Geographical Distribution Analysis of Influenza Data by Case Definitions Using Google EarthTM

The geographical distribution of influenza does not respect jurisdictional boundaries.  One of the difficulties of the present public health response system is that there is no clear agreement on how to effectively and meaningfully analyze cases geographically distributed across regions of interest, of varying sizes. Geographic Information Systems (GIS) can provide valuable insight into patterns of disease activity, from the small specific regions of interest to relatively large areas. Influenza comes with many different symptoms and while severity varies depending on the cases, it is typified and detected as an Influenza-like illness (ILI). It is important to investigate the distribution patterns of specific symptoms compared against the whole flu cases to understand what factors contribute different patterns. The authors are unaware of any prior studies investigating the distribution pattern of the influenza cases derived from individual or constellations of signs and symptoms. Google Earth^TM was utilized to display the 1,354 influenza cases diagnosed at the Mayo Clinic, Rochester, MN from 2000 to 2006 flu seasons, using valid zip codes by flu seasons and by individual symptoms. The GIS data for Cough, Fever, Sore Throat, and Dyspnea best represented the geospatial layout of the cases from the full flu season through visual inspection, and these representations were consistent with the result of distribution area analyses by ImageJ and Google Earth Pro^TM. This result also confirmed that these signs and symptoms were positive clinical predictors for defining influenza case definitions. Significantly high correlations were obtained between statistical rates of symptoms and both with the areas measured by ImageJ and Google Earth Pro^TM, which suggested distribution area method can be an encouraging analysis approach of influenza data. This would suggest that Sign and Symptom which appear earlier in the time course of an epidemic than diagnoses would be a reasonable substrate for biosurveillance.