Table showing summary of results

     We will review selected studies of intercessory prayer for health.  We will not review studies of the association of religious involvement or spirituality and better health outcomes.  For the latter, please see the article by Mueller et al.[1]  We will assume that the prayers in the studies were intended to be helpful and not harmful.  The intercessors were usually members of Protestant and Catholic groups. In most of the studies, prayer was offered at a distance and not in the presence of the patients.  Informed consent was obtained, and the studies were randomized, controlled, and double blinded, that is, the patients and researchers did not know who was or was not receiving the designated intercessory prayers.  Deviations from these features will be indicated.  The articles were found from a PubMed internet search of intercessory prayer, from the review articles of Astin et al[2], Abbot[3], Roberts et al[4], Masters et al[5], and other articles. 

     Although the first known study of intercessory prayer does not meet the above criteria, we include it for historical interest.  It was published by Francis Galton [6], who was associated more with the eugenics movement than with prayer.  Based on published registries, he compared the lifespans of several thousand people.  Those who were likely to be more prayerful and to receive more prayers like royalty and clergy did not have longer life spans than people usually less prayerful like lawyers and doctors.  Based on this and other data that he collected, he concluded that there was no statistical evidence for the efficacy of prayer.

     Joyce and Weldon[7] reported in 1965 the first study that met most of the more stringent criteria.  It had 48 patients with chronic stationary or progressive deteriorating psychological or rheumatic disease.  Patients were matched in pairs as closely as possible for sex, age, primary clinical diagnosis, marital status, and religious faith.  After matching, the pairs were randomized.  Informed consent was not obtained.  There was no significant difference in clinical or attitudinal states although there was a trend favoring the patients who received prayer, but the small size of the study was a limiting factor, i.e. type II error or false negative result for intercessory prayer or false positive null hypothesis.

     Collipp[8] reported the next study in 1969.  This study was triple blinded in the sense that even the intercessors were not told that they were participating in a study on the efficacy of prayer.  There were 18 children with acute leukemia.  Of the 10 in the prayer group, 7 were alive after 15 months.  Of 8 in the control group, 2 were alive at 15 months.  This reached statistical significance if a single outlier was dropped.  However, the conclusion is questionable because there was no matching prior to randomization, which placed both patients with unfavorable prognosis, acute myelogenous leukemia, in the control group.  The remainder of the patients had acute lymphocytic leukemia.

     Byrd[9] published a larger series in 1988 comprising 393 patients admitted to a coronary care unit.  The intercessors were “born again Christians (according to John 3:3) with daily devotional prayer and active Christian fellowship in a local church.”  The intercessors were given the patients’ first names, diagnosis, general condition, and updates.  The frequencies of complications were compared and the hospital courses were graded as good, intermediate, or bad.  The prayer group had significantly better results.  However, Roberts[10] stated that “the point at which decisions were made relating to the definitions of ‘good’, ‘intermediate’ and ‘poor was not stipulated.  It has never been clear whether these important decisions were made before or after seeing the data and whether those doing the analysis were blind to group allocation.  Bias could have influenced the result.”  Several years after the original publication, Byrd[11] indicated that the person who randomized the patients into the groups to receive and not to receive the special prayers was also collecting clinical data on the patients, thus risking unblinding of the study. And Sloan et al[12] wrote “29 outcome variables were measured, and on six the prayer group had fewer newly diagnosed ailments.  However, the six significant outcomes were not independent…”

     Also in 1988, Beutler and colleagues[13] reported on 96 patients who had uncomplicated hypertension.  The intercessors were members of the Dutch society of paranormal healing.  After matching in groups of three, there was randomization of the triplets.  Groups 2 and 3 were the usual healing and control groups.  But group 1 received laying on of hands with verbal communications.  Therefore, only groups 2 and 3 were blinded.  All three groups had significant decreases in hypertension even before the intervention began, possible Hawthorne effect, the beneficial effects of mere participation in a trial because of focused attention on the subjects.  But the results of no one group was better that any other.

     Walker et al[14] published a study of 40 alcohol abuse patients in 1997.  All the intercessors signed an agreement not to pray for specific religious conversion of the clients.  There was no significant difference in alcohol consumption with or without designated prayers.  However, patients who knew that someone was praying for them outside of the study had significantly heavier drinking at 6 months follow–up.  The authors suggested that clients that rely on a higher power or on the prayers of others may be less inclined to be active in their own recovery.

