Overall distribution of patients.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Harris WS, Gowda M, Kolb JW, et al. A Randomized, Controlled Trial of the Effects of Remote, Intercessory Prayer on Outcomes in Patients Admitted to the Coronary Care Unit. Arch Intern Med. 1999;159(19):2273–2278. doi:10.1001/archinte.159.19.2273
Intercessory prayer (praying for others) has been a common response to sickness for millennia, but it has received little scientific attention. The positive findings of a previous controlled trial of intercessory prayer have yet to be replicated.
To determine whether remote, intercessory prayer for hospitalized, cardiac patients will reduce overall adverse events and length of stay.
Randomized, controlled, double-blind, prospective, parallel-group trial.
Private, university-associated hospital.
Nine hundred ninety consecutive patients who were newly admitted to the coronary care unit (CCU).
At the time of admission, patients were randomized to receive remote, intercessory prayer (prayer group) or not (usual care group). The first names of patients in the prayer group were given to a team of outside intercessors who prayed for them daily for 4 weeks. Patients were unaware that they were being prayed for, and the intercessors did not know and never met the patients.
Main Outcome Measures
The medical course from CCU admission to hospital discharge was summarized in a CCU course score derived from blinded, retrospective chart review.
Compared with the usual care group (n=524), the prayer group (n=466) had lower mean±SEM weighted (6.35±0.26 vs 7.13±0.27; P=.04) and unweighted (2.7±0.1 vs 3.0±0.1; P=.04) CCU course scores. Lengths of CCU and hospital stays were not different.
Remote, intercessory prayer was associated with lower CCU course scores. This result suggests that prayer may be an effective adjunct to standard medical care.
FROM TIME immemorial, prayer for the sick has been a common response to the illness of a loved one. In some societies and among certain religious groups, prayer is believed to be the most important therapy that can be offered to a sick person, superseding even medical intervention. Nevertheless, intercessory prayer (praying for others) has rarely been subjected to scientific scrutiny. In 1988, Byrd1 published the results of a blinded, controlled trial of 393 patients who had been admitted to a coronary care unit (CCU) at San Francisco General Hospital, San Francisco, Calif. Patients were randomly assigned to either a usual care group, which received no organized prayer, or to an experimental, intercessory prayer group, which received remote (from outside of the hospital) prayer from persons unknown to them. Byrd reported a statistically significant beneficial effect of intercessory prayer as assessed by a summary "hospital course" score. Three recent books on spirituality and healing2-4 have noted that the Byrd study is the only published trial of intercessory prayer with clinically significant end points, and that more scientifically valid (prospective, randomized, controlled, blinded, etc) studies of prayer were needed. The purpose of the present study was to attempt to replicate Byrd's findings by testing the hypothesis that patients who are unknowingly and remotely prayed for by blinded intercessors will experience fewer complications and have a shorter hospital stay than patients not receiving such prayer.
All patients admitted to the CCU at the Mid America Heart Institute (MAHI), Kansas City, Mo, over a 12-month period were eligible for the trial (Figure 1). The only exceptions were those admitted for workup and wait-listing prior to cardiac transplantation (because of anticipated prolonged stays). Patients admitted for less than 1 day were subsequently excluded because it took up to 24 hours for intercessors to be contacted and prayer initiated. New admissions were identified in the chaplain's office on a daily basis via computer. The chaplain's secretary randomly assigned all new patients to either the usual care or prayer group based on the last digit of the medical record number; even numbers were assigned to the prayer group and odd numbers to the usual care group. This allocation scheme allowed no opportunity for bias because medical record numbers are assigned on a sequential basis to all new patients entering the hospital, regardless of how sick they are. In addition, since some patients were readmitted (having been assigned their numbers months to years previously) and some were newly admitted, no systematic assignment of the sickest patients to the odd (usual care) group was possible. Once assigned, the secretary called an intercessory prayer team leader and gave him/her the first name of the patient to be prayed for. No other information (eg, diagnosis, prognosis, age, race, socioeconomic status, or family situation) was available to the secretary; thus, it was not passed on to the intercessors. The secretary was the only person with knowledge of the assignment code, and she had no contact with the patients, the CCU staff (she did not even know where the unit was located within the hospital), the data collectors, or the statistician, all of whom were blinded throughout. After receiving the call from the secretary, the prayer team leader called the other 4 persons on his/her team and directed that the name of the new patient be entered on a log sheet provided. The intercessors were asked to pray daily for the next 28 days for "a speedy recovery with no complications" and anything else that seemed appropriate to them. A period of 28 days was chosen to ensure that prayer would continue throughout the entire hospitalization of at least 95% of patients. Some CCU patients (typically fewer than 5%) request prayer from the hospital chaplain's staff upon admission to the hospital. When made, these requests were always honored regardless of and without knowledge of group assignment. This study was approved by the hospital's institutional review board (IRB) and, in order to keep the study blinded, was exempted from the requirement to obtain informed consent (see the "Comment" section).
