Microscopy compared to culture for the diagnosis of tuberculosis in induced sputum samples: a systematic review [Review article].

BACKGROUND Resource-limited settings rely on sputum examination using microscopy to diagnose tuberculosis (TB); however, the sensitivity of the test is poor and case detection rates are low. Sputum induction is proposed as a way to improve sample collection and enhance test sensitivity. OBJECTIVE To undertake a systematic review of studies comparing microscopy and culture sensitivity in induced sputum samples. METHODS We ran duplicate searches of databases (up to August 2011) and searchable websites of major human immunodeficiency virus (HIV) and TB conferences (up to November 2010) to identify studies comparing the performance of microscopy compared to culture on induced sputum samples, with culture as the reference standard. RESULTS A total of 23 studies met our inclusion criteria. The overall success of the induction was high, ranging from 76.4% (95%CI 68.5-83.2) to 100% (95%CI 98.5-100), while adverse events associated with sputum induction were infrequent and mild. The sensitivity of microscopy compared to culture ranged from 0% to 100%; only eight studies reported on the species of mycobacterium isolated in culture. Yield was generally higher for sputum induction compared to nasopharyngeal aspiration and gastric lavage, and compared equally well to bronchoalveolar lavage and physiotherapy. DISCUSSION Sputum induction increases TB case detection and is useful for people who are negative on spontaneous smear microscopy or unable to expectorate spontaneously. It is well-tolerated by children and adults, irrespective of HIV status, and can be used where culture is not available. The use of induced sputum samples with molecular tests, such as Xpert® MTB/RIF, warrants further investigation.