Agreement Between Measures of Adherence to Isoniazid Preventive Therapy Among People With HIV in Uganda

BACKGROUND: Isoniazid (INH) preventative therapy is recommended for people with HIV (PWH) in resource-constrained settings. Valid measures are needed to assess adherence. We aimed to examine agreement between measures overall and by level of social desirability. METHODS: PWH with latent tuberculosis (TB) were recruited in Mbarara, Uganda. Past 30-day adherence was measured by the number of days with pill bottle openings using a medication event monitoring system (MEMS) and self-reported number of days pills taken. INH concentration (INH plus acetyl INH and their ratio) in hair samples was measured. We used Bland-Altman plots to examine agreement between adherence measures and calculated the area under the receiver operating characteristics curve (AUROC) to determine if INH hair concentration predicted optimal MEMS-measured adherence (≥90%). RESULTS: A total of 301 participants enrolled; 92% were virologically suppressed, and adherence was high. The median (interquartile range [IQR]) number of pill bottle openings in 30 days was 28 (24–30) compared with 30 (28–30) via self-report. The median INH concentration (IQR) was 36.2 (17.2–62.4), and the INH:acetyl ratio was 2.43 (0.99–3.92). Agreement between self-reported and MEMS adherence was greater at more optimal adherence levels. INH:acetyl INH ratio was not predictive of optimal adherence according to MEMS (AUROC, 0.62; 95% CI, 0.52–0.72) in a subset (n = 161). CONCLUSIONS: Lower MEMS adherence levels compared with self-report suggest the need for objective adherence measures. Biologic measures have potential, although in this study INH concentration was not predictive of MEMS measured adherence. More data are needed to assess the accuracy of biologic measures.