Abstract Body

High adherence was critical to the efficacy of daily oral emtricitabine/tenofovir (FTC/TDF) preexposure prophylaxis (PrEP) in clinical trials. Low adherence or early discontinuation may reduce the effectiveness of PrEP in clinical practice.

We conducted a cohort study of Kaiser Permanente Northern California members initiating PrEP from July 2012 through December 2014. Follow-up was from the first dispensing of FTC/TDF until the earliest of PrEP discontinuation (i.e., ≥120 days without medication), health plan disenrollment, HIV seroconversion, death, or end of study (June 2015). Refill adherence was calculated by dividing days’ supply dispensed by total days between first and last FTC/TDF fill during follow-up among patients with ≥2 fills. We used chi-square tests to examine low adherence (<60%, consistent with taking <4 of 7 doses per week) by age, gender, and race/ethnicity. Multivariable log-binomial regression was used to estimate risk ratios (RRs) for factors associated with discontinuation. 

Among 972 individuals who initiated PrEP, there were 850 person-years of follow-up, with a mean of 0.9 years per person. The mean age at PrEP initiation was 37 years (range 18-68), and 98% of PrEP users were men. The majority were White (65%), followed by Hispanic (11%), Asian (9.7%), and Black (4.0%). Among 915 individuals with ≥2 fills, mean adherence was 92% (median 97%; interquartile range: 90%-100%), with >80% adherence in all demographic subgroups. Only 27 (3.0%) PrEP users had <60% adherence, with a higher proportion with low adherence in patients aged <30 vs. ≥30 years (5.7% vs. 2.0%, P=0.005) and in Blacks/Hispanics vs. other racial/ethnic groups (6.6% vs. 2.3%, P=0.007); the rarity of low adherence precluded multivariable analysis of this outcome. PrEP was discontinued by 219 (23%) individuals. There were no differences in discontinuation by age or race/ethnicity, but women were over twice as likely to discontinue than men (RR 2.4, 95% confidence interval: 1.6-3.6; P<0.001). There were no seroconversions during PrEP use; however, there were 2 new HIV infections in Black and Hispanic men aged <30 years who had discontinued PrEP.

There were no HIV infections among active PrEP users during 850 person-years of follow-up, which is consistent with the high adherence observed in this population. Given the two seroconversions after PrEP discontinuation, there is a critical need for strategies to support continuation of PrEP throughout periods of HIV risk.