10/11/04

Pray get better, II

There's a new article in the Times on prayer and health. It seems there's a controversy over Federal funding for research examining a connection between the two.

I posted on this matter last year, and I am, of course, skeptical that anything would come of such research. But on the other hand, 2.3 million dollars (the amount quoted in the Times article) is a laughably small amount of money for scientific research. (For the sake of comparison, the FY 2004 budget of the National Center for Complementary and Alternative Medicine was about $118 million.)

Furthermore, sometimes extraordinary claims, for which the plausible biological mechanism isn't (yet) known, require judicious open-mindedness when framing the research question. One could do research into prayer and health with only so-so exposure definition and still get worthwhile results regarding the endpoint. In less jargony language: if we defined "prayer" as "any request of a supernatural being," wouldn't it be interesting if a study found a link between prayer and health, if the proper confounders* were controlled for, even if prayer in this instance were not especially well defined?

*A reader (hi, Mike!) asked, on a previous post, what "confounder" means. To abridge a complicated and interesting discussion, a confounder is a variable which is associated with exposure and associated with disease, but is not on the causal pathway between them. For this reason, an analysis that fails to control for a confounder will find a spurious association.

A classic example is a study done a few years ago that purported to show a statistical association between pancreatic cancer and coffee drinking. Unfortunately, it turned out that the controls (non-cases used for comparison) were other patients in the same hospital, without pancreatic cancer, who had been diagnosed with other gastrointestinal diseases. These GI patients were, reasonably enough, counseled to avoid coffee, which looked statistically (since confounding was not controlled for) as if pancreatic-cancer patients drank more coffee.

If that's not clear, someone let me know and I'll try explaining again.

No comments:

Post a Comment