Background: Adverse outcomes occurring early after antiretroviral therapy (ART) initiation are common in sub-Saharan Africa, despite reports of high levels of ART adherence in this setting. We sought to determine the relationship between very early ART adherence and early adverse outcomes in HIV-infected adults in Botswana.

Methods: This prospective cohort study of 402 ART-naïve, HIV-infected adults initiating ART at a public HIV clinic in Gaborone, Botswana evaluated the relationship between suboptimal early ART adherence and HIV treatment outcomes in the initial months after ART initiation. Early adherence during the interval between initial ART dispensation and first ART refill was calculated using pill counts. In the primary analysis patients not returning to refill and those with adherence <0.95 were considered to have suboptimal early adherence. The primary outcome was death or loss to follow-up during the first 6 months of ART; a secondary composite outcome included the primary outcome plus incident opportunistic illness (OIs) and virologic failure. We also calculated the percent of early adverse outcomes theoretically attributable to suboptimal early adherence using the population attributable risk percent (PAR%).

Results: Suboptimal early adherence was independently associated with loss to follow-up and death (adjusted OR 2.3, 95% CI 1.1-4.8) and with the secondary composite outcome including incident OIs and virologic failure (adjusted OR 2.6, 95% CI 1.4-4.7). However, of those with early adverse outcomes, less than one-third had suboptimal adherence and approximately two-thirds achieved virologic suppression. The PAR% relating suboptimal early adherence and primary and secondary outcomes were 14.7% and 17.7%, respectively.

Conclusions: Suboptimal early adherence was associated with poor outcomes, but most early adverse outcomes occurred in patients with optimal early adherence. Clinical care and research efforts should focus on understanding early adverse outcomes that occur despite optimal adherence.