Have Studies Proved That Prayer Is Ineffective?

Before even examining the results, one must understand what exactly a distant-intercession trial measures. This clarification is not an apologetic escape hatch: it is the precondition for reading the literature correctly, as the Anglican physicist-theologian John Polkinghorne (Cambridge, Fellow of the Royal Society, President of Queens' College 1988–1996) stressed — he spent his career explaining why Christian theology refuses to be tested in this form^[8].

What a distant-intercession trial actually tests

In all the significant medical trials (Byrd, Harris, Aviles, MANTRA, STEP), the protocol follows the same grammar:

1. Hospitalized patients are randomized into two or three groups.
2. Outside intercessors, often knowing only the patient's first name, receive a standardized prayer instruction.
3. The prayer lasts a few days to a few weeks.
4. An objective medical endpoint is then measured: death at 30 days, complications, length of stay, readmission rate, composite score.

This protocol may seem like common sense. It actually carries three tacit theological presuppositions that the Christian faith rejects:

• Presupposition no. 1: prayer is a doseable natural force, whose intensity or quantity produces a proportional measurable effect.
• Presupposition no. 2: God is a uniform cause, indifferent to the holiness of the intercessor, to the moral state of the patient, to the freedom of His own sovereign will.
• Presupposition no. 3: God's answer is always read in a visible medical parameter (heartbeats, biomarkers, length of stay), and never in peace of soul, conversion, forgiveness, the offering up of suffering.

None of these presuppositions is neutral. All of them amount to treating God as a causal machine — the “vending-machine theology” of Anglophone theological vocabulary^[9]. Polkinghorne expresses it in his own way, in substance: prayer is not a bundle of blank cheques drawn on a celestial Father Christmas (paraphrase of a recurring formula of the author, not verbatim)^[10].

Trent Horn: the three structural impossibilities

The Catholic apologist Trent Horn (Catholic Answers) has summed up the three structural impossibilities that make an intercession trial scientifically untestable^[11].

Impossibility no. 1 — the perfect control group does not exist. Scientific control requires total isolation of the subjects from the tested variable. No human protocol can isolate a hospitalized Western patient from all prayer — that of his family, his parish, his friends, or even the believing caregiver who silently murmurs an Our Father before the incision.

Impossibility no. 2 — the “subject” of the experiment is omniscient. God, according to the Christian faith, knows the ongoing protocol perfectly. Yet no scientific study can be valid if the “subject” knows it is being tested: it would by definition alter its behavior. Horn illustrates this with the analogy of a celebrity subjected to a fan-mail test: as soon as she knows she is being observed, she alters her response — or refuses to play, indignant at having her answer made an object of experiment (faithful summary, not verbatim)^[12].

Impossibility no. 3 — the interpretation of the results is undecidable. Whether the trial yields a positive result (an effect of prayer), a negative one (no effect) or a null one (variations indistinguishable from noise), none of the three readings truly touches the existence or the freedom of God. A positive effect could be a psychological placebo; a negative effect could mean that God refuses to play the protocol's game; a null effect may point to a multitude of causes. Science has no conceptual tools for deciding between these readings.

The illuminating analogy: testing friendship double-blind

To grasp the confusion, take a simple illustration, written for this article. Imagine wanting to prove scientifically the existence and efficacy of friendship. The protocol would be: randomize 1,000 people into two groups, ask 500 anonymous friends to send standardized affectionate messages (“I am thinking of you, I wish you well”) for two weeks to the patients in the treated group, and then measure the objective decrease in blood cortisol levels.

The experiment is technically feasible. It rigorously measures nothing of real friendship. Friendship is not a quantifiable force: it is a free relationship between persons. Whether cortisol falls or not, friendship remains what it is. Lewis draws the conclusion that applies word for word to prayer:

“Prayer is request. The essence of request, as distinct from compulsion, is that it may or may not be granted. (…) a compulsive empirical proof such as we have in the sciences can never be attained.” — C. S. Lewis, The Efficacy of Prayer, The Atlantic Monthly, January 1959^[13]

The STEP protocol's own statement of intent

Most strikingly, the authors of STEP, in their article published in the American Heart Journal, themselves acknowledge these limits — in one cautious sentence that popularization forgot:

“Private or public sources of background intercessory prayer that may have existed could have masked the effects of [our protocol's] provided prayer in this study.” — Herbert Benson et al., STEP, American Heart Journal, 2006^[14]

In other words: even the authors of the most cited negative study admit that background prayer — that of families, congregations, loved ones — inevitably contaminates all groups, including the control group^[15]. No human protocol can isolate a hospitalized Western patient from all prayer. Clinical trials therefore measure, at best, the possible marginal effect of an additional standardized prayer on top of the background noise — not the efficacy of prayer in itself.

A necessary clarification, since the reverse objection would be legitimate: this contamination does not render the comparison pointless. The trial does test something definite — the additional medical effect of a specific intervention, added on top of background prayer. What it cannot test is the supernatural efficacy of prayer itself: God remains free to answer or not a prayer instrumentalized into a protocol, and a clinical result — positive, null or negative — decides nothing of that freedom. Distinguishing the clinical result (a measured marginal effect) from supernatural efficacy (a free relationship) is not an escape hatch: it is the precondition for reading these trials honestly, in either direction.

The idea of statistically testing prayer is not a recent invention. It goes back to Francis Galton (1822–1911), cousin of Charles Darwin, pioneer of modern statistics — and inventor, incidentally, of the word eugenics^[16]. In August 1872, Galton published in the Fortnightly Review an article entitled “Statistical Inquiries into the Efficacy of Prayer” which counts as the first empirical study of the subject^[17].

The British sovereigns argument

Galton's reasoning is simple: if prayer is effective, then the people for whom the most prayer is said should enjoy the highest longevity. Yet, he observes, sovereigns are the men for whom several million subjects pray daily, for generations — “God save the Queen”^[18]. Eminent clergy, missionaries, religious officiants are likewise covered by countless public prayers.

Galton takes up a table by W. A. Guy covering already eminent men (listed in Chalmers's Biography) past the age of 30, period 1758–1843 — accidental and violent deaths excluded:

His provocative conclusion is well known:

“The sovereigns are literally the shortest lived of all who have the advantage of affluence. The prayer has therefore no efficacy. (…) The evidence I have been able to collect bears wholly and solely in that direction, and in the face of it the onus probandi lies henceforth on the other side.” — Francis Galton, Statistical Inquiries into the Efficacy of Prayer, 1872^[19]

Why the argument backfires

Galton's inquiry is fascinating as a historical document: it shows that as early as 1872, the idea of evaluating God by statistics seemed natural to a certain scientistic modernity. But the argument suffers from three fatal weaknesses that any contemporary statistician would identify within minutes^[20].

First, Galton controls for no confounding variable. Sovereigns live in a world of permanent political stress and congenital pathologies — the inbreeding of European royal families is notorious and well documented by medical genealogy^[21]. (Invoking assassinations and war deaths does not answer his sample anyway: his table excludes accidental and violent deaths.) Comparing their longevity to that of London lawyers amounts to comparing a tiny category, ultra-selected by birth, with a broad professional population.

Next, Galton himself quotes the prayer in question — “Grant her in health long to live”: it does ask for health and long life. His weakness is therefore not misquoting the request, but supposing that God would answer it mechanically, prorated by the number of people praying. No Christian theologian has ever upheld such a conception^[22].

Finally and above all, Galton presupposes what he claims to demonstrate: that God, if He exists, would act as a uniform cause triggered by the cumulative number of people praying. No Christian theologian, from St. Augustine to St. Thomas, has ever upheld such a conception^[23]. Prayer is not a heap of orisons forcing God to add so many hours of life to the recipient. Galton does not refute Christian doctrine: he refutes a caricature that no catechism ever taught.

An ideological legacy: from Victorian scientism to French-speaking skepticism

The 1872 inquiry nonetheless inaugurates a long tradition. For a century, the subject remained marginal. It resurfaced in 1988 with a modern randomized trial conducted in San Francisco, then became a structuring objection of French-speaking skeptical discourse from the 2010s onward, relayed on YouTube by various skepticism-popularization channels^[24]. The popularized objection — “science has proved that prayer does not work” — is never exactly what the studies say: it is an ideological extrapolation from a literature that is in reality heterogeneous, fragile and structurally biased.

In July 1988, cardiologist Randolph C. Byrd published in the Southern Medical Journal a prospective, randomized, double-blind trial that would lastingly mark the field. Two controlled trials had preceded it without impact — Joyce and Welldon (1965, 48 patients) and Collipp (1969, 18 leukemic children), which Byrd discusses in his introduction. The title is unambiguous: “Positive therapeutic effects of intercessory prayer in a coronary care unit population”^[25].

The San Francisco protocol

The study was conducted over ten months at San Francisco General Hospital^[26]. 393 patients admitted to the coronary care unit (CCU) were randomized into two groups: 192 were entrusted to the prayer of outside “born-again” Christian intercessors, 201 formed the control group. The intercessors received the patient's first name, but also his diagnosis and updates on his condition over the course of hospitalization. They prayed daily.

The announced results

Byrd concludes that the prayed-for group experienced, overall, a significantly more favorable hospital course than the control group. Six criteria stand out at the nominal threshold (with no correction for multiple comparisons): less congestive heart failure, fewer cardiopulmonary arrests, fewer pneumonias, less use of diuretics, antibiotics and intubation/ventilation.

