Abstract Body

Background: No gold standard exists for adherence measurement. Self-report is commonly used but it is subjective. Objective measures are more costly and difficult to implement. Research is limited on the performance of these measures for pre-exposure prophylaxis (PrEP) against HIV infection.

Methods: The Partners PrEP Ancillary Adherence Study collected the following adherence measures in 1,147 HIV-uninfected partners in serodiscordant couples from 3 Ugandan sites in the Partners PrEP Study (a randomized placebo-controlled trial of tenofovir-based PrEP collected).:
Monthly self-report
Rating– “How well have you taken your study tablets?”
Frequency– “How often do you take your study tablets?”
Percent– “What percent of the time were you able to take your study tablets?”
Objective measures
Electronic monitoring (MEMS, downloaded monthly)
Unannounced pill counts (UPC, performed monthly at home)
Plasma tenofovir level (month 6 in a randomly selected subset of 228 participants on active drug)
Rating and frequency responses were recorded on a 6-point Likert scale and converted to percents (e.g., most of the time=80%). Monthly adherence over 6 months was compared by Spearman correlation and to tenofovir level (detectable >0.32ng/ml) by regression analysis with the Huber White sandwich variance estimator.

Results: A total of 1,143 individuals were included in the analysis; median age 34 (IQR: 30-40), 606 (53%) were male. Median (IQR) adherence by each measure was as follows: rating 90% (83-93), frequency 93% (90-97), percent 97% (93-98), MEMS 97% (91-99), and UPC 98% (96-99). Correlations among self-report measures ranged from 0.61-0.67 (p<0.001). Correlations of self-report measures ranged from 0.31-0.36 with UPC (p<0.001) and 0.23-0.25 with MEMS (p<0.001). Comparisons of adherence levels with detection of tenofovir are shown in Figure 1.

Conclusions: All measures revealed high adherence. The distribution of adherence varied among types of self-report queries; the widest was seen with rating, suggesting greater potential utility for identifying individuals with adherence challenges. Correlation with MEMS and UPC was significant, but modest, likely reflecting differences in measurement techniques (e.g., daily variation vs monthly summaries). All measures significantly discriminated between detectable and undetectable tenofovir levels; however, MEMS performed best and is less resource intensive than UPC.

viewer?TAG_ACTION=SECURE_FILE_DOWNLOAD&V