     O’Loire[15] (1997) recruited 406 volunteers who responded to an advertisement.  They were studied for the effect of prayer on self esteem, anxiety, and depression.  The volunteers were divided into three groups to receive directed prayers for their benefit, non-directed prayers (God’s will be done), and no prayers.  All three groups improved in all measures, again possible Hawthorne effect; however there were no significant differences among the three groups, but those who were convinced they were in the prayer group were significantly better that those convinced they were in the control group.  The intercessors were also tested before and after the trial and were significantly better than the volunteers at pretest and post test.  The intercessors kept prayer logs, and those who prayed more improved more on testing, but the amount of prayer made no difference in the volunteers.

     Sicher et al[16] (1998) studied 40 AIDS patients.  The intercessors had diverse backgrounds: Christian, Jewish, Buddhist, Native American, shamanistic traditions, and bioenergetic and meditative healing.  Each patient was treated by 10 different practitioners.  For patients that received distant healing, there were fewer hospitalizations, fewer days of hospitalization, fewer new AIDS defining diseases, and a lower illness severity level on the Boston Health Study scale.  But the study was relatively small, and the CD4 count improvements were not significantly different between the intervention and control groups.  In spite of pair matching and randomization, all five baseline smokers and all four minorities were in the control group.

     Harris et al[17] (1999) evaluated all patients admitted to the cardiac care unit over 12 months, of which 990 out of 1019 were suitable for analysis.  The Institutional Review Board approved the proposal to proceed without informed consent, and neither patients nor medical staff was aware of the prayer study.  Their justifications were: 1. There was no known risk of receiving or not receiving additional intercessory prayers.  2. No additional data was collected on these patients.  3. Informed consent could increase anxiety in the patients.  Using an unvalidated scoring system specially designed for this trial, the patients in the intercessory prayer group were better at marginal significance (P=0.04) and the construct validity of the scoring system was questioned by Sloan[18].  For example he wondered why the need for an electro physiologic study (3 points) was three times as bad as the need for antibiotics (1 point).  And the use of the t test on the results of the scoring system was questioned by Hammerschmidt[19] because the unit increments in the scale could not be assumed to be uniform.  He also questioned the method of “randomization” based on odd and even medical record numbers.  This could have lead to unintentional unblinding.  There was no statistical significant difference for any individual complication, mean length of stay in the cardiac care unit or in the hospital.  Using the Byrd score, there was no statistical difference, but there was a trend toward better outcome in the prayer group.

     Matthews, D. et al[20] (2000) studied 40 rheumatoid arthritis patients.  The patients received in person audible intercessory prayers with laying on of hands, supplemental distant intercessory prayers, or served as controls.  In person intercessory prayer was associated with significant improvement in signs and symptoms, but there was no additional effect from supplemental distant intercessory prayer.  The improvements were not accompanied by parallel reductions in serum inflammatory markers.  The placebo effect could not be excluded, and the authors recognized that the placebo effect was common in rheumatoid arthritis.  Of course blinding was incomplete, but the evaluators were blinded.

     Harkness et al[21] (2000) chose warts, a reasonable lesion for study because they can be counted and measured and because they have a long history of being “charmed.”  The distant healers described their method as “flow/channeling interactions of energy.”  There were no significant differences in number and size of warts or in the subjective responses of the patients and controls.  The number of patients, 84, was a limiting factor.

     Matthews, W. et al[22] (2001) studied 95 patients who had end-stage renal disease on dialysis.  Roughly half of the patients were told that they would receive prayer.  The other half were told they would receive positive visualization.  But each group was divided into three smaller groups to receive distant intercessory prayer, positive visualization, or neither.  Therefore, many patients did not receive what they expected.  Analysis of the physical and psychological variables showed no treatment effects for either intervention.  However, those who expected to receive intercessory prayer reported feeling better than those who expected to receive positive visualization.  The study was limited in power because of the small number of patients.

     Aviles et al[23] (2001) reported a study of 799 Mayo Clinic patients admitted to the coronary care unit and discharged.  Patients were randomized at discharge to the intercessory prayer group or the control group.  No significant differences were detected in the medical follow-up or in the quality of life.  The authors mention several limitations:  The prayers were variable in content, amount, frequency and number of intercessors.  They could not measure background prayers or determine the receptivity of the patients to healing prayers.  The outcomes that they chose to study may not have been those affected by prayer.

     Cha et al[24] (2001) published a questionable study.  It is included here because it provoked so much controversy.  They reported 169 in vitro fertilization patients in Seoul, Korea.  The intercessors were in the U.S., Canada and Australia.  Like the Harris study, the patients and providers did not know they were participating in the study.  The end point of the study was pregnancy determined by ultrasound identification of a fetal pole and a heartbeat.  There was a significantly greater pregnancy rate in the prayer group, 50%, vs. 26% for the control group.  The overall pregnancy rate in the program during the previous year was 32.8%.  So the lower rate in the control group may have magnified the difference with the prayer group.  And the study has been clouded by the Federal grand jury indictment[25] of the second author who “pled guilty to all crimes contained in his federal indictment, thus admitting to a 20-year history of criminal, fraudulent activities,” crimes unrelated to the study.  The second author had designed the study and organized the prayer groups, but doubt was raised whether the prayer groups were organized and actually prayed for the patients.

