Adolescent girls and young women (AGYW; aged 15-24 years) in sub-Saharan Africa, the epicenter of the global HIV epidemic, have carried the primary burden of new HIV infections in this area for almost 40 years. Research has prioritized characterizing the individual predictors of HIV infection among AGYW by creating risk assessment tools that identify high-risk sub-populations for targeted HIV prevention efforts. Despite substantial efforts, there remains a disproportionate disease burden among this vulnerable population, suggesting a need to identify and assess new intervention targets beyond the individual. The objective of this dissertation is to expand our understanding of the complex relationship between the multi-level drivers of HIV infection among AGYW using advanced data science and epidemiologic methods.
This dissertation is divided into six chapters, the first of which is an introduction to the dissertation. The second chapter is a scoping review of the extant HIV-related literature that has leveraged data integration methods to combine heterogeneous, multi-level data sources. Chapters 3, 4, and 5 are empirical aims. Chapter 3 describes the development of an integrated dataset that combines information from the Population-based HIV Impact Assessment (PHIA) project, the Population and Housing Census, and the Joint United Nations Programme on HIV/AIDS (UNAIDS) Policy Indicators platform. The resulting dataset captures data at the individual, interpersonal, community, and societal levels across five sub-Saharan African countries: Cameroon, Eswatini, Malawi, Rwanda, and Uganda. Chapter 4 uses the dataset described in Chapter 3 and presents the application of causal discovery algorithms to characterize and graphically depict the pathways among individual, interpersonal, community, and societal risk factors of HIV infection among AGYW to identify the potential underlying causal mechanisms supported by the data. Chapter 5 uses the results from Chapter 4 to assess the impact of increasing the proportion of AGYW who completed secondary education on HIV prevalence using parametric g-formula. This dissertation ends with Chapter 6, which summarizes the dissertation's results and situates the findings within the broader HIV prevention literature.
A brief description of the dissertation results follows. The scoping review describes the four types of data integration methods: record linkage, multiple frame methods, imputation-based methods, and modeling techniques. I identified five thematic uses of data integration in the literature that supported the included articles’ study objectives. Those themes included using data integration to 1) describe HIVrelated etiology and prognosis; 2) develop or operationalize an HIV-related databases; 3) characterize sociodemographic, behavioral, clinical, and care risk factors; 4) estimate the population size of key or hard-to-reach populations; and 5) estimate HIV prevalence for key populations or varying geographical units. Then, using one of the described integration techniques, multiple frame methods, I present the process of developing a multi-level and -country integrated dataset that combined data from the PHIA Project, the Population and Housing Census, and the UNAIDS Policy Indicators platform. Additionally, I described the population of AGYW included in this study, as well as the different interpersonal, community, and societal environments they reside in, across Cameroon, Eswatini, Malawi, Rwanda, and Uganda.
I then applied the PC causal discovery algorithm to that dataset to elucidate the interconnectedness between individual, interpersonal, community, and societal level risk factors on HIV status among AGYW across each of the countries and overall. Community-level HIV prevalence and interpersonal sexual relationship factors consistently had direct paths to AGYW's HIV status for almost all country analyses. Additionally, there were multiple individual-level factors that had direct paths to AGYW's HIV status, and most of those variables were related to sexual behavior (e.g., number of sexual partners in the last 12 months, age of sexual debut). Additionally, there were multiple indirect paths to HIV status identified across all levels of organization. My last empirical study used the findings from Malawi and applied the parametric g-formula, to assess the impact of three hypothetical scenarios that model how increasing the proportion of AGYW who completed secondary education impacts HIV prevalence. I found that increasing the proportion of AGYW who completed secondary education from about 31% to 100% is associated with about a 26% decreased odds of HIV. The findings highlight the importance of improving educational attainment among AGYW, which will impact their life trajectory, economic prosperity, and overall autonomy.
The findings from this dissertation improve the knowledge base informing prevention interventions, thereby advancing the development of interventions that go beyond the individual to reduce the burden of HIV among AGYW. Additionally, the methods used in this dissertation provide an illustrative example of a novel and intersectional approach to assessing the multi-level determinants of health that may expand the current epidemiologic research program.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/13r2-xt03 |
| Date | January 2024 |
| Creators | Reed, Domonique Montier |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0025 seconds