Abstract

Background: Previous studies suggest that bronchoscopy and a single induced sputum sample are equally effective for diagnosing pulmonary tuberculosis.

Methods: In a prospective study of subjects with possibly active pulmonary tuberculosis, the diagnostic yield of three induced sputum tests was compared with that of bronchoscopy. Subjects either produced no sputum or (acid fast) smear negative sputum. Bronchoscopy was only performed if at least two induced sputum samples were smear negative.

Results: Of 129 subjects who completed all tests, 27 (21%) had smear negative and culture positive specimens, 14 (52%) on bronchoscopy and 26 (96%) on induced sputum (p<0.005). One patient was culture positive on bronchoscopy alone compared with 13 on induced sputum alone; 13 were culture positive on both tests. Induced sputum positivity was strikingly more prevalent when chest radiographic appearances showed any features of active tuberculosis (20/63, 32%) than when appearances suggested inactivity (1/44, 2%; p<0.005). Induced sputum costs were about one third those of bronchoscopy, and the ratio of costs of the two tests per case of tuberculosis diagnosed could be as much as 1:6.

Conclusions: In subjects investigated for possibly active or inactive tuberculosis who produce no sputum or have smear negative sputum, the most cost effective strategy is to perform three induced sputum tests without bronchoscopy. Induced sputum testing carries a high risk of nosocomial tuberculosis unless performed in respiratory isolation conditions. The cost benefits shown could be lost if risk management measures are not observed.