Background: Long-term retention of patients in Africa's rapidly expanding antiretroviral therapy (ART) programs for HIV/AIDS is essential for these programs' success but has received relatively little attention. In this paper we present a systematic review of patient retention in ART programs in sub-Saharan Africa.

Methods and findings: We searched Medline, other literature databases, conference abstracts, publications archives, and the "gray literature" (project reports available online) between 2000 and 2007 for reports on the proportion of adult patients retained (i.e., remaining in care and on ART) after 6 mo or longer in sub-Saharan African, non-research ART programs, with and without donor support. Estimated retention rates at 6, 12, and 24 mo were calculated and plotted for each program. Retention was also estimated using Kaplan-Meier curves. In sensitivity analyses we considered best-case, worst-case, and midpoint scenarios for retention at 2 y; the best-case scenario assumed no further attrition beyond that reported, while the worst-case scenario assumed that attrition would continue in a linear fashion. We reviewed 32 publications reporting on 33 patient cohorts (74,192 patients, 13 countries). For all studies, the weighted average follow-up period reported was 9.9 mo, after which 77.5% of patients were retained. Loss to follow-up and death accounted for 56% and 40% of attrition, respectively. Weighted mean retention rates as reported were 79.1%, 75.0% and 61.6 % at 6, 12, and 24 mo, respectively. Of those reporting 24 mo of follow-up, the best program retained 85% of patients and the worst retained 46%. Attrition was higher in studies with shorter reporting periods, leading to monthly weighted mean attrition rates of 3.3%/mo, 1.9%/mo, and 1.6%/month for studies reporting to 6, 12, and 24 months, respectively, and suggesting that overall patient retention may be overestimated in the published reports. In sensitivity analyses, estimated retention rates ranged from 24% in the worse case to 77% in the best case at the end of 2 y, with a plausible midpoint scenario of 50%.

Conclusions: Since the inception of large-scale ART access early in this decade, ART programs in Africa have retained about 60% of their patients at the end of 2 y. Loss to follow-up is the major cause of attrition, followed by death. Better patient tracing procedures, better understanding of loss to follow-up, and earlier initiation of ART to reduce mortality are needed if retention is to be improved. Retention varies widely across programs, and programs that have achieved higher retention rates can serve as models for future improvements.