4 – 35.9% in Ethiopia, 1.6 – 44.7% in Mozambique, 1.0 – 67.9% in Nigeria,

2.2 - 72.4% in Tanzania, 3.9 – 39.9% in Zambia, and 4.5 - 50.6% in Zimbabwe. Additionally, education did not have a statistically significant modifying effect on the positive association between household wealth and overweight in the 22 SSA countries eligible for the study. Body shape and size ideals, barriers to healthy food choices and physical activity were key themes in the research synthesis encompassing four SSA countries. Positive symbolism, including beauty, was linked to overweight and obesity in adult women. Among adolescents, although being overweight or obese was not accepted, girls were expected to be voluptuous. Body image dissatisfaction and victimization characterized the experiences of non-conforming women and girls. Barriers to healthy nutrition included migration and the food environment. Whereas, barriers to physical activity included ageism. While additional work is encouraged to validate the prevalence estimates, overweight and obesity interventions must consider whether the determinants identified in this study are relevant to their context to inform improved outcomes.

DEDICATION

This dissertation would not be possible without the unrivaled love and long-term support of my parents, Dr. Festus and Mrs. Rose Ozodiegwu. I am continually inspired by the endless sacrifices that they have made to ensure that my siblings and I attain our fullest potential. I am also motivated by the contribution they have made to improving health and education in Nigeria, specifically, Enugu, our home state and in Aku, our village. I hope that my work in global health and in Nigeria, will parallel a tenth of yours and that I can be the best daughter to you.

To my siblings, Ikenna, Chika, Tobi, the debates that we had growing up has allowed me to challenge my ideas and push the boundaries of my work. Thank you all for your love, support, words of encouragement and listening ears.

To the love of my life, Tony, there are no words to convey my appreciation for your unending support in all my pursuits.

To my friends, Fr Bede, Mary, Chinwe, Daniel, Tosin, Ruby, Micheal, Martin, Jocelyn,

Herman, Sylvester, Ena, Seun, Christian who created several special moments involving delectable food, wine, dance and jokes as I progressed through this program, I am really thankful for you all.

To all African women, I hope this dissertation project brings attention and resources to the problem of overweight and obesity in our communities.

ACKNOWLEDGEMENTS

I am grateful to Rotary International District 7570 for their financial support while I completed this dissertation project.

Many thanks to Dr. Quinn, my academic advisor and dissertation chair for her constant support from the first day I entered the doctoral program. Your patience, understanding and wise counsel is deeply appreciated. And also to Dr. Mamudu, my research mentor, your incredible support throughout this program including nominating me for the Rotary Global Grant/Skelton

Jones Scholarship has enabled me to be free from financial worries as I wrote my dissertation and I will remain forever grateful.

Dr. Doctor, or Oga Doctor, as we call you in Nigeria, thank you so much for encouraging me to pursue a doctoral degree and for supporting me through it. Your sense of humor is unsurpassed and I really enjoy working with you. Also, I am fortunate to have Dr. Littleton on my committee. Many thanks for accepting to work with me on my dissertation project even though I do not have a strong background in qualitative research. Thank you for believing in me and for the insights you shared.

Finally, I would like to thank Dr. Mercer, the most recent addition to my committee, whose work in the area of spatial models for small area estimation has been the foundation of my manuscript on estimating the prevalence of overweight in lower administrative units. Thank you so much for readily sharing your knowledge and for making my ideas come to fruition.

TABLE OF CONTENTS

Page

ABSTRACT ...... 2

DEDICATION ...... 5

ACKNOWLEDGEMENTS ...... 6

LIST OF TABLES ...... 14

LIST OF FIGURES ...... 16

LIST OF ACRONYMS ...... 18

Chapter

1. INTRODUCTION ...... 19

Problem Statement ...... 19

Specific Aims ...... 21

Significance of this Project ...... 21

Research Gaps in Overweight and Obesity Prevalence in SSA Countries ...... 21

Research Gaps in Overweight and Obesity Determinants in SSA Countries ...... 23

Organization of Chapters ...... 27

2. BACKGROUND ...... 28

Global and Regional Overweight and Obesity Burden and Trends ...... 28

Conceptualizing and Defining Overweight and Obesity...... 30

Evaluating Overweight and Obesity ...... 32

Anthropometric Indicators – Body Mass Index (BMI)...... 32

Strengths and Limitations of BMI as a Measure of Overweight and Obesity ...... 35

Anthropometric Indicators – Waist Circumference, Waist-Hip Ratio and Waist-Height

Ratio ...... 39

Body Fat Thresholds ...... 42

Consequences of Overweight and Obesity...... 43

Health and Economic Consequences ...... 43

Mechanisms Underlying the Relationship between Overweight and Obesity with Chronic

Conditions ...... 45

Obesity Paradox ...... 48

Lowest-Risk BMI...... 50

Causal Factors for Overweight and Obesity ...... 51

Energy Homeostasis...... 52

Genetic Factors ...... 54

Developmental Basis of Adult Overweight and Obesity ...... 55

Parental Undernutrition ...... 55

Parental Over-Nutrition ...... 57

Endocrine Disrupting Chemicals ...... 59

Stress ...... 61

Social and Economic Conditions ...... 63

Dietary Factors ...... 68

Sedentary Behavior ...... 70

Overweight and Obesity Interventions ...... 70

Individual-level Interventions ...... 71

Community-level Interventions ...... 72

Policy Approaches ...... 72

3. DATA SOURCES ...... 75

Demographic and Health Surveys (DHS) ...... 75

The DHS Questionnaire ...... 76

DHS Sampling Design ...... 77

DHS Geographic Data ...... 79

Strengths of the DHS ...... 79

Description and Critique of the Dissertation Dataset from the DHS ...... 80

Handling Missing Data ...... 82

Sources of Qualitative Research Studies for Research Synthesis in this Dissertation ...... 85

4. ADULT FEMALE OVERWEIGHT AND OBESITY PREVALENCE IN SEVEN SUB-

SAHARAN AFRICAN COUNTRIES: A BASELINE SUB-NATIONAL ASSESSMENT OF

INDICATOR 14 OF THE GLOBAL NCD MONITORING FRAMEWORK ...... 86

Summary ...... 87

Background ...... 87

Methods...... 87

Findings...... 87

Interpretation ...... 88

Funding ...... 88

Research in Context ...... 89

Evidence before this study ...... 89

Added value of this study ...... 89

Implications of all the available evidence ...... 90

Introduction ...... 91

Methods ...... 93

Data Sources ...... 93

Data Compilation ...... 94

Statistical Analysis ...... 95

Funding...... 99

Results ...... 99

Participants and Locations ...... 99

Findings...... 102

Discussion ...... 108

References ...... 112

Supplementary Appendix...... 117

5. THE EFFECT MODIFICATION OF EDUCATIONAL ATTAINMENT ON THE

ASSOCIATION BETWEEN HOUSEHOLD WEALTH AND FEMALE OVERWEIGHT IN

LOW AND MIDDLE-INCOME SUB-SAHARAN AFRICAN COUNTRIES: A CROSS-

SECTIONAL STUDY ...... 183

Abstract ...... 184

Background ...... 184

Methods...... 184

Results ...... 184

Conclusions ...... 185

Background ...... 186

Methods ...... 188

Data Source ...... 188

Conceptual Model ...... 188

Data Compilation ...... 189

Statistical Analysis ...... 191

Results ...... 194

Participants ...... 194

Descriptive Data...... 194

Main Results ...... 204

Discussion ...... 211

Conclusions ...... 213

List of abbreviations ...... 213

Declarations ...... 214

Ethics approval and consent to participate...... 214

Consent for publication ...... 214

Availability of data and materials ...... 214

Competing interests ...... 214

Funding ...... 215

Authors' contributions ...... 215

Acknowledgements ...... 215

References ...... 216

Supplementary Appendix...... 222

6. A QUALITATIVE RESEARCH SYNTHESIS OF CONTEXTUAL FACTORS

CONTRIBUTING TO FEMALE OVERWEIGHT AND OBESITY OVER THE LIFE

COURSE IN SUB-SAHARAN AFRICA ...... 237

Abstract ...... 238

Objective ...... 238

Methods...... 238

Results ...... 238

Significance...... 239

Introduction ...... 240

Methods ...... 242

Search Strategy ...... 242

Inclusion Criteria ...... 243

Screening...... 243

Critical Appraisal ...... 244

Data Extraction ...... 246

Data Synthesis ...... 246

Results ...... 247

Description of the included studies ...... 247

Data Synthesis Findings ...... 247

Body size and shape ideals ...... 247

Barriers to healthy eating ...... 250

Barriers to engaging in physical activity ...... 251

Discussion and Conclusion ...... 253

References ...... 259

Supplementary Appendix...... 268

7. DISCUSSION ...... 281

Summary of Main Findings ...... 281

Strengths and Limitations...... 283

Implications of the Study Findings and Recommendations for Future Work...... 284

REFERENCES ...... 286

APPENDIX: Missng Data Patterns for DHS Dataset ...... 306

VITA ...... 318

LIST OF TABLES

Table Page

3.1 Listing of Available DHS Data…………………………………………..………….81

4.1 Countries and Administrative levels..………………………………………………...95

4.2 Data Input Sample Sizes, Exclusions & National Level Overweight Prevalence for

Women, 18 Years and Older………………………………………………...………100

4.3 Misclassification Probabilities for Assessing the Locational Accuracy of the DHS

Cluster Locations For Nigeria (DHS, 2013)………………………………………...107

S4.1 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Benin………………………………………118

S4.2 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Benin…………………………………….…119

S4.3 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Ethiopia…………………………………….123

S4.4 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Ethiopia………………………………….....124

S4.5 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Mozambique……………………………….130

S4.6 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Mozambique...……………………………..131

S4.7 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Nigeria…….………………………….……136

S4.8 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Nigeria……………………………………..138

S4.9 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Tanzania...……………………………….…164

S4.10 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Tanzania……………………………………166

S4.11 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Zambia……………………………………..173

S4.12 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Zambia…………………………………..…174

S4.13 Age-Standardized Overweight Prevalence in First-Level Administrative Units

among Women, 18 Years and Older in Zimbabwe………………………………….178

S4.14 Age-Standardized Overweight Prevalence in Second-Level Administrative Units

among Women, 18 Years and Older in Zimbabwe…………...……………………..179

5.1 Description of Included Surveys and Overall Overweight Prevalence in 34 SSA

Countries…………………………………………………………………………..…195

5.2 Prevalence of Overweight and Underweight/Normal Weight Combined by Strata

of Educational Attainment and Wealth Status in 34 SSA Countries………………...200

S5.1 Prevalence of Overweight by Wealth Tertiles in 34 SSA Countries………………...223

S5.2 Odds of Overweight by Wealth Tertile in 22 SSA Countries………………………227

S5.3 Odds Ratio and 95% CIs for the Interaction of Educational Attainment on the

Association between Household Wealth and Overweight in 22 SSA Countries……230

S6.1 Description of Included Studies……………………………………………………..272

LIST OF FIGURES

Figure Page

2.1 Female Overweight and Obesity (≥ 25kg/m2) Trends by Country Income Level……....64

2.2 Male Overweight and Obesity (≥ 25kg/m2) Trends by Country Income Level………...65

4.1a Age-Standardized Prevalence of Overweight among Women, 18 Years and Older…..104

4.1b Age-Standardized Prevalence of Overweight among Women, 18 Years and Older…..105

4.1c Age-Standardized Prevalence of Overweight among Women, 18 Years and Older…..106

S4.1 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older,

Benin Communes Overlapped with Relative Uncertainty (2011 – 2012)………….….122

S4.2 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older, in

Ethiopian Zones Overlapped with Relative Uncertainty (2016)……………………….129

S4.3 Age-Standardized Overweight Prevalence among Women, 18 Years and Older, in

Mozambican Districts Overlapped with Relative Uncertainty (2011)………………....135

S4.4 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older in

Nigerian LGAs Overlapped with Relative Uncertainty (2013)………………………..163

S4.5 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older, in

Tanzanian Districts Overlapped with relative Uncertainty (2015 – 2016)…………….172

S4.6 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older, in

Zambian Districts Overlapped with Relative Uncertainty (2013 – 2014)……………..177

S4.7 Age-Standardized Prevalence of Overweight among Women, 18 Years and Older, in

Zimbabwean Districts Overlapped with Relative Uncertainty (2015)…………..…….182

5.1 Conceptual Model of the Association of Household Wealth and Overweight with

Educational Attainment and Age as Effect Modifiers…………………………………189

5.2 Plot of the Interaction of Education on the Association between Household Wealth

and Overweight in Low-Income SSA Countries……………………………………....205

5.3 Plot of the Interaction of Education on the Association between Household Wealth

and Overweight in Low-Middle Income SSA Countries……………………………...208

5.4 Plot of the Interaction of Education on the Association between Household Wealth

and Overweight in Upper-Middle Income SSA Countries……………..……………...210

S5.1 Plot of the Interaction of Education on the Association between Household Wealth

Quintiles and Overweight in Gabon……………………………………………………233

S5.2 Plot of the Interaction of Education on the Association between Household Wealth

Quintiles and Overweight in Namibia……………………………………………….....234

S5.3 Plot of the Interaction of Education on the Association between Household Wealth

Quintiles and Obesity in Gabon………………………………………………………..235

S5.4 Plot of the Interaction of Education on the Association between Household Wealth

Quintiles and Obesity in Namibia………………………………………………….…..236

6.1 PRISMA Flow Diagram………………………………………………………………. 245

LIST OF ACRONYMS

CVD Cardiovascular Disease DHS Demographic and Health Survey EDC Endocrine Disrupting Chemicals GBD Global Burden of Disease HPA hypothalamic-pituitary-adrenal HT Horvitz-Thompson INLA Integrated Nested Laplace Approximation LGA Local Government Area LMIC Low-and Middle-Income Countries MAR Missing at Random MNAR Missing not at Random NCD Non-communicable Disease SDGs Sustainable Development Goals SSA Sub-Saharan Africa/sub-Saharan African U.S. United States USAID United States Agency for International Development WHO World Health Organization

CHAPTER 1

INTRODUCTION

Problem Statement

Sub-Saharan African (SSA) countries are experiencing an upshot in overweight and obesity prevalence and trends, which disproportionately affect adult women. Data from the

World Health Organization (WHO) Global Health Observatory show that in the WHO Africa

Region, from 1975 to 2016, age-standardized adult female overweight and obesity prevalence have risen by 24% and 12% to 39% and 15%, respectively; whereas age-standardized adult male overweight and obesity in the same time period rose only by 15% and 5% to 23% and 6%, respectively (World Health Organization, 2017b).

The greater burden of overweight and obesity among African women is driving increased risk of mortality and disability from non-communicable diseases (NCDs), projected to become a leading cause of death in sub-Saharan Africa by 2030 (Marquez & Farrington, 2013), and poor pregnancy, labor, delivery and offspring outcomes. This can be attested to by an analysis of SSA countries conducted by the Global Burden of Disease (GBD) Collaborators which reveal that, except for a few countries, from 1990 to 2015, BMI-related death rates per 100,000 population and disability-adjusted life years per 100,000 population from NCDs among women showed a median increase of 36% and 42%, respectively (The GBD 2015 Obesity Collaborators, 2017).

Similarly, in a meta-analysis, Onubi et al. (2016) found that compared to normal weight pregnant women, obese pregnant women in continental Africa had roughly 2.4 times the risk of

gestational diabetes mellitus, 1.6 times the risk of pregnancy-induced hypertension, 1.8 times the risk of fetal macrosomia, and 1.6 times the risk of neonatal admission into an intensive care unit.

Further, in a modelling study, Kontis et al. (2015) concluded that obesity was the second most important risk factor that must be addressed among African women in order to achieve the voluntary global health target of a 25% decrease in premature mortality from cardiovascular diseases (CVD), cancers, chronic respiratory disease and diabetes, relative to 2010 levels, by

2025. Specific voluntary global targets related to obesity, according to the NCD Global

Monitoring Framework, is to halt the rise in obesity prevalence, which will be monitored by assessment of the age-standardized prevalence of overweight and obesity in individuals aged 18 years and older (World Health Organization, 2013). These targets were set by various countries in response to the commitments made at the 2011 United Nations High-level Meeting on NCDs

(Kontis et al., 2015). Additionally, meeting obesity-related targets will be crucial to achieving the Sustainable Development Goal (SDG) target of a one-third decrease in premature mortality from NCDs by 2030 (United Nations, 2015). However, efforts to address overweight and obesity are often targeted at children despite the research finding that over 70% of obese adults were not obese in childhood or adolescence (Simmonds, Llewellyn, Owen, & Woolacott, 2016).

Therefore, the excess burden of overweight and obesity and the elevated risk of, and rising trend in BMI-related morbidity and mortality among adult SSA women, in addition to the need to inform obesity-related indicator measurement and meet global NCD targets justify the goal of this dissertation project, which is to generate prevalence estimates of female overweight and obesity and identify potential determinants using both quantitative and qualitative approaches in order to support the development of NCD-focused interventions for adult SSA women.

Specific Aims

The current project will contribute to increased understanding of overweight and obesity in

SSA with the following specific aims:

1. Analyze the prevalence of adult female overweight and obesity in lower-level

administrative units in SSA countries using spatial models for small area estimation.

2. Analyze the effect modification of educational attainment on the relationship between

household wealth and female overweight and obesity in SSA countries through an

interaction analysis.

3. Identify, appraise and synthesize qualitative research evidence to describe contextual

factors contributing to female overweight and obesity at different life stages.

Significance of this Project

Research Gaps in Overweight and Obesity Prevalence in SSA Countries

Although available data for SSA countries describe the prevalence of adult female overweight and obesity at a regional, national level, state or provincial level, data at the lowest administrative levels (second or third-level administrative units) to inform local level priorities are often absent. Two key considerations support the need for greater precision in the characterization of adult female overweight and obesity prevalence in SSA. Firstly, most African countries are highly heterogeneous; tribal, cultural, and economic differences are not often captured by state or provincial divisions (Green, 2013). Hence, assessment of indicators at higher aggregate levels could conceal significant geographic inequities departing from the basic principle of the SDGs 2030 Agenda of leaving no one behind and the resolution of Inter-Agency and Expert Group on SDG Indicators, which state that “Sustainable Development Goal

indicators should be disaggregated, where relevant, by income, sex, age, race, ethnicity, migratory status, disability and geographic location, or other characteristics, in accordance with the Fundamental Principles of Official Statistics” (United Nations Economic and Social Council,

2016). However, while monitoring for overweight and obesity prevalence are not explicitly included in the SDG indicators, accomplishing obesity related targets in the NCD Global

Monitoring Framework will be crucial for meeting NCDs-related SDG targets and indicators

(World Health Organization, 2013). Local-level estimates will enhance ownership and participation of local stakeholders in action to achieve the aforementioned targets.

Secondly, the current environment of limited funding for international health and development calls for greater precision and efficiency in the allocation of resources, more targeted assessment of population health and dissemination and evaluation of public health programs. The importance of specificity or precision in public health estimates was recently highlighted by the Bill and Melinda Gates Foundation officials in an editorial, where they defined precision public health as “the use of data to guide interventions that benefit populations more efficiently” (Dowell, Blazes, & Desmond-Hellmann, 2016).

Methods for estimation of public health indicators with greater precision are routinely applied in high-income countries. In a study of sub-county areas in King County, Washington in the United States, Song et al. (2016) found significant geographic variation in smoking prevalence, ranging from 5 – 28%, which was used to inform county-level programs. Similar techniques will be applied to estimate female overweight and obesity prevalence in SSA

countries to fulfill specific aim one in order to empower local, national and international stakeholders to more effectively target advocacy and disease prevention and control efforts.

Research Gaps in Overweight and Obesity Determinants in SSA Countries

Sex-specific differences in overweight and obesity prevalence and trends in SSA could potentially be explained by biology, post-partum weight retention, and developmental factors.

Evidence, however, arise mostly from high-income countries and elsewhere because of the low availability of research evidence from SSA countries. For instance, women’s greater susceptibility to weight gain could be explained by the OPEN study, which found that compared to men, women living in the United States had lower energy expenditure (Lovejoy, Sainsbury, &

Stock Conference 2008 Working Group, 2009; Tooze et al., 2007). Additional evidence from the OPEN study also revealed that the risk of obesity among women further worsens with age and menopause due to further declines in energy expenditure and estrogen deficiency (Lovejoy et al., 2009).

The impact of childbearing on substantial weight gain among women is also widely studied in several high-income countries. The results from prospective longitudinal studies illustrated that first births are typically associated with persistent weight gain and that weight gain is linearly associated with preconceptual body size (Gunderson, 2009). Further, given that most SSA countries are experiencing a double burden of malnutrition – the coexistence of both undernutrition and overnutrition - sex differences in the impact of in-utero undernutrition may precipitate the elevated prevalence levels of female overweight and obesity in SSA countries. In a study conducted in South Africa, although the authors could not determine whether nutritional

deprivation occurred in-utero and continued throughout childhood, retrospective self-reports of nutritional deprivation in childhood by women was associated with obesity; whereas no corresponding effect was observed in men with similar exposures (Case & Menendez, 2009).

Similar findings have been noted in more rigorous studies in high-income countries such as a

2007 research study, which noted that females exposed to the 1944 – 1945 Dutch famine during gestation had statistically significant increases in weight and BMI and other indices of mass and mass distribution at 59 years, while males had no resultant effects at adulthood (Stein et al.,

2007).

Besides biological, pregnancy-related and developmental factors, the cultural context in which women and girls reside heighten their likelihood of becoming overweight or obese. For instance, in certain African countries, girls and women are fattened for marriage either through force-feeding, in the case of young girls, or consensual placement of into fattening-houses

(Brown & Konner, 1987; Ouldzeidoune, Keating, Bertrand, & Rice, 2013). While no prospective studies formally link pre-marital fattening with female overweight and obesity, in Mauritania where practices of force-feeding abound, the 2001 Demographic and Health Survey (DHS), a nationally representative sample survey of women 15 – 49 years, reported the highest prevalence of force-feeding among obese women (44.9 %) followed by overweight women (25.9%) (Office

National de la Statistique (ONS) [Mauritanie] and ORC Macro, 2001). These cultural practices may underlie the large sex differences in obesity prevalence in Mauritania estimated as 18.5% in adult women compared to 6.6% in adult men (World Health Organization, 2017b). Typically, force-feeding customs are based on societal norms or body image ideals that perceive overweight women as more desirable, physically mature and ready for marriage (Ouldzeidoune, Keating,

Bertrand, & Rice, 2013). However, identifying African countries where pre-marital fattening and associated cultural norms still persist is difficult because the context of overweight and obesity remain understudied with qualitative approaches.

Culture, moderated by marital status, age and socioeconomic factors, may also drive sedentary lifestyles and increased calorie intake by shaping women’s household roles. In a qualitative study conducted in Morocco, Batnitzky (2008) found that the care-taking roles conferred on women by marriage confined them to the home, leaving them with no time for leisure activities and exposing them to higher food consumption, since they are tasked with child-rearing and all cooking activities. However, in Moroccan society, women’s household roles varied with age and socioeconomic status (SES) as older women played a supervisory role and were more sedentary while younger women were the actual implementers of household duties

(Batnitzky, 2008). Also, although increasing SES was associated with fewer household duties, wealthier women with higher education espoused thinner body image ideals, hence, decreasing their likelihood of obesity (Batnitzky, 2008).

Collectively, the cited literature indicate that overweight and obesity disparities have both a biologic and sociocultural basis. However, Swinburn et al. (2011) suggests that the global obesity epidemic is better explained by changes in the global food system while variation in obesity prevalence in national and local environments can be accounted for differences by sociocultural, economic and transport environments. Secondary data allow for further study of socioeconomic and cultural factors that may explain national level differences in overweight and obesity prevalence in SSA.

Based on studies with nationally or domain representative samples from several SSA countries (Ethiopia, Ghana, Benin, Nigeria, Mali, Uganda), it is known that increasing SES among women, defined by wealth and/or education, is associated with an elevated odds of overweight and obesity (Abrha, Shiferaw, & Ahmed, 2016; Aitsi-Selmi, Bell, Shipley, &

Marmot, 2014; Dake, Tawiah, & Badasu, 2011; Gewa, Leslie, & Pawloski, 2013; Kirunda,

Fadnes, Wamani, Van den Broeck, & Tylleskar, 2015). Additionally, education does not modify the relationship of household wealth in certain SSA countries (Aitsi-Selmi et al., 2014). As postulated by Aitsi-Selmi et al. (2014), the differences in the SES-obesity relationship may be due to preponderance of food scarcity in low-income countries driving economically advantaged individuals to accumulate calories. The availability of new demographic and health surveillance data provide the impetus for further examination of the SES-obesity relationship for a wider range of SSA countries. Identifying SSA countries where education may modify the expected positive relationship between household wealth and overweight and obesity will strengthen understanding of the interrelationship of SES factors with overweight and obesity in SSA and serve to further justify additional studies and interventions. Therefore, this will be the key objective of specific aim two.

The growing availability of secondary qualitative data on public platforms also enable the study of the social and cultural world of African women to identify factors linked to weight gain in SSA. Understanding the lived experiences of SSA women, the consistency of these experiences across contexts and the life course will contribute to intervention development and may aid understanding of how SES and other factors may be driving weight gain among SSA

women. Consequently, a qualitative evidence synthesis of factors associated with overweight and obesity will be the focus of specific aim three.

Organization of Chapters

Prior to the fulfilment of the specific aims of this dissertation, a general introduction to the topic of overweight and obesity is provided. Definitions, specific measures for overweight and obesity and their historical development of as well as the epidemiology of overweight and obesity are reviewed from both a global and regional perspective in Chapter 2. This review paints a clearer picture of the problem of overweight and obesity and informs the choice of measures used in this dissertation. It also provides context for the description of the strengths and limitations of the studies undertaken in this dissertation. Chapter 3 describes the secondary data sources and datasets employed to fulfill the specific aims tied to this project. Chapters 4 through

6 present the justification and methodology for each of the individual manuscripts associated with the project specific aims. All of the studies in the individual manuscripts received ETSU

IRB approval and were not designated as research involving human subjects.

CHAPTER 2

BACKGROUND

Global and Regional Overweight and Obesity Burden and Trends

Overweight and obesity are a global health problem. The WHO estimated that in 2016, nearly 1.9 billion adults, 18 years and older, were overweight, representing 39% of all adult; of which 650 million adults ( 18 years) were obese amounting to 13% of the global adult population (World Health Organization, 2018a). The prevalence of age-standardized overweight was similar in both genders (39% of men vs. 39% of women) while age-standardized obesity was higher in women (11% of men vs. 15% of women) (World Health Organization, 2018a). A deeper analysis of obesity data from 1980 to 2015 by the GBD revealed that women generally have a higher prevalence of obesity than men for all age brackets with peak obesity prevalence present in women between the ages of 60 and 64 years and in men between the ages of 50 and 54 years (The GBD 2015 Obesity Collaborators, 2017). Further, the global obesity burden nearly tripled between 1975 and 2014, growing faster in men than women. According to the NCD Risk

Factor Collaboration (NCD-RisC) (2016), age-standardized prevalence of obesity in men soared from 3% in 1975 to 11% in 2014, and from 6% to 15% in women.

In the WHO Africa region, a political categorization of African countries, which includes

Algeria, and the majority of SSA countries, except south Sudan and Somalia, the respective population prevalence of age-standardized adult overweight and obesity was estimated as 31% and 11% (World Health Organization, 2017b). The highest prevalence of age-standardized adult overweight and obesity was observed in the Americas (62.5% and 28.6%) followed by Europe

(58.7% and 23.3%) (World Health Organization, 2017b). In WHO Africa, Americas and Europe regions, women had a greater prevalence of obesity compared to men. However, the disparity between the prevalence of obesity among men and women was greatest in the Africa region

(5.3% vs 15.6%) (World Health Organization, 2017b). Between 1975 and 2016, age- standardized overweight and obesity prevalence increased by 19.5% and 8.6% respectively in the

WHO Africa region and by 25.9% and 19.1% in the WHO Americas (World Health

Organization, 2017b). In the WHO Europe and Eastern Mediterranean regions respectively, age- standardized overweight and obesity also increased by 19.5% and 9.9%, and by 25.6% and

15.1% (World Health Organization, 2017b).

At a country level, the prevalence of age-standardized obesity among the 20 most populous countries was highest in Egypt (at 35.3%) and lowest in Vietnam (1.6%) (The GBD

2015 Obesity Collaborators, 2017). In terms of absolute numbers, the United States and China had the highest numbers of obese adults (The GBD 2015 Obesity Collaborators, 2017). The prevalence of adult obesity is estimated to have doubled in 73 countries, with many other countries seeing steady increases (The GBD 2015 Obesity Collaborators, 2017).

When the prevalence of obesity was examined by the level of country development, women were shown to have a higher prevalence than men and the prevalence of obesity increased with increasing country wealth. According to the GBD, which uses the sociodemographic index (SDI) to rank the level of development within countries or geographic locations, in 2015, at all SDI levels and for all age groups, women had a higher prevalence of obesity compared to men (The GBD 2015 Obesity Collaborators, 2017). Indeed, the highest

prevalence of obesity was observed among women 60 to 64 years living in countries with high

SDI (The GBD 2015 Obesity Collaborators, 2017). In general, a positive trend was observed between the prevalence of obesity among women, men and across SDI for all age groups, with the exception of women living in countries with a low SDI, where the prevalence of obesity after age 55 was higher than that observed for women in countries with a low-middle SDI (The GBD

2015 Obesity Collaborators, 2017).

Further, soaring obesity trends were observed in countries with low-middle SDI. From

1980 to 2015, the greatest increases in the prevalence of obesity occurred among men between the ages of 25 and 29 years who were living in countries with a low-middle SDI — from 1.1% in

1980 to 3.8% in 2015 (The GBD 2015 Obesity Collaborators, 2017). Simultaneously, obesity prevalence rose by a factor of 1.7 among both men and women living in countries with a low- middle SDI (The GBD 2015 Obesity Collaborators, 2017).

Conceptualizing and Defining Overweight and Obesity

The WHO defines overweight and obesity as “abnormal or excessive fat accumulation that may impair health” (World Health Organization, 2017c). Overweight and obesity is regarded as a disease because affected persons have an undesirable positive energy balance and weight gain that can act through various metabolic channels to negatively affect their health. Fat distribution patterns in overweight and obesity are generally described as android or gynoid

(World Health Organization, 2000). Individuals with android fat distribution may be overweight or obese but also have an excessive accumulation of abdominal fat, central or visceral obesity, whereas gynoid patterns show an even distribution of fat across the entire body (World Health

Organization, 2000). Both the amount of excess fat stored and the regional distribution of fat within the body are what determine overweight and obesity related risks and outcomes (World

Health Organization, 2000). While overweight and obesity is associated with increased risk of chronic disease related morbidity and mortality, those with visceral obesity have poorer metabolic profiles and increased chronic disease risk (Abraham, Pedley, Massaro, Hoffmann, &

Fox, 2015; Liu et al., 2010; Piché, Poirier, Lemieux, & Després, 2018).

Prior to the medicalization of excessive body fat around the 19th and 20th century, global cultural connotations of corpulence, which continue to persist in some low- and middle-countries

(LIMICs), was that of health, happiness, beauty, wealth and influence (Eknoyan, 2006; Puoane,

Tsolekile, & Steyn, 2010). As discussed by Brown & Konner (1987) in their treatise on an anthropological perspective on obesity, the preponderance of food shortages and the rarity of obesity in traditional societies may have given rise to the positive symbolism associated with large body sizes. These views were supported by Eknoyan (2006), who described how female figures in Europe were depicted as plump, up to the early 20th century, as a sign of their health and prosperity. While there may not be accessible records of pre-19th century cultural perspectives of fatness in SSA societies, Brown & Konner (1987) provide examples from the early and mid-20th century literature describing how elite pubescent girls among the Efik in

Nigeria are prepared for marriage by seclusion in fattening houses to beautify them through an increase in body size. Changing views on the significance of excess body weight in Western countries were precipitated by the writings of several prominent medical figures such as

Hippocrates and William Banting (Haslam, 2007). Moreover, early 20th century actuarial studies showing a link between overweight and obesity with increased morbidity and mortality led its

acceptance of as a health problem in the West and prompted research to characterize it (Eknoyan,

2006; Komaroff, 2016; Nuttall, 2015).

Evaluating Overweight and Obesity

Anthropometric Indicators – Body Mass Index (BMI)

Approaches to assess overweight and obesity have evolved over time and typically fall into two categories: a) those that compare an individual’s weight to the average population weight, which serves as the reference value or b) those that compare it to weight standards obtained by studying the relationship between body weight and mortality or morbidity

(Komaroff, 2016). The goals of these approaches were to provide the best indicator of body fatness, body fat-related illness, and mortality.

Tables of average weight for different age groups in Belgium developed by Adolphe

Quetelet (1796 -1874), a Belgian astronomer and statistician, and used as a standard to compare an individual’s weight to the population average fall into the first category of techniques for assessing overweight and obesity (Komaroff, 2016). Quetelet was the first to apply the Gaussian distribution to the study of human physical characteristics concluding that an individual’s weight increases as a function of the square of their height. He called this measure – W/H2- the

Anthropometric Index. (Komaroff, 2016; Nuttall, 2015).

Following this discovery, Ancel Keys (1904 – 2004), a notable physiology expert, renamed the Anthropometric Index as the Body Mass Index (BMI). Further, he examined the relationships between BMI and historical indices of relative weight (e.g. simple ratio of weight to

퐻 height, the ponderal index – ⁄3 ) as well as the relationship of these indices to a more direct √푊 measure of body fat (body density and skin fold thickness) to determine the best indicator of body fatness or relative obesity (Keys, Fidanza, Karvonen, Kimura, & Taylor, 1972). The primary criteria for selecting the most appropriate relative weight index was that it had to be highly correlated with more objective measures of body fat and independent of height, that is, individuals of standard weight for their varying heights would have the same value of the index

(standard weight for height were calculated based on a regression equation representing the linear relationship between body weight and height) (Keys et al., 1972; Khosla & Lowe, 1967).

On account of this criteria and after his analysis of a sample of 7,426 men, Keys concluded that

BMI was the preferable relative weight index because it had the lowest correlation to height and the highest correlation to body fatness compared to the ponderal index and the ratio of weight to height (Keys et al., 1972).

BMI is currently the WHO recommended measure for measuring overweight and obesity and the mostly widely-used globally. It is still defined as a measure of a person’s weight in kilograms divided by the square of their height in meters. However, its present-day adaptation began with a report from a 1995 WHO Expert Committee on Physical Status that proposed BMI cutoff points according to its relationship with mortality (World Health Organization, 1995). In the report, individuals with a BMI less than 18kg/m2 were classified as underweight, 18 –

24.99kg/m2 as normal weight, 25 – 29.99kg/m2 as grade 1 overweight, 30 – 39.99kg/m2 as grade

2 overweight, and 40kg/m2 or greater as grade 3 overweight. The committee refrained from using the term “obesity” asserting that it implied knowledge of body composition, which, they believed, was an inaccurate view since BMI did not measure fat mass or fat percentage (World

Health Organization, 1995). In addition, they stated that the absence of recognized cut-off points for fat mass and fat percentage meant that these standards could not be translated to BMI cut-off points (World Health Organization, 1995). Hence, the committee chose BMI cutoff points grounded on the inflexion points of curves showcasing the relationship between BMI and mortality. The committee’s choice was informed by a meta-analysis conducted by Troiano et al.

(1996) (in press at the time), that revealed a U-shaped relationship between BMI and mortality for non-smoking 50 year old Caucasian men followed for 30 years (World Health Organization,

1995).

Although a subsequent WHO Consultation, which launched their report in the year 2000, accepted the BMI cutoff points recommended by the WHO Expert Committee on Physical

Status, the consultation participants renamed the category for grade 1 and grade 2 overweight as obesity arguing for its value in distinguishing individuals and groups at greatest risk of morbidity and mortality (World Health Organization, 2000). Further, the WHO Expert Committee proposed an additional BMI category of 35 – 39.9 kg/m2 due to the recognition that disease management options for obesity differ above 35kg/m2 (World Health Organization, 2000).

Consequently, the current WHO recommended international BMI classification is <18.5kg/m2 –

Underweight, 18.5 – 24.9kg/m2 – Normal weight, ≥25.0kg/m2 – Overweight (Pre-obese: 25.0 –

29.9kg/m2), and ≥30kgm2 – Obese (Obese class 1: 30 – 34.9kg/m2, Obese class II: 35 –

39.9kg/m2, Obese class III: ≥40kg/m2) (World Health Organization, 2018b).

Several objectives underlie the recommendation for universal BMI cut-off points for overweight and obesity. One of them was to facilitate the comparison of overweight and obesity

within and across populations and examination of causal effects of BMI on health outcomes

(World Health Organization, 2000). At the population level, the prevalence and risk estimates could be used to guide policy action and preventive programs. However, the WHO also recommended that among populations with a predisposition to central obesity and metabolic syndrome, additional information on waist circumference should also be considered in refining actions levels based on BMI estimates (World Health Organization Expert Consultation, 2004).

Secondly, the BMI cut-off points were pictured as a potential clinical decision-making tool to identify high risk individuals for screening and treatment (World Health Organization, 2000).

Nonetheless, the WHO Expert Committee on Physical Status also noted that clinical decisions must be made in consideration of the patient’s history and the presence of other related risk factors (World Health Organization, 2000).

Strengths and Limitations of BMI as a Measure of Overweight and Obesity

The simplicity, noninvasive and inexpensive nature of measuring BMI has made it a widely adapted surrogate measure of body fat. BMI calculations only requires measuring height and weight, which can be determined easily with a stadiometer and weighing scale respectively.

Moreover, BMI has been shown to be moderately correlated to better measures of body fat such as skin fold thickness, underwater weighing and dual energy X-ray absorptiometry (Deurenberg et al., 2001; Garrow & Webster, 1985; A Keys et al., 1972), and prediction formulas are available to estimate body fat percentage from BMI values (Deurenberg, Weststrate, & Seidell,

1991; Gallagher et al., 2000). In a meta-analysis of studies, it was estimated that BMI had a 50% sensitivity to identify excess body adiposity and a 90% specificity (Okorodudu et al., 2010).

Additionally, BMI is associated with all-cause mortality and chronic disease risk (Aune et al.,

2016; The GBD 2015 Obesity Collaborators, 2017).

However, BMI does not differentiate between weight associated with fat free tissue and weight associated with body fat (World Health Organization, 2000). Moreover, given that early validation work on BMI’s correlation with body fat was carried out mostly in Caucasian men

(Fletcher, 2014; A Keys et al., 1972), differences in body composition due to gender, age, athleticism and race/ethnicity may affect both the accuracy of BMI as a predictor of the levels of body fat and the appropriateness of BMI cut-off points for overweight and obesity. A review of gender differences in body composition show that women tend to store a large proportion of their body mass as fat (Power & Schulkin, 2008). Similarly, body composition changes with age leads to deposition of more body mass as body fat (Gallagher et al., 1996). Moreover, athletes tend to have a lower percentage body fat mass and higher fat free mass compared to non-athletes (Ode,

Pivarnik, Reeves, & Knous, 2007). Unsurprisingly, multiple regression analysis indicated that the inclusion of a sex and age effect in the prediction equation of BMI on percentage body fat improved the model fit (Gallagher et al., 1996; Jackson et al., 2002). In addition, Ravaglia et al

(1999) found that body fat percentage prediction formulas based on BMI tended to overestimate body fat after age 50. Further, in a study of athletes, Ode et al (2007) found that the use of BMI misclassified 67% of male athletes as being overweight.

The appropriateness of a universal BMI cut-off point for overweight and obesity across racial and ethnic groups has been extensively debated in the research literature (Deurenberg,

2001; Deurenberg, Deurenberg-Yap, & Guricci, 2002; World Health Organization (WHO)

Expert Consultation, 2004). That is because it raises the question of whether the relationship between BMI and body fat, and the relationship between BMI and morbidity and mortality risks differ by ethnic and racial groups. However, most of the research are based on small samples

(Deurenberg, Deurenberg Yap, Wang, Lin, & Schmidt, 1999; Gallagher et al., 1996; Luke et al.,

1997), is rife with misquotes and poorly drawn up hypothesis suggesting that there are currently validated thresholds for body fat percent for defining obesity (Ho-Pham, Campbell, & Nguyen,

2011; Rahman & Berenson, 2010); and incongruous conclusions such as the study by

Deurenberg, Yap, & van Staveren (1998) that concluded that African American have a 1.3kg/m2 lower BMI compared to Caucasians for the same percentage body fat, age and gender but provides no supporting information in their analysis. At the moment, no studies fully elucidate the relationship between BMI, body fat percentage, chronic disease risk, and mortality across various ethnic and racial groups. Hence, this remains an evolving area of research.

Nonetheless, due to the finding that Asians have a higher body fat percentage than

Caucasians of the same age, sex and BMI and the disclosure by a few studies that a large proportion of Asians with BMI below 25kg/m2 possess metabolic risk factors for type 2 diabetes and cardiovascular disease, the 2002 WHO expert consultation on BMI in Asian populations, while retaining existing BMI classifications, recommended BMIs greater than or equal to

23kg/m2 as trigger points for public health action for Asians (World Health Organization (WHO)

Expert Consultation, 2004). The trigger points has the following categories: 23 – 27.5 kg/m2 and

27.5kg/m2 or higher, representing increased and high risks respectively (World Health

Organization (WHO) Expert Consultation, 2004). Still, because of insufficient evidence, the

expert panel were unable to delineate clear cut-off points for Asian populations (World Health

Organization (WHO) Expert Consultation, 2004).

A paucity of research evidence also precludes the proper elucidation of the relationship between BMI, percent body fat, morbidity and mortality risk among people of African descent, and encumbers decision-making on the appropriateness of current BMI cut-off points for targeting high-risk individuals in this group. Studies among African Americans (AAs) suggest that they have a lower percentage body fat at a similar age and BMI compared to American

Whites (Deurenberg et al., 1998; Heymsfield, Peterson, Thomas, Heo, & Schuna, 2016).

Whereas, another study showed that Nigerians have a lower percentage body fat for the same

BMI compared to AAs (Luke et al., 1997). This implies that people of African descent should have higher BMI cutoff points than Caucasians. However, suggested BMI cut-off points for obesity vary by study and outcome. For instance, in an analysis of two cohort studies conducted between 1960 and 1989, Stevens et al. (2002) suggested lower BMI cut-off points than ≥

30kg/m2 (based on rate difference between whites and blacks) for diabetes and hypertension, and higher BMI cut-offs for mortality. A cross-sectional study of 552 Southwestern Nigerians by

Raimi & Dada (2018) also suggest lower cut-offs of 24.3kg/m2 and 28.9kg/m2 for men and women respectively to identify hypertension and hypercholesterolemia. Contrastingly,

Deurenberg et al. (1998) suggested a higher BMI cut-off than ≥ 30kg/m2 to identify obese AAs based on a meta-analysis of the relationship of BMI and body fat among different ethnic groups.

Additional research focused on clarifying the interrelationships of BMI cutoff points with percent body fat, chronic disease risk and mortality in specific ethnic groups will enhance understanding of the limitations of BMI in these groups and inform methods to ameliorate it.

Anthropometric Indicators – Waist Circumference, Waist-Hip Ratio and Waist-Height Ratio

Waist circumference, waist-hip ratio and waist-height ratio are proxy measures of central, abdominal or visceral obesity. In a cross-sectional analysis of the U.S based Pennington Center

Longitudinal Study, researchers demonstrated that waist circumference has a higher correlation with visceral fat compared to BMI (0.73-0.77 vs 0.61-0.69) (Camhi et al., 2011). More recently, a comparison of anthropometric measures indicated that waist-height ratio was the best predictor of visceral adipose tissue in men and women closely followed by waist circumference and BMI, whereas, waist to hip ratio was the poorest indicator of visceral fat (Ashwell, Gunn, & Gibson,

2012).

The notion that central obesity is a significant factor in chronic disease related morbidity and mortality arose from the findings of two research studies conducted in Sweden and published in 1984. In their analysis of a cohort of 1,462 women aged 38 – 60 years, who were followed for

12 years, Lapidus et al. (1984) found that waist-hip ratio was a stronger predictor of myocardial infarction, stroke and death compared to BMI. Similarly, in a study of a cohort of 792 men aged

54 years, followed up for 13 years, Larsson et al. (1984) observed that while waist-hip ratio predicted stroke and ischemic heart disease, there was no relationship between these end points and BMI or skinfold thickness. The subsequent emergence of several research papers on this issue and the need for guidance on the use of the indices of central obesity led to the assembling

of a WHO expert consultation on waist circumference and waist hip ratio in 2008 (World Health

Organization, 2011).

The WHO Expert Panel reviewed the literature on the relationship between indices of central obesity, BMI and cardiovascular (CVD) risk factors such as type II diabetes, hypertension and dyslipidemia (World Health Organization, 2011). In addition, they considered the evidence investigating the association between these anthropometric indices and other health outcomes such as incident CVD events and mortality (World Health Organization, 2011). Key goals of their review were to compare measures of central obesity with BMI as predictors of

CVD risk and to define potential cut-off points, based on the risk of poor health outcomes, that account for ethnic and sex differences (World Health Organization, 2011). Based on the available evidence at the time, the WHO expert panel resolved that waist circumference, waist- hip ratio, and BMI were likely comparable predictors of CVD risk (World Health Organization,

2011). The absence of sufficient data precluded any determination of the usefulness of waist- height ratio. Similarly, a definitive conclusion on cut-off points for waist circumference and waist-hip ratio and variations by ethnic group was not reached because of poor data availability.

However, it was suggested that Asians would likely have lower cut-off points due to their increased risk for chronic disease at any given waist circumference and waist-hip ratio compared to Europeans (World Health Organization, 2011).

Since the dissemination of this report in 2011, the status quo remains. There is currently no WHO recommended cut-off points for waist circumference and waist-hip ratio, although the aforementioned WHO Expert Panel recommended the formation of a working group for that

purpose. Existing recommendations attributed to the WHO are not recommendations but an example provided in a previous WHO report (World Health Organization, 2011). Additionally, other recommendations for cut-off points for indices of central obesity by the International

Diabetes Foundation, United States National Cholesterol Education Program and at the country- level do not have any methodological backing (World Health Organization, 2011). Current waist circumference and waist-hip ratio cut-offs used in most publications are based on a study that found these cutoffs to have high sensitivity and specificity for correct identification of overweight and obese persons and moderate sensitivity and specificity for cardiovascular risk factors in 2,183 and 2,698 Caucasian men and women in the Netherlands (Han, van Leer,

Seidell, & Lean, 1995). The waist circumference cut-offs indicating increased risk of metabolic complications for men and women are  94 cm and  80 cm (World Health Organization, 2011).

Greater than or equal to 102 cm and 88 cm in men and women respectively are used as cut-off points for substantially increased risk of metabolic complications (World Health Organization,

2011). Waist-hip ratio cut-offs that are currently used in the research literature are 0.90 cm for men and 0.85 cm for women (World Health Organization, 2011).

In comparison to waist circumference and BMI, other findings in the literature (post-

WHO expert panel) suggest that waist-height ratio is not only a useful predictor of CVD risk but also a superior predictor. For instance, in a meta-analysis of 31 studies that used receiver operating characteristics analysis to determine the anthropometric measure with the highest discriminatory ability, waist-height ratio improved discrimination of diabetes, hypertension and

CVD by 4 – 5% over BMI whereas waist circumference improved discrimination of similar outcomes by 3% (Ashwell et al., 2012). However, this meta-analysis lacked studies from Africa.

One of the few studies conducted in Africa, specifically Cameroon, with a sample of 8,663 adults aged 25 years and older found that waist circumference was the best predictor of diabetes

(Mbanya, Kengne, Mbanya, & Akhtar, 2015). These more recent studies provide additional fodder for continued research to identify, among diverse population, the best indicators of disease risk posed by central obesity and appropriate cut-off points.

Body Fat Thresholds

Although overweight and obesity is defined as excessive fat accumulation, there are currently no validated body fat thresholds that define this phenomenon (Ho-Pham et al., 2011).

An approach to developing body fat thresholds based on body mass index (BMI) cut-off points was proffered by Gallagher et al. (2000). In their paper, Gallagher and colleagues measured a subject’s percent body fat by using values from the following evaluations: underwater weighing for body volume and density, dual-energy X-ray absorptiometry (DXA) for body fat and bone mineral mass, and deuterium dilution for total body water. Together with participant’s BMI, age, sex and ethnicity, the calculated body fat values were used to develop percent body fat prediction equations. However, the WHO warns that such body fat prediction equations are prone to substantial bias at the individual and population level because DXA, deuterium dilution, and underwater weighing are indirect methods of measuring body fat that rely on assumptions that are not always true (World Health Organization Expert Consultation, 2004). Wells & Fewtrell

(2006) provides a general overview of the assumptions underpinning indirect measures of body fat. In-vivo neutron activation analysis (IVNAA), a more direct measure of body composition is too expensive to be used in population studies (World Health Organization Expert Consultation,

2004).

The volume of visceral adipose tissue has been measured using computed tomographic scanning in various studies (Abraham et al., 2015; Liu et al., 2010; Tokunaga, Matsuzawa,

Ishikawa, & Tarui, 1983). Nevertheless, there are no existing thresholds.

Consequences of Overweight and Obesity

Health and Economic Consequences

Overweight and obesity have serious health impacts. In 2015, overweight and obesity was linked to four million global deaths and 120 million disability adjusted life years (DALYs)

(The GBD 2015 Obesity Collaborators, 2017). This represented 7.1% of the deaths from any cause and 4.9% of the DALYs from any cause with 39% of the deaths and 37% of the DALYs occurring in individuals with a BMI below 30kg/m2 (The GBD 2015 Obesity Collaborators,

2017). Among women, high BMI was the second leading cause of DALYs while it was seventh in men (Gakidou et al., 2017). From 1990 to 2015, high BMI related death rates increased by

28.3% from 41.9 deaths per 100,000 population to 53.7 deaths per 100,000 population (The

GBD 2015 Obesity Collaborators, 2017). According to the level of development, as measured by the sociodemographic index (SDI), a composite average of income rankings, educational attainment and fertility, high-middle SDI and high SDI countries had the highest and lowest age- standardized rates of BMI-related deaths and disability-adjusted life years respectively (The

GBD 2015 Obesity Collaborators, 2017). Similarly, among the 20 most populous countries,

Russia and Democratic Republic of Congo had the highest and lowest rates of BMI-related deaths and disability adjusted life-years (The GBD 2015 Obesity Collaborators, 2017).

Available evidence indicates that overweight and obesity is associated with 20 health outcomes (The GBD 2015 Obesity Collaborators, 2017). Primary among them are CVD and diabetes, the two leading causes of high-BMI related deaths. In 2015, CVD led to 2.7 million or two-thirds of BMI-related deaths and 66.3 million disability-adjusted life years to become both the leading cause of death and disability from overweight and obesity (The GBD 2015 Obesity

Collaborators, 2017). Simultaneously, diabetes was a distant second in terms of mortality accounting for 0.6 million deaths and 30.4 million disability-adjusted life-years. Overweight and obesity are also associated with other NCDs including cancers, chronic kidney disease, dyslipidemia, osteoarthritis, lower back pain, hypertension, obstructive sleep apnea; and psychological illness such as mood and anxiety disorders (Heymsfield & Wadden, 2017; The

GBD 2015 Obesity Collaborators, 2017).

Among pregnant women, preconceptual obesity combined with pre-existing diabetes is associated with an increased risk of pregnancy loss, perinatal mortality, fetal macrosomia and congenital malformations (Poston et al., 2016). Additionally, obese pregnant women with pre- existing chronic hypertension exhibit an increased risk of superimposed pre-eclampsia, caesarean section, preterm delivery, low birthweight, neonatal unit and perinatal death (Poston et al., 2016).

However, even obese pregnant women without pre-existing conditions experience poor pregnancy outcomes including gestational diabetes, pregnancy induced hypertension and preeclampsia and preterm births (Poston et al., 2016). Moreover, during childhood, adolescence and adulthood, offspring of obese women are more likely to be obese and diagnosed with cardiometabolic abnormalities such as high blood pressure, adverse lipid profiles and insulin resistance (Godfrey et al., 2017). More importantly, associations have been found between

maternal obesity and offspring risk of cardiovascular disease, coronary heart disease, type 2 diabetes, stroke, asthma and allergic disease (Godfrey et al., 2017).

Besides its health impact, obesity also leads to economic losses for individuals and society. Obesity is estimated to cost US $2 trillion annually in healthcare costs, approximately

2.8% of the world’s GDP (Dobbs Richard et al., 2014). In a systematic review of the economic burden of obesity covering 12 countries, analyses in the U.S found that adult obesity led to an annual per-capita expenditure of $6,899 and an estimated annual productivity loss of $8.65 billion (Tremmel et al., 2017). Additionally, a study in Brazil found that obesity was associated with an estimated annual direct health care costs of $269.6 million (Tremmel et al., 2017). The economic losses suffered by obese individuals could drive them and their dependents into poverty, while from a societal perspective, the productivity losses could decrease national wealth and security generating a cycle of impoverishment and violence.

Mechanisms Underlying the Relationship between Overweight and Obesity with Chronic

Conditions

Although the mechanism of action of overweight and obesity on chronic disease outcomes are complex and continue to be investigated, several pathways have proposed and, in some cases, validated in the research literature. For instance, innate immune system driven chronic low-grade inflammation, elevated levels of free fatty acids and lipid intermediates has been described as possible links between overweight and obesity, metabolic dysfunction and disease (Castro, Macedo-de la Concha, & Pantoja-Meléndez, 2017; Heymsfield & Wadden,

2017; Lumeng & Saltiel, 2011). Increasing adiposity causes enlargement of adipose tissue, accumulation of adipose tissue macrophages, and changes macrophage inflammatory profile to a proinflammatory state (Lumeng & Saltiel, 2011). Adipocytes and macrophages within adipose tissue cause excessive secretion of proinflammatory adipokines or cytokines that leads to a state of low-grade systemic inflammation in corpulent subjects (Heymsfield & Wadden, 2017).

The multiplicity of sources for free fatty acids, including the enlarged adipose tissue mass, results in high levels of plasma free-fatty acids in overweight and obese individuals

(Heymsfield & Wadden, 2017). Additionally, greater adiposity triggers enlargement of the hepatocyte liposomes, small cytoplasmic organelles situated close to mitochondria in most cells, precipitating a range of pathological conditions including non-alcoholic fatty liver disease, steatohepatitis, cirrhosis and an accumulation of lipid intermediates (Heymsfield & Wadden,

2017). Altogether, the presence of inflammatory cytokines and lipid intermediates, and elevated levels of free fatty acids contribute to impaired insulin signaling and insulin resistance in overweight and obese individuals (Heymsfield & Wadden, 2017; Lumeng & Saltiel, 2011).

Notably, these metabolic changes are also one of several pathophysiological mechanisms underlying Type II diabetes (T2D), dyslipidemia of obesity, and obesity-related liver disease

(Heymsfield & Wadden, 2017).

Visceral or central adiposity has been implicated as a major risk factor for obesity- induced hypertension and chronic kidney disease (Hall, do Carmo, da Silva, Wang, & Hall,

2015; Seravalle & Grassi, 2017). Increased blood pressure is a response to rising sodium reabsorption that occur in obese subjects due to impaired sodium excretion (Hall et al., 2015).

Physical compression of the kidney caused by excessive fat accumulation, activation of the renin-angiotensin-aldosterone system, and elevated sympathetic nervous system activity through the overproduction of leptin have been proposed as factors driving changes in sodium regulation in overweight and obese subjects (Hall et al. 2015; Heymsfield & Wadden, 2017; Seravalle &

Grassi, 2017). Figure 2 in Hall et al. (2015), summarizes the mechanisms by which visceral adiposity precipitates hypertension and chronic kidney disease.

In their review, both Alpert et al. (2016) and Kim et al. (2016) summarize the mechanisms through which overweight and obesity contributes to cardiovascular disease.

Underpinning heart failure is elevated cardiac output in overweight and obese subjects, which may act synergistically with systemic hypertension to cause left ventricular (LV) hypertrophy and subsequent LV failure (Alpert et al., 2016; S. H. Kim et al., 2016). Though less common in obesity-related heart failure, increased cardiac output could also elicit right ventricular (RV) hypertrophy and enlargement, and consequently, right ventricular failure (Alpert et al., 2016).

Additionally, accumulation of fat around the heart is linked to coronary atherosclerosis development and progression (S. H. Kim et al., 2016). Figure 1 in Alpert et al. (2016), outlines the pathophysiology of obesity cardiomyopathy.

