Background:  Active drug use is associated with poor antiretroviral adherence, but correlates of adherence in injection drug users have not been determined. Further, the prevalence of ART medication errors among injection drug users is unknown.

Methods:  We report baseline data from INSPIRE, a 4-city randomized trial of a behavioral intervention to reduce sex and injection risks and improve health care utilization and adherence among HIV-infected injection drug users. We used ACASI to assess adherence, asking subjects their precise ART prescriptions and how many pills they skipped the preceding day. Good adherence was defined as mean overall adherence of ≥ 95%. Medication errors were defined as incorrect daily doses of 1 or more ART, using both standard and alternative ART prescriptions as the reference.

Results:  Of 1161 HIV-infected Injection drug users, 55% (n = 636) were taking ART and 48% (n = 560) were taking HAART. Among those taking ART, 26% had a viral load < 400 copies/mL. Most subjects (74%, n = 473) had good adherence, which was strongly associated with having an undetectable viral load (p < 0.001). Good adherence was also associated with being a high school graduate (OR 95% CI 1.7 [1.2 to 2.5]), male sex (1.4 [1.0 to 2.0]), not being homeless (2.0 [1.4 to 3.0]), no cocaine use in the past 3 months ([1.9 [1.2 to 3.0]), and not sharing injection equipment (2.4 [1.6 to 3.5]). Several psychosocial factors were associated (p ≤ 0.01) with good adherence:  greater perceived social support, greater self-efficacy for disclosing drug use to providers and for taking HIV medications, positive attitudes toward HIV medication, better communication with providers, and not being depressed. Medication errors were made by 54% (n = 346), and were associated (p < 0.001) with detectable viremia and fewer CD4 cells. In multivariate analyses, the strongest predictors of good adherence were self-efficacy for taking HIV medications (OR_adj 95% CI 2.1 [1.5 to 2.9]), and positive attitudes toward HIV medications (1.8 [1.0 to 3.4]), and the strongest predictors of poor adherence were sharing injection equipment (2.2 [1.4 to 3.6]), and depression (1.3 [1.0 to 1.7]). The strongest independent predictors of medication errors were worse self-perceived health status (OR_adj 95% CI 1.5 [1.1 to 2.1]), poorer self-efficacy for safer drug use (1.2 [1.0 to 1.4]), and worse attitudes toward HIV medications (1.9 [1.2 to 3.1]).

Conclusions:  Adherence interventions for injection drug users should focus on reducing medication errors, improving self-efficacy and attitudes toward HIV medications, and treating depression.

Keywords: adherence