Abstract Body

Background

High-risk populations bear a disproportionate burden of HIV in sub-Saharan Africa (SSA), yet data on advanced HIV disease (AHD) in these groups are scarce. We assessed the population-level prevalences and correlates of AHD among men who have sex with men (MSM), people who inject drugs (PWID) and women engaged in commercial sex (WCS) in Malawi.

Methods

From Sep-Dec 2024, we conducted respondent-driven sampling surveys among MSM and PWID in four and three major cities, respectively, and among WCS in 10 districts. With a 6-month recall period for risk behavior, eligibility criteria included ages 18+ years and: biological male, having anal or oral sex with a man (MSM); person injecting drugs for non-medical purposes (PWID); biological female, exchanging sex for money, goods or services (WCS). Eligible, consenting participants completed an in-person interview and HIV rapid test on site. HIV viral load (VL) and CD4 testing were done among HIV-positive samples. AHD was defined as <200 CD4 cells/mm3. Treatment interruption (TxInt) was defined as ever stopping HIV treatment for 2+ weeks. Unsuppressed VL (uVL) was defined as ≥1000 copies/mL. Binomial regression models adjusted for demographics and covariates (uVL, TxInt, treatment duration) and accounted for unequal selection probabilities. We used ‘surveyglm’ in R to obtain overall weighted estimates and relative risks (RR) of AHD using 95% confidence intervals.

Results

Among 1716 MSM, 523 PWID, and 4083 WCS participants, HIV prevalence was 8.8% (MSM n=163), 11.8% (PWID n=61), and 45.3% (WCS n=1862). Median CD4 count with interquartile range (IQR) was lowest among MSM [416, IQR 254-556], followed by PWID [523, IQR 402-667] and WCS [645, IQR 468-852]. AHD prevalence was 18.7%, 6.9% and 4.3%, among MSM, PWID and WCS, respectively. In adjusted models among MSM, AHD risk was greater among those with uVL versus (v) suppressed VL [32.7% v 15.6%; RR 16.4 (2.3-117.6); p=0.006] and those who experienced TxInt v those who did not [48.0% v 2.7%; RR 19.2 (3.3-111.1); p=0.001]. Among WCS, AHD risk was greater among those with uVL v suppressed VL [18.2% v 3.0%; RR 6.2 (2.4-16.1); p<0.001]. A small PWID sample limited model reliability.

Conclusions

This first population measure of AHD among high-risk populations in SSA and Malawi shows high AHD prevalence among MSM and PWID, with TxInt and uVL as potential key factors among MSM and WCS. Monitoring to identify and prevent TxInt and continued access to VL monitoring remains critical for these populations.