Background
Advanced HIV disease (AHD) is a risk for people living with HIV (PLHIV) experiencing viral non-suppression (VNS). We determined AHD prevalence at VNS and identified factors associated with incident AHD after VNS amongst PLHIV enrolled in The African Cohort Study (AFRICOS), a multi-country longitudinal study.
Methods
We analyzed data from patient visits conducted every six months (2013-2025) with CD4 count and viral load (VL). VNS was defined as VL > 1000 copies/mL. We calculated the cumulative prevalence of AHD (CD4 < 200 cells/mm³) at VNS events. Multivariable logistic regression with generalized estimating equations was used to describe factors associated with prevalent AHD, accounting for correlation among repeated events from the same patient. Cox proportional hazards models stratified by country and with robust standard errors clustered on patient were used to assess factors associated with incident AHD after VNS.
Results
Among 3438 patients contributing 36248 visits, 1636 (47.8%) had VNS at least once during the study period. PLHIV were median age 36.2 years, 57.9% female, and country distribution was 51.8% Kenya, 11.4% Nigeria, 19.1% Tanzania, and 12.7% Uganda. Across visits where VNS was detected, 29.5% had prevalent AHD. Factors associated with prevalent AHD at VNS included log VL (aOR = 1.87,95% CI 1.54 –2.28), sex (aOR for males vs females = 1.81, 95% CI 1.25 – 2.61), and antiretroviral (ARV) regimen (aOR for protease inhibitors (PI) vs integrase strand transfer inhibitors (INSTI) = 2.23, 95% CI 1.41 – 3.55). For the Cox model, 773 distinct episodes of VNS were identified among 608 patients. Age, log viral load, time from HIV diagnosis, and sex did not show significant effects on hazard of incident AHD, but ARV therapy with PI vs INSTI was associated with increased hazard (aHR = 2.89, 95% CI 1.42 – 5.90). In a secondary analysis considering CD4 count, each 50-cell higher CD4 was protective of incident AHD (aHR 0.31, 95% CI 0.19–0.53), but the protective effect attenuated with time (time-varying aHR 1.16 per unit increase in log-time, p<0.001).
Conclusions
Nearly half of patients in the cohort experienced VNS. AHD was prevalent at almost a third of visits with VNS. High viral load, PI-based regimens, and male sex were associated with prevalent AHD, while lower CD4 counts and PI-based regimen was associated with progression to AHD after VNS. This work highlights the importance of VNS prevention and factors to consider regarding CD4 testing strategy among PLHIV with VNS.