SPH Theses and Dissertations

This dissertation undertakes the task of going beyond traditional health technology assessments in areas of reproductive and child health. In my first paper, I evaluate a reproductive health technology intervention program intended to integrate postpartum family planning services into routine maternity care in Nepal using primary data from a randomized controlled trial. I find that the integration is feasible, and it could benefit women with need for spacing or limiting their births. Expanding counseling services of higher quality could further increase access to, and uptake of postpartum contraception. The second paper investigates the discrepancies in the impact of expanding family planning access on unintended pregnancies when the heterogeneity of the population being treated is not considered. Current methods that estimate observed failure rates do not account for women selecting the method they use. Correcting for this user heterogeneity, I find that women who are not using contraception despite unmet need have lower risk of pregnancy, and the program impact of expanding family planning access is higher than currently estimated.
The final paper models and appraises the relative importance of contextual factors that could modulate the impact of scaling up Rapid Diagnostic Tests (RDTs) for malaria using a static deterministic decision model and global sensitivity analysis techniques. I find that behavioral parameters such as compliance to disease diagnosis, uptake of RDTs by providers and patients during scale up, overall effectiveness of Artemisinin Combination Therapies (the first line treatment for malaria), and epidemiological parameter of co-incidence of malaria and bacterial pneumonia are the most important factors influencing the impact and effectiveness of RDTs on under-five malaria and pneumonia burden in 24 malaria endemic Sub-Saharan African countries given their current disease epidemiology and care-seeking patterns.

This dissertation uses mixed methods to understand women’s perspectives on contraceptive access and provision, and to create and test a new population-based quantitative indicator that measures various dimensions of autonomy in family planning.
The introduction to the dissertation charts the intellectual history of the contemporary family planning movement to elucidate how we arrived at the present moment with the existing tools in our measurement toolbox.
The first paper develops a novel concept called “contraceptive autonomy,” which is defined as the various factors are necessary for a woman to decide for herself what she wants as it relates to contraceptive use, and then to realize that decision. This paper then proposes an algorithm to operationalize the measure of the three subdomains of contraceptive autonomy (free choice, full choice and informed choice) into two population-based indicators: the stand-alone autonomy score and the autonomy-adjusted contraceptive prevalence rate.
The second paper uses qualitative methods to explore women’s subjective experiences with barriers to contraceptive autonomy in a sub-Saharan African country. The analysis is based on 49 in-depth interviews with women of reproductive age and 3 key-informant interviews with health system administrators. Findings include a spectrum of coercive experiences ranging from the subtle to the overt, as well as the importance of systemic and structural causes of contraceptive coercion. The final paper is an analysis of quantitative data collected from household surveys in rural Burkina Faso to generate a proof-of-concept of the contraceptive autonomy algorithm. This analysis tests and applies the concepts and methods described in Paper 1 to a random sample of women ages 15-49 (n=2654). Findings show that lack of informed choice is the largest contributor to contraceptive non-autonomy in this setting. Logistic regression models show that predictors of contraceptive autonomy in this setting include education, nulliparity, and being married.
By creating new tools that can be used by those who implement, and evaluate contraceptive programs to better gauge the extent to which these programs respect women’s autonomy, as well as furnish an improved understanding what the major barriers to autonomy are, the ultimate goal of this dissertation is to contribute to an improved respect for women’s choices within global family planning programs.

