Comparison of timeliness and correlates of three Influenza surveillance systems

Introduction

Several systems are currently used to detect the level and spread of influenza activity in the U.S. The effectiveness, completeness, specificity and timeliness of these systems, including new one such as FLUSTAR and googleflu, has not been fully examined.

Background

Three surveillance systems are routinely used to monitor Influenza trends in Pennsylvania – ILINet, NEDSS and RODS. A 2006 study of comparing ILI, NEDSS and RODS data showed that RODS and NEDSS data are significantly correlated. However, no study has examined the correlation, timeliness, specificity and completeness of these systems and the googleflu. In this study we examine the correlation of RODS, ILI and google flu data.

Methods

We analyzed and charted data for the 2008/2009 Influenza season to asses how the three systems compare in terms of Influenza activity, timelines and completeness of data.

Results

In multivariate analyses of preliminary data and after controlling for observed covariates, our data showed that RODS and NEDSS data trends and peaks are significantly correlated while ILI was not very correlated with either. Google flu was correlated with ILI data but not so much with RODS data. The timeliest data for predicting Influenza activity was RODS data with significant difference between googleflu and ILI.

Conclusion

Different strengths of data sets can be harnessed to make an informed picture of influenza activity. RODS has the advantage of assessing timely and large case reports while ILI and NEDSS has the advantage of being specific. Google flu can be used as a compliment to ILI.