and should not come from the malnourished, often homeless population that contributes substantially to national tuberculosis mortality.

In the surveillance studies cited above the methods of tuberculin skin testing, the completeness of follow-up, and the definitions of tuberculin positivity and of tuberculosis varied, and the rigor of examination for tuberculosis may also be open to challenge. Yet certain generalizations seem justified.

First, the attack rates among tuberculin reactors is substantially lower than the oft-stated 10 percent. Even if risks observed during the first few years after infection are projected forward in linear fashion, it is hard to envision a cumulative risk as high as 10 percent. The lifetime risk estimates by Comstock and Edwards (93) suggest that perhaps 3 percent should be chosen; other studies suggest a rate closer to 5 percent.

Next, although the risk cannot be converted to annual risk in any of them, it is apparent from these studies that the risk diminishes with the passage of time. In those studies in which age-specific data are presented, the risk among adults is greatest in the young adult, postpubertal years. This is the age when many individuals enter the health care workforce, but it is not representative of the many older health care workers.

The disease risk is dramatically increased in immunocompromised individuals. The risk of tuberculosis in infected persons is substantially reduced by appropriate treatment of latent infection, and health care workers should be ideally situated for the use of such treatment. Tuberculosis mortality risk in immunocompetent health care workers with tuberculosis not due to multidrug-resistant organisms is probably close to zero.

This review could not have been accomplished without the help of others whom the author wishes to acknowledge. The reference librarians at the Health Sciences and Allen Memorial Libraries of Case Western Reserve University were helpful in finding many of the articles reviewed here, including obtaining a few by interlibrary loan. Marilyn Field, Elizabeth Epstein, and Cara Christie of the Institute of Medicine were helpful in supplying additional background material. Amy Curtis and Eugene McCray at CDC kindly gave me permission to cite some of their unpublished work. Finally, I am grateful to the members and staff of the Institute of Medicine Committee on Regulating Occupational Exposure to Tuberculosis for helpful reviews of successive drafts of this paper.