Effects of treatment interruption patterns on treatment success among patients with multidrug-resistant tuberculosis in Armenia and Abkhazia.

BACKGROUND The success of the current treatment regimen for multidrug-resistant (MDR) tuberculosis is poor partly owing to a high default rate. Many studies have explored predictors of poor outcomes, but very few have assessed the effects of treatment interruptions on treatment outcomes for MDR tuberculosis. METHODS We conducted a retrospective analysis among patients with MDR tuberculosis enrolled in 2 MDR tuberculosis programs using regimens recommended by the World Health Organization under directly observed therapy. Treatment outcomes were defined as successful if the patient was cured or completed treatment and unsuccessful if the patient died or defaulted from treatment or if treatment failed. The effect of patterns of interruptions on treatment outcomes was assessed through multivariate logistic regression. RESULTS A total of 393 patients with MDR tuberculosis were included in the study; 171 (43.5%) had a successful outcome, and 222 (56.5%) an unsuccessful outcome: 39 (9.9%) died, 56 (14.3%) had failed treatment, and 127 (32.3%) defaulted from treatment. In multivariate analysis, long interruptions (≥3 days) (adjusted odds ratio, 3.87; 95% confidence interval, 1.66-8.98) and short gaps (<10 days) between interruptions (3.94; 1.76-8.81) were independently associated with an unsuccessful treatment outcome. DISCUSSION This study shows that in a directly observed therapy-based MDR tuberculosis program, treatment interruptions at short intervals of ≥3 days directly affect treatment outcome.