[Skip to Navigation]
Sign In
Editor's Correspondence
June 26, 2000

A Randomized, Controlled Trial of Prayer?

Arch Intern Med. 2000;160(12):1871-1872. doi:

Extraordinary claims require extraordinary evidence. However, is the evidence sufficient for the claim by Harris et al1 that prayer may be an effective adjunct to standard medical care? Patients in the coronary care unit (CCU) who were randomized to receive remote, intercessory prayer (plus usual care) stayed as long in the CCU and in the hospital as patients who received usual care only. Furthermore, there were no differences between groups on 34 clinical outcome characteristics, but the prayer group had 11% lower scores on a new, unvalidated summary statistic describing clinical CCU course. The only alternative explanation that the authors discuss is chance, which they consider unlikely given one statistically significant (P=.04) difference between groups. The authors do not realize, however, that by making 34 comparisons using separate t tests with α set at .005 and another 3 with α set at .05, the chance of finding 1 significant difference is not 1 out of 25, but

Add or change institution