The relationship between overweight and obesity and psychological disorders, in particular, major depressive disorder (MDD) or depression, the single largest contributor to global disability (World Health Organization, 2017a), has been described as bidirectional

(Luppino et al., 2010; Mannan, Mamun, Doi, & Clavarino, 2016; Markowitz, Friedman, &

Arent, 2008). However, Mannan et al. (2016) showed that depression had a higher likelihood of

leading to obesity than obesity leading to depression. Markowitz et al. (2008) proposed two pathways explaining the causal effect of obesity on depression. The first one, the health concern pathway stipulates that severe obesity cause functional impairments or poor perception of personal health that could trigger a depressogenic effect. The second pathway, described as the appearance concern pathway indicates that the association between obesity and depression, especially among women of high socioeconomic class is mediated by weight-based stigma, body image dissatisfaction and dieting. To validate the appearance concern pathway, Gavin et al.

(2010) conducted a cross-sectional study with 4,543 participants from Washington and Northern

Idaho and found that body image dissatisfaction was a significant mediator of the obesity- depression association, however, the mediating effect varied by the level of educational attainment.

Obesity Paradox

The obesity paradox refers to the presence of favorable health outcomes among obese or overweight individuals relative to individuals of normal weight. The term first appeared in a study publication by Gruberg et al. (2002), who found that, contrary to their expectation and despite worse baseline clinical profile and angiographic success rates, overweight and obesity patients that underwent percutaneous coronary intervention had lower rates of periprocedural complications, cardiac-related deaths, major bleeding, and major vascular complications than normal weight patients. Subsequently, supporting studies for the so-called obesity paradox began to emerge in the research literature. For instance, Curtis et al. (2005) found that in a study of patients with stable heart failure, the risk of all-cause mortality and cardiovascular deaths were lowest among obese persons. A study by Flegal et al. (2005) revealed excess mortality was not

associated with overweight status relative to normal weight. Another study among co-morbid

T2D and cardiovascular disease patients found that individuals with grade 1 obesity had lower risk of all-cause mortality and hospitalization (Doehner et al., 2012). However, as described by

Ades & Savage (2010), these studies have several methodological gaps that preclude casual inference of the impact of overweight and obesity on chronic disease outcomes. Firstly, the aforementioned studies were vulnerable to confounding because of their retrospective nature, which limited analysis to the variables in the dataset. Secondly, the lack of prior information on the weight status of participants pre-disease hindered appropriate temporal characterization of the relationship between BMI and the studied health outcome since there was a possibility that normal weight individuals, at the time of these studies, experienced weight loss due to advanced disease.

In African countries, the obesity paradox appears to be understudied. One of the few studies that found a paradox indicated that among 91 Nigerian patients with heart failure, obese patients had higher left ventricular systolic function with the authors concluding that they had better cardiac function (Oyedeji et al., 2012). However, Tanalp (2013) commented that systolic dysfunction could not be detected by conventional ejection phase indices used by Oyedeji et al.

(2012) suggesting that methodological error may account for the surprising findings.

In recent studies with a stronger methodological design, evidence for the obesity paradox in cardiovascular disease is absent. For instance, in a cohort study with participants free of clinical disease at baseline, Khan et al. (2018) found that overweight and obese individuals had higher incidences of cardiovascular events including fatal and non-fatal myocardial infarction,

and higher incidences of cardiovascular deaths relative to normal weight individuals.

Additionally, the risk of cardiovascular disease after adjustment for confounding remained elevated in overweight and obese persons (Khan et al., 2018). Although retrospective, the study by Khan et al. (2018) pooled data across ten U.S. based cohorts with at least 10 years of follow up. Similarly, a study of 296,535 UK based participants of white European descent by

Iliodromiti et al. (2018) revealed that higher BMI, within the cutoffs for overweight and obesity, and elevated indices on other anthropometric measures (waist circumference, waist-hip ratio, and waist-height ratio) was associated with increased risk of CVD. The study by Iliodromiti et al.

(2018) was strengthened by the exclusion of individuals with CVD at baseline, the use of other anthropometric measures that are less susceptible to the impact of disease, and the adjustment of potential confounding variables such as smoking, alcohol intake and physical activity. These newer studies challenge the existence of an obesity paradox and support prevention efforts to decrease overweight or obesity in the general population.

Lowest-Risk BMI

The BMI value associated with the lowest risk all-cause mortality in a population is an important question for researchers as it aids in informing ideal BMI recommendations. Globally, this was assessed as part of a Global Burden of Disease (GBD) study and found to range between

20 -25kg/m2 (The GBD 2015 Obesity Collaborators, 2017). To conduct this analysis, the GBD authors used participant data pooled by Global BMI Mortality Collaboration et al. (2016) from

239 prospective studies conducted in Asia, Australia and New Zealand, Europe and North

America and restricted to never smokers without underlying chronic disease. Their findings is supported by an analysis of 1.46 million White adults in the National Cancer Institute

Consortium by Berrington de Gonzalez et al. (2010), who found a lowest risk BMI range of 20 -

24.9kg/m2. Similar findings of a lowest risk BMI of 20 - 24.9kg/m2 was also discovered among studies of African Americans with at least 12 years of follow-up (Cohen et al., 2014).

Additionally, Aune et al. (2016) found the nadir of the dose-response curve of the association between BMI and all-cause mortality among healthy non-smokers to be lowest within the range of 22 - 23kg/m2.

However, heterogeneous results were obtained in a national level study in Denmark and in studies of Asian populations. The Denmark study found the lowest risk BMI in a 2003 – 2013 cohort to be 27.0kg/m2 (95% CI, 26.5-27.6). Whereas, in a pooled analysis of cohort studies conducted among East Asians Chinese, Taiwanese, Singaporean, Japanese and Koreans, the BMI associated with the lowest risk of death from any cause ranged from 22.6kg/m2 to 27.5kg/m2

(Zheng et al., 2011). Among Indians and Bangladeshi subjects, surprising results were found as there was no elevated risk of all-cause mortality above a BMI of 20kg/m2. Hence, BMI values above 20kg/m2 appeared to favorably dispose towards a lower risk of mortality. These study- level variations may represent true population differences in the lowest risk BMI, may be due to chance or may emanate from study design differences.

Causal Factors for Overweight and Obesity

At the most basic level, overweight and obesity is the consequence of a positive balance between energy intake and energy expenditure. Nonetheless, available evidence indicates that overweight and obesity pathogenesis involves a variety of complex processes beyond the consumption of excess calories (Schwartz et al., 2017). Indeed, an infinite number of variables are possible

contributing factors and there exists fundamental gaps in current knowledge of the contribution and mechanism of action of identified factors at the physiological and metabolic level.

Overweight and obesity pathogenesis, however, involves gene-environment and behavioral interactions that drive/influence energy intake and expenditure. Additionally, dysfunction in energy balance due to yet unknown reasons have been noted as factors involved in pathogenesis.

In this review, it was found that majority of studies on causal factors, excluding social and economic factors, were conducted in Western countries. This may be because, until recently, overweight and obesity was not major cause of morbidity and mortality in SSA. A description of identified factors and current understanding or hypothesis of their mechanism of action ensues.

Energy Homeostasis

Weight gain and the biological defense of elevated fat mass are processes underpinning overweight and obesity development, and this is influenced by the energy homeostasis system

(Schwartz et al., 2017). The concept of energy homeostasis (the biological process that balances energy intake with energy expenditure over time) in humans is supported by studies of adaptive responses to calorie restriction. Prolonged decreases in energy intake leading to weight loss resulted in intense food cravings in normal weight young men (Keys, 1946). Additionally, in a study of both normal and obese subjects, it was observed that asynchronous calorie restriction and excess reduced and elevated energy expenditure respectively (Leibel, Rosenbaum, & Hirsch,

1995). According to Schwartz et al. (2017), these responses indicate a biological resistance to further weight loss and a tendency towards recovery of lost weight, suggesting that the physiological defense of body fat is implicated in obesity pathogenesis and weight regain after weight loss in obese individuals.

Brain-centric models of the regulation of food intake have the most compelling support

(Schwartz et al., 2017). Activation of appetite regulating neurons, located in the hypothalamus, that co-express orexigenic (stimulate hunger) and anorexigenic (suppress hunger) neuropeptides stimulate or limit feeding in response to energy balance ensuring weight stability (Schwartz et al., 2017). Therefore, dysfunctional energy homeostasis could trigger accumulation and retention of excess body fat. Nonetheless, since the hypothalamus integrates signals from the hedonic reward systems in the corticolimbic system, which is stimulated by the sight, smell and taste of food, it is possible that these reward pathways can override the homeostatic system driving the desire to consume more food despite satiation and sufficient energy stores (Greenway, 2015).

However, the exact mechanisms underlying how the reward value of food alters the energy homeostasis system leading to the preferential defense of excess body fat mass remains obscure

(Schwartz et al., 2017).

Besides the regulation of food intake, variations in basal energy expenditure may drive increased energy storage. This is exemplified by sex differences in energy homeostasis, which make women more vulnerable to body fat retention. Women have lower energy expenditure or basal metabolic rate for the same BMI due to their smaller body size and higher fat mass

(Lovejoy et al., 2009). Moreover, among women, declines in resting energy expenditure heightens with age, more so than in men (Lovejoy et al., 2009). Studies also show that women have a lower response to physical activity-related energy expenditure leading to smaller reductions in body weight and fat than in men (Lovejoy et al., 2009).

Genetic Factors

Although concordance rates among twins reared apart indicate that as much of 70% of the variance in BMI may be determined by genetic factors (Stunkard, Harris, Pedersen, &

McClearn, 1990), identified genes have weak explanatory power (Choquet & Meyre, 2011;

Schwartz et al., 2017). However, the genes discovered thus far may underlie dysfunctional energy homeostasis and feeding behaviors associated with overweight and obesity. For instance, frameshift/premature stop mutations in the leptin gene preventing normal leptin production resulted in extreme hunger, excessive food intake and severe obesity (O’Rahilly & Farooqi,

2008). Leptin, an appetite regulating hormone, is released primarily by adipose tissue to signal to the brain the status of the body fat stores, resulting in the inhibition of food intake and an increase in energy expenditure to regulate body weight (Klok, Jakobsdottir, & Drent, 2007).

Neurons that release melanocortin do so to inhibit food intake and are stimulated by leptin and insulin (Guyenet & Schwartz, 2012). Single nucleotide polymorphisms (SNPs) near the melanocortin 4 receptor (MC4R) gene is linked to hyperphagia, increased food intake and decreased satiety (Choquet & Meyre, 2011).

Genetic factors may also explain age, gender and ethnicity related differences in overweight and obesity. In a longitudinal study, age-dependent penetrance (gene expression) was observed among adult MC4R mutation carriers (37% at 20 years and 60% at > 40 years)

(Stutzmann et al., 2008). Gender specific differences in the effects of MC4R pathogenic monogenic mutations could lead to increased overweight and obesity in females than males

(Dempfle et al., 2004). Additionally, heightened overweight and obesity risk have been

associated with variants of intronic regions in SIM1 among Pima Indians but not French

Europeans (Ghoussaini et al., 2010; Traurig et al., 2009).

The influence of genetic factors on overweight and obesity risk continues to be an important area of research due to the its potential interactions with the environment and developmental factors as well as the prospect of developing gene-therapies for adiposity

(Schwartz et al., 2017).

Developmental Basis of Adult Overweight and Obesity

Results from a meta-analysis of studies that track the predictive ability of childhood obesity for adult obesity suggest that about 30% of obese children will remain obese in adulthood. Therefore, factors that promote childhood overweight and obesity may also be implicated in adult overweight and obesity. This section explores the role of development factors such as in-utero undernutrition and over-nutrition, and developmental exposure to endocrine disrupting chemicals and its impact on adult overweight and obesity.

Parental Undernutrition

Barker and colleagues proposed the best-known hypothesis that has been used to explain the connection between maternal malnutrition and risk of obesity in offspring (Barker, 2004,

2007). Known as the developmental origins of health and disease, Barker suggested that undernutrition in utero and during infancy resulted into an adaptive response, which permanently changes the structure, physiology and metabolic processes of the body leading to increased chronic disease susceptibility in adulthood (Barker, 2004, 2007). These changes occur because of

the plasticity of the embryonic and fetal stages of life as a result of high cell proliferation

(Barker, 2007). Supporting studies indicate that small-for-gestational age (SGA) infants have an increased odds of childhood overweight and obesity (Gaskins et al., 2010). SGA can result from maternal stunting, low maternal BMI during pregnancy or maternal vitamin A deficiency (Black et al., 2013). Painter et al. (2005) observed that even when infant birthweight is not impacted, maternal exposure to periods of extreme undernutrition such as famine in early gestation, resulted in higher offspring prevalence of overweight and obesity and cardiovascular risk factors in adulthood, suggesting the presence of subclinical changes due to deprivation in utero.

Evidence that the long-term effects of in utero undernutrition may differentially affect women but not men exist but have not been confirmed (Stein et al., 2007). Further, Parsons et al. (2001) delineated the relationship between birth weight and adult obesity and found that compensatory child growth among low birth weight infants was linked to increased obesity risk in adulthood.

Additionally, Parsons et al. (2001) found a linear relationship between birth weight (adjusted for gestational age) and BMI up to age 23 and a weak J-shaped relationship between birthweight and body mass index at age 33, indicating that the impact of in utero undernutrition may become apparent solely in adulthood. Based on a literature review, Parlee & MacDougald (2014) concluded that the epidemiological evidence supporting the developmental origins hypothesis indicate that: (i) maternal malnutrition during gestation is correlated with increased likelihood of developing adult obesity; (ii) the relationship between fetal birth weight and adult obesity was J- or U-shaped; (iii) post-natal compensatory growth increased the likelihood of adult obesity.

While human studies into potential mechanisms underlying maternal undernutrition, and offspring birthweight and adiposity relationship remain unavailable, animal models provide

some insights. Research on rodents indicate that intrauterine growth restriction among mothers on a low-protein diet was due to decreased transport of amino acid across the placenta, restricting the ability of rodent fetus to synthesize proteins necessary for growth and development (Jansson et al., 2006; Malandro, Beveridge, Kilberg, & Novak, 1996). Additionally, rodent pups that experience nutrient deprivation during gestation suffer from leptin resistance and dysregulation of key pathways controlling food intake, which results in hyperphagia and obesity (Desai, Gayle,

Han, & Ross, 2007; Vickers, Breier, Cutfield, Hofman, & Gluckman, 2000).

However, the evidence for the role of parental undernutrition in offspring overweight and obesity is not consensual. An umbrella review of systematic reviews and meta-analysis did not find any evidence supporting increased risk of overweight and obese among low birth weight infants (Belbasis, Savvidou, Kanu, Evangelou, & Tzoulaki, 2016). Instead the review found highly suggestive evidence of a decreased risk of overweight or obesity in adulthood among low birth weight infants (Belbasis et al., 2016). Notably, the overweight and obesity risk study among low birth weight infants by Schellong et al. (2012) included in Belbasis et al. (2016) did not differentiate between outcomes in childhood and adulthood. Additional research studies are needed to further clarify the impact of parental undernutrition on offspring adiposity at varying life stages.

Parental Over-Nutrition

Accumulating evidence suggest that maternal pre-pregnancy BMI and gestational weight gain is linked to offspring adiposity in childhood and adulthood (Sharp et al., 2015) (Gaillard,

Felix, Duijts, & Jaddoe, 2014). Large-for-gestational age, high birth weight, macrosomia and

later-life overweight/obesity are outcomes that have been associated with an elevated prepregnancy BMI; whereas, excessive gestational weight gain has been implicated in childhood overweight and obesity. In a meta-analysis by Yu et al. (2013), pre-pregnancy obesity was associated with twice the likelihood of delivering large for gestational age or high birth weight infants, and roughly three times the likelihood of fetal macrosomia. High birth weight has been found to be associated with a 46% increased risk of overweight/obesity in adulthood (Zhao,

Wang, Mu, & Sheng, 2012). Also, a meta-analysis by Tie et al. (2014) revealed that excessive gestational weight gain was associated with a 33% increased odds of childhood overweight and obesity. These findings suggest that developmental plasticity as postulated by the developmental origins of disease hypothesis may only apply to nutrient-restricted and not high calorie fetal environments. It has been hypothesized that in-utero obesogenic environments are not protective for offspring obesity because nutrient surplus conditions were rarely present in mammalian evolutionary history (Parlee & MacDougald, 2014).

Animal models exploring the relationship between maternal nutrition and offspring obesity show that fetal weight gain in obese mice fed a high calorie diet rich in fats, sugar and salts was accounted for by hyperphagia and expansion of adipose tissue mass resulting from adipocyte hypertrophy (Samuelsson et al., 2008). However, an elevated BMI is not required for a high calorie diet to affect progeny’s obesity risk as studies indicate that poor diet alone during gestation and lactation produced offspring with greater body mass index (Bayol, Farrington, &

Stickland, 2007). It has also been proposed that offspring obesity in nutrient-surplus conditions is due to placental inflammation that increases nutrient transport and fetal growth (Jansson,

Ekstrand, Björn, Wennergren, & Powell, 2002; H. N. Jones et al., 2009).

Endocrine Disrupting Chemicals

Endocrine disrupting chemicals (EDCs) have been described as exogenous compounds that can mimic or interfere with the normal action of endocrine hormones including estrogen, androgen, thyroid, hypothalamic and pituitary hormones (Elobeid & Allison, 2008; Zoeller et al.,

2012). These chemicals range from those that are found in everyday items, such as food packaging, to chemicals in cigarette, pesticides and banned substances. Examples include bisphenol A (BPA), p,p’-dichlorodiphenyltrichloroethane (DDT), and p, p’- dichlorodiphenyldichloroethylene (DDE) (Elobeid & Allison, 2008). Humans and wildlife encounter EDCs through accidental release, pollution, product leaching, volatilization

(aerosolization), as food residues after pesticide application, and migration into milk from body stores (Gore et al., 2015). Moreover, it can be dermally absorbed by application of lotions, sunscreen, soaps and other skin care products (Myers et al., 2015). EDCs are lipophilic and tend to be retained in the body and within the environment.

Interest in the link between EDCs and obesity arose from Baillie-Hamilton (2002), who posited that parallel increases in the use of EDCs and incidence of excess adiposity was suggestive of a causal relationship. With this interest emerged the “obesogen hypothesis”, aligned to the developmental origins of disease hypothesis, which stated that in utero exposure to environmental chemicals, termed obesogens, at potent levels may alter weight homeostasis and promote adipogenesis predisposing to increased overweight and obesity in later-life (Heindel,

2003; Newbold, Padilla-Banks, Snyder, & Jefferson, 2005).

Epidemiological studies have shown that prenatal EDCs exposure is related to increased adiposity risk in childhood and adulthood. Tang-Péronard et al. (2014) found that in a sample of

87 Faroese Islander overweight women, prenatal polychlorinated biphenyl levels (PCB) within the highest quartile was associated with an additional 2.07 BMI units or 5.75kg in their female progeny aged 7-year-old. In Iszatt et al. (2015), a study involving seven European cohorts, 388 nanogram/gram higher prenatal DDE concentration was associated with a 160g significant change in weight or 0.12 SD in weight-for-age z-score at 24 months of age. In support of this notion, Valvi et al. (2014) in a study of 1,285 mother-infant pairs from a Spanish birth cohort found that prenatal DDE exposure was associated with a 1.13 increased risk of rapid growth (a predictor of later obesity) at age 6 months and 1.15 increased risk of overweight. Additionally, maternal smoking during pregnancy has been tied to adiposity in childhood and adulthood

(Harris, Willett, & Michels, 2013). It has also been suggested that tributylin, an endocrine disruptor could increase adipocyte differentiation and adipose tissue amounts in adults, post- developmental exposure (Kim & Lee, 2017).

Nonetheless, among the cited studies and elsewhere in the literature, there were inconsistences in observed EDCs and adiposity relationships (Cupul-Uicab, Klebanoff, Brock, &

Longnecker, 2013; L. W. Jackson, Lynch, Kostyniak, McGuinness, & Louis, 2010; Karmaus et al., 2009; Tang-Péronard et al., 2014; Valvi et al., 2014). For instance, Valvi et al. (2014) did not observe an association between PCBs and increased adiposity, whereas this was present in subgroups analyzed by Tang-Péronard et al. (2014). Cupul-Uicab et al. (2013) found a null association between prenatal exposure to PCBs and DDE with obesity in a sample of 1,683 7- year old U.S. children. It is unclear whether these variation in study findings is due to differences

in study design factors, the inherent difficulty in measuring the body composition of EDCs, the interference of other EDCs, individual characteristics and/or complex dose-response relationships of EDCs with health outcomes, which has been described as nonmonotonic (Gore et al., 2015). Nonmonotonic dose-response relationships involve complex response patterns such as a U-shaped or inverted U-shaped curves. (Gore et al., 2015).

Moreover, it is difficult to deduce causality because it is unethical to randomize participants to the exposure and create feasible exchangeability to control for confounding.

Therefore, the strength of the existing literature needs to be assessed through an in-depth review to determine its relevance to the natural history of obesity and make recommendations for future research studies. However, it has been suggested that EDCs alter the regulation of lipids to promote adipocyte cell differentiation and lipid storage by increasing the number and size of adipocytes and interfering with the endocrine pathways involved in the control of adipose tissue development (Darbre, 2017; Spalding et al., 2008). Additionally, animal models show that the damage done by EDCs can extend beyond the generation affected in utero to their offspring

(Gore et al., 2015; Manikkam, Tracey, Guerrero-Bosagna, & Skinner, 2013).

Stress

Stress has been linked to obesity through various pathways that span cognition, behavior, physiological processes, and biochemistry (Tomiyama, 2019). Stress has been found to compromise human action that require cognition such as self-regulation, that is the ability to control one’s behavior, and self-regulatory processes such as executive functioning (Blair, 2010).

In an experiment involving a delay of gratification task, where 244 9-year old children were

offered a medium-sized plate candy immediately or a large-sized plate of candy later, it was found that those with more cumulative life stressors were less likely to delay gratification

(Evans, Fuller-Rowell, & Doan, 2012). Additionally, those same children also achieved greater

BMI increases, 4 years later (Evans et al., 2012).

Further, stress can negatively impact eating and physical activity behaviors, and sleep duration. Stress drives overconsumption, in general, especially of calorie dense foods (Adam &

Epel, 2007) and lower leisure-time physical time (Mouchacca, Abbott, & Ball, 2013). Stress is known to disrupt sleep, and it has been shown that shorter sleep duration is linked to greater risk of obesity (Åkerstedt, Kecklund, & Axelsson, 2007).

Physiologically, stress causes the activation of multiple systems including the hypothalamic-pituitary-adrenal (HPA) axis, reward processing systems, and the gut microbiome, which have implications for adiposity increases. Individual perception of stressful conditions facilitates the production of cortisol by the adrenal gland as part of a physiological cascade involving the HPA axis (Lovallo & Buchanan, 2016). Cortisol stimulates eating and fat storage, particularly in the abdomen (Epel, Lapidus, McEwen, & Brownell, 2001) (Shibli-Rahhal, Van

Beek, & Schlechte, 2006). Moreover, stress increases susceptibility to substances that activate the brain’s reward centers such as high fat, high sugar and calorie dense foods (Sinha, 2008).

With respect to the gut microbiome, existing evidence indicate that acute, chronic and prenatal stress can lead to changes in these organisms (Tomiyama, 2019). Further, it has been shown that changes in the microbiome can affect weight changes (Alcock, Maley, & Aktipis, 2014).

However, no studies directly tie stress with gut microbiome changes and increases in adiposity.

At a biochemical level, there is some evidence that suggest that leptin, the appetite suppressing hormone, and ghrelin, the appetite stimulating hormone respond to stress (Tomiyama, 2019). In a study involving exposure to a standardized laboratory stressor, individuals that produced high levels of leptin in response to the stressor, ate the least amount of high fat and high sugar foods

(Tomiyama et al., 2012). Similarly, in another laboratory stress study, it was shown that increases in ghrelin is observed after exposures to stressful conditions among emotional eaters

(Raspopow, Abizaid, Matheson, & Anisman, 2010).

Collectively, the emerging evidence suggests that stress is implicated in increases in adiposity. However, additional studies are needed to clearly delineate mechanisms of actions.

Social and Economic Conditions

Although social and economic conditions do not directly cause overweight and obesity, they certainly enable more direct overweight and obesity causal factors including changing diet composition and distribution, and physical activity patterns, hence they are called the causes of causes or social determinants of disease. In general, high income and upper-middle-income countries have higher levels of overweight and obesity, above the global average, while low- income and lower-middle-income have lower rates of overweight or obesity for both men and women (Figure 3 &4) (World Health Organization, 2017b). Ecological studies show that rising average income is positively associated with overweight and obesity rates. Using gross domestic product (GDP) per capita and gross national product per capita adjusted for purchasing power parity (GNI-PPP) respectively as a measure of socioeconomic status and data from the 2002-

2004 World Health Survey, Pampel et al. (2012) reveal that that increasing GDP per capita is

associated with a 0.4 and 0.9 BMI units increase in females and males, while Masood &

Reidpath (2017) found that per capita GNI-PPP was linked with an increased 0.5 BMI units in both sexes.

Low-income Lower-middle-income Upper-middle-income High-income Global

0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Figure 2.1. Female overweight and obesity (≥ 25kg/m2) trends by country income level. This chart was generated using data from the WHO Global Health Observatory (World Health Organization, 2017b)

However, the relationship between per capita GDP, and average BMI flattens above

$3,000 suggesting that with increasing country income, other forces become more influential

(Egger, Swinburn, & Amirul Islam, 2012). The concomitant nutritional, technological and demographic transitions to cheap processed energy-dense foods, high mechanization and an older population distribution that trail increasing national income are adduced as explanatory factors for the differences in overweight and obesity across countries (Popkin, 1998). Countries in more advanced stages of these transitions tend to have greater average BMIs (Steyn &

Mchiza, 2014).

Low-income Lower-middle-income Upper-middle-income High-income Global

0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Figure 2.2. Male overweight and obesity (≥ 25kg/m2) trends by country income level. This chart was generated using data from the WHO Global Health Observatory (World Health Organization, 2017b)

Swinburn et al. (2011) suggests that the biggest culprit in the greater global rates of overweight and obesity is changes in the food environment - unregulated production and marketing of obesogenic foods and drinks and technological changes that reduces the time cost of food - which interact with local factors to drive overconsumption and produce varying obesity rates between populations. Similarly, Offer et al. (2010) argues that elevated BMI especially in market liberalized high-income economies may be due to market freedoms that result in lower fast food prices and economic insecurity, however, confirmation of causality is hindered by the cross-sectional nature of the underlying studies and the possibility of other potential confounders. With increasing global trade liberalization propelling importation of refined fats, increased food production and processing, low and middle income countries are gradually following suit with changing food consumption patterns and rising BMI levels (Amugsi,

Dimbuene, Mberu, Muthuri, & Ezeh, 2017; Devaux & Sassi, 2013; Ford, Patel, & Narayan,

2017; Popkin, Adair, & Ng, 2012).

Studies using individual level data show that high socioeconomic status (SES) groups are more likely to be overweight and obese in lower income countries while low SES groups have a higher prevalence of overweight and obesity in higher income countries (Dinsa, Goryakin,

Fumagalli, & Suhrcke, 2012). In middle income countries, the relationship between SES and obesity exhibit mixed patterns for men but it is mostly negative for women (Dinsa et al., 2012).

These variations in the SES-obesity relationship by country stage of development is also evidenced by a recent study of the socioeconomic gradient in SSA countries, which show that the prevalence of overweight and obesity is greatest among the wealthiest and highest educated groups in most SSA countries (Neupane, Prakash, & Doku, 2016). In contrast, a study of 11

Organization for Economic Cooperation and Development (OECD) countries indicate that overwhelmingly, people of lower education, income or social class bear the larger burden of overweight and obesity, although no differences or reverse patterns were observed in men

(Devaux & Sassi, 2013).

Pampel et al. (2012) explains these differences as due partly to the monetary costs of excess foods for low and high socioeconomic groups in countries at varied stages of development. In poorer countries, food shortages put excess food out of the reach of low SES groups but within the grasp of high SES groups while in richer countries, food insecurity is characterized by inaccessibility of healthy food for low SES groups. Other studies also reach similar conclusions. Allen et al. (2017) found that in African countries and other low-income and

lower middle-income countries, high SES groups were more likely to consume processed and high fat, sugary and low-fiber foods. Whereas in the U.S. Drewnowski & Specter, (2004) and

Schwartz et al. (2017) indicate in their respective reviews that impoverished individuals overconsume high energy nutrient-poor foods because it is inexpensive. Besides the energy imbalance caused by excess food consumption among high SES groups in poorer countries and low SES groups in richer countries, physical activity patterns may differ by SES. Allen et al.

(2017) found that high-income groups in African countries and other LMICs engage in less physical activity compared to those of lesser socioeconomic status, in contrast to what is seen in higher income countries where lower SES status is associated with less physical activity

(Elhakeem, Cooper, Bann, & Hardy, 2015; Stalsberg & Pedersen, 2010). Additionally, cultural values may also play a role. In higher income countries, thinness is valued among women of higher SES and is an embodiment of social status, whereas larger body sizes are favored and symbolize high status in lower income countries (McLaren, 2007).

However, the interrelationships of SES related factors and overweight and obesity remain understudied in SSA. Aitsi-Selmi et al. (2014) only examined two SSA countries to determine the modifying role of education on the relationship between household wealth and female overweight and obesity. Further research is needed to assess role of education in diminishing the burden of overweight and obesity in SSA. Additionally, how the effect of SES factors varies with increasing information access via the internet and by age and year of birth require further study to guide the design of interventions.

Dietary Factors

While it is accepted that dietary factors are causal factors for overweight and obesity

(Swinburn et al., 2009; Swinburn, Sacks, & Ravussin, 2009), considerable debate still surrounds whether specific macronutrients, irrespective of calorie intake, are the primary culprit in the increasing global prevalence of overweight and obesity. Currently, there are two competing schools of thought on the issue. The first one - carbohydrate insulin model of obesity (CIM) - articulated by Ludwig & Ebbeling (2018) proposes that, independent of the amount of calories consumed, the primary cause of common obesity is the hypersecretion of insulin after consumption of widely available high carbohydrate diets leading to increased uptake of circulating glucose and fatty acid into adipose tissue and muscle and decreased lipolysis, which result in greater food intake and lower energy expenditure. Consequently, Ludwig & Ebbeling

(2018) advanced dietary recommendations supporting adoption of low glycemic load carbohydrate and high fat diets as a healthier alternative for combating rising BMI levels. This differs from the view held by Schwartz et al. (2017) and Hall et al. (2018) who assert the following: (i) all diets are equally fattening provided calorie over-consumption is kept constant;

(ii) although carbohydrate consumption and insulin dynamics may contribute to obesity development, obesity is a complex disease arising from a disorder of energy homeostasis with the brain as the key driver.

Indeed, in a well-referenced blog post, Stephan Guyenet, a co-author on Hall et al.,

(2018) challenged the proposals of the CIM citing contradicting human and animal studies

(Guyenet, 2018). Key among them are studies that suggest that a high-fat diet causes weight gain, more so than high carbohydrate diets (Dalby, Ross, Walker, & Morgan, 2017; Horton et al.,

1995; S. Hu et al., 2018; Lammert et al., 2000) and eating high-glycemic load diets do not lead to body weight increases (Ebbeling, Leidig, Feldman, Lovesky, & Ludwig, 2007; Lerer-Metzger et al., 1996). Based on the cited evidence, Guyenet (2018) concluded that the submissions of the

CIM had poor explanatory power for overweight and obesity. Further examination of the research literature provide support for some of the points raised by Guyenet (2018). Comparisons in weight loss studies of low carbohydrate diets versus high carbohydrate diets in the following study examples indicate no difference in weight as long as calorie consumption remains constant.

In a randomized controlled trial (RCT) of 43 obese adults, Golay et al. (1996) reported a non- significant difference in weight loss on a 15% carbohydrate diet compared to a 45% carbohydrate diet maintained for 6 weeks. Similarly, a longer RCT by Foster et al. (2003), which took place over a year among 63 obese adults, and a meta-analysis of RCTs by Hu et al. (2012) revealed no differences in body weight and waist circumference on a high vs low carbohydrate diet. These findings suggest that excess food consumption rather than macronutrient composition is implicated in overweight and obesity. However, animal studies have shown that macronutrient composition impacts plasma hormone profiles and hypothalamic gene expression, which could potentially affect overall food consumption (Kinzig, Hargrave, Hyun, & Moran, 2007; Kinzig,

Scott, Hyun, Bi, & Moran, 2005). Additionally, the low satiety fiber content of over-processed sugar-saturated carbohydrate foods could also potentially promote increased food intake

(Schwartz et al., 2017). Further studies exploring the influence of dietary composition on hypothalamic gene expression and plasma hormones in humans will shed more light on the diet- obesity relationship.

Sedentary Behavior

Sedentary behavior, which describes low energy expenditure activities requiring 1.5 or fewer metabolic equivalents (METS) such as sitting, reclining or lying, have been studied extensively in the context of television viewing time and total sitting time (Thorp, Owen,

Neuhaus, & Dunstan, 2011; Tremblay et al., 2017). One MET is defined as the rate of energy expenditure while sitting at rest, approximately an oxygen uptake of 3.5 milliliters per kilogram per minute (2018 Physical Activity Guidelines Advisory Committee, 2018). The evidence linking sedentary behavior and overweight or obesity is inconclusive. In a multivariate analysis after six years of follow-up (1992 – 1998) of 50,277 women who were normal weight at baseline, individuals who spent over 40 hours per week watching television had twice the risk of obesity compared to those who spent from zero to one hour per week (Hu, Li, Colditz, Willett, &

Manson, 2003). However, a summary of existing longitudinal studies, both RCTs and cohort studies, in the form of meta-analysis reveal conflicting results with sedentary behavior having a null association with continuous measures of body weight and BMI and a significant association with categorical measures of overweight and obesity (Campbell et al., 2018). Due to substantial heterogeneity across studies, deriving a single summary for all included studies was impossible

(Campbell et al., 2018). Hence, the existing evidence does not lead to firm conclusions on whether sedentary behavior is an independent risk factor for overweight and obesity.

Overweight and Obesity Interventions

Sustained weight loss of only 3-5% could potentially produce clinically meaningful outcomes such as decreases in the risk of developing diabetes, triglycerides and blood glucose

(Jensen et al., 2014). Further, with greater weight loss (5-10%), individuals lower their risk of

additional CVD risk factors such as high blood pressure and high low-density lipoprotein cholesterol, and low high-density lipoprotein cholesterol (Jensen et al., 2014). Additionally, the need for medication to control CVD and type 2 diabetes risk is diminished (Jensen et al., 2014).

In this section, a socioecological approach is employed to identify recommended adult overweight and obesity interventions at the individual, community and policy level.

Individual-level Interventions

At the individual level, strong evidence exist for weight loss with the adoption of on-site lifestyle interventions delivered by expert health professionals and consisting of calorie restricted diet, increased physical activity and behavior therapy (Jensen et al., 2014). Specifically, overweight or obese individuals are prescribed diets designed to elicit an energy deficit of 

500kcal/d ( 1,200 – 1,500 kcal/d total dietary calories for women; 1,500 – 1,800 kcal/d total dietary calories for men); greater aerobic activity for  150 mins/week; a structured behavior change program that enables regular self-monitoring food intake, physical activity and weight

(Jensen et al., 2014). Interventions that focus on diet or physical activity alone are not as effective as combined approaches (Chin, Kahathuduwa, & Binks, 2016). Alternatively, obese persons may undertake bariatric surgical procedures including Laproscopic adjustable gastric banding (LAGB), Laproscopic Roux-en-Y gastric bypass (RYGB), open RYGB, Biliopancreatic diversion (BPD) with and without duodenal switch and sleeve gastrectomy (Jensen et al., 2014).

Existing evidence indicate that on-site lifestyle interventions provided by a trained interventionist can induce average weight losses corresponding to 5 - 10% of initial weight at one year whereas bariatric surgery could result in a mean weight loss of 20 – 30% of initial

weight (Jensen et al., 2014). However, although both approaches are associated with weight regain in the longer term, existing evidence does not show that participants return to their pre- intervention weight on average (Jensen et al., 2014).

Community-level Interventions

The goal of community-level interventions is to harness and strengthen community assets and build the capacity needed to change overweight and obesity-related behaviors to drive weight loss (Raynor & Champagne, 2016).There is modest evidence for the implementation and effectiveness of community-level interventions for adult overweight and obesity. In a 12 month study involving lifestyle changes and a 6 month intensive weight loss program offered as group sessions, community participants recorded an average weight loss of 3.1 kg (Keyserling et al.,

2016). Another study, delivered within a local YMCA, and modeled closely with the Diabetes

Prevention Program, a 16-20 week lifestyle intervention involving classroom style meeting to build knowledge and skills for choosing healthy meals and engaging in moderate physical activity, resulted in a 6% weight loss among participants (Ackermann, Finch, Brizendine, Zhou,

& Marrero, 2008). Worksite interventions also show some promise for overweight and obesity prevention producing a BMI decrease of up 1.3kg/m2 (Morgan et al., 2011) However, no community-level interventions are recommended by academic experts or institutions.

Policy Approaches

While individual and community level interventions provide some benefit, policy approaches have a wider reach and are at the core of recommended population-level approaches for obesity prevention and control (Gortmaker et al., 2011; Sacks, Swinburn, & Lawrence,

2009). Sacks et al. (2009) delineates these policy approaches as upstream policies, which are policy actions that influence the wider socioecological environment; midstream policies, which are policy approaches that directly impact individual behavior; and downstream approaches, which describe policy action that support and enhance health services and clinical interventions.

Examples of policy approaches to obesity prevention and control include upstream approaches such as trade agreements between countries and transnational corporations, and migration policies; midstream approaches such as government funded education and campaign-based programs; and downstream approaches such as government funded programs that increase the number of dietitians and nutritionists in hospitals (Sacks et al., 2009). As part of implementing policy approaches, it has been suggested that countries exclude the provisions of the Investor-

State Dispute Settlement from trade agreements as it handicaps governments from instituting policies that are in conflict with the interests of transnational corporations (Yang, Mamudu, &

John, 2018).

Given extant knowledge showing overconsumption as the key driver of overweight and obesity (Swinburn et al., 2011), recommended governmental policy actions are mostly targeted efforts to change the food environment, food system, and diet-focused behavior change communication (Hawkes et al., 2015; Roberto et al., 2015). Examples include nutrition labelling on food packages or menu labelling in restaurants, food taxes and subsidies, and food standards for government-funded sites (Roberto et al., 2015). Several barriers to the adoption of these policies on a wider scale across countries include political pressures from industry lobby, government unwillingness to implement these policies, and lack of civil society action (Roberto et al., 2015). However, a recent systematic review of implemented food policies indicates that in

many cases there is insufficient evidence for their efficacy. For instance, studies of menu labelling reveal that they do not impact the consumer purchasing behavior (Sisnowski, Street, &

Merlin, 2017). Additionally, the introduction of subsidies for healthy foods in the US

Supplemental Nutrition Assistance Program (SNAP) in New York City had no impact on the consumption of food and vegetables (Sisnowski et al., 2017). However, the authors of systematic review indicated that their study was far from comprehensive due to a paucity of information on the evaluation of several interventions (Sisnowski et al., 2017). These findings call for instituting evaluation as a key component of policy-related interventions to inform the development of adequate evidence for obesity prevention and control.

CHAPTER 3

DATA SOURCES

Data from the Demographic and Health Survey (DHS) and existing qualitative studies will be employed for this dissertation project. This chapter describes each of these data sources and, where identifiable, their strengths, drawbacks and methods for addressing limitations.

Demographic and Health Surveys (DHS)

The DHS program collaborates with diverse countries to collect, analyze and disseminate nationally representative population health data for informing research, health policy and the allocation of national and international resources for public health interventions (United States

Agency for International Development (USAID), 2018). Funded primarily by the USAID and implemented by ICF International and other partners, since 1984 the program has supported the implementation of household and facility surveys in 90 countries across Africa, Asia, Latin

America/the Caribbean and Eastern Europe (USAID, 2018). The DHS program implements the

Demographic and Health Surveys (DHS) through interviews with women and men, collection of biomarker data including height and weight, anemia and a variety of health conditions. In addition to demographic and health data, the DHS program also collects geographic data of survey location, which can be linked to health data for spatial data analysis

The DHS is conducted either as part of standard survey or an interim survey. The standard surveys have large sample sizes, ranging from 5,000 to 60,000 households and are typically conducted every five years to allow for long-term comparisons. The interim survey, although it is nationally representative, has smaller sample sizes and is focused on collecting data

on key performance measures. This dissertation makes use of the standard survey because it contains all the variables needed for this dissertation and is available for all the countries of interest. Collected DHS data are de-identified and made available for public use via the DHS

Program website (www.dhsprogram.com).

The DHS Questionnaire

Four model questionnaires are typically used as part of the DHS standard survey to collect population health and demographic data. They include a Household Questionnaire, a

Woman’s Questionnaire, a Man’s Questionnaire, and a Biomarker Questionnaire. Many questionnaires also include standardized questionnaire modules related to the public health priorities in the implementing country. The Household Questionnaire is used to collect information on the characteristics of the household dwelling unit and its inhabitants. This information is then used to determine members of the household who are eligible for an individual interview. Eligible household members, who voluntarily consent, are interviewed with either a Woman’s or Man’s Questionnaire. DHS field workers use the biomarker questionnaire to collect biomarker data on children, women, and men in consenting households.

Since 1984, the DHS core questionnaire modules have changed with each phase of data collection. The most recent questionnaire (DHS-7) used to collect data for phase 7 of the survey encompasses the period 2013 – 2018. The core Household Questionnaire covers two topics on:

(a) the household schedule - questions on the demographic characteristics for usual residents and current visitors of the household, and (b) the household characteristics - questions on the drinking water source and treatment practices, the type of toilet facility and sanitation practices,

and household assets among others. The core Women’s Questionnaire contains questions on the following ten topics: background characteristics of the respondent, reproductive behavior and intentions, contraception, pregnancy and postnatal care, child health and nutrition, marriage and sexual activity, fertility preferences, husband’s background, gender roles, HIV/AIDS and other health issues. The core Men’s Questionnaire comprises similar but fewer topic areas than the

Women’s Questionnaire. It covers seven topic areas with questions on the man’s background characteristics, reproduction, contraception, marriage and sexual activity, fertility preferences, employment and gender roles, HIV/AIDS and other health issues. The biomarker questionnaire contains information on three topic areas, which are for assessing nutritional anthropometry in women, children and men, and anemia and HIV in women and men. Lastly, optional modules are available for questions on domestic violence, female genital cutting, maternal mortality, fistula and out-of-pocket health expenditures.

DHS Sampling Design

DHS samples are unequal probability samples typically obtained through a stratified two- stage cluster sampling process (ICF International, 2012). The DHS survey employs a multi-level stratification process, that is, the survey population is divided into stratas in several stages according to specified criteria. Subsequently, cluster sampling is first accomplished by drawing enumeration areas (EAs) from Census files. The EA samples are selected with a probability proportional to size (PPS), i.e. larger EAs have a higher probability of being selected compared to smaller EAs. In the second stage of cluster sampling, a sample of households in the list of all households in selected EAs are approached to participate in the survey. Usually, the DHS target population are women aged 15 – 49 years old, children under the age of five years old and men

within the age group of 15 – 59 years old living in ordinary residential households (ICF

International, 2012). However, in some countries, the target population may be restricted to ever- married women.

The DHS samples are designed to provide national level, residence level (urban or rural), and regionally representative estimates for the following eighteen priority indicator topic areas: total fertility and age-specific fertility rates, age at first sex, first birth, first marriage, knowledge and use of contraception, unmet need for family planning, birth spacing, antenatal care, place of delivery, assistance of skilled birth personnel during delivery, knowledge of HIV/AIDS and other STIs, high-risk sexual behavior, condom use, childhood vaccination coverage, treatment of diarrhea, fever and cough, infant and under-five mortality rates, and nutritional status. However, because the DHS prioritizes sample calculation at the design domain (sub-populations that can be identified in the sampling frame) in higher administrative or geographic units, it is difficult to guarantee precision for estimation of indicators within sub-populations in lower-level administrative units or analysis domains (sub-populations that cannot be identified in the sampling frame) (ICF International, 2012).

The DHS estimates can be affected by sampling and non-sampling errors (ICF

International, 2012). Sampling errors are related to issues of representativeness resulting from sampling a small number of eligible units from the target population rather than all eligible units.

Non-sampling errors, on the other hand result from problems that occur during data collection and processing. Sampling errors can be statistically evaluated after the survey while non- sampling errors are more difficult to evaluate at the end of a survey. The DHS dataset includes

normalized sampling weights which can be used to correct or reduce for bias introduced by non- response and other non-sampling errors (ICF International, 2012). In addition, these sampling weights keep the sample distribution close to the target population distribution and ensure valid statistical inference. However, these weights have to be denormalized to be included in pooled analysis (National Population Comission (NPC) Nigeria and ICF International, 2014)

DHS Geographic Data

Clusters or EAs that participate in the DHS surveys are georeferenced during the survey sample listing process (Burgert, Zachary, & Colston, 2013). That is, the cluster centroid coordinates are collected with Geographic Positioning Systems (GPS) receivers. The GPS readings are based on a satellite navigation system and for most clusters are accurate to less than

15 meters. Yet, the presence of obstacles such as buildings could potentially distort signals and introduce error (Burgert, Zachary, & Colston, 2013). In addition, user mistakes in the identification of cluster coordinates can also be a source of error. If a cluster does not have GPS readings, cluster coordinates can be extracted from preexisting census data provided by a national census agency.

Strengths of the DHS

 Nationally representative dataset

 Publicly available and easy to access

 Samples are carefully selected using a valid scientific process

 Contains data for several health-related variables

 Recognized and respected by policy-makers and academia as a major source of research

data

 Numerous sources of support and references for user

Description and Critique of the Dissertation Dataset from the DHS

A selection of nationally representative Demographic and Health Surveys conducted before the year 2000 from 34 countries were identified as one of the data sources for this dissertation. Of the total number of countries, 30 had geographic data (Table 1). An important limitation of the DHS geographic data is that latitude and longitude positions for all surveys are displaced to protect the confidentiality of the participants. Urban clusters are displaced randomly by 0 - 2 km while rural clusters are displaced by 0 - 5km with an additional 1% of rural clusters between 0 - 10km. This displacement is restricted such that the cluster points stay within the country and DHS survey region boundaries.

However, prior to 2009, the displacement error was not restricted to the boundaries of second-level administrative units (The Demographic and Health Surveys (DHS) Program, n.d.).

This implies that specific aim one, which focuses on estimation of female overweight and obesity prevalence within lower administrative units with spatial methods will exclude countries with geographic data before 2009. Two additional criteria for exclusion will be the level of completeness of the dataset and the availability of shared administrative boundaries. Sparse datasets cannot be used to fulfill the objective of the specific aim one because the statistical modelling involves borrowing information from areas with more information to adjust areas with less information to create smoothed prevalence maps. Hence, datasets with too sparse data will

contain insufficient information for the smoothing process. Additionally, the availability of shared boundaries is one of the requirements of the small prevalence estimation methods.

Therefore, after excluding the geographic datasets collected before 2009, datasets with large missingness (defined as 10% or more) and the dataset for Comoros due to the lack of shared boundaries between administrative units, seven country datasets (Benin, Ethiopia, Mozambique,

Nigeria, Tanzania, Zambia and Zimbabwe) in Table 1 were eligible for small area estimation.

Specific aim two - Analyze the effect modification of education on the relationship between household wealth and female overweight and obesity in SSA countries – will be accomplished by analyzing all DHS datasets from 34 countries. The issue with these datasets, however, is that many contain missing variables central to the study analysis. BMI, a key variable used to construct the outcome categories, had large missingness. Twenty-four countries have more than 10% BMI observations while 22 of these countries have 50% or more missing

BMI values (Table 3.1). This raises the question of how to deal with missing data in these datasets.

Table 1

Listing of Available DHS Data S/N Country Year Sample Size % missing BMI data Geographic Data (Yes/No) 1 Benin 2011-12 16,599 2.90 Yes 2 Burkina Faso 2010 17,087 50.52 Yes 3 Burundi 2010 9,389 51.22 Yes 4 Cameroon 2011 15,426 48.89 Yes 5 Chad 2014-15 17,719 36.22 Yes 6 Comoros 2012 5,329 2.96 Yes 7 Congo 2011-12 10,819 47.83 No 8 Congo DRC 2013-14 18,827 50.27 Yes 9 Cote d’ Ivoire 2011-12 10,060 52.15 Yes

Listing of Available DHS Data S/N Country Year Sample Size % missing BMI data Geographic Data (Yes/No) 10 Ethiopia 2016 15,683 5.28 Yes 11 Gabon 2012 8,422 34.23 Yes 12 Gambia 2013 10,233 55.58 No 13 Ghana 2014 9,396 49.45 Yes 14 Guinea 2012 9,142 48.41 Yes 15 Kenya 2014 31,079 53.60 Yes 16 Lesotho 2014 6,621 48.83 Yes 17 Liberia 2013 4,644 50.27 Yes 18 Madagascar 2008-09 17,375 51.80 Yes 19 Malawi 2015-16 24,562 67.23 Yes 20 Mali 2012-13 10,424 49.63 Yes 21 Mozambique 2011 13,745 1.06 Yes 22 Namibia 2013 10,018 48.54 Yes 23 Niger 2012 11,160 54.14 No 24 Nigeria 2013 38,948 1.58 Yes 25 Rwanda 2014-15 13,497 50.37 Yes 26 Sao Tome & 2008-09 2,615 8.76 No Principe 27 Senegal 2010-11 15,688 63.38 Yes 28 Sierra Leone 2013 16,658 52.08 Yes 29 Swaziland 2006-07 4,987 2.63 Yes* 30 Tanzania 2015-16 13,266 0.81 Yes 31 Togo 2013-14 9,480 49.08 Yes 32 Uganda 2016 8,674 68.83 Yes 33 Zambia 2013-14 16,411 1.04 Yes 34 Zimbabwe 2015 9,955 3.06 Yes *This geographic dataset will be excluded from estimation of female overweight and obesity at lower administrative levels because they were collected prior to 2009 implying that their displacement was not restricted to the second-level administrative unit

Handling Missing Data

The process for handling missing data depends on the missing data mechanism. In the literature, missing data mechanisms is termed: missing completely at random (MCAR), missing at random (MAR) and missing not at random (MNAR) (Rubin, 1976). When the missing data mechanism is MCAR, the missing data is a random sample of all cases, i.e. the observed data

pattern is sufficient for answering any research question related to the data since it is also a random sample of all cases (Sterne et al., 2009). With MAR, the missingness is dependent on the observed data but not on the unobserved data while with MNAR, the missing data does not depend on the observed data but on the unobserved data (Sterne et al., 2009). This means that for

MAR cases, the missing data pattern is predictable from the variables in the data but in the

MNAR, the missingness is not predictable or related to variables in the dataset (Bennett, 2001).

Methods to handle missing data typically focus on MAR and MNAR mechanisms because

MCAR patterns yield unbiased parameter estimates although with slight loss of statistical power

(Graham, 2009). However, MAR and MNAR will yield biased parameter estimates unless it is addressed with statistical methods (Graham, 2009).

The missing data in each dataset selected for this dissertation occurred because BMI was measured in only a sub-sample of the women that completed the survey. If this sub-sample was a random sample of the full dataset, then it is possible that the BMI values will remain representative and the missing data mechanism will be MCAR. While it is impossible to fully detect MCAR or even know the missing data pattern since we only have the observed data, it has been suggested that an MCAR assumption is not appropriate if varying missing data patterns show differences in the distribution of values across groups (UCLA Institute for Digital Research and Education, n.d.).

Bennett (2001) also indicates that statistical analysis will be biased when greater than 10% of the data is missing. Appendix 1 showcases varying distribution of values across groups in each missing data pattern category for 24 countries with greater than 10% missing data. In the examination of missing values, pregnant women. Adolescents (15 – 19 years) were also excluded because they do

not form part of the study population. According to Coffey (2015), pregnant women in SSA are estimated to gain 7kg (15.4lbs) during gestation.

The large number of missing values combined with differing distribution of mean values in the missing data group warrants methods for addressing missing data. However, because MNAR patterns are intractable, and MAR methods work even when the MAR assumption is violated, an assumption of MAR is appropriate for the missing data in this dissertation dataset (Graham, 2009).

Modern methods for dealing with missing data under the MAR assumption are likelihood- based methods such as Expectation Maximization (EM) Algorithms and Full Information

Maximum likelihood (FIML) methods; and Multiple Imputation (Bennett, 2001; Graham, 2009).

Nonetheless, multiple imputation is often the preferred method because it is less computationally intensive compared to other methods and it performs well with non-normal data. Multiple imputation has been described as a “principled data method that provides valid statistical inferences under the MAR condition” (Dong & Peng, 2013). Multiple imputation works by imputing m plausible values for the missing data to create M imputed datasets (Bennett, 2001).

The goal with creating several imputed datasets is to acknowledge the uncertainty associated with imputing missing values (Dong & Peng, 2013). The m imputed datasets are analyzed individually, and the parameter estimates are pooled together to yield a single parameter estimate and corresponding standard error. The pooled estimates reflect the variation from the imputation process with less bias than a single imputation procedure (Dong & Peng, 2013). Due to the ease of implementation, missing values in this dataset from the DHS will be handled with multiple imputation procedures.

Sources of Qualitative Research Studies for Research Synthesis in this Dissertation

To ensure access to comprehensive literature for the qualitative systematic review of studies on contextual factors associated with female overweight and obesity in SSA (specific aim

3 of the dissertation), the database search process will include the major databases – PubMed and

CINAHL – in addition to Web of Science, PROQUEST, EMBASE, PsychInfo. The choice of data sources for qualitative synthesis in this dissertation is based on a systematic review of methodological studies related to the search process for qualitative research synthesis. In his review, Booth (2016) found that number of databases that authors searched ranged from 3 to 20.

Yet, most authors reported finding the most relevant hits from MEDLINE and CINAHL. Despite the strengths of these well-known databases, Booth (2016) also advised that researchers include specialist databases containing dissertation abstracts and journals not indexed in major databases to reduce the likelihood of publication bias.

CHAPTER 4

ADULT FEMALE OVERWEIGHT AND OBESITY PREVALENCE IN SEVEN SUB-

SAHARAN AFRICAN COUNTRIES: A BASELINE SUB-NATIONAL ASSESSMENT OF

INDICATOR 14 OF THE GLOBAL NCD MONITORING FRAMEWORK

Ifeoma D. Ozodiegwu, DrPH1, Laina D. Mercer, PhD2, Megan Quinn, DrPH1, Henry V. Doctor,

PhD3, Hadii M. Mamudu, PhD4

1Department of Biostatistics and Epidemiology, East Tennessee State University, Johnson City,

Tennessee, United States of America

2Center for Vaccine Innovation and Access, Program for Appropriate Technology in Health

(PATH), Seattle, Washington, United States of America

3Department of Information, Evidence and Research, World Health Organization, Regional

Office for the Eastern Mediterranean, Cairo, Egypt

4Department of Health Services Management and Policy, East Tennessee State University

Johnson City, Tennessee, United States of America

Corresponding author: Ifeoma D. Ozodiegwu

Email: [email protected]

Phone: 4237731809

Summary

Background

Decreasing overweight and obesity prevalence requires precise data at the sub-national level to monitor progress and initiate interventions. This study aimed to estimate baseline age- standardized overweight prevalence at the lowest administrative units among women, 18 years and older, in seven African countries. The study aims are synonymous with indicator 14 of the global non-communicable disease monitoring framework.

Methods

We used the most recent Demographic and Health Survey and administrative boundaries data from the GADM. Three Bayesian hierarchical models were fitted and model selection tests implemented. The age-standardized prevalence of overweight among adult women at national, first and second administrative levels were individually reported in each country in the form of maps and tables.

Findings

Substantial variation in the age-standardized prevalence of adult female overweight was noted across several second-level administrative units. In numerous locations in Tanzania, Nigeria and

Zimbabwe, more than half of the adult female population were overweight and in one location in

Tanzania, over 72% of the adult female population were overweight. These estimates were roughly twice the national level overweight prevalence and in some cases roughly 10 – 20% greater than the overweight prevalence in first-level administrative units.

Interpretation

The observed overweight burden in subnational administrative units suggests the presence of an epidemic tantamount to the situation in more affluent economies. African countries lack the resources to effectively handle the fallout from such epidemic, therefore motivating the need for increased urgency in adopting WHO obesity-related intervention guidelines and implementing more rigorous studies to validate the study findings.

Funding

The first author’s doctoral program was partly funded by a Rotary International District 7570

Skelton Jones Scholarship

Research in Context

Evidence before this study

PubMed was searched from inception to January 24, 2018 without language restriction and using terms related to overweight, obesity, and Africa. This original search yielded 683 articles, with only four studies that provided comprehensive overweight and obesity prevalence at a sub-national level in sub-Saharan African women. These were individual studies conducted in Nigeria, Uganda,

Ghana and Mali. With the exception of Nigeria, this analysis does not include these countries due to a paucity of data. However, the first-level administrative levels for which overweight and obesity prevalence were estimated in Nigeria are still large areas, that is states with populations sizes ranging from over 500,000 to 7 million residents, which may be insufficiently informative for community and policy stakeholders at lower levels.

