Prevalence, predictors, and outcomes of early adherence after starting or changing antiretroviral therapy.

The objectives of this research were to assess prevalence and predictors of early antiretroviral therapy adherence using multiple indicators and to estimate effects of early adherence on subsequent HIV viral load and CD4+ lymphocyte responses. Study subjects were adults with HIV infection referred to an antiretroviral therapy-monitoring clinic for initiation or change in therapy between March 1998 and June 1999. The design was a prospective observational cohort involving baseline interview followed by 30 days of electronic adherence monitoring (MEMS), 30-day interview, and follow-up viral load at 1, 3, and 6 months. Adherence indicators included MEMS therapeutic coverage, observed/expected cap openings, and self-reported adherence assessed at 30 days. Of 235 consenting patients, 60 (26%) failed to complete 30 days of electronic monitoring (noncompleters). At 6 months, mean change from baseline plasma viral load was inferior among noncompleters (0.5 log vs. 1.7 log). Predictors of adherence, varying by adherence metric, included: gender, race, prior antiretroviral therapy experience, substance abuse, prior adherence behavior, health beliefs, and pharmacist prediction of adherence. Self-reported adherence was more sensitive in predicting viral load responses than MEMS-based measures and identified poor adherence at earlier time points. Approximately a quarter of consenting patients were unable to complete 30 days of MEMS monitoring, and early drop out was a poor prognostic sign. Predictors of adherence varied depending upon how adherence was measured. Differences in virologic response between patients with optimal or poor adherence may not emerge until several months after regimen change or initiation. Structured assessment of self-reported adherence is an inexpensive and useful tool to assist clinicians in monitoring adherence.