from workplace records can be very labor intensive, and address information may not be current or entered into computer databases. Because tuberculosis is negatively correlated with income, some studies have also investigated whether places of residence for workers with converted skin tests are clustered in low-income—and presumably higher-risk—areas. Others have simply used low income as a surrogate for higher risk of community exposure.


A review article by Sepkowitz (1994) summarizes studies from the 1920s through the 1950s that showed that nurses, physicians, and others working with tuberculosis patients had high rates of positive skin tests or skin test conversions compared to the rates expected in the broader community. For nursing students who were initially tuberculin skin test negative, conversion rates reported in studies in the United States and Europe ran as high as 95 to 100 percent by the time the students graduated. The reported yearly incidence of active tuberculosis ranged from 2 to 12 percent in nursing schools.4 Comparative data on nonhealth care occupations or work settings is limited, but Sepkowitz cites studies in the 1930s that reported the incidence of active tuberculosis to be 1 percent for employees of a life insurance company and 2 percent for food handlers.

Other early studies reviewed by Sepkowitz focused on medical students. One survey of those who had been medical students between 1940 and 1950 reported a tuberculosis case rate of 334 per 100,000 medical students per year. In the general population during that period, the estimated tuberculosis case rates ranged between 32 and 100 per 100,000 population per year. Another study identified a particularly high risk to medical students of participation in autopsies on those who died with active tuberculosis.


Prevalence is a measure of the probability of infection or disease in a population at a particular point in time. Incidence is the probability of new infection or disease in a specified period of time, usually a year. It is not always clear whether studies are reporting incidence rates or probabilities, which are based on different denominators (see, e.g., Kahn and Sempos [1989]). The denominator for a rate is based on the average population at risk of some event during a defined interval (reflecting reductions in the population due to the occurrence of the event, e.g., death or acquisition of disease). The denominator for a probability uses the population at risk at the beginning of the defined interval. Because the denominator of a rate reflects reductions in an at-risk population during an interval, a rate will be higher than a probability. The spread between the two increases as the level of risk increases. For example, if the annual probability of an event is 0.05 (and the event is experienced uniformly during the year), then the annual rate will be 0.051; if the probability is 0.12, the rate will be 0.128.

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement