In response to the letters of Drs. Harrington and Dr. Sobel, our data suggest that, compared with those with positive initial tuberculin reactions, the participants with positive booster reactions (defined as a PPd-T₂ reaction of 10mm and a PPd-T₂ minus PPd-T₁ reaction of 6 mm) were more likely to have had bacille Calmette–Guérin vaccination or sensitization to atypical mycobacterial antigens, both considered causes of false-positive tuberculin reactions. Dr. Sobel's data from Israeli school children support our conclusion that positive booster reactions are much less likely to represent previous tuberculous infection (that is, true-positive) than initial tuberculin reactions. In contrast to Dr. Sobel's data, some of our students with positive reactions had abnormal radiographs, so it is possible that a booster reaction represented old tuberculous infection in some. However, a history of household contact or parents who had immigrated from endemic areas, both of which were significantly associated with initial tuberculin reactions, were not associated with booster reactions. It is difficult to draw definite inferences regarding prognosis from cross-sectional data. Ferebee [1] tested middle-aged adults in mental institutions; the incidence of tuberculosis among initial tuberculin reactors was 122 of 100 000 compared with 66 of 100 000 among those whose second tuberculin test reaction was 10 mm or greater and 19 of 100 000 if both test results were negative [1]. If the risk for tuberculosis among young adults with a positive booster reaction is half the risk associated with a positive initial tuberculin reaction, this would alter the risk–benefit and cost–benefit ratios in favor of withholding preventive therapy.

We thank Dr. Stead for the clarification of our attribution of 12 mm over a previous negative test result based on conclusions drawn from Dr. Stead's two articles [2, 3]. In our study, we found that increases of as much as 25 mm occurred within only 1 week and that there was no definable cut point or size to distinguish boosting related to previous bacille Calmette–Guérin vaccination or sensitivity to PPd-Battey from presumed tuberculin infection.

The clinical dilemma of interpretation and management of a young adult with a positive booster reaction will not be resolved by changes in the criteria of interpretation. In our population, use of the increasingly stringent criteria to define a booster reaction as significant (that is, representing true tuberculous infection) simply resulted in diminished sensitivity without any improvement in specificity. We suggest that this issue will best be resolved by examining incidence of tuberculosis in a cohort of young adults who have had systematic two-step tuberculin testing.