Incremental yield of serial sputum examinations in the diagnosis of pulmonary tuberculosis in Taiwan: Findings of a pragmatic trial

Abstract

Abstract Background: Presumptive tuberculosis (TB) cases commonly had two to three sputum examinations in Taiwan. The incremental yield of serial sputum examinations has not been assessed before. Methods: In a pragmatic trial, presumptive TB patients with a frontline nucleic acid amplification test (NAAT) were classified as group A. Those without a frontline NAAT were randomized into group B frontline NAAT as intervention, and group C usual care. We investigated expected incremental yields and the number of examinations required for detection of one additional TB case from each serial sputum smear and culture. Results: Of 6835 presumptive TB cases, 395 (5.8%) were smear positive for acid-fast bacilli, and 195 (2.8%) culture positive for M tuberculosis. The expected incremental yield from a third smear was 3.5% and examination of 1712 (95% credibility interval 586e4706) third smears was required to detected one additional TB case. Sensitivity of one smear with an NAAT in group B was 46.8% (95% confidence interval 32.1%e61.9%), and that of two smears in Group C 40.0% (95% confidence interval 25.7%e55.7%). The expected incremental yield from a third culture was 8.4%, and the number of third cultures required to detect one additional TB case was 394 (95% credibility interval 231e670). Conclusions: The incremental yield of the third sputum smear was negligible. It may be reasonable to perform an NAAT, smear and culture on the first specimen and culture alone on the second. The utility of the third serial culture for the detection of additional TB case is debatable.