workers, but for some of the period they may have been at higher risk of multidrug-resistant disease. No national data report on the rate of tuberculosis infection among workers, but a small CDC demonstration project has suggested no association between job category and risk after demographic factors were taken into account. No recent occupation-specific mortality rates are available, but an analysis of older data suggests that some subgroups of health care workers may have been at higher risk of death than other workers.
In a 1995 review of data on tuberculosis among health care workers, Menzies and colleagues reviewed studies based on disease registries in several other countries. The studies reported “estimated risk ratios of 0.6 to 2.0, indicating at most only modest increased risk” of disease for health care workers compared with that for others in the community (Menzies et al., 1995, p. 92). Menzies and colleagues noted that such studies may underestimate risk because they do not standardize for age. This is important because the working population is younger and healthier than the general population. The authors also noted that cohort nor disease registry analyses are limited by “the inability to distinguish occupational from nonoccupational exposure” (Menzies et al., 1995, p. 92). The review also covered questionnaire studies from the 1950s through the 1980s that reported higher rates of disease for pathologists, certain laboratory technicians, and physicians. It noted concerns about modest response rates and possible recall biases for these studies. The authors identified no “recent” cohort studies comparing the risk of infection with M. tuberculosis among U.S. health care workers with the risk among individuals in the general community. Although they reviewed several studies reporting rates of tuberculosis infection among health care workers, the authors cited the limitations of the studies and did not present an overall assessment of infection risk in the postantibiotic era. They did not present mortality data.
Garrett and colleagues also reviewed U.S. surveys and surveillance reports for a variety of health care workers (e.g., physicians and house staff) and locales dating back to the 1960s (Garrett et al., 1999). They concluded that “available data suggest that the [annual] risk of [tuberculin skin test] conversions among hospital employees in general [i.e., for all categories of workers in nonoutbreak situations] is approximately 1 percent or less” (Garrett et al., 1999, p. 484). They did not present corresponding disease or mortality estimates. The authors noted that the data, which were sometimes inconsistent and subject to many methodologic limita-