Meet the Future of Crowdsourced Drug Studies

Those little side effect stickers pharmacies stick on pill bottles seem so quaint. Even that pamphlet listing even more side effects of and warnings about your prescription is becoming garbage, a thing to toss aside as you move toward the laptop ready to search the shit out of whatever chemical compound your doctor surely did not tell you much about, yet wants you to pay a lot of money for and put inside your body on a twice-daily basis.

A new study out from Microsoft Research and published in the British Medical Journal found that web browsing data can point to side effects of drugs well before those side effects make it to the side of a pill bottle, or are made public in any form from manufacturers and/or the FDA--or, possibly, before the manufacturers themselves are even aware of the side effects. And, in the process, providing data that is statistical, not anecdotal.

It's fairly well documented by now that doctors don't give their patients enough information, in general, about most anything in this here world of assembly line medicine. Some do, yes, but that seems to be more the exception than the rule. For those of us with a normal-ass doctor seeing patients at a rate of one every 10 minutes, there is fortunately the internet, full of a bottomless supply of good and terrible information.

The internet is still a fairly uncharted landscape when it comes to medicine, however, in terms of what it can and can't do, and should and shouldn't do. There is one aspect of providing information from some single source, but the internet also has vast capabilities for generating information. Message boards and things like askapatient.com are an example, but maybe a primitive example. These are, after all, just storehouses of anecdotal evidence.

In a way, web search information amounts to a massive-scale, uncontrolled drug study. The Microsoft study is based on web logs from millions of American internet users who had agreed to install a monitoring plug-in on their browser, which sent data back to the company's researchers (yeah, a disquieting prospect). They scanned for users searching for information on two drugs, the antidepressant paroxetine and a cholesterol lowering drug called pravastatin.

From there, the researchers looked at shared search terms for conditions and symptoms. 82 million queries were combed through (automatically) in total. Of the 6 million web users studied, one in 250 searched the name of a common drug. Which actually seems lower than one might imagine, but the researchers call "common." In any case, it was enough to find a common side effect symptom search between the two drugs: high blood sugar.

About one in 10 patients searching for both drugs went after high blood sugar. For a side effect, that's an impressive incidence. And it's important to note that this particular side effect wasn't public at the time of the study. It had been discovered, but that discovery of the interaction between the drugs was recent enough it hadn't been announced yet when the study subjects were doing their searches. Which doesn't do the patients affected in the long period between the drug's release and the discovery much good, but it's an interesting proof of concept: DIY drug safety.

It's more than drug companies having some conflicting interests when it comes to drug side effects. Side effects are by definition a highly uncontrolled thing, and there will always be something beyond what patients are told. Maybe you've spent some time in side effect hell, when how you're feeling doesn't match up to that list of stuff in pharmacy drug pamphlet. Turning to the internet is kind of the only option of confirming that this is doing that.

Which is a super-uncomfortable position to be in, relying on internet strangers for confirmation that something administered to make you well is actually making you feel worse. And, well, if you've combed a medical message board before, you're probably aware that a great many internet strangers are full of shit and super-paranoid. Search data parsing seems a way of legitimizing/verifying/confirming message board medicine. And another way to feel better.

Reach this writer at michaelb@motherboard.tv