RESUMO
SETTING: Large urban tuberculosis control program. OBJECTIVES: To determine the frequency and characteristics of treatment interruptions, and the factors associated with the different types of treatment interruptions. DESIGN: This was a case-control study using culture-positive tuberculosis (TB) patients verified in 1998-1999. Case patients included those in whom any of the following mutually exclusive categories of treatment interruption: default with return to therapy, directly observed therapy nonadherence, default without return to therapy, or multiple types of interruptions. Controls were selected randomly from the cohort. RESULTS: Overall, 6.0 percent of patients had treatment interruptions. All types of treatment interruption were associated with prolonged treatment course and decreased treatment completion rates. The median number of months to treatment interruption was 4.0 (range, 0.5-28.9 months). Two factors were significantly associated with every type of interruption: homelessness and lack of awareness of the severity of TB disease. In multivariate analysis, only lack of awareness of the severity of disease remained independently associated with all interruption types. CONCLUSION: Efforts to improve patients' understanding of TB disease and related treatment issues may be an important TB control program strategy and should be emphasized at the initiation of therapy and at intervals throughout the treatment course to minimize treatment interruption.