Longitudinal population-level HIV epidemiologic and genomic surveillance highlights growing gender disparity of HIV transmission in Uganda

Abstract

HIV incidence in eastern and southern Africa has historically been concentrated among girls and women aged 15-24 years, but as new cases decline with HIV interventions, population-level infection dynamics may shift by age and gender. Here, we integrated population-based surveillance and longitudinal deepsequence viral phylogenetics to assess how HIV incidence and the population groups driving transmission have evolved over a 15 year period from 2003 to 2018 in Uganda. HIV viral suppression increased more rapidly in women than men, resulting in 1.5-2 fold higher suppression rates in women with HIV by 2018 across age groups. Incidence declined more slowly in women than men, increasing pre-existing gender imbalance in HIV burden. Age-specific transmission flows shifted; the share of transmission to girls and women aged 15-24 years from older men declined by approximately one third, whereas the contribution of transmission to women aged 25-34 years from men aged 0-6 years older doubled from 2003 to 2018. We estimated closing the gender gap in viral suppression could have reduced HIV incidence in women by half in 2018 and ended gender disparities in incidence. This study suggests that male-targeted HIV programs to increase HIV suppression are critical to reduce incidence in women, close gender gaps in infection burden and improve men's health in Africa.

Age and gender profiles of HIV infection burden and viraemia: novel metrics for HIV epidemic control in African populations with high antiretroviral therapy coverage

“…The difference between the two populations can be explained by heterogeneities in transmission risk among individuals exhibiting viraemia. For example, phylogenetic studies in Uganda [21], Zambia [45], and Botswana [46] have estimated highest transmission risks in men aged 30-34 and women aged 20-24, and much lower for example among individuals aged more than 50. These results hence reinforce the importance of the subpopulations contributing most strongly to the burden of viraemia described in this paper.…”

“…Age and gender-specific population sizes in each survey round were estimated from line-list data of census-eligible individuals that were reported by household heads during the census. The reported ages tended to reflect grouping patterns around multiples of 5, which we disaggregated by smoothing population sizes over age (Supplementary Text S1.1 and Supplementary Figure S2[21]).…”

“…People who did not achieve viral suppression were defined as viraemic. From these data, we first estimated the population-level HIV seroprevalence (i.e., probability for an individual to be living with HIV) with a non-parametric Bayesian model over the age of individuals stratifying by gender, survey round and community type (i.e., fishing vs. inland), using the data from first-time participants as proxy for non-participating individuals (Supplementary Text, S1.2) [21]. Then, the same model was used to estimate the proportion of individuals exhibiting viraemia among people living with HIV (Supplementary TextS1.3) and among all individuals (Supplementary Text S1.4).…”