Tuesday, August 25, 2009

The Wikipedia Model Does Not Apply For Medicine

Having some limited background in the biomedical sciences (though I consider myself first a chemist), I’ve grown tired of folk remedies that I can understand have little to no plausibility. Sometimes I’ll humour the person who suggests a course of action to me, but for the most point I will ignore anecdotal recommendations in favour of real medicine. It often frustrates me to see people who believe themselves (or sell their image as) more educated on health than medical professionals.

That’s why, when I read the article, Ask Strangers for Medical Advice, coming from a pop science outlet (Wired), I just about shat a brick. The article plugs a site called CureTogether, which allows hypochondriacs everywhere to track their progress on combating whatever ails them.

If all participants were honest, diligent, and meticulous, with some way of verifying the information reported, then the idea of CureTogether could work. The problem is, there is no way to track if what people report as results is actually what is going on. Most people are not meticulous. Many are not diligent. And there’s probably enough dishonest people out there to make this site’s information next to useless.

Even the image provided by Wired highlights the failings of this method:


The chart tracks various methods used to combat acne. The credulous eye sees that “Dr. Hauschka Natural Skin Care” seems to do the best job, but ignores the fact that this value is based on only two respondents. Without the source data I can’t be sure, but the data for this point could probably fall pretty much anywhere by chance (i.e. this likely isn’t a significant result). That doesn’t stop the fact that most people will now take away the message that this is the “best” treatment. And the remedy gets free publicity from the Wired article, as well, which it does not deserve based on this data.

The author writes:

Even if some bad apples make their way into the community, it may still be a better source of information than some peer-reviewed literature, since top scientists have been caught fabricating data about medications and Elsevier has published entire fake journals dedicated to bolstering the reputation of Merck drugs.

Which I find offensive. Yes, drug companies can be bastards. But that doesn’t mean that their information is useless. The peer-review process is in place to reduce this kind of dishonesty, and it does, by replicating and revising information when new results come to light.

Wired reports this issue as science failing the common person, and so they appeal to the experience of the crowd. The problem is that one person’s opinion does not constitute proof, and combining many opinions only amplifies any errors in judgement. The plural of anecdote is not evidence. Shame on you, Wired Science.

Wired Science: Ask Strangers for Medical Advice

1 comment:

marsha said...

This is actually a little scary. I can't wait until Colbert gets his nation on this, and has them all input that drinking bleach to cure a bladder infection is effective.

God, people are stupid. Know what helps the common person? Statistically sound experiments. Jesus. Good work Shane.