The Booster Effect in Two-Step Tuberculin Testing among Young Adults in Montreal

The resurgence of tuberculosis [1], and the emergence of multiple-drug-resistant strains of bacteria [2] in North America has heightened awareness of the risks for nosocomial transmission of tuberculosis. Periodic tuberculin testing has been recommended for health care personnel who are likely to be exposed [1, 3, 4] after initial two-step testing [1, 3-6]. Two-step tuberculin testing has been recommended because a single tuberculin test may elicit little response yet stimulate an anamnestic immune response so that a second test, even when done as soon as 1 week later, may elicit a much larger, or boosted, response [5, 7, 8]. This booster phenomenon is thought to represent remote tuberculous infection among elderly persons because it is more common with increasing age [6, 9, 10]. Among residents in Arkansas nursing homes, those with a reaction to a second tuberculin test of 10 mm or more had the same risk for tuberculosis infection as did those with an initial tuberculin reaction of 10 mm or more [11]. Two-step tuberculin testing should increase sensitivity by detecting those with remote tuberculous infection but may also diminish specificity because the booster phenomenon has been associated with previous BCG vaccination [12-16] or sensitization with atypical mycobacteria [6, 17, 18]. Among young adults, who include most new hospital employees and have a low prevalence of tuberculous infection [6-8], it is unclear whether a significant booster reaction indicates remote tuberculous infection. Similarly, the risk for tuberculosis development in a young adult with a nonsignificant tuberculin reaction but a positive booster reaction is not known. Consequently, no consensus exists regarding the prognosis and definition of a significant booster reaction [4-6, 11, 19]. Quebec is unique in North America because BCG vaccinations were given en masse from 1948 to 1980. Thus, approximately 50% to 60% of young adults have received BCG vaccination [20, 21]. Fewer than 5% of nonvaccinated young adults have significant reactions to tuberculin or to atypical mycobacterial antigens [22]. We wanted to evaluate the factors associated with booster reactions and to estimate the sensitivity and predictive value of different criteria for the booster reaction in these young adults. Methods Bacille CalmetteGuerin Vaccination In Quebec from 1948 until 1980, BCG vaccinations were given en masse. More than 100 000 vaccinations were given each year, with 50% to 60% of the population vaccinated, although coverage varied from 10% to 80% in health districts within greater Montreal [21]. Vaccination was given by the scarification method using vaccine manufactured by the Institut Armand Frappier (Laval, Quebec). Infants received 3 106 viable bacilli, and all other persons were given 5 106 bacilli [20]. More than 90% of recipients had significant tuberculin reactions when tested 6 to 8 weeks later [20]. Most were vaccinated in infancy, although 20% were vaccinated at primary school entry, and fewer than 10% were vaccinated at ages 10 to 16 years. All BCG vaccinations were recorded in vaccination booklets, which were given to each child, and at a central registry at the Institut Armand Frappier. These records are still readily accessible. Participant Selection The faculties of health sciences and the student health services of six post-secondary institutions in Montreal were contacted. All students entering health professional training programs in 1989, 1990, and 1991 were eligible, except those who were on leave or receiving immunosuppressive therapy. Those with previously documented positive tuberculin tests completed questionnaires, and their previous results were recorded. The results among foreign-born students, whose BCG vaccination status could not be verified, have been reported elsewhere [23]. This study was approved by an ethics committee of the Montreal Chest Hospital. Data Collection Before skin testing, students gave informed consent and completed self-administered questionnaires regarding demographic data, BCG vaccination, previous skin tests, and potential exposure to tuberculosis. Using plastic disposable syringes and 27-gauge needles, 0.1 mL of 5 TU PPd-T (for Mycobacterium tuberculosis; Connaught Laboratories, Toronto, Canada) was injected intradermally on the volar aspect of the forearm, using the Mantoux technique. In 1991 only, students were also initially tested with 0.1 mL of purified protein derivative-Battey (PPd-B) (for M. intracellulare; Connaught Laboratories) on the volar aspect of the opposite forearm. After 48 to 72 hours, the maximal induration was defined using the ballpoint method [24], measured using machinists' calipers, and recorded in millimeters by one of four experienced readers. Those with reactions to the first test (PPd-T1) of 10 mm or more were not retested. Those with reactions to PPd-T1 of less than 10 mm were retested 1 to 4 weeks later (PPd-T2) on the