Skip navigation

Old Curiosity Shop: the power of prayer

Many people of different faiths believe that prayer can have beneficial effects on their own or others' health. Testing the power of prayer is not easy, but a randomised, double-blind trial with large numbers demands some attention.


During a 10-month period in 1982/3, patients admitted to San Francisco General Hospital coronary care unit were eligible for entry. Of these, 393 entered the trial and 57 did not want to participate after being fully informed about the nature of the project.

Intercessionory prayer was provided by "born again" Christians of several denominations. After randomisation (by computer-generated list), patients in the prayer group were prayed for by between three and seven intercessors. The intercessors were given the first name, diagnosis and general condition of the patient, with pertinent updates. Prayer took place outside the hospital daily until discharge. Intercessors prayed for a rapid recovery, and for prevention of complications and death, as well as anything else they wanted to add to the prayer.

Patients had no idea whether or not they were being prayed for. Additional prayers in either group by, for instance, family members, was not controlled for. Data on patients' condition, complications and outcome was collected blind.


There were no differences at entry between patients for any demographic variable, primary cardiac diagnosis, or noncardiac illness or complication.

Intercessionory prayer was without effect on days spent in the coronary care unit, or days in hospital, or number of discharge medications. There were 26 new problems, diagnoses or therapeutic events monitored after entry into the trial; 107 events occurred in 192 patients being prayed for and 175 occurred in 210 control patients. Six events occurred significantly less frequently with prayer - congestive heart failure, use of diuretics, cardiac arrest, pneumonia, antibiotic use and intubation or ventilation.

The clinical course of patients was scored as good, intermediate, or bad according to a scoring system. Good outcomes were more frequent in patients who were being prayed for (163/192; 85%) than in those who were not prayed for (147/201; 73%). This generated a relative benefit of 1.16 (1.05 - 1.29) and a NNT of 8.5 (5.1 - 26).


People will be able to see what they want from this trial. It was a properly randomised, double-blind trial. It had statistically significant outcomes in favour of prayer having a beneficial effect in this patient group. There were no outcomes for which the control group did better than those being prayed for, and though the effects are not great, they all go in one direction, that prayer is effective.

Doubters might point out that there may have been an element of data-dredging, because there was no prior statement of what outcomes were going to be looked at, and so we have to look less favourably on the result. They may also point out that some of the events where statistical significance was found were fairly rare, occurring in only a few percent of patients in both groups, so random chance may play a part.

The lesson is that a single trial is just that, one observation. If the effect was massive, and the trial huge, and there was an agreed and understood mechanism, then perhaps taking results from a single trial may be OK. Where these conditions are not met, then caution rules. The fact that a Cochrane review group is summarising all the literature on the effects of prayer is welcome.


  1. Positive therapeutic effects of intercessionory prayer in a coronary care unit population. Southern Medical Journal 1988 81: 826-9.

previous or next story in this issue