As data to inform evidence-based policy of new and existing HPV vaccination strategies in LMICs are limited, this dissertation focuses on the costing and evaluation of the HPV vaccine utilizing modeling and economic analyses. In Paper 1, we developed a predictive model for estimating standardized immunization delivery unit cost estimates at the country level. The predicted programmatic, economic costs per dose for routine delivery of childhood vaccines was $1.49 (95% uncertainty range: $0.33–4.87) and for HPV vaccines was $2.04 ($0.66–6.20). Country-specific costs modeled within a Bayesian meta-regression framework provide a broad indication of immunization delivery costs that may be preferable to raw country-level data without reliable, standardized review. In Paper 2, we used a multiple modeling approach to estimate the health and economic outcomes associated with HPV vaccine delivery in Uganda, a low-income country in East Africa, where cervical cancer is the leading cause of cancer among females. Using a wide range of plausible scenarios to assess what might be achieved by a campaign delivery strategy for HPV vaccination against HPV-16/18 infections, we found that campaign strategies yielded greater health benefits if campaigns occurred frequently and targeted a wide age range compared with routine HPV vaccination strategies. In Paper 3, relying on the methodology of extended cost-effectiveness analysis (ECEA) to examine the distributional and financial risk protection benefits from HPV vaccination in Ethiopia, we found that routine two-dose HPV vaccination could avert 586,000 cervical cancer deaths and 18,900 cases of catastrophic health expenditure over 2019-2118, assuming 40% vaccination coverage and 100% efficacy against HPV-16/18 with lifelong duration of protection. Approximately 30% of health benefits would accrue to the poorest quintile whereas 66% of financial risk protection benefits accrue to the poorest quintile.

Child health has been on top of the global health agenda since the Millennium Development Goal era ending in 2015. It is well recognized that a vital and productive society with a prosperous and sustainable future is built on a foundation of healthy child development. The determinants of child health are multifaceted. In this dissertation, several factors affecting children’s healthy development are explored. Using nationally representative data from Jamaica, Chapter 1 focuses on the effects of a health policy on child health, via empirically examining how the national-wide user-fee-removal policy affects children’s care seeking behavior and the financial burden their families faced, with an equity focus among children of various wealth levels. I found that in the short-term, the policy deteriorated equity of access to health care for children as the richer families appeared to benefit more; while the equity status improved fast in the medium-to-long-term as the poorer children caught up. Besides the effects of health programs, I further examine the cost of scaling up the health intervention in Chapter 2, and perform cost-effectiveness analysis (CEA) of a nutritional package program (Yingyangbao [YYB]), which is a nutrient-dense food supplement targeting infants and young children, on children’s stunting status in the context of China. Different from the traditional CEA study, I use extended cost-effectiveness analysis (ECEA) method to assess the distributional impact of YYB rollout across provinces and socioeconomic groups. Our result shows that YYB could be a pro-poor intervention that brings substantial health benefits to poor Chinese children, but with large variations across provinces from as low as ¥800 (Chongqing province) to as high as ¥23,400 (Jilin province). The determinants of child health are multi-dimensional. Many factors are beyond the reach of health sector, and require multisectoral effort to support childhood development. In Chapter 3, I assess the impact of an environmental factor on child health, via examining how prenatal exposure to sand-and-dust storms (SDS) affects children’s cognitive development using nationally representative data from China. I find prenatal SDS exposure is strongly associated with lower mathematics and word-recognition test scores, as well as additional months to begin speaking in sentences. Despite the findings of this thesis, more work should be conducted to identify mechanisms behind the findings, such as the scientific pathway for pollutant to affect cognitive development.

This dissertation is comprised of three studies that evaluate disease burden in the United States, for specific chronic and infectious diseases. These studies use mathematical modelling to synthesize empirical evidence and estimate both epidemiological and economic outcomes. In the first paper, I investigate trends in the prevalence and incidence of diabetes and diabetes diagnosis among adults ages 20 years and older in the United States, over the period 2000-2016. Using an age-stratified Markov model of undiagnosed and diagnosed diabetes, I examine trends in true incidence of diabetes and the diagnosis rates. The model is estimated using repeated cross-sectional survey data on the prevalence of undiagnosed and diagnosed diabetes from the National Health and Examination Survey (NHANES) 1988-1994 and 1999-2016. In the second paper, I develop a novel model to assess the 10-year risk of fatal-plus-non-fatal cardiovascular disease (CVD) for patients with type 2 diabetes mellitus in the United States. This model is constructed as a sex-and-cohort stratified Cox proportional-hazards model using pooled data on fatal-plus-non-fatal CVD outcomes from 5 prospective cohorts. The resulting risk prediction equation provides more accurate predictions of total CVD risk compared to current risk scores and can to be used in future comparative effectiveness and cost-effectiveness analyses to simulate outcomes of primary intervention policies targeted toward diabetic populations. In the third paper, I investigate disparities in health and economic outcomes associated with N. gonorrhoeae infection in the US in 2015. With probability tree models, I quantify the lifetime quality-adjusted life-years and costs associated with gonorrhea and its sequelae in the US and examine disparities in burden across race/ethnicity. These three studies report findings of substantive importance within each disease area. They also illustrate the utility of mathematical models for synthesizing data, estimating outcomes that would be difficult or impossible to measure empirically, and answering questions directly relevant to the goals of planning and prioritizing prevention policies.

