793. Diagnostic Accuracy of Single vs. Multiple Gene Xpert for Discontinuation of Airborne Infection Isolation in Suspected Pulmonary Tuberculosis Patients at a US Safety-Net Hospital

BACKGROUND: Patients suspected to have pulmonary tuberculosis (PTB) undergo serial sputum analysis under airborne infection isolation (AII). The US FDA approved the Cepheid GeneXpert-MTB-Rif® to support removing patients from AII. The FDA requires that “either one or two” separate sputum specimens be examined. To clarify this statement, the National Tuberculosis Controllers’ Association and the Association of Public Health Laboratories published guidelines that recommend that two sputum specimens be used and recommend that each institution examine their own data to determine whether one specimen is sufficient. Most patients in low prevalence settings do not have PTB yet are tested several times; an optimal testing strategy will reduce unnecessary isolation and related expenses. We sought to determine the diagnostic accuracy of a single vs. two sputum samples for Xpert MTB/RIF in discharging suspected PTB patients from AII. METHODS: Retrospective review of patients admitted between September 2016 to January 2018 was undertaken to identify sensitivity, specificity, positive and negative predictive values, for MTB gene Xpert in comparison to Mycobacterial culture as the gold standard. We further analyzed whether a larger number of such tests improved diagnostic yield for PTB. RESULTS: One hundred seventy-one patients, 17.5% of whom were HIV+, mostly of non-US origins (64%), provided 312 samples for Xpert MTB/RIF, of which 26 were Xpert-positive. These 26 samples came from 15 patients, 14 of whom were diagnosed using the first sample tested with Xpert MTB/RIF. Sensitivity and specificity of the first sample tested with Xpert MTB/RIF were more than those for the first two samples considered together or for all tested samples. Of these 15 positive cases, 13 were confirmed on sputum culture; 10 were positive from the first, one from the second, and two from the third sputum samples cultured. CONCLUSION: Patients suspected to have PTB at our facility can be rapidly and accurately discharged from AII after testing a single sputum sample for MTB/RIF Xpert. DISCLOSURES: All authors: No reported disclosures.