How Good Is the Tuberculin Skin Test?

Abstract

Purified protein derivative is a relatively crude material containing several different antigens. The results of a skin test with purified protein derivative are dependent on the immunologic reactivity of the recipient of the test, and the final interpretation is dependent on the individual interpreting the test. Despite these limitations, the tuberculin skin test (TST) is one of the most accurate medical tests available. Using conservative assumptions, I have estimated the sensitivity of the TST at approximately 95% and the specificity at approximately 99% to 99.5%.2 If these assumptions are correct, the TST is more accurate than most tests commonly used in clinical medicine. Despite this high degree of accuracy, the TST is not good enough to give valid results in serial skin testing programs in most hospitals and other medical facilities where the prevalence of new infection is low (including facilities ordinarily considered to be at “moderately high risk”). The table translates the conversion rate in a serial skin testing program to the actual transmission rate of tuberculosis infection assuming a sensitivity of 0.95 and a specificity of 0.99 to 0.995. If the annual conversion rate in a skin testing program is less than 1%, most conversions will be false positives and the predictive value of an individual conversion will be poor. Conversion rates above 2% increase the predictive value of the test to greater than 50%, and results in skin testing programs with annual conversions of greater that 2% probably represent mostly true transmission of tuberculous infection. Conversion rates between 1% and 2% probably represent equal numbers of false-positive and true-positive tuberculin reactions. This issue of Infection Control and Hospital Epidemiology presents four reports concerning the TST.3-6 These reports are welcome, as the results of most serial skin testing programs in medical environments go unreported. Cook et al.4 report TST results among healthcare workers in New York City. The conversion rate in their study was approximately 1.3% annually—a rate that suggests equal numbers of true-positive and false-positive conversions. The presence of some true-positive conversions is supported by the increased conversion rate in high-risk settings and among foreign-born individuals, who would be expected to have a higher than average risk in their community away from the workplace. Some of the increased conversion rate among high-risk workers was probably the result of semiannual testing in this group compared with annual testing for low-risk workers, as more frequent testing will increase the cumulative false-positive rate if such a rate is constant for each test.2 Garber et al.3 report results of skin testing programs in 19 microbiology laboratories in New York City. The incidence of TST conversion in their study of 1% probably slightly favors false conversions over true conversions,