“Based on these data, there seemed to be an effect, and that effect was presumed to be beneficial. Intercessory prayer to the Judeo-Christian God has a beneficial therapeutic effect on patients admitted to a CCU.” — Randolph C. Byrd, Southern Medical Journal, July 1988, p. 829^[27]

Why this study is not proof

The enthusiasm sparked by Byrd, notably in American Pentecostal circles, was quickly tempered by a more rigorous methodological reading, conducted notably by psychiatrist Richard P. Sloan (Columbia University) in his book Blind Faith: The Unholy Alliance of Religion and Medicine (St. Martin's Press, 2006)^[28]. Three decisive objections were raised:

1. The “sharpshooter's fallacy”: Byrd analyzed no fewer than 29 distinct clinical variables. With no statistical correction for multiple comparisons (the so-called Bonferroni correction, which consists in stiffening the significance threshold when tests are multiplied, so as not to mistake a stroke of luck for a real effect), finding a few significant variables “by chance” is unremarkable: with 29 tests at the 5% threshold, one expects on average 1.45 false positives with no real effect at all. The analysis recalls the shooter who draws the target after firing^[29].
2. No difference on the most solid endpoints: neither mortality, nor length of stay, nor the readmission rate differs significantly between the groups^[30]. These are the most demanding endpoints; but other clinically relevant events — particular complications, composite scores defined before the trial — can also measure a real effect. The relevance of an endpoint is judged trial by trial, not as a matter of principle.
3. Background prayer noise: Byrd did not measure how many control-group patients were already being prayed for by their family, parish or friends. In an American city where religious practice remains significant, this noise invalidates any clean interpretation^[31].

Byrd 1988 is therefore a suggestive but not demonstrative study. Above all, it opens the way to a research program that Anglo-Saxon religious foundations — first among them the John Templeton Foundation — would fund for twenty years^[32].

Eleven years later, in October 1999, William S. Harris and nine colleagues at the Mid America Heart Institute in Kansas City published in the Archives of Internal Medicine an expanded replication: 990 CCU patients, randomized double-blind, distant prayer for 4 weeks starting from the first name alone^[33].

A favorable composite score, but nothing on individual endpoints

Harris had a clinical course score drawn up before the trial began by three cardiologists and one internist, called MAHI-CCU (Mid America Heart Institute – Coronary Care Unit): a “composite” score, i.e. a single figure manufactured by adding together several different medical criteria — but not independently validated, which limits its scope. The main figures:

The authors' comment remains cautious:

“Remote, intercessory prayer was associated with lower CCU course scores. This result suggests that prayer may be an effective adjunct to standard medical care.” — William S. Harris et al., Archives of Internal Medicine, 25 October 1999, p. 2273^[34]

Thirteen critical letters published in the same journal

The article immediately triggered an avalanche of critical letters that the journal published as a block in its June 2000 issue — a rare event in the medical literature^[35]. The main objections:

• The MAHI-CCU score was constructed for the study before the trial, but without independent validation: its scope therefore remains limited, and Byrd's original score itself yielded nothing significant on those same data^[36].
• The most solid individual endpoints (mortality, length of stay) remain with no significant difference between groups — exactly as with Byrd^[37].
• Several letters note that the effect found is so small (0.78 points on the weighted score — 6.35 versus 7.13, i.e. −11%) that it would have no real clinical meaning, even if reproducible^[38].

To this is added an ethical objection: the study proceeded without informed consent, with an exemption granted by the ethics committee (IRB) in the name of preserving blindness — a waiver that Turner (2006) judges questionable out of respect for patients' convictions^[39].

Harris 1999 is therefore a weak replication of Byrd's signal: a borderline effect, on an unvalidated composite score, that does not hold on the most solid parameters, and without individual consent (with an IRB exemption, a waiver judged questionable by Turner).

The most disturbing episode in the scientific literature on prayer is also the least cited by skeptical popularizers. In September 2001, the Journal of Reproductive Medicine published a sensational article: “Does prayer influence the success of in vitro fertilization-embryo transfer? Report of a masked, randomized trial”, signed Kwang-Yul Cha (CHA General Hospital, Seoul), Daniel P. Wirth and Rogerio Lobo, then Chair of the obstetrics department at Columbia University^[40].

A study too good to be true

The announced protocol is spectacular: 219 Korean women undergoing in vitro fertilization are referred to the trial — 50 excluded for various reasons, 169 analyzed (88 prayed-for, 81 controls). Intercessors in the United States, Canada and Australia pray over photographs of patients. The announced results cause a sensation:

A doubling of the pregnancy rate, statistically significant at an unassailable level. The news travels around the world. Columbia proudly publicizes the study^[41].

Daniel Wirth, sentenced to five years in federal prison

As early as 2002, an investigation conducted by obstetrician-gynecologist Bruce Flamm (clinical professor at the University of California at Irvine) reveals the other side of the story^[42][43]. Co-author Daniel Wirth is not a physician: he holds a master's degree in parapsychology and a law degree. Worse: in October 2002, Wirth is indicted by a federal grand jury for fraudulent use of deceased persons' identities^[44]. On 18 May 2004, he pleads guilty to conspiracy to commit mail fraud and bank fraud in the Adelphia Communications case, involving 2.1 million dollars. On 22 November 2004, he is sentenced to five years in federal prison, three years of supervised release and over a million dollars in restitution (an Adelphia case, unrelated to the trial data, whose fabrication remains suspected but unproven).

Rogerio Lobo withdraws his name, Cha leaves Columbia

Around the same time, Rogerio Lobo publicly states that he learned of the study only six to twelve months after it was conducted — a fact confirmed as early as December 2001 by the OHRP (Office for Human Research Protections, DHHS) on Columbia's report — contributing only an editorial rereading^[45]. He leaves the chairmanship of Columbia's obstetrics department; his name is removed by erratum in October 2004.

Kwang-Yul Cha, the first author, also leaves Columbia. In 2007, he is caught in disguised republication: his Fertility and Sterility article of December 2005 (mitochondrial DNA and ovarian failure) reproduces, translated and under different author names, a Korean article of January 2004 drawn from Jeong-Hwan Kim's thesis — a journal that retracts the article (Los Angeles Times, 18 February 2007)^[46]. Bruce Flamm, professor of obstetrics-gynecology at the University of California (Irvine), showed that the protocol — despite the spectacular scale of the announced results (a doubling of the pregnancy rate) — did not withstand elementary methodological scrutiny^[47].

An article never retracted

The scandal takes an even more troubling turn: despite the criminal conviction — for a financial fraud unrelated to the trial — of one co-author and the formal withdrawal of another, the Journal of Reproductive Medicine never retracted the article (verified on PubMed, PMID 11584476: no retraction notice; only an October 2004 erratum removing Lobo's name). Flamm's critical letter, long refused, appeared only in January 2005^[48]. To this day, the 2001 Columbia study keeps appearing in some meta-analyses and circulating on superficial apologetics websites as “scientific proof of the efficacy of prayer.”

This file imposes a double reminder: integrity failures (a co-author's criminal conviction in an unrelated case, another's disguised republication) exist even in studies sometimes mobilized by apologists — as in the opposing literature; and rigor in citation is a Catholic requirement before being an academic one. An Ignispedia article mentions Cha-Wirth-Lobo 2001 only to refute it — never to rely on it.

In December 2001, at the same time as the Columbia scandal, Jorge Aviles and the cardiology team at the Mayo Clinic published in Mayo Clinic Proceedings a rigorously designed trial: 799 patients discharged from the CCU between July 1997 and October 1999, each entrusted to five intercessors praying at least once a week for 26 weeks^[49].

The primary composite endpoint — death, cardiac arrest, cardiovascular rehospitalization, revascularization, cardiac emergency visits — was reached by 25.6% of the prayer group versus 29.3% of the control group^[50]. The difference is not significant (OR 0.83; p = 0.25). None of the subgroup analyses (high risk, low risk) reaches the threshold. The authors soberly conclude:

“As delivered in this study, intercessory prayer had no significant effect on medical outcomes after hospitalization in a coronary care unit.” — Jorge Aviles et al., Mayo Clinic Proceedings, December 2001, p. 1192^[51]

The Mayo Clinic study is generally regarded as the best negative trial of the first generation: solid protocol, reasonable sample, long intercession duration (26 weeks), a team with no particular religious conflict of interest^[52]. If one negative study had to be cited, it would be this one, not STEP — whose informed group showed an excess of complications that the authors say they cannot explain.

In July 2005, The Lancet — the most prestigious British medical journal — published MANTRA II (Monitoring and Actualisation of Noetic Trainings), a 2×2 factorial multicenter trial conducted by Mitchell W. Krucoff (Duke University) on 748 patients undergoing elective angioplasty or catheterization in nine American centers^[53].

The protocol's originality: distant prayer is entrusted to several groups of intercessors from four traditions — Christian denominations, Jewish, Muslim and Buddhist — alongside another arm evaluating MIT therapy (music, imagery, touch)^[54]. After 11 September 2001, twelve second-tier groups are added to pray for the intercessors themselves (“higher dose of prayer”). This diversity explicitly pursues the idea of avoiding favoring one religion over others — a methodological choice that immediately renders the protocol theologically incoherent from a Christian, Jewish or Muslim standpoint, whose theologies of prayer differ radically.

The main result is unequivocal:

“Neither masked prayer nor MIT therapy significantly improved clinical outcome after elective catheterisation or percutaneous coronary intervention.” — Mitchell W. Krucoff et al., MANTRA II, The Lancet, vol. 366, 16 July 2005, p. 211^[55]

This verdict concerns the primary endpoint (in-hospital events). As a secondary finding, 6-month mortality is lower with MIT therapy (HR 0.35; p = 0.016) and two-tier prayer shows a non-significant favorable trend; simple masked prayer shows nothing. Note: 89% of patients knew someone was praying for them outside the protocol.

The editorial accompanying the article — entitled “MANTRA II: measuring the unmeasurable?” — stresses precisely the epistemological difficulty of testing prayer like a drug^[56].

