Work

Longitudinal patterns of antiretroviral treatment interruptions in a high burden, low-resource setting in South Africa: a retrospective cohort study

Year: 2025

Type: article

Abstract: Background Non-retention undermines the potential of antiretroviral therapy (ART) to improve individual and public health HIV outcomes in South Africa, where routine ART care is largely provided through primary care clinics or differentiated service delivery models (such as adherence clubs). We aimed to describe long-term patterns of treatment interruption and re-engagement in a large routine cohort. Methods In this retrospective cohort study, we used demographic, mortality, dispensing, and laboratory data from the Provincial Health Data Centre, which includes data from all facilities in the Western Cape province, South Africa (52 hospitals and 354 primary care clinics). We included adults (age 15–85 years) who sought care in Khayelitsha or Gugulethu (which are low-resource, high HIV burden settings), and initiated ART under universal test-and-treat guidelines, implemented from Sept 1, 2016, but before Sept 30, 2021 (to allow >1 year of potential follow-up to data cutoff on Sept 30, 2022). The main outcome was time to first treatment interruption (>90 days late for an expected visit, based on treatment refill data) or censoring (death or data cutoff) after ART initiation and survival analysis explored patterns of recurrent interruption and re-engagement. Cox proportional hazards regression evaluated associations of treatment interruption with demographics (age, sex, area of residence, and access to care), clinical characteristics (comorbidities, pregnancy, hospitalisation, tuberculosis, or COVID-19), and ART history (year of initiation, CD4 cell count, and regimen at initiation). Findings 68 888 people were included in the cohort, of whom 47 631 (69·1%) were female, 21 189 (30·8%) were male, and 17 078 (24·8%) were aged 25 years or younger, with a median follow-up of 4 years (IQR 3–5). By 6 years of follow-up after ART initiation, cumulative incidence of treatment interruption was 71·8% (95% CI 71·4–72·3), with a median time to interruption of 4·9 months (IQR 0·9–13·0) from initiation. The cumulative incidence of re-engagement after ART interruption at 6 years was 75·6% (95% CI 74·8–76·4), after a median of 7·9 months (IQR 4·4–17·1) after first treatment interruption. Of the 40 384 interruption events with a subsequent return to care, 29 967 (74·2%) events had sufficient follow-up to evaluate outcomes a year after restart, among which 15 062 (50·3%) were virally suppressed (≤1000 copies per mL). The magnitude of association between treatment interruptions and demographic or clinical characteristics was small. Interpretation Treatment interruptions were common and an expected part of long-term ART use, particularly soon after ART initiation and restart. However, most out of care were individuals at low risk, suggesting the need for more nuanced ways to identify who requires support. Most people returned to care, but ultimately had poor viral suppression, revealing gaps in the way primary care is organised to support sustained viral suppression despite frequent cycling in and out of care. Funding Clarendon Fund and St John's College.

Source: The Lancet Primary Care

DOI: 10.1016/j.lanprc.2025.100077