UCSF

BACKGROUND:Objective adherence metrics for tenofovir (TFV) disoproxil fumarate/emtricitabine(FTC)-based PrEP were critical for interpretation of efficacy in PrEP clinical trials, and there is increasing interest in using drug levels to tailor interventions for reengagement and adherence. Point-of-care immunoassays (POC) for TFV, which examine short-term adherence, are in development. However the ability of poor short-term and long-term adherence to predict future PrEP non-retention is unknown. SETTING:Secondary data analysis of a large, prospective multi-site U.S. PrEP demonstration project METHODS:: An adjusted Cox-proportional hazards model examined the relationship of dried blood spot (DBS) levels of FTC-triphosphate (FTC-TP) or TFV-diphosphate (TFV-DP), measures of short-term and long-term PrEP adherence, respectively, with future study non-retention. RESULTS:Overall, 294 individuals (median age 33 years) contributed drug levels within the U.S. PrEP demonstration project. By study end, 27% were lost to follow-up, 25% had at least one undetectable FTC-TP level indicating poor short-term adherence, and 29% had a drug level indicating sub-optimal long-term adherence (TFV-DP<700 fmol/punch). The strongest factor associated with future study non-retention using a binary drug-level cut-off was an undetectable DBS FTC-TP level [adjusted hazard ratio (AHR) 6.3; 95% confidence interval (CI) 3.8-10.2]. Sub-optimal long-term adherence based on low DBS TFV-DP levels was also associated with non-retention (AHR 4.3; 95% CI 2.4-7.6). CONCLUSIONS:Both short and long-term metrics of PrEP adherence are strongly associated with future loss to follow-up in a U.S. demonstration project study. Short-term metrics of adherence, once available at the point-of-care, could be used to direct real-time tailored retention and adherence interventions.