Health systems aim to improve population health. Despite global efforts, millions of children still die every year from vaccine preventable diseases and undernutrition attributed deaths. Moreover, about 293,000 maternal deaths occurred in 2013. The sources of these deaths include various inequalities such as vaccine coverage, nutritional status, and health services utilization. In order to make progress toward mortality reduction, we need to address the sources that are most likely to affect the poorest. One strategy is conditional cash transfers (CCTs), which provides cash payments in exchange for compliance with health-related conditionality. This dissertation explores evidence from two large randomized experiments in Indonesia, PKH (a large-scale household CCT) and Generasi (a large-scale incentivized community block grant). Chapter two investigates whether PKH improves vaccination coverage among poor children. After two years of implementation, the results show that PKH leads to significant increase in vaccination coverage for all vaccine types among children younger than 12 months old. The evidence also suggests that PKH is equity enhancing by reducing the differences in vaccination coverage between children living in more and less supply-ready areas and children of more and less educated mothers.
Chapter three investigates the impact of PKH and Generasi on child food intake. The results show that both programs increase child food intake particularly for protein-rich items. They increase milk and fish intake up to 19% and 14% compared to the control group means, respectively. Improving child nutrition outcomes, PKH reduces wasting and severe wasting up to 41% and Generasi reduces the prevalence of severely underweight children up to 47%, compared to the control areas. Chapter four explores how PKH and Generasi help improve determinants of maternal mortality among poor women. Evidence shows different results between the two programs with Generasi produces more positive impact in many aspects of determinants. Both programs, however, are unlikely to have a large effect on maternal mortality due to factors that might significantly reduce the program’s effectiveness. For instance, while the programs improved utilization, they did so at community-based facilities, which are not appropriate for delivery services in the case of obstetric emergency.

Malaria and HIV are the two diseases that cause the most disability and loss of life in sub-Saharan Africa. Treatments play a critical role in ameliorating the impact of these diseases on patients and their communities. This dissertation focuses on three elements of treatment: The impact of HIV treatment on childhood education, integration of HIV care and sexual and reproductive healthcare, and adherence to treatment.

In chapter one, we assessed the impact of adult HIV treatment on the educational attainment of children in the same household through a regression discontinuity analysis. Adult HIV treatment results in large gains in the educational attainment of children in the same household (intention-to-treat [ITT]: 0.30 years, p=0.017; complier average causal effect [CACE]: 1.17, p=0.036). This increase in educational attainment is important for increasing individual and community well-being and human capital.

In chapter two, we estimated the association between moving through the HIV treatment cascade and contraceptive use through a bivariate probit analysis. We find large increases in condom use among HIV-positive women as they move through the treatment cascade, with increases of 22.8 percentage points (p<0.001) among women on ART for 4-7 years relative to women who are unaware of their HIV-positive status. This finding is a positive indication of the integration of HIV care and reproductive healthcare.

In chapter three, we conducted a randomized trial to assess the impact of text message reminders on adherence to antimalarial treatment and in chapter four we estimated whether patient factors predict antimalarial treatment adherence and modify the impact of text reminders using logistic regression models. We found that a short, simple text message reminder increases adherence to antimalarial treatment (OR: 1.45, p=0.030) and that no patient factors we assessed statistically significantly predicted antimalarial treatment adherence. While adherence to treatment remains difficult to predict, short, simple text message reminders could increase antimalarial treatment adherence.

The global response to HIV is in a transition state, moving from the rapid roll out of HIV treatment services to sustained delivery and expansion into new populations. This dissertation analyzes three scenarios in which improved insights into data collection, analysis, and policy are likely to yield improved health outcomes in the context of HIV/AIDS in rural KwaZulu-Natal, South Africa. Firstly, we investigate HIV health system gaps using a longitudinal transition-by-transition cascade of care, a concept which breaks down the pathway from infection to therapeutic response into discrete steps. We demonstrate biases due to transition time in the standard cross-sectional approaches, and recast the cascade using patient flow between stages using survival analysis. We find that the strongest bottleneck in the cascade is becoming linked to care, and that these transitions are likely to be slowing over time. Secondly, we conduct a field experiment to identify whether list randomization can reduce social biases in survey responses related to HIV, using known truths for comparison. List randomization is a survey method intended to mitigate respondent biases by shrouding patient responses in a longer list of questions. We pilot this method using surveillance-derived known status for HIV status. We find list randomization did not accurately reveal known true statuses and behaviors. Finally, we investigate whether the qualification structure of the South African disability grant program, in which qualification for some requires having a low CD4 count, may have negatively impacted HIV health in certain circumstances by disincentivizing antiretroviral therapy adherence. We assess this empirically using a difference-in-difference of recovery rates between disability grant recipients and non-recipients specific to the qualification threshold, finding that manipulation of recovery rates due to the threshold is likely to have occurred.

This dissertation is comprised of three studies that, together, explore the links between: 1) access to and use of maternal and child health care, family planning, and reproductive health services; 2) fertility and maternal and child health outcomes; and 3) longer-term measures of well-being, in developing country contexts. I utilize both theoretical and empirical methods to explore these relationships of interest. In the first study, I use a large multi-country microeconomic data set to assess the associations between physical access to services and utilization of maternal and child health services as well as child mortality outcomes. I find that living close to a health facility is associated with increased utilization of maternal health services, namely receipt of antenatal care and in-facility delivery, as well as improved child survival, particularly in the neonatal period. In the second study, I adopt a macrosimulation approach to investigate the long-run health and economic effects of a decline in fertility. I construct a model that describes the interrelated evolution of economic and demographic outcomes under a “baseline” scenario, in which fertility declines slowly over time, and I compare these outcomes under a simulated “alternative” scenario in which fertility declines more rapidly. I calibrate the model parameters using findings from well-identified microeconomic studies, and I use baseline data from Nigeria to compare the model’s predictions for each of the key outcomes under the two fertility scenarios. Through this modeling exercise, I show that a decline in fertility creates the potential for a demographic dividend and a window of opportunity for economic growth; moreover, the magnitude of this growth factor may be substantially larger than what has been estimated to date. In the third study, I use population-based survey data to identify children growing up in healthy environments in low- and middle-income countries, and I compare the height distribution of these children to the height distribution of the international growth reference sample that was established by the World Health Organization. I find that observed differences in child height across populations are likely not due to innate or genetic differences, but are more likely to reflect children’s continued exposure to resource-poor environments, poor maternal education, and lack of access to health and sanitation.

This dissertation includes three papers in the field of education and health. This first paper provides estimates of the effects of secondary schooling on HIV infection and childbearing. A 1996 policy reform in Botswana changed the grade structure of secondary school and led to sharp increases in educational attainment among affected birth cohorts. Exploiting cohort-specific exposure and differential impact by birth village as ‘natural experiments’, I find that the reform decreased HIV infection and childbearing. Turning to mechanisms, I find that schooling had no effect on HIV knowledge; however it influenced sexual debut, norms and behaviors, labor force participation, and literacy. In HIV endemic settings, estimates of the returns to secondary schooling may be underestimated due to the exclusion of health benefits. The second paper uses data from the Tanzania Censuses to investigate whether additional schooling in children affects their parent’s old-age survival. To do so, I exploit quasi-random variation resulting from a policy reform that initiated a drive towards Universal Primary Education in the mid-1970s. I show that the reform caused a large increase in school enrollment. Additional primary schooling in children as a result of the reform induced large reductions in the probability of parental death by the time exposed child cohorts had reached age 40. These survival gains highlight the large societal benefits of human capital investment, particularly in low-resource settings. In the third paper, I exploit the 1980 Zimbabwe education reform, which rapidly expanded access to secondary schools. Exploiting cohort-specific exposure, I assess whether additional schooling among parents affected the risk of undernutrition in their children under five. I find no evidence of a causal effect of parental schooling on undernutrition in their children at the national level. Among urban and wealthier households, additional maternal schooling had a protective effect against the risk of wasting in their children. Parental schooling may play a more muted role in child investment decisions than previously suggested. This essay emphasizes the need for directed health investments to improve child development in low-resource settings.

In anticipation of population aging, governments are implementing policies that seek to improve elderly well‑being and protect them from financial hardship. This dissertation begins by evaluating the health effects of two popular pension reforms in Latin America- privatization of social security and non‑contributory pensions. In the first essay, I evaluate the effect of social security privatization on survival. Using initial fund selection as an instrumental variable for pension wealth at legal retirement age, I find that increases in pension wealth reduce old age mortality and lead to large differences in life expectancy across income quintiles. In the second essay, I study the impact of a non‑contributory pension program on health care utilization patterns and find that receiving cash transfer pensions increases overall health care utilization, but more importantly leads to a shift from low‑quality to high‑quality services. Despite increased utilization, there is no evidence of higher expenditures, which can be partially explained by the increased uptake of public health insurance by those receiving non‑contributory pensions. The first two essays exemplify the link that exists between economic well‑being and health, a relationship that has been made explicit in the international agenda with the Sustainable Development Goals. Nevertheless, there are few national welfare measures that combine household‑level economic well‑being and longevity. The last essay develops a proof of concept measure adjusting quality of life for poverty, that is analogous and complementary to the Disability‑Free Life Expectancy index. The proposed index, Poverty‑Free Life Expectancy, estimates the average years an individual can expect to live out of poverty and ranges considerably across countries with females losing larger shares of life than males, offsetting the gender gap in life expectancy. This essay shows that, as household data becomes increasingly available in the developing world, more precise measures of quality of life should be be desisgned and adopted to guide policy making.

Maternal and child mortality, particularly during the neonatal period, are among the most challenging global health issues of this era. This burden disproportionately affects the poorest populations, across and within countries. And although many of these deaths would be avertable, improvements in most countries have been slow. This dissertation explores three main research questions: (1) what is the effect of maternal health on infant outcomes?; (2) what survival gains could be attained through improved interventions, across the continuum of care?; and (3) how do health system characteristics affect the potential impact and cost-effectiveness of such interventions? The first paper uses decision modeling to evaluate how increased use of family planning and of improved intrapartum care could reduce maternal deaths in Nepal—as well as the cost-effectiveness of doing so, and of accompanying interventions to achieve these targets. The second paper estimates the potential impact of administering interventions from the Safe Childbirth Checklist at health facilities in India, and how “real world” implementation might see different results due to health system characteristics. Lastly, the third paper examines child survival outcomes following a maternal death in Ethiopia, using a long-term household-level longitudinal dataset. Together, these papers aim to provide new insights on approaches to reducing the high level of mortality among women and children.

This dissertation is comprised of three studies that examine three global and national-level policies, and apply different quantitative analyses to improve the research base that informs these policies, with the aim of ultimately improving the designs of existing health policies. Chapter 2 examines the UNAIDS’ goal to eliminate AIDS by 2030. It combines survival analysis of a longitudinal dataset and a Markov model of progression through different stages of HIV care cascade, and find that the mathematical models that informed the UNAIDS’ policy overestimates the health benefits that could be realized in real life. Chapter 3 examines South Africa’s Integrated Chronic Disease Management model, using regression models I conclude that how different types of multimorbidity affects the care patients receive should be considered when designing care delivery in order to provide coherent and efficient care. Chapter 4 assesses the target set by the Global Vaccine Action Plan, which aims to improve health equity through providing equal access to vaccines. I developed a methodology to quantify the impact of different vaccine coverage scenarios with respect to household income that take into account the distribution of other risk factors. I conclude in this chapter that merely ensuring equal access to vaccines will not reduce health outcome gaps across income quintiles because of the differences in the distribution of risks and the treatment provided.

