Trends in Drug-Resistant Mycobacterium tuberculosis in New York City, 1991–2003

Background. Two drug-resistance surveys showed a very high prevalence of drug resistance among isolates obtained from patients with tuberculosis in 1991 and 1994 in New York, New York.Methods. A cross-sectional survey in April 1997 and a survey of incident cases in April–June 2003 were conducted. The trend in the proportion of drug resistance in the 4 surveys was examined separately for prevalent and incident cases. Risk factors for drug resistance in incident cases were also assessed.Results. The number of patients was 251 in the 1997 survey and 217 in the 2003 survey. Among prevalent cases, the percentage of cases with resistance to any antituberculosis drug decreased from 33.5% in 1991 to 23.8% in 1994 and to 21.5% in 1997 (P < .001, by test for trend); cases of multidrug-resistant tuberculosis also decreased significantly, from 19% in 1991 to 6.8% in 1997 (P < .001, by test for trend). Among incident cases in the 4 surveys, the decrease in resistance to any antituberculosis drugs was not statistically significant; however, the decrease in multidrug-resistant tuberculosis (from 9% in 1991 to 2.8% in 2003) was statistically significant (P = .002, by test for trend). However, in 2003, a worrisome increase in incident cases of multidrug-resistant tuberculosis (an increase of 23%) was seen among previously treated patients with pulmonary tuberculosis not born in the United States. Human immunodeficiency virus infection, a strong predictor for drug resistance in 1991 and 1994, was not associated with drug resistance in subsequent surveys.Conclusions. Intensive case management, including directly observed therapy, adherence monitoring, and periodic medical review to ensure appropriate treatment for each patient, should be sustained to prevent acquired drug resistance.