Hospitalized persons with suspected pulmonary tuberculosis (PTB) are placed in airborne isolation to prevent nosocomial infection, as recommended by the Centers for Disease Control and Prevention (CDC). There is significant evidence that clinicians overuse this resource due to an abundance of caution when confronted with a patient with possible PTB. Many researchers have developed predictive tools based on clinical and radiographic data to assist clinicians in deciding which patients to place in respiratory isolation. We assessed the isolation practices for an urban hospital serving a large immigrant population and then retrospectively applied seven previously derived prediction models of isolation of PTB to our population.Our current clinical practice results in 76% of patients with PTB being placed in isolation on admission. However, 208 patients without PTB were placed in isolation unnecessarily for a total of 584 days. Four models had sensitivities greater than 90%, and two models had sensitivities of 100%. The use of these models would have potentially saved more than 150 days of patient isolation per year.

No abstract

No abstract