November 16, 2008

Deconstructing Scientific Articles

I spent some time this evening reading thoughtful, well-reasoned critiques of a few recent medical studies. On his blog, Genomics, Evolution, and Pseudoscience, Steven Salzberg describes 5 problems about the recent report that Cresor can result in "a 'dramatic risk reduction' in heart attack risk for men."

In Serious doubts about new study of statins and heart disease, Salzberg summarizes the studies: "[they] claimed that people with normal cholesterol levels could get significant health benefits [by taking Crestor]. If true, [these two studies] impl[y] that millions more people should start taking statins to protect themselves against heart attacks." He adds "[t]his new finding is rife with problems, despite the breathless news reporting about it" and goes on to describe 5 of them:
  1. "Both studies were funded by AstraZeneca, the drug company that sells Crestor," although Salzberg is quick to add that this is clearly disclosed in both articles.
  2. The lead author of both studies is Paul Ridker, who owns the patent on the primary test for C-reactive protein (CRP) levels, and he stands to benefit financially if more people are tested for CRP. Again, Salzberg calls this only eyebrow-worthy, as this is clearly stated in both articles.
  3. Although it suggests that a seven-variable method is more predictive than the traditional five-variable model, "the Circulation study didn’t report separately on the effect of CRP and family history of heart disease."
  4. "[T]he NEJM [New England Journal of Medicine] study actually reports a very small benefit: ... you’d have to treat 95 people for 2 years with statins to prevent 1 heart attack."
  5. "The patients in the NEJM study were randomly divided into two groups, treatment (Crestor) and placebo [and] there are 3 critical variables where the two groups are not identical."
It's a fascinating read, because Salzberg knowledgably disputes the articles' claims (and includes a few snarky notes about how the media covered the story).

If you like that post, you might like these:I like these posts because they debunk bad science ... and I also humbly suggest that they may be helpful in teaching science students how to evaluate scientific articles.

For the socially networked: I found this blog through my FriendFeed buddy Jean-Claude Bradley, who follows Lars Juhl Jensen, who linked to Salzberg's blog.

No comments: