Using data on an influenza B outbreak to evaluate a syndromic surveillance system--Israel, June 2004

• Published in: MMWR. Morbidity and mortality weekly report Vol. 54; no. 33; p. S191
• Main Authors: Shalev, Y, Aharonowitz, G, Dichtiar, R, Cohen, E, Bromberg, M, Marom, R, Peled-Leviatan, T, Shalev, V, Kaufman, Zalman, Havkin, O, Green, M, Lavi, C, Shemer, J
• Format: Newsletter Journal Article
• Language: English
• Published: Atlanta U.S. Government Printing Office 26.08.2005; U.S. Center for Disease Control
• Subjects: Health aspects; Influenza; Security systems industry
• ISSN: 0149-2195; 1545-861X

Introduction: Since 2002, as part of a national biologic terrorism preparedness program, the Israel Center for Disease Control (ICDC) has been developing a syndromic surveillance system based on data from community clinics and hospital emergency departments. Selected analytic tools are being evaluated for possible integration within this system. Objectives: This study evaluated the performance of the What's Strange About Recent Events (WSARE) algorithm for anomaly pattern detection when applied to records of daily patient visits to clinics of a local health maintenance organization (HMO). Methods: Data from an influenza B outbreak that occurred in June 2004 in an elementary school in a small (population: approximately 7,000 persons) Israeli town were used. WSARE searches for groups with specific characteristics (e.g., a recent pattern of place, age, and diagnosis associated with illness that is anomalous when compared with historic patterns). The data set used was limited to 1) patients living in the county where the outbreak occurred; 2) a 35-day period during May--June 2004; and 3) records containing International Classification of Diseases, Ninth Revision (ICD-9) codes for signs, symptoms, and syndromes associated with infectious morbidity. On average, the data set included 510 records/day. Besides ICD-9 codes, data included date of visit to clinic, day of week, city/town code, and patient's age. Results: Two successive significant anomalies (p<0.0001) were detected in the HMO data set that could signal the influenza outbreak, both sharing three constituents: 1) the town code; 2) the age category of affected children; and 3) the ICD-9 code for viral infection, which was the most prevalent diagnosis assigned by HMO physicians identified in an investigation by the regional health department. Had the data been available for real-time analysis, the first anomaly could have been detected on day 2, when the outbreak was first reported to public health officials. Conclusion: A centralized, comprehensive surveillance system can rapidly detect localized, fast-developing outbreaks. Although early detection is hard to achieve in this instance, timely and reliable information produced by syndromic surveillance is of great value in supporting outbreak management and placing it in the context of the background morbidity in the country. However, had the outbreak occurred in winter, detection would have been more complex. When the outbreak data were superimposed on winter background, only a single significant two-constituents anomaly was detected at day 2 of the simulated outbreak, lacking the information to target on the specific age group of the schoolchildren.