The broad, longterm objective of this project is to improve the allocation of scarce tuberculosis (TB) isolation beds for patients who present to a hospital Emergency Department (ED) with suspected TB. By providing an accurate classification of sputum acid-fast bacilli (AFB) positive or negative TB from the ED, we would decrease the costs of construction and management of in-hospital isolation bed units, and decrease the risk to health care workers (HCWs) and other patients to TB transmission. We hypothesize that the recommendation for isolation or no isolation after a 12 hour encounter in an Emergency Department Observation Unit (EDOU) is more accurate than routine ED recommendation, and has high concordance rates with the more prolonged inpatient diagnostic assessment period, during which the patient is maintained in TB isolation. The first specific aim is to compare the sensitivity, specificity, positive predictive value and negative predictive value of direct microscopic detection of AFB when performed on 3 serial sputum samples obtained within 12 hours in an EDOU versus 3 early morning sputum samples obtained during consecutive hospital days, The second specific aim is to determine the accuracy of a standardized clinical assessment obtained in the ED that considers patient history, risk factors, physical exam and chest radiograph at presentation in predicting the diagnosis of active pulmonary TB. The third specific aim is to compare the sensitivity and specificity of the diagnosis of TB in the EDOU versus In-Patient Isolation Unit (IPIU) assessing the diagnostic capability of the EDOU as a rapid identifier of new TB cases allowing more precise indications for hospitalization and the use of TB isolation beds. We will evaluate independent variables obtained in the EDOU to examine which are most closely associated with either smear positive or culture positive TB. We will perform multivariate logistic regression data in order to develop a model for diagnosis of both culture proven and infectious TB.
