Countries with substantial Muslim populations are experiencing rapid changes in HIV prevalence. HIV testing and circumcision, as biomedical interventions, are the focus of this dissertation since biomedical strategies are the among the most efficacious HIV interventions. This dissertation examines the relationship of multilevel effects to HIV stigma, HIV risk behavior, and HIV status with two evidence-based HIV prevention intervention strategies (HIV testing and male circumcision) and a third HIV prevention intervention strategy (female circumcision) that is highly disputed, via three separate and distinct papers. This study is theoretically guided by the Ecological Perspective and the Social Network Conceptual Model. The sample for the first paper on Central Asia includes Kazakhstan (n=14,310), Kyrgyzstan (n=6,493), Uzbekistan (n=13,404), and Tajikistan (n=4,677), for a total n=38,884. The second paper sample is drawn from India: 65,356 men between the ages of 15 and 54. The third paper sample is drawn from Mali: 14,583; all of these participants are ever-married women of reproductive age (15-49 years old). Multilevel modeling was used in all three papers. This innovative methodology produced empirical evidence for the association of context with the behavior of the individual. A finding consistent in all three papers is that: context does matter. This dissertation examines context in terms of family and community membership. Specifically, the context of different levels of stigma and family/community membership impacts individuals' HIV testing and circumcision. In Central Asia, HIV stigma at the individual, family, and community levels is significantly associated with decreased HIV testing uptake and receipt of HIV test results. HIV stigma is associated with male circumcision status (i.e., whether a male is circumcised or uncircumcised) on individual, family, and community levels in India. In Mali, female circumcision was significantly associated with increased odds of HIV positive status, and circumcision status was not associated with HIV risk behavior. Family and community membership was also associated with HIV status and HIV risk behavior in Mali. The findings of the dissertation have important implications for practice, policy, and research.

Early Childhood Caries (ECC) is a highly prevalent disease afflicting approximately 28 percent of children in the U.S. under the age of 6 years (Bruce A Dye et al., 2007). ECC is a serious condition that can have profound health implications, including altered physical appearance, impaired ability to chew and speak, diminished quality of life, and increased risk for both oral and systemic health conditions (Colares and Feitosa, 2003; B. L. Edelstein, Vargas, and D, 2006; Norman Tinanoff and Reisine, 2009). Early identification of risk and prompt, targeted intervention is essential to overcoming the rising rates of ECC. The Diet and Early Childhood Caries (DECC) study was designed to evaluate a novel risk assessment tool, MySmileBuddy (MSB), in a predominantly Spanish speaking, low income, urban population. MSB serves as an interactive platform for education and goal setting for ECC prevention and a comprehensive ECC risk assessment tool that incorporates questions evaluating diet, feeding practices, general attitudes and beliefs, fluoride use, and family history. A large component of the MSB tool is devoted to the assessment of dietary risk factors related to ECC via inclusion of a modified 24-hour dietary recall. A primary aim of the DECC study was to establish concurrent criterion validity by evaluating if MSB diet and comprehensive scores were associated with physical evidence of risk (i.e., oral mutans levels, decalcifications, visible plaque, and ECC status). Additionally the DECC study aimed to examine associations between physical evidence of caries risk and overall frequency of oral exposures, length of exposure time, and body mass index-for-age (BMI/age). Lastly, the DECC study was designed to assess the preliminary impact of the MSB intervention on recollection of stated goals and progress toward achievement of targeted ECC-related behavior changes one month post-intervention. In 108 parent/child (caregiver/child) dyads, the MSB diet risk scores were found to be significantly associated with early stage indicators of caries risk, specifically oral mutans levels (p less than 0.05), and borderline associated (p less than 0.1) with visible plaque levels. The MSB comprehensive risk score was also found to be significantly associated with both oral mutans and visible plaque (p less than 0.05). Children with high MSB risk scores (diet and comprehensive) were more likely to have higher levels of oral mutans, and more likely to have higher levels of visible plaque compared to children with lower scores. Physical indicators of caries risk were not associated with other factors included in the DECC study (i.e., frequency of oral exposures and intake of individual food/beverage categories, length of oral exposure time, and BMI/age weight status). Preliminary data from the one-month follow-up suggests that the majority of parents/caregivers were able to recall their MSB goal and were beginning to initiate diet- and other dental-related changes at home. Overall, these findings suggest that the MSB tool may be a valid tool for predicting known physical precursors to caries and may be an effective avenue for behavior change. While these preliminary findings are encouraging, larger and longer-term studies will be necessary to determine the ultimate utility of MSB in predicting the ECC experience in children.

