Mad Science Monday, 9/7/2009

I'm under the weather today, so I'm going to keep this week short. This week's article also isn't "mad science," per se, but simply "science" that makes me mad. It also happens to be about drug studies, so I thought it was fitting to give it a look while I'm sick.

What makes me mad isn't so much the study, but that it gets worse every time it's passed through another filter on the web. Today's "Placebos Are Getting More Effective" headline on Slashdot drove me over the edge.

Placebos are not getting more effective. Several factors are combining to make the placebo effect larger compared to the "real" drug in the same studies, but it isn't that something magical is happening with placebos.

First, the studies are getting better. For example, imagine if you were studying a drug in the 1930s (in a world where 1930s researchers knew to do placebo-controlled studies), and this drug was supposed to decrease the incidence of lung cancer. You would create two groups, a placebo control group and an experimental group, making sure to balance for factors you expected to affect the results--age, gender, etc. By chance you might end up with more smokers in your control group than in your experimental group (because why bother controlling for that, if you don't think it has anything to do with cancer?). After your study, you'd likely find that your experimental group had a lower incidence of lung cancer, and thus that your placebo had very little affect compared to your drug. Of course, if you did that same study today, you'd be able to balance your groups for all kinds of known factors, including genetic risks for lung cancer, not just for the smoking bit. More and more, any improvement in your experimental group vs the random improvement of your placebo-controlled group would decrease, which you could choose to see as your placebo magically getting stronger. That's not what it is, though; you're just doing better science. See this great article over at Mind Hacks for more on this side of the effect.

Second, we're getting better at making placebos. We know strange things about human psychology, such as the wondrous bits in the graphic about half-way down the page on Wired's version of this news. We can make the placebo green in an anti-anxiety study, for example, because green pills work better for anxiety medicine (or we can at least make the placebo and the real drug the same color). That doesn't mean something magical is happening, either; it means we know how to harness psychology to boost the effectiveness of the pills, even if the medicine doesn't actually do anything beyond what the placebo does.

Third, the medicines being tested are, very often, just marginal improvements (or potential improvements) on existing drugs. We don't see as much of an effect because there isn't much of an effect to see.

So, if you see the headline I'm expecting this to morph into, something about placebos proving that medicine is unnecessary or some other similar nonsense, be sure to take it with a grain of salt. The pharmaceutical industry is still making improvements to our health, it's just doing so with better scientific practices.