Blumberg, Sotir, and colleagues studied nosocomial transmission of tuberculosis infection and skin test conversions among employees at Grady Hospital in Atlanta, Georgia, a public hospital admitting about 200 tuberculous patients annually during the first half of the 1990s before and after implementation of intensified infection control measures (33,34). Six-month conversion rates fell from 118/3,579 (3.3 percent; calculated ARI = 6.49 percent/year) in the first 6 months of 1992 to 23/5,153 (0.4 percent; calculated ARI = 0.89 percent/year) during the first 6 months of 1994. During the latter period, the conversion rate was not related to job status but was positively correlated with black race and low economic status. Subsequently, Blumberg and associates studied tuberculin skin test conversions among house staff in the Emory University Affiliated Hospitals Training Program (35). These interns and residents spend approximately half of their training time at Grady Hospital. As noted above, expanded infection control measures were implemented at Grady Hospital in 1992, and tuberculin test conversion rates were compared for the 6-month period at the initiation of these measures with the rates for the subsequent 4.5 years. The rate fell from 6.0 per 100 person-years to 1.1 per 100 person-years (p <0.001). Rates were significantly higher for house officers in medicine and obstetrics/gynecology than for those other clinical departments. Graduates of foreign medical schools had higher conversion rates than American graduates. As at St. Clare’s Hospital and the Cabrini Medical Center in New York, the implementation of control measures was thought to have had an impact on transmission of infection to the health care workers at Grady Hospital.
An important study of tuberculosis in New York City health care workers was conducted using restriction fragment length polymorphism (RFLP) DNA fingerprinting techniques (36). In 1992–1994 among six New York City hospitals where no recognized nosocomial outbreaks of tuberculosis occurred, isolates from 20 cases of tuberculosis occurring in health care workers were available for typing. Of the 20, 8 were nurses or nurses aides, 7 were physicians, and the remaining 5 were not in patient contact positions. The tuberculous health care workers from whom the fingerprinted organisms were isolated did not differ from those from 181 nonhealth care workers similarly studied with respect to age, sex, country of birth, race, and HIV infection status. The fingerprinting technique allowed the identification of clusters of patients all infected with the same strain of M. tuberculosis. Overall, 87 of 201 isolates fingerprinted in New York during the period of the study were clustered, indicating that they represented recent transmission of currently circulating strains. Among health care workers, clustered strains were found in six of the seven physicians and in eight of the nine HIV-infected workers (all occupations). This suggests that physicians and HIV-infected persons were particularly susceptible to occupational infection.