Background: Patients with suspected pulmonary tuberculosis (PTB) are usually placed in respiratory isolation awaiting three sputum smear microscopy results for acid-fast bacilli (3AFB). GeneXpert MTB/RIF (Xpert) on a pooled sample from two sputa may allow for more rapid de-isolation.
Objectives: To compare the sensitivity and negative predictive value (NPV) of Xpert performed on a single pooled sputum sample ('pooled Xpert') to 3AFB, in order to exclude PTB in patients placed in respiratory isolation.
Methods: Hospital inpatients in respiratory isolation for possible PTB were enrolled prospectively. Three expectorated sputum samples were obtained for smear microscopy. Two of the same samples had 0.5 ml removed from each and pooled for pooled Xpert. The diagnostic accuracy of pooled Xpert and 3AFB were assessed and compared to liquid culture at 8 weeks as the reference standard.
Results: Of 56 participants, nine (16.1%) were diagnosed with PTB. Compared to liquid culture, pooled Xpert had a sensitivity of 88.9% (95% confidence interval (CI) 57-99%) and NPV of 97.9% (95% CI 89-99%). 3AFB had a sensitivity of 66.7% (95% CI 35-88%) and NPV of 93.5% (95% CI 83-98%).
Conclusions: A single pooled Xpert was non-inferior to 3AFB, with a strong trend towards greater sensitivity and better NPV. These findings support the use of a single pooled Xpert as an effective rapid screening approach for ruling out PTB in low incidence settings. Its value in high incidence settings and optimal combination with smear microscopy and culture warrant further evaluation.
Keywords: Detection; Non-tuberculous mycobacteria; Respiratory isolation; Tuberculosis diagnosis; Xpert MTB/RIF.
Copyright © 2019 The Author(s). Published by Elsevier Ltd.. All rights reserved.