Olson DR, Viboud C, Simonsen L.: Reassessing google flu trends data for detection of seasonal and pandemic influenza: a comparative epidemiological s

[tetano]

PLoS Comput Biol. 2013 Oct;9(10):e1003256. doi: 10.1371/journal.pcbi.1003256. Epub 2013 Oct 17.
Reassessing google flu trends data for detection of seasonal and pandemic influenza: a comparative epidemiological study at three geographic scales.
Olson DR, Konty KJ, Paladini M, Viboud C, Simonsen L.
Source

New York City Department of Health and Mental Hygiene, New York, New York, United States of America.
Abstract

The goal of influenza-like illness (ILI) surveillance is to determine the timing, location and magnitude of outbreaks by monitoring the frequency and progression of clinical case incidence. Advances in computational and information technology have allowed for automated collection of higher volumes of electronic data and more timely analyses than previously possible. Novel surveillance systems, including those based on internet search query data like Google Flu Trends (GFT), are being used as surrogates for clinically-based reporting of influenza-like-illness (ILI). We investigated the reliability of GFT during the last decade (2003 to 2013), and compared weekly public health surveillance with search query data to characterize the timing and intensity of seasonal and pandemic influenza at the national (United States), regional (Mid-Atlantic) and local (New York City) levels. We identified substantial flaws in the original and updated GFT models at all three geographic scales, including completely missing the first wave of the 2009 influenza A/H1N1 pandemic, and greatly overestimating the intensity of the A/H3N2 epidemic during the 2012/2013 season. These results were obtained for both the original (2008) and the updated (2009) GFT algorithms. The performance of both models was problematic, perhaps because of changes in internet search behavior and differences in the seasonality, geographical heterogeneity and age-distribution of the epidemics between the periods of GFT model-fitting and prospective use. We conclude that GFT data may not provide reliable surveillance for seasonal or pandemic influenza and should be interpreted with caution until the algorithm can be improved and evaluated. Current internet search query data are no substitute for timely local clinical and laboratory surveillance, or national surveillance based on local data collection. New generation surveillance systems such as GFT should incorporate the use of near-real time electronic health data and computational methods for continued model-fitting and ongoing evaluation and improvement.

PMID:
24146603
[PubMed - in process]

Free full text

http://www.ncbi.nlm.nih.gov/pubmed/24146603

 

[Vibrant62]

Re: Olson DR, Viboud C, Simonsen L.: Reassessing google flu trends data for detection of seasonal and pandemic influenza: a comparative epidemiological study at three geographic scales

I do have one observation here

If we have a virus of a low incidence but a high severity, a more severe picture will emerge at a public health monitoring level as more individuals will seek medical care. in other words, search trend volumes could be low as the absolubte number of people infected could be low, but a high proportion of those infected could be severely ill and need medical help. This could be why the first wave of pH1N1 was 'missed'? But perhaps not.

If we have a virus of high incidence but low severity, more people will self treat (search the internet for information) and not seek a doctor consultation and therefore fewer people will appear in the public health data relative to search trends.

The only way to resolve this potential for discrepancy would be random sampling of blood samples collected for other health testing each year to look for recent influenza infection (reterospective seroconversion study) and seeing if there is a greater correlation with public health data or search trends (or somewhere in between).

Public health needs data to show the potential burden on a flu outbreak on the health care system, but this is always going to be a multiplier of incidence x strain severity, and there still isnt anything that can accurately predict this - its a simple case of managing the patients that turn up at the door.

I still wonder if search data could be most accurately used for determining a baseline incidence of flu (i.e a denominator value), and that linking this to public health data might give an index of severity: i.e a search trend volume divided by a public health index might give a measure of relative virus severity. I am sure someone will do a trial at some point, and it would be interesting to see what the results are.