RESUMO
BACKGROUND: Adherence to tuberculosis preventive treatment (TPT) is an important determinant of clinical benefit. We assessed the association of participant behaviors early in TPT with subsequent discontinuation. METHODS: We used data from a phase 3 randomized trial and the preceding phase 2 trial to compare 4 months of rifampin to 9 months of isoniazid for TPT. We excluded participants whose providers discontinued TPT due to adverse events or tuberculosis disease. We analyzed 4 outcomes: discontinuing TPT within the first month of treatment, discontinuing TPT between the first and second month, discontinuing TPT after the second month, and completing treatment but not per protocol. We analyzed the association of outcomes with regimen and participant characteristics and 4 behavioral predictors of discontinuation recorded at the month 1 and month 2 follow-up visits: reporting symptoms of intolerance, missing >20% of doses, rescheduling appointments, and not bringing their medication bottle. RESULTS: Overall, 6656 participants were included (phase 3, 5848; phase 2, 808), of whom 4318 (64.9%) completed treatment per protocol. Participant characteristics were inconsistently associated with discontinuation. Phase 3 trial participants with 1, 2, or 3-4 behavioral predictors at the month 1 follow-up had 5.0 (95% confidence interval, 3.6-6.7), 18.6 (13.3-26.1), and 79.4 (38.2-165.0), respectively, higher odds of discontinuing before the second month. The corresponding number of predictors at the month 2 follow-up had 1.8 (1.4-2.2), 4.7 (3.6-6.2), and 7.4 (4.6-11.9) higher odds of discontinuing before completing treatment; phase 2 findings were similar. CONCLUSIONS: Four behavioral predictors recorded early in therapy were more strongly associated with subsequent discontinuation than participant characteristics, particularly when more than 1 behavioral predictor was recorded. Clinical Trials Registration. NCT00170209; NCT00931736.