ISOLATION OF PATIENTS WITH SUSPECTED TB AND IMPLICATIONS OF CLINICAL PREDICTION SYSTEMS AT A UNIVERSITY HOSPITAL IN BROOKLYN

Abstract

SESSION TITLE: TB or Not TB: That Is the Question: Better Ways to Diagnose Tuberculosis SESSION TYPE: Original Investigations PRESENTED ON: 10/22/2019 10:45 AM - 11:45 AM PURPOSE: Patients admitted to hospitals, with elevated clinical suspicion for active pulmonary mycobacteria tuberculosis (PTB) infection are routinely placed on respiratory isolation to prevent the spread of the disease, especially to other patients in accordance with Center for Disease Control (CDC) guidelines. The increased concern for nosocomial transmission led to aggressive respiratory isolation for suspected cases resulting significant, as much as 92-fold, overuse of isolation rooms. Many investigators have developed prediction systems for PTB based on clinical and radiographic findings, though many are not validated in external populations. METHODS: We analyzed the isolation practices for a university hospital in Brooklyn (SUNY Downstate) serving an immigrant population and assessed how the implementation of various prediction scoring systems for PTB would improve our practice. Our current policy regarding isolation is based on individual clinicians’ judgements. We retrospectively analyzed all adult patients placed on respiratory isolation for suspicion of PTB from 2016-2017. Patients who were treated for culture negative TB, presented with a diagnosis of PTB from another facility, or treated for purely extrapulmonary TB without suspicion of pulmonary involvement were excluded from the analysis. RESULTS: Our current clinical practice results in 97% of patients being place on isolation necessarily. Patients with culture proven PTB had with greater frequency cough (69 vs 50%), hemoptysis (24 vs 12%), subjective fevers (55 vs 36%), weight loss (76 vs 24%), night sweats (34 vs 13%), and upper lung field disease on chest X-ray (69 vs 19%). They tended to be older (59 vs 48 years-old) and more often born abroad (76% vs 20%). Acid fast bacilli (AFB) were present on at least one sputum smear in 72% of patients and present on the first sputum specimen in 52% of patients with culture proven PTB. Two clinical prediction models demonstrated 100% sensitivity (Academic Emergency medicine 1997 and Archives of Internal Medicine 2000) while three more had greater than 90% sensitivity in our population. Implementation of one of the clinical prediction models into clinical practice would spare patients more than 100 days per year of unnecessary respiratory isolation. CONCLUSIONS: Clinical and radiographic based prediction systems may be used to supplement clinical judgement and assist in determining patients at high risk for pulmonary tuberculosis for the purposes of respiratory isolation. While our data is suggestive of potential benefits for reducing the burden of isolation of patients and hospitals, clinicians should carefully weigh the risk of missing active pulmonary tuberculosis as the public health consequences may be serious. CLINICAL IMPLICATIONS: Implementation of clinical prediction models for assessing the risk of tuberculosis may save many patients from unnecessary respiratory isolation. DISCLOSURES: No relevant relationships by Bryan Chesen, source=Web Response No relevant relationships by Joshua Davidson, source=Web Response No relevant relationships by Samir Kumar, source=Web Response No relevant relationships by Daniel Shayowitz, source=Web Response