Against the background of this contradictory literature, the John Templeton Foundation co-funds, at a total cost of 2.4 million dollars, the most ambitious trial ever conducted^[57]. Coordinated by Herbert Benson (Mind/Body Medical Institute, Harvard) and published in April 2006 in the American Heart Journal, the STEP trial (Study of the Therapeutic Effects of Intercessory Prayer) includes 1,802 patients undergoing heart surgery in six American hospitals, recruited from January 1998 to November 2000^[58].

The three-arm protocol

STEP's innovation is its three-arm structure, designed to distinguish the effect of prayer itself from the psychological effect of knowing one is prayed for:

The intercessors come from three groups: two Catholic communities — Saint Paul's Monastery (Saint Paul, Minnesota) and the Community of Teresian Carmelites (Worcester, Massachusetts) — and Silent Unity (Lee's Summit, Missouri), the prayer ministry of the Unity movement, stemming from American New Thought, and not a Christian Church^[59]. For 14 days, each added to his or her usual prayers the same intention — “a successful surgery with a quick, healthy recovery and no complications” — with durations varying widely by intercessor, from thirty seconds to several hours a day^[60].

The results: 52% / 51% / 59%

The primary endpoint — occurrence of a complication within 30 days — gives the following figures:

The figures bring out the decisive methodological point: the only clear gap concerns not concealed prayer, but the informed group that knew with certainty a prayer would certainly be applied to it. The three lessons to retain:

1. No significant difference between group 1 (prayed for unawares) and group 2 (not prayed for) — one point apart^[61]. Discreet, standardized intercessory prayer has, in this protocol, no measurable medical effect.
2. Group 3 — the one that knew with certainty it was being prayed for — shows more complications than group 1 (59% versus 52%, relative risk RR 1.14 — i.e. 14% more complications — with a 95% confidence interval of 1.02 to 1.28, that is the range compatible with the data at the usual threshold)^[62]. This was one of the two comparisons planned by the protocol, with a corrected threshold of 0.025. Faced with this result, the authors write that they have “no clear explanation” and consider a chance effect; the idea of “performance anxiety” comes from press commentary, not the article.
3. 30-day mortality and major events do not differ between the three groups^[63].

“Intercessory prayer itself had no effect on complication-free recovery from CABG, but certainty of receiving intercessory prayer was associated with a higher incidence of complications.” — Herbert Benson et al., STEP, American Heart Journal, vol. 151, no. 4, April 2006, p. 934^[64]

The limits acknowledged by the authors themselves

Three pages later, in the same publication, the authors themselves list the methodological difficulties (which media coverage largely passed over in silence)^[65]:

• Background prayer noise: “private or public sources of background intercessory prayer may have masked the effects of the provided prayer.”
• Standardization: “the provided prayer may have differed in content and intensity from the prayer usually offered by families and friends.”
• Duration: 14 days may be insufficient to measure a spiritual transformation.
• Seriously yet stably ill patients: all patients are at the preoperative stage of a scheduled surgical procedure — a situation where Western medical care already achieves a high success rate. The margin within which any additional effect could appear is minuscule.

The journalistic reading: the “dangerous prayer” effect

The media reprise of STEP — which retained the most spectacular result, the excess complications in group 3 — reflects a tendentious journalistic reading of a one-off result, itself open to alternative psychological interpretations^[66]. Benson's real conclusion is:

• group 1 vs. group 2: no effect of prayer (51–52% complications);
• group 3: an excess of complications that the authors themselves say they cannot explain (chance possible).

The gap between this technical conclusion and its alarmist reprise perfectly illustrates the framing bias to which negative religious studies fall victim in the general press: the most spectacular result is retained (adverse effect), the authors' caution is forgotten (no clear explanation, chance possible — not inverted providence), and it is turned into a militant hook.

From 2006 onward, the critical mass of published trials allows statistical synthesis. Three major meta-analyses now structure the field.

Hodge 2007: a small but significant effect

David R. Hodge (Arizona State University) published in March 2007 in Research on Social Work Practice a rigorous meta-analysis including the main available studies. The overall result: a small but statistically significant effect in favor of prayer (g = −0.171 — g measures the effect size, here small — with p = 0.015 — p measures how far the data depart from the no-effect hypothesis, below the usual 5% threshold, not the probability that the result is due to chance)^[67].

“The meta-analysis showed a small but significant effect size for the use of intercessory prayer (g = −0.171, p = 0.015).” — David R. Hodge, Research on Social Work Practice, vol. 17, no. 2, March 2007, p. 174^[68]

Hodge nevertheless himself classifies intercessory prayer in the “experimental intervention” category (per Division 12 criteria of the American Psychological Association), i.e. not yet evidence-based^[69]. He recalls that well-conducted trials, such as STEP, are not significant, but that the overall body of data remains fragile and probably slightly favorable.

Masters et al. 2006: no effect apart from the controversial study

Conversely, Kevin Masters (University of Colorado) published in August 2006 in the Annals of Behavioral Medicine a random-effects meta-analysis of 14 controlled studies^[70] — a few months before Hodge's competing synthesis (March 2007). Verdict: overall effect g = 0.100, not different from zero. Excluding the controversial 2001 Cha-Wirth-Lobo study (sensitivity analysis), the effect falls to g = 0.012.

