Tuberculin skin test is a method to examine M. tuberculosis infection and has been used all over the world. But various factors make it difficult to understand testing results. In 2000, the American Thoracic Society recommended that skin test results should be decided by considering risk factors of the tested. In Korea, high tuberculosis infection rate and BCG vaccination rate make it difficult to differentiate current infection, past infection, and no infection by the skin test. This study was attempted to examine a negative predictive value of the skin test to understand how the skin test acts on deciding administration of anti-tuberculosis drug. METHOD: From Mar. 1 to Jul. 31 in 2001, the test was performed for patients hospitalized in Department of Internal Medicine, Hallym University College of Medicine, Chunchon, Korea by administering Tuberculin PPD RT23 2 TU (0.1 ml)to them that has been currently used in Korea based on Mantoux method. They were decided to be infected with tuberculosis bacilli by following diagnostic standard: 1) tuberculosis bacilli was cultured in sputum by microbiological diagnostic standard or Acid-fast bacilli was proven on a microscopic examination or 2) tuberculosis bacilli was not proven in the aforesaid microbiological test by clinical diagnostic standard, while there was opinion or symptom suitable for tuberculosis by radiographic or histological standard so the doctor decided to apply the tuberculosis treatment.

In this study, total 210 patients except 20 patients (8.7%) among 230 hospitalized patients were evaluated. Their average age was 60 16.8 years, and male-female rate was 1.28: 1 (male: 118, female: 92). Number of patient, who was diagnosed and decided as tuberculosis, was 53(25.2%). Pulmonary tuberculosis was found in 45 patients (84.9%); 22 patients were decided to be positive in the Acid-fast bacilli smear test by microbiological examination (culture positive: 13, culture negative: 9), and 23 patients were decided to be tuberculosis patients by clinical diagnosis standard. Tuberculosis pleuritis was found in 8 patients (15.1%); 4 patients were diagnosed and decided by histological standard, and 4 patients were decided and treated by clinical standard. In differentiating patients into 'Negative' and 'Positive' by the skin test standard of the American Thoracic Society, negative predictive value 92.3%, positive predictive value 47.3%, sensitivity and specificity were 83%, 68.8%, respectively.

In hospitalized respiratory patients, there was high negative predictive vlaue 92.3% by tuberculin skin test, therefore skin test would be a important factor for deciding administration of anti-tuberculosis drug on negative skin test patient.