Added value of this study

This study complements the existing literature by generating estimates of adult female overweight and obesity prevalence at second-level administrative levels within seven African countries

(Benin, Ethiopia, Mozambique, Nigeria, Tanzania, Zambia and Zimbabwe). Its findings are the first granular analysis of overweight in the region and will enhance precision in the surveillance of female overweight and obesity by global, national, and local stakeholders in Sub-Saharan

Africa, providing baseline data at lower administrative levels to track progress towards meeting global obesity targets, which will enable efficient allocation of resources for further research and optimal intervention planning and implementation.

Implications of all the available evidence

The high levels of adult female overweight observed in several second-level administrative units and the moderate burden in first-level administrative units within the studied countries underscores the need for increased attention to these issues. A critical step would involve adoption and implementation of the WHO diet and physical activity recommendations, including the utilization of taxation measures to fund overweight and obesity prevention efforts in African countries.

Additionally, leveraging ongoing undernutrition programs and research to intervene and provide expanded understanding of underlying issues will be crucial to meeting global obesity and NCD targets. The findings of this study offer stakeholders the opportunity to accomplish research and intervention efforts more cost-effectively.

Introduction

The probability of premature death (i.e. between the ages of 30 years to 70 years) from major non- communicable diseases (NCDs) - cardiovascular disease, cancers, chronic respiratory diseases and diabetes – is projected to increase in Africa and decrease elsewhere.1 This implies that meeting the

Sustainable Development Goal (SDG) 3.4 target of a one-third decline in premature mortality due to NCDs by 2030 in Africa 2 and the World Health Organization (WHO) Global Action Plan target of a 25% relative reduction in premature mortality from cardiovascular diseases, cancer, diabetes and chronic respiratory diseases by 20253 will be impossible without precise action to address related risk factors.

To reverse the present trajectory of NCD-related mortality in Africa, voluntary global targets for obesity and other NCD risk factors must be met.1 The global obesity target agreed upon by policymakers in 2013 was to halt the rise in its prevalence within their countries by 2025, relative to the baseline in 2010.3 Progress towards this target will be monitored by assessment of the age- standardized prevalence of overweight and obesity in individuals aged 18 years and older, amongst other indicators.3 Reaching this target is especially important because obesity is the second largest contributor to non-communicable disease mortality among African women.1 Nonetheless, the lack of granular or micro geographic prevalence data to enhance precise planning and action by local stakeholders, especially in the context of the health system stress brought on by the double burden of malnutrition facing many African countries, could potentially interfere with efforts to achieve

NCD-related and other health system goals, warranting this study.

The extant literature shows that adult women are disproportionately affected by overweight and obesity in sub-Saharan Africa (SSA). Age-standardized estimates from the WHO indicate that in

2016, roughly 23% and 39% of adult African men and women (18+ years), respectively were overweight while 6% and 15% of adult African men and women were obese.4 Besides the obvious risk of major NCDs posed by the increased prevalence in high body mass index (BMI) in the region, overweight and obese SSA women are, among other things, likely to face an increased risk of pregnancy-related complications, including cesarean delivery, preeclampsia, gestational diabetes, spontaneous abortion, and stillbirth.5

Although recent and well-designed subnational analysis of overweight and obesity prevalence with various methodologies in SSA women provide some insight into the heterogeneity present at varying administrative levels, the majority of subnational studies are limited in the administrative regions covered and the methods applied.6–9 For instance a study conducted in Uganda used a local

Getis-Ord Gi* statistic to provide combined estimates of overweight and obesity prevalence at the district level, however, this method does not account for the large sampling variance introduced by small sample sizes and the complex survey sampling design of the country’s Demographic and

Health Survey.8 Further, another study provide combined state-level estimates for overweight and obesity in Nigeria, which are useful at the state-level, but are not informative for efficient resource allocation and intervention planning at the lowest administrative levels.7

Hierarchical Bayesian (HB) models, the method applied in this paper, involves borrowing strength across geographical space in small areas, that is areas with insufficient data for indicator estimation.10 Small sample sizes in HB model estimation are addressed by explicit accounting of

large sampling variance and harnessing of spatial and non-spatial relationships.11 The hierarchical smoothing and the model acknowledgement of the design weights included with complex survey samples have been shown in simulation studies to decrease the variance associated with the estimates and bias from non-response and non-random sampling respectively.10 Notably, HB models have been used previously to estimate region-specific child mortality rates in Tanzania.12

Therefore, this study aimed to provide baseline estimates of age-standardized prevalence of overweight (≥ 25kg/m2) among women, 18 years and older, at second-level administrative units in seven SSA nations.

Methods

Data Sources

The DHS are household-based sample surveys of women aged 15 – 49 years that are collected in over 90 countries and designed to provide representative indicator estimates at national and sub- national levels.13 For sampling purposes, countries were divided into survey regions, which are sub-national units that correspond to existing administrative units such as states or provinces. From each region, enumeration units or clusters were randomly sampled to get a representative estimate for that region. Within each cluster, a group of households were randomly selected for inclusion in the survey. Within the selected survey households, 20 to 30 women, depending on whether it was rural or urban cluster, were selected for inclusion in the survey and information such as household sociodemographic status, reproductive history and anthropometrics are collected.13

Anthropometric data were obtained by trained personnel who measure and record participants’

weight using the SECA 874 digital scale to the nearest 0.01kg and height with the Shorr height board.14 Once collected, these measurements were recorded in the biomarker questionnaire.14

The DHS geographic data consists of the latitude and longitude positions collected with a positional accuracy of 15 meters or less from the center of populated areas within surveyed clusters.15 To ensure the confidentiality of participants, random displacement of urban clusters occurred up to two kilometers while rural clusters were displaced up to five kilometers, with 1% of rural clusters displaced up to 10 kilometers.15 However, the displacement is restricted to stay within the country and the DHS survey region.15 In the seven countries examined in this research study, we only included data containing explicit information on the GPS displacement. This information was not available prior to 2016 for Ethiopia, 2015 for Tanzania, and 2015 for

Zimbabwe. With the exception of Nigeria, displacement of survey clusters in all the included countries were restricted to the relevant second-level administrative divisions at the time of the data compilation by the DHS program.

Administrative boundaries data for mapping the surveyed clusters at the level of estimation was acquired from GADM, the database of global administrative areas (https://gadm.org/data.html).

Data Compilation

For the purposes of this study, participants of interest were overweight or obese women, 18 years or older, in line with the global NCD indicator 14 for monitoring the voluntary global overweight and obesity-related target.3 Pregnant women were excluded due to pregnancy-related weight changes over a short time period.16 Adolescents younger than 18 years were also excluded because

they are not the participants of interest. Overweight and obesity, henceforth called overweight,

was defined as a BMI of ≥25kg/m2.17

The number and names of administrative units considered in this study were based on the GADM

and varied by country as seen in Table 1 below.

Table 1: Countries and Administrative levels No Country Year of Number and Name of Number and Name of Second- Survey First-level level Administrative Units Administrative Units 1. Benin 2011 – 2012 12 Department 76 Communes 2. Ethiopia 2016 9 Regional States and 2 79 Zones Chartered Cities 3. Mozambique 2011 10 Provinces 129 Districts 4. Nigeria 2013 36 States and the Federal 774 Local Government Areas Capital Territory (LGA) 5. Tanzania 2015 – 2016 30 Regions 169 Districts 6. Zambia 2013 – 2014 10 Provinces 72 Districts 7. Zimbabwe 2015 10 Provinces 60 Districts

DHS cluster locations without longitude and latitude positions were excluded from the analysis.

Using their longitude and latitude positions, the DHS surveyed clusters for individual countries

were mapped onto the administrative units to obtain the associated first and second-level

administrative units. The resulting country datasets were joined to the DHS individual data using

the DHS cluster variables present in both datasets as the linking variable. This final dataset was

used in the statistical analysis.

Statistical Analysis

Analysis was completed on an individual country basis by fitting three separate Bayesian

hierarchical models with between area variation modelled as having a spatial structured with a

Besag-York- Mollie (BYM) model, a BYM2 model (a scaled version of the BYM model) or as spatially unstructured with the IID model 18 using the Integrated Nested Laplace Approximation

(INLA) package in R.22 Similar to previous work,19 in the first stage, the estimated prevalence of overweight, y, in each second-level administrative unit, i, were calculated as the ratio of total count (푇̂푖) of overweight individuals and the total population (푁̂푖) accounting for the design weights in both the estimator and variance as specified by the Horvitz-Thompson (HT) formula.

Subsequently, design weighted estimates were age-standardized using the WHO World Standard

Population to ensure comparability with prevalence estimates generated by the WHO.20

푦 푇̂푖 푖푘 푃̂푖 = ⁄ , 푇̂푖 = ∑ 푁푖 푘∈푠푖 휋푖푘

Further, to develop the hierarchical smoothing model, the estimated HT area level proportions were defined as the empirical logistic transform to constrain the probability to lie in (0, 1). Thus, the likelihood was taken as the asymptotic distribution

푣푎푟 (푃̂ ) 퐿 푃푖 푖 푦푖 |푃푖 ~ 푁 (log [ ⁄ ] ) 1 − 푃푖 ̂2 2 푃푖 (1 − 푃̂푖)

In second stage, the HB models were fitted to the logit transformed area-level data to allow partial pooling of data between areas with sparse and large samples. The independent random effects models (IID model) which included spatially unstructured random effects was of the form

푃푖 2 2 log [ ⁄ ] = 훽0 + ∈푖, ∈푖 |휎푣 ~푖.푖.푑. 푁(0, 휎푣 ) 1 − 푃푖

Whereas, the spatial random effects model (BYM model or BYM2 model) was of the form

1 휎2 푃푖 2 푢 log [ ⁄1 − 푃 ] = 훽0 + ∈푖+ 푈푖, 푈푖|푈푗, 𝑖 ≠ j, 휎푢 ~푁( ∑ 푈푗 , ) 푖 푛푖 푛푖 푖~푗

where 훽0 is the overall or national level mean log odds of overweight in the absence of random effects, ∈푖, the spatially unstructured random effects are normal and independently distributed with

2 mean, 0, and an unknown variance, 휎푣 . The default flat INLA hyperprior, Gamma (1, 5e-5) was

1 placed on the precision, 2 , of ∈푖 in the IID model, and hyperprior, Gamma (1, 5e-4) on the 휎푣

1 precision, 2 , of ∈푖 in the BYM model. On the other hand, 푈푖, the spatially structured random 휎푣

2 effects are conditional on their neighbors, 푈푗 and their unknown variance, 휎푢 . It is based on an intrinsic conditional autoregressive model (ICAR) and follows a normal distribution with a mean

2 1 휎푢 defined as the mean of its neighbors, ∑푖~푗 푈푗, and unknown variance scaled by its neighbors, . 푛푖 푛푖

1 The default flat INLA hyperprior, Gamma (1, 5e-4) was placed on the precision, 2 , of 푈푖. The 휎푢

2 variance of the spatially unstructured random effects, 휎푣 , captures the between area-variability in

2 the residual log-odds, and 휎푈, the variance of the between-area spatial random effects captures spatial dependence.21

While BYM2 maintains both the spatially unstructured and spatially structured random effects, they are both standardized such that their variance is equal to one and hyperpriors placed on the precision, 휏푏 and the mixing parameter, ∅, which quantifies the degree of mixing between both random effects. Therefore, instead of the sum of the random effects, 푏, being equal to its structured component, 푈푖 , and unstructured components, ∈푖, it is reparametrized as

1 푏 = (√1 − ∅ ∈푖+ √∅ 푈푖) √휏푏

Penalized complexity (PC) precision hyper priors, pc.prec (1, 0.01) was placed on the precision,

휏푏 , and PC hyperpriors, phi (0.5, 0.5) on the mixing parameter, ∅.

The INLA provides the median, 2.5th and 97.5th percentile of the posterior distribution for the logit of the female overweight prevalence estimates. The inverse logit transformation was used to generate female overweight prevalence estimates and 95% confidence intervals (CIs) for individual second-level administrative units in each of the seven countries. Model comparison and selection was undertaken with the deviance information criteria (DIC) and the Watanabe-Akaike

Information Criteria (WAIC).23,24

To assess the impact of the DHS random displacement on the locational accuracy of the DHS cluster locations for Nigeria, the maximum probability covariate (MPC) selection method was used to determine the most probable second-level administrative unit by placing a buffer, equal to the displacement radius, around each DHS cluster location and misclassification probabilities were estimated through a simulation based process, which employs a Monte Carlo integration.25

Finally, age-standardized overweight estimates for individual countries were mapped on to their respective second-level administrative units. To enable visualization of the heterogeneity of overweight estimates at the second-level administrative, age-standardized estimates of overweight prevalence estimates at first-level administrative units were generated using the HT formula and mapped and presented side-by-side with the second-level administrative unit estimates. Since the

DHS in the studied countries were representative at the first-administrative levels, the HT formula provided representative estimates.

Funding

The first author’s doctoral program was partly funded by a Rotary International District 7570

Skelton Jones Scholarship. However, the funders had no involvement in the study design, data analysis and interpretation, and manuscript preparation associated with this research study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results

Participants and Locations

The number of participants and locations surveyed by the DHS program differed by country

(Table 2). Nigeria, followed by Benin, had the largest sample size and numbers of DHS survey/cluster locations. After excluding adolescents and pregnant women, missing BMI data in each individual country dataset remained roughly at 5% or less. Due to the small proportion of missing data, its impact was deemed negligible and it was deleted from the analysis sample.

With the exception of Tanzania and Zimbabwe, a small number of DHS clusters did not have longitude and latitude positions (Table 2). These clusters with missing longitude and latitude values did not form part of the analysis. The order of magnitude of sample sizes and cluster locations by country remained the same after exclusions.

Table 2: Data Input Sample Sizes, Exclusions & National Level Overweight Prevalence for Women, 18 Years and Older DHS cluster data Sample Size Individual Level (excluded due DHS Data Individual DHS Data missing National Level Original/Analysis Exclusion Reasons longitude/latitude Overweight % Country Sample Size (Number) data) Prevalence (95% CI) Benin 16,599/12,811 Adolescents (1,869) 746 (4) 28.3 Pregnant women (1,547) (22.9 – 33.7) Missing BMI data (372)

Ethiopia 15,683/11,862 Adolescents (2,094) 622 (21) 8.2 Pregnant women (1,092) (3.6 – 12.8) Missing BMI data (635)

Mozambique 13,745/10,393 Adolescents (1,970) 609 (1) 17.8 Pregnant women (1,265) (10.9 – 24.8) Missing BMI data (117)

Nigeria 38,948/29,304 Adolescents (4,946) 889 (7) 26.9 Pregnant women (4,256) (23.1 – 30.6) Missing BMI data (442)

Tanzania 13,266/10,369 Adolescents (1,749) 608 (0) 31.1 Pregnant women (1,062) (24.6 – 35.5) Missing BMI data (86)

Zambia 16,411/12,809 Adolescents (2,168) 721 (2) 25.2 Pregnant women (1,297) (18.0 – 32.5) Missing BMI data (137)

100

Table 2: Data Input Sample Sizes, Exclusions & National Level Overweight Prevalence for Women, 18 Years and Older DHS cluster data Sample Size Individual Level (excluded due DHS Data Individual DHS Data missing National Level Original/Analysis Exclusion Reasons longitude/latitude Overweight % Country Sample Size (Number) data) Prevalence (95% CI) Zimbabwe 9,955/7,787 Adolescents (1,348) 400 (0) 38.3 Pregnant women (569) (33.5– 44.2) Missing BMI data (251)

101

Findings

Age-standardized overweight prevalence in the study population varied across countries and within countries at the first and second-level administrative levels. Overweight prevalence at the national level was greatest in Zimbabwe at 38.3% (95% CI: 33.5 – 44.2), followed by Tanzania at 31.1% (24.6 – 35.5) and lowest in Ethiopia at 8.2% (3.6 – 12.8) (Table 2). Overweight prevalence at first-level administrative units ranged from 9.9 – 42.0% in Benin, 3.6 – 33.3% in

Ethiopia, 6.2 – 43.9% in Mozambique, 10.6 – 46.7% in Nigeria, 14.2 – 49.9% in Tanzania, 11.8

– 38.5% in Zambia, and from 30.8% - 50.6% in Zimbabwe (Figures 1a-c & Supplementary

Appendix). One first-level administrative unit in Benin, four in Ethiopia, three in Mozambique, seven in Nigeria, four in Tanzania, one in Zambia and two in Zimbabwe, had ten percent greater overweight prevalence than their respective national averages.

Estimates of overweight prevalence at second-level administrative units revealed the existence of further heterogeneity at lower scales. Overweight prevalence ranged from 7.5 – 42% in Benin,

1.4 – 35.9% in Ethiopia, 1.6 – 44.7% in Mozambique, 1.0 – 67.9% in Nigeria, 2.2 - 72.4% in

Tanzania, 3.9 – 39.9% in Zambia, and 4.5 - 50.6% in Zimbabwe. For instance, in Alibori, a

Department (first-level administrative unit) in Benin with an overweight prevalence of 20.4%

(13.8 – 27.0%), overweight prevalence within Kandi, one of its Communes (second-level administrative unit) was as high as 32.8% (28.5 – 37.4%) and in another, Banikoara, overweight prevalence was as low as 11.1% (8.3 – 14.6%). Equivalently, while Ethiopia had the lowest prevalence of overweight at the national level, the Southern Nations, Nationalities and Peoples, a

Regional State (first-level administrative unit), which had some of the lowest proportions of overweight women in a Zone (second-level administrative unit) also contained the Zone, Burji,

102

with the highest overweight prevalence at 35.9% (35.5 – 36.2%). In Nigeria, Adamawa, a state

(first-level administrative unit) with an overweight prevalence of 24.7% (13.8 – 35.6%), had an

LGA (second administrative level) - Jada – with overweight prevalence at 40.1% (39.5 – 40.8%) and another – Demsa- with a prevalence of 7.6% (7.3 – 8.0%)

Moreover, several second-level administrative units had overweight prevalence that were over twice the prevalence the national overweight prevalence level in their respective nations. This represented two districts in Tanzania, 14 LGAs in Nigeria; six districts in Mozambique, and five zones in Ethiopia. A detailed list of age-standardized overweight prevalence and 95% CIs among women, 18 years and older, at second-level administrative units in the study countries are provided in the supplementary appendix.

103

104

105

106

Table 3 presents the misclassification probabilities for the 889 DHS clusters included in the estimation of overweight prevalence in Nigeria’s second-level administrative units or LGAs. Our results showed that approximately 1.5% of the cluster points had more than a 50% probability of being misclassified. This implies that these clusters were more likely to be located elsewhere than the current location used for the estimation. However, the impact of such misclassification on the accuracy of second-level administrative unit estimates were reasoned as potentially low given the small number of points.

Table 3: Misclassification Probabilities for Assessing the Locational Accuracy of the DHS cluster locations for Nigeria (DHS, 2013) Misclassification Probability Number of Cluster Points Percent 0 369 41.5 0.00001 - 0.1 286 32.2 0.10001 - 0.2 79 8.9 0.200001 - 0.3 63 7.1 0.300001 - 0.4 45 5.1 0.400001 - 0.5 34 3.8 >0.5 13 1.5 Total 889

Figures 1 – 7 in the supplementary appendix as well as the corresponding tables present the relative uncertainty associated with each second-level administrative unit overweight prevalence estimates. Relative uncertainty was calculated as the ratio of the difference between the upper and lower 95% CI and the mean. Cutoff points for relative uncertainty were 20% and 50%. The proportion of second-level administrative units with high relative uncertainty, that is, above

50%, were 18.4% in Benin, 63.3% in Ethiopia, 33.3% in Mozambique, 42.3% in Nigeria, 23.7% in Tanzania, 44.4% in Zambia, and 10% in Zimbabwe. The prevalence of overweight in second-

107

level administrative units with high relative uncertainty varied widely in individual countries and were upward of 25.2% in Benin, 19.1% in Ethiopia, 30.8% in Mozambique, 50% in Nigeria,

46.7% in Tanzania, 30.4% in Zambia, and 37% in Zimbabwe.

Discussion

To the best of our knowledge, this is the first study to provide evidence of highly varying disparities in the age-standardized prevalence of adult female overweight and obesity across second-level administrative units in seven African countries and a baseline against which progress towards halting the rise in the prevalence of adult female overweight can be measured at the local level. In Tanzania, we noted a location in which 72% of the adult female population were overweight or obese, and several locations in Tanzania, Nigeria and Zimbabwe where more than half of the adult female population were overweight or obese, reminiscent of the situation observed in more affluent countries, which tends to have greater burdens of overweight and obesity, especially among women,26,27 and underscoring the need to prioritize overweight and obesity-related interventions in these settings.

However, although the micro-data offered by this study empowers local actors to engage in the prevention of overweight and obesity in their communities, the complexity of the problem requires a multisectoral approach involving stakeholders at various higher administrative levels and development agencies operating nutrition programs in these countries. WHO recommends several diet and physical activity policy and program options ranging from policy measures to reduce trans-fat, saturated fat and excess calorie intake to engagement with food service providers to improve the availability, affordability, acceptability of healthier food, and promotion

108

of healthy food in public institutions.3 Additionally, a recent publication identified four potential entry points and provided specific recommendations for leveraging existing undernutrition programs to accommodate overweight and obesity prevention.28 While the recently concluded

Third United Nations High Meeting on NCDs is expected to renew momentum towards uptake of overweight and obesity interventions, the 2017 Non-communicable Diseases Progress

Monitor indicates that none of the countries included in this study fully complies with WHO diet and physical activity policy and program recommendations.29 Nonetheless, this is not unique to the countries in this study as a 2017 WHO global survey reported that under 20% of African countries had an operational policy, strategy or action plan for the prevention of overweight and obesity.30 Moreover, several research gaps exists on key contextual determinants of weight gain in women such as the drivers of food choice and physical inactivity in SSA countries28 suggesting the low priority placed on issues and hinting at a low probability of meeting global obesity targets and associated NCD targets.

By providing disaggregated analyses on overweight and obesity burden and highlighting high burden areas, the findings of this study could potentially raise the profile of this issue in national health agendas and ensure efficiency in resource allocation. Similar to the tactics employed in tobacco control in Western countries, high-burden lower administrative units could become a testing ground for interventions prior to state and national level scale up. Given the limited funding environment, a potential source of funding for such interventions include earmarked taxes on alcohol, tobacco and sugar-sweetened beverages, which remain under-utilized in several

African countries.30 Additionally, high burden overweight second-level administrative units can

109

be ruled out as a focus of programs primarily targeting undernutrition among adult women allowing the design and uptake of double-burden interventions in such locations.

Failure to address the large burden of overweight in the identified second-level administrative units places individuals and communities at risk of morbidity and economic losses from NCDs.

Cardiovascular diseases and diabetes, the top two leading causes of BMI-related mortality27 require substantial individual and health system resources to treat and manage, which are desperately lacking in African countries.

Enhanced precision in the estimation of overweight and obesity within lower administrative units in African countries can be achieved through the collection and availability of more publicly available micro data. While the Demographic and Health Survey remains an important resource, it is not designed to be representative at lower scales, prompting the statistical models employed in this study. However, besides this obvious limitation of our study, we also assumed that the population of women at the survey location remained statistic during the period of the study.

Although the World Pop project provides migration data, these are restricted to first-level administrative units and were not suitable for use in this study.31 Moreover, our estimates could potentially have been improved through the inclusion of area-level covariate data such as indicators of socioeconomic status and level of urbanization, which have been shown to be associated with increased burden of overweight,32,33 but this information was not publicly available. Thirdly, since administrative boundary data was obtained from the GADM, we could not verify whether differences existed between the GADM data and the status quo in individual

110

countries. A final limitation to note in our study is that our modeling process assumed information sufficiency in some areas in order to adjust estimates in low information areas.

In conclusion, we noted variable disparities in the age-standardized prevalence of adult female overweight within second-level administrative units in seven African countries while providing sub-national baseline data to partly satisfy measurement of indicator 14 of the global NCD monitoring framework. Overweight in adult women has become a health problem of substantial magnitude in some sub-national administrative units and requires immediate prioritization. The data provided by this study will be critical in prioritizing resources to address this emerging problem and tracking progress towards meeting global obesity and NCD targets.

111

References

1. Kontis V, Mathers CD, Bonita R, et al. Regional contributions of six preventable risk

factors to achieving the 25 × 25 non-communicable disease mortality reduction target: a

modelling study. Lancet Glob Heal. 2015;3(12):e746-e757. doi:10.1016/S2214-

109X(15)00179-5

2. United Nations. Transforming Our World: The 2030 Agenda for Sustainable

Development.; 2015. doi:10.1007/s13398-014-0173-7.2

3. World Health Organization. Global Action Plan for the Prevention and Control of

Noncommunicable Diseases 2013-2020.; 2013. doi:978 92 4 1506236

4. World Health Organization. Global Health Observatory Data Repository. WHO.

http://apps.who.int/gho/data/?theme=main. Published 2017.

5. Onubi OJ, Marais D, Aucott L, Okonofua F, Poobalan AS. Maternal obesity in Africa: a

systematic review and meta-analysis. J Public Health (Bangkok). 2016;38(3):e218-e231.

doi:10.1093/pubmed/fdv138

6. Ofori-Asenso R, Agyeman AA, Laar A, Boateng D. Overweight and obesity epidemic in

Ghana-a systematic review and meta-analysis. BMC Public Health. 2016;16(1):1239.

doi:10.1186/s12889-016-3901-4

7. Kandala N-B, Stranges S. Geographic variation of overweight and obesity among women

in Nigeria: a case for nutritional transition in sub-Saharan Africa. PLoS One.

2014;9(6):e101103. doi:10.1371/journal.pone.0101103

8. Turi KN, Christoph MJ, Grigsby-Toussaint DS. Spatial distribution of underweight,

overweight and obesity among women and children: results from the 2011 Uganda

Demographic and Health Survey. Int J Environ Res Public Health. 2013;10(10):4967-

112

4981. doi:10.3390/ijerph10104967

9. Gewa CA, Leslie TF, Pawloski LR. Geographic distribution and socio-economic

determinants of women’s nutritional status in Mali households. Public Health Nutr.

2013;16(9):1575-1585. doi:10.1017/S136898001200451X

10. Chen C, Wakefield J, Lumely T. The use of sampling weights in Bayesian hierarchical

models for small area estimation. Spat Spatiotemporal Epidemiol. 2014;11:33-43.

doi:10.1016/J.SSTE.2014.07.002

11. Dwyer-Lindgren L, Kakungu F, Hangoma P, et al. Estimation of district-level under-5

mortality in Zambia using birth history data, 1980–2010. Spat Spatiotemporal Epidemiol.

2014;11:89-107. doi:10.1016/J.SSTE.2014.09.002

12. Mercer LD, Wakefield J, Pantazis A, Lutambi AM, Masanja H, Clark S. Space-Time

Smoothing of Complex Survey Data: Small Area Estimation for Child Mortality. Ann

Appl Stat. 2015;9(4):1889-1905. doi:10.1214/15-AOAS872

13. ICF International. Demographic and Health Survey: Sampling and Household Listing

Manual. Calverton, Maryland; 2012.

14. ICF International. MEASURE DHS Biomarker Field Manual. Calverton, Maryland,

U.S.A; 2012.

https://dhsprogram.com/pubs/pdf/DHSM7/DHS6_Biomarker_Manual_9Jan2012.pdf.

Accessed October 20, 2017.

15. Burgert-Brucker CR, Colston J, Roy T, Zachary B. Geographic Displacement Procedure

and Georeferenced Data Release Policy for the Demographic and Health Surveys. DHS

Spatial Analysis Reports No. 7. Calverton, Maryland; 2013.

16. Coffey D. Prepregnancy body mass and weight gain during pregnancy in India and sub-

113

Saharan Africa. Proc Natl Acad Sci U S A. 2015;112(11):3302-3307.

doi:10.1073/pnas.1416964112

17. World Health Organization. Obesity: Preventing and Managing the Global Epidemic.

Report of a WHO Consultation. Vol 894.; 2000.

http://www.ncbi.nlm.nih.gov/pubmed/11234459. Accessed January 10, 2016.

18. Riebler A, Sørbye SH, Simpson D, Rue H. An Intuitive Bayesian Spatial Model for

Disease Mapping That Accounts for Scaling.; 2016. www.r-inla.org. Accessed February 4,

2019.

19. Mercer L, Wakefield J, Chen C, Lumley T. A comparison of spatial smoothing methods

for small area estimation with sampling weights. Spat Stat. 2014;8:69-85.

doi:10.1016/j.spasta.2013.12.001

20. Ahmad OB, Boschi-pinto C, Lopez AD, Murray CJ, Lozano R, Mie I. Age

Standardization of Rates: A New WHO Standard.; 2001.

https://www.who.int/healthinfo/paper31.pdf.

21. Besag J, York J, Mollié A. Bayesian image restoration, with two applications in spatial

statistics. Ann Inst Stat Math. 1991;43(1):1-20. doi:10.1007/BF00116466

22. Rue H, Martino S, Chopin N. Approximate Bayesian inference for latent Gaussian models

by using integrated nested Laplace approximations. J R Stat Soc Ser B Stat Methodol.

2009. doi:10.1111/j.1467-9868.2008.00700.x

23. Spiegelhalter DJ, Best NG, Carlin BP, Van Der Linde A. Bayesian Measures of Model

Complexity and Fit. Vol 64.; 2002.

https://rss.onlinelibrary.wiley.com/doi/pdf/10.1111/1467-9868.00353. Accessed

September 22, 2018.

114

24. Watanabe S. Asymptotic Equivalence of Bayes Cross Validation and Widely Applicable

Information Criterion in Singular Learning Theory. Vol 11.; 2010.

http://www.jmlr.org/papers/volume11/watanabe10a/watanabe10a.pdf. Accessed

September 22, 2018.

25. Warren JL, Perez-Heydrich C, Burgert CR, Emch ME. Influence of Demographic and

Health Survey Point Displacements on Point-in-Polygon Analyses. Spat Demogr.

2016;4(2):117-133. doi:10.1007/s40980-015-0015-z

26. OECD. Overweight or obese population. doi:10.1787/86583552-en

27. The GBD 2015 Obesity Collaborators. Health Effects of Overweight and Obesity in 195

Countries over 25 Years. N Engl J Med. 2017;377(1):13-27.

doi:10.1056/NEJMoa1614362

28. Jaacks LM, Kavle J, Perry A, Nyaku A. Programming maternal and child overweight and

obesity in the context of undernutrition: current evidence and key considerations for low-

and middle-income countries. Public Health Nutr. 2017;20(7):1286-1296.

doi:10.1017/S1368980016003323

29. WHO. Noncommunicable Diseases Progress Monitor, 2017. Geneva; 2017.

http://apps.who.int/iris/bitstream/handle/10665/258940/9789241513029-

eng.pdf?sequence=1.

30. World Health Organization. Assessing National Capacity for the Prevention and Control

of Noncummunicable Diseases: Report of the 2017 Global Survey. Geneva; 2018.

https://apps.who.int/iris/bitstream/handle/10665/276609/9789241514781-eng.pdf?ua=1.

31. Tatem AJ. WorldPop, open data for spatial demography. Sci Data. 2017;4:170004.

doi:10.1038/sdata.2017.4

115

32. Pampel FC, Denney JT, Krueger PM. Obesity, SES, and economic development: A test of

the reversal hypothesis. Soc Sci Med. 2012;74(7):1073-1081.

doi:10.1016/j.socscimed.2011.12.028

33. Jones AD, Acharya Y, Galway LP. Urbanicity Gradients Are Associated with the

Household- and Individual-Level Double Burden of Malnutrition in Sub-Saharan Africa. J

Nutr. 2016;146(6):1257-1267. doi:10.3945/jn.115.226654

116

Supplementary Appendix

ADULT FEMALE OVERWEIGHT AND OBESITY PREVALENCE IN 7 SUB-SAHARAN

AFRICAN COUNTRIES: A BASELINE SUB-NATIONAL ASSESSMENT OF INDICATOR

14 OF THE GLOBAL NCD MONITORING FRAMEWORK

117

Benin

Table 1: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Benin

Country Departments % Overweight Prevalence Lower Upper 95% CI 95% CI

Benin Alibori 20.4 13.8 27.0 Atakora 9.9 7.9 11.8 Atlantique 33.6 27.2 40.0 Borgou 24.2 14.3 34.0 Collines 31.0 26.5 35.6 Donga 13.9 9.8 18.0 Kouffo 16.7 12.5 20.9 Littoral 42.0 41.2 42.8 Mono 33.5 28.8 38.3 Ouémé 36.7 30.3 43.1 Plateau 27.4 16.1 38.7 Zou 28.1 24.1 32.0

118

Table 2: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Benin Department Communes % Overweight Lower Upper 95% %Relative Prevalence 95% CI CI Uncertainty

Alibori Banikoara 11.1 8.3 14.6 57.6 Alibori Gogounou 17.7 17.4 18.1 4.1 Alibori Kandi 32.8 28.5 37.4 27.0 Alibori Karimama 11.6 11.2 12.1 7.7 Alibori Malanville 25.1 18.9 32.6 54.6 Alibori Segbana 17.6 16.4 18.8 13.7 Atakora Cobly 11.9 8.4 16.6 69.1 Atakora Natitingou 9.1 5.3 15.1 108.2 Atakora Toucountouna 10.8 4.2 26.4 206.4 Atakora Boukoumbé 20.9 20.8 21.0 1.3 Atakora Kérou 20.9 20.8 21.0 1.3 Atakora Kouandé 20.9 20.8 21.1 1.3 Atakora Matéri 20.9 20.8 21.0 1.3 Atakora Péhunco 20.9 20.8 21.1 1.3 Atakora Tanguiéta 20.9 20.8 21.1 1.3 Atlantique Abomey-Calavi 36.8 34.4 39.3 13.2 Atlantique Allada 19.5 15.6 24.0 43.0 Atlantique Ouidah 34.5 29.7 39.8 29.3 Atlantique Toffo 24.4 23.1 25.7 11.0 Atlantique Tori-Bossito 35.0 32.6 37.5 14.1 Atlantique Kpomassè 21.0 20.9 21.1 0.8 Atlantique Sô-Ava 20.9 20.8 21.1 1.3 Atlantique Zè 20.9 20.8 21.1 1.3 Borgou N'Dali 19.9 12.0 30.7 94.1 Borgou Nikki 24.6 21.0 28.7 31.3 Borgou Parakou 37.1 36.0 38.1 5.8 Borgou Tchaourou 27.6 24.6 30.7 22.4 Borgou Bembéréké 20.9 20.8 21.1 1.3 Borgou Kalalé 20.8 20.7 21.0 1.3 Borgou Pèrèrè 20.9 20.8 21.1 1.3 Borgou Sinendé 20.9 20.8 21.1 1.3 Collines Savalou 30.5 26.0 35.3 30.5 Collines Bantè 20.9 20.8 21.1 1.3 Collines Dassa-Zoumè 20.9 20.8 21.1 1.3

119

Collines Glazoué 20.9 20.8 21.1 1.3 Collines Ouèssè 20.9 20.8 21.1 1.3 Collines Savè 20.9 20.8 21.1 1.3 Donga Bassila 14.0 9.2 20.6 82.0 Donga Copargo 7.5 6.8 8.3 20.4 Donga Djougou 16.5 12.3 21.8 57.5 Donga Ouaké 20.6 20.5 20.7 1.3 Kouffo Djakotomey 8.7 7.5 10.0 29.1 Kouffo Dogbo 19.1 13.0 27.4 75.3 Kouffo Lalo 19.8 11.1 32.6 108.5 Kouffo Toviklin 16.6 10.9 24.6 82.6 Kouffo Aplahoué 20.9 20.8 21.1 1.3 Kouffo Klouékanmè 20.9 20.8 21.1 1.3 Littoral Cotonou 42.0 41.2 42.8 3.8 Mono Bopa 32.7 29.5 36.2 20.3 Mono Grand-Popo 38.0 37.0 39.1 5.4 Mono Athiémé 20.9 20.8 21.1 1.3 Mono Comè 20.9 20.8 21.1 1.3 Mono Houéyogbé 21.0 20.9 21.2 1.3 Ouémé Adjarra 20.9 20.8 21.1 1.3 Ouémé Adjohoun 20.9 20.7 21.0 1.3 Ouémé Aguégués 20.9 20.8 21.1 1.3 Ouémé Akpro- 20.9 20.8 21.1 1.3 Missérété Ouémé Avrankou 20.9 20.8 21.1 1.3 Ouémé Bonou 20.9 20.8 21.1 1.3 Ouémé Dangbo 20.9 20.8 21.1 1.3 Ouémé Porto-Novo 21.2 21.0 21.3 1.3 Ouémé Sèmè-Kpodji 20.9 20.8 21.1 1.3 Plateau Ifangni 25.2 16.1 37.2 83.4 Plateau Adja-Ouèrè 20.9 20.8 21.1 1.3 Plateau Kétou 20.9 20.8 21.0 1.3 Plateau Pobè 20.9 20.8 21.1 1.3 Plateau Sakété 20.9 20.8 21.1 1.3

120

Zou Abomey 31.5 28.7 34.4 18.1 Zou Agbangnizoun 20.0 19.5 20.6 5.4 Zou Bohicon 33.1 28.2 38.4 30.7 Zou Djidja 24.8 18.6 32.4 55.7 Zou Ouinhi 32.9 27.7 38.5 32.6 Zou Za-Kpota 23.9 17.3 32.1 61.8 Zou Zagnanado 16.5 16.2 16.8 3.8 Zou Zogbodomey 31.7 27.4 36.3 28.1 Zou Covè 20.9 20.8 21.1 1.2

121

122

Ethiopia

Table 3: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Country Regional States and Chartered Cities % Overweight Lower Upper 95% Prevalence 95% CI CI

Ethiopia Addis Abeba 33.3 32.4 34.3 Afar 9.4 3.1 25.1 Amhara 3.6 1.5 8.7 Benshangul-Gumaz 7.4 2.7 18.7 Dire Dawa 23.8 10.4 45.6 Gambela Peoples 9.8 4.2 21.0 Harari People 22.9 11.0 41.5 Oromia 8.0 3.1 19.1 Somali 18.6 12.9 26.0 Southern Nations, Nationalities and 5.9 3.2 10.9 Peoples Tigray 6.6 1.7 22.4

123

Table 4: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Regional Zones % Lower Upper % Relative States and Overweight 95% CI 95% CI Uncertainty Chartered Prevalence Cities Addis Abeba Addis Abeba 33.3 32.4 34.2 5.5 Afar Afar Zone 1 10.5 6.2 17.1 103.9 Afar Afar Zone 2 4.8 3.0 7.4 91.8 Afar Afar Zone 3 9.1 3.2 23.5 224.3 Afar Afar Zone 4 8.4 3.4 19.5 192.5 Afar Afar Zone 5 4.4 4.0 4.8 16.3 Amhara Agew Awi 3.9 1.2 11.8 271.6 Amhara Argoba 6.2 1.2 26.1 404.8 Amhara Bahir Dar 13.4 12.9 13.8 6.7 Special Zone Amhara Debub Gondar 1.5 1.4 1.5 8.3 Amhara Debub Wollo 6.3 3.2 11.9 138.5 Amhara Mirab Gojjam 2.6 1.3 5.1 148.9 Amhara Misraq Gojjam 3.7 1.3 10.0 237.6 Amhara North Shewa 6.8 4.6 9.8 77.6 Amhara Oromia 6.4 1.5 23.3 342.2 Amhara Semen Gondar 4.1 2.4 6.9 111.2 Amhara Semen Wello 5.3 1.6 16.3 275.5 Amhara Wag Himra 2.1 2.0 2.2 10.4 Benshangul- Asosa 6.0 2.0 16.7 243.0 Gumaz Benshangul- Kemashi 6.2 2.5 14.6 193.0 Gumaz Benshangul- Metekel 8.8 5.8 13.3 84.5 Gumaz Dire Dawa Dire Dawa 18.2 8.7 34.4 141.4 Gambela Agnuak 9.0 3.9 19.6 173.7 Peoples Gambela Majang 8.1 7.1 9.2 26.4 Peoples Gambela Nuer 7.5 2.9 17.9 200.8 Peoples Harari People Hareri 19.1 9.7 34.5 129.6 Oromia Arsi 9.4 7.6 11.6 42.5 Oromia Bale 10.0 3.6 25.1 214.1

124

Table 4: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Regional Zones % Lower Upper % Relative States and Overweight 95% CI 95% CI Uncertainty Chartered Prevalence Cities Oromia Borena 8.3 8.0 8.6 7.7 Oromia Debub Mirab 7.1 2.6 17.9 214.8 Shewa Oromia Guji 4.5 4.4 4.7 6.0 Oromia Horo Guduru 8.0 7.7 8.4 8.5 Oromia Ilubabor 9.6 3.4 25.2 228.1 Oromia Jimma 1.9 1.8 1.9 6.8 Oromia Kelem Wellega 5.2 4.9 5.4 9.5 Oromia Mirab Arsi 7.8 4.4 13.4 116.7 Oromia Mirab Hararghe 8.7 2.7 24.4 249.7 Oromia Mirab Shewa 1.4 0.7 2.6 142.8 Oromia Mirab Welega 6.1 2.1 16.4 236.9 Oromia Misraq Harerge 1.9 1.7 2.1 16.5 Oromia Misraq Shewa 11.3 4.8 24.5 173.9 Oromia Misraq Wellega 3.1 2.8 3.3 16.7 Oromia North Shewa 6.7 4.6 9.8 77.6 Somali Afder 9.1 7.0 11.7 51.4 Somali Doolo 18.2 17.7 18.7 5.1 Somali Fafan 18.8 9.3 34.6 134.5 Somali Jarar 24.3 20.4 28.7 34.3 Somali Korahe 12.8 12.1 13.5 11.0 Somali Liben 12.7 6.6 23.3 131.8 Somali Nogob 10.0 2.4 33.5 312.8 Somali Shabelle 13.5 12.4 14.6 16.3 Somali Siti 8.9 2.2 29.5 308.7 Southern Alaba 6.7 1.6 23.8 332.8 Nations, Nationalities and Peoples Southern Alle 7.8 1.8 28.0 334.5 Nations, Nationalities and Peoples Southern Amaro 8.9 2.1 30.3 317.8 Nations,

125

Table 4: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Regional Zones % Lower Upper % Relative States and Overweight 95% CI 95% CI Uncertainty Chartered Prevalence Cities Nationalities and Peoples Southern Basketo 6.9 1.4 28.1 384.0 Nations, Nationalities and Peoples Southern Bench Maji 6.1 5.9 6.3 6.4 Nations, Nationalities and Peoples Southern Burji 35.9 35.5 36.2 2.1 Nations, Nationalities and Peoples Southern Dawro 2.2 2.1 2.3 10.2 Nations, Nationalities and Peoples Southern Debub Omo 4.8 4.6 5.0 9.8 Nations, Nationalities and Peoples Southern Derashe 13.7 12.8 14.7 13.6 Nations, Nationalities and Peoples Southern Gamo Gofa 6.2 6.1 6.3 3.9 Nations, Nationalities and Peoples Southern Gedeo 4.1 1.3 12.5 271.7 Nations, Nationalities and Peoples Southern Gurage 2.3 2.2 2.3 5.9 Nations, Nationalities and Peoples

126

Table 4: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Regional Zones % Lower Upper % Relative States and Overweight 95% CI 95% CI Uncertainty Chartered Prevalence Cities Southern Hadiya 6.1 4.9 7.6 43.2 Nations, Nationalities and Peoples Southern Keffa 4.1 2.8 6.0 78.7 Nations, Nationalities and Peoples Southern Kembata 6.7 6.4 7.0 10.2 Nations, Tembaro Nationalities and Peoples Southern Konso 8.7 2.1 29.3 313.4 Nations, Nationalities and Peoples Southern Konta 9.1 8.8 9.4 6.7 Nations, Nationalities and Peoples Southern Sheka 7.3 6.8 7.7 12.4 Nations, Nationalities and Peoples Southern Sidama 5.8 2.5 12.7 175.8 Nations, Nationalities and Peoples Southern Silti 7.2 2.3 20.8 255.4 Nations, Nationalities and Peoples Southern Wolayita 7.0 2.3 19.5 248.3 Nations, Nationalities and Peoples Southern Yem 5.4 1.2 21.6 375.7 Nations,

127

Table 4: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Ethiopia

Regional Zones % Lower Upper % Relative States and Overweight 95% CI 95% CI Uncertainty Chartered Prevalence Cities Nationalities and Peoples Tigray Debubawi 7.2 2.9 17.0 194.6 Tigray Mehakelegnaw 5.8 2.1 15.3 229.0 Tigray Mi'irabawi 3.9 2.2 6.6 112.5 Tigray Misraqawi 5.1 1.8 13.2 225.2 Tigray Semien 5.3 1.9 13.6 221.1 Mi'irabaw

128

129

Mozambique

Table 5: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Mozambique Country Provinces % Overweight Lower 95% Upper 95% Prevalence CI CI Mozambique Cabo Delgado 13.1 7.1 19.2 Gaza 24.6 22.6 26.5 Inhambane 28.8 21.6 36.0 Manica 15.8 15.5 16.2 Maputo 30.5 20.5 40.5 Maputo City 43.9 41.0 46.9 Nampula 15.4 9.3 21.4 Nassa 12.3 4.7 20.0 Sofala 11.9 3.4 20.4 Tete 11.7 4.4 19.0 Zambezia 6.2 2.8 9.6

130

Table 6: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Mozambique

Province Districts % Lower Upper % Relative Overweight 95% CI 95% CI Uncertainty Prevalence Cabo Delgado Ancuabe 11.7 11.4 11.9 3.9 Cabo Delgado Balama 14.5 13.8 15.1 9.4 Cabo Delgado Macomia 3.8 2.9 5.0 56.8 Cabo Delgado Mecufi 29.2 28.4 29.9 5.1 Cabo Delgado Meluco 8.7 1.5 37.8 418.7 Cabo Delgado Mocimboa da 18.9 13.6 25.6 63.6 Praia Cabo Delgado Montepuez 11.7 7.5 17.8 88.0 Cabo Delgado Mueda 12.2 10.9 13.6 21.9 Cabo Delgado Muidumbe 17.8 17.1 18.6 8.3 Cabo Delgado Namuno 9.6 9.4 9.9 5.2 Cabo Delgado Nangade 8.7 1.5 37.8 418.7 Cabo Delgado Palma 5.5 5.1 6.0 15.7 Cabo Delgado Pemba 30.8 20.3 43.8 76.5 Cabo Delgado Quissanga 8.7 1.5 37.8 418.7 Cabo Delgado Chiúre 8.6 8.4 8.9 5.3 Gaza Bilene 25.7 24.2 27.3 12.1 Gaza Chibuto 21.9 20.4 23.4 13.7 Gaza Chicualacuala 9.5 3.8 21.9 191.2 Gaza Chigubo 8.7 1.5 37.8 418.7 Gaza Mabalane 8.7 1.5 37.8 418.7 Gaza Mandlakazi 28.2 27.4 29.1 5.8 Gaza Massangena 8.7 1.5 37.8 418.7 Gaza Massingir 15.1 14.4 15.8 9.1 Gaza Xai-Xai 27.0 18.0 38.5 76.0 Gaza Chókwè 8.7 8.5 8.9 5.3 Gaza Guijá 8.7 8.5 8.9 5.3 Inhambane Funhalouro 8.7 8.4 9.0 6.4 Inhambane Govuro 14.5 14.1 14.8 5.0 Inhambane Homoine 30.1 25.0 35.8 35.9 Inhambane Inharrime 39.4 33.5 45.7 30.8 Inhambane Inhassoro 15.9 15.1 16.7 10.4 Inhambane Jangamo 38.5 33.5 43.8 26.8

131

Table 6: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Mozambique

Province Districts % Lower Upper % Relative Overweight 95% CI 95% CI Uncertainty Prevalence Inhambane Mabote 8.7 1.5 37.8 418.7 Inhambane Massinga 32.5 28.0 37.3 28.6 Inhambane Morrumbene 29.1 28.7 29.5 2.8 Inhambane Panda 13.5 13.2 13.7 3.7 Inhambane Vilanculos 21.7 20.8 22.5 7.8 Inhambane Zavala 34.7 34.4 35.1 2.3 Manica Barue 12.2 9.1 16.3 59.0 Manica Gondola 13.2 10.8 16.1 40.2 Manica Guro 6.9 6.6 7.3 9.5 Manica Machaze 17.6 17.2 18.0 4.3 Manica Macossa 30.2 29.0 31.4 7.9 Manica Manica 13.8 8.6 21.5 93.2 Manica Mossurize 25.1 24.2 26.0 7.1 Manica Sussundenga 19.1 18.6 19.6 5.0 Manica Tambara 13.4 12.5 14.3 13.4 Maputo Boane 39.0 37.5 40.6 8.0 Maputo Magude 29.6 27.2 32.2 16.6 Maputo Marracuene 27.8 26.9 28.8 6.8 Maputo Moamba 28.9 15.3 47.9 113.0 Maputo Namaacha 27.7 26.4 29.1 9.8 Maputo Manhiça 8.7 8.5 8.9 5.3 Maputo Matutuíne 8.8 8.6 9.0 5.3 Maputo City Boane 39.0 37.5 40.6 8.0 Maputo City Maputo 44.7 42.6 46.8 9.5 Nampula Angoche 10.4 9.7 11.2 14.4 Nampula Erati 20.2 18.2 22.5 21.0 Nampula Lalaua 10.3 9.8 10.9 10.0 Nampula Malema 12.4 11.6 13.2 12.6 Nampula Meconta 6.9 3.6 13.0 136.3 Nampula Mecuburi 4.6 4.4 4.8 8.1 Nampula Memba 3.6 3.2 3.9 19.8 Nampula Mogovolas 3.8 3.4 4.1 18.1 Nampula Moma 13.8 7.1 25.2 131.1

132

Table 6: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Mozambique

Province Districts % Lower Upper % Relative Overweight 95% CI 95% CI Uncertainty Prevalence Nampula Monapo 13.0 5.6 27.5 167.7 Nampula Mongincual 15.0 13.4 16.7 21.5 Nampula Mossuril 9.0 8.3 9.7 15.7 Nampula Muecate 40.5 38.1 43.0 11.9 Nampula Murrupula 14.7 14.0 15.4 9.8 Nampula Nacala Velha 15.7 14.9 16.6 10.7 Nampula Namapa 14.1 8.8 21.9 93.3 Nampula Nampula 22.1 12.4 36.4 108.7 Nampula Ribaue 8.6 4.5 15.9 133.7 Nassa Cuamba 21.6 20.9 22.4 7.2 Nassa Lago 9.8 9.5 10.1 6.7 Nassa Lichinga 18.6 10.8 30.2 104.2 Nassa Majune 6.2 5.9 6.5 8.4 Nassa Mandimba 21.1 19.8 22.4 12.0 Nassa Marrupa 8.1 3.4 18.2 183.6 Nassa Mavago 8.7 7.6 10.0 27.9 Nassa Mecanhelas 5.5 5.1 6.0 17.1 Nassa Mecula 8.7 1.5 37.8 418.7 Nassa Metarica 7.1 6.5 7.8 17.8 Nassa Muembe 10.8 10.1 11.7 14.8 Nassa N'gauma 9.5 8.7 10.2 16.0 Nassa Nipepe 8.7 1.5 37.8 418.7 Nassa Sanga 12.2 11.8 12.5 5.8 Nassa Maúa 8.6 8.4 8.9 5.3 Sofala Buzi 12.2 10.5 14.2 30.4 Sofala Caia 7.1 2.8 17.0 199.2 Sofala Chemba 3.2 2.8 3.6 25.7 Sofala Cheringoma 8.7 1.5 37.8 418.7 Sofala Chibabava 6.4 6.1 6.8 11.0 Sofala Dondo 21.9 16.0 29.2 60.2 Sofala Gorongosa 7.7 5.8 10.1 56.1 Sofala Machanga 10.1 9.7 10.6 9.7 Sofala Maringue 2.7 2.5 2.9 11.5

133

Table 6: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Mozambique

Province Districts % Lower Upper % Relative Overweight 95% CI 95% CI Uncertainty Prevalence Sofala Marromeu 6.1 5.3 7.1 29.3 Sofala Muanza 8.7 1.5 37.8 418.7 Sofala Nhamatanda 6.9 3.4 13.6 147.6 Tete Cahora Bassa 8.1 2.9 20.7 219.2 Tete Changara 19.1 8.5 37.7 152.8 Tete Chifunde 11.7 11.2 12.1 8.0 Tete Chiuta 5.7 5.4 5.9 9.6 Tete Macanga 8.7 1.5 37.8 418.7 Tete Magoe 12.2 11.8 12.5 6.1 Tete Maravia 4.7 4.2 5.3 23.0 Tete Moatize 11.0 5.9 19.9 126.9 Tete Mutarara 3.6 3.4 3.7 7.7 Tete Tsangano 13.3 12.8 13.9 8.5 Tete Zumbu 9.1 8.6 9.5 10.3 Tete Angónia 8.7 8.5 8.9 5.3 Zambezia Alto Molocue 6.3 5.9 6.7 12.9 Zambezia Chinde 5.3 1.9 13.4 218.3 Zambezia Gile 7.7 6.6 8.9 29.5 Zambezia Gurue 3.1 2.6 3.7 35.9 Zambezia Ile 3.1 2.9 3.4 14.7 Zambezia Inhassunge 8.7 1.5 37.8 418.7 Zambezia Lugela 9.0 8.6 9.5 9.5 Zambezia Maganja da 1.6 1.5 1.6 11.0 Costa Zambezia Milange 5.9 2.8 11.8 153.5 Zambezia Mocuba 4.8 1.4 15.1 283.3 Zambezia Mopeia 8.7 1.5 37.8 418.7 Zambezia Morrumbala 2.5 2.1 3.0 34.6 Zambezia Namacurra 8.7 1.5 37.8 418.7 Zambezia Namarroi 3.8 3.5 4.1 15.6 Zambezia Nicoadala 14.2 7.8 24.4 116.7 Zambezia Pebane 11.8 11.5 12.0 4.8

134

135

Nigeria

Table 7: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Nigeria Country States % Overweight Lower Upper Prevalence 95% CI 95% CI

Nigeria Abia 33.2 27.8 38.6 Adamawa 24.7 13.8 35.6 Akwa Ibom 34.8 32.7 37.0 Anambra 39.9 33.9 45.9 Bauchi 13.4 5.7 21.1 Bayelsa 37.2 33.7 40.8 Benue 19.9 15.5 24.3 Borno 17.0 9.4 24.6 Cross River 30.8 23.3 38.2 Delta 30.6 24.2 36.9 Ebonyi 19.1 16.4 21.7 Edo 37.6 35.8 39.4 Ekiti 34.9 34.0 35.8 Enugu 35.4 31.4 39.3 Federal Capital 42.8 28.7 57.0 Territory Gombe 14.5 7.1 21.9 Imo 39.3 36.8 41.7 Jigawa 14.2 8.8 19.6 Kaduna 25.4 11.9 39.0 Kano 19.1 7.7 30.4 Katsina 10.6 6.1 15.0 Kebbi 18.5 12.8 24.2 Kogi 30.3 23.4 37.2 Kwara 34.8 28.7 40.9 Lagos 46.7 45.3 48.0 Nassarawa 24.6 14.7 34.5 Niger 27.0 17.3 36.7 Ogun 36.2 33.7 38.8 Ondo 34.6 27.6 41.7 Osun 32.4 26.3 38.6 Oyo 31.7 22.8 40.5 Plateau 29.1 14.6 43.6 Rivers 42.1 26.4 57.8

136

Table 7: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Nigeria Country States % Overweight Lower Upper Prevalence 95% CI 95% CI

Sokoto 13.9 6.3 21.6 Taraba 26.3 18.3 34.2 Yobe 25.5 16.5 34.6 Zamfara 13.1 8.7 17.5