“There is no scientifically discernable effect for IP as assessed in controlled studies. (…) we recommend that further resources not be allocated to this line of research.” — Kevin S. Masters et al., Annals of Behavioral Medicine, vol. 32, no. 1, August 2006, p. 21^[71]

Cochrane 2009: “neither recommend nor advise against”

The most respected arbitration in the medical field belongs to the Cochrane Collaboration, an independent organization for systematic meta-analyses. Its review “Intercessory prayer for the alleviation of ill health”, updated in April 2009, includes 10 randomized trials totaling 7,646 patients^[72].

The overall results:

Of the 33 complications examined, three marginally reach significance (cardiac arrest, major surgery, Swan-Ganz) — a figure compatible with chance alone for 33 uncorrected comparisons (about 1.7 expected on average), not proof of an effect^[73]. The final conclusion is quoted verbatim:

“These findings are equivocal and (…) the evidence does not support a recommendation either in favour or against the use of intercessory prayer. We are not convinced that further trials of this intervention should be undertaken.” — Roberts, Ahmed, Davison, Cochrane Database of Systematic Reviews, 2009^[74]

The review adds that it is not convinced further trials of this intervention are worthwhile, preferring to see resources allocated to other health questions — a resource-management opinion, not proof of inefficacy.

It is this verdict — equivocal — that French-speaking skepticism improperly translates as “science has proved that prayer does not work.” The Cochrane review is more modest: it says that the data allow no conclusion, in either direction.

Beyond the statistical verdict, distant-intercession trials suffer from structural limits that the critical literature has documented precisely. These limits are not apologetic excuses: they are published in the world's greatest medical journals.

The background-prayer problem

This is the most implacable objection, formulated notably by Richard P. Sloan (Columbia) in The Lancet as early as 1999, and taken up in Perspectives in Biology and Medicine in 2006^[75][76]. Every Western patient hospitalized for coronary bypass is, almost by definition, the object of non-randomized prayers: his family prays, his friends pray, his parish puts him on the Sunday intentions, the Catholic nurse murmurs inwardly, the Christian surgeon crosses himself before the incision^[77]. In the United States as in France, the background noise of prayer is massive and uncontrollable.

Sloan calculated that with a background-to-protocol prayer ratio of 3 to 1, 7,132 patients per group would have to be recruited to detect statistically an effect of size g = 0.2 — more than 14,000 patients in total, beyond the reach of any realistic medical trial on the subject^[78]. Even in the more favorable scenario where background prayer equals protocol prayer (1:1 ratio), 766 patients per group would still be needed. These figures hold under those assumptions (effect size, usual thresholds): they show the practical unrealism of a perfectly controlled trial, not a mathematical impossibility.

In other words: the perfectly controlled trial is practically unfeasible. No “prayer-free” group can exist without isolating the patient from all human society — which would violate every ethical rule of medical research.

Composite scores and multiple comparisons

As Sloan and Ramakrishnan showed, several large studies test many variables without systematic correction for multiple comparisons^[79]. Byrd tested 29 variables, Harris 35 adverse events (Table 3 of the article, excluding composite scores and lengths of stay). Without correction (Bonferroni), finding a few significances is mechanical: for 29 comparisons at α = 5%, one expects on average 1.45 positive results under the null hypothesis — an expectation, not an exact number^[80]. Two clarifications are nevertheless in order: Harris's weighted score had been drawn up before the trial, and STEP planned two primary comparisons with a corrected threshold of 0.025. The critique of after-the-fact tinkering therefore does not apply indistinctly to all studies.

A standardized prayer is not a Christian prayer

The STEP protocol is explicit: intercessors receive the instruction to pray “for a successful surgery with a quick, healthy recovery and no complications”^[81]. Is this formula theologically adequate? A serious Christian does not ask God only for a clinical outcome; he asks for faithfulness in trial, inner peace, forgiveness, the good use of illness as a path of sanctification. Reducing prayer to a medical formulation is already to have truncated it at the root.

“And in praying do not heap up empty phrases as the Gentiles do; for they think that they will be heard for their many words. Do not be like them, for your Father knows what you need before you ask him.” — Jesus Christ, Gospel according to St. Matthew 6:7–8^[82]

The role of the John Templeton Foundation

A little-commented fact in France: the STEP trial (2.4 million dollars, funded mostly by the John Templeton Foundation — created in 1987 by the Presbyterian investor John M. Templeton (1912–2008) — with hospital co-funding from the Baptist Memorial Health Care of Memphis)^[83] is the only prayer trial whose Templeton funding is documented down to the dollar. The foundation (an endowment of about 3.4 billion dollars, some 134 million in annual grants across all programs) is neither the established funder of Byrd 1988, Harris 1999 or Aviles 2001, nor the sole funder of the field: the American federal government has devoted more than 2.3 million dollars to prayer research since 2000^[84].

