ABSTRACT
BACKGROUND: The performance of Xpert-MTB/RIF, including the newer Xpert Ultra test, for the diagnosis of pleural TB is poor (~28 - 38%). There are no data on patient characteristics that portend a positive Xpert-Ultra test in pleural fluid. These characteristics could be useful for selecting patients for Xpert-Ultra testing, thus maximising benefits of a positive test, while minimising cost. OBJECTIVES: To determine the clinical, radiological, microbiological and biochemical characteristics associated with Xpert-Ultra positivity in patients with suspected pleural tuberculosis. METHODS: We performed a subgroup analysis of a prospective observational cohort (N=165 patients with suspected pleural TB) evaluating same-day diagnostic tools for pleural tuberculosis. Forty-nine patients with confirmed pleural tuberculosis (culture and/or histology) were included in the final analysis. RESULTS: Of the 49 participants, 17 (35%) were female and 9 (18.4%) were HIV-infected. In the multivariate analysis including demographic, radiological and pleural fluid test characteristics, there were no independent predictors of Xpert-Ultra positivity. However, when pleural fluid test results were excluded, and when only rapidly ascertainable pre-test factors (demographic and radiologic variables) were considered, the multivariable analysis showed that only a chest X-ray (CXR) suggestive of active TB (cavities, consolidation and hilar lymphadenopathy) was associated with Xpert-Ultra positivity (p=0.021). Notably, only 22% (n=11/49) of the participants had a CXR suggestive of active TB and of these, 73% (n=8/11) had a positive Xpert-Ultra result. CONCLUSION: CXR features suggestive of active TB are significantly associated with a positive Xpert-Ultra test result on pleural fluid. These data inform clinical practice in resource-poor settings.