137

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Abia Aba North 36.6 35.9 37.4 4.2 Abia Aba South 36.6 36.0 37.3 3.6 Abia Arochukwu 16.1 15.8 16.5 4.3 Abia Bende 26.3 26.1 26.6 1.8 Abia Ikwuano 31.1 12.0 59.9 154.1 Abia Isiala-Ngwa North 31.3 30.8 31.8 3.2 Abia Isiala-Ngwa South 25.8 25.5 26.1 2.0 Abia Isuikwuato 35.3 34.4 36.2 5.3 Abia Oboma Ngwa 21.2 20.8 21.5 3.2 Abia Ohafia 26.5 23.2 30.2 26.6 Abia Osisioma Ngwa 31.4 12.2 60.2 153.0 Abia Ugwunagbo 32.5 12.1 62.7 155.6 Abia Ukwa East 43.9 43.5 44.4 2.1 Abia Ukwa West 32.6 12.2 62.8 155.1 Abia Umu-Nneochi 31.5 11.8 61.4 157.5 Abia Umuahia North 50.8 49.9 51.6 3.3 Abia Umuahia South 35.7 35.1 36.3 3.3 Adamawa Demsa 7.6 7.3 8.0 8.7 Adamawa Fufore 17.5 17.0 18.0 5.6 Adamawa Ganye 22.7 7.4 52.0 196.1 Adamawa Girei 16.6 11.2 23.9 76.5 Adamawa Gombi 41.1 40.4 41.8 3.2 Adamawa Guyuk 14.6 6.0 31.4 174.4 Adamawa Hong 34.1 33.6 34.7 3.2 Adamawa Jada 40.1 39.5 40.8 3.2 Adamawa Lamurde 11.4 11.1 11.8 6.3 Adamawa Madagali 12.9 3.2 39.5 281.3 Adamawa Maiha 24.1 7.7 54.5 194.3 Adamawa Mayo-Belwa 24.0 23.5 24.5 3.9 Adamawa Michika 18.7 4.9 50.9 246.0 Adamawa Mubi North 31.7 22.2 42.9 65.4 Adamawa Mubi South 24.7 5.8 64.1 235.6 Adamawa Numan 17.5 5.8 42.2 208.5 Adamawa Shelleng 14.0 13.7 14.4 4.8

138

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Adamawa Song 22.2 7.8 48.9 185.4 Adamawa Toungo 23.0 6.3 57.1 221.1 Adamawa Yola North 42.9 42.1 43.6 3.4 Adamawa Yola South 44.4 43.9 44.9 2.3 Akwa Abak 20.6 20.3 20.9 2.7 Ibom Akwa Eastern Obolo 27.0 8.6 59.4 188.0 Ibom Akwa Eket 48.6 48.3 49.0 1.5 Ibom Akwa Esit Eket 27.6 27.0 28.3 4.8 Ibom Akwa Essien Udim 35.0 34.3 35.7 3.9 Ibom Akwa Etim Ekpo 28.7 10.0 59.5 172.5 Ibom Akwa Etinan 29.4 10.1 60.6 171.5 Ibom Akwa Ibeno 28.8 9.8 60.0 174.6 Ibom Akwa Ibesikpo Asutan 30.9 11.0 61.9 164.9 Ibom Akwa Ibiono Ibom 44.8 43.8 45.7 4.2 Ibom Akwa Ika 30.8 10.3 63.2 171.5 Ibom Akwa Ikono 31.7 12.2 60.7 153.0 Ibom Akwa Ikot Abasi 13.5 13.1 13.9 6.0 Ibom Akwa Ikot Ekpene 48.2 47.5 49.0 3.1 Ibom Akwa Ini 20.5 19.7 21.2 7.4 Ibom Akwa Itu 38.6 38.1 39.2 2.7 Ibom Akwa Mbo 16.5 16.2 16.8 3.3 Ibom

139

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Akwa Mkpat Enin 27.9 9.7 58.1 173.4 Ibom Akwa Nsit Atai 39.8 39.1 40.6 3.7 Ibom Akwa Nsit Ibom 27.3 26.9 27.7 2.9 Ibom Akwa Nsit Ubium 17.4 17.1 17.7 3.2 Ibom Akwa Obot Akara 24.0 23.6 24.3 3.0 Ibom Akwa Okobo 23.0 22.3 23.9 7.0 Ibom Akwa Onna 30.5 9.9 63.8 176.7 Ibom Akwa Oron 31.4 30.9 31.9 3.1 Ibom Akwa Oruk Anam 25.6 8.7 55.6 183.3 Ibom Akwa Udung Uko 26.0 7.4 60.9 205.5 Ibom Akwa Ukanafun 13.5 13.1 13.9 5.7 Ibom Akwa Uruan 32.3 12.1 62.3 155.3 Ibom Akwa Urue Offong/Oruko 26.2 8.2 58.5 192.1 Ibom Akwa Uyo 43.9 42.1 45.8 8.6 Ibom Anambra Aguata 45.3 44.6 46.0 3.1 Anambra Anambra East 49.0 40.3 57.7 35.6 Anambra Anambra West 33.8 13.3 63.0 147.0 Anambra Anaocha 31.1 30.6 31.5 2.7 Anambra Awka North 40.4 17.1 68.8 127.9 Anambra Awka South 52.4 51.7 53.1 2.6 Anambra Ayamelum 19.1 18.7 19.5 4.1 Anambra Dunukofia 45.2 44.6 45.7 2.3 Anambra Ekwusigo 44.4 43.6 45.2 3.6 Anambra Idemili North 40.9 40.1 41.7 3.9

140

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Anambra Idemili South 58.1 57.5 58.7 1.9 Anambra Ihiala 26.1 25.3 27.0 6.4 Anambra Njikoka 67.9 67.0 68.7 2.5 Anambra Nnewi North 44.1 15.7 76.9 139.0 Anambra Nnewi South 53.2 52.3 54.0 3.2 Anambra Ogbaru 18.9 18.7 19.1 2.5 Anambra Onitsha North 36.7 36.2 37.2 2.7 Anambra Onitsha South 38.7 14.3 70.6 145.3 Anambra Orumba North 45.7 44.9 46.4 3.2 Anambra Orumba South 33.2 32.6 33.8 3.6 Anambra Oyi 25.8 25.6 26.0 1.7 Bauchi Alkaleri 19.1 18.5 19.7 6.2 Bauchi Bauchi 22.0 12.3 36.1 108.6 Bauchi Bogoro 18.8 5.2 49.3 235.1 Bauchi Dambam 18.3 18.0 18.7 3.7 Bauchi Darazo 12.1 3.9 31.5 228.3 Bauchi Dass 20.1 4.8 55.3 251.2 Bauchi Gamawa 2.5 2.5 2.6 6.2 Bauchi Ganjuwa 12.6 12.3 13.0 5.4 Bauchi Giade 12.6 3.6 35.8 254.2 Bauchi Itas/Gadau 6.0 5.8 6.2 8.0 Bauchi Jama'are 9.8 2.4 31.7 300.3 Bauchi Katagum 7.7 7.5 8.0 6.1 Bauchi Kirfi 13.5 4.2 36.0 235.0 Bauchi Misau 34.7 34.2 35.1 2.5 Bauchi Ningi 13.3 13.1 13.5 3.3 Bauchi Shira 6.0 5.8 6.1 6.5 Bauchi Tafawa-Balewa 15.1 14.9 15.3 2.8 Bauchi Toro 21.1 18.7 23.8 24.4 Bauchi Warji 11.6 11.4 11.8 3.4 Bauchi Zaki 3.7 3.6 3.8 6.1 Bayelsa Brass 42.1 33.6 51.0 41.2 Bayelsa Ekeremor 31.7 29.9 33.5 11.5 Bayelsa Kolokuma/Opokuma 33.9 10.7 68.5 170.7

141

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Bayelsa Nembe 49.2 48.5 49.9 2.9 Bayelsa Ogbia 40.7 40.0 41.5 3.7 Bayelsa Sagbama 32.0 12.5 60.8 150.9 Bayelsa Southern Ijaw 31.4 30.9 31.9 3.1 Bayelsa Yenegoa 38.5 29.2 48.8 50.8 Benue Ado 8.7 5.4 13.6 95.4 Benue Agatu 20.0 19.5 20.6 5.5 Benue Apa 22.3 7.4 50.9 194.9 Benue Buruku 7.5 7.2 7.8 7.8 Benue Gboko 21.8 12.1 36.2 110.5 Benue Guma 20.1 7.1 45.3 189.8 Benue Gwer East 26.1 25.5 26.7 4.6 Benue Gwer West 23.2 8.0 51.0 185.4 Benue Katsina-Ala 19.7 19.1 20.3 5.9 Benue Konshisha 21.1 7.4 47.4 189.2 Benue Kwande 28.2 28.0 28.4 1.3 Benue Logo 20.8 6.8 48.6 200.7 Benue Makurdi 40.7 40.1 41.2 2.7 Benue Obi 6.1 5.7 6.6 14.8 Benue Ogbadibo 33.3 32.4 34.3 5.9 Benue Ohimini 16.6 16.0 17.3 7.9 Benue Oju 28.9 28.5 29.3 2.9 Benue Okpokwu 54.2 53.9 54.6 1.3 Benue Oturkpo 28.0 22.9 33.7 38.6 Benue Tarka 9.7 9.4 9.9 4.6 Benue Ukum 22.4 22.0 22.8 3.6 Benue Ushongo 21.3 7.2 48.3 193.3 Benue Vandeikya 31.2 30.8 31.7 3.2 Borno Abadam 13.9 13.5 14.3 5.7 Borno Askira/Uba 19.4 6.5 45.4 200.8 Borno Bama 12.4 3.4 36.4 265.7 Borno Bayo 17.6 5.5 43.8 217.5 Borno Biu 17.1 5.8 40.7 204.3 Borno Chibok 16.8 4.5 46.7 250.8

142

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Borno Damboa 13.1 12.9 13.4 3.8 Borno Dikwa 12.9 12.7 13.3 4.6 Borno Gubio 13.1 3.9 35.7 242.5 Borno Guzamala 13.1 4.1 35.2 237.1 Borno Gwoza 3.3 3.1 3.6 13.2 Borno Hawul 21.2 20.8 21.7 4.1 Borno Jere 33.0 32.5 33.5 2.9 Borno Kaga 15.1 4.7 39.2 229.1 Borno Kala/Balge 13.2 3.1 42.3 296.4 Borno Konduga 24.4 23.5 25.3 7.4 Borno Kukawa 13.8 3.5 41.2 273.7 Borno Kwaya Kusar 22.9 22.7 23.2 2.3 Borno Mafa 15.7 5.0 39.8 221.0 Borno Magumeri 16.0 15.4 16.6 7.4 Borno Maiduguri 19.3 4.5 55.1 261.7 Borno Marte 14.5 3.9 41.2 256.9 Borno Mobbar 6.8 6.5 7.0 7.1 Borno Monguno 16.8 13.4 20.9 44.6 Borno Ngala 13.7 3.2 43.5 293.4 Borno Nganzai 9.5 9.2 9.8 6.8 Borno Shani 18.4 6.3 43.1 200.4 Cross Abi 21.7 7.2 50.0 197.0 River Cross Akamkpa 30.8 30.5 31.2 2.4 River Cross Akpabuyo 40.1 29.6 51.6 54.9 River Cross Bakassi 36.5 6.8 82.2 206.5 River Cross Bekwarra 23.1 7.7 52.0 191.8 River Cross Biase 25.5 9.5 52.8 169.9 River Cross Boki 22.0 21.6 22.4 3.7 River Cross Calabar South 33.7 12.3 64.7 155.6 River

143

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Cross Calabar Municipal 54.1 52.7 55.6 5.5 River Cross Etung 44.6 44.0 45.1 2.4 River Cross Ikom 36.0 35.5 36.5 2.9 River Cross Obanliku 35.5 34.8 36.3 4.2 River Cross Obubra 7.9 7.5 8.2 8.4 River Cross Obudu 36.2 35.6 36.8 3.4 River Cross Odukpani 33.7 33.0 34.4 4.1 River Cross Ogoja 19.7 19.4 20.1 3.9 River Cross Yakurr 33.2 32.9 33.6 2.3 River Cross Yala 15.8 15.2 16.4 7.3 River Delta Aniocha North 34.1 30.9 37.4 19.0 Delta Aniocha South 33.6 12.9 63.2 149.7 Delta Bomadi 30.2 26.8 33.9 23.5 Delta Burutu 25.8 25.5 26.1 2.4 Delta Ethiope West 23.8 23.5 24.1 2.6 Delta Ethiope East 31.3 30.8 31.7 2.9 Delta Ika North East 33.1 12.5 63.3 153.4 Delta Ika South 29.7 17.2 46.2 97.5 Delta Isoko North 20.7 20.2 21.1 4.6 Delta Isoko South 29.3 9.7 61.4 176.4 Delta Ndokwa East 31.3 12.4 59.5 150.4 Delta Ndokwa West 30.6 11.6 59.7 157.1 Delta Okpe 23.1 22.9 23.3 1.8 Delta Oshimili North 34.8 12.8 66.0 152.7 Delta Oshimili South 54.6 54.4 54.8 0.8 Delta Patani 22.2 22.0 22.5 2.1 Delta Sapele 35.0 33.9 36.1 6.4

144

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Delta Udu 35.4 12.5 67.7 156.1 Delta Ughelli North 40.1 39.6 40.7 2.7 Delta Ughelli South 23.6 23.5 23.7 1.2 Delta Ukwuani 22.5 22.2 22.7 2.4 Delta Uvwie 33.3 32.9 33.7 2.5 Delta Warri North 32.6 13.1 60.7 145.8 Delta Warri South-West 43.5 43.1 43.9 1.9 Delta Warri South 59.8 59.5 60.1 0.9 Ebonyi Abakaliki 7.8 7.6 7.9 4.3 Ebonyi Afikpo North 24.8 24.3 25.3 4.0 Ebonyi Afikpo South 25.3 8.9 53.9 177.8 Ebonyi Ebonyi 45.6 45.0 46.2 2.5 Ebonyi Ezza North 13.2 12.5 13.8 9.8 Ebonyi Ezza South 16.6 16.2 17.0 4.6 Ebonyi Ikwo 8.1 7.9 8.4 6.6 Ebonyi Ishielu 11.6 6.7 19.3 108.9 Ebonyi Ivo 22.8 14.3 34.3 87.5 Ebonyi Izzi 3.5 3.4 3.7 8.4 Ebonyi Ohaozara 20.5 20.3 20.7 2.1 Ebonyi Ohaukwu 14.3 13.9 14.7 5.7 Ebonyi Onicha 33.1 32.8 33.3 1.4 Edo Akoko-Edo 33.8 13.6 62.3 144.4 Edo Egor 42.0 38.2 45.9 18.5 Edo Esan Central 37.0 36.3 37.7 3.8 Edo Esan North-East 37.5 37.1 37.8 1.8 Edo Esan South-East 30.5 30.0 30.9 3.2 Edo Esan West 41.2 35.2 47.6 30.0 Edo Etsako Central 32.8 12.0 63.6 157.4 Edo Etsako East 29.8 16.9 47.0 101.0 Edo Etsako West 34.1 33.4 34.7 3.9 Edo Iguegben 34.7 13.7 64.0 144.8 Edo Ikpoba-Okha 36.7 35.9 37.4 4.0 Edo Oredo 49.1 48.4 49.8 2.9 Edo Orhionmwon 36.2 35.5 36.8 3.7

145

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Edo Ovia North-East 33.8 33.2 34.5 4.0 Edo Ovia South-West 29.3 29.0 29.7 2.7 Edo Owan East 42.2 41.7 42.6 2.1 Edo Owan West 32.8 32.4 33.2 2.6 Edo Uhunmwonde 35.7 14.3 64.9 141.7 Ekiti Ado-Ekiti 45.3 44.5 46.0 3.2 Ekiti Efon 9.5 8.9 10.2 13.3 Ekiti Ekiti East 39.4 35.9 43.0 17.8 Ekiti Ekiti South-West 6.1 5.8 6.3 8.1 Ekiti Ekiti West 30.1 29.3 30.8 5.0 Ekiti Emure 33.7 12.4 64.7 155.2 Ekiti Gbonyin 36.3 30.6 42.4 32.7 Ekiti Ido-Osi 32.3 11.5 63.6 161.1 Ekiti Ijero 34.4 33.7 35.1 4.2 Ekiti Ikere 41.8 41.2 42.5 3.0 Ekiti Ikole 41.3 39.0 43.7 11.4 Ekiti Ilejemeje 33.1 11.1 66.2 166.4 Ekiti Irepodun/Ifelodun 27.3 26.8 27.8 3.9 Ekiti Ise/Orun 34.8 13.3 64.8 148.1 Ekiti Moba 33.5 32.7 34.3 4.7 Ekiti Oye 34.5 33.9 35.1 3.4 Enugu Aninri 26.4 9.1 56.5 179.4 Enugu Awgu 39.1 33.2 45.4 31.2 Enugu Enugu East 43.8 42.6 45.0 5.5 Enugu Enugu North 46.3 45.3 47.3 4.3 Enugu Enugu South 30.4 29.4 31.4 6.5 Enugu Ezeagu 48.0 47.4 48.6 2.6 Enugu Igbo-Etiti 38.5 38.0 39.1 2.9 Enugu Igbo-Eze-North 28.3 27.6 29.0 4.9 Enugu Igbo-Eze-South 40.5 40.1 40.8 1.8 Enugu Isi-Uzo 8.1 7.9 8.3 4.9 Enugu Nkanu East 8.1 7.8 8.4 7.3 Enugu Nkanu West 36.8 36.2 37.4 3.3 Enugu Nsukka 36.8 26.0 49.0 62.5

146

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Enugu Oji-River 38.7 15.9 67.7 133.8 Enugu Udenu 28.2 9.8 58.6 172.8 Enugu Udi 31.4 30.9 31.8 3.0 Enugu Uzo-Uwani 24.1 23.3 24.8 6.3 Federal Abaji 33.2 32.3 34.2 5.6 Capital Territory Federal Municipal Area 58.6 58.2 58.9 1.1 Capital Council Territory Federal Bwari 35.8 20.3 55.0 96.9 Capital Territory Federal Gwagwalada 37.0 35.3 38.6 8.8 Capital Territory Federal Kuje 25.6 25.3 25.8 1.9 Capital Territory Federal Kwali 21.2 20.8 21.5 3.5 Capital Territory Gombe Akko 20.5 14.7 27.9 64.5 Gombe Balanga 20.9 12.6 32.6 95.2 Gombe Billiri 29.2 26.0 32.6 22.8 Gombe Dukku 1.0 0.9 1.0 6.4 Gombe Funakaye 13.5 13.0 13.9 6.8 Gombe Gombe 14.7 3.9 42.0 259.4 Gombe Kaltungo 13.8 13.5 14.1 4.9 Gombe Kwami 6.4 4.3 9.5 80.6 Gombe Nafada 8.6 3.4 20.5 198.4 Gombe Shomgom 19.1 6.2 45.8 208.0 Gombe Yamaltu/Deba 32.1 31.6 32.6 3.1 Imo Aboh-Mbaise 38.8 38.0 39.7 4.3 Imo Ahiazu-Mbaise 45.7 45.0 46.5 3.3 Imo Ehime-Mbano 44.1 37.0 51.5 32.7 Imo Ezinihitte 35.7 13.4 66.6 149.2

147

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Imo Ideato South 25.1 22.7 27.6 19.8 Imo Ideato North 50.4 49.5 51.3 3.6 Imo Ihitte/Uboma 36.9 14.5 66.8 141.7 Imo Ikeduru 36.9 36.4 37.4 2.7 Imo Isiala Mbano 36.7 13.9 67.7 146.4 Imo Isu 35.9 12.4 68.8 157.3 Imo Mbaitoli 36.5 14.7 65.7 139.6 Imo Ngor-Okpala 33.4 13.3 62.1 146.1 Imo Njaba 36.0 12.8 68.3 154.3 Imo Nkwerre 35.2 12.1 68.1 159.2 Imo Nwangele 35.0 11.4 69.1 164.9 Imo Obowo 33.6 32.9 34.3 4.1 Imo Oguta 23.5 23.1 23.9 3.2 Imo Ohaji/Egbema 20.4 17.0 24.2 35.5 Imo Okigwe 30.6 30.2 31.1 2.9 Imo Orlu 36.9 14.1 67.5 144.7 Imo Orsu 48.9 48.2 49.6 2.9 Imo Oru East 33.1 31.8 34.4 7.7 Imo Oru West 41.8 30.6 54.0 55.9 Imo Owerri Municipal 39.7 11.6 77.0 164.4 Imo Owerri North 31.8 27.0 37.1 31.7 Imo Owerri West 63.6 62.9 64.3 2.2 Imo Onuimo 36.7 13.8 67.8 147.2 Jigawa Auyo 9.1 2.9 25.8 250.5 Jigawa Babura 2.9 2.8 3.1 8.2 Jigawa Biriniwa 12.5 3.8 34.1 241.0 Jigawa Birnin Kudu 12.5 4.2 31.9 222.5 Jigawa Buji 10.9 3.2 31.1 255.3 Jigawa Dutse 18.9 17.3 20.6 17.4 Jigawa Gagarawa 6.7 2.0 20.4 275.4 Jigawa Garki 8.0 2.5 22.8 252.8 Jigawa Gumel 5.4 1.3 20.0 348.4 Jigawa Guri 13.6 4.4 35.0 225.8 Jigawa Gwaram 25.7 25.6 25.9 1.0

148

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Jigawa Gwiwa 6.1 1.7 19.8 297.3 Jigawa Hadejia 9.8 9.6 10.1 4.8 Jigawa Jahun 6.9 6.7 7.1 4.9 Jigawa Kafin Hausa 9.0 2.7 25.6 255.5 Jigawa Kaugama 7.6 7.3 7.8 6.3 Jigawa Kazaure 6.5 1.9 20.0 279.6 Jigawa Kiri Kasama 10.2 3.1 29.1 254.6 Jigawa Kiyawa 10.5 3.1 29.9 254.3 Jigawa Maigatari 2.3 2.2 2.5 14.1 Jigawa Malam Madori 10.0 2.9 29.5 265.7 Jigawa Miga 9.5 9.3 9.7 4.4 Jigawa Ringim 19.2 10.4 32.6 115.4 Jigawa Roni 7.2 2.0 22.7 288.7 Jigawa Sule Tankarkar 5.0 4.8 5.2 7.0 Jigawa Taura 12.2 12.0 12.3 2.7 Jigawa Yankwashi 5.9 1.4 20.7 328.2 Kaduna Birnin Gwari 19.3 18.8 19.9 5.6 Kaduna Chikun 39.5 38.7 40.4 4.4 Kaduna Giwa 18.5 18.0 19.1 6.0 Kaduna Igabi 27.9 14.0 48.0 121.5 Kaduna Ikara 16.7 5.6 40.5 209.8 Kaduna Jaba 12.4 12.0 12.8 6.1 Kaduna Jema'A 21.8 7.7 48.4 187.1 Kaduna Kachia 26.6 9.7 54.9 170.1 Kaduna Kaduna North 28.3 8.5 62.9 192.1 Kaduna Kaduna South 35.5 34.6 36.3 4.8 Kaduna Kagarko 28.8 19.3 40.6 74.1 Kaduna Kajuru 24.7 8.3 54.1 185.6 Kaduna Kaura 18.8 5.6 47.2 220.8 Kaduna Kauru 19.3 18.9 19.7 4.0 Kaduna Kubau 6.5 6.3 6.7 6.1 Kaduna Kudan 18.6 6.1 44.9 208.2 Kaduna Lere 21.1 20.5 21.8 5.9 Kaduna Makarfi 17.8 5.2 46.2 230.5

149

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Kaduna Sabon Gari 27.1 26.7 27.5 2.9 Kaduna Sanga 42.7 34.7 51.0 38.1 Kaduna Soba 19.5 19.0 19.9 4.9 Kaduna Zangon Kataf 23.7 23.2 24.3 4.7 Kaduna Zaria 21.3 6.6 50.9 208.1 Kano Ajingi 14.4 4.6 36.9 224.7 Kano Albasu 15.3 4.7 39.9 229.3 Kano Bagwai 11.6 3.7 30.6 232.3 Kano Bebeji 17.0 5.3 42.9 221.3 Kano Bichi 8.3 8.0 8.6 8.1 Kano Bunkure 17.1 16.7 17.4 4.5 Kano Dala 27.5 27.0 27.9 3.2 Kano Dambatta 6.7 6.5 6.8 4.0 Kano Dawakin Kudu 16.4 5.3 40.8 216.8 Kano Dawakin Tofa 20.3 20.0 20.6 3.2 Kano Doguwa 16.7 5.5 41.0 211.8 Kano Fagge 18.3 5.7 45.5 216.9 Kano Gabasawa 11.4 3.7 30.4 233.7 Kano Garko 15.3 15.0 15.7 4.8 Kano Garun Malam 16.9 5.2 43.0 224.7 Kano Gaya 18.3 17.8 18.9 6.1 Kano Gezawa 14.4 4.4 38.2 233.7 Kano Gwale 20.4 6.5 48.4 204.9 Kano Gwarzo 12.9 4.0 34.5 237.0 Kano Kabo 13.2 4.3 34.2 226.6 Kano Kano Municipal 18.8 4.9 51.4 248.1 Kano Karaye 15.0 4.5 39.8 235.1 Kano Kibiya 16.4 4.8 43.1 233.8 Kano Kiru 22.5 22.1 22.9 3.5 Kano Kumbotso 17.2 5.8 41.1 206.1 Kano Kunchi 8.7 2.6 25.4 260.9 Kano Kura 16.7 4.8 44.2 236.1 Kano Madobi 22.5 22.0 23.0 4.5 Kano Makoda 8.8 2.8 24.8 251.0

150

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Kano Minjibir 5.7 5.4 6.0 11.1 Kano Nassarawa 32.5 31.8 33.1 4.0 Kano Rano 18.0 17.6 18.3 4.1 Kano Rimin Gado 4.9 4.8 5.0 4.0 Kano Rogo 16.8 5.4 41.3 214.0 Kano Shanono 11.5 3.6 31.4 241.4 Kano Sumaila 15.3 4.4 41.2 240.5 Kano Takai 14.6 4.6 37.5 225.2 Kano Tarauni 18.9 6.0 46.1 212.5 Kano Tofa 9.1 8.8 9.3 5.6 Kano Tsanyawa 19.3 19.2 19.4 1.2 Kano Tudun Wada 16.3 5.6 38.9 203.6 Kano Ungogo 33.0 32.4 33.7 4.0 Kano Warawa 14.8 14.6 15.0 2.6 Kano Wudil 15.7 5.2 38.7 213.7 Katsina Bakori 15.5 4.9 39.2 221.7 Katsina Batagarawa 16.6 16.4 16.8 2.4 Katsina Batsari 12.6 12.2 12.9 5.3 Katsina Baure 5.7 1.4 20.5 337.2 Katsina Bindawa 7.9 2.3 23.5 268.0 Katsina Charanchi 9.8 3.0 27.8 254.0 Katsina Dandume 17.3 5.1 44.6 228.6 Katsina Danja 16.8 5.4 41.9 217.6 Katsina Dan Musa 11.6 3.7 31.4 237.7 Katsina Daura 5.1 1.2 19.1 348.0 Katsina Dutsi 5.4 1.5 17.8 299.2 Katsina Dutsin Ma 10.2 3.1 28.5 249.3 Katsina Faskari 16.1 15.4 16.7 8.0 Katsina Funtua 15.0 6.8 29.9 154.2 Katsina Ingawa 4.9 4.7 5.0 4.9 Katsina Jibia 8.7 8.5 8.9 4.9 Katsina Kafur 15.0 4.7 38.9 228.2 Katsina Kaita 9.2 2.8 26.2 254.2 Katsina Kankara 19.2 18.8 19.7 4.9

151

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Katsina Kankiya 9.7 3.1 26.4 240.9 Katsina Katsina 33.9 33.3 34.5 3.5 Katsina Kurfi 8.4 8.3 8.6 4.4 Katsina Kusada 8.8 2.5 26.4 270.2 Katsina Mai'Adua 5.2 1.2 19.5 354.4 Katsina Malumfashi 13.3 13.1 13.6 3.8 Katsina Mani 3.2 3.0 3.3 9.1 Katsina Mashi 2.7 2.6 2.9 8.0 Katsina Matazu 5.3 5.2 5.4 5.2 Katsina Musawa 6.5 6.3 6.7 5.8 Katsina Rimi 8.6 2.5 26.0 272.1 Katsina Sabuwa 17.7 17.3 18.1 4.9 Katsina Safana 14.2 14.0 14.3 2.7 Katsina Sandamu 5.5 1.4 19.3 323.3 Katsina Zango 5.6 1.2 21.2 358.4 Kebbi Aleiro 24.8 24.4 25.3 3.7 Kebbi Arewa Dandi 12.7 4.1 33.2 229.1 Kebbi Argungu 1.9 1.9 2.0 7.8 Kebbi Augie 15.2 7.3 29.1 143.5 Kebbi Bagudu 14.4 4.4 38.0 232.4 Kebbi Birnin Kebbi 15.9 7.6 30.6 144.6 Kebbi Bunza 18.3 18.0 18.7 3.9 Kebbi Dandi 30.9 30.5 31.3 2.7 Kebbi Wasagu-Danko 15.6 5.2 38.3 212.5 Kebbi Fakai 12.2 4.0 32.1 230.7 Kebbi Gwandu 23.5 15.8 33.4 74.8 Kebbi Jega 19.7 19.2 20.2 5.1 Kebbi Kalgo 14.1 4.0 39.3 249.5 Kebbi Koko-Besse 6.1 5.9 6.3 6.8 Kebbi Maiyama 13.8 4.5 35.2 222.9 Kebbi Ngaski 19.5 19.2 19.8 3.1 Kebbi Sakaba 33.3 32.5 34.1 4.7 Kebbi Shanga 3.5 3.3 3.7 11.4 Kebbi Suru 14.2 14.0 14.3 1.9

152

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Kebbi Yauri 12.8 4.0 34.3 237.2 Kebbi Zuru 11.7 7.6 17.7 85.3 Kogi Adavi 38.8 38.2 39.3 2.9 Kogi Ajaokuta 33.9 33.1 34.7 4.8 Kogi Ankpa 19.5 17.7 21.5 19.1 Kogi Bassa 32.8 32.0 33.5 4.6 Kogi Dekina 28.8 27.7 30.0 7.9 Kogi Ibaji 27.1 26.7 27.5 2.9 Kogi Idah 30.6 9.5 64.9 180.6 Kogi Igalamela-Odolu 32.7 31.8 33.7 5.6 Kogi Ijumu 39.0 30.1 48.7 47.7 Kogi Kabba/Bunu 54.2 53.6 54.9 2.4 Kogi Koton-Karfe 17.9 17.6 18.2 3.1 Kogi Lokoja 41.1 40.1 42.2 5.2 Kogi Mopa-Muro 36.4 11.8 71.0 162.9 Kogi Ofu 27.8 10.4 56.0 164.3 Kogi Ogori/Magongo 33.2 8.7 72.1 191.2 Kogi Okehi 23.1 20.9 25.5 19.9 Kogi Okene 32.6 12.0 63.0 156.7 Kogi Olamaboro 18.5 14.8 23.0 44.1 Kogi Omala 17.5 17.1 18.0 5.6 Kogi Yagba East 34.3 13.4 63.8 147.0 Kogi Yagba West 32.6 12.6 61.9 151.4 Kwara Asa 33.1 32.4 33.7 4.0 Kwara Baruten 17.4 17.1 17.7 3.6 Kwara Edu 31.1 25.3 37.5 39.2 Kwara Ekiti 33.6 12.5 64.1 153.6 Kwara Ifelodun 41.1 39.5 42.8 8.0 Kwara Ilorin East 46.4 45.6 47.2 3.3 Kwara Ilorin South 36.6 13.1 68.9 152.5 Kwara Ilorin West 44.5 43.8 45.2 3.1 Kwara Irepodun 28.2 27.8 28.5 2.5 Kwara Isin 33.6 10.6 68.3 171.7 Kwara Kaiama 23.3 8.4 50.4 180.1

153

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Kwara Moro 26.0 25.6 26.5 3.7 Kwara Offa 37.1 36.7 37.6 2.4 Kwara Oke Ero 33.4 12.4 63.8 154.0 Kwara Oyun 36.9 36.1 37.6 4.2 Kwara Pategi 14.2 14.0 14.4 3.0 Lagos Agege 42.4 14.8 75.8 144.0 Lagos Ajeromi/Ifelodun 45.6 16.5 78.0 134.9 Lagos Alimosho 42.9 34.0 52.2 42.4 Lagos Amuwo Odofin 45.1 44.4 45.7 2.9 Lagos Apapa 47.3 46.7 47.9 2.5 Lagos Badagry 36.0 35.3 36.6 3.7 Lagos Epe 23.1 22.0 24.2 9.5 Lagos Eti Osa 48.2 47.0 49.3 4.9 Lagos Ibeju Lekki 48.7 47.9 49.4 3.0 Lagos Ifako/Ijaye 50.0 37.0 63.0 51.8 Lagos Ikeja 41.3 40.8 41.7 2.2 Lagos Ikorodu 46.3 45.7 46.9 2.8 Lagos Kosofe 49.4 49.2 49.5 0.5 Lagos Lagos Island 47.0 46.4 47.7 2.9 Lagos Lagos Mainland 58.1 57.6 58.6 1.7 Lagos Mushin 56.6 56.1 57.2 2.0 Lagos Ojo 38.6 38.0 39.1 2.9 Lagos Oshodi/Isolo 44.3 43.7 45.0 3.0 Lagos Shomolu 52.8 51.7 53.9 4.2 Lagos Surulere 42.8 42.1 43.4 2.9 Nasarawa Akwanga 34.0 20.7 50.3 87.3 Nasarawa Awe 21.1 7.5 46.7 185.9 Nasarawa Doma 24.7 24.3 25.2 3.8 Nasarawa Karu 28.0 27.8 28.3 1.8 Nasarawa Keana 20.2 6.1 49.4 214.6 Nasarawa Keffi 47.6 46.6 48.7 4.3 Nasarawa Kokona 24.0 23.5 24.6 4.7 Nasarawa Lafia 22.8 13.5 35.8 97.9 Nasarawa Nasarawa 19.0 18.5 19.4 4.9

154

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Nasarawa Nasarawa Egon 24.5 9.7 49.6 163.0 Nasarawa Obi 11.8 11.4 12.2 6.6 Nasarawa Toto 26.5 8.7 57.8 185.0 Nasarawa Wamba 25.5 24.8 26.2 5.5 Niger Agaie 28.8 17.6 43.4 89.4 Niger Agwara 14.7 3.8 42.6 263.3 Niger Bida 39.8 39.4 40.3 2.3 Niger Borgu 12.9 12.8 13.1 2.2 Niger Bosso 35.6 35.0 36.2 3.4 Niger Chanchaga 29.8 5.2 76.7 240.4 Niger Edati 28.6 7.5 66.5 206.4 Niger Gbako 18.1 17.8 18.5 3.6 Niger Gurara 26.2 25.7 26.7 3.9 Niger Katcha 14.1 13.6 14.6 7.5 Niger Kontagora 28.5 17.9 42.1 84.7 Niger Lapai 29.9 11.1 59.4 161.4 Niger Lavun 23.2 22.9 23.4 1.9 Niger Magama 18.3 6.1 43.8 205.8 Niger Mariga 19.3 6.9 43.5 188.8 Niger Mashegu 23.1 8.6 49.1 175.0 Niger Mokwa 47.6 46.9 48.3 2.9 Niger Munya 28.4 9.4 60.3 179.1 Niger Paikoro 44.9 44.3 45.5 2.8 Niger Rafi 28.0 27.5 28.6 4.0 Niger Rijau 11.1 7.8 15.6 70.6 Niger Shiroro 22.8 22.0 23.6 7.0 Niger Suleja 36.9 13.9 68.0 146.6 Niger Tafa 61.0 60.3 61.6 2.0 Niger Wushishi 30.8 30.4 31.2 2.6 Ogun Abeokuta South 35.7 34.9 36.4 4.2 Ogun Abeokuta North 31.8 12.1 61.2 154.5 Ogun Ado Odo/Ota 32.6 32.0 33.3 4.2 Ogun Yewa North 28.4 27.9 28.9 3.6 Ogun Yewa South 11.6 11.1 12.1 8.7

155

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Ogun Ewekoro 33.4 32.9 33.9 3.0 Ogun Ifo 37.4 15.8 65.5 133.0 Ogun Ijebu North East 36.7 12.1 71.2 160.9 Ogun Ijebu East 45.5 45.1 45.9 1.8 Ogun Ijebu North 35.8 14.4 64.9 140.8 Ogun Ijebu Ode 42.5 41.9 43.2 3.0 Ogun Ikenne 39.8 39.2 40.3 2.6 Ogun Imeko Afon 21.0 20.5 21.6 5.1 Ogun Ipokia 25.4 25.1 25.7 2.6 Ogun Obafemi Owode 47.1 46.3 47.9 3.3 Ogun Odeda 34.5 13.4 64.2 147.4 Ogun Odogbolu 40.3 39.7 40.8 2.7 Ogun Ogun Waterside 29.8 10.1 61.7 172.9 Ogun Remo North 37.5 14.2 68.5 144.6 Ogun Shagamu 37.1 30.2 44.5 38.5 Ondo Akoko North East 30.5 30.0 31.1 3.5 Ondo Akoko South East 34.2 12.0 66.5 159.6 Ondo Akoko South West 30.7 30.5 30.9 1.4 Ondo Akoko North West 27.6 24.2 31.4 26.0 Ondo Akure North 49.7 43.7 55.8 24.4 Ondo Akure South 46.6 45.7 47.4 3.6 Ondo Ese Odo 27.6 9.0 59.4 182.5 Ondo Idanre 31.9 31.5 32.3 2.6 Ondo Ifedore 11.8 11.5 12.3 6.7 Ondo Ilaje 21.5 20.9 22.2 6.1 Ondo Ile Oluji/Okeigbo 45.3 38.1 52.7 32.3 Ondo Irele 19.9 19.5 20.4 4.5 Ondo Odigbo 17.6 17.1 18.2 5.8 Ondo Okitipupa 14.9 14.2 15.5 8.7 Ondo Ondo East 30.3 10.9 60.8 164.8 Ondo Ondo West 39.0 38.4 39.7 3.1 Ondo Ose 46.3 45.2 47.3 4.6 Ondo Owo 33.1 32.5 33.7 3.4 Osun Atakunmosa East 28.6 28.3 29.0 2.5

156

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Osun Atakunmosa West 28.6 11.4 55.8 155.1 Osun Ayedade 28.4 27.9 28.9 3.7 Osun Ayedire 30.0 11.3 59.2 159.6 Osun Boluwaduro 26.0 25.4 26.6 4.9 Osun Boripe 32.3 11.6 63.4 160.2 Osun Ede North 27.6 8.9 59.8 184.6 Osun Ede South 28.2 10.1 57.8 169.4 Osun Egbedore 13.8 13.3 14.3 7.2 Osun Ejigbo 26.0 22.4 29.9 28.7 Osun Ife East 29.7 9.5 62.7 179.3 Osun Ife North 36.7 29.5 44.6 41.2 Osun Ife South 22.8 22.4 23.2 3.6 Osun Ife Central 30.2 9.0 65.6 187.3 Osun Ifedayo 39.6 39.0 40.1 2.7 Osun Ifelodun 50.7 50.3 51.2 1.8 Osun Ila 30.8 11.8 59.8 155.5 Osun Ilesha East 27.3 26.6 28.0 5.0 Osun Ilesha West 26.5 8.6 57.9 186.0 Osun Irepodun 34.5 33.9 35.1 3.5 Osun Irewole 21.1 20.6 21.7 5.5 Osun Isokan 30.2 9.2 64.8 184.3 Osun Iwo 32.9 32.1 33.6 4.5 Osun Obokun 25.7 25.0 26.5 5.8 Osun Odo Otin 33.7 13.2 62.9 147.7 Osun Ola Oluwa 30.8 10.3 63.4 172.1 Osun Olorunda 48.4 48.0 48.9 1.9 Osun Oriade 21.3 20.7 22.1 6.5 Osun Orolu 24.7 24.2 25.3 4.8 Osun Osogbo 29.2 10.6 59.0 166.1 Oyo Afijio 32.3 12.9 60.6 147.8 Oyo Akinyele 50.8 50.1 51.6 2.9 Oyo Atiba 26.8 9.9 54.9 168.4 Oyo Atisbo 22.6 22.1 23.0 4.0 Oyo Egbeda 29.8 29.1 30.4 4.4

157

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Oyo Ibadan North East 39.3 39.0 39.6 1.7 Oyo Ibadan North West 38.3 37.8 38.8 2.5 Oyo Ibadan North 36.6 13.1 68.9 152.4 Oyo Ibadan South East 29.5 28.4 30.6 7.5 Oyo Ibadan South West 36.2 13.5 67.3 148.6 Oyo Ibarapa Central 29.7 10.2 61.0 171.0 Oyo Ibarapa East 31.8 11.3 63.1 162.7 Oyo Ibarapa North 20.7 20.0 21.5 7.6 Oyo Ido 47.9 47.4 48.4 2.0 Oyo Irepo 14.6 14.1 15.2 7.6 Oyo Iseyin 39.2 38.7 39.8 3.0 Oyo Itesiwaju 27.1 9.5 56.7 174.2 Oyo Iwajowa 26.2 9.2 55.6 176.6 Oyo Kajola 20.5 11.2 34.6 113.7 Oyo Lagelu 32.2 31.7 32.7 3.2 Oyo Ogbomosho North 33.3 32.7 33.9 3.5 Oyo Ogbomosho South 30.8 10.1 63.8 174.3 Oyo Ogo Oluwa 29.5 10.9 58.9 162.5 Oyo Olorunsogo 26.2 9.5 54.6 171.9 Oyo Oluyole 35.9 14.5 64.9 140.5 Oyo Ona ara 33.1 12.4 63.4 153.9 Oyo Orelope 23.1 7.6 52.5 194.2 Oyo Ori Ire 29.9 11.3 58.9 159.1 Oyo Oyo East 29.7 10.3 60.8 170.0 Oyo Oyo West 31.4 30.6 32.3 5.5 Oyo Saki East 25.8 18.5 34.6 62.4 Oyo Saki West 22.8 6.5 55.7 216.2 Oyo Surulere1 31.8 12.6 60.2 149.6 Plateau Barkin Ladi 24.6 8.5 53.1 181.5 Plateau Bassa1 52.0 51.3 52.7 2.6 Plateau Bokkos 25.1 8.8 53.9 179.6 Plateau Jos East 26.4 9.3 55.3 174.5 Plateau Jos North 47.5 46.7 48.3 3.3 Plateau Jos South 49.9 49.6 50.2 1.3

158

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Plateau Kanam 20.3 11.9 32.5 101.2 Plateau Kanke 14.7 14.4 15.0 3.8 Plateau Langtang North 23.6 14.4 36.2 92.2 Plateau Langtang South 15.2 14.8 15.6 5.2 Plateau Mangu 38.0 30.9 45.6 38.8 Plateau Mikang 15.3 15.0 15.6 3.6 Plateau Pankshin 19.9 6.9 45.5 193.9 Plateau Qua'an Pan 22.1 7.5 49.6 190.5 Plateau Riyom 2.0 1.8 2.1 14.0 Plateau Shendam 11.6 11.3 11.9 4.5 Plateau Wase 22.5 21.8 23.3 6.5 Rivers Abua/Odua 31.8 31.5 32.1 2.1 Rivers Ahoada East 25.0 24.7 25.4 2.6 Rivers Ahoada West 24.8 24.3 25.3 4.1 Rivers Akuku Toru 38.0 12.7 72.0 156.1 Rivers Andoni 34.2 11.0 68.7 168.5 Rivers Asari-Toru 39.2 11.5 76.3 165.0 Rivers Bonny 52.0 45.4 58.6 25.4 Rivers Degema 37.9 37.2 38.5 3.5 Rivers Eleme 49.3 48.4 50.1 3.4 Rivers Emuoha 44.8 43.6 46.0 5.4 Rivers Etche 34.6 13.6 64.1 146.0 Rivers Gokana 30.5 29.8 31.2 4.4 Rivers Ikwerre 37.3 13.4 69.5 150.4 Rivers Khana 36.9 35.9 37.9 5.4 Rivers Obio/Akpor 61.5 60.8 62.2 2.3 Rivers Ogba/Egbema/Ndon 32.2 13.0 60.3 146.8 i Rivers Ogu Bolo 36.5 13.6 67.8 148.5 Rivers Okrika 41.7 16.3 72.3 134.4 Rivers Omumma 18.5 17.7 19.4 9.2 Rivers Opobo/Nkoro 32.5 31.9 33.1 4.0 Rivers Oyigbo 44.6 43.7 45.5 4.1 Rivers Port-Harcourt 54.6 53.9 55.3 2.5 Rivers Tai 13.2 12.6 13.8 8.8

159

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Sokoto Binji 8.6 8.3 8.9 6.3 Sokoto Bodinga 11.4 3.4 32.2 251.6 Sokoto Dange Shuni 5.2 5.0 5.3 5.2 Sokoto Gada 10.1 9.8 10.5 6.0 Sokoto Goronyo 11.9 3.7 32.3 241.2 Sokoto Gudu 10.8 3.0 31.9 266.8 Sokoto Gwadabawa 6.8 6.6 6.9 5.0 Sokoto Illela 16.0 15.7 16.4 4.1 Sokoto Isa 13.3 13.0 13.6 5.1 Sokoto Kebbe 12.5 4.0 33.3 234.1 Sokoto Kware 12.7 4.2 32.5 222.1 Sokoto Rabah 12.2 4.0 31.9 228.9 Sokoto Sabon Birni 12.5 3.2 38.2 280.0 Sokoto Shagari 11.5 3.4 32.4 251.8 Sokoto Silame 4.4 4.2 4.5 7.6 Sokoto Sokoto North 39.0 38.6 39.3 1.8 Sokoto Sokoto South 14.4 4.2 39.5 245.4 Sokoto Tambuwal 7.9 7.8 8.1 4.5 Sokoto Tangaza 7.9 7.6 8.2 7.3 Sokoto Tureta 12.5 3.9 33.4 236.5 Sokoto Wamakko 22.4 16.2 30.2 62.4 Sokoto Wurno 11.5 11.4 11.7 2.2 Sokoto Yabo 8.0 7.9 8.2 4.0 Taraba Ardo-Kola 28.1 14.9 46.6 113.0 Taraba Bali 17.0 16.8 17.3 3.0 Taraba Donga 23.7 23.2 24.2 4.2 Taraba Gashaka 25.4 7.8 58.0 197.4 Taraba Gassol 21.0 20.4 21.5 5.0 Taraba Ibi 34.0 33.8 34.2 1.3 Taraba Jalingo 20.4 5.7 52.1 227.1 Taraba Karim Lamido 20.3 18.6 22.2 17.7 Taraba Kurmi 26.6 9.0 57.1 181.2 Taraba Lau 16.9 16.4 17.4 6.1 Taraba Sardauna 36.8 34.2 39.5 14.2

160

Table 8: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Nigeria States Local Government % Lower Upper % Relative Areas Overweight 95% CI 95% CI Uncertainity Prevalence Taraba Takum 43.8 43.6 44.0 0.9 Taraba Ussa 17.1 16.8 17.4 3.3 Taraba Wukari 34.1 33.6 34.5 2.6 Taraba Yorro 22.5 22.2 22.9 2.8 Taraba Zing 4.5 4.4 4.7 7.2 Yobe Bade 15.7 4.6 42.0 238.8 Yobe Borsari 16.3 15.8 16.8 6.4 Yobe Damaturu 14.0 4.2 37.3 237.0 Yobe Fika 12.1 4.0 31.6 227.6 Yobe Fune 8.1 7.9 8.2 3.8 Yobe Geidam 13.9 4.3 36.3 231.0 Yobe Gujba 21.2 12.6 33.6 98.8 Yobe Gulani 13.8 4.6 34.5 216.9 Yobe Jakusko 12.8 5.5 27.1 168.4 Yobe Karasuwa 33.8 30.8 37.1 18.6 Yobe Machina 15.3 4.4 41.5 242.0 Yobe Nangere 10.8 3.3 30.0 247.6 Yobe Nguru 26.7 17.4 38.6 79.4 Yobe Potiskum 11.3 2.8 35.8 290.2 Yobe Tarmua 13.5 4.3 35.4 230.2 Yobe Yunusari 15.6 4.5 41.7 239.3 Yobe Yusufari 36.4 36.2 36.6 1.0 Zamfara Anka 14.6 4.5 38.4 232.2 Zamfara Bakura 16.0 15.5 16.4 5.4 Zamfara Birnin Magaji- 10.8 10.6 10.9 3.4 Kiyaw Zamfara Bukkuyum 20.2 19.8 20.5 3.5 Zamfara Bungudu 15.2 14.6 15.8 7.7 Zamfara Gummi 13.0 4.3 33.3 222.0 Zamfara Gusau 20.4 11.7 33.2 105.7 Zamfara Kaura Namoda 13.9 7.3 24.8 125.7 Zamfara Maradun 13.2 4.5 32.9 215.0 Zamfara Maru 11.9 9.5 14.7 43.5 Zamfara Shinkafi 13.4 13.0 13.8 5.7 Zamfara Talata Mafara 14.1 4.1 38.9 247.4

161

162

163

Tanzania

Table 9: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Tanzania Country Regions % Overweight Lower 95% Upper 95% CI Prevalence CI Tanzania Arusha 34.3 23.3 45.3 Dar es 49.9 49.2 50.6 Salaam Dodoma 23.1 20.3 25.8 Geita 21.2 18.0 24.3 Iringa 31.3 8.9 53.7 Kagera 14.2 9.8 18.7 Katavi 22.0 13.3 30.6 Kigoma 19.7 6.7 32.8 Kilimanjaro 46.8 39.6 54.0 Lindi 28.9 21.3 36.5 Manyara 23.1 19.9 26.2 Mara 16.8 12.0 21.6 Mbeya 39.2 21.6 56.8 Morogoro 38.1 30.9 45.3 Mtwara 31.7 16.9 46.5 Mwanza 24.9 20.4 29.3 Njombe 29.9 12.5 47.3 Pemba 30.7 22.7 38.7 North Pemba 39.9 31.3 48.4 South Pwani 39.9 29.9 49.8 Rukwa 22.3 9.4 35.1 Ruvuma 22.5 9.8 35.2 Shinyanga 27.0 13.7 40.4 Simiyu 16.6 15.3 17.9 Singida 24.4 17.6 31.1 Tabora 27.8 16.9 38.7 Tanga 41.0 32.1 49.9 Zanzibar 40.7 39.1 42.3 North Zanzibar 44.7 44.1 45.3 South and Central

164

165

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Arusha Arusha 34.9 24.1 47.6 67.4 Arusha Arusha Urban 38.8 38.1 39.6 3.8 Arusha Karatu 40.1 39.5 40.7 3.0 Arusha Longido 8.7 8.3 9.1 8.9 Arusha Meru 39.6 38.5 40.7 5.6 Arusha Monduli 57.3 52.6 61.8 16.2 Arusha Ngorongoro 21.7 9.1 43.3 157.4 Dar es Ilala 53.9 53.2 54.6 2.8 Salaam Dar es Kinondoni 46.2 45.4 47.0 3.5 Salaam Dar es Temeke 50.8 50.1 51.5 2.8 Salaam Dodoma Bahi 33.8 32.3 35.4 9.0 Dodoma Chamwino 18.6 12.1 27.4 82.5 Dodoma Chemba 23.1 22.4 23.8 6.0 Dodoma Dodoma Urban 17.2 12.2 23.8 67.7 Dodoma Kondoa 26.1 17.3 37.4 77.0 Dodoma Kongwa 28.6 27.2 30.0 9.8 Dodoma Mpwapwa 20.6 19.9 21.3 6.4 Geita Bukombe 22.7 11.3 40.5 128.5 Geita Chato 19.9 19.5 20.4 4.2 Geita Geita 21.9 21.5 22.4 3.9 Geita Mbogwe 27.4 18.9 38.0 69.6 Geita Nyang'wale 22.4 6.7 53.7 209.8 Iringa Iringa Rural 14.4 14.0 14.8 5.5 Iringa Iringa Urban 53.7 53.2 54.1 1.7 Iringa Kilolo 35.7 23.7 49.8 73.1 Iringa Mafinga Township 42.8 41.8 43.8 4.7 Authority Iringa Mufindi 16.2 8.7 28.2 120.6 Kagera Biharamulo 2.7 2.6 2.8 6.3 Kagera Bukoba Rural 18.8 17.3 20.3 15.6 Kagera Bukoba Urban 36.6 36.0 37.3 3.5 Kagera Karagwe 11.1 10.9 11.3 4.0 Kagera Kyerwa 13.2 12.8 13.6 6.5

166

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Kagera Missenyi 19.1 18.6 19.6 5.0 Kagera Muleba 14.7 12.1 17.7 38.6 Kagera Ngara 8.4 8.0 8.8 9.9 Katavi Mlele 17.9 17.5 18.3 4.4 Katavi Mpanda 21.6 15.3 29.5 65.8 Katavi Mpanda Urban 32.8 27.8 38.3 32.2 Kigoma Buhigwe 21.6 20.3 23.0 12.4 Kigoma Kakonko 2.7 2.6 2.9 11.5 Kigoma Kasulu 22.2 20.9 23.4 11.4 Kigoma Kasulu Township 28.1 19.9 38.2 65.1 Authority Kigoma Kibondo 12.8 6.8 22.9 125.2 Kigoma Kigoma Rural 20.6 20.2 21.1 4.4 Kigoma Kigoma Urban 47.2 46.3 48.1 3.9 Kigoma Uvinza 11.0 10.6 11.4 6.6 Kilimanjaro Hai 44.6 42.6 46.6 9.0 Kilimanjaro Moshi Rural 51.9 50.7 53.1 4.5 Kilimanjaro Moshi Urban 55.5 54.8 56.2 2.6 Kilimanjaro Mwanga 36.2 35.6 36.7 2.8 Kilimanjaro Rombo 29.1 22.7 36.4 46.9 Kilimanjaro Same 49.6 39.2 60.0 42.0 Kilimanjaro Siha 53.4 51.6 55.2 6.8 Lindi Kilwa 24.5 22.0 27.2 21.0 Lindi Lindi Rural 28.7 28.1 29.3 4.1 Lindi Lindi Urban 60.8 59.9 61.8 3.0 Lindi Liwale 30.8 30.4 31.2 2.5 Lindi Nachingwea 32.2 29.9 34.5 14.4 Lindi Ruangwa 19.7 11.0 32.7 110.0 Manyara Babati 33.6 33.3 34.0 2.0 Manyara Babati Urban 47.2 46.7 47.8 2.3 Manyara Hanang 24.5 23.1 26.0 11.9 Manyara Kiteto 18.8 18.3 19.2 4.8 Manyara Mbulu 13.9 11.2 17.1 42.9 Manyara Simanjiro 10.7 10.3 11.0 6.9 Mara Bunda 20.1 13.4 29.2 78.4

167

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Mara Butiama 22.4 21.5 23.3 8.0 Mara Musoma Rural 24.6 24.0 25.3 5.4 Mara Musoma Urban 31.0 30.3 31.8 4.8 Mara Rorya 3.3 3.2 3.4 5.6 Mara Serengeti 2.2 2.1 2.3 11.0 Mara Tarime 13.5 8.8 20.3 84.4 Mbeya Chunya 38.4 33.1 44.1 28.6 Mbeya Ileje 26.8 26.1 27.4 4.9 Mbeya Kyela 51.4 43.7 58.9 29.6 Mbeya Mbarali 26.1 24.8 27.6 10.8 Mbeya Mbeya Rural 16.6 15.9 17.3 8.1 Mbeya Mbeya Urban 64.5 63.3 65.8 3.9 Mbeya Mbozi 32.8 20.2 48.4 85.8 Mbeya Momba 28.1 12.1 52.5 143.7 Mbeya Rungwe 55.7 54.5 56.8 4.1 Mbeya Tunduma 29.7 8.1 66.8 197.9 Morogoro Gairo 42.7 33.1 52.9 46.4 Morogoro Kilombero 35.6 32.1 39.4 20.5 Morogoro Kilosa 27.1 25.6 28.7 11.4 Morogoro Morogoro Rural 33.7 25.8 42.6 49.8 Morogoro Morogoro Urban 57.4 53.5 61.2 13.4 Morogoro Mvomero 30.2 29.9 30.6 2.6 Morogoro Ulanga 16.7 15.8 17.8 11.9 Mtwara Masasi 30.9 30.4 31.4 3.4 Mtwara Masasi Township 31.0 12.3 59.1 151.0 Authority Mtwara Mtwara Rural 23.5 23.2 23.9 3.0 Mtwara Mtwara Urban 47.7 47.0 48.4 2.9 Mtwara Nanyumbu 25.2 24.6 25.8 4.6 Mtwara Newala 32.4 18.9 49.5 94.5 Mtwara Tandahimba 29.8 23.7 36.8 43.9 Mwanza Ilemela 49.4 48.7 50.2 2.9 Mwanza Kwimba 18.0 12.0 26.0 77.7 Mwanza Magu 21.3 21.0 21.5 2.2 Mwanza Misungwi 13.6 13.2 14.0 6.2

