Google searches: Real health behaviors in real time

Did the "S-CHIP tax" really motivate people to quit smoking? (Photo: AaronC/Flickr)

In April 2009, the U.S. federal cigarette excise tax was raised from $0.39 to $1.01 per pack as part of Congress’s reauthorization of the State Children’s Health Insurance Program. This well-intentioned “SCHIP tax” was meant to encourage people to quit smoking.

But a different story emerges from an analysis of Google searches.

Searches related to smoking cessation immediately rose 50 percent — but were back at the baseline level within weeks. Meanwhile, searches for tax-free online cigarettes rose dramatically, increasing 300 percent over baseline when the tax was first implemented. Search volume remained almost 60 percent higher throughout the subsequent year.

Florida provides a dramatic example. Here are results of a search on searches for “online cigarettes” from 2008 through 2010. The first 2009 peak coincides with enactment of the SCHIP tax; the second dovetails with a state cigarette tax increase, enacted on July 1.

This study doesn’t merely show the power of addiction, or the power of the Internet in subverting public policy. It also demonstrates the power of tapping into Google searches to learn more about people’s health behaviors – quickly and in real time.

“It’s behaviors that kill people,” says John Ayers, of the Children’s Hospital Informatics Program (CHIP), who led the study in collaboration with Kurt Ribisl of the University of North Carolina at Chapel Hill. “Traditional surveillance for health behaviors only happens annually – you get this one snapshot every year. Now we can get real-time measurements that are sensitive to changes in the environment.”

You can do this yourself. The Google Insights tool makes it possible for just about anyone to practice health behavioral surveillance. “Others can easily replicate our findings with minimal statistical training,” says Ayers, who is now a doctoral candidate at Johns Hopkins. “It’s a technique that any researcher could use within a couple of weeks.”

Ayers and John Brownstein, also of CHIP, have just completed another study correlating economic factors like foreclosure rates and unemployment with online searches about mental distress. Here’s my own search – depression symptoms, United States, 2004-present.

Searches can easily cross international boundaries. Last year, Brownstein and Ben Reis of CHIP found that states and countries with lower abortion rates and less permissive abortion policies had higher search volumes for the word “abortion.” (The converse was also true. )

Real-time intervention

There’s a great benefit to being able to capture behavioral information on the Internet: You get a chance to try to change it.

“The fact that we can identify people searching allows us to reach them and link them to evidence-based programs that are already online,” Ayers says.

So if you Google for tax-free cigarettes, ads for smoking cessation programs could pop up on your Search Results page. Or if you’re Googling to avoid taxes on sugar-sweetened beverages, you’d see ads for weight-loss interventions.

Take that smoker who really does want to quit. Currently, when smokers search on “cessation,” they’re apt to click on a fad approach, like hypnosis or electronic cigarettes, Ayers has found. But maybe an effective ad could redirect smokers to medically proven interventions.

Or maybe people Googling “quick weight loss” could be steered to more gradual approaches that include exercise (any guesses about that pronounced peak in August 2009 below?)


Because Google advertising is relatively inexpensive and quick to implement and monitor, it’s easy to test what kinds of ads are most effective and what links motivate users to actually change their behaviors. This spring Ayers and Brownstein hope to do just that.

”It’s a whole new world for health research,” says Brownstein, who, in full disclosure, receives some grant support from Google.org. “You have zero cost, you’re not fighting for data, you have no human-subjects issues, and you have millions of data points at your fingertips instantaneously as events unfold. Within seconds you can get an initial result from populations you’d never be able to survey.”