The intercessors were recruited by the investigators via contacts in the local community. In order to be an intercessor, an individual did not need to be of any particular denomination, but he/she did need to agree with the following statements: "I believe in God. I believe that He is personal and is concerned with individual lives. I further believe that He is responsive to prayers for healing made on behalf of the sick." Once identified, the intercessors were organized into 15 teams of 5 members (a total of 75), each with 1 person designated as the team leader. Intercessors were randomly assigned to teams; those within a given team did not know the others in the same team, and prayer was offered individually, not in groups.
Patient demographics and admission diagnoses were obtained from the hospital computer system. All patient charts were reviewed retrospectively by a blinded physician/investigator to collect information regarding comorbid conditions at the time of admission, length of CCU and hospital stay, and clinical outcomes. The latter were defined as all new diagnoses, events, or procedures occurring at least 24 hours after admission to the CCU (to allow time for organized prayer to begin) until discharge or death. Thus, if a patient who presented to the emergency department with an acute myocardial infarction was catheterized, revascularized, and then admitted to the CCU, these events/procedures were not recorded as new CCU events. On the other hand, if after the first day in the CCU, a patient developed unstable angina, had a coronary angiogram, and had a subsequent revascularization procedure, all of these were recorded as new events.
Since prayer was offered for a speedy recovery with no complications, it was anticipated that the effect of prayer was unlikely to be evident in any specific clinical outcome category (eg, the need for antibiotics, the development of pneumonia, or the extension of infarction), but would only be seen in some type of global score. Review of the medical literature revealed no previously validated and standardized statistic to quantitate severity of outcomes in critically ill cardiovascular patients. Severity of illness or comorbidity scales, such as the Acute Physiology and Chronic Health Evaluation (APACHE) score5 and Charlson scale,6 do exist, but these are prognostic tools designed to predict major health outcomes for individual patients; they are not designed to summarize a CCU course. Accordingly, before the trial began, 3 experienced cardiologists and 1 internist from MAHI and the University of Missouri–Kansas City School of Medicine developed a weighted and summed scoring system called the MAHI-CCU score (Table 1). The MAHI-CCU score is a continuous variable that attempts to describe outcomes from excellent to catastrophic. For example, if, after the first day in the CCU, a patient developed unstable angina (1 point), was treated with antianginal agents (1 point), was sent for heart catheterization (1 point), underwent unsuccessful revascularization by percutaneous transluminal coronary angioplasty (3 points), and went on to coronary artery bypass graft surgery (4 points), his weighted MAHI-CCU score would be 10. Another patient might have developed a fever and received antibiotic treatment (1 point) but experienced no other problems and been discharged from the hospital with a score of 1. A third patient might have suffered a cardiac arrest (5 points) and died (6 points), for a total weighted score of 11 points. In addition to the weighted MAHI-CCU scores, a nonweighted MAHI-CCU score was calculated that was simply a count of events, procedures, and/or prescriptions after CCU admission. For the examples above, the unweighted MAHI-CCU scores would have been 5, 1, and 2, respectively. To evaluate interrater reproducibility for the MAHI-CCU score, 10 physicians (5 cardiologists and 5 cardiology fellows) blindly scored 11 randomly selected CCU patient charts. The raters were in agreement (mean±SD) 96%±3% of the time. Finally, for comparison, the Hospital Course Score used by Byrd1 was also calculated. The Byrd score broadly categorizes each patient's progress after CCU admission as good, intermediate, or bad.