This is not an argument against the scientific quality of the studies — Templeton also funds research in cosmology and philosophy without interfering^[85]. But no quantified synthesis, with denominator and period, supports claiming that world literature was built “90%” by and for a religious foundation: that share is undocumented.

Just another drug? — Turner's 2006 philosophical critique

In August 2006, in the Journal of Medical Ethics, philosopher Derek D. Turner (Connecticut College) published a systematic critique of randomized prayer trials under the programmatic title: “Just another drug? A philosophical assessment of randomised controlled studies on intercessory prayer”^[86]. His thesis in one sentence: prayer cannot be studied like a drug, and doing so is both epistemologically vain and ethically questionable.

Turner identifies three structural flaws complementary to Sloan's:

• The consent and volunteer-bias flaw: several studies (Harris notably) proceeded without consent, with an ethics-committee exemption — a waiver Turner judges questionable out of respect for patients' religious convictions and because of possible adverse effects. Patients who agree to participate moreover form a group self-selected by its spiritual receptivity.
• The subversion flaw: consistent believing intercessors cannot help praying also for the control group (“slipping in an extra request: “Lord, take care of those in the control group too””), which empties the trial of meaning — all the more so since background prayer is inevitable.
• The internal-tension flaw: the scientist reasons correlation-first, while the intercessor reasons causation-first; demanding the former posture from intercessors renders the experiment conceptually incoherent.

“It therefore makes no ethical or methodological sense to study remote, intercessory prayer as if it were just another drug.” — Derek D. Turner, Journal of Medical Ethics, vol. 32, no. 8, August 2006, p. 487^[87]

Turner does not say prayer is ineffective: he says that the randomized-trial format is not the appropriate epistemological tool for measuring it. That is exactly the thesis of Polkinghorne, Lewis, Trent Horn — and of St. Thomas Aquinas.

Beyond the limits of the studies themselves, the popular objection “science has proved the inefficacy of prayer” rests on a chain of characteristic cognitive biases. Naming them allows a point-by-point answer.

“Cherry-picking” the negative results

Some skeptical popularizers retain above all STEP 2006 as THE scientific “proof,” and forget or minimize:

• Byrd 1988, Harris 1999 and Hodge 2007 — slightly favorable results;
• the equivocal conclusion of Cochrane 2009 — which decides in neither direction;
• the methodological critiques published in The Lancet, JAMA, the Journal of Medical Ethics — which invalidate the very claim to conclude^[88][89].

Cherry-picking consists in selecting the data that confirm the preferred hypothesis and discarding those that contradict it^[90]. It is a well-identified sophism in scientific methodology — and it is found in several skeptical reworkings of this file, which retain STEP while passing over Byrd, Harris or Cochrane's equivocal verdict in silence.

Tendentious journalistic reading

The 2006 media coverage of STEP illustrates this framing risk: reports retained the most spectacular result (the excess complications in the informed group) while leaving aside the authors' caution^[91]. A “counter-intuitive” result (prayer would be dangerous!) travels faster than an equivocal result (the data allow no conclusion).

The category confusion: proving non-existence by absence of effect

The underlying objection assumes that “if prayer were effective, it could be measured.” But this is circular reasoning: it presupposes precisely what is under debate — namely that prayer belongs to measurable physical causality. If it belongs to another order of reality (a free interpersonal relationship with a transcendent God), then its absence in a randomized protocol proves nothing more than the absence of a fine wine's taste in a chromatographic analysis.

The philosopher Alvin Plantinga (Notre Dame) defends a neighboring thesis: theological statements form a proper epistemological category, not reducible to measurable empirical statements — judging one reality with the instruments of another order, like weighing a color on a scale^[92].

The Popperian falsifiability criterion

Karl Popper (1902–1994) laid down that the mark of a scientific claim is its falsifiability: for a statement to be scientific, one must be able to imagine an observation that would contradict it^[93]. But is the statement “Christian prayer is effective” falsifiable in the Popperian sense?

The honest answer is no — not by default, but because this statement never claimed to be a scientific one. It is a theological claim about a free relationship between free persons. Just as the statement “Peter loves Mary” is not falsifiable by a cortisol assay, the statement “God hears His children's prayer” is not falsifiable by a randomized trial on the 30-day cardiac complication rate.

The skeptical objection thus commits an inversion: it reproaches prayer for not being scientific, when prayer never claimed to be. This reproach amounts to asking a mathematical theorem to bear fruit or a work of art to prove its truth double-blind.

The methodological limits exposed by the skeptics themselves — Sloan, Turner, Flamm, Masters — converge toward a more radical conclusion, which Christian theology has stated for twenty centuries: a randomized trial cannot, by nature, measure Christian prayer. Not because prayer is ineffective, but because it is not of the same order of reality as a drug.

St. Thomas Aquinas: praying does not change God, praying orients us toward Him

The classic answer of Catholic theology goes back to St. Thomas Aquinas (1225–1274), in the Summa Theologiae, IIa-IIae, question 83 “De oratione”^[94]. Question 83 is precisely that of a 13th-century skeptic: what is the use of praying, if God already knows everything, and has already decided from all eternity what must happen?