Global commitment to the Sustainable Development Goals to improve maternal and newborn health is premised on and motivated by the understanding that universal health coverage of comprehensive, high quality maternal health services could be a path towards improved maternal and newborn health. This has propelled scientific inquiry towards understanding and improving the quality of facility-based maternal and newborn care globally. This dissertation assesses the trends in quality of maternal and newborn care across several low- and middle-income countries (LMICs), as well as the impact of poor quality of care on maternal health outcomes. This dissertation is guided by the World Health Organization (WHO) framework for quality of care for pregnant women and newborns and comprises of three distinct empirical studies on quality of facility-based maternal and newborn care. The first study uses a nationally representative sample of public health facilities at the district level across India to explore the public health system’s capacity for basic delivery and newborn services. The second study assesses socioeconomic disparities in access to high quality maternal care in Kenya by combining a nationally representative health system assessment with population-representative demographic data to describe and quantify the geographic distribution of poverty and maternal care quality. The third, and final, study examines the relationship between quality of care during childbirth and severe maternal outcomes using a large cross-sectional survey of over 300,000 births across 357 health facilities in 29 countries in Africa, Asia, Latin America and the Middle East. These studies use innovative approaches and underutilized data sources to extend empirical methods for studying quality of maternal and newborn care to ultimately monitor quality improvement and make progress towards meeting the global goals for reducing the burden of poor maternal and newborn health. The results underscore the need to generate context-specific evidence in LMICs to monitor and improve quality of care for mothers and newborns, and to target programs to reach health facilities with poor quality and health outcomes.

This dissertation explores questions related to financing and delivery of health services under the Sustainable Development Goals (SDGs) and is comprised of three standalone papers. In Paper 1, using the case of Armenia, I explore the health and financial impact of higher cigarette prices (achieved through higher excise taxes) using extended cost-effectiveness analysis (ECEA) methods. Specifically, I model the potential impact of tobacco tax hikes on measures of health, individual and government expenditures, and financial risk protection. In addition, I use qualitative methods to examine the agenda setting of tobacco taxation. The ECEA results suggest large health and health-related financial benefits of tobacco tax increases, with the health benefits concentrated among the poor. The findings from qualitative interviews with key stakeholders indicate that fiscal constraints faced by the government and influence of external pressure created a window of opportunity for tobacco taxes to be placed on the policy agenda. In Paper 2, I investigate the impact of measles supplementary immunization activities (SIAs) on utilization of selected maternal and child health (MCH) services in low- and middle-income countries (LMICs). This is the first quantitative multi-country study to examine this relationship. Based on the findings, measles SIAs do not appear to significantly influence the utilization of MCH services at health facilities, except for care seeking related to a child’s cough for which I do find a substantial reduction in utilization during the period of SIA implementation. In Paper 3, I examine the efficiency of primary health care facilities in four sub-Saharan African countries and the degree to which the incorporation of technical quality changes facility estimates of technical efficiency. This is one of the first studies to incorporate measures of technical quality in efficiency analysis in LMICs. While the country average efficiency scores using quality-adjusted output are similar to the unadjusted efficiency scores, I observe significant changes in individual facility efficiency scores. This suggests that quality is particularly important if the goal of efficiency analysis is to identify best and worst performers rather than to simply measure average efficiency among a set of facilities.

