Objectives: This study examined whether data routinely available in emergency departments could be used to improve isolation decisions for tuberculosis patients.
Methods: In a large emergency department in New York City, we compared the exposure histories of tuberculosis culture-positive and culture-negative patients and used these data to develop a rapid decision instrument to predict culture-positive tuberculosis. The screen used only data that are routinely available to emergency physicians.
Results: The method had high sensitivity (.96) and moderate specificity (.54).
Conclusions: The method is easily adaptable for a broad range of settings and illustrates the potential benefits of applying basic epidemiologic methods in a clinical setting.