Antiretroviral treatment programs in sub-Saharan Africa have high rates of early mortality and loss to follow-up. Switching to second-line regimens is often delayed because of limited access to laboratory monitoring.
Retrospective analysis was performed of a cohort of adults who initiated a standard first-line antiretroviral treatment at 5 public sector sites in 3 African countries. Monitoring included routine CD4 cell counts, human immunodeficiency virus RNA measures, and records of whether appointments were kept. Incidence and predictors of death, loss to follow-up, and switch to second-line regimens were analyzed by time-to-event approaches.
A total of 3749 patients were analyzed; at baseline, 37.1% were classified as having World Health Organization disease stage 3 or 4, and the median CD4 cell count was 192 cells/µL. First-line regimens were nevirapine based in 96.5% of patients; 17.7% of patients attended <95% of their drug pickup appointments. During 4545 person-years of follow-up, mortality was 8.6 deaths per 100 person-years and was predicted by lower baseline CD4 cell count, lower hemoglobin level, and lower body mass index (calculated as weight in kilograms divided by the square of height in meters); more-advanced clinical stage of infection; male sex; and more missed drug pickup appointments. Dropouts (which accrued at a rate of 2.1 dropouts per 100 person-years) were predicted by a lower body mass index, more missed visits and missed drug pickup appointments, and later calendar year. Incidence of switches to second-line regimens was 4.9 per 100 person-years; increased hazards were observed with lower CD4 cell count and earlier calendar year at baseline. In patients who switched, virological failure was predicted by combined clinical and CD4 criteria with 74% sensitivity and 30% specificity.
In an antiretroviral treatment program employing comprehensive monitoring, the probability of switching to second-line therapy was limited. Regular pickup of medication was a predictor of survival and was also strongly predictive of patient retention.