Evaluating Syndromic Signals from Ambulatory Care Data in Four States W. Katherine Yih, 1 Candace Fuller, 2 Dawn Heisey-Grove, 3 John Hsu, 4 Benjamin A. Kruskal, 5 Michael Leach, 6 James Nordin, 7 Jessie Patton-Levine, 8 Ella Puga, 8 Edward Sherwood, 9 Irene Shui, 1 Richard Platt 1 1 Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, 2 Minnesota Department of Health, 3 Massachusetts Department of Public Health, 4 Kaiser Foundation Research Institute, 5 Harvard Vanguard Medical Associates, 6 San Mateo County Health Department, 7 HealthPartners Re- search Foundation, 8 Austin/Travis County health department, 9 Williamson County health department OBJECTIVE In this interim report we describe the signals detected by a real-time ambulatory care-based syndromic sur- veillance system and discuss their relationship to true outbreaks of illness. BACKGROUND The utility of syndromic surveillance systems to aug- ment health departments’ traditional surveillance for naturally occurring disease has not been prospec- tively evaluated. METHODS A previously described syndromic surveillance sys- tem based on ambulatory care data [1,2] has been in operation in four areas since February/March 2007. The locales (data-providers, health departments) are: greater Boston, Massachusetts (Harvard Vanguard Medical Associates, Massachusetts Department of Public Health); greater Minneapolis-St. Paul, Minne- sota (HealthPartners, Minnesota Department of Health); greater Austin, Texas (Austin Regional Clinic and Austin Diagnostic Clinic, Austin/Travis Co. and Williamson Co. health departments); and San Mateo County, California (Kaiser Permanente North- ern California, San Mateo Co. health department). Ten syndromes defined by ICD9 codes and similar to ones developed by a CDC-DoD working group are tracked. Signal detection is by the space-time permu- tation scan statistic available in SaTScan [3], using one year of historical data, a scanning window of 14 days, and a maximum circle size of 50% of the popu- lation. Health departments are automatically notified of signals with at least 3 cases and recurrence inter- vals (RI) exceeding 365 days. Data on the alerts are automatically collected and stored by the system’s data center; health depart- ments collect data on all clusters of infectious disease that come to their attention and provide these data to the data center via the system’s website in uniform databases developed by the Massachusetts Depart- ment of Public Health. RESULTS In the first 8 months of operation, 10 signals had re- currence intervals exceeding the 365-day threshold: State Syn- drome Ob- served Ex- pected # zip codes RI in days CA Hemor. 10 2 9 667 TX Neuro. 7 1 32 2,000 TX Up. GI 8 1 14 455 MN Hemor. 5 0 9 500 MN ILI 3 0 1 385 MA Asthma 5 0 5 370 MA Lesions 38 15 34 3,333 MA Neuro. 4 0 5 769 MA Rash 4 0 2 1,667 MA Up. GI 27 9 16 1,000 Diagnostic information was reviewed for the 10 sig- nals. None of the 10 corresponded to clusters of con- cern to the respective health department. The statisti- cally most unusual signal, lesions in MA, consisted of uncomplicated cases of tick bite. During the sur- veillance period, the Minnesota Department of Health registered 36 non-institutional GI clusters in the Twin Cities surveillance area, including several norovirus outbreaks with dozens ill and an outbreak of E. coli O157:H7 from a pig roast with 26 ill. Sensitivity and positive predictive value to date are zero or incalcu- lable. CONCLUSIONS During its first 8 months, this ambulatory care-based surveillance system has not detected clusters of con- cern to public health. Surveillance and evaluation will continue for one full year. REFERENCES [1] Platt R, Bocchino C, Caldwell B, Harmon R, Kleinman K, Lazarus R, Nelson AF, Nordin JD, Ritzwoller DP. Syndromic surveillance using minimum transfer of identifiable data: the ex- ample of the National Bioterrorism Syndromic Surveillance Dem- onstration Program. J Urban Health 2003;80(Suppl. 1):i25–i31. [2 Daniel JB, Heisey-Grove D, Gadam P, Yih WK, Mandl K, DeMaria Jr. A, Platt R. Connecting health departments and pro- viders: syndromic surveillance’s last mile. Morbidity and Mortal- ity Weekly Report 2005;54 (suppl):147-150. [3] Kulldorff M, Heffernan R, Hartman J, Assunção R, Mostashari F. A space-time permutation scan statistic for the early detection of disease outbreaks. PLoS Medicine, 2:e59, 2005. Advances in Disease Surveillance 2007;4:210