Using clinical data in BioSense to detect pneumonia cases

BioSense is a national program initiated by CDC and is intended to improve the nation's capabilities for conducting near real-time biosurveillance. More than 350 hospitals are currently sending chief complaint and ICD-9 discharge code data to BioSense. In addition, BioSense receives radiology text reports and microbiological laboratory data from more than 20 facilities.

Pneumonia is a major contributor to morbidity and mortality and several Category A Bioterrorism agents / diseases, such as plague, tularemia, and inhalational anthrax can manifest clinically as pneumonia. Also, pneumonia is a common manifestation of avian influenza. Thus, detection of pneumonia is an important function of BioSense. Most previous research on detecting pneumonia through administrative databases and natural language processing programs has been conducted as single institutions.

The purpose of our study is to determine the contribution of radiology text reports and patient chief complaints for monitoring pneumonia in BioSense. In study 1, selected chest radiograph reports will be evaluated for pneumonia by a clinician's reading, a simple key-word look-up program, and natural language processing programs. The clinician's reading will be performed using pre-defined criteria and will be used as the gold standard. In study 2, the radiology results will be compared to the gold standard derived from ICD-9 discharge diagnoses. In study 3, patient chief complaints will be compared with discharge diagnoses. These studies will help to understand the utility of BioSense data to detect pneumonia and of the BioSense system in monitoring a disease of major public health importance.