Poor concordance between objective and reported measures of pre-exposure prophylaxis (PrEP) adherence and factors associated with over-reported adherence among adolescent girls and young women (AGYW) in Kenya

BACKGROUND: Adherence is crucial for PrEP effectiveness but challenging to determine in real-world PrEP programs and resource-limited settings. We investigated concordance between an objective measure of adherence based on tenofovir-diphosphate (TFV-DP) concentrations within dried blood spot (DBS) samples and self-reported adherence, and identified cofactors of over-reported adherence among AGYW in Kenya.
METHODS: We enrolled participants, randomly selected among AGYW, aged 18-24 years, who were participating in the PEPFAR DREAMS-PrEP program in Kisumu and Homa Bay counties. Participants underwent interviews and provided DBS samples at two timepoints approximately three months apart between June 2019 and January 2020. We assessed adequate adherence, defined as TFV-DP '¥700 fmol/punch or reporting 4+ doses in past week. We used Cohen's Kappa statistic and McNemar's test to assess agreement and discrepancy between adequate DBS-based and self-reported adherence by timepoint. We used bivariate mixed effects logistic regression models to explore cofactors of over-reported adherence, defined as reporting 4+ doses in past week with TFV-DP <700 fmol/punch.
RESULTS: Among 359 AGYW, 93.6% and 62.8% reported current PrEP use at Timepoint1 and Timepoint2, respectively. Among those reporting current PrEP use, 88.7% reported 4+ doses in past week and 6.3% had adequate DBS-based adherence at Timepoint1. At Timepoint2, 83.3% reported 4+ doses in past week and 4.4% had adequate DBS-based adherence. Adequate DBS-based and self-reported adherence differed markedly (p-value <0.001); agreements between the two measures was only slightly better than chance (Kappa <0.1) at both timepoints. The proportion of AGYW who over-reported adherence was 82.7% at Timepoint1 and 78.8% at Timepoint2. Over-reported adherence was positively associated with reporting partner aware of PrEP use (OR=1.70, 95%CI: 1.05-2.74), preference for safe space over clinic for PrEP service (OR=1.74, 95%CI: 1.02-2.99) and moderate/severe depression as measured by PHQ-9 scale (OR=1.97, 95%CI: 1.05-3.69), and negatively associated with reporting moderate/high perceived chance of getting HIV (OR=0.41, 95%CI: 0.21-0.82) and partner with positive/unknown HIV status (OR=0.48, 95%CI: 0.29-0.78).
CONCLUSIONS: AGYW in this study had very low adherence to daily oral PrEP and considerably over-estimated their PrEP adherence in self-reports. AGYW at high risk of HIV need more assistance to achieve adequate levels for effective PrEP.