Positive practice environments promote the health, safety and well-being of health workers by supporting professionalism, acknowledging performance and ensuring opportunities for professional growth. According to the International Collaborating Partners of the Positive Practice Environments Campaign, positive practice environments foster a motivated, productive and high-performing pool of workers who deliver high quality care. Research shows that achieving positive practice environments for health care professionals is problematic worldwide. In Tanzania, the capacity of human resources for health managers to implement deliberate and strategic action to plan for and effectively deploy health workers, as well as safeguard their practice environments, is severely diminished. The basic strategic human resources management (SHRM) components model was used to guide understanding of the relationship between people management practices and the practice environment. This dissertation examined the people management practices that have been implemented in Tanzania and their impact on the practice environment experienced by associate clinicians (ACs), nurses and midwives who deliver emergency obstetric care (EmOC). Interview and survey data obtained from frontline ACs, nurses and midwives and members of Council Health Management Teams (CHMTs) tasked with planning for and deploying health human resources offered provider and managerial perspectives of the practice environment within which EmOC is delivered in Tanzania. The eight people management practices specified in the basic SHRM components model were applied in varying extents across the 48 districts assessed. Partial implementation contributed to loss of motivation and undermined the performance of ACs, nurses and midwives who deliver the essential interventions constituting EmOC in Tanzania.

The overall aim of the proposed dissertation is to enhance understanding of the impact of the community and neighborhood in preventing violence against women, and how women who have been displaced from their communities may be at increased risk of violence. This three-paper dissertation utilized secondary data sources from two studies of IPV against women in Uganda: the SASA! Study and the Ugandan Demographic and Health Study (UDHS). The first paper used quantitative data from the baseline of the SASA! study (a cluster randomized controlled trial of a community-based intervention to prevent violence against women and HIV/AIDS, called SASA!), a representative sample of community members in two districts in Kampala. This study hypothesized that women who live in neighborhoods with higher levels of collective efficacy to prevent IPV would be at decreased risk of experiencing male-perpetrated IPV. Using a multi-level logistics model, there was no significant neighborhood effect on intimate partner violence related to collective efficacy or otherwise. However, women with higher levels of self-efficacy to prevent IPV against others were significantly less likely to experience physical IPV themselves. Other fixed effect factors, including younger age, no education, higher number of children, having no electricity, not earning an income, and partner's daily alcohol use significantly predicted women's risk of IPV. Potential research and practice implications will be discussed. The second paper utilized secondary analyses of the impact of displacement on IPV against women from the Demographic and Health Survey, a representative community sample of women throughout Uganda. Using propensity score matching, this study attempts to determine the causal effect of displacement on women's experiences of intimate partner violence. Given that assumptions hold, the results indicate that women who are displaced in northern Uganda are less likely to experience IPV than if they had not been displaced. Potential explanations for these findings, such as the renegotiation of gender during displacement and the impact of the humanitarian Cluster Approach, will be discussed. The third paper is an in-depth qualitative study using secondary analysis of focus groups with community leaders in Kampala Uganda, also from the baseline of the SASA! study. Key findings using framework analysis of focus group discussions with religious leaders, sengas/traditional aunties, health care workers, police and local council leaders suggest a widely held justification for violence against women based on an underlying cultural belief in men's authority over women and expectations on women. The belief in men's power over women manifests in three, interrelated themes: men's authority, blaming women, and controlling women's sexuality. Few dissenting voices argued against violence against women for reasons related to the impact on the children and the need for women and men to live with peace and happiness in the home. Overall, despite numerous justifications for violence against women, community leaders expressed a strong sense of responsibility in responding to violence against women, particularly in life threatening situations. Suggested strategies for intervening in situations of violence against women in the home included recruiting elders, talking to the men about the violence, calling upon help from local council leaders, and reporting to the police. These suggested strategies were not, however, without underlying sentiments of men's authority and associated risks faced by community leaders. Community leaders also expressed a sense of responsibility in helping organize community members for prevention activities, although they did not see their role as leaders or facilitators of these efforts.

