more precise form has been retained. In many instances, no time interval was given, and in these cases percent is used. Because many of these conversions to percentages are based on small absolute numbers, the actual numbers are given in those cases to allow the reader to note this fact.
The annual risk of infection (ARI) with M. tuberculosis, the tubercle bacillus, is central to any consideration of the occupational tuberculosis risk of health care workers. Ideally, one would like to know this figure not only for employees but also for the community in which they reside. Data upon which the ARI for American populations can be calculated are generally lacking, however, and many of the reports of infections in health care workers do not provide relevant time intervals. In this review, ARI is expressed as percent per year, the usage of most workers who have dealt with this subject.
It is acknowledged that the term “annual risk” as it is used here is imprecise—annual probability or likelihood would be more accurate terms—but its use is widely established in the published literature on this subject, and it is used here for consistency with that literature. It is also true that such calculations on an annual basis ignore the fact that the pool of individuals considered may change during a year. However, data are almost always lacking in the studies reviewed here for estimation of the more accurate use of person-years at risk. Errors thus introduced are small at low levels of risk and do not affect the conclusions drawn in this review.
Two types of ARI are reported here: (a) calculated annual risks, when the data permit such direct calculation, and (b) estimated annual risks based on tuberculosis incidence when direct calculation is not possible.
If one accepts the development of tuberculin hypersensitivity as a reliable index of primary tuberculous infection, then it is a straightforward task to calculate the ARI with M. tuberculosis expressed as percent per year from serial skin testing data. The algebraic formula for this calculation is
ARI = [1 − (Qb/Qa)1/(b−a)] * 100,
where Q is 1 − P, P is the probability of being infected at a given age or year a or b, a is the initial age or year of observation, and b is the second year or age of observation (10). In the text of the present review paper, ARI thus calculated is referred to as the “calculated ARI.”