OBJECTIVES: We aimed to assess the extent of treatment interruption caused by efavirenz-associated ADEs.

METHODS: A case-control study of efavirenz recipients who did, versus did not (control) develop adverse drug events (ADE), and who were matched for baseline CD4 + at a ratio of 1:1.3 was conducted. Antiretroviral -naïve patients who were started on efavirenz were followed up retrospectively, and their records scrutinized every month for 2 years. Demographic and clinical predictors of treatment interruption were computed using Cox proportional hazard models. Kaplan- Meier curves were plotted to assess time to treatment interruption for the two groups. Clinical endpoints were: i) efficacy -improved CD4 + counts and/or viral load (VL) suppression, ii) safety -absence of treatment-limiting toxicities, and iii) durability - no interruption until follow-up ended.

RESULTS: Both groups had comparable CD4 + counts at baseline (p = 0.15). At t = 24-months, VL in both groups were suppressed to undetectable levels (<20 copies/mL) while median CD4 + was 353 cells/µL (IQR: 249-460). The mean time on treatment was 23 months (95% CI, 22.3 -23.4) in the control group without ADE and 20 months (95% CI, 18.9 - 21.6) in the ADE group (p = 0.001). Kaplan-Meier plots demonstrated that 59.5% of patients who experienced ≥ 1 ADE versus 81% of those who did not experience any ADE were estimated to continue treatment for up to 24 months with no interruption (p = 0.001). Most interruptions to EFV treatment occurred in the presence of opportunistic infections and these were detected within the first 5 months of treatment initiation. Independent predictors which negatively impacted the dependent variable i.e., treatment durability, were intravenous drug use (adjusted hazard ratio, aHR 2.17, 95% CI, 1.03-4.61, p = 0.043), presence of ≥ 1 opportunistic infection(s) (aHR 2.2, 95% CI, 1.13-4.21, p = 0.021), and presence of ≥ 1 serious ADE(s) (aHR 4.18, 95% CI, 1.98-8.85, p = 0.00).