opposite forearm using the same techniques for testing and reading. Bacille CalmetteGurin vaccination was verified from childhood vaccination booklets, through the registry of the Institut Armand Frappier, or both. The accuracy of this information was previously verified [22]. All students with PPd-T1 or PPd-T2 reactions of 10 mm or more were referred to the Montreal Chest Hospital for chest roentgenogram and further evaluation. Roentgenogram reports were reviewed and classified as consistent with previous tuberculous infection if there was mention of granulomata, hilar calcifications, or apical pleuroparenchymal scarring. Data Analysis Initial Tuberculin Reaction For students who had initial dual testing with PPd-T and PPd-B in 1991, the results of the reactions to the two antigens were adjusted as described below. This adjustment was based on evidence that U.S. military recruits with simultaneous reactions to PPd-B that were larger than to PPd-T had a low risk for tuberculosis, unless the reaction to PPd-T was 12 mm or more [25]. If PPd-T1 was less than PPd-B and PPd-T1 was less than 12 mm, then the reaction to PPd-T1 was assumed to represent cross-reactivity and the result was adjusted to zero; if PPd-T1 was 12 mm or more, no adjustment was made, regardless of size of PPd-B reaction; if PPd-T1 was equal to or greater than PPd-B, then PPd-B was assumed to represent cross-reactivity and the result was adjusted to zero. PPd-T1 was considered positive if the reaction measured 10 mm or more (for students in 1991, after adjustment as above), and PPd-B was considered positive if the reaction measured 5 mm or more. Booster Reaction A dichotomous outcome was defined as positive if PPd-T2 was 10 mm or more and measured at least 6 mm more than PPd-T1 [5, 6, 14, 15, 26, 27]. Boosting Effect A continuous outcome measured in millimeters was defined as PPd-T2 minus PPd-T1. This was used to analyze the biologic determinants of boosting and to examine the frequency distribution of change in reaction size associated with different factors. Using these definitions, associations were tested for statistical significance using the chi-square test for categorical and the Student t-test for continuous independent variables [28]. All data were analyzed using a personal computer and SAS (SAS Institute, Cary, North Carolina) and BMDP software (BMDP, Statistical Software, Los Angeles, California) for logistic regression. A P value less than 0.05 was accepted as evidence of statistical significance [28]. Confidence intervals (95%) for proportions were calculated as suggested by Colton [28]; adjusted odds ratios were calculated from logistic regression as suggested by Kleinbaum and Kupper [29]. Sensitivity, specificity, and positive predictive values were calculated as suggested by Sackett [30]; measures of agreement were calculated according to methods described by Feinstein [31]. Results Participants Of 2399 eligible students, 1961 (81.7%) completed the two-step testing protocol (Table 1). This report was confined to the 1542 Canadian-born students whose BCG vaccination status could be verified (from vaccination booklets for 1004 participants and from the records of the Institut Armand Frappier for 538 participants). Records of 332 students who had vaccination booklets were also traced through the central registry to verify accuracy and concordance of these two sources. Agreement between the two sources was 84% ( statistic = 0.68). In all, 591 (38%) students had received BCG vaccination, of whom 64% were vaccinated only once in infancy, 24% once between the ages of 2 and 10 years (median, 6 years), and 12% twice, primarily in infancy and between the ages of 5 and 7 years. The mean (SD) age of the participants was 21.4 4.1 years and 84% were female, reflecting the preponderance of female students in many health professional training programs. Table 1. Characteristics of the Study Participants* Skin Testing Of the 1542 participants, 107 (6.6%) had positive reactions to PPd-T1, were referred for evaluation, and were not retested. Of the 440 students tested in 1991, 26 (5.9%) had reactions to PPd-B measuring 5 mm or more, and 7 (1.6%) had reactions measuring 10 mm or more. Among the 1435 students who had a second test (PPd-T2), initial reactions to PPd-T1 had an important effect on boosting, as shown in Table 2. The proportion with positive booster reactions increased with larger size of initial reaction, as did the mean booster effect measured in millimeters. However, of the 74 students with a positive booster reaction, 49 (66%) had an initial reaction to PPd-T1 measuring 0 mm. Table 2. Effect of Initial Tuberculin Reaction on Booster Reaction* Effect of Bacille CalmetteGurin Vaccination After correction for cross-reactivity to PPd-T1, BCG vaccination did not have any significant effect on reactions to PPd-B. Of the 591 students vaccinated with BCG, 84 (14.2%) had positive PPd-T1 reactions, compared with 22 of the 951 students (2.3%) who were not vaccinated (chi-square test; P