Comparing Flu Surveillance: Google Trends vs. CDC Data
Author Information
Author(s): Justin R. Ortiz, Hong Zhou, David K. Shay, Kathleen M. Neuzil, Ashley L. Fowlkes, Christopher H. Goss
Primary Institution: University of Washington
Hypothesis
How does Google Flu Trends data compare with national surveillance data for laboratory-confirmed influenza infections?
Conclusion
Google Flu Trends is highly correlated with rates of influenza-like illness but less so with laboratory-confirmed influenza cases.
Supporting Evidence
- Google Flu Trends estimates were highly correlated to CDC surveillance for ILI, with a mean correlation coefficient of 0.97 during the 2007–08 influenza season.
- The correlation between Google Flu Trends and CDC Virus Surveillance was 0.72, indicating a moderate relationship.
- Most influential outlier observations occurred during the 2003–04 influenza season, affecting the correlation results.
Takeaway
This study looked at how well Google searches for flu symptoms match actual flu cases. It found that while Google can give a good idea of flu trends, it's not as accurate for confirmed cases.
Methodology
Data from Google Flu Trends and CDC surveillance systems were compared using Pearson's correlation coefficients over several influenza seasons.
Potential Biases
Potential biases include variations in physician testing practices and patient health-seeking behavior during different influenza seasons.
Limitations
The study relies on convenience samples and may be affected by changes in health-seeking behavior and internet search patterns over time.
Statistical Information
Confidence Interval
95% CI: 0.64, 0.79
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website