ABSTRACT
BACKGROUND: We evaluated an appropriate number of sputum specimens and usage of mycoba c-terial tests (AFB stain, culture and direct PCR) for the diagnosis of pulmonary tuberculosis. METHODS: We reviewed retrospectively the results of mycobacterial tests (AFB stain 8,216, culture 4,728, direct PCR 345) for 8,216 sputum specimens from 1,520 patients with pulmonary tuberculosis from September, 2000 to March, 2004. RESULTS: The cumulative detection rates of acid-fast bacilli by AFB stain was 77.6% in the first specimen, 91.9% in the second, 98.0% in the third and 100% in any specimen submitted later than the third. The respective figures were 78.6%, 93.1%, 97.1% and 100% for AFB culture, and 83.6%, 95.9%, 98.6% and 100% for direct PCR. Three consecutive sputum specimens were submitted only in 14.0% for AFB stain and 13.4% for culture. The quality of sputum graded by Gram stain was poor in 72.2% for AFB stain, 73.1% for culture and 80.8% for direct PCR. The quality of sputum was found to affect significantly the positive rates of AFB stain (P = 0.000) and culture (P = 0.038), but not the rate of direct PCR (P = 0.607). CONCLUSION: Three sequential sputum specimens are needed to diagnose pulmonary tuberculosis effectively and the importance of sputum quality and appropriate use of mycobacterial tests should be emphasized.
