Monday, October 13, 2008

Did a meta-analysis published in the Lancet really vindicate homeopathy?

The audacity of con artists knows few limits. In response to my recent Epinions beat-up of Oscillococcinum, a "homeopathic" sugar pill made and marketed by French quack cure behemoth Boiron, Inc., a representative of the company, one Alissa Gould, posted fifteen studies which supposedly prove the efficiacy of homeopathy or Oscillococcinum. Some of these studies were merely low quality garbage published in faux journals set up to promote quackery (Journal of Alternative and Complementary Medicine", British Homeopathy Journal), others were inconclusive and more than one actually supported the opposite of what Ms Gould claims.

The one that caught my attention was a 2005 study by Shang et al, "Are the clinical effects of homoeopathy placebo effects? Comparative study of placebo-controlled trials of homoeopathy and allopathy", Lancet 366, and not only because I was shocked that the editors at such a prestigious journal let the use of "allopathy" as a synonym for mainstream medicine make it to press. (Whether "allopathy" was ever practiced is questionable, but it is clear that modern medicine is not based on some tawdry "cure disease with its opposite" maxim.)

Shang and co-authors compared 110 placebo-controlled trials of homeopathy and 110 matched trials of evidence-based medicine, finding a trend toward bias in small trials of both, but significant evidence from the large trials that conventional medicine has specific effects, whereas homeopathy is similar to placebo. The paper is a bit difficult to read--I can't think of another field in which those "funnel plots" are used--but worthwhile not only for its frank assessment of homeopathy but also as an exemplar of method.

Take-home lesson?: Beware Boiron employees pushing studies: they don't limit their hucksterism to the sugar-pill trade.