Assessing the District of Columbia Syndromic Surveillance Systems during the 2009 H1N1 Outbreak

BACKGROUND:  The 2009 H1N1 influenza pandemic served as a reminder of the importance of a robust disease surveillance system to monitor the progression of outbreaks to provide a timely response.  The District of Columbia had several systems in place during the outbreak, including the Electronic Surveillance System for the Early Notification of Community Based Epidemics (ESSENCE), and DC Sentinel Surveillance.  This report assessed the performance of DC surveillance systems against one another as well other regional and national disease surveillance systems during the H1N1 outbreak.  

METHODS:  DC ESSENCE, ANCR ESSENCE, DC Sentinel Surveillance, and CDC ILINet data were extracted for the period of March 28, 2009 to March 27, 2010 to generate percent of ED or clinic visits for influenza-like-illness (ILI), defined as having a fever and cough and/or sore throat.  For the regional ANCR ESSENCE, percent of OTC pharmacy sales for thermometers along with percent of sales for flu/cold OTC medication were also generated.  For Google Flu Trends the number of flu related searches per 100,000 physician visits was extracted for the national and DC levels.  Percentages were compared using Pearson’s correlation and root mean squared error (RMSE) to assess fit of the various curves generated by weekly percentages across systems.

RESULTS:  Percent ILI from DC ESSENCE was found to be most strongly correlated with the trends from regional ANCR ESSENCE, DC Google Flu Trends, and regional CDC Regional ILINet data with correlations of 0.9868, 0.91359, and 0.92557 respectively, and RMSE of 0.39209, 0.98835, and 0.92016 respectively.  the DC sentinel surveillance system did not even remotely resemble the trends of all the other systems, with a correlation of 0.07531 and RMSE of 3.20195, and was outperformed by even thermometer sales, with a correlation of 0.88988 and RMSE of 1.10882.  

CONCLUSIONS:  All assessed DC systems aside from DC Sentinel Surveillance were able to track ILI activity as well as other regional and national systems.    The lack of sentinel sites means that non-reporting significantly impacts quality of data, and demonstrates the need for strong data sources for disease surveillance and the use of multiple systems to complement one another.