Overreporting of adherence to hepatitis C direct-acting antiviral therapy and sustained virologic response among people who inject drugs in the HERO study

Background Self-reported adherence to direct-acting antivirals (DAAs) to treat hepatitis C virus (HCV) among persons who inject drugs (PWID) is often an overreport of objectively measured adherence. The association of such overreporting with sustained virologic response (SVR) is understudied. This study among PWID aimed to determine a threshold of overreporting adherence that optimally predicts lower SVR rates, and to explore correlates of the optimal overreporting threshold. Methods This study analyzed per-protocol data of participants with adherence data (N = 493) from the HERO (Hepatitis C Real Options) study. Self-reported and objective adherence to a 12-week DAA regimen were measured using visual analogue scales and electronic blister packs, respectively. The difference (Delta) between self-reported and objectively measured adherence was calculated. We used the Youden index based on receiver operating characteristic (ROC) curve analysis to identify an optimal threshold of overreporting for predicting lower SVR rates. Factors associated with the optimal threshold of overreporting were identified by comparing baseline characteristics between participants at/above versus those below the threshold. Results The self-reported, objective, and Delta adherence averages were 95.1% (SD = 8.9), 75.9% (SD = 16.3), and 19.2% (SD = 15.2), respectively. The >= 25% overreporting threshold was determined to be optimal. The SVR rate was lower for >= 25% vs. < 25% overreporting (86.7% vs. 95.8%, p <.001). The factors associated with >= 25% Delta adherence were unemployment; higher number of days and times/day of injecting drugs; higher proportion of positive urine drug screening for amphetamine, methamphetamine, and oxycodone, and negative urine screening for THC (tetrahydrocannabinol)/cannabis. Conclusions Self-reported DAA adherence was significantly greater than objectively measured adherence among PWID by 19.2%. Having >= 25% overreported adherence was associated with optimal prediction of lower SVR rates. PWID with risk factors for high overreporting may need to be more intensively managed to promote actual adherence.