Layout 1


ISDS Annual Conference Proceedings 2012. This is an Open Access article distributed under the terms of the Creative Commons Attribution-
Noncommercial 3.0 Unported License (http://creativecommons.org/licenses/by-nc/3.0/), permitting all non-commercial use, distribution, and
reproduction in any medium, provided the original work is properly cited.

ISDS 2012 Conference Abstracts

Comparing Prescription Sales, Google Trends and CDC
Data as Flu Activity Indicators
Avinash Patwardhan* and David Lorber

Clinical Affairs, Walgreen Co., Deerfield, IL, USA

Objective
To examine if the prescription sales data from a large retail phar-

macy chain in the US were comparable to Google Flu trends and
CDC’s US ILI Network data as flu activity indicator.

Introduction
In a 2007 survey of public health officials in the United States, In-

ternational Society for Disease Surveillance found that only 7% used
pharmacy prescription sales data for surveillance (1). There have been
many reports suggesting effective use of prescription sales data in
syndromic surveillance (2, 3, 4, 5). Community pharmacies can pro-
vide a valuable supplementary tool for syndromic surveillance of in-
fectious diseases.

Methods
We extracted five years of de-identified prescription sales data

from the proprietary pharmacy computer system of a large retail phar-
macy chain in the United States. The prescriptions were written for
the common drugs for the treatment of influenza: Amantadine, Os-
eltamivir, Rimantadine, and Zanamivir. We acquired Google Flu
trends national aggregate counts data that represented the estimates
of the ILI cases per 100,000 physician visits. We acquired CDC
ILINET data for 2007. We calculated Pearson ‘r’ between our data
and Google and CDC data. We also created comparable trends graphs
after converting the counts of the influenza scripts and the counts of
the Google estimated ILI cases to logarithmic scale.

Results
The Pearson ‘r’ between the aggregate counts of scripts for all the

four drugs and the Google estimates of the ILI cases for years 2007,
2008, 2009, 2010, and 2011 were 0.85 (95% CI, 0.75-0.91), 0.92
(95% CI, 0.86-0.95), 0.91(95% CI, 0.85-0.95), 0.88 (95% CI, 0.80-
0.93), and 0.87 (95% CI, 0.78-0.92) and 0.92 (95% CI, 0.90-0.94)
for years 2007 through 2011 together. The Pearson ‘r’ between the
aggregate counts of scripts and the CDC % unweighted ILI (2007)
was 0.97 (95% CI, 0.95-0.98).

Conclusions
A strong to very strong correlation between prescription sales data

and Google Flu trends and CDC’s ILI activities data for influenza
suggests that the former can serve as a good and valid influenza ac-
tivity indicator. Community pharmacies with large footprints might
want to proactively build automated syndromic surveillance systems
with near real time reporting capabilities and explore the possibility
of building predictive models (3) and of forming partnerships with
public health agencies for syndromic surveillance.

Pearson ’r’ between prescription sales data and Google Flu trends and CDC ILI
data

Keywords
public health; syndromic surveillance; Flu; prescription sales; activ-
ity or early warning indicator

Acknowledgments

We wish to acknowledge Criselda Valencia (Lead Analyst Support & De-
ployment Data Warehouse at Walgreen Co.), Erma Villarreal (IT Man-
ager - Support & Deployment Data Warehouse at Walgreen Co.), and
Zhongwen Huang (Senior Analyst Outcomes & Analytics at Walgreen
Co.) for pulling data for us.

References

1. The International Society for Disease Surveillance (ISDS) Available:
http://www.syndromic.org/ Accessed May 2, 2012. 2. Chadwick D
(2009) The Rhode Island Department of Health. The First Statewide
System for Tracking Disease Using Prescription Data. Press Release
Archives. State of Rhode Island. Department of Health. Available:
http://www.health.ri.gov/news/archives/2009/ Published October 27,
2009. Accessed May 2, 2012. 3. Sugawara T, Ohkusa Y, Ibuka Y,
Kawanohara H, Taniguchi K, Okabe N (2012) Real-time Prescription
Surveillance and its Application to Monitoring Seasonal Influenza
Activity in Japan J Med Internet Res. 14(1):e14. 4. Vergu E, Grais RF,
Sarter H, et al. (2006) Medication sales and syndromic surveillance,
France. Emerging infectious diseases. 12(3):416. 5. van den Wijn-
gaard C, van Pelt W, Nagelkerke N, Kretzschmar M, Koopmans M
(2011) Evaluation of syndromic surveillance in the Netherlands: its
added value and recommendations for implementation. Euro Surveill.
16(9):19806.

*Avinash Patwardhan
E-mail: avinash.patwardhan@walgreens.com

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * 5(1):e43, 2013