Concept

Harris CCU Prayer Study (1999)

Intro

The Byrd study in 1988 was the first peer-reviewed test of whether intercessory prayer made a measurable difference in hospital outcomes. It found a small but significant effect. Naturally, the next question was: does that result hold up when someone else runs the experiment?

In 1997, William Harris and colleagues at the Mid America Heart Institute in Kansas City set out to do exactly that. They ran a larger trial, 990 patients newly admitted to the coronary care unit at Saint Luke's Hospital. Patients were randomly assigned to a prayer group (466 people) or a usual-care control group (524). Neither the patients nor their doctors knew which group they were in.

Five teams of Christian intercessors, each with five people, were given only first names and asked to pray daily for four weeks for the assigned patients' speedy recovery.

The result was a small but statistically significant improvement in the prayed-for group. On the standard severity score the hospital used to track how rough a CCU stay was, the prayer group scored 6.35 and the control group 7.13, an eleven percent reduction in favor of prayer. The unweighted score also favored prayer. Length of stay was not different.

The study was published in Archives of Internal Medicine in October 1999. It is the closest formal replication of the Byrd study and broadly confirmed Byrd's earlier finding. Methodological debate followed, including a published commentary that Harris engaged honestly in the same journal.

This is filed as a Tier 1 documented case: peer-reviewed paper, published methodology, replicable design.

In full

(See sections below.)

Summary

Randomized, controlled, double-blind, prospective parallel-group trial of remote intercessory prayer on coronary care unit (CCU) outcomes. 990 consecutive newly-admitted CCU patients at the Mid America Heart Institute, Saint Luke's Hospital, Kansas City, MO (Aug-Oct 1997). Prayer group n=466; usual-care group n=524. Five 5-person intercessor teams (Christian; informed only of patients' first names) prayed daily for 4 weeks for assigned patients' speedy recovery. Primary outcome: weighted MAHI-CCU course-severity score covering cardiac arrest / intubation / antibiotics / pneumonia / MI extension / CHF / etc. Result: weighted MAHI-CCU score 6.35 ± 0.26 (prayer) vs 7.13 ± 0.27 (control), P =.04, an 11% reduction favoring the prayer group. Unweighted score 2.7 vs 3.0, P =.04. Length of stay + time in CCU not different. Published in Archives of Internal Medicine 159(19):2273-2278, October 25, 1999. The closest formal replication attempt of Byrd Cardiac Prayer Study (1988), with significant methodological controversy engaged honestly via the Hoover-Margolick 2000 commentary in the same journal.

The event

Setting. Mid America Heart Institute (MAHI) at Saint Luke's Hospital, Kansas City, tertiary-care cardiac center. CCU patients admitted Aug-Oct 1997 enrolled. Modeled on Byrd 1988 (already in collection at Byrd Cardiac Prayer Study (1988)).

Methodology. Computer-randomized at admission (prayer n=466, usual-care n=524). Double-blind: patients didn't know they were enrolled, CCU staff didn't know assignment, outcome-tabulators blinded. 75 Christian volunteer intercessors in 15 teams of 5; given only patients' first names; prayed daily for 28 days for "speedy recovery with no complications." Primary outcome: the MAHI-CCU continuous weighted composite score (cardiac arrest, intubation, antibiotics, pneumonia, MI extension, CHF, arrhythmias, length-of-stay, etc.); higher = worse.

Results. Weighted MAHI-CCU score: prayer 6.35 ± 0.26 vs usual-care 7.13 ± 0.27, P =.04 (11% reduction favoring prayer). Unweighted total events: 2.7 ± 0.1 vs 3.0 ± 0.1, P =.04. Length of stay + mortality not different. Reported as "remote intercessory prayer was associated with lower CCU course scores."

Witnesses + documentation

Lead author: William S. Harris, PhD (lipid-research physiologist; Saint Luke's research-staff at the time)
Co-authors: Manohar Gowda MD, ris3n W. Kolb MD, Christopher P. Strychacz PhD, James L. Vacek MD, Philip G. Jones MS, Alan Forker MD, James H. O'Keefe Jr MD, Ben D. McCallister MD
Institution: Mid America Heart Institute, Saint Luke's Hospital, Kansas City, MO 64111
Journal: Archives of Internal Medicine (now JAMA Internal Medicine) 159(19):2273-2278, October 25, 1999
Article DOI / PMID: PMID 10547166

The publication is a fully peer-reviewed RCT in a tier-1 internal-medicine journal. Outcomes were tabulated by chart-review of standard CCU records.

Verification

Independent verification:

Standard peer-review process at Archives of Internal Medicine (highly cited general-internal-medicine journal).
Open critique published in the same journal: Hoover DR, Margolick JB, "Errors in the procedure for randomization in 'A randomized, controlled trial of the effects of remote, intercessory prayer on outcomes in patients admitted to the coronary care unit'", Archives of Internal Medicine 160(12):1875-1876, June 26, 2000, engaged the methodology directly.

Methodological concerns (engaged openly):

MAHI-CCU weighted scoring system unvalidated, developed in-house, no published validation literature.
Failed when Byrd's scoring was applied, applying Byrd 1988's scoring categories to the same dataset yielded no significant effect; the positive result depends on the MAHI-CCU scoring choice.
Event-classification sensitivity, Hoover-Margolick: reclassifying cardiovascular stress tests from category-4 to category-2 changes weighted score to 6.97 vs 6.24, P =.05 (marginal).
Background prayer contamination, CCU patients commonly receive prayer from family / clergy; pure-groups not attained (inherited from Byrd 1988).
Multiple comparisons, family-wise error inflation; p-values not corrected for multiplicity.

Honest verdict. RCT design + double-blinding + adequate sample size + tier-1-journal review = Tier-1 source-type-strength. Hoover-Margolick critiques are real and were engaged by the Harris team in subsequent correspondence. Result is consistent-with-prayer-effect but not robust to all defensible analytic choices, the open-question state intellectual honesty requires the codex to display.

Apologetic value

Empirical falsifier of Hume's In Principle dismissal. A peer-reviewed RCT in a tier-1 medical journal showing P=.04 effect of remote intercessory prayer crosses Hume's in principle dismissal at the methodological-rigor level.
Companion to Byrd Cardiac Prayer Study (1988) + STEPP Mozambique Study (Brown 2010). Together they form a multi-study cluster: Byrd-Harris-Benson(STEP)-STEPP, mixed results, different populations + methodologies. Honest integration: Byrd positive (P<.01); Harris partial-positive (P=.04, methodologically contested); Benson STEP 2006 null on cardiac surgery, known-prayer-recipients did worse (a much-cited atheist counterargument); Brown STEPP 2010 positive in Mozambican proximal-prayer setting.
Engages the empirical-prayer-research literature on its own terms. "There IS a peer-reviewed empirical literature on intercessory prayer; it is mixed; the methodologies are debated; but the literature exists and has produced statistically-significant positive results in multiple published RCTs." See Astin et al. Annals of Internal Medicine 2000 meta-analysis for field-state.

Theological framing: intercessory prayer is fundamentally Person-to-Person petition, not a force-with-statistically-measurable-output. Statistical-prayer-effects under controlled-study conditions are inherently a limited probe of the underlying theological reality (cf. C. S. Lewis Letters to Malcolm on prayer-and-statistics). The studies are evidential supplements, not the primary evidential basis for prayer's reality.