Hospital financial stress is a common problem. From 1995-2010 15-30% of hospitals every year were classified as financially distressed based on a negative total profit margin. However, very few studies to date have assessed the impact of hospital financial distress on quality of care indicators or patient outcomes. A comprehensive and systematic literature review was conducted to assess the state of the science on hospital financial distress and quality of care or patient outcomes. A total of three peer-reviewed articles were found that used a measure of hospital financial distress as an exposure with a quality of care indicator or health-related factor as an outcome. However, despite the limited research on this topic, the evidence suggested there was an important relationship between hospital financial distress and patient outcomes. What became clear after the systematic literature review was that not only had very few studies ever assessed the relationship between hospital financial distress and patient outcomes, but there was also limited evidence on the validity of various measures of hospital financial distress. An analysis conducted as part of this dissertation assessed the validity of profit margin, as well as two composite measures, the Financial Strength Index and the modified z-index, to see how well they predicted hospital closure. Overall, despite the added complexity of the Financial Strength Index and the modified z-index compared to profit margin, all three measures performed similarly with respect to predicting hospital closure. Finally, profit margin, as a measure of hospital financial distress was used to predict the receipt of immediate breast reconstruction. Women undergoing mastectomy at hospitals under high levels of hospital financial distress were significantly less likely to receive immediate breast reconstruction compared to women treated at hospitals under minimal to no financial distress, adjusting for important confounders. This means hospital financial distress is an important, although unwarranted, determinant of breast cancer care.

This dissertation aims to add to the growing literature on the risks and mechanisms in early life that may be associated with later bipolar disorder (BP), expanding the understanding of when and why divergences from typical developmental course occur in BP, if they do. To do so, it utilizes prospectively obtained, serologically documented prenatal biomarkers and clinically documented prenatal and perinatal risk factors, as well as premorbid measures of neurocognitive functioning, in a well-defined birth cohort followed up for BP. This offers a unique opportunity to test some of the evidence as to whether BP is a neurodevelopmental illness. The first paper is a systematic literature review of the neurodevelopmental hypothesis of BP. This review focuses on three developmental time points: prenatal and perinatal exposures, premorbid and prodromal symptom development, and neurocognitive functioning prior to onset. The second paper focuses on two specific putative prenatal and perinatal risk factors for BP: T. gondii and oxytocin to induce labor. The third paper assesses cognition, using both the BP case-control study and the full birth cohort to assess risks for BP and the potential that cognitive impairment reflects a mediator or endophenotype of later BP. Taken as a whole, the findings suggest support for the neurodevelopmental hypothesis of BP and indicate some potentially specific risks for BP.