Baseline variables and specific medical outcomes were analyzed by χ2 analysis and the Fisher exact test for categorical data. Byrd scores were analyzed by the Cochran-Armitage test for trends7; t tests were used to compare continuous variables (eg, age, length of stay, and MAHI-CCU scores). A difference with a 2-tailed P<.05 was accepted as statistically significant, except for comorbid conditions upon admission (Table 2) and individual events/procedures occurring during the CCU stay (Table 3). For these 2 data sets, P<.005 was required for statistical significance because of the multiple comparisons evaluated.8 Data are presented as means±SEs. All analyses were carried out blindly on an intention-to-treat basis using SAS, version 6.12 (SAS Institute, Cary, NC).
The intercessors represented a variety of Christian traditions, with 35% listing their affiliations as nondenominational, 27% as Episcopalian, and the remainder as other Protestant groups or Roman Catholic. The intercessors were predominantly women (87%), and their mean age was 56 years. All reported at least weekly church attendance and daily prayer habits (prior to the study). A review of intercessor log sheets indicated that prayer (by at least 1 intercessor) began within 1.2±0.05 days after admission to the CCU. All intercessors who were ultimately going to pray for a given patient began doing so within 1.6±0.16 days after CCU admission.
A total of 1019 patients were admitted to the CCU during the period of the trial. After elimination of 6 patients who were waiting for cardiac transplantation, 1013 were randomized (Figure 1), 484 (48%) to the prayer group and 529 (52%) to the usual care group. This difference in sample sizes was most likely caused by chance (P=.18). After subsequent removal of those patients who spent less than 24 hours in the CCU, 524 remained in the usual care group and 466 in the prayer group. Comorbid conditions upon admission were similar for each group (Table 2). Men and women were equally represented in the usual care and prayer groups (66% vs 61% men, respectively; P=.10), and the mean age was 66 years for both groups.
The primary predefined end point in this trial was the weighted MAHI-CCU score (Table 4). We found an 11% reduction in scores in the prayer group (6.35±0.26) compared with the usual care group (7.13±0.27) (P=.04). Using the unweighted MAHI-CCU score, which simply counted elements in the original scoring system without assigning point values, the prayer group had 10% fewer elements (P=.04) than the usual care group. There were no statistically significant differences between groups for any individual component of the MAHI-CCU score (Table 3). Mean lengths of stay in the CCU and in the hospital (after initiation of prayer) were not different (Table 4), and median hospital stay was 4.0 days for both groups. There were 2 patients in the prayer group whose hospital stays were approximately twice as long (137 and 161 days) as those of any other patient in the study. Without these 2 patients, length of hospital stay for the prayer group dropped from 6.48±0.54 days to 5.84±0.31 days. Neither was significantly different from the length of stay in the usual care group (5.97±0.29 days). There was no significant difference between groups using Byrd's hospital course score (Table 5).
Using a severity-adjusted outcomes score, we found lower overall adverse outcomes for CCU patients randomized to the prayer group compared with those randomized to the usual care group. Lengths of CCU stay and hospital stay after initiation of prayer were not affected. These findings are consistent with those of Byrd,1 who reported that intercessory prayer for hospitalized patients lowered the hospital course score but did not significantly affect length of stay.
