Background: Many adults with pulmonary tuberculosis are unable to expectorate. Gastric washing, sputum induction using nebulized hypertonic saline, and bronchoscopy with bronchoalveolar lavage have all been used to obtain specimens for diagnosis, but to our knowledge, the timing and volume of induced sputum have not been well studied, and these 3 methods have not been compared.
Methods: The study recruited consecutive adult inpatients with chest radiography findings suggestive of tuberculosis who were unable to expectorate. Subjects provided 3 induced sputum samples for culture on day 1 and additional samples on days 2 and 3. In addition, gastric washing specimens were collected on days 1, 2, and 3. A proportion of subjects with negative smear results underwent bronchoalveolar lavage.
Results: The study recruited 140 subjects. Among 107 subjects who provided 3 gastric washing specimens and at least 3 induced sputum specimens, 43% had cultures positive for Mycobacterium tuberculosis. Use of 3 induced sputum samples detected more cases than did use of 3 gastric washings (39% vs. 30%; P=.03). Among 79 subjects with culture results for all 5 induced sputum specimens, there was no difference in yield between samples obtained by induced sputum induction performed in a single day or that performed over 3 days (34% vs. 37%; P=.63). There was no association between sputum volume and positive culture results. No additional cases were diagnosed in the 21 patients who underwent bronchoscopy.
Conclusions: Use of 3 induced sputum samples was more sensitive than use of 3 gastric washings for diagnosis of tuberculosis in patients who could not expectorate spontaneously. Use of bronchoscopy with bronchoalveolar lavage did not increase diagnostic sensitivity. Samples could be collected in 1 day, allowing for faster diagnosis, faster initiation of treatment, and shorter hospital stay.