168

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Mwanza Nyamagana 18.2 17.9 18.6 4.2 Mwanza Sengerema 28.3 28.0 28.6 2.3 Mwanza Ukerewe 21.6 14.1 31.6 81.4 Njombe Ludewa 31.5 19.6 46.5 85.3 Njombe Makambako 31.5 19.0 47.6 90.7 Township Authority Njombe Makete 26.6 26.3 26.9 2.5 Njombe Njombe 7.0 6.7 7.3 8.6 Njombe Njombe Urban 46.7 31.5 62.6 66.6 Njombe Wanging'ombe 22.0 21.9 22.2 1.5 Pemba Micheweni 17.8 17.4 18.2 4.7 North Pemba Wete 40.7 36.3 45.2 22.0 North Pemba Chake 38.0 24.3 54.0 77.9 South Pemba Mkoani 39.0 36.3 41.8 14.0 South Pwani Bagamoyo 45.7 42.0 49.5 16.5 Pwani Kibaha 38.8 32.6 45.4 32.8 Pwani Kibaha Urban 59.6 59.2 60.0 1.4 Pwani Kisarawe 21.2 20.1 22.5 11.3 Pwani Mafia 72.4 71.7 73.2 2.2 Pwani Mkuranga 17.4 15.2 19.8 26.7 Pwani Rufiji 39.5 26.7 53.9 69.0 Rukwa Kalambo 14.7 14.5 14.9 3.0 Rukwa Nkasi 29.9 27.8 32.0 14.2 Rukwa Sumbawanga Rural 17.7 8.2 34.0 145.5 Rukwa Sumbawanga Urban 29.7 18.0 44.9 90.3 Ruvuma Mbinga 16.4 9.7 26.4 101.3 Ruvuma Namtumbo 29.2 20.8 39.2 63.3 Ruvuma Nyasa 27.2 7.8 62.2 200.2 Ruvuma Songea Rural 13.1 12.8 13.4 4.7 Ruvuma Songea Urban 41.5 40.4 42.5 5.0 Ruvuma Tunduru 18.5 12.4 26.9 78.2 Shinyanga Kahama 19.5 17.6 21.5 20.2

169

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Shinyanga Kahama Township 44.5 38.2 51.0 29.0 Authority Shinyanga Kishapu 16.2 15.7 16.8 7.0 Shinyanga Shinyanga Rural 18.4 18.1 18.6 2.8 Shinyanga Shinyanga Urban 40.7 39.4 42.0 6.2 Simiyu Bariadi 18.0 17.8 18.2 2.6 Simiyu Busega 14.6 5.9 31.8 177.8 Simiyu Itilima 15.2 11.6 19.7 53.0 Simiyu Maswa 17.7 17.3 18.0 4.3 Simiyu Meatu 15.5 14.6 16.4 12.1 Singida Ikungi 20.6 18.2 23.3 24.9 Singida Iramba 21.8 20.9 22.8 8.7 Singida Manyoni 19.5 11.5 31.2 100.7 Singida Mkalama 31.5 30.9 32.1 3.6 Singida Singida Rural 18.3 18.0 18.5 2.8 Singida Singida Urban 34.8 23.2 48.5 72.7 Tabora Igunga 33.8 24.2 45.1 61.8 Tabora Kaliua 8.5 8.0 9.1 13.5 Tabora Nzega 15.7 15.3 16.1 4.8 Tabora Sikonge 32.0 18.8 48.9 94.0 Tabora Tabora Urban 37.4 32.1 43.1 29.4 Tabora Urambo 20.5 20.1 20.8 3.4 Tabora Uyui 36.9 36.3 37.6 3.6 Tanga Handeni 26.9 26.4 27.5 4.3 Tanga Handeni Township 61.9 60.6 63.2 4.2 Authority Tanga Kilindi 35.7 34.6 36.9 6.4 Tanga Korogwe 28.8 28.5 29.2 2.4 Tanga Korogwe Township 54.8 43.9 65.2 39.0 Authority Tanga Lushoto 17.5 17.2 17.9 3.6 Tanga Mkinga 34.3 33.9 34.7 2.2 Tanga Muheza 59.0 57.9 60.0 3.5 Tanga Pangani 19.0 18.4 19.6 6.1 Tanga Tanga 46.2 45.3 47.2 4.0

170

Table 10: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Tanzania Regions Districts % Lower 95% Upper Relative Overweight CI 95% CI Uncertainty Prevalence Zanzibar Kaskazini 'A' 38.8 38.4 39.2 2.2 North Zanzibar Kaskazini 'B' 42.9 38.7 47.3 20.2 North Zanzibar Kati 48.8 46.8 50.7 7.9 South and Central Zanzibar Kusini 38.3 36.5 40.1 9.6 South and Central Zanzibar Magharibi 49.9 44.0 55.9 23.7 West Zanzibar Mjini 44.9 43.9 46.0 4.7 West

171

172

Zambia

Table 11: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older in Zambia Country Province % Overweight Lower 95% Upper 95% CI Prevalence CI Zambia Central 24.5 14.0 35.1 Copperbelt 33.2 28.0 38.5 Eastern 21.8 16.9 26.8 Luapula 13.7 8.7 18.8 Lusaka 38.5 36.4 40.5 Muchinga 15.5 7.7 23.2 North-Western 16.5 6.2 26.9 Northern 12.9 6.8 19.0 Southern 25.2 15.6 34.7 Western 11.8 3.6 19.9

173

Table 12: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Zambia Province District % Lower 95% Upper 95% % Relative Overweight CI CI Uncertainty Prevalence Central Chibombo 29.1 19.6 40.9 73.0 Central Kabwe 32.7 31.8 33.7 5.8 Central Kapiri Mposhi 19.9 15.7 24.9 46.4 Central Mkushi 22.7 21.6 23.9 10.2 Central Mumbwa 23.0 17.6 29.4 51.1 Central Serenje 13.9 8.8 21.2 88.6 Copperbelt Chililabombwe 39.9 31.7 48.7 42.6 Copperbelt Chingola 30.4 21.6 40.9 63.5 Copperbelt Kalulushi 30.0 27.8 32.2 14.8 Copperbelt Kitwe 34.3 33.9 34.7 2.3 Copperbelt Luanshya 37.8 35.2 40.6 14.2 Copperbelt Lufwanyama 26.2 25.4 27.0 6.0 Copperbelt Masaiti 15.9 15.6 16.2 4.0 Copperbelt MPongwe 28.9 26.6 31.3 16.5 Copperbelt Mufulira 28.0 26.8 29.3 8.7 Copperbelt Ndola 37.1 36.0 38.2 6.0 Eastern Chadiza 16.7 14.8 18.8 24.1 Eastern Chipata 21.8 15.6 29.7 64.8 Eastern Katete 21.1 15.2 28.5 63.5 Eastern Lundazi 16.9 13.0 21.7 51.6 Eastern Mambwe 8.5 8.4 8.7 4.0 Eastern Nyimba 29.2 28.6 29.7 3.7 Eastern Petauke 27.5 26.4 28.5 7.7 Luapula Chiengi 3.9 3.8 4.0 4.1 Luapula Kawambwa 10.8 7.5 15.1 70.7 Luapula Mansa 21.8 19.2 24.6 24.7 Luapula Milenge 14.1 9.3 20.9 81.4 Luapula Mwense 9.9 9.0 10.9 18.7 Luapula Nchelenge 11.4 6.9 18.3 100.1 Luapula Samfya 16.6 13.1 20.8 46.3 Lusaka Chongwe 31.1 30.3 31.8 4.9 Lusaka Kafue 35.0 33.6 36.4 7.9 Lusaka Luangwa 21.5 21.0 22.1 5.3 Lusaka Lusaka 39.6 38.7 40.6 4.9

174

Table 12: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Zambia Muchinga Chama 12.9 9.0 18.1 70.5 Muchinga Chinsali 13.8 9.1 20.3 81.2 Muchinga Isoka 12.5 9.1 16.9 62.0 Muchinga Mpika 16.0 11.0 22.6 72.4 Muchinga Nakonde 21.7 16.9 27.4 48.7 Northern Chilubi 13.8 13.5 14.0 3.5 Northern Kaputa 9.4 5.6 15.5 105.5 Northern Kasama 13.9 8.8 21.3 89.7 Northern Luwingu 10.7 7.9 14.3 60.1 Northern Mbala 13.8 13.1 14.5 9.9 Northern Mporokoso 17.9 17.5 18.2 4.0 Northern Mpulungu 13.5 7.3 23.6 120.1 Northern Mungwi 9.4 8.0 11.1 33.3 North- Chavuma 9.0 2.5 27.2 273.5 Western North- Kabompo 13.5 10.6 17.0 47.6 Western North- Kasempa 12.0 7.5 18.4 90.8 Western North- Mufumbwe 12.8 10.5 15.5 38.6 Western North- Mwinilunga 13.8 9.6 19.5 72.2 Western North- Solwezi 25.7 19.0 33.7 57.2 Western North- Zambezi 9.7 4.2 20.6 169.1 Western Southern Choma 22.4 15.6 31.3 70.1 Southern Gwembe 13.8 13.4 14.1 4.7 Southern Itezhi-Tezhi 22.0 21.4 22.7 6.2 Southern Kalomo 14.5 9.5 21.5 82.2 Southern Kazungula 22.7 21.8 23.6 8.1 Southern Livingstone 39.9 39.0 40.8 4.6 Southern Mazabuka 37.8 31.8 44.2 32.8 Southern Monze 22.6 21.1 24.2 13.9 Southern Namwala 16.7 15.1 18.5 20.3 Southern Siavonga 25.0 18.7 32.7 55.9 Southern Sinazongwe 18.6 12.1 27.6 83.6 Western Kalabo 10.4 8.5 12.7 40.2

175

Table 12: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older in Zambia Western Kaoma 14.2 9.0 21.8 90.2 Western Lukulu 11.5 7.3 17.6 90.2 Western Mongu 12.5 7.0 21.5 115.1 Western Senanga 7.9 4.7 12.8 101.7 Western Sesheke 14.9 10.4 20.9 70.3 Western Shangombo 6.7 6.4 7.0 9.2

176

177

Zimbabwe

Table 13: Age-Standardized Overweight Prevalence in First-Level Administrative Units among Women, 18 Years and Older Country Provinces % Overweight Lower 95% Upper 95% CI Prevalence CI Zimbabwe Bulawayo 48.6 47.9 49.3 Harare 50.6 50.0 51.3 Manicaland 35.4 27.9 43.0 Mashonaland Central 30.9 25.1 36.7 Mashonaland East 36.6 31.2 41.9 Mashonaland West 34.9 27.2 42.6 Masvingo 35.3 27.1 43.4 Matabeleland North 30.8 25.6 36.1 Matabeleland South 33.6 27.2 39.9 Midlands 36.2 27.6 44.8

178

Table 14: Age-Standardized Overweight Prevalence in Second-Level Administrative Units among Women, 18 Years and Older Province District % Lower 95% Upper % Relative Overweight CI 95% CI Uncertainty Prevalence Bulawayo Bulawayo 48.6 47.8 49.3 3.0 Harare Harare 50.6 49.9 51.3 2.8 Manicaland Buhera 24.0 23.4 24.6 4.8 Manicaland Chimanimani 33.4 33.1 33.7 1.8 Manicaland Chipinge 39.7 35.5 44.1 21.8 Manicaland Makoni 30.0 23.1 37.9 49.3 Manicaland Mutare 40.5 32.0 49.6 43.7 Manicaland Mutasa 33.6 33.4 33.8 1.3 Manicaland Nyanga 40.0 31.8 48.8 42.5 Mashonaland Bindura 37.0 24.4 51.8 74.0 Central Mashonaland Centenary 23.7 23.4 24.0 2.6 Central Mashonaland Guruve 23.6 23.1 24.1 4.1 Central Mashonaland Mazowe 36.2 35.3 37.1 5.2 Central Mashonaland Mount 30.5 30.1 30.9 2.4 Central Darwin Mashonaland Rushinga 27.5 26.4 28.6 8.0 Central Mashonaland Shamva 19.6 18.6 20.6 10.2 Central Mashonaland Chikomba 40.4 38.6 42.2 8.8 East Mashonaland Goromonzi 37.3 31.4 43.6 32.6 East Mashonaland Marondera 46.2 40.8 51.8 23.8 East Mashonaland Mudzi 36.8 31.7 42.3 28.7 East Mashonaland Murehwa 32.6 28.8 36.6 23.9 East Mashonaland Mutoko 30.2 27.2 33.3 20.5 East Mashonaland Seke 28.6 27.6 29.7 7.5 East Mashonaland UMP 31.6 30.6 32.6 6.3 East

179

Mashonaland Wedza 31.3 30.4 32.2 5.8 East Mashonaland Chegutu 40.2 38.8 41.7 7.3 West Mashonaland Hurungwe 26.6 21.5 32.4 40.9 West Mashonaland Kadoma 38.3 31.6 45.5 36.1 West Mashonaland Kariba 32.7 19.4 49.5 92.0 West Mashonaland Makonde 32.4 22.3 44.4 68.1 West Mashonaland Zvimba 43.0 40.8 45.3 10.6 West Masvingo Bikita 37.4 37.1 37.6 1.2 Masvingo Chiredzi 35.4 28.4 43.0 41.2 Masvingo Chivi 26.3 16.4 39.2 86.8 Masvingo Gutu 35.3 34.2 36.3 6.0 Masvingo Masvingo 40.3 30.8 50.7 49.4 Masvingo Mwenezi 36.9 31.1 43.1 32.5 Masvingo Zaka 26.1 25.4 26.8 5.3 Matabeleland Binga 19.1 14.0 25.3 59.4 North Matabeleland Bubi 43.3 42.3 44.3 4.4 North Matabeleland Hwange 37.1 29.7 45.2 41.5 North Matabeleland Lupane 39.7 37.8 41.6 9.7 North Matabeleland Nkayi 24.5 24.1 24.9 3.3 North Matabeleland Tsholotsho 29.6 28.9 30.3 4.9 North Matabeleland Umguza 30.9 24.9 37.6 41.1 North Matabeleland Beitbridge 40.5 31.1 50.7 48.6 South Matabeleland Bulilima 24.6 23.7 25.4 6.7 South (North) Matabeleland Gwanda 36.4 35.2 37.7 6.9 South Matabeleland Insiza 23.1 21.9 24.3 10.2 South

180

Matabeleland Mangwe 30.9 23.3 39.8 53.2 South (South) Matabeleland Matobo 34.8 34.3 35.3 2.9 South Matabeleland Umzingwane 40.9 40.3 41.6 3.1 South Midlands Chirumhanzu 4.5 4.3 4.8 12.1 Midlands Gokwe North 25.7 24.8 26.6 6.8 Midlands Gokwe South 26.8 24.9 28.9 14.7 Midlands Gweru 50.6 46.4 54.9 16.7 Midlands Kwekwe 36.0 28.1 44.9 46.6 Midlands Mberengwa 24.6 24.2 25.0 3.4 Midlands Shurugwi 46.3 37.5 55.4 38.6 Midlands Zvishavane 50.0 48.3 51.7 6.7

181

182

CHAPTER 5

THE EFFECT MODIFICATION OF EDUCATIONAL ATTAINMENT ON THE

ASSOCIATION BETWEEN HOUSEHOLD WEALTH AND FEMALE OVERWEIGHT IN

LOW AND MIDDLE-INCOME SUB-SAHARAN AFRICAN COUNTRIES: A CROSS-

SECTIONAL STUDY

Ifeoma D. Ozodiegwu, DrPH1, Henry V. Doctor, PhD2, Megan Quinn, DrPH1, Laina D. Mercer,

PhD3, Hadii M. Mamudu, PhD4

1Department of Biostatistics and Epidemiology, East Tennessee State University, Johnson City,

Tennessee, United States of America

2Department of Information, Evidence and Research, World Health Organization, Regional

Office for the Eastern Mediterranean, Cairo, Egypt

3Center for Vaccine Innovation and Access, Program for Appropriate Technology in Health

(PATH), Seattle, Washington, United States of America

4Department of Health Services Management and Policy, East Tennessee State University

Johnson City, Tennessee, United States of America

Emails

HVD: [email protected]

MQ: [email protected]

LDM: [email protected]

HMM: [email protected]

Corresponding author: Ifeoma D. Ozodiegwu

Email: [email protected]

183

Abstract

Background

There is limited evidence on the interrelationship of socioeconomic factors with overweight in several sub-Saharan African (SSA) countries. This study examined the interaction of education on the relationship between household wealth and adult female overweight in 22 SSA countries.

Methods

We used recent Demographic Health and Health Surveys data from women 18 years and older in

SSA countries. Overweight was defined as a body mass index of ≥ 25kg/m2. Household wealth index tertiles was the exposure and educational attainment, the effect modifier. Potential confounders included place of residence, ethnicity, age and parity. Descriptive analysis was conducted in 34 countries to identify those with sufficient sample size for subgroup analysis.

Consequently, logistic regression models were fitted for 22 SSA countries. Missing data was addressed with multiple imputation and results were adjusted for multiple comparisons error.

Findings were presented based in tables and charts by country income status.

Results

Overweight prevalence varied by country with the lowest levels observed in Madagascar and the highest in Swaziland. Among those with secondary or higher education, the richest tertile tended to bear the greatest overweight burden. We did not find evidence of the multiplicative effect modification of education on the association between household wealth and overweight in either low or middle income SSA countries examined in this study. Our conclusions remained

184

unchanged after sensitivity analysis for upper middle-income SSA countries with obesity as the outcome.

Conclusions

While it is possible that educational attainment may influence the impact of wealth status on overweight or obesity in SSA countries, we were unable to detect this using our study data. The strength of this study is the use of a nationally representative dataset for the analysis. Further research to understand conditions under which the relationship of household wealth index with overweight depends on educational attainment is warranted.

Keywords: Overweight, sub-Saharan Africa, socioeconomic status, interaction.

185

1 Background

2 Among women, overweight and obesity has been shown to have a positive association with

3 socioeconomic status (SES) in low-income countries and vice-versa in middle income countries

4 (1). Nonetheless, the interrelationships of SES-related factors remain understudied in both low

5 and middle sub-Saharan African (SSA) countries. Although available evidence suggest an

6 absence of multiplicative effect modification of educational attainment on the positive

7 relationship between household wealth and obesity seen in low income countries with the reverse

8 in middle income countries, the conclusions of that study are based solely on data from Nigeria

9 and Benin (2). The study conclusions assume homogeneity in the examined associations for

10 numerous African countries and provides little insight for in-country stakeholders across the

11 African continent. Additionally, given the low prevalence of obesity in both countries, detection

12 of an effect modification in a multiplicative model may be implausible (3).

14 The availability of recent Demographic and Health Survey (DHS) data for the formerly studied

15 countries and other SSA countries serves as impetus for a re-examination of these relationships

16 to monitor determinants, identify vulnerable groups and develop optimal interventions for

17 individual countries. However, unlike Aitsi-Selmi et al. (2012) & (2014) that defined obesity (≤

18 30kg/m2) as the primary outcome, overweight (≤ 25kg/m2) was used as the primary outcome in

19 this study to ensure that the sub-group analyses that form part of this interaction study is

20 adequately powered. The inclusion of overweight participants is considered to be important for

21 this study because, contrary to the notion that overweight is protective, several recent rigorous

22 studies indicate that being overweight is associated with an increased risk of cardiovascular

23 morbidity and mortality, hence, debunking the notion of an obesity paradox (5–7). Moreover,

186

24 SSA countries are experiencing an uptick in overweight with the greatest burden borne by adult

25 women (8). For instance, data from the World Health Organization Global Health Observatory

26 revealed that from 1975 to 2016 age-standardized overweight prevalence among SSA women

27 rose by 24% to 39% versus 15% to 23% in men (8). Additionally, a few African countries fall

28 in the lower-middle income and upper middle-income category (9), providing an opportunity to

29 compare findings with low-income countries.

31 Evidence of the negative effect modification of education on the positive association between

32 household wealth and adult female overweight will support the case for the development of

33 health promotion interventions to decrease overweight and the case for the health benefits of

34 education in SSA in general, which could lead to continued funding and expansion of

35 educational opportunities for women and girls. Currently, international aid for education has

36 stagnated since 2010 and it estimated that by 2030 many low and lower middle incomes will face

37 an education financing shortfall of US$39 billion (10). In this study, it was hypothesized that

38 higher levels of educational attainment will attenuate the positive statistical association between

39 wealth index and female overweight. This is possible given the observed positive effect of

40 education on other health outcomes related to women. In two studies involving several low and

41 middle income countries, increasing educational attainment was associated with decreased infant

42 mortality (11,12). The relationship between education and health is postulated to be mediated by

43 a variety of factors including personal traits such as personal control, increased income or

44 earnings, and social support (13). Consequently, this study aimed to examine the effect

45 modification of educational attainment on the associated with household wealth and adult female

46 overweight in 34 SSA countries with a multiplicative model.

187

47 Methods

48 Data Source

49 The DHS are nationally and regionally representative cluster sample surveys of women aged 15

50 and older, collected in 90 countries within individual households (14,15). During sampling,

51 countries were divided into survey regions, which are sub-national units that correspond to

52 existing administrative units such as states or provinces. Cluster sampling occurred in two stages.

53 From each region, random sampling of enumeration units or clusters were conducted. Within

54 each enumeration unit, a random sample of households were selected for inclusion in the survey.

55 Information such as household sociodemographic status and women’s reproductive history were

56 self-reported. Trained personnel collected anthropometric data by measuring and recording

57 participants’ weight using the SECA 874 digital scale to the nearest 0.01kg and height with the

58 Shorr height board. These anthropometric measurements were recorded in the biomarker

59 questionnaire (16).

61 Conceptual Model

62 It is well-established that dietary factors and the level of physical activity are more direct

63 individual causal factors implicated in overweight and obesity (17–19). Therefore, although we

64 acknowledged country-level heterogeneity, we hypothesized that these variables mediate the

65 relationship between household wealth and overweight in African countries and do not require

66 statistical adjustment. Additionally, while evidence for the relationship between parity and

67 obesity is not substantive in African countries (20), based on studies that show that first birth is

68 associated with weight gain (21), we hypothesized that parity was a potential confounder in this

69 study as it could be influenced by household wealth. We also hypothesized that the relationship

188

70 of household wealth and overweight will be confounded by place of residence (urban or rural

71 residence) (22), age (23), and by ethnicity-related factors such cultural practices and political

72 hegemony (24–26). Education was hypothesized as an effect modifier. These relationships were

73 tested in a statistical model, which is described in detail in the methods section. Figure 1 is a

74 diagrammatic representation of this conceptual model.

76 Figure 1: Conceptual Model of the Association of Household Wealth and Overweight with

77 Educational Attainment and Age as Effect Modifiers

79 Household Overweight 80 Wealth

82 Ethnicity, place of

83 residence, parity, age

85 Overweight outcomes by educational attainment 86

89 Data Compilation

90 For the purposes of this study, participants of interest were women 18 years or older. Pregnant

91 women were excluded due to gestational weight changes (27). Additionally, individuals younger

92 than 18 years were also excluded. We defined overweight and obesity, henceforth called

189

93 overweight, using the World Health Organization Adult Body Mass Index (BMI) classification

94 of ≥25kg/m2. The outcome – BMI categories - was coded as “0” or underweight/normal weight

95 for BMI <24.9kg/m2 and “1” or overweight for BMI ≥25kg/m2.

97 The exposure of interest, the household wealth index, is a composite measure of household living

98 standard calculated by the DHS program using principal component analysis that is based on

99 household ownership of specific assets such as television and bicycles, materials for household

100 construction, and types of water sources and sanitation facilities (28). The DHS dataset contained

101 wealth index factor scores which were ranked into three wealth index tertiles with labels of “1”

102 for the poor tertile, “2” for the middle tertile,”3” for the rich tertile. (15). Education level, the

103 potential effect modifier formerly placed into four categories of “no education”, “primary

104 education, “secondary education” and “higher education” was recoded as “0” or no/primary

105 education and “1” or secondary/higher education. The breakdown of wealth index and education

106 used in this study was to ensure sufficient power or number of participants within each strata of

107 these potential effect modifiers to enable the interaction analysis.

108

109 Potential confounders included age, ethnicity, place of residence and parity. Each country dataset

110 contained the age in years for individual participants and it was used in this form in the study

111 analysis. Ethnicity variables varied by country and for the purposes of the analysis were based on

112 data collected by the DHS program, which were present in each country’s dataset. In Nigeria and

113 Zimbabwe, where there were more than 60 ethnicity categories with several categories

114 containing fewer than ten respondents, and eight other countries (Comoros, Liberia, Rwanda,

115 Tanzania, Zimbabwe, Lesotho, Swaziland, Namibia) where ethnicity data was not collected,

190

116 ethnicity was not adjusted for as a confounder in the analysis. The same categories of place of

117 residence in the original DHS dataset – “urban” and “rural”- was used in the analysis. Parity, a

118 discrete variable that lists the number of children ever born by respondents was used in its

119 original form in the analysis.

120

121 Statistical Analysis

122 Analysis of complex surveys can be undertaken either with a design-based or model-based

123 approach to inference (29). Design-based approaches account for the complex sampling

124 mechanism using survey-weighted estimators, which do not rely on distributional assumptions of

125 randomness in the observed outcome to generate model parameters (29). Rather, randomness in

126 the observed outcome is introduced through the random sampling design (29). Whereas in

127 model-based approaches the complex survey mechanism is accounted for through accurate

128 specification of the underlying probability model to make the sampling mechanism

129 uninformative (29). Pure design-based modeling approaches allow for only descriptive inference

130 while pure model-based approaches allow for analytic inference (29).

131

132 The SAS software provides a hybrid complex survey procedure that integrates both the model

133 based and the design approaches allowing for both descriptive and analytic inferences, wherein

134 disproportionate sampling are adjusted by weighting coefficient estimates and stratification and

135 clustering are accounted for in the variance estimation using Taylor linearization (29,30).

136 Multilevel models, a model-based approach, often used with DHS data was not employed in this

137 analysis because there is no explicit scientific interest in modeling within cluster variance in the

138 relationship of SES variables and overweight outcomes (31). Additionally, multilevel models

191

139 require availability of covariates such as cluster level, household level and scaled individual

140 weights that capture all the information on the sampling mechanism (31). This sort of

141 information is not available in the DHS, hence, the hybrid approach incorporated in the SAS

142 software to account for complex survey features was used in the regression models in this

143 manuscript.

144

145 Statistical analysis was conducted using SAS software version 9.2 (32). First, descriptive

146 analysis was used to explore the interrelationships between overweight, household wealth index,

147 and educational attainment in the 34 countries initially considered for inclusion in this study.

148 Country datasets with less than 10 overweight and underweight respondents in any cell within

149 the strata of education and wealth did not undergo further analysis because such small sample

150 sizes were deemed to underpowered to detect the true effect, would not maintain the balance of

151 randomization equivalent to the true population of the subgroup, and were insufficient to

152 compute the true odds of overweight based on the normal approximation to the binomial

153 distribution (33,34). As a result, the final analysis included 22 countries.

154

155 After the descriptive analysis, SAS software multiple imputation procedure, PROC MI, was used

156 to create multiply imputed datasets for datasets with greater than 10% missing values, using a

157 seed value to ensure replication and adjusting for the sample weights and sample stratification.

158 Fully conditional specification method, the imputation technique applied in this study has been

159 previously described (35). At the end of the imputation process, multicollinearity of the

160 independent variables were assessed (variance inflation factor < 10, tolerance > 0.2) and a simple

161 logistic regression model followed by a multiple logistic regression model was used to test the

192

162 association of household wealth, as a categorical term, with overweight after adjustment for

163 education, ethnicity, parity, age and place of residence in each imputed and non-imputed

164 datasets. Country datasets where household wealth did not have a statistically significant

165 relationship in at least one wealth strata were not analyzed further. Finally, a multiple logistic

166 regression model was used to conduct an interaction analysis with educational attainment as the

167 effect modifier, household wealth as the exposure and overweight as the outcome with

168 adjustment for ethnicity, where applicable, parity, age, and place of residence. The final logistic

169 regression model for the analysis was in the form seen below.

푃푖 170 log [ ⁄ ] = 훽0 + 훽1푋1 + 훽2푋2 + 훽3푋3 + 퐵4 푋4 + 훽5 푋5 + 훽6 푋6 + 훽7푋1 ∗ 푋2 1 − 푃푖

171 + 훽8푋1 ∗ 푋3

172 Where 훽0 = 𝑖푛푡푒푟푐푒푝푡, 훽1 =

173 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 ℎ표푢푠푒ℎ표푙푑 푤푒푎푙푡ℎ 푡푒푟푡𝑖푙푒 푓표푟 푠푒푐표푛푑푎푟푦 표푟 ℎ𝑖푔ℎ푒푟 푒푑푢푐푎푡𝑖표푛, 훽2 =

174 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푚𝑖푑푑푙푒 ℎ표푢푠푒ℎ표푙푑 (퐻퐻)푤푒푎푙푡ℎ 푡푒푟푡𝑖푙푒, 훽3 =

175 푐표푒푓푓푐𝑖푒푛푡 푓표푟 푟𝑖푐ℎ 퐻퐻 푤푒푎푙푡ℎ 푡푒푟푡𝑖푙푒, 훽4 = 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푎푔푒, 훽4 =

176 푣푒푐푡표푟 표푓 푐표푒푓푓𝑖푐𝑖푒푛푡푠 푓표푟 푒푡ℎ푛𝑖푐𝑖푡푦 , 훽5 = 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푝푙푎푐푒 표푓 푢푟푏푎푛 푟푒푠𝑖푑푒푛푐푒,

177 훽6 = 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푝푎푟𝑖푡푦 푎푛푑 훽7 =

178 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푡ℎ푒 𝑖푛푡푒푟푎푐푡𝑖표푛 표푓 푠푒푐표푛푑푎푟푦 표푟 ℎ𝑖푔ℎ푒푟 푒푑푢푐푎푡𝑖표푛 푎푛푑 퐻퐻 푤푒푎푙푡ℎ 𝑖푛 푡ℎ푒

179 푚𝑖푑푑푙푒 푡푒푟푡𝑖푙푒, 훽8 =

180 푐표푒푓푓𝑖푐𝑖푒푛푡 푓표푟 푡ℎ푒 𝑖푛푡푒푟푎푐푡𝑖표푛 표푓 푠푒푐표푛푑푎푟푦 표푟 ℎ𝑖푔ℎ푒푟 푒푑푢푐푎푡𝑖표푛 푎푛푑 퐻퐻 𝑖푛 푡ℎ푒 푟𝑖푐ℎ 푡푒푟푡𝑖푙푒

181

182 Finally, the parameter estimates, and the standard errors from each imputed dataset were

183 combined using the SAS software procedure, PROC MIANALYZE to generate averaged

184 estimates. The strata-level 95% CIs and p-values for the parameter estimates were adjusted for

193

185 multiple comparisons using a previously described technique (35). Standard errors were also

186 adjusted for the complex sample design and for the variability resulting from the multiple

187 imputation process. Finally, the odds ratios and 95% confidence intervals (CIs) of the

188 relationship of household wealth and overweight within strata of education level, the interaction

189 variables, were generated as an exponent of the parameter estimates and their 95% CIs

190 respectively.

191

192 Results

193 Participants

194 Table 1 details the surveys that were included in the study, the total DHS sample prior to

195 exclusions and identification of missing cases, complete and missing BMI cases, excluded

196 observations and reasons for exclusions, and overweight prevalence for all the thirty-four

197 country datasets initially considered in this study. The percentage of missing BMI data is the

198 ratio of the missing BMI to the sum of the complete and missing BMI cases. Missing BMI cases

199 ranged from 0.8% to 68.6%. All included surveys were the most recent DHS surveys available

200 for public use and were conducted prior to 2005. The presentation of countries in Table 1 and

201 subsequent tables were grouped by income and this was based on the World Bank classification

202 of countries for the year of the survey (9).

203

204 Descriptive Data

205 Besides the survey description, Table 1 also shows the national level overweight prevalence in

206 women, 18 years and older by country. Madagascar had the lowest overweight prevalence at

207 7.1% while the highest overweight prevalence was observed in Swaziland at 56.6%.

194

Table 1: Description of Included Surveys and Overall Overweight Prevalence in 34 SSA Countries Country & Survey Total DHS Missing BMI Exclusions Overweight Year sample data (% of total Prevalence /complete BMI sample) & 95% CI BMI cases Low Income Countries Benin 2011 - 2012 16,599/12,811 372 (2.8) Adolescents 28.9 (1,869) Pregnant women (27.9 - 30.0) (1,547)

Burkina Faso 2010 17,087/6,617 6,731 (50.4) Adolescents 12.2 (2,104) Pregnant women (11.1 - 13.2) (1,635)

Burundi 2010 9,389/3,411 3,397 (49.9) Adolescents 8.4 (1,440) Pregnant women (7.5 - 9.3) (920)

Chad 2014 - 2015 17,719/8,367 4,655 (35.7) Adolescent 12.6 (2,411) Pregnant women (11.6 - 13.6) (2,286)

Comoros 2012 5,329/4,102 122 (2.9) Adolescents 41.0 (779) Pregnant women (38.7 - 43.3) (326)

Congo DRC 2013 - 18,827/7,031 7,168 (50.5) Adolescents 17.3 2014 (2,376) Pregnant women (15.3 - 19.3) (2,252)

Ethiopia 2016 15,683/11,862 635 (5.1) Adolescents 8.22 (2,094) Pregnant women (7.1 - 9.3) (1,092)

Gambia 2013 10,233/3,603 4,457 (55.3) Adolescents 24.9 (1,374)

195

Guinea 2012 9,142/3,623 3,476 (49.0) Adolescents 21.1 (1,153) Pregnant women (19.2 - 23.1) (2,043)

Liberia 2013 9,239/3,654 3,648 (50.0) Adolescents 29.7 (1,192) Pregnant women (27.0 - 32.5) (745)

Madagascar 2008 - 17,375/6,575 6,575 (51.8) Adolescents 7.1 2009 (2,439) Pregnant women (6.2 - 8.0) (1,304)

Malawi 2015 - 2016 24,562/6,443 13,264 (67.3) Adolescents 23.1 (3,170) Pregnant women (21.7 - 24.4) (1,685)

Mali 2012 - 2013 10,424/ 4,153 4,069 (49.5) Adolescents 19.1 (1,096) Pregnant women (17.6 - 20.7) (1,106)

Mozambique 2011 13,745/10,393 117 (1.1) Adolescents 18.1 (1,970) Pregnant women (17.0 - 19.2) (1,265)

Niger 2012 11,160/3,930 4,694 (54.4) Adolescents 19.3 (1,155) Pregnant women (17.6 - 21.0) (1,381)

196

Table 1: Description of Included Surveys and Overall Overweight Prevalence in 34 SSA Countries Country & Survey Total DHS Missing BMI Exclusions Overweight Year sample data (% of total Prevalence /complete BMI sample) & 95% CI BMI cases Rwanda 2014 - 2015 13,497/5,361 5,450 (50.4) Adolescents 22.5 (1,743) Pregnant women (21.2 - 23.8) (943)

Sierra Leone 2013 16,658/6,269 6,653 (51.5) Adolescents 20.3 (2,416) Pregnant women (18.5 - 22.0) (1.320)

Tanzania 2015 - 2016 13,266/10,369 86 (0.8) Adolescents 31.4 (1,749) Pregnant women (29.9 - 33.0) (1,062)

Togo 2013 - 2014 9,480/3,875 3,796 (49.5) Adolescents 33.3 (1,019) Pregnant women (31.5 -35.2) (790)

Uganda 2016 8,674/2,044 4,474 (68.6) Adolescent 20.5 (1,245) Pregnant women (18.0 - 23.0) (911)

Zimbabwe 2015 9,955/7,758 251 (3.1) Adolescent 38.9 (1,348) Pregnant women (37.3 - 40.4) (569) Lower-middle Income Countries Cameroon 2011 15,426/6,091 5,804 (48.8) Adolescents (2,142) 35.1 Pregnant women (33.5 - (1,389) 36.6)

Congo 2011- 2012 10,819/4,414 4,072 (48.0) Adolescents (1,326) 29.3 Pregnant women (26.9 - (1,007) 31.7)

197

Table 1: Description of Included Surveys and Overall Overweight Prevalence in 34 SSA Countries Country & Survey Total DHS Missing BMI Exclusions Overweight Year sample data (% of total Prevalence /complete BMI sample) & 95% CI BMI cases Cote d' Ivoire 2011 - 10,060/3,763 4,163 (52.5) Adolescents (1,178) 27.7 2012 Pregnant women (25.7 - (956) 29.7)

Ghana 2014 9,396/3,845 3,785 (49.6) Adolescents (1,101) 44.3 Pregnant women (42.0 - (665) 46.6)

Kenya 2014 31,079/11,762 13,581 (53.6) Adolescents (3,769) 35.8 Pregnant women (34.4 - (1,967) 37.2)

Lesotho 2014 6,621/2,764 2,656 (49.0) Adolescents (949) 49.3 Pregnant women (47.0 - (252) 51.7)

Nigeria 2013 38,948/29,304 442 (1.49) Adolescents (4,946) 27.6 Pregnant women (26.6 - (4,256) 28.6)

Senegal 2010 - 2011 15,688/4,491 7,870 (63.7) Adolescents (2,113) 23.9 Pregnant women (21.8 - (1,214) 26.0)

Sao Tome & Principe 2,615/1,901 185 (8.9) Adolescents (319) 36.6 2008 - 2009 Pregnant women (33.8 - (210) 39.4)

Swaziland 2006 - 2007 4,987/3,855 110 (2.8) Adolescents (782) 56.6 Pregnant women (54.8 - (240) 58.5)

Zambia 2013 - 2014 16,411/12,809 137 (1.1) Adolescent (2,168) 25.1 Pregnant women (24.0 - (1,297) 26.1)

198

Table 1: Description of Included Surveys and Overall Overweight Prevalence in 34 SSA Countries Country & Survey Total DHS Missing BMI Exclusions Overweight Year sample data (% of total Prevalence /complete BMI sample) & 95% CI BMI cases Upper-middle Income Countries Gabon 2012 8,422/4,265 2,251 (34.5) Adolescents (1,121) 48.7 Pregnant women (46.2 - (785) 51.3) Namibia 2013 10,018/4,407 3,994 (47.5) Adolescents (1,072) 35.2 Pregnant women (33.4 - (545) 37.0) 208

209 In Table 2, the prevalence of overweight within strata of education attainment and wealth tertiles

210 are presented. Overall, overweight prevalence in the educational attainment strata of individuals

211 with secondary or higher education was greatest in the rich tertile. In the combined strata of

212 individuals with no or primary education, the rich tertile also had the highest prevalence of

213 overweight, with the exception of 13 countries such as Comoros where the poor tertile had the

214 greatest overweight prevalence. The supplementary appendix also provides additional descriptive

215 data of overweight prevalence by wealth index in individual countries.

216

217

218

219

220

199

Table 2: Prevalence of Overweight and Underweight/Normal Weight Combined by Strata of Educational Attainment and Wealth Status in 34 SSA Countries* Country Wealth No/Primary Education Secondary/ Higher Tertiles - Number (%) Education - Number (%) Under/normal Overweight Under/normal Overweight weight weight Low income Countries Benin Poor 3,453 (42.1) 720 (22.1) 130 (6.8) 20 (2.4) Middle 2,799 (34.8) 1,038 (33.7) 385 (21.0) 81 (9.5) Rich 1, 524 (23.0) 1,074 (44.2) 1,022 (72.2) 565 (88.1)

Burkina Faso Poor 2,133 (41.8) 106 (18.8) 33 (5.4) 0 (0) Middle 1,977 (37.3) 131 (21.2) 56 (7.7) 3(1.6) Rich 1,155 (21.0) 345 (60.0) 479 (86.9) 197(98.4)

Burundi Poor 1,096 (43.7) 40 (21.1) 35 (10.3) 1 (1.7) Middle 990 (40.6) 64 (36.2) 72 (23.5) 7 (8.9) Rich 488 (15.7) 130 (42.7) 350 (66.1) 138 (89.5)

Chad Poor 2,476 (40.5) 226 (28.7) 126 (17.2) 14 (7.3) Middle 2,347 (34.3) 190 (24.0) 155 (20.9) 26 (9.0) Rich 1,763 (25.1) 388 (47.3) 489 (61.8) 167 (83.6)

Comoros Poor 570 (50.8) 414 (40.2) 287 (22.4) 96 (10.1) Middle 364 (32.6) 325 (35.7) 429 (34.1) 247 (35.7) Rich 183 (16.7) 220 (24.1) 541 (43.5) 413 (54.2)

Congo DRC Poor 1,727 (43.2) 157 (25.7) 455 (16.8) 23 (2.3) Middle 1,436 (41.7) 176 (33.2) 701 (30.1) 72 (10.3) Rich 464 (15.1) 145 (41.0) 1,178 (53.1) 497 (87.4)

Ethiopia Poor 3,681 (34.7) 152 (16.0) 173 (5.5) 4 (0.4) Middle 3,332 (56.2) 191 (34.4) 472 (36.0) 36 (7.2) Rich 1,325 (9.1) 569 (49.6) 1361 (58.5) 566 (92.4)

Gambia Poor 825 (40.3) 189 (27.4) 223 (17.7) 29 (6.7) Middle 678 (37.7) 200 (35.0) 235 (24.5) 58 (21.2) Rich 281 (22.1) 181 (37.6) 485 (57.8) 219 (72.2)

200

Table 2: Prevalence of Overweight and Underweight/Normal Weight Combined by Strata of Educational Attainment and Wealth Status in 34 SSA Countries* Country Wealth No/Primary Education Secondary/ Higher Tertiles - Number (%) Education - Number (%) Under/normal Overweight Under/normal Overweight weight weight Guinea Poor 1,093 (45.8) 95 (17.0) 31 (6.9) 2 (1.0) Middle 905 (37.4) 210 (36.9) 96 (20.1) 22 (14.0) Rich 437 (16.7) 256 (46.1) 319 (73.0) 157 (85.1)

Liberia Poor 960 (37.1) 223 (23.2) 80 (5.2) 20 (3.3) Middle 707 (33.8) 230 (24.5) 196 (15.0) 46 (7.8) Rich 345 (29.1) 234 (52.2) 389 (79.8) 224 (89.0)

Madagascar Poor 2,078 (44.5) 51 (22.0) 104 (7.1) 1 (0.4) Middle 1,644 (42.9) 81 (39.0) 480 (30.5) 19 (7.9) Rich 539 (12.6) 83 (39.0) 1,222 (62.4) 273 (91.7)

Mali Poor 1,219 (43.6) 165 (25.1) 17 (4.6) 1 (1.0) Middle 1,100 (40.0) 206 (32.5) 73 (23.0) 8 (8.0) Rich 539 (16.4) 335 (42.3) 330 (72.4) 160 (91.0)

Malawi Poor 1,722 (49.5) 254 (30.2) 172 (16.0) 24 (4.2) Middle 1,326 (36.4) 356 (36.8) 349 (26.3) 82 (15.9) Rich 569 (14.1) 355 (33.0) 746 (57.6) 488 (79.9)

Mozambique Poor 3,174 (56.1) 244 (21.6) 60 (5.0) 4 (0.8) Middle 2,486 (34.8) 562 (38.9) 353 (28.1) 69 (11.3) Rich 904 (9.1) 762 (39.5) 1,100 (66.9) 675 (87.9)

Niger Poor 1,180 (45.4) 168 (28.3) 13 (10.2) 0 (0) Middle 1,095 (41.6) 212 (31.6) 31 (19.1) 8 (8.7) Rich 512 (13.1) 371 (40.1) 198 (70.6) 137 (91.3)

Rwanda Poor 1483 (46.0) 231 (27.8) 104 (13.3) 15 (3.9) Middle 1,172 (37.0) 275 (33.5) 219 (28.6) 59 (16.6) Rich 602 (17.0) 359 (38.7) 518 (58.1) 324 (79.5)

Senegal Poor 1,222 (30.3) 188 (15.6) 67 (7.5) 8 (2.6)

201

Table 2: Prevalence of Overweight and Underweight/Normal Weight Combined by Strata of Educational Attainment and Wealth Status in 34 SSA Countries* Country Wealth No/Primary Education Secondary/ Higher Tertiles - Number (%) Education - Number (%) Under/normal Overweight Under/normal Overweight weight weight Middle 1,000 (31.0) 274 (27.1) 201 (21.4) 33 (10.1) Rich 750 (39.7) 326 (57.3) 319 (71.1) 103 (87.3)

Sierra Leone Poor 1,656 (46.4) 256 (28.9) 172 (13.2) 18 (4.8) Middle 1,369 (38.5) 328 (35.7) 330 (24.0) 57 (13.1) Rich 630 (15.2) 341 (35.3) 802 (62.8) 310 (82.1)

Tanzania Poor 2,730 (50.2) 521 (22.7) 177 (11.8) 37 (2.9) Middle 1,941 (32.5) 788 (32.8) 520 (27.4) 211 (12.6) Rich 823 (17.3) 873 (44.4) 935 (60.8) 813 (84.4)

Togo Poor 992 (40.4) 158 (16.1) 143 (12.8) 20 (3.5) Middle 671 (40.2) 273 (35.7) 242 (29.8) 81 (18.2) Rich 282 (19.4) 326 (48.1) 393 (57.4) 294 (78.3)

Uganda Poor 595 (47.3) 40 (17.2) 38 (8.7) 5 (2.6) Middle 435 (40.5) 109 (49.5) 131 (35.5) 20 (13.0) Rich 150 (12.2) 95 (33.3) 260 (55.7) 166 (84.4)

Zimbabwe Poor 949 (74.4) 334 (56.0) 989 (34.9) 371 (18.5) Middle 299 (21.6) 232 (33.8) 1,201 (35.9) 873 (35.4) Rich 81 (4.0) 94 (10.2) 1,111 (29.2) 1,253 (46.2)

Lower-middle Income Countries Cameroon Poor 1,424 (59.7) 302 (28.5) 207 (11.2) 80 (5.3) Middle 761 (30.3) 418 (43.5) 560 (32.7) 306 (24.2) Rich 235 (10.1) 251 (28.0) 797 (56.2) 750 (70.5)

Congo Poor 870 (36.1) 99 (15.9) 410 (9.2) 79 (3.9) Middle 550 (39.1) 110 (38.1) 697 (26.8) 164 (13.4) Rich 162 (24.7) 73 (46.0) 744 (64.0) 456 (82.8)

202

Table 2: Prevalence of Overweight and Underweight/Normal Weight Combined by Strata of Educational Attainment and Wealth Status in 34 SSA Countries* Country Wealth No/Primary Education Secondary/ Higher Tertiles - Number (%) Education - Number (%) Under/normal Overweight Under/normal Overweight weight weight Cote d' Ivoire Poor 1,079 (43.9) 170 (20.4) 44 (9.1) 5 (1.5) Middle 823 (33.0) 256 (31.9) 147 (27.7) 33 (13.0) Rich 450 (23.1) 329 (47.8) 267 (63.2) 160 (85.6)

Ghana Poor 759 (56.5) 169 (22.9) 280 (16.7) 61 (4.4) Middle 311 (33.1) 228 (41.7) 456 (39.8) 305 (27.5) Rich 84 (10.4) 150 (35.4) 431 (43.5) 611 (68.2)

Kenya Poor 2,891 (45.1) 509 (18.7) 424 (12.2) 102 (4.3) Middle 1,711 (36.7) 875 (38.9) 1,010 (34.6) 409 (20.2) Rich 695 (18.2) 752 (42.3) 1,196 (53.2) 1,188 (75.5)

Lesotho Poor 436 (59.4) 234 (38.5) 182 (20.2) 99 (10.2) Middle 168 (27.5) 192 (36.9) 281 (35.7) 290 (35.5) Rich 64 (13.1) 108 (24.6) 287 (44.1) 423 (54.3)

Nigeria Poor 7,571 (61.7) 1,192 (33.9) 835 (8.8) 157 (2.9) Middle 3,901 (29.8) 1,489 (40.6) 3,282 (35.0) 1,096 (20.9) Rich 1,038 (8.5) 846 (25.5) 4,529 (56.2) 3,368 (76.1)

Sao Tome and Poor 388 (43.0) 182 (31.1) 47 (14.1) 8 (4.8) Principe Middle 337 (38.3) 165 (29.9) 85 (19.6) 46 (20.1) Rich 140 (18.7) 160 (39.0) 215 (66.3) 128 (75.1)

Swaziland Poor 413 (61.7) 391 (47.1) 265 (26.9) 225 (17.7) Middle 182 (28.7) 295 (35.3) 366 (37.3) 447 (35.8) Rich 83 (9.7) 166 (17.7) 376 (35.9) 646 (46.6)

Zambia Poor 3,163 (54.5) 396 (26.7) 628 (15.6) 69 (4.0) Middle 2,036 (32.9) 567 (36.8) 1,326 (29.0) 354 (16.3)

203

Table 2: Prevalence of Overweight and Underweight/Normal Weight Combined by Strata of Educational Attainment and Wealth Status in 34 SSA Countries* Country Wealth No/Primary Education Secondary/ Higher Tertiles - Number (%) Education - Number (%) Under/normal Overweight Under/normal Overweight weight weight Rich 545 (12.5) 436 (36.9) 2,040 (55.5) 1,242 (79.8) Upper-middle Income Countries Gabon Poor 652 (40.8) 266 (16.0) 354 (9.5) 171 (5.5) Middle 261 (34.7) 257 (42.8) 529 (31.5) 374 (26.5) Rich 104 (24.6) 160 (41.2) 575 (59.0) 562 (68.0)

Namibia Poor 425 (69.7) 111 (35.6) 519 (32.1) 131 (12.3) Middle 184 (25.7) 139 (42.5) 582 (33.3) 357 (34.2) Rich 31 (4.6) 55 (21.9) 529 (34.6) 550 (53.5) Note: Percentages may not sum up to 100 due to rounding 221

222 Main Results

223 In all the 22 countries included in the interaction analyses; household wealth was positively

224 associated with overweight (see table 2 in the supplementary appendix). Figures 1 – 3 illustrates

225 the results of the interaction analysis after adjustment for age, parity, place of residence and

226 ethnicity in the 22 countries with sufficient sample size for analytical procedures. We found no

227 evidence of the multiplicative interaction of education in relation to the association between

228 household wealth index and overweight in any of the countries examined. However, the point

229 estimates for the odds ratios of overweight in the rich tertile among those with secondary or

230 higher education in Cameroon, Ghana, Nigeria, Swaziland, Gabon and Namibia were diminished

231 by one or more units compared to those with no or primary education. For the middle tertile, this

232 difference of at least one unit was only observed in Gabon. Detailed odds ratio estimates and

233 95% CI for the interaction analysis are presented in the supplementary appendix

204

234 Benin (DHS, 2011-12) Chad (DHS, 2014 - 2015) 235 4.5 4 4 3.5 Rich Tertile Rich Tertile 236 3.5 (None*) 3 (None) 3 237 Rich Tertile 2.5 Rich Tertile 2.5 (Secondary**) (Secondary) 2

2 Middle Tertile Middle Tertile 95% CI 238 95%CI 1.5 (None) 1.5 (None) 1 Middle Tertile 1 Middle Tertile 239 (Secondary) (Secondary)

0.5 0.5

Odds Ratio Overeight of and Odds Ratio of Overweight Odds Ratio Overweight of and 0 0 240

241 Benin (DHS, 2011-12) 5 Congo DRC (DHS, 2013 - 2014) 242 Comoros (DHS, 2012)Rich 4 4 Tertile(None*) 16 Rich Tertile 243 3.5 14 3 (Secondary**)Rich Tertile Rich Tertile 3 (None) 12 (None) Middle Tertile 244 2 Rich Tertile 95% CI 2.5 (None)Rich Tertile 10 (Secondary) (Secondary) 245 1 2 Middle Tertile 8

(Secondary)Middle Tertile Middle Tertile 95%CI 95% CI 1.5 (None) 6 (None) 0 246 Odds Ratio Overweight of and 1 Middle Tertile 4 Middle Tertile (Secondary) 0.5 (Secondary) 2

Odds Ratio of Overweight Odds Ratio Overweight of and 247 Odds Ratio Overweight of and 0 0

248 Figure 2: Plot of the Interaction of Education on the Association between Household Wealth and Overweight in Low-Income SSA Countries (Note: 249 *no/primary education; **secondary/higher education; analysis was adjusted for ethnicity, where applicable, parity, age, and place of residence)

250

205

251 Liberia (DHS, 2013) Gambia (DHS, 2013) 7 252 4.5 4 6 Rich Tertile 253 Rich Tertile (None) 3.5 (None) 5 3 Rich Tertile Rich Tertile 4 (Secondary) 254 2.5 (Secondary) 3 Middle Tertile 2 Middle Tertile 95% CI (None) 255 95CI % 1.5 (None) 2 Middle Tertile Middle Tertile 1 1 (Secondary) 256 (Secondary)

0.5 Odds Ratio Overweight of and

Odds Ratio of Overweight Odds Ratio Overweight of and 0 0 257

258

259 Malawi (DHS, 2015 - 2016) Rwanda (DHS, 2014 - 2015) 7 260 7 6 Rich Tertile 6 261 (None) Rich Tertile 5 (None) Rich Tertile 5 262 4 (Secondary) Rich Tertile 4 (Secondary) 3 Middle Tertile 95%CI 3 Middle Tertile 263 (None) 95%CI 2 (None) Middle Tertile 2 Middle Tertile 1 (Secondary)

264 1 (Secondary) Odds Ratio of Overweight Odds Ratio Overweight of and

0 Odds Ratio Overweight of and 0 265

266 Figure 2 continued 267

206

268 Sierra Leone (DHS, 2013) Tanzania (DHS, 2015 - 2016) 269 4.5 9 4 8 Rich Tertile Rich Tertile 270 3.5 (None) 7 (None) 3 271 Rich Tertile 6 Rich Tertile 2.5 (Secondary) 5 (Secondary) 2 Middle Tertile 4

95% CI Middle Tertile 272 (None) 95%CI 1.5 3 (None) Middle Tertile 1 2 Middle Tertile 273 (Secondary) (Secondary)

0.5 1

Odds Ratio of Overweight Odds Ratio Overweight of and Odds Ratio of Overweight Odds Ratio Overweight of and 274 0 0

275

276 Togo (DHS, 2013 - 2014) Zimbabwe (DHS, 2015) 277 9 7 8 Rich Tertile 6 7 Rich Tertile 278 (None) (None) 6 5 Rich Tertile Rich Tertile 279 5 (Secondary) 4 (Secondary) 4 Middle Tertile 95%Ci 3 Middle Tertile 280 3 (None) 95%CI (None) 2 2 Middle Tertile Middle Tertile (Secondary)

281 1 1 (Secondary) Odds Ratio of Overweight Odds Ratio Overweight of and 0 Odds Ratio Overweight of and 0 282

283 Figure 2 continued

284

207

285 Cameroon (DHS, 2011) Congo (DHS, 2011 - 2012) 286 6 7

6 287 5 Rich Tertile Rich Tertile (None) (None) 5 4 Rich Tertile Rich Tertile 288 (Secondary) 4 (Secondary) 3

Middle Tertile 3 Middle Tertile 95%CI 289 95%CI 2 (None) (None) 2 Middle Tertile Middle Tertile

290 1 (Secondary) 1 (Secondary) Odds Ratio of Overweight Odds Ratio Overweight of and 0 Odds Ratio Overweight of and 0 291

292

293 Ghana (DHS, 2014) Kenya (DHS, 2014) 294 14 6 12 Rich Tertile 5 Rich Tertile 295 (None) 10 (None) Rich Tertile 4 Rich Tertile 296 8 (Secondary) (Secondary) 3

6 Middle Tertile CI Middle Tertile 95%CI (None) 297 2 (None) 4 Middle Tertile Middle Tertile

298 2 (Secondary) 1 (Secondary) Odds Ratio of Overweight Odds Ratio Overweight of and 0 0 299 oddsRatio Overweight of and 95

300 Figure 3: Plot of the Interaction of Education on the Association between Household Wealth and Overweight in Lower-middle Income SSA Countries (Note: *no/primary education; **secondary/higher education; analysis was adjusted for ethnicity, where applicable, parity, age, and place of residence) 301

208

302 Lesotho (DHS, 2014) Nigeria (DHS, 2013) 303 6 7

6 304 5 Rich Tertile Rich Tertile (None) (None) 5 4 Rich Tertile Rich Tertile 305 (Secondary) 4 (Secondary) 3

Middle Tertile 3 Middle Tertile 95%CI 306 95%CI 2 (None) (None) 2 Middle Tertile Middle Tertile

307 1 (Secondary) 1 (Secondary) Odds Ratio of Overweight Odds Ratio Overweight of and Odds Ratio of Overweight Odds Ratio Overweight of and 0 0 308

309

310 Swaziland (DHS, 2006 -2007) Zambia (DHS, 2013 - 2014) 311 8 8 7 7 312 Rich Tertile Rich Tertile 6 (None) 6 (None) 5 Rich Tertile 5 Rich Tertile 313 (Secondary) (Secondary) 4 4

Middle Tertile Middle Tertile 95%CI 314 3 (None) 95%CI 3 (None) 2 Middle Tertile 2 Middle Tertile 315 (Secondary) (Secondary)

1 1 Odds Ratio of Overweight Odds Ratio Overweight of and 0 Odds Ratio Overweight of and 0 316

317 Figure 3 Continued

318

209

319 Gabon (DHS, 2012) Namibia (DHS, 2013) 320 8 14 7 12 321 Rich Tertile Rich Tertile (None) (None) 6 10 322 5 Rich Tertile Rich Tertile (Secondary) 8 (Secondary) 4

Middle Tertile 6 Middle Tertile 95%CI 323 95%CI 3 (None) (None) 4 2 Middle Tertile Middle Tertile 324 (Secondary) (Secondary)

1 2 Odds Ratio of Overweight Odds Ratio Overweight of and Odds Ratio of Overweight Odds Ratio Overweight of and 0 0 325

326 Figure 4: Plot of the Interaction of Education on the Association between Household Wealth and Overweight in Upper-middle Income sub-Saharan African 327 Countries (Note: *no/primary education; **secondary/higher education; analysis was adjusted for ethnicity, where applicable, parity, age, and place of residence) 328

329

330

331

332

333

334

335

210

336 Discussion

337 In contrast to our study hypothesis, we noted a lack of effect modification of education on the

338 association between household wealth index and adult female overweight within the 22 SSA

339 countries examined in this study. Our study findings signify that wealthy individuals have an

340 equal likelihood of overweight irrespective of educational attainment, given that the underlying

341 relationship between education and wealth is multiplicative. Our findings for low-income

342 countries are similar to previous work in which obesity was used as the outcome (2). However, it

343 was not in consonance with previous work where significant interactions were found in middle-

344 income countries (2,4).

