| Issue/Objective |
UNAIDS 95-95-95 is often operationalized as if expanding ART coverage will yield durable viral suppression. We tested whether failure to approach 95% suppression is driven primarily by rebound (loss of suppression) rather than insufficient re-suppression. |
| Methodology/Approach |
We analyzed longitudinal HIV surveillance linked to clinical viral load
(VL) testing from the Africa Health Research Institute (AHRI) population cohort
in uMkhanyakude District, KwaZulu-Natal, South Africa. Suppression was defined as VL ≤1,550 copies/mL using the last VL in each calendar year. We estimated annual suppression prevalence among VL-observed person-years (2011-2023; excluding 2020) and year-to-year transitions among individuals with VL observed in consecutive years: rebound and re-suppression. We fit a four-regime, two-state Markov model (pre-UTT ≤2014; UTT pulse 2015→2016; post-UTT 2016-2018; post-COVID 2021-2022) and projected suppression from 2023 to 2030 under policy levers that increase re-suppression ("coverage") and/or reduce rebound ("retention"). Uncertainty used 95% Wilson intervals and bootstrap projection intervals (n=800) |
| Results |
Among 13,339 people living with HIV, we analyzed 33,032 VL-observed person-years and 10,094 consecutive VL year-pairs. Viral suppression prevalence increased from 0.300 (2011) to 0.701 at UTT rollout (2016) and was 0.659 in 2023. Rebound dropped transiently during the UTT pulse (2015→2016) but remained high post-COVID (2021-2022; 245 per 1,000 suppressed), with suppression inflow largely offset by outflow. Under post-COVID baseline dynamics, coverage-only improvements could not reach 95% by 2030 even at maximum re-suppression. Achieving 95% required an 89.7% rebound reduction (95% CI 87.8-91.4) or an 80.8% joint improvement in both rebound and re-suppression (95% CI 78.0-83.1) |
| Discussion/Conclusion |
These findings reframe the third 95 as a retention target. Coverage-only scale-up cannot achieve 95% suppression; reducing rebound from suppression is the binding lever. Programs should prioritize interruption prevention and rapid restart and re-engagement pathways, and implement flow-based cascade monitoring that tracks suppression inflow and outflow simultaneously. |
| Presenters and Affiliations |
Gashaija Absolomon Centre for Impact, Innovation and Capacity building for Health Information Systems and Nutrition (CIIC-HIN), Kigali, Rwanda |
