RESUMEN
BACKGROUND: Current therapy for latent TB infection (LTBI) is long, and requires close follow-up. This results in sub-optimal adherence-the major reason for failure of therapy. METHODS: In an open label randomised trial comparing 4 months of rifampicin with 9 months of isoniazid, the proportion and regularity of doses taken, measured with an electronic monitoring system (MEMS), and provider estimates of adherence in the first month of therapy, were assessed as predictors of treatment completion. RESULTS: Of 104 patients analysed, 86 took more than 80% of doses within the expected interval, 11 took more than 80% of doses but over a longer time interval than usually allowed, and seven did not complete treatment. Treatment completion was associated with the number of doses taken, and the variability of intervals between doses during the first month of treatment. CONCLUSIONS: Adherence in the first month, based on the number of doses and variability of times when taken, could be useful to predict completion of LTBI therapy. Interventions could be targeted to patients with suboptimal adherence in the first month.