The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #69636

Rebecca Roberts, MD¹, Robert James Rydman, PhD², Brandon Zagorski, MPH³, Pamela Diaz, MD⁴, and Nicole Cohen, MD⁴. (1) Department of Emergency Medicine, Stroger Hospital of Cook County, 1900 West Polk, 10th Floor, Chicago, IL 60612, 312-633-3225, rroberts@ccbh.org, (2) Department of Emergency Medicine, Cook County Hospital/Rush University, 1900 W. Polk St. 10th floor, Chicago, IL 60612, (3) Dept Emerg Med, Cook County Hospital, 1900 W. Taylor Street, 10th Floor, Chicago, IL 60612, (4) Chicago Department of Health - CDC, Ogden and Taylor, Chicago, IL 60612

Background: Ideally, regional surveillance requires pooling of disease or syndrome rates from multiple EDs. These sources capture different data elements that might not measure similar events. We tested empirically derived syndromic algorithms for detecting West Nile Virus encephalitis as part of a CDC funded B2 project. Correlations between different algorithms were measured to assess the feasibility of combining disparate data. Methods: Successive limited comparisons were performed using electronic data from an urban public hospital with 120,000 ED visits. Data included complaints, vitals, test results, diagnoses, and disposition. Algorithm #1 was based on literature review and expert opinion from the Chicago DPH. To decrease noise from mild illness, lumbar puncture (LP) served as a surrogate for physician assessment of high illness severity. Algorithm #2 was based on factors empirically associated with diagnoses of, or LP results consistent with viral meningitis, or encephalitis, but where LP was deemed unwarranted. Algorithm #3 was based on inconsistencies between clinical findings and diagnosis reflecting undiagnosed encephalitis. Algorithm #4 was hospitalization for encephalitis, meningitis, or fever with headache. Results: 332,527 visits were analyzed (Jan 2000 - Dec 2002). All algorithms detected significant increases in case rates in late summer in all 3 years and were chronologically correlated over the 3-year time course; (p < 0.01). Conclusions: Multiple surveillance algorithms based on chief complaints, vitals, discharge diagnoses or test results correlated well over time in our ED. Future studies are warranted to evaluate the performance of pooled data from different ED surveillance systems for detection of regional events.

Learning Objectives:

Participants will learn:
• how to use empiric data to create multiple surveillance algorithms for syndromes that frequently escape detection.
• methods for combining published data, expert opinion, and empiric results to enhance syndrome surveillance efforts
• how to evaluate different methods of syndrome surveillance

Keywords: Data/Surveillance, Emergency Department/Room

Presenting author's disclosure statement:
I have a significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.
Relationship: This work was done in collaboration with the Chicago Department of Public Health, and was funded by a subcontract with them. The CDC B2 Bioterrorism projects provided the funding.