Racial differences in virologic failure associated with adherence and quality of life on efavirenz-containing regimens for initial HIV therapy: results of ACTG A5095.

Abstract

Blacks had higher rates of virologic failure than whites on efavirenz-containing regimens in the AIDS Clinical Trials Group (ACTG) A5095 study; preliminary analyses also suggested an association with adherence. We rigorously examined associations over time among race, virologic failure, 4 self-reported adherence metrics, and quality of life (QOL). ACTG A5095 was a double-blind placebo-controlled study of treatment-naive HIV-positive patients randomized to zidovudine/lamivudine/abacavir versus zidovudine/lamivudine plus efavirenz versus zidovudine/lamivudine/abacavir plus efavirenz. Virologic failure was defined as confirmed HIV-1 RNA >or=200 copies/mL at >or=16 weeks on study. The zidovudine/lamivudine/abacavir arm was discontinued early because of virologic inferiority. We examined virologic failure differences for efavirenz-containing arms according to missing 0 (adherent) versus at least 1 dose (nonadherent) during the past 4 days, alternative self-reported adherence metrics, and QOL. Analyses used the Fisher exact, log rank tests, and Cox proportional hazards models. The study population included white (n = 299), black (n = 260), and Hispanic (n = 156) patients with >or=1 adherence evaluation. Virologic failure was associated with week 12 nonadherence during the past 4 days for blacks (53% nonadherent failed vs. 25% adherent; P < 0.001) but not for whites (20% nonadherent failed vs. 20% adherent; P = 0.91). After adjustment for baseline covariates and treatment, there was a significant interaction between race and week 12 adherence (P = 0.02). In time-dependent Cox models using self-reports over time to reflect recent adherence, there was a significantly higher failure risk for nonadherent subjects (hazard ratio [HR] = 2.07; P < 0.001). Significant race-adherence interactions were seen in additional models of adherence: missing at least 1 medication dose ever (P = 0.04), past month (P < 0.01), or past weekend (P = 0.05). Lower QOL was significantly associated with virologic failure (P < 0.001); there was no evidence of an interaction between QOL and race (P = 0.39) or adherence (P = 0.51) in predicting virologic failure. There was a greater effect of nonadherence on virologic failure in blacks given efavirenz-containing regimens than in whites. Self-reported adherence and QOL are independent predictors of virologic failure.