Longitudinal Performance of Self-Reported ART Adherence Compared With Tenofovir Diphosphate in Dried Blood Spots in South Africa

Background: Self-reported ART adherence is widely used in resource-limited settings but may not accurately reflect cumulative drug exposure. Few longitudinal studies in sub-Saharan Africa have directly compared self-report with pharmacologic adherence measures, specifically tenofovir diphosphate concentrations (TFV-DP) in dried blood spots (DBS). Setting: Observational cohort study conducted in four public-sector ART clinics in Cape Town, South Africa. Methods: We analyzed data from 250 adults living with HIV, followed monthly for 13 visits. Participants reported missed ART doses in the prior 30 days and provided DBS for TFV-DP concentration. Longitudinal associations between missed doses and TFV-DP were examined using generalized estimating equations. Receiver operating characteristic (ROC) analyses evaluated the ability of SR to identify low TFV-DP in DBS (<699 fmol/punch). Logistic GEE models assessed the association between self-reported nonadherence and low TFV-DP. Results: Most participants were female (78%), median age 34 years (IQR, 27–42). Median TFV-DP concentrations declined with increasing self-reported missed doses: 1206 (IQR, 943–1442), 1009 (IQR, 773–1364), and 859 (IQR, 542–1194) fmol/punch for 0, 1–2, and ≥3 missed doses, respectively. Each additional missed dose was associated with lower TFV-DP concentrations in adjusted models (β = −29.2 fmol/punch per missed dose; p<0.001). Reporting ≥2 missed doses was associated with higher odds of low TFV-DP in longitudinal models (adjusted OR 1.50; p<0.001). Conclusions: Self-report missed doses correlated with TFV-DP concentrations over time but lacked sensitivity for detecting low drug exposure. Self-report remains pragmatic but may be optimized when paired with objective biomarkers to identify those at risk for treatment failure.