345

346 To investigate whether the differences in the study outcome were due to differences in

347 methodology, we reanalyzed the data for Gabon and Namibia, two upper-middle income SSA

348 countries. In the sensitivity analysis and similar to previous work (2,4), we used household

349 wealth index as a continuous variable, with and without adjustment for ethnicity, in Gabon where

350 the ethnicity variable was available, and without corrections for multiple comparisons, and

351 arrived at the same conclusion (see supplementary appendix). Moreover, we also conducted

352 additional sensitivity analysis with Gabon and Namibia using obesity (obesity prevalence in

353 Gabon and Namibia stood at 20.9% and 14.8% respectively) as the outcome in a binary logistic

354 regression and our conclusions did not change (supplementary appendix). This suggested that the

355 observed variability in study outcome may result from national differences which may influence

356 an individual’s ability to properly navigate the food environment. While it is unknown why these

357 hypothesized differences exist, a potential factor could be the greater priority placed on

358 overweight and obesity prevention in the middle-income countries examined in previous work

211

359 (2,4). Therefore, it is possible that improving levels of educational attainment in SSA countries

360 may be a feasible strategy for female overweight and obesity prevention but only under certain

361 conditions. This has been noted in tobacco control where the introduction of the 1964 Surgeon’s

362 Report, the general consensus that smoking was linked to greater lung cancer and mortality risk,

363 and the introduction of tobacco control efforts led to declines in smoking, with a dose-response

364 effect depending on the level of educational attainment (36–38). Studies that examine the

365 availability, adoption and implementation of overweight and obesity prevention policies and

366 programs in middle-income SSA countries compared to other middle-income countries where

367 educational attainment has been found to interact with household wealth in a multiplicative

368 model may shed some light. Nonetheless, the absence of multiplicative interaction in our study

369 implies the potential presence of additive interaction (39). Whether this additive interaction is of

370 public health significance, it is beyond the scope of this study.

371

372 Besides our examination of effect modification, our study shows that some low-income

373 economies in SSA also exhibit a high prevalence of female overweight exceeding 30%. This

374 includes countries like Togo, Tanzania, Comoros and Zimbabwe. Additionally, variations exist

375 in the socioeconomic group with the greater prevalence of overweight by country. For instance,

376 in both Togo and Tanzania, overweight is concentrated among the richest tertile irrespective of

377 educational attainment, whereas in Zimbabwe, Comoros, and Swaziland, the poorest tertile

378 among those with no or primary education and the richest tertile among those with secondary or

379 higher education have the highest prevalence of overweight in their respective strata. These

380 findings need to be taken into account in the design of overweight and obesity policies and

381 programs in individual countries.

212

382 Finally, these study findings should be interpreted in the light of several limitations. First, the

383 cross-sectional nature of our study limits any claims of causality. Second, since detecting

384 interactions require large sample sizes, it is possible that the sample sizes within the subgroups

385 were inadequate for the purposes of the study. Third, some of the countries included in this study

386 had a large proportion of missing data and although we tried to address this issue using

387 imputation techniques, the potential for bias still remains. Fourth, the multiplicative effect

388 modification of education may be present only in college-educated individuals, however, we

389 were unable to create stand-alone categories for this sub-population due to their small numbers,

390 which would hinder the examination of an interaction effect.

391

392 Conclusions

393 In this study, we examined the multiplicative effect modification of education on the positive

394 relationship between household wealth index and overweight in low and middle income SSA

395 countries and found a null effect. Although our findings were in consonance with previous work

396 in low-income countries, it contradicts previous claims for middle-income countries. Our

397 sensitivity analysis did not change our conclusions suggesting that the observed divergence from

398 previous work may be due to contextual differences related to educational attainment. Further

399 research to document and test the impact of these contextual factors on the influence of

400 education on overweight and obesity outcomes is warranted.

401

402 List of abbreviations

403 DHS – Demographic and Health Survey

404 SES – Socioeconomic Status

213

405 SSA –sub-Saharan Africa

406

407 Declarations

408 Ethics approval and consent to participate

409 The research protocol for this study was reviewed by the East Tennessee State University

410 Institutional Review Board and did not require ethical approval because it was designated a non-

411 human subject research

412

413 Consent for publication

414 Not applicable

415

416 Availability of data and materials

417 The datasets generated and/or analyzed during the current study are available in the DHS

418 program repository, [https://dhsprogram.com/]

419

420 Competing interests

421 The authors declare that they have no competing interests

422

423

424

214

425 Funding

426 The author’s doctoral program is partly funded by Rotary International through Rotary District

427 7570. Rotary International had no involvement in the design of the study, data collection,

428 analysis, and interpretation, and in writing the manuscript.

429

430 Authors' contributions

431 I.D.O conceptualized the study, analyzed and interpreted the study data, and wrote the first draft

432 of the study. All other authors contributed to the interpretation and writing of the study.

433

434 Acknowledgements

435 The authors would like to acknowledge Dr. Joe Brown of Georgia Institute of Technology, who

436 provided valuable suggestions during the conception of this study.

215

437 References

438 1. Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. Obesity and socioeconomic status in

439 developing countries: A systematic review. Obes Rev. 2012;13(11):1067–79.

440 2. Aitsi-Selmi A, Bell R, Shipley MJ, Marmot MG. Education modifies the association of

441 wealth with obesity in women in middle-income but not low-income countries: an

442 interaction study using seven national datasets, 2005-2010. PLoS One. 2014;9(3):e90403.

443 3. Neupane S, Prakash KC, Doku DT. Overweight and obesity among women: analysis of

444 demographic and health survey data from 32 Sub-Saharan African Countries. BMC Public

445 Health. 2016 Jan;16:30.

446 4. Aitsi-Selmi A, Chandola T, Friel S, Nouraei R, Shipley MJ, Marmot MG. Interaction

447 between education and household wealth on the risk of obesity in women in Egypt. PLoS

448 One. 2012;7(6):e39507.

449 5. Iliodromiti S, Celis-Morales CA, Lyall DM, Anderson J, Gray SR, Mackay DF, et al. The

450 impact of confounding on the associations of different adiposity measures with the

451 incidence of cardiovascular disease: a cohort study of 296 535 adults of white European

452 descent. Eur Heart J. 2018 May 1;39(17):1514–20.

453 6. Khan SS, Ning H, Wilkins JT, Allen N, Carnethon M, Berry JD, et al. Association of

454 Body Mass Index With Lifetime Risk of Cardiovascular Disease and Compression of

455 Morbidity. JAMA Cardiol. 2018 Apr 1;3(4):280.

456 7. The GBD 2015 Obesity Collaborators. Health Effects of Overweight and Obesity in 195

457 Countries over 25 Years. N Engl J Med. 2017;377(1):13–27.

458 8. World Health Organization. Global Health Observatory Data Repository [Internet]. WHO.

459 2017. Available from: http://apps.who.int/gho/data/?theme=main

216

460 9. World Bank. World Bank Country and Lending Groups [Internet]. 2018 [cited 2018 Feb

461 25]. Available from: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-

462 world-bank-country-and-lending-groups

463 10. UNESCO. Aid to education is stagnating and not going to countries most in need: Global

464 Education Monitoring Report [Internet]. Paris: UNESCO; 2017 [cited 2019 Feb 5].

465 Available from: https://unesdoc.unesco.org/ark:/48223/pf0000249568

466 11. Fuchs R, Pamuk E, Lutz W. Education or wealth: which matters more for reducing child

467 mortality in developing countries? Vienna Yearb Popul Res [Internet]. 2010 [cited 2019

468 Feb 5];8:175–99. Available from: http://hw.oeaw.ac.at?arp=0x0024fc44

469 12. Pamuk ER, Fuchs R, Lutz W. Comparing Relative Effects of Education and Economic

470 Resources on Infant Mortality in Developing Countries. Popul Dev Rev [Internet]. 2011

471 Dec 1 [cited 2019 Feb 5];37(4):637–64. Available from:

472 http://doi.wiley.com/10.1111/j.1728-4457.2011.00451.x

473 13. Zimmerman E, Woolf S, Haley A. Understanding the Relationship Between Education

474 and Health: A Review of the Evidence and an Examination of Community Perspectives.

475 [Internet]. Agency for Healthcare Research and Quality. 2015 [cited 2019 May 2].

476 Available from: https://www.ahrq.gov/professionals/education/curriculum-

477 tools/population-health/zimmerman.html

478 14. ICF International. Demographic and Health Survey: Sampling and Household Listing

479 Manual. Calverton, Maryland; 2012.

480 15. United States Agency for International Development (USAID). The Demographic and

481 Health Surveys Program [Internet]. 2018. Available from: https://dhsprogram.com/

482 16. ICF International. MEASURE DHS Biomarker Field Manual [Internet]. Calverton,

217

483 Maryland, U.S.A; 2012 [cited 2017 Oct 20]. Available from:

484 https://dhsprogram.com/pubs/pdf/DHSM7/DHS6_Biomarker_Manual_9Jan2012.pdf

485 17. Swinburn BA, Sacks G, Lo SK, Westerterp KR, Rush EC, Rosenbaum M, et al.

486 Estimating the changes in energy flux that characterize the rise in obesity prevalence. Am

487 J Clin Nutr [Internet]. 2009 Jun [cited 2018 Sep 24];89(6):1723–8. Available from:

488 http://www.ncbi.nlm.nih.gov/pubmed/19369382

489 18. Swinburn B, Sacks G, Ravussin E. Increased food energy supply is more than sufficient to

490 explain the US epidemic of obesity. Am J Clin Nutr [Internet]. 2009 Dec 1 [cited 2018

491 Sep 24];90(6):1453–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19828708

492 19. Jensen MD, Ryan DH, Apovian CM, Ard JD, Comuzzie AG, Donato KA, et al. 2013

493 AHA/ACC/TOS Guideline for the Management of Overweight and Obesity in Adults.

494 Circulation [Internet]. 2014 Jun 24 [cited 2018 Aug 20];129(25 suppl 2):S102–38.

495 Available from:

496 http://circ.ahajournals.org/lookup/doi/10.1161/01.cir.0000437739.71477.ee

497 20. Onubi OJ, Marais D, Aucott L, Okonofua F, Poobalan AS. Maternal obesity in Africa: a

498 systematic review and meta-analysis. J Public Health (Bangkok) [Internet]. 2016 Sep

499 [cited 2017 Feb 2];38(3):e218–31. Available from:

500 http://www.ncbi.nlm.nih.gov/pubmed/26487702

501 21. Gunderson EP. Childbearing and obesity in women: weight before, during, and after

502 pregnancy. Obstet Gynecol Clin North Am [Internet]. 2009 Jun [cited 2018 Sep

503 2];36(2):317–32, ix. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19501316

504 22. Ford ND, Patel SA, Narayan KMV. Obesity in Low- and Middle-Income Countries:

505 Burden, Drivers, and Emerging Challenges. Annu Rev Public Health [Internet]. 2017 Mar

218

506 20 [cited 2018 Aug 2];38(1):145–64. Available from:

507 http://www.ncbi.nlm.nih.gov/pubmed/28068485

508 23. Lovejoy JC, Sainsbury A, Stock Conference 2008 Working Group. Sex differences in

509 obesity and the regulation of energy homeostasis. Obes Rev [Internet]. 2009 Mar [cited

510 2018 Aug 31];10(2):154–67. Available from:

511 http://www.ncbi.nlm.nih.gov/pubmed/19021872

512 24. Frantz JM, Ngambare R. Physical activity and health promotion strategies among

513 physiotherapists in Rwanda. Afr Health Sci [Internet]. 2013 Mar [cited 2018 Dec

514 21];13(1):17–23. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23658563

515 25. Ibrahim FM, Jegede AS. Tradition and Limits: Polemical Construction of Body Size

516 Among the Yoruba of Southwestern Nigeria. J African Am Stud [Internet]. 2017 Jun 2

517 [cited 2018 Dec 21];21(2):236–55. Available from:

518 http://link.springer.com/10.1007/s12111-017-9360-x

519 26. Berman BJ. Ethnicity, patronage and the African state: The politics of uncivil nationalism.

520 Afr Aff (Lond) [Internet]. 1998 Jul 1 [cited 2019 Jan 10];97(388):305–41. Available from:

521 https://academic.oup.com/afraf/article-lookup/doi/10.1093/oxfordjournals.afraf.a007947

522 27. Coffey D. Prepregnancy body mass and weight gain during pregnancy in India and sub-

523 Saharan Africa. Proc Natl Acad Sci U S A [Internet]. 2015 Mar 17 [cited 2017 Oct

524 20];112(11):3302–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25733859

525 28. United States Agency for International Development (USAID). Standard Recode Manual

526 for DHS 6 [Internet]. 2013. Available from:

527 https://www.dhsprogram.com/pubs/pdf/DHSG4/Recode6_DHS_22March2013_DHSG4.p

528 df

219

529 29. Sterba SK. Alternative Model-Based and Design-Based Frameworks for Inference From

530 Samples to Populations: From Polarization to Integration. Multivariate Behav Res

531 [Internet]. 2009 Nov 1 [cited 2018 Nov 21];44(6):711–40. Available from:

532 http://www.ncbi.nlm.nih.gov/pubmed/20411042

533 30. SAS. SAS/STAT Software: Survey Analysis [Internet]. [cited 2018 Nov 21]. Available

534 from: https://support.sas.com/rnd/app/stat/procedures/SurveyAnalysis.html

535 31. West BT, Beer L, Gremel GW, Weiser J, Johnson CH, Garg S, et al. Weighted Multilevel

536 Models: A Case Study. Am J Public Health [Internet]. 2015 Nov [cited 2018 Nov

537 21];105(11):2214–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26378830

538 32. SAS Institute. The SAS System for Windows. Cary, NC: SAS Institute; 2011.

539 33. Wang R, Lagakos SW, Ware JH, Hunter DJ, Drazen JM. Statistics in Medicine —

540 Reporting of Subgroup Analyses in Clinical Trials. N Engl J Med [Internet]. 2007 Nov 22

541 [cited 2019 Jan 3];357(21):2189–94. Available from:

542 http://www.nejm.org/doi/abs/10.1056/NEJMsr077003

543 34. Pagano M, Gauvreau K. Principles of biostatistics. Duxbury; 2000.

544 35. Berglund PA. Multiple Imputation Using the Fully Conditional Specification Method: A

545 Comparison of SAS®, Stata, IVEware, and R. In SAS Institute; 2015 [cited 2018 Jun 17].

546 p. 1–17. Available from: https://support.sas.com/resources/papers/proceedings15/2081-

547 2015.pdf

548 36. U.S. Department of Health and Human Services. The Health Consequences of Smoking -

549 50 Years of Progress: A Report of the Surgeon General [Internet]. 2014 [cited 2019 Jan

550 10]. Available from: www.cdc.gov/tobacco

551 37. Gilman SE, Martin LT, Abrams DB, Kawachi I, Kubzansky L, Loucks EB, et al.

220

552 Educational attainment and cigarette smoking: a causal association? Int J Epidemiol

553 [Internet]. 2008 Jun [cited 2019 Jan 10];37(3):615–24. Available from:

554 http://www.ncbi.nlm.nih.gov/pubmed/18180240

555 38. Pierce JP, Fiore MC, Novotny TE, Hatziandreu EJ, Davis RM. Trends in Cigarette

556 Smoking in the United States. JAMA [Internet]. 1989 Jan 6 [cited 2019 Jan 10];261(1):56.

557 Available from:

558 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.1989.03420010066034

559 39. Kamangar F. Effect modification in epidemiology and medicine. Arch Iran Med

560 [Internet]. 2012 Sep [cited 2019 Jan 10];15(9):575–82. Available from:

561 http://www.ncbi.nlm.nih.gov/pubmed/22924379

562

563

564

565

221

566 Supplementary Appendix 567

568 THE EFFECT MODIFICATION OF EDUCATIONAL ATTAINMENT ON THE

569 ASSOCIATION BETWEEN HOUSEHOLD WEALTH AND OVERWEIGHT IN WOMEN IN

570 LOW AND MIDDLE-INCOME SUB-SAHARAN AFRICAN COUNTRIES: A CROSS-

571 SECTIONAL STUDY

572

573

574

575

576

577

578

579

580

581

582

583

584

585

222

Table 1: Prevalence of Overweight by Wealth Tertiles in 34 SSA Countries Overweight - N Country Wealth Tertiles Under/normal weight - N (%) (%) Low-Income Countries Benin Poorest 3,583 (35.5) 740 (7.9) Middle 3,184 (32.3) 1,119 (28.5) Rich 2,564 (32.2 1,639 (53.6)

Burkina Faso Poorest 2,166 (38.4) 106 (13.8) Middle 2,033 (34.5) 134 (16.1) Rich 1,635 (27.0) 543 (70.1)

Burundi Poorest 1,131 (40.1) 41 (15.4) Middle 1,062 (38.8) 71 (28.2) Rich 838 (21.2) 268 (56.4)

Chad Poorest 2,602 (37.6) 240 (23.8) Middle 2,502 (32.6) 216 (20.6) Rich 2,252 (29.7) 555 (55.5)

Comoros Poorest 861 (36.8) 513 (27.8) Middle 795 (33.3) 573 (35.6) Rich 725 (29.9) 635 (36.7)

Congo DRC Poorest 2,182 (31.8) 180 (12.5) Middle 2,137 (36.7) 248 (20.2) Rich 1,642 (31.5) 642 (67.3)

Ethiopia Poorest 3,366 (30.3) 156 (10.2) Middle 3,804 (53.1) 227 (24.3) Rich 2,686 (16.5) 1,135 (65.4)

Gambia Poorest 1,048 (31.5) 218 (19.4) Middle 913 (32.6) 258 (29.7) Rich 766 (35.9) 400 (50.9)

Guinea Poorest 1,124 (39.5) 97 (12.6) Middle 1,001 (34.6) 232 (30.6) Rich 756 (25.8) 413 (56.8)

223

Table 1: Prevalence of Overweight by Wealth Tertiles in 34 SSA Countries Overweight - N Country Wealth Tertiles Under/normal weight - N (%) (%)

Liberia Poorest 1,040 (25.4) 243 (15.1) Middle 903 (26.9) 276 (17.7) Rich 734 (47.8) 458 (67.3)

Madagascar Poorest 2,182 (33.9) 52 (10.0) Middle 2,124 (39.4) 100 (21.7) Rich 1,761 (26.6) 356 (68.3)

Malawi Poorest 1,894 (41.3) 278 (20.8) Middle 1,675 (33.9) 438 (29.2) Rich 1,315 (24.8) 843 (50.0)

Mali Poorest 1,236 (39.0) 166 (20.9) Middle 1,173 (38.0) 214 (28.3) Rich 869 (23.0) 495 (50.8)

Mozambique Poorest 3,234 (48.7) 248 (15.3) Middle 2,839 (33.8) 631 (30.6) Rich 2,004 (17.5) 1,437 (54.1)

Niger Poorest 1,193 (43.5) 168 (24.9) Middle 1,128 (40.3) 220 (28.8) Rich 713 (16.2) 508 (46.3)

Rwanda Poorest 1,587 (39.7) 246 (20.8) Middle 1,391 (35.4) 334 (28.6) Rich 1,120 (24.9) 683 (50.6)

Senegal Poorest 1,289 (26.2) 196 (13.1) Middle 1,201 (29.2) 307 (23.7) Rich 1,069 (44.6) 429 (63.2)

Sierra Leone Poorest 1,828 (38.0) 274 (21.3) Middle 1,699 (34.9) 385 (28.6) Rich 1,432 (27.1) 651 (50.0)

224

Table 1: Prevalence of Overweight by Wealth Tertiles in 34 SSA Countries Overweight - N Country Wealth Tertiles Under/normal weight - N (%) (%)

Tanzania Poorest 2,907 (42.7) 558 (17.2) Middle 2,461 (31.5) 999 (27.2) Rich 1,758 (25.7) 1,686 (55.7)

Togo Poorest 1,135 (31.8) 178 (11.7) Middle 913 (37.0) 354 (29.6) Rich 675 (31.2) 620 (58.7)

Uganda Poorest 633 (37.3) 45 (11.7) Middle 566 (39.2) 129 (35.6) Rich 410 (23.4) 261 (52.7)

Zimbabwe Poorest 1,938 (47.2) 705 (26.9) Middle 1,500 (31.5) 1,105 (35.0) Rich 1,192 (21.4) 1,347 (38.1) Lower-middle Income Countries Cameroon Poorest 1,631 (41.1) 382 (15.6) Middle 1,321 (31.2) 724 (32.8) Rich 1,032 (27.7) 1,001 (51.6)

Congo Poorest 1,280 (18.7) 178 (6.5) Middle 1,247 (31.2) 274 (18.9) Rich 906 (50.2) 529 (74.6)

Cote d'Ivoire Poorest 1,123 (37.7) 175 (16.3) Middle 970 (32.0) 289 (27.7) Rich 717 (30.3) 489 (56.0)

Ghana Poorest 1,039 (34.2) 230 (10.0) Middle 767 (36.8) 533 (31.8) Rich 515 (28.9) 761 (58.3)

Kenya Poorest 3,315 (32.1) 611 (11.6) Middle 2,721 (35.9) 1,284 (29.7) Rich 1,891 (32.1) 1,940 (58.6)

225

Table 1: Prevalence of Overweight by Wealth Tertiles in 34 SSA Countries Overweight - N Country Wealth Tertiles Under/normal weight - N (%) (%)

Lesotho Poorest 618 (37.4) 333 (21.3) Middle 449 (32.1) 482 (36.0) Rich 351 (30.5) 531 (42.7)

Nigeria Poorest 8,406 (40.9) 1,349 (16.4) Middle 7,183 (31.9) 2,585 (29.5) Rich 5,567 (27.2) 4,214 (54.1)

Sao Tome & Principe Poorest 435 (32.7) 190 (23.1) Middle 422 (31.7 ) 211 (26.9) Rich 355 (35.6) 288 (50.0)

Swaziland Poorest 678 (40.9) 616 (29.1) Middle 548 (33.8) 742 (35.6) Rich 459 (25.3) 812 (35.4)

Zambia Poorest 3,793 (38.7) 466 (14.6) Middle 3,365 (31.3) 921 (25.7) Rich 2,585 (30.0) 1,679 (59.7) Upper-middle Income Countries Gabon Poorest 1,006 (17.9) 437 (8.5) Middle 790 (32.3) 631 (31.2) Rich 679 (49.8) 722 (60.3)

Namibia Poorest 944 (41.5) 242 (17.2) Middle 766 (31.4) 496 (35.9) Rich 560 (27.1) 605 (46.9) 586

587

588

589

226

Table 2: Odds of Overweight by Wealth Tertile in 22 SSA Countries Country Overweight - Odds Ratio (95% CI) Low-income Countries Benin Poorest Ref Middle 1.63 (1.42 - 1.87) Rich 2.69 (2.31 - 3.14)

Chad Poorest Ref Middle 1.08 (0.82 - 1.42) Rich 1.61 (1.28 - 2.04)

Comoros Poorest Ref Middle 1.56 (1.27 - 1.91) Rich 1.79 (1.44 - 2.23)

Congo DRC Poorest Ref Middle 1.27 (0.95 - 1.72) Rich 3.91 (2.83 - 5.40)

Gambia Poorest Ref Middle 1.39 (1.10 - 1.75) Rich 1.74 (1.19 - 2.54)

Liberia Poorest Ref Middle 1.19 (0.93 - 1.52) Rich 3.16 (2.16 - 4.60)

Malawi Poorest Ref Middle 1.49 (1.25 - 1.79) Rich 3.00 (2.46 - 3.68)

Rwanda Poorest Ref Middle 1.43 (1.18 - 1.73) Rich 3.05 (2.56 - 3.62)

Sierra Leone Poorest Ref Middle 1.36 (1.11 - 1.66) Rich 2.51 (1.94 - 3.25)

227

Table 2: Odds of Overweight by Wealth Tertile in 22 SSA Countries Country Overweight - Odds Ratio (95% CI) Tanzania Poorest Ref Middle 2.12 (1.82 - 2.46) Rich 5.20 (4.13 - 6.55)

Togo Poorest Ref Middle 1.67 (1.29 - 2.16) Rich 3.53 (2.21 - 5.66)

Zimbabwe Poorest Ref Middle 1.96 (1.66 - 2.31) Rich 3.18 (2.51 - 4.02) Lower-middle Income Countries Cameroon Poorest Ref Middle 1.94 (1.60 - 2.36) Rich 3.34 (2.62 - 4.25)

Congo Poorest Ref Middle 1.40 (1.07 - 1.83) Rich 3.03 (2.22 - 4.16)

Ghana Poorest Ref Middle 2.56 (2.06 - 3.18) Rich 5.88 (4.34 - 7.96)

Kenya Poorest Ref Middle 1.87 (1.64 - 2.14) Rich 3.85 (3.33 - 4.45)

Lesotho Poorest Ref Middle 1.92 (1.48 - 2.49) Rich 2.64 (1.84 - 3.80)

Nigeria Poorest Ref Middle 2.19 (1.92 - 2.51) Rich 4.19 (3.60 - 4.87)

228

Table 2: Odds of Overweight by Wealth Tertile in 22 SSA Countries Country Overweight - Odds Ratio (95% CI)

Zambia Poorest Ref Middle 2.03 (1.73 - 2.37) Rich 4.95 (4.08 - 6.02)

Swaziland Poorest Ref Middle 1.81 (1.43 - 2.28) Rich 2.59 (2.04 - 3.29) Upper-middle Income Countries Gabon Poorest Ref Middle 1.88 (1.35 - 2.62) Rich 2.51 (1.80 - 3.50)

Namibia Poorest Ref Middle 2.56 (2.09 - 3.15) Rich 4.11 (3.17 - 5.33) 590

591

592

593

594

595

229

Table 3: Odds Ratios and 95% CIs for the Interaction of Educational Attainment on the Association between Household Wealth and Overweight in 22 SSA Countries** Country Wealth Quintile No/Primary Education Secondary/Higher Education Overweight - Odds Overweight - Odds Ratio Ratio (95% CI) (95% CI) Low-income Countries Benin Poorest Ref Ref Middle 1.65 (1.34 - 2.03) 1.06 (0.43 - 2.64) Rich 2.47 ( 2.16 - 3.48) 1.78 (0.78 - 4.10) Chad Poorest Ref Ref Middle 1.08 (0.73 - 1.58) 1.07 (0.34 - 3.38) Rich 1.65 (1.18 - 2.31) 1.30 (0.50 - 3.37) Comoros* Poorest Ref Ref Middle 1.37 (0.96 - 1.97) 2.07 (1.18 - 3.65) Rich 1.76 (1.12 - 2.75) 2.14 (1.27 - 3.59) Congo DRC Poorest Ref Ref Middle 1.37 (0.88 - 2.12) 1.87 (0.82 - 4.29) Rich 3.68 (2.27 - 5.97) 6.90 (3.17 - 15.01) Gambia Poorest Ref Ref Middle 1.33 (0.94 - 1.88) 1.68 (0.83 - 3.41) Rich 1.74 (1.04 - 2.92) 1.95 (0.96 - 3.99) Liberia* Poorest Ref Ref Middle 1.18 (0.84 - 1.66) 1.14 (0.48 - 2.71) Rich 3.22 (1.95 - 5.30) 2.85 (1.31 - 6.20) Malawi Poorest Ref Ref Middle 1.47 (1.15 - 1.87) 1.67 (0.85 - 3.27) Rich 3.02 (2.29 - 3.99) 3.12 (1.67 - 5.81) Rwanda* Poorest Ref Ref Middle 1.41 (1.08 - 1.84) 1.56 (0.70 - 3.47) Rich 3.08 (2.38 - 3.99) 3.04 (1.44 - 6.39) Sierra Leone

230

Table 3: Odds Ratios and 95% CIs for the Interaction of Educational Attainment on the Association between Household Wealth and Overweight in 22 SSA Countries** Country Wealth Quintile No/Primary Education Secondary/Higher Education Overweight - Odds Overweight - Odds Ratio Ratio (95% CI) (95% CI) Low-income Countries Poorest Ref Ref Middle 1.35 (1.01 -1.81) 1.24 (0.57 - 2.73) Rich 2.65 (1.84 - 3.82) 2.01 (0.97 - 4.16) Tanzania* Poorest Ref Ref Middle 2.13 (1.68 - 2.70) 1.61 (0.73 - 3.53) Rich 5.46 (3.80 - 7.84) 3.60 (1.72 - 7.53) Togo Poorest Ref Ref Middle 1.67 (1.17 - 2.83) 1.59 (0.70 - 3.59) Rich 3.67 (2.04 - 6.60) 3.16 (1.30 - 7.67) Zimbabwe* Poorest Ref Ref Middle 2.13 (1.51 - 3.00) 1.87 (1.39 - 2.52) Rich 3.52 (1.92 - 6.47) 3.05 (2.09 - 4.47) Lower-middle Income Countries Cameroon Poorest Ref Ref Middle 2.03 (1.51 - 2.72) 1.53 (0.96 - 2.44) Rich 3.76 (2.49 - 5.68) 2.76 (2.49 - 5.68) Congo Poorest Ref Ref Middle 1.60 (0.97 - 2.64) 1.21 (0.69 - 2.12) Rich 3.15 (1.74 - 5.74) 2.78 (1.62 - 4.75) Ghana Poorest Ref Ref Middle 2.59 (1.77 - 3.79) 2.41 (1.45 - 4.01) Rich 6.73 (3.86 - 11.75) 5.35 (3.18 - 9.01) Kenya Poorest Ref Ref Middle 1.96 (1.61 - 2.39) 1.52 (1.04 - 2.21) Rich 4.07 (3.18 - 5.21) 3.13 (2.19 - 4.49)

231

Table 3: Odds Ratios and 95% CIs for the Interaction of Educational Attainment on the Association between Household Wealth and Overweight in 22 SSA Countries** Country Wealth Quintile No/Primary Education Secondary/Higher Education Overweight - Odds Overweight - Odds Ratio Ratio (95% CI) (95% CI) Low-income Countries Lesotho* Poorest Ref Ref Middle 1.94 (1.26 - 2.99) 1.85 (1.18 - 2.91) Rich 2.80 (1.56 - 5.03) 2.52 (1.48 - 4.27) Nigeria+ Poorest Ref Ref Middle 2.29 (1.82 - 2.87) 1.59 (1.11 - 2.29) Rich 4.50 (3.42 - 5.91) 3.06 (2.12 - 4.42) Swaziland* Poorest Ref Ref Middle 1.94 (1.28 - 2.94) 1.61 (1.02 - 2.54) Rich 4.01 (2.19 - 7.34) 2.13 (1.41 - 3.22) Zambia+ Poorest Ref Ref Middle 2.02 (1.52 - 2.67) 1.94 (1.23 - 3.05) Rich 5.27 (3.73 - 7.44) 4.48 (2.77 - 7.24) Upper-middle Income Countries Gabon Poorest Ref Ref Middle 2.84 (1.75 - 4.60) 1.81 (1.07 - 3.04) Rich 3.94 (2.10 - 7.39) 2.18 (1.36 - 3.51) Namibia* Poorest Ref Ref Middle 2.68 (1.71 - 4.18) 2.47 (1.76 - 3.48) Rich 5.47 (2.58 - 11.59) 3.90 (2.65 - 5.75) **Analysis was adjusted for age, parity, residence and ethnicity *ethnicity variable not available in DHS dataset for adjustment +Ethnicity variables had too many level/categories and contextual information to combine levels was unavailable 596

232

597 Sensitivity Analysis

598 This sensitivity analysis was conducted to determine whether the differences in the study outcome between this study and previous

599 work (2,4) was due to differences in methodology. The reanalysis was undertaken for Gabon and Namibia with overweight and

600 obesity respectively as the outcomes and we arrived at the same conclusions.

601 Results for Overweight

602 Odds Ratio of Overweight with 1-unit Increase in Odds Ratio of Overweight with 1-unit Increase in 603 Wealth Quintile in Gabon, DHS 2012 (adjusted for Wealth Quintile in Gabon, DHS 2012 (no ethnicity) adjustment for ethnicity) 604 1.8 2

605 1.6 1.8 1.6 606 1.4 1.4 1.2 607 No/Primary 1.2 No/Primary 1 Education Education 1 608 0.8 Secondary/Higher Secondary/Higher Education 0.8 Education 0.6 609 0.6 0.4 0.4 610

0.2 Odds Ratio of Overweight Odds Ratio Overweight of and CI 95% Odds Ratio of Overweight Odds Ratio Overweight of and CI 95% 0.2 611 0 0

612 Figure 1: Plot of the Interaction of Education on the Association between Household Wealth Quintiles and Overweight in Gabon 613

233

614

615 Odds Ratio of Overweight with 1-unit increase in Wealth Quintile in Namibia, DHS 2013 (no adjustment for ethnicity 616 2.5 617

618 2

619 1.5 No/Primary Education

620 Secondary/Higher 1 Education 621

622 0.5 Odds Ratio of Overweight Odds Ratio Overweight of and CI 95% 623 0

624

625 Figure 2: Plot of the Interaction of Education on the Association between Household Wealth Quintiles and Overweight in Namibia

626

627

628

629

630

234

631 Results for Obesity

632 Results for Obesity Odds Ratio of Obesity with 1-unit Increase in Odds Ratio of Obesity with 1-unit Increase in 633 Wealth Quintile in Gabon, DHS 2012 (no Wealth Quintile in Gabon, DHS 2012 (adjusted for adjustment for ethnicity) ethnicity) 634 2 1.8

1.8 1.6 635 1.6 1.4 1.4 636 1.2 1.2 No/Primary No/Primary 637 Education 1 Education 1 Secondary/Higher 0.8 Secondary/Higher 638 0.8 Education Education 0.6 0.6 639 0.4

0.4 Odds Ratio obesity of and 95%CI Odds Ratio of Obesity Odds Ratio Obesity of and CI 95% 640 0.2 0.2 0 0

Figure 3: Plot of the Interaction of Education on the Association between Household Wealth Quintiles and Obesity in Gabon

235

641

Odds Ratio of Obesity with 1-unit Increase in Wealth Quintile in Namibia, DHS 2013 (no adjustment for ethnicity) 2.5

1.5 No/Primary Education

Secondary/Higher 1 Education

0.5 Odds Ratio of Obesity Odds Ratio Obesity of and CI 95%

Figure 4: Plot of the Interaction of Education on the Association between Household Wealth Quintiles and Obesity in Namibia

236

CHAPTER 6

A QUALITATIVE RESEARCH SYNTHESIS OF CONTEXTUAL FACTORS

CONTRIBUTING TO FEMALE OVERWEIGHT AND OBESITY OVER THE LIFE COURSE

IN SUB-SAHARAN AFRICA

Ifeoma D. Ozodiegwu, DrPH1, Mary Ann Littleton, PhD2, Megan Quinn, DrPH1, Richard

Wallace, PhD3, Christian Nwabueze MBBS1, Oluwaseun Famojuro MBBS1, Hadii M. Mamudu,

PhD4

1Department of Biostatistics and Epidemiology, East Tennessee State University, Johnson City,

Tennessee, United States of America

2Department of Community and Behavioral Health, East Tennessee State University, Johnson

City, Tennessee, United States of America

3Quillen College of Medicine Library, East Tennessee State University Johnson City, Tennessee,

United States of America

4Department of Health Services Management and Policy, East Tennessee State University

Johnson City, Tennessee, United States of America

Corresponding author: Ifeoma D. Ozodiegwu Email: [email protected]

237

Abstract

Objective

Adult women are disproportionately burdened by overweight and obesity in Sub-Saharan

African (SSA) countries and there is limited understanding of the sociocultural context of this problem. In this qualitative research synthesis, we aimed to surface contextual influences that potentially predispose adult women and adolescent girls to overweight and obesity.

Methods

PubMed, CINAHL, PsychInfo, PROQUEST, EMBASE, Web of Science were searched to locate qualitative research articles conducted in sub-Saharan African countries beginning in year 2000.

After assessment for eligibility and critical appraisal, 16 studies were included in the synthesis.

Textual data and quotes was synthesized using the methods proposed by the Joan Briggs Institute and a thematic analysis framework.

Results

The synthesized studies were conducted in South Africa, Ghana, Kenya and Botswana. The three overarching themes across these studies were body size and shape ideals, barriers to healthy eating, and barriers to physical activity with cultural and social factors as cross-cutting influences within the major themes. Culturally, the ideal African woman was expected to be overweight or obese, and voluptuous and this was associated with their identity. While overweight and obesity was not acceptable among adolescent girls, they also desired to be voluptuous. Healthy food choices among women and girls were hampered by several factors including the affordability

238

and peer pressure. Both adult women and girls experienced ageism and institutional discrimination as barriers to physical activity.

Significance

This is the first qualitative research synthesis to amplify the voices of women and girls in SSA countries highlighting the challenges they face in maintaining a healthy weight. Sociocultural, institutional and peer-related factors were powerful forces shaping body size preferences, food choices and participation in physical activity. Our study findings provide insights for the design of contextually appropriate interventions and lay the foundation for further research studies.

239

Introduction

The growing prevalence of overweight and obesity among Sub-Saharan African (SSA) women is a cause for concern. From 1975 to 2016, age-standardized adult female overweight and obesity prevalence rose by 24% and 12% to 39% and 15% respectively; whereas age-standardized adult male overweight and obesity in the same time period rose only by 15% and 5% to 23% and 6%

(1). Although biological differences may explain a portion of the disproportionate burden of overweight and obesity among SSA women (2), extant evidence suggest that the sociocultural context matters. For instance, cultural ideals that regard overweight and obese women as beautiful influence cultural practices such as pre-marital fattening or force-feeding performed to hasten physical development and marriage in adolescent girls (3). In an analysis of the 2001

Demographic and Health from Mauritania, where force-feeding is practiced, the authors found that obese women reported a greater prevalence of force-feeding (4). Greater comprehension and documentation of contextual factors that predispose women to overweight and obesity in SSA countries is critical to the development of interventions and well-informed research hypothesis.

Moreover, identifying factors contributing to overweight and obesity in SSA is essential to supporting efforts to halt the rise in obesity, one of the global voluntary targets outlined in the

World Health Organization (WHO) Global Action Plan for the Prevention and Control of Non- communicable Diseases (NCD) (5). Additionally, given the link between overweight and obesity, various chronic conditions, and premature mortality (6,7), achieving progress for obesity-related targets will likely feed into the accomplishment of other NCD-related targets including the Sustainable Development Goal 3.4, which is to reduce premature mortality from

NCDs by one-third (8).

240

One of the overarching principles of the aforementioned WHO Global Action Plan is the recognition that a life-course approach to the prevention and control of NCDs is crucial (5).

Adopting a life-course lens permits the identification of potential adiposity risk factors and prevention needs at multiple life stages. This is consequential because adolescence is potentially a critical period for targeting interventions to reduce the likelihood of later life adiposity and

NCDs. A systematic review and meta-analysis of studies suggests that roughly 80% and 70% of obese adolescents remain obese in adulthood and after age 30 respectively (9), indicating that efforts to decrease the burden of adult female overweight and obesity must also target contributing factors in adolescents. A systematic presentation of overweight and obesity determinants in adolescent girls and adult women in SSA will enable ease in the identification of parallels and divergence to inform research, programs and policy.

Using a qualitative research synthesis approach, also known as meta-aggregation, permits consolidation and concise presentation of evidence from various qualitative studies (10). The complete focus on qualitative studies in a meta-aggregation aids deeper exploration and comprehension of complex human phenomenon in order to interpret this phenomenon through the lens of the lived experiences of individuals in their natural setting (11). Additionally, the consistency of human experience and social context is explored and program planners gain insight into influences on human behavior and information to design and implement contextually appropriate interventions. For instance, a Cochrane Systematic Review used a qualitative research synthesis method to assess how context affects the implementation of lay health worker programs so as to improve maternal and child health care access (12). Subsequently, the results of this review was used by the WHO to develop guidelines with suggestions to guide

241

implementation in various contexts (13,14). Further, the qualitative research synthesis methodology allows qualitative research to be subject to critique, informing and improving research practice (10).

Qualitative evidence is especially critical in overweight and obesity prevention at this time because a recent evaluation study suggested that environmentally focused policy interventions, often touted as the most-effective obesity prevention measure, are falling short of expectations in high-income countries (15). Insufficient consideration of contextual factors in designing and implementing policy solutions may account for this surprising finding. Therefore, as SSA countries progress towards embracing policy interventions for the prevention of overweight and obesity, other metabolic conditions, and NCDs in general, the findings of this study could possibly inform that work. As such, this qualitative evidence synthesis aims to identify contextual factors that influence increasing overweight and obesity among adolescent girls and adult females in SSA.

Methods

Search Strategy

An exhaustive search of six databases was conducted in PubMed, CINAHL, PsychInfo,

PROQUEST, EMBASE, Web of Science - using a combination of search terms including

“overweight”, “obesity”, “Africa”, at various time points between November 18th, 2018 and

January 13th, 2019 to identify journal articles and dissertations related to the study aims. An example of the search strategy for PubMed and CINAHL can be viewed in the Appendix. Search findings were subsequently filtered using terms such as “qualitative” and restricted to studies

242

conducted from year 2000 to ensure that they were relevant to the current time period. Due to the difficulty in narrowing the search in PROQUEST, search results were limited to the first five pages.

Inclusion Criteria

Solely primary qualitative studies written in English language were included in this study. Using the PICo mnemonic proposed by the Joanna Briggs Institute (JBI) (10), the inclusion criteria were outlined as follows:

1. Population: Healthy adolescent and adult females without regard for educational status,

social status and ethnicity. Adolescents were defined as individuals aged ten to nineteen

in accordance with the WHO standards (16). Studies with male participants were

excluded except when the results of the study were reported separately for females, or

when the male participants report their perceptions of factors associated with female

overweight and obesity.

2. Context: Studies must be conducted within SSA countries.

3. Phenomena of Interest: contextual factors contributing to female overweight and obesity.

Screening

The database search yielded 533 records. After the removal of duplicates and studies conducted prior to 2000 (n=87), a total of 446 records remained. Two reviewers evaluated the titles and abstracts against the inclusion criteria resulting in the exclusion of 419 records. The full-text of the remaining 27 articles were independently evaluated for eligibility by two reviewers using the

243

inclusion criteria. A total of 9 articles were excluded leaving 18 articles for inclusion in the critical appraisal.

Critical Appraisal

Two reviewers independently evaluated the methodologic appropriateness, quality and rigor of the remaining 18 articles using the Critical Appraisal Skills Program tool for Qualitative research. The tool provides 10 questions for assessing the following: the clarity of the research aims and the findings, the appropriateness of the methodology, recruitment, data collection, data analysis, and ethical procedures, and researcher’s reflexivity. However, due to the dependence of most of the appraisal criteria on journal reporting requirements and page limits, it was decided that articles will be only be excluded if the response was negative for the following questions : 1) was there a clear statement of the aims of the research? 2) is the qualitative methodology appropriate? 3) have ethical issues been taken into consideration? 4) is there a clear statement of findings? Based on the critical appraisal exclusion criteria, two articles were eliminated because they lacked information on ethical approval for their studies. Consequently, 16 articles were utilized for the qualitative evidence synthesis. The reporting of this review followed the principles recommended in the Preferred Reporting in Systematic Reviews and Meta-Analyses

(PRISMA) statement (16) and a flow diagram is presented in Figure 1.

244

Figure 1: PRISMA Flow Diagram

Records identified through

database searching (n = 533)

Identification

Duplicates removed (n = 87) Records excluded based on title Records screened for title and abstract and abstract screening (n = 419) Screening (n = 446)

Articles excluded (n = 9)  Conference abstract (3)  Results of the female participants not reported separately (2)

Full-text articles assessed for  Quantitative study (1) ity eligibility  Mixed-method study with results (n = 27) of the qualitative study was not

reported (1) Eligibil  Unrelated to the study aim (1)  Insufficient reporting of the qualitative study method and results (1)

Studies included in critical Articles excluded (n = 2) appraisal (n = 18)  No information on ethical approval

Critical Critical Appraisal

Studies included in qualitative research synthesis (n = 16)

Included

245

Data Extraction

A pre-piloted standard form was used by one reviewer to extract the key study details, the demographic characteristics of the participants (where available), and the main study findings for all 16 included studies. Using the method outlined by JBI (10), we considered findings to be the demographic information included in the article and results of a thematic analysis with unequivocal level of plausibility, defined as findings that cannot be disputed because they are accompanied by illustrative quotes and/or text that are congruent with the study questions.

Repeated readings of the thematic analysis was undertaken by two reviewers and agreement on the validity of the extraction by a single reviewer was reached with a second reviewer after comparison with the material in the included articles. In accordance with the focus of this research synthesis, pertinent study findings extracted from each article were specific to cultural and social norms, behavioral and ecological risk factors, perceived barriers and attitudes reported by study participants that potentially predispose them to overweight and obesity.

Data Synthesis

In line with some of the recommendations of the JBI (10) and using a thematic analysis framework (17), extracted findings were categorized by concept similarity. Themes were arrived at through a consensus process between two reviewers. Themes were identified by assessing the prevalence of a concept across multiple studies. Finally, synthesized findings were created by summarizing and re-conceptualizing the information within each category and relating them to an appropriate theme.

246

Results

Description of the included studies

Fourteen out of the sixteen studies included in this meta-aggregation were primarily exploratory qualitative studies while two studies combined exploratory qualitative methods with quantitative methods (mixed-methods studies). Twelve studies were conducted in South Africa (18–29), two studies in Ghana (30,31) and the remaining two studies were conducted in Botswana (32) and

Kenya respectively (33). The Botswana and Kenya studies were PhD dissertations while the rest of the articles were journal articles. Focus group discussions and semi-structured in-depth interviews were the primary methods used for data collection and thematic analysis was used for textual analysis in all the included studies. Sample sizes varied according to the data collection methods ranging from eight participants in in-depth interviews to 60 participants in focus group discussions. Although two studies were based on the same sample both were included in this synthesis because they provided differing levels of detail on the findings of the study (18,21). In the presentation of the findings of the qualitative synthesis, the references for each quote was placed after the quote. The supplementary table provides the main findings of each study and further details.

Data Synthesis Findings

Body size and shape ideals

There are cultural expectations for adult SSA women to have large and voluptuous bodies

(20,26,29,30). Notably, men perceived the ideal body weight for women to be overweight or obese, whereas women perceived that of men to be normal weight or overweight (20). One woman clearly described these cultural expectations of female body sizes "according to our

247

values and culture, it is important for a woman to have a large body. It makes you to be respected” (Obese South African woman) (20). Women perceived fatness to be part of the

African cultural identity (30). Moreover, women also encountered pressure from family and friends to attain normative cultural body sizes. "My auntie has been asking me "why, don't your parents give you food to eat? Come to my house to eat" She says as a woman I should eat and put on some flesh to look good. Every time she sees me she says the same thing" (Normal weight

Ghanaian woman) (30).

Women who were larger were regarded as more attractive "Our culture says that we are supposed to be fat. You must have structure, you must be beautiful. In other words, our culture does not allow women to be thin. They say that someone who is fat is sexier” (South African focus group) (29). However, there were thresholds to fatness and women also had to be voluptuous – large hips and bosoms, and small midsections – in order to be considered attractive

"They want to see features as far as the Ghanaian woman is concerned. At least with some hips to keep the "coca-cola" shape kind at least to appear nice in your outfit. Yeah, I think that is it but not so fat" (Normal weight Ghanaian Woman) (30). Additionally, being overweight was regarded by some women as a sign of wealth, happiness, eating a great diet and the absence of problems "If a person is fat (overweight) we usually assume she is happy and has (lots of money). It is evident that he/she eats nicely, and a lot, and not having problems.... ( South

African Overweight Woman) (20).

Consequently, women experienced body image dissatisfaction and desired to gain more weight

"This is not my 'normal' weight. I would be happy and I will look more attractive, if I can gain

248

more weight" (Normal weight South African woman) (20). However, there were few dissenting voices with some participants expressing a disinterest in being overweight or obese and others suggesting that cultural acceptance for overweight and obesity is changing with increasing knowledge of the associated health risks (20,29).

Nonetheless, these cultural ideals of beauty, the positive symbolism associated with overweight and community experiences with morbidity from HIV/AIDS led to a perceived link between weight loss and HIV or AIDS infection and weight gain with recovery “These days, most people, especially the youth, are losing weight because of the HIV or AIDS…” "If you have AIDS and suddenly gain weight, people believe that you are well again. I know a girl who is fat and pretty. Now you cannot say she was sick" (South African women) (26). As such, women were resistant to losing weight to avoid being perceived as someone with a HIV infection even though they were aware of the health conditions linked to being overweight “She would not be happy [to lose weight] because in our community it would be perceived that there is something wrong with her. She would think that the community is looking at her. So she would feel insecure” (South

African focus group) (29). HIV infected persons were perceived negatively and experienced stigma, hence, weight retention was a way of protecting their reputation “In this community, people with HIV or AIDS are believed to sleep around and no-one wants people to say filthy things like that about them” (South African focus group) (26).

Similar but more complicated phenomenon was observed among adolescents. Thinness signified

HIV infection and influenced the willingness of obese girls to lose weight (19,23,32).

Additionally, there was a strong peer influence on body size and shape ideals. As described by

249

one adolescent girl those who do not fit into the preferred curvaceous ideal were shamed and labeled as HIV infected persons "Sometimes you find that the other girls is having curves and you don't have it. If she saw you she is going to call you by all names saying that you don't have curves, you're thin as if you're HIV positive. you don't have buttocks”(South African adolescent girl) (19). However, unlike adult women, overweight and obesity was not socially accepted and girls with such body sizes were also bullied "…….usually when you around this from 6 to 10 when your body is this big you are more, as an adolescent you are more, you are going to be teased like you are likely to get your feelings hurt…..they are going to tease you because you have grown too fat, they isolate you” (Botswanan adolescent girl) (32). Further, muscularity in girls was linked with poverty and muscular girls were bullied "oh, they [the boys] laugh at them

[the girls]. They told them that they have 'mapotirsi' [muscle], and they have a hard skin because they work hard at home (South African adolescent girl) (19). Nonetheless, parental perceptions also played a role in shaping adolescent body size preferences "My mother prefers a fat person to a slender person because she said when a person is slender it looks like that person is ill so because of that I ended up feeling proud for being fat…” (South African adolescent girl) (23).

Barriers to healthy eating

Both adult women and adolescents experienced barriers to healthy eating (18,21,24,27,29,32).

For Congolese and Zimbabwean migrant women in South Africa being away from home prevented them from making their healthy traditional meals, which were often plant based, leading them to compensate with fast foods "There in Zimbabwe I used to eat our own traditional foods. [....] Yes, but with this one it was totally different. [with this pregnancy] I wanted sweet stuff, fast foods. I almost ate fast food for this one I didn't cook for myself. I was

250

lazy even here at home. I think it is because.... it is difficult for me to have our own traditional food. So I had to force myself to have an alternative" (Zimbabwean woman) (24). For other women, the affordability of fruits and vegetables was a barrier to eating a more balanced diet

“The cucumber and veggies and stuff are expensive. All the food with fat are cheap, vetkoek, derms, pork” (South African woman) (29).

Adolescents experienced peer pressure to conform to the food choices of their friends "For example when you are a group of friends and you are eating your healthy meal of like green salad and then everybody is having like pies, fizzy drinks, people will be looking like 'Wow!

Wow! Eish!"(Botswanan adolescent girl) (32). Additionally, adolescents lacked autonomy in meals eaten at home, which were not always healthy (18). Moreover, the food environment at schools attended by adolescents comprised of unhealthy foods, fruits were expensive and the free meals provided by schools were reported to be unpleasant "During first school break we buy kotas and cold drink, the next one crisps, the other one during study time we buy sweets and chocolate"(South African adolescent girl) (21). "I don't feel good about the free food we get at school, because they don't cook well. After eating it, I have stomach cramps, so we decided to stop eating the free food at school. If we don't have money for lunch, we just walk around the schoolyard until lunch is over; if we have some money we buy vetkoek and niknaks (from vendors)…..." (South African adolescent girl (27).

Barriers to engaging in physical activity

Adult women and adolescents encountered numerous personal, cultural, community level and institutional barriers to engaging in physical activity (18–20,22,25,28,29,31,33). Time

251

constraints, the potential for self-injury and lack of self-efficacy were some of the personal concerns raised by adult women. "It takes a lot of time in the day and we don't have that time………." "I try to do exercises that are safe for me, for my age, but I don't always know what

I can and cannot do……….." (Ghanaian women) (31).

For adolescents, it was perceived that the potential for sexual violence was an important safety issue “There are people who are observing when we go to work or other places; when it is a woman and she is left alone, they are thinking of going there to rape her” (South African Male

Youth Leader) (19). Pregnant women believed that physical activity could potentially harm them or their unborn child "Maybe if you run a lot you can make the baby sore or if you change direction all of a sudden or you run into something you can hurt your baby" (South African woman) (22). They also reported receiving inadequate physical activity guidance from health care professionals (22). The paucity of recreational facilities for physical activity was a common complaint among both adult and adolescent women (18,22,29). However, even those that had access to a gym lacked the knowledge necessary to use the equipment and found it daunting to ask questions because it was mostly frequented by athletic young men (31). Moreover, it was generally perceived that exercise was an activity for children and participation by adolescents and adult women was atypical. “They want to exercise to be fit or for sport. But those who are looking think they are crazy sweating, running in the sun looking crazy while we are drinking our beer.” "It's not important what they are doing because they look at the person and undermine them and think that this person is too old for the thing that he/she is doing” (South

African female focus groups) (29). Household chores and domestic responsibilities was the mostly widely accepted form of physical activity (25,33).

252

Similar age-based discrimination in physical activity was also reported to exist in schools "Even at school when you want to participate in athletics, they will tell you that you are over-age, so you become discouraged” (16 – 19 year old South African adolescent girls’ focus group) (19).

Additionally, institutions and key stakeholders perpetuated gender inequalities in physical activity participation through inadequate maintenance and poor support for female sporting events. “Our soccer pitch for boys is good, but for girls it's not good because the [netball] ground is inside the school yard and their place is too small.” “Mr. 'M' is an owner of S

[business] and he is sponsoring all the schools. He concentrates on the under-21s only, and he is doing it for boys only. Girls are suffering when it comes to competition and they don't have enough games, their games are very scarce...... To have sponsor for girls it is scarce, especially for netball or ladies soccer. Girls are left behind but for boys everything is going well” (South African Male Sports Teachers) (19). Finally, overweight adolescent girls were on the receiving end of weight bias from school officials who negatively perceived their sporting abilities and restricted from participating in school events (28).

Discussion and Conclusion

To the best of our knowledge, this is the first qualitative research synthesis surfacing the contextual factors that predispose adult women and adolescent girls to overweight and obesity in sub-Saharan African countries. Our findings were grouped into three themes consisting of body shape and size ideals, barriers to health eating, and barriers to physical activity. However, we found that cultural and social norms were connecting webs shaping attitudes, perceptions, and behaviors related to body size preferences, food choices and recreational physical activity for

253

women and girls. Cultural conception of ideal adult female bodies as large and/or voluptuous was consistent across studies in the four countries featured in this synthesis (20,30,32,33), and seemingly across both low and high socioeconomic status (30,32) despite the low HIV prevalence in Ghana (1.7%) and Kenya (4.8%) (1), suggesting its possible long-standing nature within SSA countries. As illustrated by Eknoyan, positive connotations of female adiposity are not novel and were highly prevalent in the West up until the early 20th century (34). Eknoyan argued that such ideas resulted from the evolutionary advantage bestowed by excess fat in conditions of famine (34). Therefore, given that many SSA countries remain in low-income status or transitioned from low income to middle-income status in recent years (35), it is unsurprising that such ideas still persist. However, such cultures become detrimental as they come in contact with the increasing globalization of food systems coupled with the massive marketing of obesogenic foods, and the technological transition aiding the preponderance of sedentary lifestyles (36). Therefore, it is expected that the burden of overweight and obesity in

African countries will continue to soar unless urgent action is taken. This prediction is supported by findings from the African American community where large body size ideals exist and non-

Hispanic black women have the greatest obesity prevalence among all racial groups (37–39).

Additionally, the presence of HIV stigma and the notion that thinness and weight loss were associated with HIV infection within SSA countries will exacerbate the situation as we noted that this led to increased weight retention among overweight and obese women and girls and victimization of thin individuals (19,26,29).

Among adolescents, however, we noted a lower social acceptance of overweight and obesity expressed through weight-related victimization involving bullying and teasing from their peers

254

(19,23,32). Weight-related victimization, also termed weight stigma, have also been observed in the United States (40,41). In a study conducted among high school teens, the majority of respondents reported that overweight students were teased in a mean way, ignored, avoided and ostracized from social events (41). According to researchers, it is believed that overweight and obese individuals are lazy and weight stigma is considered a beneficial incentive for weight loss, which may explain the social stigma associated with adiposity (42–44). Nonetheless, this is in contrast to the research literature that shows that weight stigma precipitates deleterious effects such as binge eating, social isolation, decreased physical activity, and weight gain, worsening adiposity and the propensity to engage in healthy behaviors (45,46). Besides individual perception, however, the influx of western ideals may also account for the negative perception of overweight and obesity by adolescents. Swami (47) calls it the “globalization of the thin ideal” but given that thinness is typically associated with illness or HIV in South Africa and Botswana, the two countries from which studies of adolescents were included in this synthesis, it is possible that western body size ideals lies in tension with cultural ideals leading to somewhat “middle ground” in adolescent body size and shape ideals.

Addressing the cultural factors and social pressures linked to overweight requires the design and adoption of culturally sensitive interventions. The PEN-3 model is a theoretical framework used to foreground culture in the development, implementation and evaluation of health interventions

(48). For instance, in a depression-prevention intervention for urban African American and

Latino Adolescents, the PEN-3 model guided the identification of individual, interpersonal and community level barriers and promoters of depression and this information was applied to improve the visual and linguistic appeal of the program publications (49). With regards to our

255

study findings, the PEN-3 model could be applied to develop interventions that addresses cultural and social conceptions of ideal body sizes, to deconstruct myths surrounding the link between weight loss and HIV, as well as promote an environment where all community members are accepted irrespective of HIV and weight status. Moreover, peer bullying interventions that tackles body shaming and weight-related victimization in adolescents are warranted. However, novel interventions are needed as extant interventions show a small to medium effect on weight- biased attitudes and beliefs (49).

Our research synthesis also revealed a nutrition transition from traditional to convenience foods propelled by multiple barriers to making healthy food choices. Key factors shaping food choices included migration, cost, peer pressure, and the food environment (21,24,27,29,32).

Additionally, women and girls also encountered barriers to physical activity including gender- based discrimination in funding for female sporting events, ageism, the possibility of sexual violence and other safety concerns, ageism, and a paucity of facilities that support women’s engagement in physical activity (18,19,22,33). The WHO Global Action Plan recommends the development of national guidelines and provides several evidence-based policy and program options to promote healthy diets and physical activity (5), which has been endorsed by Member

States including SSA countries. The Food and Agriculture Organization also stresses action to improve information access on the nutritional content and health consequences of different foods to enhance better decision making in food purchases (50). Implementing these recommendations to achieve significant change will involve incorporating achievement of the targets of the

Sustainable Development Goal Target 5.9, which is to “adopt and strengthen sound policies and enforceable legislation for the promotion of gender equality and the empowerment of all women

256

and girls at all levels” (8). Rising overweight and obesity prevalence is intractable, for even advanced economies, with no reversals in the trajectory recorded in any country (51).

Nonetheless, given that several SSA countries are still in the nascent stages of the epidemic, an opportunity exists to tackle the problem before it spirals out of control.

To do this effectively requires collaboration with community members. Community-based participatory research methods provides opportunities for community members to participate in the design of interventions and it has been found to be effective in the prevention and control of obesity in African-American communities (37,48). The participation of the formal and informal food industry, institutions and health professionals is imperative. This can be achieved through policy initiatives that provide incentives for food systems to offer affordable, healthy and nutritious meals with the goal of propagating better food choices among women and girls, institutional action to create opportunities for women to participate in physical activity from childhood and onward, and the provision of appropriate nutrition and physical activity counsel by health care professionals especially among pregnant women who at risk of postnatal weight retention (47).

A key limitation of this synthesis is that only four SSA countries were examined which hampers the transferability of the findings to countries outside of those included in this study. It is therefore advised that caution should be exercised in the applications of the findings to a different SSA country. Additionally, the categorization of socioeconomic status by investigators in the included studies was not clear. However, the insights gained from this synthesis is useful for examining parallels in a unique setting.

257

In conclusion, this research synthesis of qualitative studies identified contextual factors that predispose adolescent girls and adult women to overweight and obesity in SSA countries. Body size and shape ideals, food choices and physical activity participation were influenced by several determinants including sociocultural factors, migration, ageism and institutional factors, self- efficacy and peer pressure. We also found a number of adverse effects from the institutionalization of body size and shape ideals including weight-related victimization among adolescents and stigmatization of thinness due to a perceived link with HIV. Community-based participatory research, policy design and implementation and multi-sectoral efforts to prevent and control female overweight and obesity in SSA communities is warranted.

258

References

1. World Health Organization. Global Health Observatory Data Repository [Internet]. WHO.

2017. Available from: http://apps.who.int/gho/data/?theme=main

2. Lovejoy JC, Sainsbury A, Stock Conference 2008 Working Group. Sex differences in

obesity and the regulation of energy homeostasis. Obes Rev [Internet]. 2009 Mar [cited

2018 Aug 31];10(2):154–67. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/19021872

3. Ouldzeidoune N, Keating J, Bertrand J, Rice J. A description of female genital mutilation

and force-feeding practices in Mauritania: implications for the protection of child rights

and health. PLoS One [Internet]. 2013 [cited 2018 Apr 21];8(4):e60594. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/23593257

4. Office National de la Statistique (ONS) [Mauritanie] and ORC Macro. Enquête

Démographique et de Santé Mauritanie 2000-2001. Calverton, Maryland, USA; 2001.

5. World Health Organization. Global action plan for the prevention and control of

noncommunicable diseases 2013-2020. [Internet]. World Health Organization. 2013.

Available from: http://apps.who.int/iris/bitstream/10665/94384/1/9789241506236_eng.pdf

6. Heymsfield SB, Wadden TA. Mechanisms, Pathophysiology, and Management of

Obesity. Longo DL, editor. N Engl J Med [Internet]. 2017 Jan 19 [cited 2018 May

31];376(3):254–66. Available from: http://www.nejm.org/doi/10.1056/NEJMra1514009

7. The GBD 2015 Obesity Collaborators. Health Effects of Overweight and Obesity in 195

Countries over 25 Years. N Engl J Med. 2017;377(1):13–27.

8. United Nations. Transforming our world: The 2030 agenda for sustainable development

[Internet]. Agenda. 2015. Available from:

259

https://sustainabledevelopment.un.org/content/documents/7891Transforming Our World.

pdf

9. Simmonds M, Llewellyn A, Owen CG, Woolacott N. Predicting adult obesity from

childhood obesity: a systematic review and meta-analysis. Obes Rev [Internet]. 2016 Feb

[cited 2018 Nov 18];17(2):95–107. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/26696565

10. Lockwood C, Munn Z, Porritt K. Qualitative research synthesis: Methodological guidance

for systematic reviewers utilizing meta-aggregation. Int J Evid Based Healthc [Internet].

2015 Sep [cited 2019 Jan 13];13(3):179–87. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/26262565

11. Ulin PR, Robinson ET, Tolley EE. Qualitative methods in public health : a field guide for

applied research. Jossey-Bass; 2005. 318 p.

12. Glenton C, Colvin C, Carlsen B, Swartz A, Lewin S, Noyes J, et al. Barriers and

facilitators to the implementation of lay health worker programmes to improve access to

maternal and child health: qualitative evidence synthesis. In: Glenton C, editor. Cochrane

Database of Systematic Reviews [Internet]. Chichester, UK: John Wiley & Sons, Ltd;

2013 [cited 2018 Sep 29]. Available from:

http://doi.wiley.com/10.1002/14651858.CD010414

13. World Health Organization. WHO recommendations: optimizing health worker roles to

improve access to key maternal and newborn health interventions through task shifting

[Internet]. Geneva, Switzerland ; 2012 [cited 2018 Sep 29]. Available from:

www.who.int/reproductivehealth

14. Noyes J, Booth A, Cargo M, Flemming K, Garside R, Hannes K, et al. Cochrane

260

Qualitative and Implementation Methods Group guidance series—paper 1: introduction. J

Clin Epidemiol [Internet]. 2018 May [cited 2018 Sep 29];97:35–8. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/29242094

15. Sisnowski J, Street JM, Merlin T. Improving food environments and tackling obesity: A

realist systematic review of the policy success of regulatory interventions targeting

population nutrition. Fürnsinn C, editor. PLoS One [Internet]. 2017 Aug 4 [cited 2018

Aug 25];12(8):e0182581. Available from:

http://dx.plos.org/10.1371/journal.pone.0182581

16. WHO. Global accelerated action for the health of adolescents ( AA-HA!): guidance to

support country implementation [Internet]. Geneva; 2017 [cited 2019 Jan 13]. Available

from: http://apps.who.int/iris/bitstream/handle/10665/255415/9789241512343-

eng.pdf?sequence=1

17. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol [Internet].

2006 Jan [cited 2019 Jan 23];3(2):77–101. Available from:

http://www.tandfonline.com/doi/abs/10.1191/1478088706qp063oa

18. Sedibe MH, Feeley AB, Voorend C, Griffiths PL, Doak CM, Norris SA. Narratives of

urban female adolescents in South Africa: dietary and physical activity practices in an

obesogenic environment. South African J Clin Nutr [Internet]. 2014;27(3):114–9.

Available from:

https://search.ebscohost.com/login.aspx?direct=true&AuthType=cookie,ip,url,uid,athens&

db=ccm&AN=103900200&site=ehost-live

19. Kinsman J, Norris SA, Kahn K, Twine R, Riggle K, Edin K, et al. A model for promoting

physical activity among rural South African adolescent girls. Glob Health Action

261

[Internet]. 2015;8:15. Available from: %3CGo

20. Okop KJ, Mukumbang FC, Mathole T, Levitt N, Puoane T. Perceptions of body size,

obesity threat and the willingness to lose weight among black South African adults: a

qualitative study. BMC Public Health [Internet]. 2016 Dec 29;16(1):365. Available from:

https://login.iris.etsu.edu:3443/login?url=https://search.proquest.com/docview/192474018

6?accountid=10771

21. Voorend CGN, Norris SA, Griffiths PL, Sedibe MH, Westerman MJ, Doak CM. “We eat

together; today she buys, tomorrow I will buy the food”: adolescent best friends’ food

choices and dietary practices in Soweto, South Africa. Public Health Nutr [Internet].

2013;16(3):559–67. Available from:

https://login.iris.etsu.edu:3443/login?url=https://search.proquest.com/docview/127213928

4?accountid=10771

22. Muzigaba M, Kolbe-Alexander TL, Wong FN. The Perceived Role and Influencers of

Physical Activity Among Pregnant Women From Low Socioeconomic Status

Communities in South Africa. J Phys Act Health [Internet]. 2014;11(7):1276–83.

Available from: %3CGo

23. Puoane T, Tsolekile L, Steyn N. Perceptions about body image and sizes among Black

African girls living in Cape Town. Ethn Dis [Internet]. 2010;20(1):29–34. Available from:

http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L35989

2003

24. Hunter-Adams J, Rother H-A. Pregnant in a foreign city: A qualitative analysis of diet and

nutrition for cross-border migrant women in Cape Town, South Africa. Appetite. 2016

Aug;103:403–10.

262

25. Phillips EA, Comeau DL, Pisa PT, Stein AD, Norris SA. Perceptions of diet, physical

activity, and obesity-related health among black daughter-mother pairs in Soweto, South

Africa: a qualitative study. BMC Public Health [Internet]. 2016 Dec 9;16(1):750.

Available from:

http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L61720

8387

26. Matoti-Mvalo T, Puoane T. Perceptions of body size and its association with HIV/AIDS.

South African J Clin Nutr [Internet]. 2011;24(1):40–5. Available from:

https://search.ebscohost.com/login.aspx?direct=true&AuthType=cookie,ip,url,uid,athens&

db=ccm&AN=104881536&site=ehost-live

27. Sedibe HM, Kahn K, Edin K, Gitau T, Ivarsson A, Norris SA, et al. Qualitative study

exploring healthy eating practices and physical activity among adolescent girls in rural

South Africa. BMC Pediatr [Internet]. 2014 Aug;14(1):211. Available from:

http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L60031

2661

28. Bodiba P, Madu SN, Ezeokana JO, Nnedum OAU. The relationship between body mass

index and self-concept among adolescent black female university students. Curationis

[Internet]. 2008 Mar;31(1):77–84. Available from:

http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L35228

3518

29. Draper CE, Davidowitz KJ, Goedecke JH. Perceptions relating to body size, weight loss

and weight-loss interventions in black South African women: a qualitative study. Public

Health Nutr. 2016 Feb;19(3):548–56.

263

30. Tuoyire DA, Kumi-Kyereme A, Doku DT, Amo-Adjei J. Perceived ideal body size of

Ghanaian women: “Not too skinny, but not too fat”. Women Health. 2018;58(5):583–97.

31. Tuakli-Wosornu YA, Rowan M, Gittelsohn J. Perceptions of physical activity, activity

preferences and health among a group of adult women in urban Ghana: a pilot study.

Ghana Med J. 2014 Mar;48(1):3–13.

32. Brown CD. Adolescent and adult perceptions of adolescent diet, physical activity, body

size, and obesity prevention in Botswana [Internet]. ProQuest Dissertations and Theses.

[Ann Arbor]: University of Pennsylvania; 2014. Available from:

https://login.iris.etsu.edu:3443/login?url=https://search.proquest.com/docview/154589589

3?accountid=10771

33. Mugo A. Obesity Among Women in Rural Kenya: Knowledge, Beliefs, and Perceptions

[Internet]. ProQuest Dissertations and Theses. [Ann Arbor]: Walden University; 2016.

Available from:

https://login.iris.etsu.edu:3443/login?url=https://search.proquest.com/docview/175887027

3?accountid=10771

34. Eknoyan G. A History of Obesity, or How What Was Good Became Ugly and Then Bad.

Adv Chronic Kidney Dis [Internet]. 2006 Oct [cited 2018 Mar 16];13(4):421–7. Available

from: http://www.ncbi.nlm.nih.gov/pubmed/17045228

35. World Bank. World Bank Country and Lending Groups [Internet]. 2018 [cited 2018 Feb

25]. Available from: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-

world-bank-country-and-lending-groups

36. Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT, Moodie ML, et al. The

global obesity pandemic: shaped by global drivers and local environments. Lancet

264

[Internet]. 2011 Aug 27 [cited 2018 Aug 1];378(9793):804–14. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/21872749

37. Hales CM, Carroll MD, Fryar CD, Ogden CL. Prevalence of Obesity Among Adults and

Youth: United States, 2015–2016. U.S. Centers for Diseas Control and Prevention,

National Center for Health Statistics. 2017.

38. Robinson SA, Webb JB, Butler-Ajibade PT. Body Image and Modifiable Weight Control

Behaviors Among Black Females: A Review of the Literature. Obesity [Internet]. 2012

Feb 14 [cited 2019 Feb 8];20(2):241–52. Available from:

http://www.ncbi.nlm.nih.gov/pubmed/21494225

39. Talleyrand RM, Gordon AD, Daquin J V., Johnson AJ. Expanding Our Understanding of

Eating Practices, Body Image, and Appearance in African American Women: A

Qualitative Study. J Black Psychol [Internet]. 2017 Jul 11 [cited 2019 Feb 8];43(5):464–

92. Available from: http://journals.sagepub.com/doi/10.1177/0095798416649086

40. Pearce MJ, Boergers J, Prinstein MJ. Adolescent Obesity, Overt and Relational Peer

Victimization, and Romantic Relationships. Obes Res [Internet]. 2002 May [cited 2019

Feb 7];10(5):386–93. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12006638

41. Puhl RM, Luedicke J, Heuer C. Weight-Based Victimization Toward Overweight

Adolescents: Observations and Reactions of Peers. J Sch Health [Internet]. 2011 Nov 1

[cited 2019 Feb 7];81(11):696–703. Available from: http://doi.wiley.com/10.1111/j.1746-

1561.2011.00646.x

42. Pont SJ, Puhl R, Cook SR, Slusser W. Stigma Experienced by Children and Adolescents

With Obesity. Pediatrics. 2017;

43. Puhl RM, Heuer CA. Obesity stigma: Important considerations for public health. Am J

265

Public Health. 2010;

44. Flint SW, Nobles J, Gately P, Sahota P, Association for the Study of Obesity, Obesity

Empowerment Network, et al. Weight stigma and discrimination: a call to the media.

lancet Diabetes Endocrinol [Internet]. 2018 Mar 1 [cited 2019 Feb 7];6(3):169–70.

Available from: http://www.ncbi.nlm.nih.gov/pubmed/29475497

45. Puhl R, Suh Y. Health Consequences of Weight Stigma: Implications for Obesity

Prevention and Treatment. Curr Obes Rep [Internet]. 2015 Jun 1 [cited 2019 Feb

7];4(2):182–90. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26627213

46. Tomiyama AJ, Carr D, Granberg EM, Major B, Robinson E, Sutin AR, et al. How and

why weight stigma drives the obesity ‘epidemic’ and harms health. BMC Med [Internet].

2018 Dec 15 [cited 2019 Feb 7];16(1):123. Available from:

https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-018-1116-5

47. Swami V. Cultural Influences on Body Size Ideals. Eur Psychol [Internet]. 2015 Jan [cited

2019 Feb 8];20(1):44–51. Available from: https://econtent.hogrefe.com/doi/10.1027/1016-

9040/a000150

48. Iwelunmor J, Newsome V, Airhihenbuwa CO. Framing the impact of culture on health: a

systematic review of the PEN-3 cultural model and its application in public health

research and interventions. Ethn Health [Internet]. 2014 Feb [cited 2019 Feb 7];19(1):20–

46. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24266638

49. Saulsberry A, Corden ME, Taylor-Crawford K, Crawford TJ, Johnson M, Froemel J, et al.

Chicago Urban Resiliency Building (CURB): An Internet-Based Depression-Prevention

Intervention for Urban African-American and Latino Adolescents. J Child Fam Stud

[Internet]. 2013 Jan 14 [cited 2019 Feb 7];22(1):150–60. Available from:

266

http://link.springer.com/10.1007/s10826-012-9627-8

50. Staatz J, Hollinger F. West African Food Systems and Changing Consumer Demands.

2016;(4). Available from: https://www.oecd-ilibrary.org/content/paper/b165522b-en

51. Swinburn BA, Kraak VI, Allender S, Atkins VJ, Baker PI, Bogard JR, et al. The Global

Syndemic of Obesity, Undernutrition, and Climate Change: The Lancet Commission

report. Lancet [Internet]. 2019 Jan [cited 2019 Jan 30];0(0). Available from:

https://linkinghub.elsevier.com/retrieve/pii/S0140673618328228

267

Supplementary Appendix

A QUALITATIVE RESEARCH SYNTHESIS OF CONTEXTUAL FACTORS

CONTRIBUTING TO FEMALE OVERWEIGHT AND OBESITY OVER THE LIFE COURSE

IN SUB-SAHARAN AFRICA

268

Examples of Search Terms

PubMed (results were filtered by “Qualitative Research”)

((("Obesity"[Mesh]) OR "Overweight"[Mesh]) AND "Africa South of the

Sahara"[Mesh]) Filters: Publication date from 2000/01/01

(("Nutritional Status"[Mesh] AND ("2000/01/01"[PDat] : "3000/12/31"[PDat] ))) AND "Africa

South of the Sahara"[Mesh] Filters: Publication date from 2000/01/01

(((("Policy"[Mesh] AND ("2000/01/01"[PDat] : "3000/12/31"[PDat] ))) AND "Obesity"[Mesh])

OR "Overweight"[Mesh]) AND "Africa South of the Sahara"[Mesh] Filters: Publication date from 2000/01/01

((((("Exercise"[Mesh] AND ("2000/01/01"[PDat] : "3000/12/31"[PDat] ))) AND

("Female"[Mesh] AND ( "2000/01/01"[PDat] : "3000/12/31"[PDat] ))) AND "Obesity"[Mesh])

OR "Overweight"[Mesh]) AND "Africa South of the Sahara"[Mesh] Filters: Publication date from 2000/01/01

(("Food"[Mesh] AND ("2000/01/01"[PDat] : "3000/12/31"[PDat] ))) AND "Africa South of the

Sahara"[Mesh] Filters: Publication date from 2000/01/01

(((("Attitude"[Mesh]) AND ((((overweight [Title/Abstract]) OR "Overweight"[Mesh]) OR obesity[Title/Abstract]) OR Adiposity[Title/Abstract])) AND (("Africa South of the

Sahara"[Mesh]) OR Africa[TIAB]))) AND "Qualitative Research"[Mesh]

(("Attitude"AND ((((overweight[Title/Abstract]) OR "Overweight"OR obesity[Title/Abstract])

OR Adiposity[Title/Abstract])) AND (("Africa South of the Sahara"OR Africa[TIAB])

269

(("Perception"AND ((((overweight[Title/Abstract]) OR "Overweight"OR obesity[Title/Abstract]) OR Adiposity[Title/Abstract])) AND (("Africa South of the Sahara"OR

Africa[TIAB])

(((("Perception"AND ((((overweight[Title/Abstract]) OR "Overweight"OR obesity[Title/Abstract]) OR Adiposity[Title/Abstract])) AND (("Africa South of the Sahara"OR

Africa[TIAB]))) AND "Qualitative Research"

(((((social norms[MH] OR "cultural norms"[TIAB] OR "cultural norm"[TIAB] OR culture[MH]

OR anthropology, cultural[MH] OR social values[MH] OR "social values"[TIAB] OR "social value"[TIAB])) AND ((((overweight[Title/Abstract]) OR "Overweight"[Mesh]) OR obesity[Title/Abstract]) OR Adiposity[Title/Abstract])) AND (("Africa South of the

Sahara"[Mesh]) OR Africa[TIAB]))) AND "Qualitative Research"[Mesh]

CINAHL

(Obesity OR Overweight AND Africa AND Qualitative) (Limiters - Full Text; Published Date:

20000101-; Geographic Subset: Africa; Language: English; Age Groups: Child: 6-12 years,

Adolescent: 13-18 years, Adult: 19-44 years

Expanders - Also search within the full text of the articles

Search modes - Boolean/Phrase)

(Attitude OR attitudes OR perspective OR orientation OR position OR Inclination OR opinion)

AND (overweight OR obesity OR Adiposity OR obese) AND Africa AND "Qualitative

Research"

270

(Perception OR image OR impression OR conception) AND (overweight OR obesity OR

Adiposity OR obese) AND Africa AND “Qualitative Research"

(“Social norms” OR "cultural norms" OR "cultural norm" OR culture OR “cultural anthropology” OR "social values" OR "social value") AND (overweight OR obesity OR

Adiposity OR obese) AND Africa AND "Qualitative Research"

(Knowledge OR information OR wisdom) AND (overweight OR obesity OR Adiposity OR obese) AND Africa AND "Qualitative Research"

271

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings Adolescents had a preference for convenience foods because it was enjoyable, accessible, affordable, filling and it can be shared with friends. Bringing lunch to school was seen as embarrassing for their age and because it was a mostly traditional food. Lunch from home was only acceptable if it was convenience meals such as burgers. Fruit sold at school was of poor quality and expensive. Fast foods or convenience foods was Exploratory also consumed at home and vegetables were not always Qualitative available for home meals. Caregivers often made the study/Duo semi- decision on what is consumed at home. There was lack structured Low SES, of opportunities and support for recreational physical interview of Urban, Grade activity for young women in the community. Safety and adolescent and 12, time were barriers to participating in physical activity. best friend/ Note Adolescent & However, few participants actively included vegetables Sedibe (2014)/South taking/Thematic 58 (29 Adult (15 - 21 during cooking and engaged in running, cycling and Africa/ Soweto analysis pairs) years) street dancing Low & High SES, Urban, Primary, secondary & Interviewees had a preference for larger body sizes university compared to images of western women and it was education. expected that all African women should look larger than Qualitative Most had western women. Ideal bodies had to be voluptuousness study/Semi- completed with a flat midsection. The perceived ideal body shape structured secondary looked attractive in outfits. Women encountered Tuoyire interviews/Audio- education, pressure from peers and family members to attain ideal (2018)/Ghana/Accra & recorded/Thematic Adult (33 +/- body size and as a result may increase their Tamale analysis 36 9.2 years) consumption of food.

272

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings Muscular girls were bullied by boys. Muscularity in girls is linked with poverty because it results from engaging in physical activity that is found among low SES (e.g. walking to school, doing housework). There were contrasting views of body size ideals. Ideal bodies also had to curvaceous. Gender inequalities existed in the opportunities to participate in sporting events. Boys had better facilities for sporting events. Boys also had Low SES, sports events that were better organized and funded. Rural, Concerns about sexual violence may also prevent Qualitative 51 Secondary women from engaging in physical activity. study/Focus group adolescent education, Additionally, physical activity was viewed as the discussions/Audio girls and 7 Adolescent domain of younger girls by peers and authority figures Kinsman (2015)/South recorded/Thematic adult key (13 - 19 and some respondents associated physical activity with Africa/Agincourt analysis informants years) health risks. Mixed methods Women were hindered from participating in physical study/Focus activity due to time and safety concerns. Older women groups/Semi- Urban, 40% were intimidated by fitness equipment and the male- structured in- college dominated gym. There were no physical activity role depth interviews educated, models for older women. Women were interested in the Tuakli-Wosornu /Thematic Adult (58 - 71 social elements of fitness and preferred group-based (2014)/Ghana/ Accra analyses Unavailable years) activities because of social interaction. Women were aware of the benefits of PA during Exploratory Low SES, pregnancy but had concerns about causing harm to their qualitative Low baby or themselves. Pregnancy discomforts also study/Focus educational prevented women from engaging in physical activity. Muzigaba (2014)/South groups/Audio- attainment, Other barriers to physical activity during pregnancy Africa/ Western Cape recorded/Thematic Adolescents included a lack of time due to household and work Province analysis 34 and Adults responsibilities, lack of facilities for exercising,

273

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings (17 - 36 neighborhood safety, the perception that exercise years) requires a gym and high SES and the lack of knowledge about places that organize exercises for pregnant women. The interviewees also would have liked to receive guidance and permission from a healthcare professional to engage in exercise. Fatness was associated with genetics, health, happiness, and wealth. Thinness was associated with infectious disease such as HIV, AIDS and or tuberculosis. Obese girls held the opinion that obesity was preferable because it allows one to engage in tasking activities, obese people look more respectable and obesity signifies good health. Contrastingly, thin girls were of the view that fatness was linked to chronic conditions like diabetes and hypertension. Obese girls also indicated that they had difficulty finding clothing in Low SES, their size. Despite the noted disadvantages of obesity, Urban, the authors found that interviewees were resistant to the Qualitative primary and idea of losing weight to avoid being stigmatized as study/Focus high school, someone with HIV/AIDS or TB. Losing weight is group/Audio- Adolescents linked to illness. Fat girls were often bullied and parents Puoane (2010)/South recorded/Thematic (10 - 18 had mixed views about fatness and thinness in their Africa/Cape Town analysis 60 years) children

274

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings Migrants from Congolese and Zimbabwe generally desire traditional foods from their home country such as leafy green, ground nuts and dried fish but were unable to access such foods in their new location because they were expensive and required forethought. Somalis also expressed difficulty accessing local foods from their home country with comparable quality to what is found in Somalia. Women ate fast foods during pregnancy to satisfy cravings and typically, pregnancy was the only 23 time their significant other was willing to cover the participants expenses related to fast food consumption. Fast foods for the in- was used to compensate for their longing for traditional Qualitative depth foods from their home countries. Women were study/In-depth interviews encouraged to satisfy any cravings they had during interviews/Focus and 48 Low SES, pregnancy. Fast foods was also considered healthy Hunter-Adams groups/audio- participants Urban, Adult because it could help combat nausea. The women felt (2016)/South Africa/ Cape recorded/thematic in the focus (20 - 40 that the nutrition advice given in the clinics was not Town analysis groups years) appropriate to their situation. Daughters had knowledge of health eating but did not think it was priority. Healthy food choices may be Ranged from motivated by adverse health experiences. Weight gain Grade 6 to was a sign of happiness and maturity. For some Bachelors, mothers, providing fast food for their children was a Adults means of providing a better upbringing for their Qualitative (daughters children. A possible protective factor for the daughters study/In-depth (24.2 yrs., was that they ate smaller or normal sized meals. For interviews/audio- 17 SD=0.04), some mothers, the reverse was the case, a better Phillips (2016)/South recorded/thematic daughters, mothers (53 upbringing for their children meant giving them healthy Africa/ Soweto analysis 15 mothers yrs. SD=4.9)) food. Participants viewed performing household duties

275

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings as a form of exercise. Mothers discouraged their daughters form participating in physical activity because they felt it was for younger children. Additionally, mothers were of the view that exercising in the gym should not be a priority of their daughters as it was a waste of money. As mother's got older, they become more health conscious and seek out healthy foods. Mothers also embraced exercise only as a response to a health issue. Mothers preferred large body sizes (US sizes 14 - 18) compared to daughters (US sizes 2 - 10). Many mothers were motivated to raise their children differently than how they were raised. HIV was highly prevalent in the community and infected people experience weight loss. Hence, thinness was associated with HIV. HIV was a highly stigmatized condition and no one wants to be associated with it. When HIV infected people gain weight, community Mixed methods members' outlook on them becomes positive and are study/Focus perceived as cured. Being overweight was a preventive Matoti-Mvalo group/Audio- mechanism against stigma from HIV. Being voluptuous (2011)/South Africa/Cape recorded/Note Low SES, was seen as part of the culture and greater value was Town taking 20 Urban, Adult placed on women with such body frame. While participants rightfully pointed out that vegetables Qualitative were a healthy food choice, it appeared that they did not study/Duo-in- always have a clear idea of what represents adequate depth Rural, nutrition. Animal products seems to be preferred and interview/audio- Adolescents daily consumption is viewed as a status symbol. Sedibe (2014)/South recorded/Thematic (16 - 19 Additionally, there was a preference for junk food Africa/Agincourt analysis 11 pairs years) because it was filling. The taste of vegetables was a

276

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings deterrent to their consumption. Older female students (grade 12) are discouraged from participating in sports. They agreed that physical exercise promotes good health.

Mixed methods Larger body size was viewed as part of being African. study /Focus Overweight people have no room to participate in group/audio- sporting activities and may be discriminated against. recorded/Tape Having a smaller waist was seen as more beautiful. Bodiba (2008)/South recorded Thematic Some overweight participants expressed concern about Africa analysis 9 Adolescents the inability to wear the clothes they would like to wear. Being overweight or obese was acceptable and seen as beautiful by some participants. However, there were opposing views among some participants who stated that they would not like to be overweight or obese. All participants regarded obese people as being "too fat". The general view was that overweight/obesity was not a sign of wealth and that it places one at risk of morbidity and causes physical limitations. There were negative perceptions of slender individuals as well as they were viewed as having a disease. Individuals that lose weight for health reasons were also perceived negatively. Women that diet were seen as snobbish by some community members. Highly processed food, fatty Qualitative foods and certain animal products were cheap but study/Semi- Low SES, vegetables were expensive. Those that were financially structured focus Urban, Adults insecure could not buy healthy food. Overconsumption Draper (2015)/South groups/Thematic (24 - 51 of high energy staple foods was also implicated in Africa/Cape Town analysis 21 years) weight gain. People who exercise in the open received

277

There was a cultural and normative expectation for women to be overweight. Losing weight was perceived as a sign of ill health and stress. Being overweight was seen as a sign of happiness and wealth. Being overweight was acceptable as long as one does not exceed a certain threshold. Nonetheless, some younger women challenged the acceptability of overweight due to its negative consequences. Some believed that stress, lack of exercise, low SES and poor access to fruits and vegetables led to overweight. The majority of overweight and obese women underestimated their sizes, chose silhouettes that were smaller than their true Low SES, size and expressed satisfaction with their body size. Descriptive 60% had high However, normal weight and overweight women were qualitative school or dissatisfied with their body size and believed they Study/Semi- college would be more attractive if they gained more weight. Okop (2016)/South structured focus education, Obese people were more likely to express a desire to Africa/Langa (near Cape groups/thematic Adults (35 - lose weight while the reverse was the case for Town) analysis 36 70 years) overweight individuals. Barriers to physical activity

278

Adolescent girls preferred purchasing calorie dense foods with poor nutritive value from school rather than bring lunch from home. Few students reported eating fruits and the ones sold in school were deemed Exploratory expensive. Adolescent girls jointly made decisions qualitative Grade 12, about the food choice when they are together at the study/Duo- Adolescents mall. It often involves alternating between each other's interviewing (mean age = food preferences or the decision will be made by the Voorend (2012)/South technique/audio- 58 (29 18 years person paying for the meal. The girls dieted because Africa/Soweto recorded pairs) (SD=1.2)) they wanted to look slim. Adolescent health decisions regarding physical activity, diet and obesity are influenced by time, place and company. When adolescents are away from home, they Qualitative have the opportunity to purchase sugar sweetened descriptive drinks and fast foods, which are preferred. Being around study/Focus friends can influence food choice and not conforming to group/note- Low and High prevailing choice will receive disapproval from friends. Brown taking/Thematic SES, Female participants indicated that while they preferred a (2014)/Botswana/Gaborone analysis 18 Adolescent healthy body weight, they wanted to be voluptuous so

279

Table 1: Description of Included Studies Author Study Type/Data Sample size (Year)/Country/City or Collection/Data (females Town Analysis only) Demographics Main findings that they "are able to wear nice clothes." Family members and others in Botswana viewed obese individuals as attractive, wealthy and strong. However, among adolescents, obesity was stigmatized. Equally, being thin was viewed as negative. Thinness and rapid weight loss was associated with diseases such as HIV infection. Adolescents perceived that their parents would be supportive of them engaging in obesity prevention and weight loss. Qualitative study Some of the participants correctly described obesity but /semi-structured a few had an incorrect descriptions of obesity. Physical in-depth activity is viewed as a pursuit of the young. The only interviews/audio- Rural, Adult form of physical activity that appeared acceptable were Mugo (2016)/Kenya/ recorded (20 - 45 household-related duties. Women with big and Subukia /Thematic analysis 8 years) voluptuous bodies were regarded as healthy

280

CHAPTER 7

DISCUSSION

Summary of Main Findings

This dissertation project encompassed three specific aims, which were fulfilled in three research manuscripts. Prior to the development of the manuscripts, an extensive literature reviews was completed to identify research gaps that informed the research questions, the methods and the study interpretation. For specific aim one, which was to analyze the prevalence of adult female overweight and obesity in second level administrative units, the related study revealed significant area-level variation among women, 18 years and older in the seven SSA countries examined. Overweight and obesity prevalence combined ranged from 7.5 – 42% in

Benin, 1.4 – 35.9% in Ethiopia, 1.6 – 44.7% in Mozambique, 1.0 – 67.9% in Nigeria, 2.2 -

72.4% in Tanzania, 3.9 – 39.9% in Zambia, and 4.5 - 50.6% in Zimbabwe. At a higher scale, that is first level administrative units, this level of variation was not clearly discernable as overweight prevalence varied from 9.9 – 42.0% in Benin, 3.6 – 33.3% in Ethiopia, 6.2 – 43.9% in Mozambique, 10.6 – 46.7% in Nigeria, 14.2 – 49.9% in Tanzania, 11.8 – 38.5% in Zambia, and from 30.8% - 50.6% in Zimbabwe. Study findings included tables and maps listing the prevalence of overweight in each second-level administrative unit and the level of uncertainty for each estimate. Collectively, the findings of this study provides baseline data congruent with indicator 14 of the Global NCD monitoring framework, which is critical to monitor progress towards achievement of the 2025 global NCD target for obesity.

The focus of specific aim two was to analyze the multiplicative effect modification of education on the association between household wealth and overweight in low and middle-

281

income SSA countries and a corresponding research study involving 22 low and middle income

SSA countries was undertaken to accomplish it. Consequently, although we found a positive association between household wealth and overweight, there was no multiplicative effect modification in the presence of education in any of the countries examined. Because the lack of multiplicative effect modification contrasted with findings from other middle income countries elsewhere, a sensitivity analysis was conducted replicating the same methods used in previous work and the conclusions remained the same.

The goal of specific aim three was to identify, appraise and synthesize qualitative research evidence to describe contextual factors contributing to female weight gain at different life stages. In the study linked to this specific aim, the findings of the included studies was consolidated into three themes namely, body size and shape ideals, barriers to healthy eating, and barriers to physical activity. Social and cultural factors had cross-cutting influences on the key themes. Body size ideals varied for adult women and adolescents. Although both groups desired to be voluptuous, the adults preferred to be overweight or obese. This preference was reinforced by cultural and social expectations that associated adult female overweight or obesity with positive attributes. Overweight and obese adolescents were victimized and there appeared to be a preference for normal weight but voluptuous body type among adolescent. Thinness were perceived negatively and related to HIV infection. Specific barriers to healthy food choices included migration, affordability of food, peer pressure and the food environment. Whereas, specific barriers to physical activity included time constraints, self-efficacy, lack of recreational facilities and the potential for sexual violence.

282

Strengths and Limitations

The research study on estimating the prevalence of adult female overweight within second-level administrative units and the other study that examined the multiplicative effect modification of education on the association between household wealth and overweight was strengthened by the use of nationally representative datasets from individual countries. The

Demographic and Health Survey Program, the data sources for the quantitative research projects, collects data using a valid scientific process. However, although the datasets for the prevalence study was not representative at second-level administrative units, a reasonable statistical model was applied to adjust the raw estimates and levels of uncertainty for the estimates were presented. A key drawback for the study on the effect modification of education on the association between household wealth and overweight was the cross-sectional nature of the study, which precludes statements of causality.

The final study in this dissertation project, the qualitative research synthesis is strengthened by the exhaustive search undertaken to identify research articles, which suggests that the study findings is a suitable summary of the existing literature on the topic. Nonetheless, because the study results are only based on a synthesis of studies from four SSA countries (South

Africa, Kenya, Ghana, Botswana), it is not recommended that the findings are transferred beyond those settings.

In general, since all the research studies in this dissertation project are based on secondary data, the findings may not capture contemporary developments.

283

Implications of the Study Findings and Recommendations for Future Work.

The micro-level analysis of female overweight prevalence showcased substantial level of heterogeneity x-raying the importance of data analysis at lower scales as an important line of action in disease control and prevention. Overweight is emerging as a significant public health problem in SSA and key stakeholders in the arena of NCD prevention must prioritize the identified high burden administrative units for additional research and intervention planning.

Research studies are necessary to validate the findings of this study with ground truth and resources are warranted to develop better models with increased precision for small area analysis with survey data as well as collect more micro data to improve model predictions. A key object of intervention planning is formative research and community based participatory research to identify the determinants of overweight and obesity and outline culturally-sensitive solutions.

Addressing the cultural and social aspects of female overweight and obesity will be challenging especially as it relates to the stigmatization of thinness, the cultural conception of adult female overweight as beautiful, and the victimization of overweight and obese adolescents.

However, the findings of this dissertation project suggests some preliminary solutions.

The absence of an effect modification by education implies the need to assess nutrition-related knowledge and skills garnered at primary, secondary and tertiary education levels. Designing a curriculum that teaches individuals to make healthy food choices and regulating the food environment at educational institutions are low hanging fruit for the prevention of female overweight and obesity. This is supported by the qualitative research synthesis, which showed that adolescent girls preferred consuming convenience foods at school due to affordability and peer influence. Such food preferences develop into the lifelong habits leading to increased

284

adiposity in adulthood. Regulation of the food environment can also apply to individuals outside of the school system.

Another important measure would be closing the apparent gender gap in female physical activity. This can commence from the provision of institutional support and resources for adolescent females to engage in physical activity, eliminating female ageism and any form of discrimination from physical activity and action to support the rise of female physical activity role models to shift the negative perception associated with female physical activity.

Finally, more studies on this topic are needed across SSA countries. The paucity of evidence will preclude stakeholders from developing a comprehensive and effective NCD prevention and control strategy, and hinder the exchange of cross-country ideas and resources across SSA boundaries.

285

REFERENCES

2018 Physical Activity Guidelines Advisory Committee. (2018). 2018 Physical Activity Guidelines Advisory Committee Scientific Report. Washington DC . Retrieved from https://health.gov/paguidelines/second- edition/report/pdf/PAG_Advisory_Committee_Report.pdf Abraham, T. M., Pedley, A., Massaro, J. M., Hoffmann, U., & Fox, C. S. (2015). Association Between Visceral and Subcutaneous Adipose Depots and Incident Cardiovascular Disease Risk Factors. Circulation, 132(17), 1639–1647. https://doi.org/10.1161/CIRCULATIONAHA.114.015000 Abrha, S., Shiferaw, S., & Ahmed, K. Y. (2016). Overweight and obesity and its sociodemographic correlates among urban Ethiopian women: evidence from the 2011 EDHS. BMC Public Health, 16, 636. https://doi.org/10.1186/s12889-016-3315-3 Ackermann, R. T., Finch, E. A., Brizendine, E., Zhou, H., & Marrero, D. G. (2008). Translating the Diabetes Prevention Program into the Community: The DEPLOY Pilot Study. American Journal of Preventive Medicine, 35(4), 357–363. https://doi.org/10.1016/J.AMEPRE.2008.06.035 Adam, T. C., & Epel, E. S. (2007). Stress, eating and the reward system. Physiology & Behavior, 91(4), 449–458. https://doi.org/10.1016/j.physbeh.2007.04.011 Ades, P. A., & Savage, P. D. (2010). The obesity paradox: perception vs knowledge. Mayo Clinic Proceedings, 85(2), 112–114. https://doi.org/10.4065/mcp.2009.0777 Aitsi-Selmi, A., Bell, R., Shipley, M. J., & Marmot, M. G. (2014). Education modifies the association of wealth with obesity in women in middle-income but not low-income countries: an interaction study using seven national datasets, 2005-2010. PloS One, 9(3), e90403. https://doi.org/10.1371/journal.pone.0090403 Åkerstedt, T., Kecklund, G., & Axelsson, J. (2007). Impaired sleep after bedtime stress and worries. Biological Psychology, 76(3), 170–173. https://doi.org/10.1016/J.BIOPSYCHO.2007.07.010 Alcock, J., Maley, C. C., & Aktipis, C. A. (2014). Is eating behavior manipulated by the gastrointestinal microbiota? Evolutionary pressures and potential mechanisms. BioEssays, 36(10), 940–949. https://doi.org/10.1002/bies.201400071 Allen, L., Williams, J., Townsend, N., Mikkelsen, B., Roberts, N., Foster, C., & Wickramasinghe, K. (2017). Socioeconomic status and non-communicable disease behavioural risk factors in low-income and lower-middle-income countries: a systematic review. The Lancet. Global Health, 5(3), e277–e289. https://doi.org/10.1016/S2214- 109X(17)30058-X Alpert, M. A., Lavie, C. J., Agrawal, H., Kumar, A., & Kumar, S. A. (2016). Cardiac effects of obesity: Pathophysiologic, clinical, and prognostic consequences-A review. Journal of Cardiopulmonary Rehabilitation and Prevention. https://doi.org/10.1097/HCR.0000000000000147

286

Amugsi, D. A., Dimbuene, Z. T., Mberu, B., Muthuri, S., & Ezeh, A. C. (2017). Prevalence and time trends in overweight and obesity among urban women: an analysis of demographic and health surveys data from 24 African countries, 1991-2014. BMJ Open, 7(10), e017344. https://doi.org/10.1136/bmjopen-2017-017344 Ashwell, M., Gunn, P., & Gibson, S. (2012). Waist-to-height ratio is a better screening tool than waist circumference and BMI for adult cardiometabolic risk factors: systematic review and meta-analysis. Obesity Reviews, 13(3), 275–286. https://doi.org/10.1111/j.1467- 789X.2011.00952.x Aune, D., Sen, A., Prasad, M., Norat, T., Janszky, I., Tonstad, S., … Vatten, L. J. (2016). BMI and all cause mortality: systematic review and non-linear dose-response meta-analysis of 230 cohort studies with 3.74 million deaths among 30.3 million participants. BMJ (Clinical Research Ed.), 353, i2156. https://doi.org/10.1136/BMJ.I2156 Baillie-Hamilton, P. F. (2002). Chemical Toxins: A Hypothesis to Explain the Global Obesity Epidemic. The Journal of Alternative and Complementary Medicine, 8(2), 185–192. https://doi.org/10.1089/107555302317371479 Barker, D. J. P. (2004). Developmental origins of adult health and disease. Journal of Epidemiology and Community Health, 58(2), 114–115. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14729887 Barker, D. J. P. (2007). The origins of the developmental origins theory. Journal of Internal Medicine, 261(5), 412–417. https://doi.org/10.1111/j.1365-2796.2007.01809.x Batnitzky, A. (2008). Obesity and household roles: gender and social class in Morocco. Sociology of Health & Illness, 30(3), 445–462. https://doi.org/10.1111/j.1467- 9566.2007.01067.x Bayol, S. A., Farrington, S. J., & Stickland, N. C. (2007). A maternal “junk food” diet in pregnancy and lactation promotes an exacerbated taste for “junk food” and a greater propensity for obesity in rat offspring. British Journal of Nutrition, 98(4), 843–851. https://doi.org/10.1017/S0007114507812037 Belbasis, L., Savvidou, M. D., Kanu, C., Evangelou, E., & Tzoulaki, I. (2016). Birth weight in relation to health and disease in later life: an umbrella review of systematic reviews and meta-analyses. BMC Medicine, 14(1), 147. https://doi.org/10.1186/s12916-016-0692-5 Bennett, D. A. (2001). How can I deal with missing data in my study? Australian and New Zealand Journal of Public Health, 25(5), 464–469. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11688629 Berrington de Gonzalez, A., Hartge, P., Cerhan, J. R., Flint, A. J., Hannan, L., MacInnis, R. J., … Thun, M. J. (2010). Body-Mass Index and Mortality among 1.46 Million White Adults. New England Journal of Medicine, 363(23), 2211–2219. https://doi.org/10.1056/NEJMoa1000367 Black, R. E., Victora, C. G., Walker, S. P., Bhutta, Z. A., Christian, P., de Onis, M., … Maternal and Child Nutrition Study Group. (2013). Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet (London, England), 382(9890), 427– 451. https://doi.org/10.1016/S0140-6736(13)60937-X

287

Blair, C. (2010). Stress and the Development of Self-Regulation in Context. Child Development Perspectives, 4(3), 181–188. https://doi.org/10.1111/j.1750-8606.2010.00145.x Booth, A. (2016). Searching for qualitative research for inclusion in systematic reviews: a structured methodological review. Systematic Reviews, 5(1), 74. https://doi.org/10.1186/s13643-016-0249-x Brown, P. J., & Konner, M. (1987). An anthropological perspective on obesity. Annals of the New York Academy of Sciences, 499, 29–46. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/3300488 Burgert, C. R., Zachary, B., & Colston, J. (2013). Incorporating Geographic Information into Demographic and Health Surveys: A Field Guide to GPS Data Collection. Calverton, Maryland, USA. Camhi, S. M., Bray, G. A., Bouchard, C., Greenway, F. L., Johnson, W. D., Newton, R. L., … Katzmarzyk, P. T. (2011). The relationship of waist circumference and BMI to visceral, subcutaneous, and total body fat: sex and race differences. Obesity (Silver Spring, Md.), 19(2), 402–408. https://doi.org/10.1038/oby.2010.248 Campbell, S. D. I., Brosnan, B. J., Chu, A. K. Y., Skeaff, C. M., Rehrer, N. J., Perry, T. L., & Peddie, M. C. (2018). Sedentary Behavior and Body Weight and Composition in Adults: A Systematic Review and Meta-analysis of Prospective Studies. Sports Medicine, 48(3), 585– 595. https://doi.org/10.1007/s40279-017-0828-6 Case, A., & Menendez, A. (2009). Sex differences in obesity rates in poor countries: Evidence from South Africa. Economics & Human Biology, 7(3), 271–282. https://doi.org/10.1016/j.ehb.2009.07.002 Castro, A. M., Macedo-de la Concha, L. E., & Pantoja-Meléndez, C. A. (2017). Low-grade inflammation and its relation to obesity and chronic degenerative diseases. Revista Médica Del Hospital General de México, 80(2), 101–105. https://doi.org/10.1016/J.HGMX.2016.06.011 Chin, S.-H., Kahathuduwa, C. N., & Binks, M. (2016). Physical activity and obesity: what we know and what we need to know*. Obesity Reviews, 17(12), 1226–1244. https://doi.org/10.1111/obr.12460 Choquet, H., & Meyre, D. (2011). Genetics of Obesity: What have we Learned? Current Genomics, 12(3), 169–179. https://doi.org/10.2174/138920211795677895 Coffey, D. (2015). Prepregnancy body mass and weight gain during pregnancy in India and sub- Saharan Africa. Proceedings of the National Academy of Sciences of the United States of America, 112(11), 3302–3307. https://doi.org/10.1073/pnas.1416964112 Cohen, S. S., Park, Y., Signorello, L. B., Patel, A. V., Boggs, D. A., Kolonel, L. N., … Palmer, J. R. (2014). A Pooled Analysis of Body Mass Index and Mortality among African Americans. PLoS ONE, 9(11), e111980. https://doi.org/10.1371/journal.pone.0111980 Cupul-Uicab, L. A., Klebanoff, M. A., Brock, J. W., & Longnecker, M. P. (2013). Prenatal Exposure to Persistent Organochlorines and Childhood Obesity in the U.S. Collaborative Perinatal Project. Environmental Health Perspectives, 121(9), 1103–1109.

288

https://doi.org/10.1289/ehp.1205901 Curtis, J. P., Selter, J. G., Wang, Y., Rathore, S. S., Jovin, I. S., Jadbabaie, F., … Krumholz, H. M. (2005). The Obesity Paradox. Archives of Internal Medicine, 165(1), 55. https://doi.org/10.1001/archinte.165.1.55 Dake, F. A., Tawiah, E. O., & Badasu, D. M. (2011). Sociodemographic correlates of obesity among Ghanaian women. Public Health Nutrition, 14(07), 1285–1291. https://doi.org/10.1017/S1368980010002879 Dalby, M. J., Ross, A. W., Walker, A. W., & Morgan, P. J. (2017). Dietary Uncoupling of Gut Microbiota and Energy Harvesting from Obesity and Glucose Tolerance in Mice. Cell Reports, 21(6), 1521–1533. https://doi.org/10.1016/j.celrep.2017.10.056 Darbre, P. D. (2017). Endocrine Disruptors and Obesity. Current Obesity Reports, 6(1), 18–27. https://doi.org/10.1007/s13679-017-0240-4 Dempfle, A., Hinney, A., Heinzel-Gutenbrunner, M., Raab, M., Geller, F., Gudermann, T., … Hebebrand, J. (2004). Large quantitative effect of melanocortin-4 receptor gene mutations on body mass index. Journal of Medical Genetics, 41(10), 795–800. https://doi.org/10.1136/jmg.2004.018614 Desai, M., Gayle, D., Han, G., & Ross, M. G. (2007). Programmed Hyperphagia Due to Reduced Anorexigenic Mechanisms in Intrauterine Growth-Restricted Offspring. Reproductive Sciences, 14(4), 329–337. https://doi.org/10.1177/1933719107303983 Deurenberg, P. (2001). Universal cut-off BMI points for obesity are not appropriate. The British Journal of Nutrition, 85(2), 135–136. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11280336 Deurenberg, P., Andreoli, A., Borg, P., Kukkonen-Harjula, K., de Lorenzo, A., van Marken Lichtenbelt, W., … Vollaard, N. (2001). The validity of predicted body fat percentage from body mass index and from impedance in samples of five European populations. European Journal of Clinical Nutrition, 55(11), 973–979. https://doi.org/10.1038/sj.ejcn.1601254 Deurenberg, P., Deurenberg-Yap, M., & Guricci, S. (2002). Asians are different from Caucasians and from each other in their body mass index/body fat per cent relationship. Obesity Reviews : An Official Journal of the International Association for the Study of Obesity, 3(3), 141–146. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12164465 Deurenberg, P., Deurenberg Yap, M., Wang, J., Lin, F. P., & Schmidt, G. (1999). The impact of body build on the relationship between body mass index and percent body fat. International Journal of Obesity, 23(5), 537–542. https://doi.org/10.1038/sj.ijo.0800868 Deurenberg, P., Weststrate, J. A., & Seidell, J. C. (1991). Body mass index as a measure of body fatness: age- and sex-specific prediction formulas. The British Journal of Nutrition, 65(2), 105–114. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2043597 Deurenberg, P., Yap, M., & van Staveren, W. (1998). Body mass index and percent body fat: a meta analysis among different ethnic groups. International Journal of Obesity, 22(12), 1164–1171. https://doi.org/10.1038/sj.ijo.0800741 Devaux, M., & Sassi, F. (2013). Social inequalities in obesity and overweight in 11 OECD

289

countries. The European Journal of Public Health, 23(3), 464–469. https://doi.org/10.1093/eurpub/ckr058 Dinsa, G. D., Goryakin, Y., Fumagalli, E., & Suhrcke, M. (2012). Obesity and socioeconomic status in developing countries: A systematic review. Obesity Reviews, 13(11), 1067–1079. https://doi.org/10.1111/j.1467-789X.2012.01017.x Dobbs Richard, Sawers Corrine, Thompson Fraser, Manyika James, Woetzel Jonathan, Child Peter, McKenna Sorcha, S. A. (2014). Overcoming obesity: An initial economic analysis. McKinsey Global Institute. Doehner, W., Erdmann, E., Cairns, R., Clark, A. L., Dormandy, J. A., Ferrannini, E., & Anker, S. D. (2012). Inverse relation of body weight and weight change with mortality and morbidity in patients with type 2 diabetes and cardiovascular co-morbidity: An analysis of the PROactive study population. International Journal of Cardiology, 162(1), 20–26. https://doi.org/10.1016/J.IJCARD.2011.09.039 Dong, Y., & Peng, C.-Y. J. (2013). Principled missing data methods for researchers. SpringerPlus, 2(1), 222. https://doi.org/10.1186/2193-1801-2-222 Dowell, S. F., Blazes, D., & Desmond-Hellmann, S. (2016). Four steps to precision public health. Nature, 540(7632), 189–191. https://doi.org/10.1038/540189a Drewnowski, A., & Specter, S. (2004). Poverty and obesity: the role of energy density and energy costs. The American Journal of Clinical Nutrition, 79(1), 6–16. https://doi.org/10.1093/ajcn/79.1.6 Ebbeling, C. B., Leidig, M. M., Feldman, H. A., Lovesky, M. M., & Ludwig, D. S. (2007). Effects of a Low–Glycemic Load vs Low-Fat Diet in Obese Young Adults. JAMA, 297(19), 2092. https://doi.org/10.1001/jama.297.19.2092 Egger, G., Swinburn, B., & Amirul Islam, F. M. (2012). Economic growth and obesity: An interesting relationship with world-wide implications. Economics & Human Biology, 10(2), 147–153. https://doi.org/10.1016/J.EHB.2012.01.002 Eknoyan, G. (2006). A History of Obesity, or How What Was Good Became Ugly and Then Bad. Advances in Chronic Kidney Disease, 13(4), 421–427. https://doi.org/10.1053/j.ackd.2006.07.002 Elhakeem, A., Cooper, R., Bann, D., & Hardy, R. (2015). Childhood socioeconomic position and adult leisure-time physical activity: a systematic review. The International Journal of Behavioral Nutrition and Physical Activity, 12, 92. https://doi.org/10.1186/s12966-015- 0250-0 Elobeid, M. A., & Allison, D. B. (2008). Putative environmental-endocrine disruptors and obesity: a review. Current Opinion in Endocrinology, Diabetes and Obesity, 15(5), 403– 408. https://doi.org/10.1097/MED.0b013e32830ce95c Epel, E., Lapidus, R., McEwen, B., & Brownell, K. (2001). Stress may add bite to appetite in women: a laboratory study of stress-induced cortisol and eating behavior. Psychoneuroendocrinology, 26(1), 37–49. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11070333

290

Evans, G. W., Fuller-Rowell, T. E., & Doan, S. N. (2012). Childhood cumulative risk and obesity: the mediating role of self-regulatory ability. Pediatrics, 129(1), e68-73. https://doi.org/10.1542/peds.2010-3647 Flegal, K. M., Graubard, B. I., Williamson, D. F., & Gail, M. H. (2005). Excess Deaths Associated With Underweight, Overweight, and Obesity. JAMA, 293(15), 1861. https://doi.org/10.1001/jama.293.15.1861 Fletcher, I. (2014). Defining an epidemic: the body mass index in British and US obesity research 1960-2000. Sociology of Health & Illness, 36(3), 338–353. https://doi.org/10.1111/1467-9566.12050 Ford, N. D., Patel, S. A., & Narayan, K. M. V. (2017). Obesity in Low- and Middle-Income Countries: Burden, Drivers, and Emerging Challenges. Annual Review of Public Health, 38(1), 145–164. https://doi.org/10.1146/annurev-publhealth-031816-044604 Foster, G. D., Wyatt, H. R., Hill, J. O., McGuckin, B. G., Brill, C., Mohammed, B. S., … Klein, S. (2003). A Randomized Trial of a Low-Carbohydrate Diet for Obesity. New England Journal of Medicine, 348(21), 2082–2090. https://doi.org/10.1056/NEJMoa022207 Gaillard, R., Felix, J. F., Duijts, L., & Jaddoe, V. W. V. (2014). Childhood consequences of maternal obesity and excessive weight gain during pregnancy. Acta Obstetricia et Gynecologica Scandinavica, 93(11), 1085–1089. https://doi.org/10.1111/aogs.12506 Gakidou, E., Afshin, A., Alemu Abajobir, A., Hassen Abate, K., Abbafati, C., Abbas, K. M., … Factors Collaborators, R. (2017). Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990 - 2016: a systematic analysis for the Global Burden of Disease Study 2016. https://doi.org/10.1016/S0140-6736(17)32366-8 Gallagher, D., Heymsfield, S. B., Heo, M., Jebb, S. A., Murgatroyd, P. R., & Sakamoto, Y. (2000). Healthy percentage body fat ranges: an approach for developing guidelines based on body mass index. The American Journal of Clinical Nutrition, 72(3), 694–701. https://doi.org/10.1093/ajcn/72.3.694 Gallagher, D., Visser, M., Sepúlveda, D., Pierson, R. N., Harris, T., & Heymsfield, S. B. (1996). How useful is body mass index for comparison of body fatness across age, sex, and ethnic groups? American Journal of Epidemiology, 143(3), 228–239. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8561156 Garrow, J. S., & Webster, J. (1985). Quetelet’s index (W/H2) as a measure of fatness. International Journal of Obesity, 9(2), 147–153. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/4030199 Gaskins, R. B., Lagasse, L. L., Liu, J., Shankaran, S., Lester, B. M., Bada, H. S., … Roberts, M. (2010). Small for gestational age and higher birth weight predict childhood obesity in preterm infants. American Journal of Perinatology, 27(9), 721–730. https://doi.org/10.1055/s-0030-1253555 Gavin, A. R., Simon, G. E., & Ludman, E. J. (2010). The association between obesity, depression, and educational attainment in women: The mediating role of body image dissatisfaction. Journal of Psychosomatic Research, 69(6), 573–581.

