Background: This study investigated the appropriate number of sputum specimens and the usage patterns of tuberculosis tests (acid-fast staining, mycobacterial culture, and direct PCR) in the diagnosis of pulmonary tuberculosis.

Methods: We retrospectively analyzed the results of tuberculosis tests (8,216 acid-fast staining, 4,728 mycobacterial cultures, and 345 direct PCR) on 8,216 sputum specimens from 1,520 tuberculosis patients diagnosed through patient history and positive mycobacterial culture from September 2000 to March 2004.

Results: The first sputum specimen was positive in 77.6% of cases by acid-fast staining, with additional detection rates of 14.3%, 4.1%, and 4.0% for the second, third, and subsequent specimens, respectively. For mycobacterial culture, the rates were 78.6%, 14.6%, 4.0%, and 2.9%, and for direct PCR, the rates were 83.6%, 12.3%, 2.7%, and 1.4%. When three sputum specimens were submitted, the detection rates were 14.0% for acid-fast staining and 13.4% for mycobacterial culture. For direct PCR, when four or more specimens were tested, the detection rate was 9.9%. Specimens of inadequate quality accounted for 72.2% of acid-fast staining, 73.1% of mycobacterial culture, and 80.8% of direct PCR tests. The quality of sputum specimens was related to the detection rates in acid-fast staining (P = 0.000) and mycobacterial culture (P = 0.038), but not in direct PCR (P = 0.607).

Conclusion: For the diagnosis of pulmonary tuberculosis, it is appropriate to perform staining and culture on three consecutive sputum specimens. Education on the correct use of tuberculosis tests and proper specimen collection is deemed necessary.

(Korean J Clin Microbiol 2005;8(1):74-81)