Background:  Efavirenz (EFV) and nevirapine (NVP) are 2 non-nucleoside reverse transcriptase inhibitors (NNRTI) that are increasingly used as first-line agents in ART roll-out programs in resource-limited settings. Prior research from clinical trials has indicated that EFV- and NVP-based ART have comparable efficacy. However, the relative effectiveness of these agents has not been evaluated in a large-scale field program setting.

Methods:  We studied 2821 HIV-infected South African adults started on EFV- or NVP-based ART through a private-sector HIV/AIDS disease management program. The primary endpoint was sustained virologic suppression (<400 copies/mL) throughout follow-up. Secondary endpoints included time to initial viral load suppression and time to subsequent virologic failure (>400 copies/mL). Cox proportional hazards regression and the log-rank test were used to model the effects of baseline variables and medication adherence (based on pharmacy claims), on time to viral suppression or failure.

Results: Of the total, 1822 (64.6%) patients were on EFV-based and 999 (35.4%) on NVP-based regimens. The mean age (SD) at HAART initiation was 37.0 (7.7) years; 1775 patients (62.9%) were female, and 2734 (96.9%) were black Africans. The median (IQR) follow-up period was 2.2 (1.7 to 2.7) years. Patients on EFV were more likely to achieve 100% adherence than those on NVP (38.2% vs 30.1%, p <.001). Furthermore, among patients achieving >70% adherence, those on EFV were more likely to achieve 100% suppression than those on NVP (69% vs 31%, p <0.001). Variables significantly associated with shorter time to viral suppression in multivariate analysis (hazard ratio, 95% confidence interval) were female gender (1.17, 1.06 to 1.28), baseline viral load ≤10⁵copies/mL (1.28, 1.18 to 1.40), EFV-based regimen (1.20, 1.10 to 1.32), and strict adherence (3.79, 3.13 to 4.58, comparing 100% vs <50% adherence). Predictors for shorter time to virologic failure after initial suppression included low baseline CD4⁺ count (1.60, 1.22 to 2.10, comparing ≤50 vs >200 cells/µL), baseline viral load >10⁵ copies/mL (1.39, 1.14 to 1.70), NVP-based regimen (1.43, 1.16 to 1.75), and poor adherence (1.42, 1.15 to 1.73, per 10% decrease in adherence).

Conclusions:  Use of EFV led to faster viral suppression and slower viral rebound than NFV in this South African population, and was associated with better adherence. The mechanism for these findings deserves further exploration.