A method for estimating the prevalence of tuberculosis infection

Abstract

The prevalence of tuberculous infection in a population is generally estimated from calculating the proportion of tested individuals who react with at least 10 mm of induration to 5 TU of PPD-S tuberculin. Reactions due to infection with atypical mycobacteria, however, may cause the prevalence to be overestimated. This paper is concerned with an alternative method of estimating the prevalence of infection with Mycobacterium tuberculosis. The method utilizes population distributions of reaction size by dividing study populations into two groups--individuals with and without known exposure to tuberculosis. The mathematical model developed here removes the effect of atypical infections and provides a truer picture of tuberculous infection. Data from a Navy recruit population demonstrate the use of the model with the result that among recruits with no known exposure to tuberculosis, the estimated prevalence is reduced by about one-half. Among recuits with known exposure to tuberculosis, there is essentially no difference between the two methods. Important advantages in using this method are that probabilities of true infection by induration size are generated, and that itis less sensitive to variations caused by differences in reading techniques and in tuberculin potencies. Furthermore, it is applicable to other diseases if the underlying assumptions are met.