291

https://doi.org/10.1016/j.jpsychores.2010.05.001 Gewa, C. A., Leslie, T. F., & Pawloski, L. R. (2013). Geographic distribution and socioeconomic determinants of women’s nutritional status in Mali households. Public Health Nutrition, 16(9), 1575–1585. https://doi.org/10.1017/S136898001200451X Ghoussaini, M., Stutzmann, F., Couturier, C., Vatin, V., Durand, E., Lecoeur, C., … Meyre, D. (2010). Analysis of the SIM1 Contribution to Polygenic Obesity in the French Population. Obesity, 18(8), 1670–1675. https://doi.org/10.1038/oby.2009.468 Global BMI Mortality Collaboration, E. Di, Di Angelantonio, E., Bhupathiraju, S., Wormser, D., Gao, P., Kaptoge, S., … Hu, F. (2016). Body-mass index and all-cause mortality: individual-participant-data meta-analysis of 239 prospective studies in four continents. Lancet (London, England), 388(10046), 776–786. https://doi.org/10.1016/S0140- 6736(16)30175-1 Godfrey, K. M., Reynolds, R. M., Prescott, S. L., Nyirenda, M., Jaddoe, V. W. V, Eriksson, J. G., & Broekman, B. F. P. (2017). Influence of maternal obesity on the long-term health of offspring. The Lancet. Diabetes & Endocrinology, 5(1), 53–64. https://doi.org/10.1016/S2213-8587(16)30107-3 Golay, A., Allaz, A. F., Morel, Y., de Tonnac, N., Tankova, S., & Reaven, G. (1996). Similar weight loss with low- or high-carbohydrate diets. The American Journal of Clinical Nutrition, 63(2), 174–178. https://doi.org/10.1093/ajcn/63.2.174 Gore, A. C., Chappell, V. A., Fenton, S. E., Flaws, J. A., Nadal, A., Prins, G. S., … Zoeller, R. T. (2015). EDC-2: The Endocrine Society’s Second Scientific Statement on Endocrine- Disrupting Chemicals. Endocrine Reviews, 36(6), E1–E150. https://doi.org/10.1210/er.2015-1010 Gortmaker, S. L., Swinburn, B. A., Levy, D., Carter, R., Mabry, P. L., Finegood, D. T., … Moodie, M. L. (2011). Changing the future of obesity: science, policy, and action. Lancet (London, England), 378(9793), 838–847. https://doi.org/10.1016/S0140-6736(11)60815-5 Graham, J. W. (2009). Missing Data Analysis: Making It Work in the Real World. Annual Review of Psychology, 60(1), 549–576. https://doi.org/10.1146/annurev.psych.58.110405.085530 Green, E. (2013). Explaining African ethnic diversity. International Political Science Review, 34(3), 235–253. https://doi.org/10.1177/0192512112455075 Greenway, F. L. (2015). Physiological adaptations to weight loss and factors favouring weight regain. International Journal of Obesity (2005), 39(8), 1188–1196. https://doi.org/10.1038/ijo.2015.59 Gruberg, L., Weissman, N. J., Waksman, R., Fuchs, S., Deible, R., Pinnow, E. E., … Lindsay, J. (2002). The impact of obesity on the short-term and long-term outcomes after percutaneous coronary intervention: the obesity paradox? Journal of the American College of Cardiology, 39(4), 578–584. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11849854 Gunderson, E. P. (2009). Childbearing and obesity in women: weight before, during, and after pregnancy. Obstetrics and Gynecology Clinics of North America, 36(2), 317–32, ix.

292

https://doi.org/10.1016/j.ogc.2009.04.001 Guyenet, S. J. (2018). Why the carbohydrate-insulin model of obesity is probably wrong: A supplementary reply to Ebbeling and Ludwig’s JAMA article. Retrieved October 8, 2018, from http://www.stephanguyenet.com/why-the-carbohydrate-insulin-model-of-obesity-is- probably-wrong-a-supplementary-reply-to-ebbeling-and-ludwigs-jama-article/ Guyenet, S. J., & Schwartz, M. W. (2012). Regulation of Food Intake, Energy Balance, and Body Fat Mass: Implications for the Pathogenesis and Treatment of Obesity. The Journal of Clinical Endocrinology & Metabolism, 97(3), 745–755. https://doi.org/10.1210/jc.2011- 2525 Hall, J. E., do Carmo, J. M., da Silva, A. A., Wang, Z., & Hall, M. E. (2015). Obesity-induced hypertension: interaction of neurohumoral and renal mechanisms. Circulation Research, 116(6), 991–1006. https://doi.org/10.1161/CIRCRESAHA.116.305697 Hall, K. D., Guyenet, S. J., & Leibel, R. L. (2018). The Carbohydrate-Insulin Model of Obesity Is Difficult to Reconcile With Current Evidence. JAMA Internal Medicine, 178(8), 1103. https://doi.org/10.1001/jamainternmed.2018.2920 Han, T. S., van Leer, E. M., Seidell, J. C., & Lean, M. E. (1995). Waist circumference action levels in the identification of cardiovascular risk factors: prevalence study in a random sample. BMJ (Clinical Research Ed.), 311(7017), 1401–1405. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8520275 Harris, H. R., Willett, W. C., & Michels, K. B. (2013). Parental smoking during pregnancy and risk of overweight and obesity in the daughter. International Journal of Obesity (2005), 37(10), 1356–1363. https://doi.org/10.1038/ijo.2013.101 Haslam, D. (2007). Obesity: a medical history. Obesity Reviews, 8(s1), 31–36. https://doi.org/10.1111/j.1467-789X.2007.00314.x Hawkes, C., Smith, T. G., Jewell, J., Wardle, J., Hammond, R. A., Friel, S., … Kain, J. (2015). Smart food policies for obesity prevention. Lancet (London, England), 385(9985), 2410– 2421. https://doi.org/10.1016/S0140-6736(14)61745-1 Heindel, J. J. (2003). Endocrine Disruptors and the Obesity Epidemic. Toxicological Sciences, 76(2), 247–249. https://doi.org/10.1093/toxsci/kfg255 Heymsfield, S. B., Peterson, C. M., Thomas, D. M., Heo, M., & Schuna, J. M. (2016). Why are there race/ethnic differences in adult body mass index-adiposity relationships? A quantitative critical review. Obesity Reviews, 17(3), 262–275. https://doi.org/10.1111/obr.12358 Heymsfield, S. B., & Wadden, T. A. (2017). Mechanisms, Pathophysiology, and Management of Obesity. New England Journal of Medicine, 376(3), 254–266. https://doi.org/10.1056/NEJMra1514009 Ho-Pham, L. T., Campbell, L. V, & Nguyen, T. V. (2011). More on body fat cutoff points. Mayo Clinic Proceedings, 86(6), 584; author reply 584-5. https://doi.org/10.4065/mcp.2011.0097 Horton, T. J., Drougas, H., Brachey, A., Reed, G. W., Peters, J. C., & Hill, J. O. (1995). Fat and carbohydrate overfeeding in humans: different effects on energy storage. The American

293

Journal of Clinical Nutrition, 62(1), 19–29. https://doi.org/10.1093/ajcn/62.1.19 Hu, F. B., Li, T. Y., Colditz, G. A., Willett, W. C., & Manson, J. E. (2003). Television Watching and Other Sedentary Behaviors in Relation to Risk of Obesity and Type 2 Diabetes Mellitus in Women. JAMA, 289(14), 1785. https://doi.org/10.1001/jama.289.14.1785 Hu, S., Wang, L., Yang, D., Li, L., Togo, J., Wu, Y., … Speakman, J. R. (2018). Dietary Fat, but Not Protein or Carbohydrate, Regulates Energy Intake and Causes Adiposity in Mice. Cell Metabolism, 0(0). https://doi.org/10.1016/j.cmet.2018.06.010 Hu, T., Mills, K. T., Yao, L., Demanelis, K., Eloustaz, M., Yancy, W. S., … Bazzano, L. A. (2012). Effects of Low-Carbohydrate Diets Versus Low-Fat Diets on Metabolic Risk Factors: A Meta-Analysis of Randomized Controlled Clinical Trials. American Journal of Epidemiology, 176(suppl_7), S44–S54. https://doi.org/10.1093/aje/kws264 ICF International. (2012). Demographic and Health Survey: Sampling and Household Listing Manual. Calverton, Maryland. Iliodromiti, S., Celis-Morales, C. A., Lyall, D. M., Anderson, J., Gray, S. R., Mackay, D. F., … Sattar, N. (2018). The impact of confounding on the associations of different adiposity measures with the incidence of cardiovascular disease: a cohort study of 296 535 adults of white European descent. European Heart Journal, 39(17), 1514–1520. https://doi.org/10.1093/eurheartj/ehy057 Iszatt, N., Stigum, H., Verner, M.-A., White, R. A., Govarts, E., Palkovicova Murinova, L., … OBELIX. (2015). Prenatal and Postnatal Exposure to Persistent Organic Pollutants and Infant Growth: A Pooled Analysis of Seven European Birth Cohorts. Environmental Health Perspectives, 123(7), 730–736. https://doi.org/10.1289/ehp.1308005 Jackson, A., Stanforth, P., Gagnon, J., Rankinen, T., Leon, A., Rao, D., … Wilmore, J. (2002). The effect of sex, age and race on estimating percentage body fat from body mass index: The Heritage Family Study. International Journal of Obesity, 26(6), 789–796. https://doi.org/10.1038/sj.ijo.0802006 Jackson, L. W., Lynch, C. D., Kostyniak, P. J., McGuinness, B. M., & Louis, G. M. B. (2010). Prenatal and postnatal exposure to polychlorinated biphenyls and child size at 24 months of age. Reproductive Toxicology, 29(1), 25–31. https://doi.org/10.1016/j.reprotox.2009.09.008 Jansson, N., Pettersson, J., Haafiz, A., Ericsson, A., Palmberg, I., Tranberg, M., … Jansson, T. (2006). Down-regulation of placental transport of amino acids precedes the development of intrauterine growth restriction in rats fed a low protein diet. The Journal of Physiology, 576(3), 935–946. https://doi.org/10.1113/jphysiol.2006.116509 Jansson, T., Ekstrand, Y., Björn, C., Wennergren, M., & Powell, T. L. (2002). Alterations in the activity of placental amino acid transporters in pregnancies complicated by diabetes. Diabetes, 51(7), 2214–2219. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12086952 Jensen, M. D., Ryan, D. H., Apovian, C. M., Ard, J. D., Comuzzie, A. G., Donato, K. A., … Yanovski, S. Z. (2014). 2013 AHA/ACC/TOS Guideline for the Management of Overweight and Obesity in Adults. Circulation, 129(25 suppl 2), S102–S138. https://doi.org/10.1161/01.cir.0000437739.71477.ee

294

Jones, H. N., Woollett, L. A., Barbour, N., Prasad, P. D., Powell, T. L., & Jansson, T. (2009). High-fat diet before and during pregnancy causes marked up-regulation of placental nutrient transport and fetal overgrowth in C57/BL6 mice. The FASEB Journal, 23(1), 271–278. https://doi.org/10.1096/fj.08-116889 Karmaus, W., Osuch, J. R., Eneli, I., Mudd, L. M., Zhang, J., Mikucki, D., … Davis, S. (2009). Maternal levels of dichlorodiphenyl-dichloroethylene (DDE) may increase weight and body mass index in adult female offspring. Occupational and Environmental Medicine, 66(3), 143–149. https://doi.org/10.1136/oem.2008.041921 Keys, A. (1946). Human starvation and its consequences. Journal of American Dietetic Association, 22, 582–587. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/20991087 Keys, A., Fidanza, F., Karvonen, M. J., Kimura, N., & Taylor, H. L. (1972). Indices of relative weight and obesity. Journal of Chronic Diseases, 25(6), 329–343. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/4650929 Keyserling, T. C., Samuel-Hodge, C. D., Pitts, S. J., Garcia, B. A., Johnston, L. F., Gizlice, Z., … Ammerman, A. S. (2016). A community-based lifestyle and weight loss intervention promoting a Mediterranean-style diet pattern evaluated in the stroke belt of North Carolina: the Heart Healthy Lenoir Project. BMC Public Health, 16, 732. https://doi.org/10.1186/s12889-016-3370-9 Khan, S. S., Ning, H., Wilkins, J. T., Allen, N., Carnethon, M., Berry, J. D., … Lloyd-Jones, D. M. (2018). Association of Body Mass Index With Lifetime Risk of Cardiovascular Disease and Compression of Morbidity. JAMA Cardiology, 3(4), 280. https://doi.org/10.1001/jamacardio.2018.0022 Khosla, T., & Lowe, C. R. (1967). Indices of obesity derived from body weight and height. British Journal of Preventive & Social Medicine, 21(3), 122–128. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6033482 Kim, J. T., & Lee, H. K. (2017). Childhood obesity and endocrine disrupting chemicals. Annals of Pediatric Endocrinology & Metabolism, 22(4), 219–225. https://doi.org/10.6065/apem.2017.22.4.219 Kim, S. H., Després, J.-P., & Koh, K. K. (2016). Obesity and cardiovascular disease: friend or foe? European Heart Journal, 37(48), 3560–3568. https://doi.org/10.1093/eurheartj/ehv509 Kinzig, K. P., Hargrave, S. L., Hyun, J., & Moran, T. H. (2007). Energy balance and hypothalamic effects of a high-protein/low-carbohydrate diet. Physiology & Behavior, 92(3), 454–460. https://doi.org/10.1016/j.physbeh.2007.04.019 Kinzig, K. P., Scott, K. A., Hyun, J., Bi, S., & Moran, T. H. (2005). Altered Hypothalamic Signaling and Responses to Food Deprivation in Rats Fed a Low-Carbohydrate Diet. Obesity Research, 13(10), 1672–1682. https://doi.org/10.1038/oby.2005.205 Kirunda, B. E., Fadnes, L. T., Wamani, H., Van den Broeck, J., & Tylleskar, T. (2015). Population-based survey of overweight and obesity and the associated factors in peri-urban and rural Eastern Uganda. BMC Public Health, 15, 1168. https://doi.org/10.1186/s12889- 015-2506-7

295

Klok, M. D., Jakobsdottir, S., & Drent, M. L. (2007). The role of leptin and ghrelin in the regulation of food intake and body weight in humans: a review. Obesity Reviews, 8(1), 21– 34. https://doi.org/10.1111/j.1467-789X.2006.00270.x Komaroff, M. (2016). For Researchers on Obesity: Historical Review of Extra Body Weight Definitions. Journal of Obesity, 2016, 1–9. https://doi.org/10.1155/2016/2460285 Kontis, V., Mathers, C. D., Bonita, R., Stevens, G. A., Rehm, J., Shield, K. D., … Ezzati, M. (2015). Regional contributions of six preventable risk factors to achieving the 25 × 25 non- communicable disease mortality reduction target: a modelling study. The Lancet Global Health, 3(12), e746–e757. https://doi.org/10.1016/S2214-109X(15)00179-5 Lammert, O., Grunnet, N., Faber, P., Bjørnsbo, K. S., Dich, J., Larsen, L. O., … Quistorff, B. (2000). Effects of isoenergetic overfeeding of either carbohydrate or fat in young men. The British Journal of Nutrition, 84(2), 233–245. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11029975 Lapidus, L., Bengtsson, C., Larsson, B., Pennert, K., Rybo, E., & Sjöström, L. (1984). Distribution of adipose tissue and risk of cardiovascular disease and death: a 12 year follow up of participants in the population study of women in Gothenburg, Sweden. British Medical Journal (Clinical Research Ed.), 289(6454), 1257–1261. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6437507 Larsson, B., Svärdsudd, K., Welin, L., Wilhelmsen, L., Björntorp, P., & Tibblin, G. (1984). Abdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: 13 year follow up of participants in the study of men born in 1913. British Medical Journal (Clinical Research Ed.), 288(6428), 1401–1404. https://doi.org/10.1136/BMJ.288.6428.1401 Leibel, R. L., Rosenbaum, M., & Hirsch, J. (1995). Changes in Energy Expenditure Resulting from Altered Body Weight. New England Journal of Medicine, 332(10), 621–628. https://doi.org/10.1056/NEJM199503093321001 Lerer-Metzger, M., Rizkalla, S. W., Luo, J., Champ, M., Kabir, M., Bruzzo, F., … Slama, G. (1996). Effects of long-term low-glycaemic index starchy food on plasma glucose and lipid concentrations and adipose tissue cellularity in normal and diabetic rats. The British Journal of Nutrition, 75(5), 723–732. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8695599 Liu, J., Fox, C. S., Hickson, D. A., May, W. D., Hairston, K. G., Carr, J. J., & Taylor, H. A. (2010). Impact of Abdominal Visceral and Subcutaneous Adipose Tissue on Cardiometabolic Risk Factors: The Jackson Heart Study. The Journal of Clinical Endocrinology & Metabolism, 95(12), 5419–5426. https://doi.org/10.1210/jc.2010-1378 Lovallo, W. R., & Buchanan, T. W. (2016). Stress hormones in psychophysiological research: Emotional, behavioral, and cognitive implications. In Handbook of Psychophysiology, Fourth Edition (pp. 465–494). Cambridge University Press. https://doi.org/10.1017/9781107415782.021 Lovejoy, J. C., Sainsbury, A., & Stock Conference 2008 Working Group. (2009). Sex differences in obesity and the regulation of energy homeostasis. Obesity Reviews, 10(2), 154–167.

296

https://doi.org/10.1111/j.1467-789X.2008.00529.x Ludwig, D. S., & Ebbeling, C. B. (2018). The Carbohydrate-Insulin Model of Obesity. JAMA Internal Medicine, 178(8), 1098. https://doi.org/10.1001/jamainternmed.2018.2933 Luke, A., Durazo-Arvizu, R., Rotimi, C., Prewitt, T. E., Forrester, T., Wilks, R., … Cooper, R. S. (1997). Relation between Body Mass Index and Body Fat in Black Population Samples from Nigeria, Jamaica, and the United States. American Journal of Epidemiology, 145(7), 620–628. https://doi.org/10.1093/oxfordjournals.aje.a009159 Lumeng, C. N., & Saltiel, A. R. (2011). Inflammatory links between obesity and metabolic disease. The Journal of Clinical Investigation, 121(6), 2111–2117. https://doi.org/10.1172/JCI57132 Luppino, F. S., de Wit, L. M., Bouvy, P. F., Stijnen, T., Cuijpers, P., Penninx, B. W. J. H., & Zitman, F. G. (2010). Overweight, Obesity, and Depression. Archives of General Psychiatry, 67(3), 220. https://doi.org/10.1001/archgenpsychiatry.2010.2 Malandro, M. S., Beveridge, M. J., Kilberg, M. S., & Novak, D. A. (1996). Effect of low-protein diet-induced intrauterine growth retardation on rat placental amino acid transport. American Journal of Physiology-Cell Physiology, 271(1), C295–C303. https://doi.org/10.1152/ajpcell.1996.271.1.C295 Manikkam, M., Tracey, R., Guerrero-Bosagna, C., & Skinner, M. K. (2013). Plastics Derived Endocrine Disruptors (BPA, DEHP and DBP) Induce Epigenetic Transgenerational Inheritance of Obesity, Reproductive Disease and Sperm Epimutations. PLoS ONE, 8(1), e55387. https://doi.org/10.1371/journal.pone.0055387 Mannan, M., Mamun, A., Doi, S., & Clavarino, A. (2016). Is there a bi-directional relationship between depression and obesity among adult men and women? Systematic review and bias- adjusted meta analysis. Asian Journal of Psychiatry, 21, 51–66. https://doi.org/10.1016/j.ajp.2015.12.008 Markowitz, S., Friedman, M. A., & Arent, S. M. (2008). Understanding the Relation Between Obesity and Depression: Causal Mechanisms and Implications for Treatment. Clinical Psychology: Science and Practice, 15(1), 1–20. https://doi.org/10.1111/j.1468- 2850.2008.00106.x Marquez, P. V, & Farrington, J. L. (2013). The Challenge of Non-Communicable Diseases and Road Traffic Injuries in Sub-Saharan Africa: An Overview. Washington, DC: The World Bank. https://doi.org/10.2337/db12-0849/-/DC1 Masood, M., & Reidpath, D. D. (2017). Effect of national wealth on BMI: An analysis of 206,266 individuals in 70 low-, middle- and high-income countries. PLOS ONE, 12(6), e0178928. https://doi.org/10.1371/journal.pone.0178928 Mbanya, V. N., Kengne, A. P., Mbanya, J. C., & Akhtar, H. (2015). Body mass index, waist circumference, hip circumference, waist–hip-ratio and waist–height-ratio: Which is the better discriminator of prevalent screen-detected diabetes in a Cameroonian population? Diabetes Research and Clinical Practice, 108(1), 23–30. https://doi.org/10.1016/j.diabres.2015.01.032

297

McLaren, L. (2007). Socioeconomic Status and Obesity. Epidemiologic Reviews, 29(1), 29–48. https://doi.org/10.1093/epirev/mxm001 Morgan, P. J., Collins, C. E., Plotnikoff, R. C., Cook, A. T., Berthon, B., Mitchell, S., & Callister, R. (2011). Efficacy of a workplace-based weight loss program for overweight male shift workers: The Workplace POWER (Preventing Obesity Without Eating like a Rabbit) randomized controlled trial. Preventive Medicine, 52(5), 317–325. https://doi.org/10.1016/j.ypmed.2011.01.031 Mouchacca, J., Abbott, G. R., & Ball, K. (2013). Associations between psychological stress, eating, physical activity, sedentary behaviours and body weight among women: a longitudinal study. BMC Public Health, 13(1), 828. https://doi.org/10.1186/1471-2458-13- 828 Myers, S. L., Yang, C. Z., Bittner, G. D., Witt, K. L., Tice, R. R., & Baird, D. D. (2015). Estrogenic and anti-estrogenic activity of off-the-shelf hair and skin care products. Journal of Exposure Science & Environmental Epidemiology, 25(3), 271–277. https://doi.org/10.1038/jes.2014.32 National Population Comission (NPC) Nigeria and ICF International. (2014). Nigeria Demographic and Health Survey 2013. Abuja, Nigeria and Rockville, Maryland. Retrieved from https://dhsprogram.com/pubs/pdf/FR293/FR293.pdf NCD Risk Factor Collaboration (NCD-RisC). (2016). Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19·2 million participants. Lancet (London, England), 387(10026), 1377–1396. https://doi.org/10.1016/S0140-6736(16)30054-X Neupane, S., Prakash, K. C., & Doku, D. T. (2016). Overweight and obesity among women: analysis of demographic and health survey data from 32 Sub-Saharan African Countries. BMC Public Health, 16, 30. https://doi.org/10.1186/s12889-016-2698-5 Newbold, R. R., Padilla-Banks, E., Snyder, R. J., & Jefferson, W. N. (2005). Developmental exposure to estrogenic compounds and obesity. Birth Defects Research Part A: Clinical and Molecular Teratology, 73(7), 478–480. https://doi.org/10.1002/bdra.20147 Nuttall, F. Q. (2015). Body Mass Index: Obesity, BMI, and Health: A Critical Review. Nutrition Today, 50(3), 117–128. https://doi.org/10.1097/NT.0000000000000092 O’Rahilly, S., & Farooqi, I. S. (2008). Human obesity as a heritable disorder of the central control of energy balance. International Journal of Obesity, 32(S7), S55–S61. https://doi.org/10.1038/ijo.2008.239 Ode, J. J., Pivarnik, J. M., Reeves, M. J., & Knous, J. L. (2007). Body mass index as a predictor of percent fat in college athletes and nonathletes. Medicine and Science in Sports and Exercise, 39(3), 403–409. https://doi.org/10.1249/01.mss.0000247008.19127.3e Offer, A., Pechey, R., & Ulijaszek, S. (2010). Obesity under affluence varies by welfare regimes: The effect of fast food, insecurity, and inequality. Economics & Human Biology, 8(3), 297– 308. https://doi.org/10.1016/j.ehb.2010.07.002 Office National de la Statistique (ONS) [Mauritanie] and ORC Macro. (2001). Enquête

298

Démographique et de Santé Mauritanie 2000-2001. Calverton, Maryland, USA. Okorodudu, D. O., Jumean, M. F., Montori, V. M., Romero-Corral, A., Somers, V. K., Erwin, P. J., & Lopez-Jimenez, F. (2010). Diagnostic performance of body mass index to identify obesity as defined by body adiposity: a systematic review and meta-analysis. International Journal of Obesity, 34(5), 791–799. https://doi.org/10.1038/ijo.2010.5 Onubi, O. J., Marais, D., Aucott, L., Okonofua, F., & Poobalan, A. S. (2016). Maternal obesity in Africa: a systematic review and meta-analysis. Journal of Public Health, 38(3), e218–e231. https://doi.org/10.1093/pubmed/fdv138 Ouldzeidoune, N., Keating, J., Bertrand, J., & Rice, J. (2013). A description of female genital mutilation and force-feeding practices in Mauritania: implications for the protection of child rights and health. PloS One, 8(4), e60594. https://doi.org/10.1371/journal.pone.0060594 Oyedeji, A. T., Balogun, M. O., Akintomide, A. O., Sunmonu, T. A., Adebayo, R. A., & Ajayi, O. E. (2012). The “obesity Paradox” in Nigerians with heart failure. Annals of African Medicine, 11(4), 212–216. https://doi.org/http://dx.doi.org/10.4103/1596-3519.102851 Painter, R. C., Roseboom, T. J., & Bleker, O. P. (2005). Prenatal exposure to the Dutch famine and disease in later life: An overview. Reproductive Toxicology, 20(3), 345–352. https://doi.org/10.1016/j.reprotox.2005.04.005 Pampel, F. C., Denney, J. T., & Krueger, P. M. (2012). Obesity, SES, and economic development: A test of the reversal hypothesis. Social Science and Medicine, 74(7), 1073– 1081. https://doi.org/10.1016/j.socscimed.2011.12.028 Parlee, S. D., & MacDougald, O. A. (2014). Maternal nutrition and risk of obesity in offspring: The Trojan horse of developmental plasticity. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease, 1842(3), 495–506. https://doi.org/10.1016/j.bbadis.2013.07.007 Parsons, T. J., Power, C., & Manor, O. (2001). Fetal and early life growth and body mass index from birth to early adulthood in 1958 British cohort: longitudinal study. BMJ (Clinical Research Ed.), 323(7325), 1331–1335. https://doi.org/10.1136/BMJ.323.7325.1331 Piché, M.-E., Poirier, P., Lemieux, I., & Després, J.-P. (2018). Overview of Epidemiology and Contribution of Obesity and Body Fat Distribution to Cardiovascular Disease: An Update. Progress in Cardiovascular Diseases, 61(2), 103–113. https://doi.org/10.1016/j.pcad.2018.06.004 Popkin, B. M. (1998). The nutrition transition and its health implications in lower-income countries. Public Health Nutrition, 1(01), 5–21. https://doi.org/10.1079/PHN19980004 Popkin, B. M., Adair, L. S., & Ng, S. W. (2012). Global nutrition transition and the pandemic of obesity in developing countries. Nutrition Reviews, 70(1), 3–21. https://doi.org/10.1111/j.1753-4887.2011.00456.x Poston, L., Caleyachetty, R., Cnattingius, S., Corvalán, C., Uauy, R., Herring, S., & Gillman, M. W. (2016). Preconceptional and maternal obesity: epidemiology and health consequences. The Lancet Diabetes & Endocrinology, 4(12), 1025–1036. https://doi.org/10.1016/S2213- 8587(16)30217-0

299

Power, M. L., & Schulkin, J. (2008). Sex differences in fat storage, fat metabolism, and the health risks from obesity: possible evolutionary origins. British Journal of Nutrition, 99(05), 931–940. https://doi.org/10.1017/S0007114507853347 Puoane, T., Tsolekile, L., & Steyn, N. (2010). Perceptions about body image and sizes among Black African girls living in Cape Town. Ethnicity & Disease, 20(1), 29–34. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/20178179 Rahman, M., & Berenson, A. B. (2010). Accuracy of current body mass index obesity classification for white, black, and Hispanic reproductive-age women. Obstetrics and Gynecology, 115(5), 982–988. https://doi.org/10.1097/AOG.0b013e3181da9423 Raimi, T. H., & Dada, S. A. (2018). Lower BMI cut-off than the World Health Organization- based classification is appropriate for Nigerians. Journal of Diabetes and Endocrinology, 9(1), 1–10. https://doi.org/10.5897/JDE2017.0117 Raspopow, K., Abizaid, A., Matheson, K., & Anisman, H. (2010). Psychosocial stressor effects on cortisol and ghrelin in emotional and non-emotional eaters: Influence of anger and shame. Hormones and Behavior, 58(4), 677–684. https://doi.org/10.1016/j.yhbeh.2010.06.003 Ravaglia, G., Forti, P., Maioli, F., Boschi, F., Cicognani, A., & Gasbarrini, G. (1999). Measurement of Body Fat in Healthy Elderly Men: A Comparison of Methods. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 54(2), M70– M76. https://doi.org/10.1093/gerona/54.2.M70 Raynor, H. A., & Champagne, C. M. (2016). Position of the Academy of Nutrition and Dietetics: Interventions for the Treatment of Overweight and Obesity in Adults. Journal of the Academy of Nutrition and Dietetics, 116(1), 129–147. https://doi.org/10.1016/j.jand.2015.10.031 Roberto, C. A., Swinburn, B., Hawkes, C., Huang, T. T.-K., Costa, S. A., Ashe, M., … Brownell, K. D. (2015). Patchy progress on obesity prevention: emerging examples, entrenched barriers, and new thinking. The Lancet, 385(9985), 2400–2409. https://doi.org/10.1016/S0140-6736(14)61744-X Rubin, D. B. (1976). Inference and missing data. Biometrika, 63(3), 581–592. https://doi.org/10.1093/biomet/63.3.581 Sacks, G., Swinburn, B., & Lawrence, M. (2009). Obesity Policy Action framework and analysis grids for a comprehensive policy approach to reducing obesity. Obesity Reviews, 10(1), 76– 86. https://doi.org/10.1111/j.1467-789X.2008.00524.x Samuelsson, A.-M., Matthews, P. A., Argenton, M., Christie, M. R., McConnell, J. M., Jansen, E. H. J. M., … Taylor, P. D. (2008). Diet-Induced Obesity in Female Mice Leads to Offspring Hyperphagia, Adiposity, Hypertension, and Insulin Resistance: A Novel Murine Model of Developmental Programming. Hypertension, 51(2), 383–392. https://doi.org/10.1161/HYPERTENSIONAHA.107.101477 Schellong, K., Schulz, S., Harder, T., & Plagemann, A. (2012). Birth Weight and Long-Term Overweight Risk: Systematic Review and a Meta-Analysis Including 643,902 Persons from 66 Studies and 26 Countries Globally. PLoS ONE, 7(10), e47776.

300

https://doi.org/10.1371/journal.pone.0047776 Schwartz, M. W., Seeley, R. J., Zeltser, L. M., Drewnowski, A., Ravussin, E., Redman, L. M., & Leibel, R. L. (2017). Obesity Pathogenesis: An Endocrine Society Scientific Statement. Endocrine Reviews, 38(4), 267–296. https://doi.org/10.1210/er.2017-00111 Seravalle, G., & Grassi, G. (2017). Obesity and hypertension. Pharmacological Research, 122, 1–7. https://doi.org/10.1016/J.PHRS.2017.05.013 Sharp, G. C., Lawlor, D. A., Richmond, R. C., Fraser, A., Simpkin, A., Suderman, M., … Relton, C. L. (2015). Maternal pre-pregnancy BMI and gestational weight gain, offspring DNA methylation and later offspring adiposity: findings from the Avon Longitudinal Study of Parents and Children. International Journal of Epidemiology, 44(4), 1288–1304. https://doi.org/10.1093/ije/dyv042 Shibli-Rahhal, A., Van Beek, M., & Schlechte, J. A. (2006). Cushing’s syndrome. Clinics in Dermatology, 24(4), 260–265. https://doi.org/10.1016/j.clindermatol.2006.04.012 Simmonds, M., Llewellyn, A., Owen, C. G., & Woolacott, N. (2016). Predicting adult obesity from childhood obesity: a systematic review and meta-analysis. Obesity Reviews, 17(2), 95– 107. https://doi.org/10.1111/obr.12334 Sinha, R. (2008). Chronic stress, drug use, and vulnerability to addiction. Annals of the New York Academy of Sciences, 1141, 105–130. https://doi.org/10.1196/annals.1441.030 Sisnowski, J., Street, J. M., & Merlin, T. (2017). Improving food environments and tackling obesity: A realist systematic review of the policy success of regulatory interventions targeting population nutrition. PLOS ONE, 12(8), e0182581. https://doi.org/10.1371/journal.pone.0182581 Song, L., Mercer, L., Wakefield, J., Laurent, A., & Solet, D. (2016). Using Small-Area Estimation to Calculate the Prevalence of Smoking by Subcounty Geographic Areas in King County, Washington, Behavioral Risk Factor Surveillance System, 2009–2013. Preventing Chronic Disease, 13, 150536. https://doi.org/10.5888/pcd13.150536 Spalding, K. L., Arner, E., Westermark, P. O., Bernard, S., Buchholz, B. A., Bergmann, O., … Arner, P. (2008). Dynamics of fat cell turnover in humans. Nature, 453(7196), 783–787. https://doi.org/10.1038/nature06902 Stalsberg, R., & Pedersen, A. V. (2010). Effects of socioeconomic status on the physical activity in adolescents: a systematic review of the evidence. Scandinavian Journal of Medicine & Science in Sports, 20(3), 368–383. https://doi.org/10.1111/j.1600-0838.2009.01047.x Stein, A. D., Kahn, H. S., Rundle, A., Zybert, P. A., van der Pal–de Bruin, K., & Lumey, L. (2007). Anthropometric measures in middle age after exposure to famine during gestation: evidence from the Dutch famine. The American Journal of Clinical Nutrition, 85(3), 869– 876. https://doi.org/10.1093/ajcn/85.3.869 Sterne, J. A. C., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., … Carpenter, J. R. (2009). Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ, 338(jun29 1), b2393–b2393. https://doi.org/10.1136/bmj.b2393

301

Stevens, J., Juhaeri, Cai, J., & Jones, D. W. (2002). The effect of decision rules on the choice of a body mass index cutoff for obesity: examples from African American and white women. The American Journal of Clinical Nutrition, 75(6), 986–992. https://doi.org/10.1093/ajcn/75.6.986 Steyn, N. P., & Mchiza, Z. J. (2014). Obesity and the nutrition transition in Sub-Saharan Africa. Annals of the New York Academy of Sciences, 1311(1), 88–101. https://doi.org/10.1111/nyas.12433 Stunkard, A. J., Harris, J. R., Pedersen, N. L., & McClearn, G. E. (1990). The Body-Mass Index of Twins Who Have Been Reared Apart. New England Journal of Medicine, 322(21), 1483– 1487. https://doi.org/10.1056/NEJM199005243222102 Stutzmann, F., Tan, K., Vatin, V., Dina, C., Jouret, B., Tichet, J., … Meyre, D. (2008). Prevalence of melanocortin-4 receptor deficiency in Europeans and their age-dependent penetrance in multigenerational pedigrees. Diabetes, 57(9), 2511–2518. https://doi.org/10.2337/db08-0153 Swinburn, B. A., Sacks, G., Hall, K. D., McPherson, K., Finegood, D. T., Moodie, M. L., & Gortmaker, S. L. (2011). The global obesity pandemic: shaped by global drivers and local environments. The Lancet, 378(9793), 804–814. https://doi.org/10.1016/S0140- 6736(11)60813-1 Swinburn, B. A., Sacks, G., Lo, S. K., Westerterp, K. R., Rush, E. C., Rosenbaum, M., … Ravussin, E. (2009). Estimating the changes in energy flux that characterize the rise in obesity prevalence. The American Journal of Clinical Nutrition, 89(6), 1723–1728. https://doi.org/10.3945/ajcn.2008.27061 Swinburn, B., Sacks, G., & Ravussin, E. (2009). Increased food energy supply is more than sufficient to explain the US epidemic of obesity. The American Journal of Clinical Nutrition, 90(6), 1453–1456. https://doi.org/10.3945/ajcn.2009.28595 Tanalp, A. C. (2013). “Obesity paradox” or methodologic error. Ann Afr Med, 12(3), 191. https://doi.org/10.4103/1596-3519.117633 Tang-Péronard, J. L., Heitmann, B. L., Andersen, H. R., Steuerwald, U., Grandjean, P., Weihe, P., & Jensen, T. K. (2014). Association between prenatal polychlorinated biphenyl exposure and obesity development at ages 5 and 7 y: a prospective cohort study of 656 children from the Faroe Islands. The American Journal of Clinical Nutrition, 99(1), 5–13. https://doi.org/10.3945/ajcn.113.066720 The Demographic and Health Surveys (DHS) Program. (n.d.). GPS Data Collection: Methodology. Retrieved April 14, 2018, from https://dhsprogram.com/What-We-Do/GPS- Data-Collection.cfm The GBD 2015 Obesity Collaborators. (2017). Health Effects of Overweight and Obesity in 195 Countries over 25 Years. New England Journal of Medicine, 377(1), 13–27. https://doi.org/10.1056/NEJMoa1614362 Thorp, A. A., Owen, N., Neuhaus, M., & Dunstan, D. W. (2011). Sedentary Behaviors and Subsequent Health Outcomes in Adults. American Journal of Preventive Medicine, 41(2), 207–215. https://doi.org/10.1016/j.amepre.2011.05.004

302

Tie, H.-T., Xia, Y.-Y., Zeng, Y.-S., Zhang, Y., Dai, C.-L., Guo, J. J., & Zhao, Y. (2014). Risk of childhood overweight or obesity associated with excessive weight gain during pregnancy: a meta-analysis. Archives of Gynecology and Obstetrics, 289(2), 247–257. https://doi.org/10.1007/s00404-013-3053-z Tokunaga, K., Matsuzawa, Y., Ishikawa, K., & Tarui, S. (1983). A novel technique for the determination of body fat by computed tomography. International Journal of Obesity, 7(5), 437–445. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6642855 Tomiyama, A. J. (2019). Stress and Obesity. Annual Review of Psychology, 70(1), annurev- psych-010418-102936. https://doi.org/10.1146/annurev-psych-010418-102936 Tomiyama, A. J., Schamarek, I., Lustig, R. H., Kirschbaum, C., Puterman, E., Havel, P. J., & Epel, E. S. (2012). Leptin concentrations in response to acute stress predict subsequent intake of comfort foods. Physiology & Behavior, 107(1), 34–39. https://doi.org/10.1016/j.physbeh.2012.04.021 Tooze, J. A., Schoeller, D. A., Subar, A. F., Kipnis, V., Schatzkin, A., & Troiano, R. P. (2007). Total daily energy expenditure among middle-aged men and women: the OPEN Study. The American Journal of Clinical Nutrition, 86(2), 382–387. https://doi.org/10.1093/ajcn/86.2.382 Traurig, M., Mack, J., Hanson, R. L., Ghoussaini, M., Meyre, D., Knowler, W. C., … Baier, L. J. (2009). Common Variation in SIM1 Is Reproducibly Associated With BMI in Pi ma Indians. Diabetes, 58(7), 1682–1689. https://doi.org/10.2337/db09-0028 Tremblay, M. S., Aubert, S., Barnes, J. D., Saunders, T. J., Carson, V., Latimer-Cheung, A. E., … Chinapaw, M. J. M. (2017). Sedentary Behavior Research Network (SBRN) – Terminology Consensus Project process and outcome. International Journal of Behavioral Nutrition and Physical Activity, 14(1), 75. https://doi.org/10.1186/s12966-017-0525-8 Tremmel, M., Gerdtham, U.-G., Nilsson, P., Saha, S., Tremmel, M., Gerdtham, U.-G., … Saha, S. (2017). Economic Burden of Obesity: A Systematic Literature Review. International Journal of Environmental Research and Public Health, 14(4), 435. https://doi.org/10.3390/ijerph14040435 Troiano, R. P., Frongillo, E. A., Sobal, J., & Levitsky, D. A. (1996). The relationship between body weight and mortality: a quantitative analysis of combined information from existing studies. International Journal of Obesity and Related Metabolic Disorders : Journal of the International Association for the Study of Obesity, 20(1), 63–75. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8788324 UCLA Institute for Digital Research and Education. (n.d.). Multiple Imputation in SAS Part 1. Retrieved April 14, 2018, from https://stats.idre.ucla.edu/sas/seminars/multiple-imputation- in-sas/mi_new_1/ United Nations. (2015). Transforming our world: The 2030 agenda for sustainable development. Agenda. https://doi.org/10.1007/s13398-014-0173-7.2 United Nations Economic and Social Council. (2016). Report of the Inter-Agency and Expert Group on Sustainable Development Goal Indicators. Retrieved from http://ggim.un.org/knowledgebase/KnowledgebaseArticle51479.aspx

303

United States Agency for International Development (USAID). (2018). The Demographic and Health Surveys Program. Retrieved from https://dhsprogram.com/ Valvi, D., Mendez, M. A., Garcia-Esteban, R., Ballester, F., Ibarluzea, J., Goñi, F., … Vrijheid, M. (2014). Prenatal exposure to persistent organic pollutants and rapid weight gain and overweight in infancy. Obesity, 22(2), 488–496. https://doi.org/10.1002/oby.20603 Vickers, M. H., Breier, B. H., Cutfield, W. S., Hofman, P. L., & Gluckman, P. D. (2000). Fetal origins of hyperphagia, obesity, and hypertension and postnatal amplification by hypercaloric nutrition. American Journal of Physiology-Endocrinology and Metabolism, 279(1), E83–E87. https://doi.org/10.1152/ajpendo.2000.279.1.E83 Wells, J. C. K., & Fewtrell, M. S. (2006). Measuring body composition. Archives of Disease in Childhood, 91(7), 612–617. https://doi.org/10.1136/adc.2005.085522 World Health Organization. (2000). Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organization technical report series (Vol. 894). Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11234459 World Health Organization. (2011). Waist Circumference and Waist-Hip Ratio Report of a WHO Expert Consultation. Waist Circumference and Waist–Hip Ratio: Report of a WHO Expert Consultation. https://doi.org/10.1038/ejcn.2009.139 World Health Organization. (2013). Global action plan for the prevention and control of noncommunicable diseases 2013-2020. World Health Organization. https://doi.org/978 92 4 1506236 World Health Organization. (2017a). Depression and other common mental disorders: global health estimates. World Health Organization. https://doi.org/CC BY-NC-SA 3.0 IGO World Health Organization. (2017b). Global Health Observatory Data Repository. Retrieved from http://apps.who.int/gho/data/?theme=main World Health Organization. (2017c). Obesity and overweight. Retrieved November 21, 2017, from http://www.who.int/mediacentre/factsheets/fs311/en/ World Health Organization. (2018a). Obesity and overweight: Key Facts. Retrieved August 20, 2018, from http://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight World Health Organization. (2018b). What is overweight and obesity? Retrieved April 23, 2018, from http://www.who.int/dietphysicalactivity/childhood_what/en/ World Health Organization (WHO). (1995). Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. World Health Organization Technical Report Series. https://doi.org/854 World Health Organization (WHO) Expert Consultation. (2004). Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. The Lancet, 363(9403), 157–163. https://doi.org/10.1016/S0140-6736(03)15268-3 Yang, J. S., Mamudu, H. M., & John, R. (2018). Incorporating a structural approach to reducing the burden of non-communicable diseases. Globalization and Health, 14(1), 66. https://doi.org/10.1186/s12992-018-0380-7

304

Yu, Z., Han, S., Zhu, J., Sun, X., Ji, C., & Guo, X. (2013). Pre-Pregnancy Body Mass Index in Relation to Infant Birth Weight and Offspring Overweight/Obesity: A Systematic Review and Meta-Analysis. PLoS ONE, 8(4), e61627. https://doi.org/10.1371/journal.pone.0061627 Zhao, Y., Wang, S.-F., Mu, M., & Sheng, J. (2012). Birth weight and overweight/obesity in adults: a meta-analysis. European Journal of Pediatrics, 171(12), 1737–1746. https://doi.org/10.1007/s00431-012-1701-0 Zheng, W., McLerran, D. F., Rolland, B., Zhang, X., Inoue, M., Matsuo, K., … Potter, J. D. (2011). Association between Body-Mass Index and Risk of Death in More Than 1 Million Asians. New England Journal of Medicine, 364(8), 719–729. https://doi.org/10.1056/NEJMoa1010679 Zoeller, R. T., Brown, T. R., Doan, L. L., Gore, A. C., Skakkebaek, N. E., Soto, A. M., … Vom Saal, F. S. (2012). Endocrine-Disrupting Chemicals and Public Health Protection: A Statement of Principles from The Endocrine Society. Endocrinology, 153(9), 4097–4110. https://doi.org/10.1210/en.2012-1422

305

APPENDIX Missing Data Patterns for DHS Dataset

1. Burkina Faso Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealth_ age education BMI Frequency Percent wealth age education BMI

1.0544133.030 1 X X X X 5274 49.26 8 338 0.072431 1.994501

1.0835733.154 2 X X X . 5432 50.73 9 087 0.073085 .

2.0000030.000 3 X X . . 1 0.01 0 000 . .

2. Burundi Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns

Group Means

Group wealth_ age education BMI Frequency Percent wealth age education BMI

0.8511833.29419 1.96602 1 X X X X 2325 48.81 3 4 0.104946 2

0.8847433.52255 2 X X X . 2438 51.19 2 9 0.108696 .

306

3. DR Congo Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealth_ age education BMI Frequency Percent wealth age education BMI

1.08262 33.024 1 X X X X 5313 49.93 8 092 0.363072 2.021080

1.11582 32.781 2 X X X . 5327 50.07 5 491 0.376384 .

4. Congo-Brazzaville Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealth age education BMI Frequency Percent wealth age education BMI

33.151 1 X X X X 3601 52.38 0.942516 069 0.557623 2.111358

33.545 2 X X X . 3274 47.62 0.941662 816 0.560476 .

307

5. Cote D' Ivoire Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealth age education BMI Frequency Percent wealth age education BMI

0.92190 32.8494 1 X X X X 2830 47.43 8 70 0.120495 2.217668

0.95154 32.8565 2 X X X . 3137 52.57 6 51 0.129742 .

6. Cameroon Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealth age education BMI Frequency Percent wealth age education BMI

1 X X X X 4479 51.51 1.06965 32.896 0.372181 2.316142 8 852

2 X X X . 4217 48.49 1.03746 33.154 0.377520 . 7 138

308

7. Gabon Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealt age educati BMI Frequen Percent wealth age education BMI h on cy

1 X X X X 3317 66.21 1.0006 33.9 0.559542 2.4094 03 372 06 93

2 X X X . 1693 33.79 1.0112 33.5 0.585351 . 23 410 51

8. Ghana Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Frequenc Percen Group wealth age education BMI y t wealth age education BMI

1.07878 1 X X X X 2805 50.41 8 34.795722 0.472371 2.396078

1.07212 2 X X X . 2759 49.59 8 34.696629 0.480609 .

309

9. Gambia Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealt age educati BM Frequen Percen wealt age educatio BMI h on I cy t h n

1 X X X X 2495 45.80 1.113 32.313 0.23847 2.14549 427 427 7 1

2 X X X . 2953 54.20 1.134 32.567 0.23772 . 440 220 4

10. Guinea Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati BM Frequen Percen wealth age educatio BMI p h on I cy t n

1 X X X X 2746 50.75 0.9978 33.269 0.10196 2.11107 15 483 6 1

2 X X X . 2665 49.25 0.9958 33.296 0.11369 . 72 060 6

310

11. Kenya Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt ag educati BM Frequen Percen wealth age educatio BMI p h e on I cy t n

1 X X X X 9358 46.69 0.9535 33.192 0.31096 2.26138 16 990 4 1

2 X X X . 10686 53.31 0.9610 33.297 0.31667 . 71 586 6

12. Liberia Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealt age educati BMI Frequen Perce wealt age educatio BMI h on cy nt h n

1 X X X X 3083 50.58 0.921 33.11 0.23418 2.23321 505 4175 7 4

2 X X X . 3012 49.42 0.961 33.43 0.23605 . 819 1607 6

311

13. Lesotho Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati BMI Frequen Percen wealt age educatio BMI p h on cy t h n

1 X X X X 2100 52.02 1.054 33.00 0.49857 2.50523 762 5238 1 8

2 X X X . 1937 47.98 1.083 33.11 0.50490 . 118 0996 4

14. Madagascar Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealt age educati BM Percent wealt age educatio BMI h on I Frequenc h n y

1 X X X X 5074 48.28 0.857 33.40 0.29030 1.83563 509 0276 4 3

2 X X X . 5436 51.72 0.885 33.70 0.30261 . 762 6402 2

312

15. Mali Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati BMI Frequency Percen wealt age educatio BMI p h on t h n

1 X X X X 3300 50.62 0.857 32.2 0.09181 2.13818 879 542 8 2 42

2 X X X . 3219 49.38 0.873 32.0 0.09692 . 874 245 5 42

16. Malawi Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Frequenc Group wealth age education BMI y Percent wealth age education BM

0.87029 32.10802 1 X X X X 5258 32.86 3 6 0.240396 2.197794

0.88299 32.32197 2 X X X . 10743 67.14 4 7 0.230383 .

313

17. Niger Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealth age educati BM Frequen Percen wealt age educatio BMI p _ on I cy t h n

1 X X X X 3159 45.75 0.945 32.1 0.06362 2.13042 552 547 8 1 96

2 X X X . 3733 54.06 0.998 32.2 0.08679 . 928 582 3 37

3 X X . X 5 0.07 1.600 30.2 . 1.80000 000 000 0 00

4 X X . . 8 0.12 1.750 32.0 . . 000 000 00

314

18. Namibia Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt ag educatio BM Frequen Percen wealth age educatio BMI p h e n I cy t n

1 X X X X 2675 47.79 1.0807 33.7 0.69682 2.33644 48 1626 2 9 2

2 X X X . 2922 52.21 1.0749 33.6 0.70533 . 49 5366 9 2

19. Rwanda Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Percen Group wealth age education BMI Frequency t wealth age education BMI

0.97026 33.92847 1 X X X X 3733 49.35 5 6 0.137691 2.195553

0.97990 34.01122 2 X X X . 3832 50.65 6 1 0.136743 .

315

20. Sierra Leone Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati BM Frequen Percen wealt age educatio BMI p h on I cy t h n

1 X X X X 4908 48.95 1.049 33.24 0.17909 2.1574 307 8778 5 98

2 X X X . 5119 51.05 1.073 32.99 0.19203 . 256 8242 0

21. Senegal Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati BM Frequen Percen wealth age educatio BMI p h on I cy t n

1 X X X X 3246 36.09 1.0043 32.26 0.08133 2.06469 13 3093 1 5

2 X X X . 5747 63.91 0.9989 31.77 0.07795 . 56 7971 4

316

22. Chad Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Group wealt age educati BM Frequen Percen wealth age educatio BMI h on I cy t n

1 X X X X 3246 36.09 1.0043 32.26 0.08133 2.06469 13 3093 1 5

2 X X X . 5747 63.91 0.9989 31.77 0.07795 . 56 7971 4

23. Togo Missing Data Pattern with Categorical BMI

The MI Procedure

Missing Data Patterns Group Means

Grou wealt age educati B Frequen Percen wealt age educatio BMI p h on MI cy t h n

1 X X X X 6671 64.52 1.034 32.61 0.08214 1.92909 328 2952 7 6

317

VITA

IFEOMA DOREEN OZODIEGWU

Education DrPH, Epidemiology, East Tennessee State University, Johnson City, May 2019

MPH, Health Services Administration, East Tennessee State University, Johnson City, May 2012

B.Sc., Biochemistry, Enugu State University of Science and Technology, Enugu, Nigeria, December 2007

Professional Experience Graduate Teaching and Research Assistant, Department of Epidemiology, East Tennessee State University, 2015 – 2019

Regional Monitoring and Evaluation Consultant, Control of Neglected Tropical Diseases (NTD), ENVISION Global Health Division RTI International; Enugu, Nigeria, 2015 – 2015

Management Information Systems Consultant, Global Fund Malaria Project, Society for Family Health (SFH); Enugu, Nigeria, 2014 - 2015

Publications Liu Y, Ozodiegwu ID, Yu Y, Hess R, Bie R. (2017). An Association of Health Behaviors with Depression and Metabolic Risks: Data from 2007 to 2014 U.S. National Health and Nutrition Examination Survey. Journal of Affective Disorders, Vol 217, 190-196.

Liu Y, Ozodiegwu ID, Nickel J.C, Wang K, R, Iwasaki L. (2017). Self-reported Health and Behavioral Factors are Associated with Metabolic Syndrome in Americans aged 40 and over. Preventive Medicine Reports, Vol 7, 193-197

Quinn M, Caldara G, Collins K, Owens H, Ozodiegwu ID, Loudermilk E, Stinson J. D. (2017). Methods for Understanding Childhood Trauma: Modifying the Adverse Childhood Experiences International Questionnaire for Cultural Competency. International Journal of Public

318

Health https://doi.org/10.1007/s00038-017-1058-2

Ariyo O, Ozodiegwu ID, Doctor HV. (2017). The Influence of the Social and Cultural Environment on Maternal Mortality in Nigeria: Evidence from the 2013 Demographic and Health Survey. PLOS One https://doi.org/10.1371/journal.pone.0190285

Honors and Awards Rotary International. 2018 Rotary Global Grant/Skelton Jones Scholarship Recipient

East Tennessee State University. 2017 Sherrod Graduate Student Scholarship Award for Excellence in Research

319