Concept

Byrd Cardiac Prayer Study (1988)

Intro

Does prayer actually do anything? In 1985, a cardiologist named Randolph Byrd at San Francisco General Hospital set out to test that question scientifically. He used the gold-standard tool that medicine uses for testing drugs: a randomized double-blind controlled trial.

Byrd took 393 patients in the cardiac care unit and randomly assigned them to two groups. One group of 192 had a team of Christian intercessors praying for them daily, by name, from outside the hospital. The other group of 201 received the same standard medical care but no prayer in the study's protocol. Neither the patients, the doctors, nor the nurses knew who was in which group.

Over the course of the study, the prayed-for group had statistically significantly better outcomes. Patients in the control group needed ventilators, antibiotics, and water pills more often. The prayed-for group had a smoother hospital course on a standard severity score.

Byrd published his results in 1988 in the Southern Medical Journal. The paper, "Positive Therapeutic Effects of Intercessory Prayer in a Coronary Care Unit Population," became the first peer-reviewed randomized trial of prayer and kicked off a whole research literature. Later studies have been mixed: Harris's 1999 replication confirmed the effect, the Benson STEP study in 2006 found nothing, and the STEPP Mozambique study in 2010 found dramatic effects for in-person prayer.

The Byrd study is filed here as a Tier 1 documented case because the evidence is a peer-reviewed paper, the methodology is published, and the data are independently checkable.

In full

(See sections below.)

Summary

Randolph C. Byrd, a board-certified cardiologist at San Francisco General Hospital, conducted a prospective randomized double-blind clinical trial measuring the effect of distant intercessory prayer on cardiac care unit (CCU) outcomes between August 1985 and August 1986. 393 patients were randomized to either an "intercessory prayer group" (192 patients prayed for daily by participating Christian intercessors outside the hospital) or a "control group" (201 patients receiving standard care without protocol-mandated prayer). The prayer group exhibited statistically-significantly lower severity-scores on the post-entry hospital course (P <.01); control patients required ventilatory assistance, antibiotics, and diuretics significantly more frequently. Byrd published the results in Southern Medical Journal 81(7):826-829 (July 1988): "Positive Therapeutic Effects of Intercessory Prayer in a Coronary Care Unit Population." The study became the landmark first-randomized-controlled-trial of intercessory prayer and initiated the modern empirical-prayer-research literature (Harris 1999 replication, STEP 2006 Benson null-result, STEPP 2010 STEPP Mozambique Study (Brown 2010) all in the lineage Byrd opened).

The event

The study design:

Patient enrollment: 393 consecutive CCU admissions over a 10-month period (Aug 1985 - Aug 1986) at San Francisco General Hospital. Patients were randomly assigned by computer to intercessory-prayer-group (n=192) or control (n=201).
Blinding: double-blind, neither patients nor attending physicians knew which group a patient was in. Byrd's primary research-coordinator Janet Greene knew assignments (this is later identified as a methodological-blinding issue by Posner's critique; see Verification below).
Intercessor protocol: participating Christian intercessors received the patient's first name + diagnosis + general clinical-status. They were instructed to pray daily for the patient's "rapid recovery and prevention of complications and death." Each patient had 3-7 intercessors. Intercessors prayed in their own homes; no contact with the hospital, the patients, or each other was structured.
Outcome measures: 26 outcome variables measured during hospitalization, including new diagnoses, complications, therapeutic interventions, and final clinical status. A composite "hospital course severity score" was calculated.
Statistical analysis: chi-square + t-test comparisons; Bonferroni-corrected significance levels.

The published results:

Composite severity score: significantly favored the prayer group (P <.01)
Specific outcome differences favoring prayer group (with Bonferroni-corrected P-values): congestive heart failure (P =.03), use of diuretics (P =.05), cardiopulmonary arrest (P =.02), pneumonia (P =.03), use of antibiotics (P =.005), intubation (P =.002)
No significant differences in: length of CCU stay, total days hospitalized, number of deaths

Witnesses + documentation

Lead investigator: Randolph C. Byrd, MD, cardiologist; affiliated with San Francisco General Hospital at study time. Member of American College of Cardiology.
Study coordinator: Janet Greene, RN; data-collection responsibility.
Peer-reviewed publication: Southern Medical Journal 81(7):826-829 (July 1988). The journal is a credible academic medical-journal published by the Southern Medical Association. Article is PubMed-indexed (PMID: 3393937), indicating standard medical-literature acceptance criteria were met.
Replication studies: William S. Harris et al. (1999) in Archives of Internal Medicine, replicated Byrd's protocol with similar positive results in 990 CCU patients at MidAmerica Heart Institute, Kansas City (P =.04 for composite-score). Posner's critique addressed in Harris's published work.
Subsequent contrary study: Benson et al. (2006) STEP study in American Heart Journal, large multi-center RCT (1,802 patients) showing null results on cardiac-bypass-surgery outcomes; in fact, patients who KNEW they were being prayed for had MORE complications (suggesting a "performance anxiety" effect). The STEP null result is significant for the broader empirical-literature.
Continuing-tradition study: STEPP Mozambique Study (Brown 2010), Brown et al. Southern Medical Journal 2010, positive results in Mozambican proximal-prayer (laying-on-of-hands) context.

