Does Google Know When You’re Sick? A Penn Study Counts the Clicks

Anyaberkut/Getty Images.

We’ve all done it: You come down with a cough or you’ve got a never-ending stomach ache, so you head straight to WebMD and search for your symptoms. You shop around for a few maladies, and suddenly you’re reading about the worst-case-scenario disease and asking yourself, Could I really have bubonic plague?

Knowing this tendency — 72 percent of all American adults use Google to search health-related questions — Jeremy Asch and a team of researchers at Penn Medicine set out to study internet search trends from patients who ended up in Penn’s emergency room. The team asked more than 700 patients to share their recent Google search data with the team — about 100 of whom said yes. The researchers pored over the data after the fact, like benevolent Big Brother, finding, among other things, that 53 percent of the subjects searched for information about the chief complaint that eventually brought them to the hospital. Asch and I spoke over the phone about the study’s findings, its implications for future healthcare, and the logistical challenges of using search results to predict ER visits.

From your vantage point, is it necessarily a good thing that nearly three-quarters of Americans are turning to Google for health questions? There’s a lot of misinformation out there.
I’m not sure if it’s a good or bad thing. Our goal here isn’t to dissuade people from doing things that they’re already going to do. I don’t think that we’re ever going to stop people from looking for more information online. But what we can do is try to figure out what those questions were that they wanted to know outside of the hospital environment and try to answer them in a different way. No matter what information we’re going to give patients, they’re still going to go home and Google questions to try to find more information.

One of the main findings in your study was that among emergency room visitors, the frequency about health-related queries went from about 6 percent of all searches to 15 percent in the week prior to the ER visit, which strikes me as very self-evident: People are sick and then they show to up to the ER eventually.
I think there is kind of that initial reaction to think, Well, of course. People are going to go to the hospital and because they’re having some sort of condition, that’s going to increase the amount of questions they’re having online. We can maybe look at that data to find more information about what questions a patient is asking before they make the decision to come to the hospital.

This isn’t what you’d consider a representative sample given that it’s one particular emergency room in one environment. But can you extrapolate any findings from this beyond what you have for these hundred something people?
What I’m curious to see is, now that we know patients are searching for these things beforehand, what are the questions that they’re asking? Is there some way that we can anticipate that need and say that patients who come in with this sort of condition to the hospital tend to ask these questions about their care before they come in?

How would that work in actual practice? That would require Google providing you search results for everyone in the Philadelphia area Googling about their health concerns. How realistic is that?
So, that’s not really the avenue. We’re not going to ask these patients in order to collect their data in real-time all the time. But ways that we can provide that information for them and say, “If you came in with this, were you having these symptoms?” Maybe there’s some sort of alert set up like, if you Google these symptoms, maybe this is time to go to the hospital.

I was going to ask you specifically if you envisioned a future where you Google a symptom and then you get a notification on your phone about considering going to the ER. Is that one of the applications you’d see from this research?
I don’t really see it happening in that way. I don’t know Google is the way that’s going to happen.

Hospitals tend to create the narratives themselves and say, “Here are the questions patients normally ask. And here’s the information we’re going to give you.” But after that’s all done, what are the questions you actually wanted to know? What didn’t you remember? Throughout your daily life, what are the questions that arise that you feel need to be answered? Figuring out what those are is just the first step.

This conversation has been edited and condensed for clarity.