Unreported Side Effects of Drugs Found Using Internet Data, Study Finds | Hacker News Reader

Unreported Side Effects of Drugs Found Using Internet Data, Study Finds

nytimes.com

58 pointserickhill13 comments

I have hung around a few health related forums and left a few in disgust. The echo chamber for unproven alternative remedies, untested theories and conspiracy theories totally dominates over rational discussion of evidence based medicine. I am afraid trawling this data is just going to tell practitioners that vaccines cause autism, lots of stuff about crazy diets and detoxing and the importance of organic and non-gmo produce. And that would be the least crazy stuff.

I agree about the poor data, but I don't think this is an issue chiefly because you are trawling for an untested hypothesis, which you can then test using more robust methods. This is more about what to investigate, which is nearly infinite in theory. If this points you in the right directions, all the better..

The signal is probably stronger in what they search for than in what they read or post in forums (which far fewer people do than search for things, anyway).

Exactly! I imagine they have some way of filtering this out that NYtimes hasn't gone into. But who knows?

I am a bit concerned about how this will play out with the various drug conspiracies. For example many many people search up antivaxx literature under things like "vaccines autism". How do they distinguish real results from popular myth results? The two drugs they mention may very well have the negative interaction that they suggest. But, how do they distinguish this from say if a site like Natural News were to publish some crazy idea about a drug causing a spike in people searching for the terms together?

Does anyone have the link to the original article? I can't find it on the JAMIA website, and there is no link on the NYT site either

I'm assuming the original research they're building on is this: http://jamia.bmj.com/content/19/1/79.abstract

However, I'm not able to figure out what the newest article in this month's journal is, based on the most recent Table of Contents.

And even so, the article (including the original) is behind a paywall as per usual. :/

Does anyone know what the search toolbar is they're using? The article claims they tracked 82 million drug related searches.

The most interesting question to me: how can I get the dataset they used.

I like this article, but I just want to point something out

   "With the aid of Microsoft researchers, he was able to acquire
    anonymized data taken from a software toolbar installed in Web 
    browsers by users who allowed their search histories to be 
    collected."

I'm thinking the users who consented to having their search histories exposed were like how date rape victims "consent" to ruffies. Hey they drank from the glass right?

This is not kosher.

Thanks for the wildly inappropriate analogy.

No, this is like allowing Google to collect your location information and use it to show traffic data on their Maps. From your own quote: by users who allowed their search histories to be collected.

If you have ever wondered what Microsoft does with toolbar query data, I suggest you search for the term "hiybbprqag"

Reply on news.ycombinator.com