Worldwide HIV Virulence Evolution in Response to Changes in Prevalence and Treatment Coverage

Sarah Stansfield | 2019

Advisor: Steven M. Goodreau

Research Area(s): Global Health, Infectious Diseases, Social Determinants of Health


Introduction Whether worldwide HIV virulence has been increasing, decreasing, or remaining constant through time is still debated. Modeling work has suggested that prevalence and treatment coverage within countries may impact HIV virulence evolution at the population level, but these factors have not yet been considered in data analyses of HIV virulence changes. Additionally, disparities in HIV burden, including in prevalence and treatment coverage, exist between black and white men who have sex with men (MSM) in the US and worldwide. If differences in prevalence and treatment coverage impact mean population HIV virulence, this impact could be seen in disparities in virulence levels between these groups. Methods I utilized the NESCENT-CASCADE dataset, which combines data from 32 HIV cohorts representing individuals from 184 countries around the world. This contained individual-level HIV virulence marker data as well as other individual-level covariates. Prevalence data came from the Institute of Health Metrics and Evaluation and treatment coverage data from the World Bank. Because different treatment initiation guidelines would likely affect virulence evolution differently, these analyses only consider the time period in which treatment was based on CD4 guidelines. I utilized a multilevel modelling approach to allow for random effects at the country level and also utilized univariate and multivariable linear regression analyses to examine differences in HIV virulence with prevalence and treatment coverage differences in heterosexuals. As prevalence and treatment coverage were assessed on the country level, they are more likely to be representative of rates in the heterosexual population than the MSM population, given the much larger size of the former. Therefore, we only included heterosexuals in these analyses. I utilized linear regression and a Welch two-sample t-test to examine differences in virulence between black and white MSM in the US and Europe. Results The proportion of variance attributable to the country of origin was so small that it indicated that this effect was not meaningful. Higher prevalence was significantly associated with higher virulence in both univariate and multivariable analyses (p<0.001). Higher CD4-based treatment coverage was also significantly associated with lower virulence level in the multivariable analysis (p=0.001). No differences in virulence marker level were found between black and white MSM (p=0.556) but slight differences in virulence change though time were seen. Conclusion Consistent with previous modeling findings, a comparative analysis of 32 HIV cohorts finds that prevalence and treatment coverage both impact HIV virulence at the population level in heterosexuals. These factors should be considered when examining virulence levels through time. More data should be analyzed to determine if this effect extends to other groups, such as MSM.