This great post from the folks over at The Incidental Economist reviews a paper from the Journal of General Internal Medicine (PMID: 21792695), whose punchline is as follows:
For both physicians and patients who are trying to evaluate the mortality benefit of two drugs with absolute mortality rates of 6% (“old” drug) vs. 4% (“new” drug), the absolute mortality benefit is the most accurate measure, relative mortality reduction is the most persuasive (in favor of the “new”, lower-mortality drug) and absolute survival benefit blurred any difference between the two drugs.
What does that mean for clinical practice? If you’re a drug rep, always present data in terms of the largest relative difference in outcomes. If you’re a skeptic about new drugs, present the smallest relative difference (since 94% survival vs. 96% survival was viewed as “no difference” but 4% mortality vs. 6% mortality is viewed as different).
But what if you’re just an average everyday doc? How do other students handle this in their clinical settings?
[This is not news: Kahaneman and Tversky z”l covered this 30-40 years ago, and much of current behavioral economics deals with variations on this theme.]
What I still don’t know, after much reading, is what the hell I should do about it? I do always strive to present absolute risk reduction numbers because personally I think those are the only easily understandable numbers. Your thoughts, oh mighty readership (I think there are 2-3 of you out there), would be most welcome …