Background: Important management decisions depend on results of the tuberculin skin test. However, the test is subject to several potential errors, and its reliability has not been adequately studied.
Objective: To ascertain the reliability of tuberculin skin testing.
Design: Cross-sectional study.
Setting: University hospital.
Participants: 96 persons who received a tuberculin skin test.
Measurements: Ballpoint-pen and palpation measures of induration.
Results: Global intra- and interobserver reliability coefficients of the ballpoint-pen technique were high. Five percent of the time, however, a second measurement by the same observer could be at least 2.7 mm less to 3.0 mm more than the first measurement and the measurement from the second observer could be at least 3.4 mm less to 3.7 mm more than the measurement from the first observer. This could lead to the reclassification of a positive test result as negative or vice versa. The area of imprecision was 38% less broad for the ballpoint-pen technique than for the palpation technique.
Conclusion: Reading of tuberculin skin tests may frequently result in misclassifications when measurements are close to the cutoff point that separates negative from positive results.