This dissertation study examines the associations between neighborhood economic conditions and children's probability of dropping out of high school before completion and female teenagers' likelihood of giving birth before age 20. This dissertation study makes two major contributions to the current literature. First, by taking a longitudinal view of neighborhood socioeconomic situations, this research has demonstrated the advantage and importance of examining the impact of socioeconomic situations in which children are embedded during their entire childhood. Comparing the results from this study with those from using point-in-time measures of neighborhood conditions, I have found estimates of neighborhood effects using longitudinal measures are larger and more efficient. Second, unlike prior research that has focused on neighborhood poverty, this study examines three important dimensions of neighborhood economic conditions: poverty, affluence, and economic segregation by using the index of concentration at the extremes. Each of the dimensions has different impact on children's probabilities of quitting high school early and becoming teenage mothers. Neighborhood poverty is curvilinearly related to children's likelihood of dropping out of high school while neighborhood affluence and ICE have linear impact on children's educational attainment. For teenage childbirth outcome, effects of all three economic dimensions are linear. Substantial racial differences in response to neighborhood economic impact have been discovered. Results confirm the prior findings that white children are more responsive to neighborhood affluence. Holding constant individual and family characteristics, and influence from neighborhood racial composition, black children may fare better in their academic achievement than white children. This study fails to provide substantial support for relative deprivation and competition mechanisms of neighborhood economic influence. The neighborhood impact is mainly channeled through social isolation avenue. Family economic conditions and the educational attainment of family heads have strong impact on both of the children's outcomes. Residential mobility has negative impact on children's school performance but not on their health risk of teenage childbirth.

Black women in the United States suffer disproportionately from a number of chronic and acute diseases. Not only do black women suffer from these diseases, they also have poorer outcomes and higher levels of morbidity and mortality than others. Years of biomedical and social science research have identified various permutations of patient-level factors, including cultural mistrust and genetic predisposition, to explain the existence of this race and gender-specific reality. However, few studies have looked at physician-level influences on poor health results for black women. Through the use of face-to-face and internet-based instruments combining patient vignettes and closed and open-ended treatment questions, the Implicit Association Test (IAT) and a series of measures of explicit discriminatory beliefs, this study used the case of Systemic Lupus Erythematosus (SLE) to illustrate how and whether decisions about medical treatment and follow-up for black women patients differ from decisions made for white women patients in varying degrees according to: the patient´s race; the severity of symptoms; the physician's perception of the patient's personal characteristics (e.g., personality, mood); the physician's demographic characteristics; and the physician´s score on the discriminatory belief measures named above. The study has collected data from 94 rheumatologists. The study enabled the assessment of differences in treatment recommendations for women SLE patients presenting with symptoms of lupus nephritis (LN), a fairly common organ involvement for people with SLE. The only difference between the patients seen by the doctors through the study was their race, black or white, so the study asked whether treatment recommendations were significantly different for black patients as compared to white patients. The study also asked whether these treatment decisions could be predicted by scores on a series of measures of implicit (unconscious) and explicit racial bias. Overall, this study did not find evidence that physicians recommended less optimal treatment to black patients. Several possible reasons for the non-significant findings are discussed. The thesis recommends further study and intervention into the identification and treatment of early symptoms of disease among black women to reduce the incidence of avoidable morbidity and mortality in this population. These studies should as well take into account the possibility that over-compensatory behaviors may be exhibited by physician study participants who suspect that their racism or discriminatory beliefs may be revealed through their responses. New methods to obscure explicit and implicit discriminatory measures and to reduce the threat of racism for respondents should be explored.

