Pulmonary tuberculosis: Evaluation of current diagnostic strategy.

OBJECTIVE To identify tools that will result in faster diagnosis, making the current pulmonary tuberculosis strategy more efficient Patients and methods: A 4-year (2015 - 2018) retrospective study. The gold standard for diagnosis was a positive culture from a respiratory specimen. All sputum, fibroscopy and post-fibroscopy specimens (for smear negative patients) were collected. Each specimen was analyzed through smear examination and culture. All nucleic acid amplification testing results were included. Analyses looked at the incremental yield of positive cases of each successive specimen collection, and time to diagnosis. RESULTS A total of 354 patients had at least one positive culture. Sputum allowed a diagnosis in 92% of cases (including a gain in sensitivity of around 7% for the third sputum specimen), with 160 smear-positive patients (45%). Among smear-negative patients, 109 underwent a fibroscopy procedure (culture sensitivity of 75%), and 59 had a post-fibroscopy specimen collected, which together identified the rest of the patients (8%). Molecular testing was used in 237 specimens. Median time to diagnosis was 11 days, which was significantly reduced among smear-negative patients when molecular testing was used (p<0.001). Shortening the delay between sputum specimen collections did not alter procedure sensitivity. CONCLUSIONS We identified several aspects of the French tuberculosis diagnosis algorithm that could be improved, and posed the basis for a prospective study. Centers in higher incidence areas could benefit from a dedicated, predefined procedure exploring suspicions of tuberculosis. A high suspicion score of tuberculosis could drive the reasoned use of molecular testing in such settings.