Abstract Body

Background: Gross inaccuracies in participant-reported adherence has dampened or eliminated confidence in self-report as a method to characterize PrEP use. Sophisticated measurement strategies available in clinical trials, however, are unlikely candidates for adoption in clinical practice, contributing to concerns about PrEP roll-out. Self-report of self-selected open-label PrEP use has not been characterized to date and may differ from observations in efficacy trials. We examined the performance of self-report in this specific context using self-reported recent PrEP use among iPrEx open label extension (OLE) participants contrasted to drug detection.

Methods: All iPrEx OLE participants choosing to receive PrEP at the 11 geographically-diverse research sites were informed that a blood specimen collected at some point during their first 12 weeks of receiving PrEP would be evaluated for drug levels and results would be shared with them. Recall of dates of last 3-doses taken was collected via interview at all on-drug visits. Drug levels were determined using blood plasma by liquid chromatography tandom MS having a lower limit of quantification of 10 ng/ml, reflecting dosing in the past 3 days. Self-report (at least one dose in the past 3-days vs no dosing) was compared to drug detection (detected versus not detected) using binary logistic regression.

Results: 1172 participants had drug levels matched to self-report. The vast majority reported having had at least one dose in the past three days (84%) and of these 83% had detectable drug (PPV). Among the 16% reporting not having dosed, 82% had no drug detected (NPV). Sensitivity of self-report was 96%; specificity was 48%. Self-report was highly associated with drug level, OR 21.57 [14.42-32.26], which retained significance when controlling for site or whether the report was of taking or not taking PrEP. Age interacted with self-report and drug detection; among those reporting recent dosing, age positively associated with having drug detected.

Conclusions: In this study where participants had the choice of receiving open label PrEP, reports of recent PrEP use had high association and moderate concordance with drug detection. Participants misclassified as “adherent’ via self-report tended to be younger, suggesting the need to develop alternative assessment strategies for younger PrEP users. In contrast to data from PrEP efficacy trials, results suggest that use of self-report in practice may be a valuable tool.