     Leibevici[26] (2001) published a debatable study of retrospective intercessory prayer.  It is included here because it seemed to be a study intended to show that intercessory prayer should not be tested in controlled trials[27].  They studied 3393 patients with bloodstream infections during the period 1990 to 1996, but the prayers were not offered until 2000.  There was no significant difference in mortality, but the length of stay and duration of fever were significantly shorter in the prayer group. P=0.01 and 0.04 respectively.  The rationale for retrospective prayer was that God is not limited by linear time.  In a letter to the editor, Hettiaratchy and Heansley[28] commented that the length of stay was the most significant finding.  Although the median length of stay was similar in the control and intervention groups, the control group had some very long stays that skewed the results.  They also stated that the “paper proves power of statistics, not prayer”.

     Krucoff et al[29] (2005) reported on 748 patients having elective percutaneous coronary intervention or elective catheterization with possible coronary intervention.  The distant intercessors were Christian, Muslim, Jewish, and Buddhist.  There was also music, imagery, and touch therapy (MIT).  Patients were randomized to prayer only, MIT only, both and neither.  There were no statistically significant differences among the four groups with the exception of lower mortality at 6 months for MIT therapy than no MIT therapy.

     We end the review with the major study by Benson et al (2006)[30].  This was a large, well designed, well executed, multicenter study.  It enrolled 1802 patients admitted for coronary artery bypass graft surgery, cost 2.4 million dollars, and took 10 years to complete.  The authors learned from and avoided the mistakes of previous studies.  The patients were from Oklahoma, Massachusetts, Washington DC, Tennessee, Minnesota, and Florida.  The investigators were from five different medical centers.  For outcomes, they chose postoperative complications defined by the Society of Thoracic Surgeons Adult Cardiac Surgery Database and ‘Major events” defined by the New York State Cardiac Surgery Reporting System.  The intercessors were from three Christian groups, two Catholic and one Protestant.  They were contacted in a consistent manner and agreed to add to their prayers the phrase “for a successful surgery with a quick, healthy recovery and no complication”.  All investigators and patients were blinded except for group 3 patients who were intentionally unblinded and knew they would receive the special intercessory prayers.  Groups 1 and 2 were uncertain about receiving intercessory prayers; group 1 did, and group 2 did not.  59% of group 3 had at least one complication compared with 52% of group 1 (relative risk 1.14, 95% confidence interval 1.02-1.29, P=.025).  51% of group 2 had at least one complication.  18 % of group 1 had a major event versus 13% of group 2 (relative risk 1.18, 95% confidence interval 1.03-1.35, P=.027).  14% of group 3 had a major event.  Mortalities were 3% of group 1, 2% of group 2, and 2% of group 3.  All of the assessments for complications, major events, death were for the period within 30 days of surgery. After excluding patients with incomplete data, the group comparisons were similar.

     The authors felt that the excess complications in group 3 and the excess major events in group 1 may be chance findings.  However, in an accompanying editorial, Krucoff et al[31] suggested that even benevolent intercessory prayer may be harmful.  Bethea[32], one of the co-authors of this study said “…being aware of the strangers’ prayers…may have caused some of the patients a kind of performance anxiety. It may have made them uncertain, wondering am I so sick they had to call in their prayer team?”  In a much smaller study, Walker et al[33] found that patients who knew that someone was praying for them outside of the study had significantly heavier drinking at 6 month follow-up.

      Dusek, second STEP author, said that major event outcomes were introduced half way through the study because the independent Data Safety Monitoring Board suggested that they include secondary outcome measures.  They did not want to emphasize the difference between groups 1 and 2 in major events because they were secondary outcomes.  A possible reason for the unfavorable findings in groups 1 and 3 was the randomization of more patients with unfavorable baseline features to groups 1 and 3, but the article stated that multivariate analysis disclosed no important differences at baseline.

     The authors mentioned some limitations of the study:  The intercessors received limited information and no feedback about the patients and did not communicate with the patients or families.  They recognized that patients could pray for themselves and friends and family could pray for them.  The “non-study prayer” could drown out any effect from the study intercessors.

Author can be reached for comment at:

comments@intercessoryprayerstudies.com