Geolocalization of Influenza Outbreak Within an Acute Care Population: A Layered-Surveillance Approach

We seek to use a novel layered-surveillance approach to localize influenza clusters within an acute care population. The first layer of this system is a syndromic surveillance screen to guide rapid polymerase chain reaction testing. The second layer is geolocalization and cluster analysis of these patients. We posit that any identified clusters could represent at-risk populations who could serve as high-yield targets for preventive medical interventions.

This was a prospective observational surveillance study. Patients were screened with a previously derived clinical decision guideline that has a 90% sensitivity and 30% specificity for influenza. Patients received points for the following signs and symptoms within the past 7 days: cough (2 points), headache (1 point), subjective fever (1 point), and documented fever at triage (temperature >38°C [100.4°F]) (1 point). Patients scoring 3 points or higher were indicated for influenza testing. Patients were tested with Xpert Flu (Cepheid, Sunnyvale, CA), a rapid polymerase chain reaction test. Positive results were mapped with ArcGIS (ESRI, Redlands, CA) and analyzed with kernel density estimation to create heat maps.

There were 1,360 patients tested with Xpert Flu with retrievable addresses within the greater Phoenix metro area. One hundred sixty-seven (12%) of them tested positive for influenza A and 23 (2%) tested positive for influenza B. The influenza A virus exhibited a clear cluster pattern within this patient population. The densest cluster was located in an approximately 1-square-mile region southeast of our hospital.

Our layered-surveillance approach was effective in localizing a cluster of influenza A outbreak. This region may house a high-yield target population for public health intervention. Further collaborative efforts will be made between our hospital and the Maricopa County Department of Public Health to perform a series of community vaccination events before the next influenza season. We hope these efforts will ultimately serve to reduce the burden of this disease on our patient population, and that this system will serve as a framework for future investigations locating at-risk populations.