Styblo has observed that there is an empiric and relatively constant relationship between the incidence of smear-positive tuberculosis and the annual risk of infection (11). He estimated that the ratio between this incidence and the ARI ranged between 50 and 60 in a variety of populations, pricipally those of high tuberculosis prevalence. Daniel and Debanne reasoned that this relationship could be used in reverse to estimate ARI when good case reporting was available but tuberculin skin test data were absent. They tested this hypothesis using data from white male U.S. naval recruits and found that in this population the ratio of incidence of tuberculosis of all forms to ARI was approximately 150 (10). The disparity between this figure and the lower figure of Styblo may result in part from the use of low-incidence populations, but the largest reason for the difference rests with the use by Daniel and Debanne of rates for all forms of tuberculosis, whereas Styblo considered only single-sputum smear-positive, pulmonary tuberculosis. In the current review paper estimation of ARIs is based on a ratio of 150, and ARI thus derived is referred to as the “estimated ARI.” While this method is imprecise, it is often the only means available to judge ARI in American populations.
In doing this review, three limitations were deliberately imposed. First, papers published prior to 1970 were used only to a limited degree and then only to provide historical context. Second, limited use was made of reports of outbreaks, for these accounts are usually anecdotal in character. Third, most papers describing studies done in other countries were excluded because both health care occupational sites and attitudes toward occupational risks in most other countries differ substantially from those in the United States.
A major limitation in this review and in the entire body of knowledge that it approaches rests with the definitions of tuberculous infection and of tuberculous disease. This subject is separately addressed in a paper authored by John B. Bass Jr., and is included in this report as Appendix B.
If one is to examine the occupational risk of tuberculous infection among health care workers in relation to the communities in which they reside, then it is first important to try to determine the annual risk of infection in the general American public. Tuberculin testing data upon