Background Young Black men who have sex with men (MSM) continue to be disproportionately at risk for HIV and other sexually transmitted infections. Substance use - the use of alcohol and other mood-altering drugs - before and during sexual encounters has long been connected with the HIV epidemic, especially in young Black MSM. Substance use can decrease inhibitions and lead to poor decision-making skills, especially in younger men with less sexual experience. Internalized homophobia - a facet of minority stress - may be a particularly important factor that influences substance use before or during sexual behavior, as young MSM may use drugs before sex to escape the psychological distress induced by engaging in the social taboo of sex with other men. In addition, resilience factors may play a key role in blocking the relationship between internalized homophobia and substance use before or during sex. This dissertation aimed to examine relationships between internalized homophobia, psychological distress, and substance use before or during sexual behavior. This dissertation also aimed to construct a model of resilience, as well as determine whether resilience may act as a moderator in the relationship between internalized homophobia and substance use before or during sex. Methods Young Black men who have sex with men between the ages of 18 and 34 participated in two phases of this study: 1) a cross-sectional survey (n = 228) and 2) a longitudinal sex diary (n = 153). In the cross-sectional survey, participants answered demographic questions as well as questions about their substance use behavior in the 2 months prior to baseline; they completed the Connor-Davidson Resilience Scale (CD-RISC), the Mastery Scale, and the Social Support from Friends and Social Support from Parents scales. A subset of participants was then followed for 8 weeks. Every week, they reported on their sexual behaviors, substance use behaviors before or during a sexual encounter, and their psychological distress using the Kessler Psychological Distress Scale (K10). Logistic regression was used to analyze the cross-sectional data. Multilevel logistic regressions and multilevel generalized structural equation models were used to analyze the structured diary data. Results Men with higher levels of internalized homophobia were also more likely to have used alcohol before or during sex during the eight weeks of the study, but not other substances. Men who had higher levels of internalized homophobia also had higher levels of weekly psychological distress, but psychological distress was not related to drug use. Four factors - hardiness, mastery, peer support and maternal support - were related to resilience. This construct of resilience did not moderate the relationship between internalized homophobia and substance use before or during sexual intercourse. However, peer support alone did moderate the relationship between these two variables - men who had higher levels of peer support had a weaker relationship between internalized homophobia and alcohol use. Conclusions The findings of this dissertation suggest that stigma, in the form of internalized homophobia, may be an important structurally-related factor that influences alcohol use before or during sex in young Black MSM. The results also highlight the importance of considering protective resilience factors that may weaken this relationship; however, the way in which resilience works in this relationship may be complex. Although the idea of a composite construct of resilience was supported, this composite construct did not moderate the relationship between internalized homophobia and drug use. However, peer support alone did, lending credence to the idea of resilience as a complex construct whose separate indicators may moderate relationships differently. This research has valuable implications for designing HIV and substance use prevention interventions in young Black MSM.

This dissertation contains three essays on development economics in the areas of education and health in developing countries. The first chapter evaluates the impacts of girls' education support program on human capital development of 3,997 female students (9th ~ 11th grades) at 33 secondary schools located outside Lilongwe, Malawi. We find that female students treated with one-year tuition and monthly cash stipends are more likely to attend school and have better test scores. We also find that cognitive ability in the treatment group increases by 0.215 standard deviations, and those treated also display higher aspirations for educational achievement. Moreover, there is a significant improvement in time preference (increased patience). In the second chapter, written jointly with Cristian Pop-Eleches and Hyuncheol Bryant Kim, we address two questions: 1) How to promote demand for male circumcision and 2) What is the role of peer effects in demand for male circumcision. We randomly provided free male circumcision and transportation voucher to male students in 124 classrooms across 33 secondary schools near Lilongwe, Malawi. Using a two-step randomized design, we first assigned classrooms into three groups (100% Treatment, 50% Treatment, or No Treatment classrooms) and then also randomly selected half of male students in 50% Treatment classrooms for treatment. We find that our intervention substantially increased the demand for male circumcision by on average 14.2 percentage points (243%). We also find evidence of peer effects since untreated students in 50% Treatment classrooms were 3.8 percentage points (79%) more likely to get circumcised than students in No Treatment classrooms. Finally, we provide evidence of important reinforcement effects when close friends within the same classroom receive the intervention together. The third chapter, written jointly with Cristian Pop-Eleches and Hyuncheol Bryant Kim, explores complementarities of three HIV/AIDS prevention interventions: HIV/AIDS Education, Male Circumcision for boys, and Girls' Education Support aimed at keeping girls in school. The study is based in 33 secondary schools near Lilongwe, Malawi and we focus on the behaviors within the existing 124 classrooms in these schools. Our research design to study the complementarities of these interventions is based on the randomized allocation of the different mix of interventions across classrooms. Our preliminary results indicate limited evidence of complementarities among the three interventions.