Setting: During the 1980s, New York City experienced a rapid increase of tuberculosis cases, more than 40% of which were human immunodeficiency virus (HIV)-associated.
Objective: To better define the molecular epidemiology of tuberculosis in New York City.
Design: We collected an isolate from every patient in New York City with a positive culture for Mycobacterium tuberculosis, including both incident and prevalent cases, in April 1991. Restriction fragment length polymorphism (RFLP) analysis using IS6110 was performed and the clinical, demographic, epidemiologic, and drug susceptibility patterns of patients were correlated with RFLP results.
Results: Of 441 patients, 12 (3%) had laboratory, clinical, and RFLP evidence of falsely positive cultures. The remaining 429 patients had 252 distinct RFLP patterns. Patients with clustered 1-3 band isolates did not share demographic or drug susceptibility patterns. Eliminating these patients from the analysis, 344 patients remained, of whom 126 (37%) belonged to one of 31 clusters ranging in size from 2-17 patients (median cluster size = 3). Clustering was more common among patients with multidrug-resistant isolates (53%), African Americans (44%), and the homeless (49%), but was not associated with HIV infection or acquired immune deficiency syndrome (AIDS), Multidrug-resistance, being African American, and homelessness remained independently associated with clustering in multivariate analysis. Of 79 patients in clusters of > or = 4 patients, 25 (32%) had identifiable epidemiologic linkages; 17 (74%) of these patients, and 6% of all cases, were documented to have been nosocomially associated.
Conclusion: A small but non-negligible proportion (3%) of New York City patients had falsely positive cultures for M. tuberculosis as a result of laboratory error. More than one third of all patients and most patients with multidrug-resistance in April 1991 had clustered RFLP patterns, suggesting recent transmission of M. tuberculosis. Homelessness, multidrug-resistance, and being African American independently increased the risk of clustering. Most of the identified epidemiologic linkages and 6% of all cases resulted from transmission in hospitals.