Background: Isolation of patients suspected for pulmonary tuberculosis is guided by serial sputum smears. This can result in isolation for days for patients with noncontagious tuberculosis. To determine whether a single sample negative for Mycobacterium tuberculosis complex at polymerase chain reaction (PCR) can guide isolation.
Methods: We retrospectively evaluated sputum samples analyzed for M. tuberculosis complex at the International Reference Laboratory of Mycobacteriology, Copenhagen, Denmark in 2002-2011. We selected culture-confirmed tuberculosis cases with ≥3 samples within 14 days before or after the initial culture-positive sample. We repeated the process for those with ≥2 samples within 28 days. The primary outcome was PCR-negative, smear-positive patients.
Results: We included 53 533 sputum samples from 20 928 individuals; 1636 had culture-confirmed tuberculosis. Of these, 856 had ≥3 sputum samples analyzed within the 28 days, and 482 had ≥1 PCR result. Nine patients (2.5% of smear-positive patients) were smear positive/PCR negative; 8 of the 9 had a smear-positive result in only 1 of 3 samples, and 5 had a low smear grade. Of 722 patients with 2 samples, 7 (1.3% of smear-positive patients) were smear positive/PCR negative. Overall, none were smear positive for the sample that produced the negative PCR result.
Conclusions: Primary PCR identified >97% of serial smear-positive cases. The majority of the missed cases had low-grade smears. Nevertheless, the occurrence of smear-positive/PCR-negative cases underlines the importance of increasing the quantity and quality of samples. Moreover, it is important that samples analyzed with PCR are cultured, owing to higher-sensitivity drug susceptibility testing, differential diagnosis, and surveillance.
Keywords: in-hospital transmission; nucleic acid amplification; polymerase chain reaction; respiratory isolation; tuberculosis.
© The Author 2015. Published by Oxford University Press on behalf of the Infectious Diseases Society of America. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.