Delivery of health care services that meet evidence-based standards and are responsive to people’s health needs is core to achieving universal health coverage (UHC). Non-communicable diseases (NCDs) have now overtaken communicable diseases as the number one cause of mortality and morbidity in a majority of low- and middle-income countries (LMICs). Despite the availability of guidelines and growing access to clinics, large numbers of people in LMICs do not receive high-quality health care services. Measurement of health system quality is also problematic, as nationally representative data on quality of care and disparities in quality across population subgroups and social groups are limited, and quality itself is poorly understood.
My dissertation addresses health system quality in the treatment of chronic diseases. It is constituted of three papers that analyze each one element underlying quality: measurement, improvement, and equity. In paper one, I evaluate predictive validity of a measure of process quality; in paper two, I model a strategy for system improvements through integration of care; and in paper three, I analyze the distribution in good quality care to inform health system policies in middle-income countries. The first two study settings are located in sub-Saharan Africa, and the third study setting is China, a middle-income country that can provide useful lessons for health system improvement in low-income settings of sub-Saharan Africa. Specifically, the first study uses data from a cohort of HIV patients in Tanzania to develop and validate facility-level process measures of quality of HIV care. The second study uses data from a nationally representative survey of NCD risk factors (hypertension, diabetes mellitus, high cholesterol) in Uganda to model the impact and cost-effectiveness of integrating basic NCD care within HIV treatment delivery services to improve the health outcomes of people living with HIV receiving antiretroviral therapy. The third study uses data from nationally representative surveys of health facilities in China to examine the variation in quality of care provided to patients admitted to hospitals with acute myocardial infarction. Together, these three analyses contribute evidence towards formulating strategies to address the urgent problem of low-quality and inequitable care for chronic diseases in LMICs.

Understanding the behavioral responses and external economic consequences of health and development interventions is essential for creating successful, effective, and sustainable policy. My research utilizes both applied econometric techniques and behavioral economics theory to examine some of the psychological and external effects of specific health interventions and general health state utility elicitation. Utilizing a combination of primary and secondary data as well as both experimental and quasi- experimental methods, I evaluate external welfare consequences of early antiretroviral (ART) initiation on household food security and demonstrate new behavioral biases in health state utility assessment with direct application to health policy prioritization. The first chapter of this research employs a regression discontinuity design over longitudinal data from the Africa Health Research Institute in KwaZulu-Natal, South Africa to measure the causal impact of early ART initiation on household food security. The study examines the direct impact of ART on household food security, a fundamental development indicator, to demonstrate the implications of expanding treatment as prevention. This project finds that for a one to two year period following ART initiation, there is a significant increase in the probability of reported household food insecurity, which afterward diminishes to zero. This finding is consistent with the literature on economic recovery post-ART initiation and improves upon previous observational studies by utilizing a quasi-experimental structure to more precisely target valid counterfactuals for those initiating ART utilizing the 200 cells/μL CD4 rationing threshold. Results from this study can be used to help inform policy-makers of relevant potential external effects of scaling up ART initiation on other aspects of household socioeconomic welfare within a treatment-as-prevention paradigm. The second and third chapters deal with the theoretical formulation and empirical testing of two behavioral biases in health state utility elicitation utilizing a quality-adjusted life years (QALY) approach. The second chapter presents a qualitative study where subjects are asked to respond to standard gamble and time-tradeoff preference elicitation questions over two reduced health states, chronic severe depression and complete blindness. Using the post-survey cognitive interviews, and tools from salience theory in behavioral economics, I identify and formalize models for two types of biases: salient impact bias and salient treatment bias. The third chapter takes the qualitative results and theory formulated in chapter two and empirically tests the impact of utilizing a potential impact checklist on responses to standard gamble questions over 4,000 individuals in Dar es Salaam, Tanzania. This study utilizes an experimental structure and randomly assigns survey respondents to receive or not receive a prompt immediately before responding to the preference elicitation questions. Results show that for all participants prompting aligns responses from the standard gamble technique with those obtained in a pre-prompting time-tradeoff question. Additionally, subgroups demonstrated in the qualitative research to have different types of salient impact biases were all shown significantly to reduce cognitive biases. These results demonstrate that prompting for relevant impact dimensions can reduce the effect of cognitive biases on health state preference elicitation surveys and should be employed to more accurately assess quality weights for reduced health states applied to effectiveness measurement.

