The most award winning
healthcare information source.
TRUSTED FOR FOUR DECADES.
Tuberculin or Mantoux testing is frequently used to establish a diagnosis of tuberculosis (TB) infection. Most often it is the gold standard upon which the decision to employ chemoprophylaxis is based.1 Recent revisions in the cutoffs for positivity have made the accuracy and reproducibility of this test even more important. In low-risk populations (healthy non-exposed adults), a 15 mm induration is required, whereas in those with a high probability for acquiring TB, the cutoff is only 5 mm (HIV-infected and close TB contacts). Although the dosing and methodology applying the tuberculin test have been standardized, the methodology for reading it has not.2,3 Both the palpation technique and the "ball point pen" technique are recommended, and the reading is frequently performed by different and sometimes relatively untrained personnel.3 The "pen" method involves the drawing of a line beginning 1-2 cm away from the area of reaction; the pen stops at the site of the first resistance due to induration, with the procedure repeated on the other side. The distance between the opposite lines of demarcated induration is then measured. To assess the repeatability and sources of error in determining the magnitude of induration, Pouchot et al evaluated both intra- and inter-observer variability and compared the reliability of the ball point pen technique to the palpation method. Ninety-six individuals were enrolled; those who had received bacille Calmette-Guerein vaccine were given preference. Ten units of standard tuberculin were injected on the volar surface of the forearm (Mantoux); technique and readings were done on day three. Two experienced investigators performed two repeated tests by "pen" and one by palpation.
There were 27 (28%) non-reactors. In these individuals, there was no intra-observer variability. Reliability of the second "pen" observation was highest for the 5 mm cutoff, compared to the 10 mm and 15 mm groups, and would have resulted in 12% of patients being reclassified. Ninety-five percent confidence limits were between 2.7 mm less than or 3.0 mm greater than the first measure. Between observers, the confidence limits were significantly higherfrom 3.4 mm to 3.7 mm in either direction. In 12% of cases, test results would have been reclassified. All reliability measures were slightly lower for the palpation technique, with a much broader area of imprecision (5 mm lower to 17 mm higher), resulting in reclassification of 17% of patients. There was only moderate agreement between "pen" and palpation methodology.
The tuberculin test is the most commonly used test in the evaluation of tuberculosis. It is used to establish infection (not necessarily disease) in at-risk patients. A "positive" PPD usually determines the need for preventive therapy with cutoffs for positivity depending on the pre-test probability of being infected (i.e., the higher the risk, the lower the cutoff). Significant concerns about the use of the standard Mantoux technique persist and include low levels of reactivity in immunosuppressed patients who may be at highest risk and technique variability. It is often administered by personnel with low levels of training. Pouchot et al now report on an additional area of concern--that of variability in reading test results. As the authors point out, intra-observer variability may be the most important factor, as only one examiner usually tests a given patient. As indicated by both measures of reliability and intra-observer confidence limits, the "pen" method was more reproducible, with the palpation method having 38% broader confidence limits. Other studies have also demonstrated significant variability of the palpation methods.1-3 Regardless of the method, significant reclassification resulted when the test was read a second time either by the same or different observer.
What is the take-home message of this simple but important study? First, the ball point pen method should be the preferred technique for tuberculin testing and should probably result in more accurate diagnosis. Second, all those involved in the diagnosis and management of tuberculosis (physicians and their extenders) should be aware of the wide confidence intervals for tuberculin test reading and interpret with caution values that are close to positive cutoff. The tuberculin test remains imperfect, but, so far, it’s the best we have.
1. American Thoracic Society. Diagnostic standards and classification of tuberculosis. Am Rev Respir Dis 1990;142:725-735.
2. Loudon RG, et al. Variation in tuberculin test reading. Am Rev Respir Dis 1963;87:852-861.
3. Bouros D, et al. The role of inexperience in measuring tuberculin skin reaction (Mantoux test) by the pen or palpation technique. Respir Med 1992;86:219-223.