The first six months after antiretroviral therapy (ART) initiation is recognized as a high-risk period for interruptions in care that may compromise viral suppression. Recent evidence from South Africa suggested that more than 40% of patients interrupt care for greater than 28 days during this early treatment period. The quantitative relationship between early treatment interruptions and viral suppression at 6 and 12 months had not been fully characterized prior to this analysis.
Researchers conducted a retrospective cohort study using South Africa's national ART database, TIER.Net, drawing data from 24 public sector facilities across four provinces. Adults aged 18 years or older who initiated ART between 1 January 2018 and 7 November 2024 and who had at least 14 months of follow-up were included.
Engagement in care was classified separately for months 0–6 and months 7–12. Three categories were defined using completed and scheduled visit dates:
Viral suppression (VS) was defined as HIV RNA <50 copies/mL. The analysis adjusted for demographic and clinical covariates including age, sex, ART initiation year, ART regimen, engagement pattern, and baseline CD4 count.
Modified Poisson regression models were used to estimate adjusted risk ratios (aRRs) for achieving VS at 6 and 12 months. The models controlled for the prespecified covariates listed above. Confidence intervals reported in the source reflect precision around the aRR estimates.
The cohort comprised 57,553 participants. Two-thirds (66%) of the cohort were female and the median age was 33 years. Engagement patterns varied over time: at 6 months, 49% of participants were continuously engaged and 22% were cyclically engaged; at 12 months, continuous engagement declined to 42% while cyclical engagement declined to 17%.
At 6 months, 54% of participants who were continuously engaged had achieved VS (<50 copies/mL) compared with 34% of participants who were cyclically engaged. The adjusted risk ratio for VS at 6 months comparing continuous to cyclical engagement was 1.60 (95% CI 1.55–1.64).
At 12 months, 56% of continuously engaged participants were suppressed versus 40% of those with cyclical engagement. The adjusted risk ratio at 12 months for continuous versus cyclical engagement was 1.38 (95% CI 1.34–1.42).
In addition to engagement patterns, the analysis found associations between VS and several other factors reported in the source. Viral suppression was more likely among participants on dolutegravir-based regimens, those who initiated ART in later years within the study period, participants with baseline CD4 counts greater than 200 cells/µL, female participants, and older individuals. The preprint reports these associations but does not provide additional mechanistic explanations beyond the observed relationships.
The study concludes that even relatively brief interruptions in ART during the first year of treatment are associated with substantially lower rates of viral suppression at both 6 and 12 months. Given the high prevalence of early interruptions observed in prior South African research and the differences in suppression across engagement categories in this large cohort, preventing early interruptions is identified as a programmatic priority to improve treatment outcomes.
The authors caution that this work is a preprint and has not been peer reviewed; the source explicitly notes it should not be used to guide clinical practice without additional validation.
Ethical approvals were obtained from the Boston University Institutional Review Board (H-38115) and the Human Research Ethics Committee of the University of the Witwatersrand (M190445). The authors declared no competing interests.
Data used in the study are owned by the South Africa National Department of Health and cannot be shared directly by the authors; requests for data must be made through the National Health Research Database (https://nhrd.health.gov.za/). Funding for the work is reported from the Gates Foundation (INV-031690). Supplementary materials and links to data and code are provided in the preprint record.
Limitations explicitly reported in the source text beyond that this is a preprint were not detailed in the abstract; additional methodological limitations, potential biases, or sensitivity analyses (if present) are described in the full preprint and supplementary materials linked in the source record.