In an editorial about the effect of vitamin-mineral supplements in the prestigious American Journal of Clnicial Nutrition, the author, Donald McCormick, a professor of nutrition at Emory University, writes:
This study is a meta-analysis of randomized controlled trials that were previously reported. Of 2311 trials identified, only 16 met the inclusion criteria.
That’s throwing away a lot of data! Maybe, just maybe, something could be learned from other 2295 randomized controlled trials?
5 Replies to “The Emperor’s New Clothes: Meta-Analysis”
It would be interesting to do a second meta-analysis that is more or totally inclusive, and compare results.
I take the meta-analysis to mean that if you want to discount the benefits of antioxidants, you must discard 2295 out of 2311 controlled trials, or 99.3% of them. What other proposed therapy is robust enough to stand up to such treatment? Probably not even aspirin.
Nathan, the discarding of the “bad” studies was done, I’m sure, according to widely-accepted rules. It wasn’t tailored to this particular situation. But it is truly bizarre that this sort of thing is considered reasonable.
“The a priori defined exclusion criteria were as follows: 1) no data in humans, 2) no original data, 3) observational epidemiologic study, 4) no random allocation, 5) no vitamin-mineral supplement as intervention, and 6) no subclinical marker of atherosclerosis as outcome”
Seeing as how supplementation is a subject so prone to bias, I’m not sure it’s such a bad thing. Do we need 100 new minerals claiming to be the cure to heart disease? A thousand new snake-oil salesman to hock them, because they’re unregulated and so easily obtained? What makes you think a simple dose of vitamins deserves to cure one of the biggest killers out there? Maybe it wouldn’t have hurt to list the major culling criterion, but they started out with one huge filter: looking for effects on heart-disease.
“I’m not sure it’s such a bad thing.” Is there any other branch of science in which 99% of expensive hard-to-collect data is thrown out?
Comments are closed.