Introduction The majority of cases of pulmonary tuberculosis (pTB) are diagnosed by microscopy and culture of sputum. When a patient is unable to produce sputum spontaneously, further procedures are required to obtain suitable samples for examination. There is debate about the relative merits of sputum induction (SI) and fiberoptic bronchoscopy (FOB; Brown M et al, 2007, Anderson C et al, 1995). SI is less invasive, cheaper, and unlikely to cause cross-infection, whereas FOB allows visualisation of the bronchial tree and other pathologies—especially cancer. At our hospitals, a diagnostic algorithm was devised to reduce the need for FOB for the diagnosis of pTB. Clinicians only requested bronchoscopy when three induced sputum samples were negative, unsuccessful, or contraindicated.

Method A retrospective cohort study, from 1 January 2008 to 31 December 2010. Patients undergoing SI for suspected pTB were identified from physiotherapy records, and the bronchoscopy database was interrogated for mycobacterial requests.

Results 521 induced sputum samples were sent from 214 patients. In total, 28 patients were diagnosed with pTB, 16 (57%) of whom were smear-positive. Non-tuberculous mycobacteria (NTM) were grown from four patients. Microbiological results of SI are listed in Abstract P53 table 1. Tuberculosis was diagnosed on the first sample in 25 cases (89%), the second in one (4%) case, and the third in two cases (7%). However, only 57% had at least three samples taken. Of 472 patients who underwent SI or FOB, 14 (3.0%) had both. Of those 14, three (21%) had positive samples for mycobacteria, two of which were Mycobacterium tuberculosis. All 14 had concordant culture results, whereas one case had a discordant smear result. In this case Mycobacterium malmoense was grown at both FOB and SI, although only the FOB sample was smear-positive.

Comment In this cohort, FOB carried out after SI did not increase the diagnostic yield. Fewer than 2% of those undergoing SI went on to have FOB, which suggests that clinicians were satisfied with SI and did not feel that bronchoscopy was required. However a systematic review and meta-analysis, and a larger, prospective study would be desirable.