Setting: A large urban teaching hospital in the southeast of Paris.
Objective: Primary surveillance of nosocomial transmission of tuberculosis (TB) by systematic restriction fragment length polymorphism analysis (RFLP) of isolates (n = 161) recovered from smear-positive pulmonary TB patients identified from 1 March 1993 to 28 February 1994, and from all TB patients (with any form of tuberculous infection) identified from 1 March 1994 to 30 April 1995.
Results: Systematic RFLP analysis revealed 12 clusters of patients (n = 40) infected by strains of Mycobacterium tuberculosis showing matching RFLP patterns. None of the isolates were multidrug-resistant. Compared with non-clustered patients, clustered patients were more likely to be homeless (55% vs 19%, P < or = 0.001), or Africans living in hostels for migrant workers (20% vs 6%, P = 0.01), and had fewer previous admissions to hospital (12% vs 28%, P = 0.05). Further epidemiological investigations showed that the clustered TB cases actually resulted not from nosocomial transmission, but from transmission in the community, very likely in homeless shelters and hostels for migrant workers.
Conclusion: No nosocomial transmission of TB was identified among the patients included during the study period. Systematic RFLP analysis using hospital-based sampling can detect the spread of TB in specific environments in the community where transmission is occurring.