Tuberculosis (TB) remains a major threat to human health, and TB diagnostic methods remain unsatisfactory. Nucleic acid amplification tests (NAATs) show higher sensitivity compared with culture for the diagnosis of pulmonary TB (PTB). However, NAATs are expensive and cannot be easily implemented outside major medical centers. To improve the sensitivity of NAATs for PTB diagnosis, we investigated the predictive factors that might optimize NAAT utilization.

A total of 1263 patients with suspected PTB were enrolled for evaluation. The sensitivity, specificity, and accuracy of methods including smear-microbiology, culture of Mtb and NAAT for Mycobacterium tuberculosis (Mtb) detection in sputum and bronchoalveolar lavage fluid samples were compared. Odds ratios and 95% confidence intervals were used to assess variables that might be associated with positive NAAT results for sputum and bronchoalveolar lavage fluid from patients with suspected PTB.

NAAT showed higher sensitivity for Mtb detection (61.1%) when compared with smear (9.0%) and Mtb culture (47.8%). We found that an elevated erythrocyte sedimentation rate, the presence of cavities, and positive interferon-γ release assay (IGRA) results were indicative of positive Mtb detection by NAAT. Moreover, individuals who had all three of these characteristics showed an 86% diagnostic positivity for PTB from Mtb detection by NAAT.

Our study suggests that an elevated erythrocyte sedimentation rate, a positive IGRA result, and the presence of pulmonary cavities are helpful factors for predicting positive Mtb detection by NAAT. Patients with the three positive clinical markers should undergo NAAT for Mtb detection because they are the most likely individuals to be bacteriologically confirmed as having TB.