Decreases in Self-Reported ART Adherence Predict HIV Viremia Among Pregnant and Postpartum South African Women

Introduction: Routine HIV viral load (VL) monitoring is recommended for patients on antiretroviral therapy, but frequent VL testing, required in pregnant and postpartum women, is often not feasible. Self-reported adherence can be valuable, but little is known about its longitudinal characteristics. Methods: We followed women living with HIV from antiretroviral therapy initiation in pregnancy through 18-month postpartum in Cape Town, South Africa, with repeated measurement of VL and self-reported adherence using a 3-item scale. We used generalized estimating equations [with results presented as odds ratios (ORs) with 95% confidence intervals (CIs)] to investigate the association between viremia and change in adherence over pairs of consecutive visits. Results: Among 2085 visit pairs from 433 women, a decrease in self-reported adherence relative to the previous visit on any of the 3 self-report items, or the combined scale, was associated with VL >50 and >1000 copies per milliliter. The best-performing thresholds to predict VL >50 copies per milliliter were a single-level decrease on the Likert response item “how good a job did you do at taking your HIV medicines in the way that you were supposed to?” (OR 2.08, 95% CI: 1.48 to 2.91), and a decrease equivalent to ≥5 missed doses or a one-level decrease in score on either of 2 Likert items (OR 1.34, 95% CI: 1.06 to 1.69). Conclusions: Longitudinal changes in self-reported adherence can help identify patients with viremia. This approach warrants consideration in settings where frequent VL monitoring or other objective adherence measures are not possible.