The STEP Intercessory Prayer Trial

Every earlier prayer trial had run into the same design problem. If you tell patients they may be prayed for and then pray for half of them, you cannot distinguish an effect of prayer from an effect of expectation, and if you tell them nothing you cannot study expectation at all. The STEP design used three groups. The first was told they might or might not receive prayer, and did. The second was told the same, and did not. The third was told they certainly would receive prayer, and did. Comparing the first two isolates prayer with expectation held constant; comparing the first and third isolates the effect of certainty.

Intercessors at three Christian congregations received first names and initials and prayed for each patient for fourteen days after surgery, using an agreed form of words. The primary outcome was any complication within thirty days. The trial enrolled patients at six hospitals and reported 1,802 in the analysis, which makes it far larger than anything attempted before or since in this area. The intercessors were drawn from three Christian congregations, two Catholic and one Protestant, and were given only a first name and the initial of a surname, which is roughly what a prayer list in an ordinary congregation contains. The trial deliberately did not attempt to specify how much prayer each patient received in total, since patients also have families, and the authors were explicit that background prayer could not be controlled and that the design therefore compares more prayer against less rather than prayer against none.

Prayer made no difference. Complication rates in the group prayed for and the group not prayed for, both uncertain of their status, were effectively identical. The unexpected part was the third arm: patients told with certainty that they were being prayed for had a somewhat higher complication rate than those left uncertain. The investigators declined to interpret this causally, and were right to. It is one comparison in one trial, it was not the primary hypothesis, and the plausible explanations, from performance anxiety to the possibility that being singled out signals a worse prognosis, are speculation. The trial is nonetheless the most useful entry in the intercessory prayer literature because of how it was built rather than what it found. Byrd in 1988 and Harris in 1999 had both reported positive effects using composite outcome measures assembled from many individual endpoints, a technique that generates apparent effects readily.

A Cochrane review of the field concluded that the accumulated trials did not justify further work of this kind. There is also a theological objection, made from inside religious traditions rather than against them, that a controlled trial treats petition as a mechanism to be metered, which is not what most of those traditions take prayer to be.