Although there was a trend toward better outcomes in the prayer group using the Byrd score, the difference between groups was not statistically significant. Other than the fact that it is a categorical instead of a continuous statistic, we have no explanation as to why the Byrd score did not detect a difference between groups and the MAHI-CCU score did. There were, however, several important differences between the 2 study designs that may have contributed to this discrepancy. First, the present study was conducted under completely blinded conditions, with neither patients nor medical staff aware that a study was being conducted. In Byrd's trial, the staff and patients were fully aware that the study was in progress, although nobody knew which patients were receiving "study" prayer. Another difference was in the kinds of patients enrolled. In the present trial, informed consent was not sought and thus patients were not prescreened for their willingness to be prayed for. Of the 450 patients invited to participate in the Byrd study, 57 (12.7%) refused to do so "for personal reasons or religious convictions" or were otherwise unwilling to give consent. This indicates that only "prayer-receptive" patients were included in his final cohort. Finally, in Byrd's study, the intercessors were given a considerable amount of information about the patient (eg, diagnoses, general conditions, and updates as their status changed), and they prayed only until the patient left the unit. These factors could have produced a heightened intensity of or commitment to prayer in Byrd's intercessors. In contrast, our intercessors were asked to pray for 28 days regardless of what happened to the patient, and our intercessors received no feedback regarding patient progress during this time. Whether this affected their commitment to continued prayer is not known.
As noted above, both patients and staff were completely blinded not only to assignment of treatment groups, but to the very existence of the trial. This was possible because the hospital's IRB granted the study an exemption from the requirement to obtain informed consent. Since this may be viewed as problematic by some, the reasons supporting this decision will be discussed in some detail. First, it was agreed that there was no known risk associated with receiving remote, intercessory prayer, and no known risk for the patients in the usual care group associated with not receiving extra prayer. Second, no additional data were collected on the patients in this study beyond those that are normally collected for all patients in the hospital. Third, and perhaps most important, the very process of obtaining informed consent could conceivably have caused increased anxiety in some patients. For example, had they known about the study, the possibility of not being in the prayer group might have greatly distressed some patients. For nonreligious or antireligious patients, having to accept or reject the offer of prayer (especially considering the gravity of their illness) might have been very challenging. The policy of the US Department of Health and Human Services for the protection of human subjects states that the
IRB may waive the requirement for the investigator to obtain a signed consent form for some or all subjects if it finds . . . that the research presents no more than minimal risk of harm to subjects and involves no procedures for which written consent is normally required outside of the research context.9
Scientifically, a study design with complete blinding was preferred because it eliminated any possibility of bias, and enrolling all patients in the study increased its generalizability. In light of all these factors, an exemption was granted.
In evaluating the results of this trial, it is important to note that we were most likely studying the effects of supplementary intercessory prayer. Since at least 50% of patients admitted to this hospital state that they have a religious preference, it is probable that many if not most patients in both groups were already receiving intercessory and/or direct prayer from friends, family, and clergy during their hospitalization. Thus, there was an unknowable and uncontrollable (but presumed similar) level of "background" prayer being offered for patients in both groups; whatever impact that group assignment had on healing was over and above any influence background prayer may have had.
Neither this study nor that of Byrd provided any mechanistic explanation for the possible benefits of intercessory prayer. However, others have speculated as to what they might be10; they generally fall into 2 broad categories: natural or supernatural explanations. The former explanation would attribute the beneficial effects of intercessory prayer to "real" but currently unknown physical forces that are "generated" by the intercessors and "received" by the patients; the latter explanation would be, by definition, beyond the ken of science. However, this trial was designed to explore not a mechanism but a phenomenon. Clearly, proof of the latter must precede exploration of the former. By analogy, when James Lind, by clinical trial, determined that lemons and limes cured scurvy aboard the HMS Salisbury in 1753, he not only did not know about ascorbic acid, he did not even understand the concept of a "nutrient." There was a natural explanation for his findings that would be clarified centuries later, but his inability to articulate it did not invalidate his observations.
