
Image by billaday via Flickr
One of the many hats I sometimes wear is reviewer of submitted articles to a major medical journal. Maybe I’m a bit of a geek, but I take this role very seriously: I don’t want to be the guy who gave the thumbs-up to a study, only to have it be published and found to be complete garbage. By way of example, I’m sure the folks who reviewed the article that ultimately led to the downfall of a certain British doctor are feeling rather uncomfortable right now.
An area that I pay particular attention to is the statistics used in the study. A huge pet peeve of mine, which I’m sure you’ll recognize if you’ve read any of my previous posts, is the reporting of relative risk to make an intervention seem more beneficial than it actually is. Another is playing up the positive results while either downplaying the negative results, or ignoring them altogether.
A recent study suggests that my compulsions may be warranted.
More than half of 72 reports examined by French and British researchers had dressed up their conclusions to make it seem as if new treatments were beneficial, even though they weren’t according to the statistics in the report.
via Negative research often spun to look good: study | Reuters.
To me, this is incredibly disturbing. By allowing reports to be published with significant “spin”, we’re misleading the public and practitioners alike, sometimes with significant negative effects. For example:
When a clinical study was published in 2000 comparing Vioxx with an older drug, the authors downplayed the risk of heart attack: Instead of reporting the five-fold increase in risk, they chose to frame it as a protective effect of the older drug.
via Negative research often spun to look good: study | Reuters.
This is why I’ve instituted a curriculum in statistics within the residency program I oversee. Physicians need to be able to critically review studies and have the tools to analyze the conclusions of the researchers. We need to empower individuals to honestly question research, even published research in peer-reviewed journals.
Now if you’ll excuse me, I have an article to review.