Study product adherence measurement in the iPrEx placebo-controlled trial: concordance with drug detection

Abstract

Objective: To evaluate the concordance between adherence estimated by self-report (in-person interview or computer-assisted self-interview), in-clinic pill counts, and pharmacy dispensation records and drug detection among participants in a placebo-controlled pre-exposure prophylaxis HIV prevention trial (iPrEx).

Design: Cross-sectional evaluation of 510 participants who had drug concentration data and matched adherence assessments from their week-24 study visit.

Methods: Self-reported adherence collected through (1) interview and (2) computer-assisted self-interview surveys, (3) adherence estimated by pill count, and (4) medication possession ratio was contrasted to having a detectable level of drug concentrations [either tenofovir diphosphate (TFV-DP) or emtricitabine triphosphate (FTC-TP)], as well as to having evidence of consistent dosing (tenofovir diphosphate ≥ 16 fmol/10⁶ cells), focusing on positive predictive values, overall and by research site.

Results: Overall, self-report and pharmacy records suggested high rates of product use (over 90% adherence); however, large discrepancies between these measures and drug detection were noted, which varied considerably between sites (positive predictive values from 34% to 62%). Measures of adherence performed generally well in the US sites but had poor accuracy in other research locations. Medication possession ratio outperformed other measures but still had relatively low discrimination.

Conclusions: The sizable discrepancy between adherence measures and drug detection in certain regions highlights the potential contribution of factors that may have incentivized efforts to seem adherent. Understanding the processes driving adherence reporting in some settings, but not others, is essential for finding effective ways to increase accuracy in measurement of product use and may generalize to promotion efforts for open-label pre-exposure prophylaxis.