Although we cannot know why we obtained the results we did, we can comment on what our data do not show. For example, we have not proven that God answers prayer or that God even exists. It was intercessory prayer, not the existence of God, that was tested here. All we have observed is that when individuals outside of the hospital speak (or think) the first names of hospitalized patients with an attitude of prayer, the latter appeared to have a "better" CCU experience. Although our findings would be expected to occur by chance alone only 1 out of 25 times that such an experiment was conducted, chance still remains a possible explanation of our results.
Interest in alternative or complementary medicine is growing rapidly in this country,11,12 and prayer "therapy" falls into this category. Two recent books3,4 have focused on the health benefits of a patient's own spiritual orientation. Each has documented that church membership/attendance is associated with improved medical outcomes.13-15 People who believe in God and pray during illness have been reported to have better health outcomes than people who do not.16-18 For some, faith is an effective means of stress reduction, which has itself been shown to reduce cardiac morbidity.19 Some of these benefits may derive from favorable hormonal, autonomic, and immunologic20,21 responses to the emotional reassurance that belief can provide. Nevertheless, the present trial was designed to study the impact not of personal spirituality, but of prayer offered for patients regardless of their spiritual orientation.
Other studies besides Byrd's have explored the impact of intercessory prayer on health outcomes. O'Laoire22 examined the effects of intercessory prayer on self-esteem, anxiety, and depression in 406 subjects (who received either no prayer, directed prayer, or nondirected prayer) and in the 90 intercessors. There were no specific benefits detected for the prayer groups. A pilot study of the effects of intercessory prayer on 40 recovering alcoholics likewise reported no clinical benefit.23 Finally, in a 6-month trial of "distant healing" in patients with acquired immune deficiency syndrome, Sicher et al24 found statistically significant benefits for the intervention group (fewer new illnesses, physician visits, hospitalizations, and days of hospitalization; lower illness severity scores; and improved mood scores). These studies illustrate the broadening scope of interest in remote therapies and suggest that scientifically valid, properly controlled studies can be carried out in this emerging arena.
The principal limitation of this study was defining the end point measure (ie, determining some way to quantify how well a patient did during a CCU stay). The score we devised, although intuitive and evenly applied to both groups, has not been validated. (It should be noted that the Byrd score is likewise an unvalidated measure of CCU outcomes.) It is not immediately obvious how any score could be validated given the fact that there is no known criterion standard summary statistic with which we could compare the MAHI-CCU score. The fact that there were significantly fewer total events in the prayer group suggests that the observed difference between groups was not an artifact of the scoring system. Another limitation lies in interpreting the clinical significance of a 10% difference in MAHI-CCU scores. Since the score itself is only an estimate of overall CCU course, there is no known way to ascribe a clinical significance to it, other than to say that as a group, the patients in the prayer group "did 10% better." The score should be viewed only as a summary statistic designed to detect the impact of a mild global intervention on overall health in large groups, not in individual patients.
In conclusion, using the MAHI-CCU scoring system, we found that supplementary, remote, blinded, intercessory prayer produced a measurable improvement in the medical outcomes of critically ill patients. Our findings support Byrd's conclusions despite the fact that we could not document an effect of prayer using his scoring method. With 2 randomized, controlled trials now suggesting the possible benefits of intercessory prayer, further studies using validated and standardized outcome measures and variations in prayer strategy are warranted to explore the potential role of prayer as an adjunct to standard medical care.
Accepted for publication May 3, 1999.
The authors are indebted to Tracy Isaacson, the secretary to Chaplain Kolb, for assigning patients to study groups and contacting the intercessors and to Shubha Gowda, MD, for helping with chart review. The support of the associate chaplains at Saint Luke's Hospital was also much appreciated. The encouragement and support of Richard G. Hastings, Douglas Willhoite, MD, and Robert and Marie Evans were greatly appreciated, as was the critical review provided by John Spertus, MD. Finally, we are indebted to the intercessors for faithfully praying for the patients in this study.
Corresponding author: William S. Harris, PhD, Lipid Research Laboratory, Saint Luke's Hospital, 4401 Wornall Rd, Kansas City, MO 64111 (e-mail: email@example.com).