Background: Poor adherence to antituberculosis treatment is the most important obstacle to tuberculosis control.
Purpose: To identify and analyze predictors and consequences of nonadherence to antituberculosis treatment.
Patients and methods: Retrospective study of a citywide cohort of 184 patients with tuberculosis in New York City, newly diagnosed by culture in April 1991-before the strengthening of its control program-and followed up through 1994. Follow-up information was collected through the New York City tuberculosis registry. Nonadherence was defined as treatment default for at least 2 months.
Results: Eighty-eight of the 184 (48%) patients were nonadherent. Greater nonadherence was noted among blacks (unadjusted relative risk [RR] 3.0, 95% confidence interval [CI] 1.1 to 8.6, compared with whites), injection drug users (RR 1.5, 95% CI 1.1 to 2.0), homeless (RR 1.4, 95% CI 1.0 to 1.8), alcoholics (RR 1.4, 95% CI 1.0 to 1.9), and HIV-infected patients (RR 1.4, 95% CI 1.1 to 1.9); also, census-derived estimates of household income were lower among nonadherent patients (P = 0.018). In multivariate analysis, only injection drug use and homelessness predicted nonadherence, yet 46 (39%) of 117 patients who were neither homeless nor drug users were nonadherent. Nonadherent patients took longer to convert to negative culture (254 versus 64 days, P < 0.001), were more likely to acquire drug resistance (RR 5.6, 95% CI 0.7 to 44.2), required longer treatment regimens (560 versus 324 days, P < 0.0001), and were less likely to complete treatment (RR 0.5, 95% CI 0.4 to 0.7). There was no association between treatment adherence and all-cause mortality.
Conclusions: In the absence of public health intervention, half the patients defaulted treatment for 2 months or longer. Although common among the homeless and injection drug users, the problem occurred frequently and unpredictably in other patients. Nonadherence may contribute to the spread of tuberculosis and the emergence of drug resistance, and may increase the cost of treatment. These data lend support to directly observed therapy in tuberculosis.