Comparing Predictive Ability of Two Objective Adherence Measures in a Community-Based Cohort on Antiretroviral Therapy in South Africa: Tenofovir Diphosphate Concentrations and Electronic Adherence Monitors

Abstract

Background: Electronic adherence (EA) and tenofovir diphosphate (TFV-DP) in dried blood spots (DBS) are objective measures of antiretroviral therapy (ART) adherence. We characterized the association between these measures in a prospective cohort of persons with HIV (PWH) on ART. Setting: Four primary health clinics in Cape Town, South Africa. Methods: We enrolled 250 virally suppressed PWH receiving tenofovir-based ART. We collected EA data, monthly viral load, and TFV-DP in DBS for 12 months. We used logistic regression to estimate the adjusted odds ratio (aOR) and 95% confidence intervals (CIs) for future viral breakthrough (VB) (>400 copies/mL) for each adherence measure. Receiver operating characteristics (ROCs) provided the predictive power of these measures. Results: Participants had a median (IQR) age of 34 (27–42); 78% were women. Twenty-one (8%) developed VB. Logistic regression showed that when percent EA and TFV-DP concentrations increased, the odds of VB decreased. This relationship was consistent at the time of VB (aOR of 0.41 [95% CI: 0.25 to 0.66] for TFV-DP and aOR of 0.64 [95% CI: 0.54 to 0.76] for EA) and for up to 2 months before VB. Both adherence measures predicted future VB at both 1 month and 2 months before viral load measurement. Conclusion: We established that 2 objective adherence measures, EA and TFV-DP in DBS, have a positive association with, and are both strongly predictive of, VB in a community-based South African cohort on ART. Future research is needed to determine the feasibility of implementing these adherence measures in resource-limited settings to facilitate adherence interventions.