Abstract Objective: To compare four laboratory methods in the diagnosis of pulmonary tuberculosis. Methods: Respiratory secretion specimens were collected from 160 patients suspected of having pulmonary tuberculosis. Direct testing for Mycobacterium tuberculosis was carried out using Ziehl-Neelsen staining, auramine staining, culture on Löwenstein-Jensen (LJ) medium and polymerase chain reaction (PCR). The strains isolated were identified by means of a radiometric method using p-nitro-alpha-acetylamino-beta-hydroxypropiophenone (NAP) and classical methods. The sensitivity of the methods was compared to the gold standard for the diagnosis of pulmonary tuberculosis, based on clinical, radiological and microbiological criteria. Results: Of the 160 patients, 142 were diagnosed with pulmonary tuberculosis according to the gold standard. The sensitivity of Ziehl-Neelsen staining, auramine staining, culture on LJ medium and PCR was 54.2%, 58.4%, 67.6% and 77.5%, respectively, when compared with the diagnostic criterion adopted. All four methods presented 100% specificity. In the identification of mycobacteria, there was high (96.8%) concordance between PCR and the radiometric method using NAP. The sensitivity of PCR was 50.8% in samples with negative sputum smear microscopy results and 98.8% in those with positive results. The sensitivity of PCR was lower in specimens with negative results in sputum smear microscopy and culture than in those with positive results (25.6% and 99.0%, respectively). Conclusions: We found PCR to be a promising method for the diagnosis of pulmonary tuberculosis, even in paucibacillary specimens. Simultaneous identification and faster results are additional advantages of this method.

Keywords: Tuberculosis, pulmonary/diagnosis; Culture media; Polymerase chain reaction; Sputum/microbiology.