Objective: to determine adherence of an indigent African HIV-infected cohort initiating antiretroviral therapy (ART); to identify predictors of incomplete adherence (< 95%) and virologic failure (> 400 HIV RNA copies/ml). Design: Prospective monitoring of adherence in a poor HIV-positive cohort, attending a public sector hospital and receiving ART through phase III studies. Methods: Adherence to ART was determined over 48 weeks by counting tablet-returns. Logistic regression models including age, WHO HIV stage, home language, socio-economic status, complexity and type of regimen were fitted to determine predictors of incomplete adherence and virologic failure at 48 weeks. Results: 289 patients were recruited between January 1996 and May 2001. Median (mean) adherence of the cohort was 93.5% (87.2%). Three times daily dosing [risk ratio (RR), 3.07; 95% confidence interval (CI), 1.40-6.74], speaking English (RR, 0.41; 95% CI, 0.21-0.80) and age (RR, 0.97; 95% CI, 0.94-0.99) were independent predictors of incomplete adherence. Socio-economic status, sex and HIV stage did not predict adherence. Independent predictors of virologic failure included baseline viral load (RR, 2.57; 95% CI, 1.57-4.22) and three times daily dosing (RR, 2.64; 95% CI, 1.23-5.66), incomplete adherence (RR, 1.92; 95% CI, 1.10-3.57), age (RR, 0.96; 95% CI, 0.92-0.99) and dual nucleoside therapy (RR, 2.69; 95% CI, 1.1 7-6.15). Conclusion: The proportion of individuals achieving viral suppression matched results from the developing world. Speaking the same language as site staff and simplified dosing frequency were beneficial. Socio-economic status had no impact on adherence and should not be used as a limitation to ART access. © 2003 Lippincott Williams & Wilkins.
