In sub-Saharan Africa (SSA), the region with the world’s highest HIV prevalence, antiretroviral therapy (ART) has recently reduced AIDS-related deaths. However, death rates still remain unacceptably high, as do the numbers of new HIV infections. HIV treatment is also often provided by stand-alone HIV clinics, distinct from other healthcare services. As the people living with HIV (PLWH) start to survive into older age, adequately addressing emerging challenges, including but not limited to age-related diseases and disabilities, will require more integration for HIV- and non-HIV services. Whereas ART, especially early ART for all HIV-infected patients, is the single most important intervention against HIV-related mortality and morbidity, the persistence of some complications (e.g., HIV-associated neurocognitive disorder [HAND]) and high rates of residual mortality in the ART era suggest that additional interventions are needed.
In the first chapter of my dissertation, I present a comparison of the cognitive scores of PLWH in a rural community in South Africa and HIV-negative comparators in the same community. Although adverse cognitive outcomes have been reported in clinical studies PLWH, no prior population-based studies in SSA have compared cognitive functioning among PLWH and HIV-negative comparators. The baseline data of participants in the “Health and Aging in Africa: A Longitudinal Study of an INDEPTH Community in South Africa” (HAALSI), a population-based cohort in rural Agincourt, Mpumalanga, northeast South Africa, were analyzed. Participants, who are men and women aged at least 40 years. Cognitive scores on a conventional instrument assessing orientation, immediate and delayed recall, and numeracy for 4,560 participants, and a novel instrument (the Oxford Cognitive Screen [OCS]-Plus), assessing memory, language, visual-spatial ability, and executive functioning for 1,997 participants were used to measure cognitive functioning. We used linear regression to compare composite cognitive scores between PLWH and HIV-negative participants, and among PLWH, anti-retroviral therapy (ART) users versus non-users. Multilevel (random-intercepts) linear models assessed domain-specific associations on the OCS-Plus. Estimates were adjusted for age, gender, education, country of birth, father’s occupation, household asset index, and ever-consumed alcohol. Of 4,560 participants, 1,048 were HIV positive (719 used ART). PLWH averaged 0.06 (95% CI: 0.01 to 0.12) standard deviation units higher scores on the conventional cognitive battery than HIV-negative participants. However, PLWH averaged 0.03 (95% CI, -0.08 to 0.03) standard deviation units lower scores on the OCS-plus instrument, the result not reaching statistical significance. Among PLWH, ART use did not predict cognition. There were no domain-specific cognitive effects from HIV or ART using the OCS-Plus measures. PLWH thus had higher cognitive function scores than HIV-negative comparators on a conventional cognitive assessment but not on a novel measure designed for low-literacy settings.
In the second chapter, I present an assessment of the association between HIV status and ART and age-related disability in the HAALSI baseline sample. We specifically evaluated whether these associations depend on body mass index (BMI). Although antiretroviral therapy (ART) use could mitigate risk of age-related disability among people living with HIV (PLWH), ART often causes weight-gain and could counter-intuitively increase risk of disability through elevated BMI. The baseline data of 4552 individuals (1040 with HIV and 3512 without HIV) were analyzed. Primary predictors were HIV status and ART use. The outcome was disability in at least one of five basic ADLs (walking across the room, getting up from bed, dressing, bathing and using the toilet). BMI, calculated from measured weights and heights, was considered in categories of underweight (<18.5kg/m2), normal BMI (18.5 to <25), overweight (25 to <30) and obese (≥30). We estimated prevalence differences in the outcome comparing PLWH to participants without HIV, and among PLWH, ART users to non-users. Additional models in PLWH compared virally suppressed ART users vs. unsuppressed ART users vs. ART non-users. We assessed BMI effects using its interaction with HIV and ART use. All regression models were adjusted for age and sex; more comprehensively adjusted models added education, father’s occupation, country of origin, and alcohol use. Among the 4552 study participants, 11.9 % reported at least one ADL disability. PLWH had lower prevalence of obesity (21% among PLWH vs 30% in participants without HIV) but higher prevalence of underweight (7.4% among PLWH vs 4.6% among participants without HIV). Among PLWH, those who were underweight had 13.3 percentage points (95% CI: 3.8 to 22.8) higher prevalence of an ADL disability than those with normal BMI. Those with obesity had 2.9 percentage points (95% CI: -2.3 to 8.1) higher prevalence of an ADL disability than those with normal BMI, the result not reaching statistical significance. Among the 69% of PLWH who were ART users, those with underweight had 13.7 percentage points (95% CI: 2.8 to 24.4) higher prevalence of ADL disability than normal-weight ART users. Those with obesity had 1.2 percentage points higher prevalence of an ADL disability (95% CI: -4.3 to 6.9) higher prevalence of an ADL than those with normal BMI, the result not statistically significant. Underweight participants thus had increased prevalence of ADL disability, but overweight and obese individuals did not. We find no evidence that weight increases associated with ART use are likely to increase disability.
In the third and final chapter, I assess the sufficiency of ART in eliminating excess mortality from Kaposi’s sarcoma (KS), an HIV-associated cancer, among HIV-infected adults initiating ART in Uganda. Despite KS being among the most common adult malignancies in the region following the onset of the HIV epidemic, approaches to its therapeutic management have largely been extrapolated from high-resource settings such as the US. In particular, antiretroviral therapy (ART) is often used alone for the initial management of persons with KS who do not have immediately life-threatening complications without any direct evidence in African patients of the effectiveness of this strategy. Among HIV-infected participants in Uganda who were initiated on ART, we compared those with biopsy-confirmed KS to those without KS. We used a directed acyclic graph (DAG) to identify relevant confounders, which were measured identically in both groups, in order to test the hypothesis that the participants with KS have excess mortality. Survival was determined over 4 years with dedicated attention to decrease loss to follow-up by actively tracking lost participants in the community. We evaluated 224 participants with KS and 683 participants without KS who were initiated on ART between 2005 and 2013. Males were 37%, median values were; age 34 (IQR 28 to 40), CD4+ T-cell count 158 cells/mm3 (IQR 76 to 263), and plasma HIV RNA level 5.2 log10copies/ml (IQR 4.6 to 5.6). In the unadjusted analysis, mortality at 1 year was 18.8% in those with KS and 3.9% in those without KS; at 4 years, mortality was 30.4 % and 8.2% respectively. After adjustment using proportional hazards regression for age, sex, asset holding, history of tuberculosis, history of cryptosporidial diarrhea, history of esophageal candidiasis, physical health summary score, body mass index, hemoglobin, CD4+ T cell count, plasma HIV RNA level, and calendar date of ART initiation, participants with KS had a 5.0-fold (95% CI 2.5-10.0; p <0.001) higher rate of death in the first year after start of ART than those without KS and a 2.9-fold (95% CI 1.2-7.0, p=0.020) higher rate of death thereafter. In a prototypical patient, the absolute difference in risk of death among those with KS +20% at 4 years. Even in this population of patients with KS who did not have immediately life-threatening complications, use of ART alone did not eliminate excess mortality from KS.
The findings of this work do suggest the need for additional interventions over and above ART to address HIV-related morbidity and mortality in SSA. We did not find evidence that PLWH are likely to have increased burden of cognitive impairment or disability. However, future studies should explore cognitive and disability trajectories in longitudinal data to further evaluate associations with HIV and ART use, especially early ART, to guide intervention.