ABSTRACT
OBJECTIVE: To assess the utility of various cut-off points of tuberculin skin test in making a diagnosis of tuberculosis in patients with respiratory symptoms. METHODS: Tuberculin skin test was conducted on consecutive new patients attending chest clinic for various respiratory symptoms. All subjects were then investigated to establish diagnosis, and categorized into tuberculous and nontuberculous groups. Receiver operating characteristic (ROC) curve was plotted to evaluate discrimination by tuberculin skin test. Sensitivity, specificity and predictive value were also calculated at various cut-off points. RESULTS: Of 250 patients, 59 (23.6%) had tuberculosis on clinical and microbiological criteria (other than the tuberculin test). Sensitivity and specificity of tuberculin test at readings greater than 5, 10 and 15 mm were 0.8136 and 0.7068, 0.6271 and 0.8901, and 0.2034 and 0.9738 respectively. Area under ROC curve for this test was 0.80. CONCLUSION: A cut-off point of 10 mm is likely to be useful in supporting a diagnosis of tuberculosis in patients with strong clinical suspicion of tuberculosis, in other patients, 15 mm cut-off may be more suitable.