Verification

This is a peer-reviewed-paper source-type entry. Verification operates differently than for medical-bureau-ratification entries (Lourdes / Vatican), the study is verified by:

Peer-review process: Southern Medical Journal editorial review confirmed the methodology meets standard medical-literature criteria
Independent replication attempts: Harris 1999 replicated Byrd's protocol with similar positive results; STEP 2006 produced a null result on a different protocol (cardiac surgery rather than CCU admissions); the broader literature shows mixed results
Methodological transparency: Byrd published the full protocol, allowing independent critique and replication

Skeptical engagement (Gary P. Posner MD, Free Inquiry 1990 + Skeptical Inquirer):

Posner's published methodological critiques of Byrd 1988 are substantive and worth engaging openly:

The "pure groups" problem: Byrd himself acknowledged in the published paper that "some of the patients in both groups would be prayed for by people not associated with the study; this was not controlled for. Therefore, 'pure' groups were not attained in this study." Posner correctly notes this dilutes the comparison, the control group was likely receiving SOME prayer (from family, churches, etc.) at unknown rates.
Blinding integrity: Posner notes that Janet Greene, the study coordinator on whom Byrd depended for data collection, knew the patient assignments and interacted regularly with patients. This is a real concern, Byrd's response is that the data-coding was performed at sufficient remove from the assignments to mitigate the bias, but the critique is methodologically valid.
Outcome-variable independence: Posner argues that some of the 26 outcome variables are not independent (e.g., "antibiotic use" and "pneumonia diagnosis" are causally-linked), inflating the apparent effect-size. Multiple-comparison concerns are real; the Bonferroni correction Byrd applied addresses some but not all of this concern.
Effect-size limitation: the LARGEST outcome differences (length of stay, total days, mortality) showed NO significant difference. The significant differences were in lower-severity intermediate-outcomes. This pattern is consistent with either real-but-modest prayer-effect OR with chance distribution of the 26 variables.

These critiques don't refute the study's empirical results, but they raise the bar for confidence, replication is required to rule out methodological-artifact explanations. Harris 1999 partially replicated; STEP 2006 produced null results on a different protocol; the broader literature is mixed. The HONEST scientific assessment: Byrd 1988 provides genuine empirical evidence consistent with intercessory-prayer-effect, but the empirical literature as a whole has not converged on a robust positive result.

Theological framing: The Christian theological response to mixed empirical results is consistent: God is not a vending machine; statistical-prayer-effects measured under controlled-study conditions are inherently limited (the subject of prayer is a Person who responds to relationships, not a force that operates on stimulus-response). C.S. Lewis's Letters to Malcolm engages this directly. The empirical studies provide one form of evidence; the broader Christian-theological-experience tradition provides others. Neither alone is decisive.

Apologetic value

Anti-Hume "in principle" deployment: the Byrd study is peer-reviewed published evidence in a credible medical journal of a positive intercessory-prayer-effect on measurable medical outcomes. Hume's framework requires that all evidence for miracles be of a kind he can dismiss; peer-reviewed published RCTs are structurally not of that kind. Even if the study has methodological-limitations (which Posner identifies), the empirical bar is FAR HIGHER than what Hume's framework presupposes.
Anti-strict-naturalism deployment: the study provides empirical data inconsistent with strict-naturalist expectations. If naturalism is true, prayer should have NO measurable physiological effect. Byrd 1988's positive result + Harris 1999 replication + STEPP 2010 in different context provide cumulative empirical pressure on strict-naturalism.
Diversification of miracle-evidence sources: the Miracles collection benefits from source-type diversity. Lourdes-Bureau and Vatican-canonization cases provide medically-vetted individual-cure documentation; peer-reviewed-prayer-studies (Byrd 1988, Brown 2010) provide population-level statistical evidence. Different evidential modes; cumulative apologetic case.
Deployment with intellectual honesty: this entry should be deployed WITH the skeptical engagement (Posner's critiques + STEP 2006 null result). The honest deployment STRENGTHENS apologetic credibility because it shows engagement with the actual scientific literature rather than selective-quotation. The dishonest "Byrd 1988 PROVED prayer works!" deployment is counterproductive, it invites legitimate skeptical pushback.