This thesis explores cultural, social and economic determinants of HIV risk across three settings in sub-Saharan Africa. I use mixed methods of inquiry to examine how these contextual features inform HIV transmission, testing, and prevention. My research is motivated by the growing appreciation that public health challenges require not only accurate and clear measurement, but a nuanced understanding of how health phenomena are perceived and how these perceptions shape individual behaviors and population health. To this end, the methodological tools and theoretical constructs to situate assessment within its particular social context are critical to an accurate diagnosis of the causes of important public health challenges. Paper 1 examines the extent to which conjugal relationship structure, in particular long-term unmarried partnering, is associated with HIV risk in KwaZulu-Natal South Africa. Using longitudinal panel data I estimate HIV incidence among, those who are not in conjugal relationships, those who are in long-term unmarried conjugal relationships, and those who are in married conjugal relationships. Although being married in a conjugal relationship is associated with a significant decrease in HIV risk, those in long-term, unmarried conjugal relationships experienced significantly increased HIV risk. Paper 2 explores how men in rural Malawi understand their access to and participation in HIV testing. Using qualitative analysis of data obtained from interviews, I contextualize the key factors men identify as important to how they rationalize non-testing and highlight the role that testing by partner proxy plays in how they weigh the pressures for and against testing. Paper 3 offers insight into the strategies that female commercial sex workers use to assess their own HIV risk, and how they apply this information in their participation in the commercial sex market in Kampala Uganda. Using qualitative analysis of data obtained through interviews, I propose a framework based on the strategies used by many sex workers to balance the risks and benefits associated with their work, in particular in navigating the condomless sex market. In each case, the study methodology and data selected complement existing knowledge to add theoretical insight. Individuals’ perception of their health risk, and their subjective assessment of the factors that they consider important, inform their navigation of the HIV risk environment.

Measuring health through surveys is challenging. Participants may respond in a socially favorable but untruthful way, and responses across respondents may be difficult to compare. List experiments and anchoring vignette techniques have been proposed to improve survey measures, and this dissertation applied qualitative and quantitative methods to evaluate either the usefulness or validity of those techniques. The first paper explored how list-based questions perform compared to direct questions when measuring two behaviors, smoking and intravenous infusion use. The difference-in-differences between the two survey methods and two behaviors was non-significant. List experiments might introduce downward biases rather than alleviate them due to cognitive difficulty in responding. The second paper applied cognitive interviewing to the design of vignettes among Chinese students with objectively different visual acuities. Ten major problems were identified regarding vignette comprehension, judgment, and responses. Respondents rated vignette character’s vision differently from their own, demonstrating a norm of being “strict with oneself and lenient towards others” in an Asian context.
The third paper assessed the validity of three vignette survey methods (i.e., indirect comparison between self and vignettes, direct comparison between self and vignettes, and self-assessment primed by vignettes) in measuring distance vision. Surprisingly, it was found that vignette methods were not improvements over self-assessment, and indirect comparison performed the worst.
These three studies shed light on a series of cognitive difficulties related to advanced survey techniques. Traditional self-assessment may be as useful as the list experiment and more valid than some anchoring vignette techniques in these studies’ context.

Non-communicable diseases (cardiovascular diseases, cancers, chronic respiratory diseases, diabetes, and mental illnesses) and associated risk factors (unhealthy diets, physical inactivity, harmful use of alcohol, physical inactivity) are on the rise in developing countries, posing a threat to the health and financial systems of emerging economies. In response, international organizations and Ministries of Health alike have started to tackle chronic diseases and associated risk factors with policies and treatment programs. Yet to this day, the body of evidence for best practices regarding the monitoring, prevention, and control of non-communicable diseases in low- and middle-income countries remains small. This doctoral thesis adds to this body of evidence. The first paper of my thesis assesses the impact of a national tobacco control program in high schools in Chile. Specifically, it evaluates the effectiveness and makes several policy recommendations based on the findings. My second dissertation paper assesses the modifying effect of a change in anti-retroviral treatment among HIV-positive subjects in KwaZulu-Natal, South Africa on cardiovascular disease risk factors of high body mass index and high blood pressure. The third paper is based on a randomized controlled trial assessing the effectiveness of a social-network-based diabetes and weight management program in Jordan.