St. Thomas's answer — the Respondeo of article 2 — has become the canonical formula of Catholic teaching on prayer:

“We pray, not that we may change the Divine disposition, but that we may impetrate that which God has disposed to be fulfilled by our prayers.” — St. Thomas Aquinas, Summa Theologiae, IIa-IIae, q. 83, a. 2^[95]

The reasoning is luminous: divine Providence has decided not only which effects would occur in history, but also by which causes and in which order. Human prayer is one of those secondary causes willed by God from all eternity^[96]. To pray is therefore not to twist God's arm: to pray is precisely the means by which God, from all eternity, decided to give us certain goods.

The objection “if God has already decided everything, praying is useless” rests on a logical confusion between the order of ends and the order of causes. St. Thomas dissolves it in one paragraph.

Garrigou-Lagrange: prayer as a willed secondary cause

In the 20th century, the Dominican theologian Réginald Garrigou-Lagrange (1877–1964) — one of the greatest Thomist commentators — updated this doctrine in his treatise Providence (English edition: Providence, Herder, 1937; French original La Providence et la confiance en Dieu, Desclée de Brouwer, 1932)^[97]. His synthetic formula deserves meditation:

According to Garrigou-Lagrange, God's will is absolutely immutable, and it is precisely in this immutability that the ultimate guarantee of prayer's efficacy is found: far from opposing it, divine immutability vouches for it (paraphrase)^[98]. True prayer — offered with the required conditions (asking for oneself, the goods necessary for salvation, with piety and perseverance, according to St. Thomas, IIa-IIae q. 83 a. 15, and the promise “Ask, and it will be given you,” Mt 7:7) — is infallibly efficacious because God so decreed and does not revoke what He decreed (paraphrase)^[99].

Garrigou-Lagrange offers an agricultural comparison of great clarity: prayer stands to the obtaining of spiritual gifts as seed stands to harvest — no harvest without sowing, no interior life without prayer: God prepared a material seed for the material harvest, and a spiritual seed, including prayer, for the spiritual harvest (paraphrase)^[100]. No one says it is useless to sow because grain grows according to the laws of nature willed by God: likewise, prayer is the concomitant act by which man enters into the providential plan.

St. Alphonsus Liguori: “The great means of prayer”

The Doctor of the Church St. Alphonsus Liguori (1696–1787) devoted an entire treatise — The Great Means of Prayer (1759) — to teaching that prayer is not one means among others of obtaining salvation, but the necessary means willed by God to give His graces^[101]. His central argument reverses the question:

“He who prays will be saved; he who does not pray will be damned.” — St. Alphonsus Liguori, The Great Means of Prayer, 1759 (famous formula summing up his thesis)^[102]

The whole difference between the saints and the damned, he teaches, is that the damned do not pray and the saints pray (paraphrase).

For St. Alphonsus, asking God for scientific proof of prayer's efficacy before praying is to invert exactly the movement of grace: grace comes through prayer; refusing prayer in the name of doubt is refusing the channel through which God decided to answer.

St. Augustine: praying to dilate our desire

Eight to nine centuries before St. Thomas, St. Augustine (354–430) had already formulated the same intuition in his Letter 130 to Proba, written around 411–412^[103]. Proba, a wealthy Roman widow, asks him why one must pray since God knows our needs. Augustine's answer is a masterpiece of spirituality:

“For our Lord and our God does not need us to make our wishes known to Him, for He cannot be ignorant of them; but He wishes our desire to be exercised in prayer, that we may be able to receive what He is preparing to give.” — St. Augustine, Letter 130 to Proba, ch. VIII, § 17^[104]

Praying serves not to inform God: God knows. Praying serves not to force God: God is free. Praying serves to transform the one who prays himself — to make him desire more deeply what God wills to give him, to widen the receptacle of his soul so that it can receive grace.

“In this faith, hope, and love, we pray always, with unceasing desire.” — St. Augustine, Letter 130 to Proba^[105]

If such is the real object of Christian prayer — the inner transformation of the one who prays, the dilation of desire, conformity to God's will — then a randomized trial measuring the 30-day CCU readmission rate does not measure what it claims to measure. It asks the wrong question of the wrong variable.

“You shall not tempt the Lord your God”: Dt 6:16 and Mt 4:7

Beyond the philosophical question, Sacred Scripture passes an even more radical judgment on clinical intercession trials. In the book of Deuteronomy, Moses reminds the people of Israel of the absolute rule of their relationship with God:

“You shall not put the LORD your God to the test, as you tested him at Massah.” — Deuteronomy 6:16^[106]

The Hebrew uses the verb nāsāh — to test, to try, to put to the proof^[107]. The context is enlightening: at Massah, in the desert, the Hebrews summoned God to supply them with water in order to verify whether He was truly with them (Ex 17:1–7). This demand is not a trusting supplication: it is an experimental testing of divine faithfulness. And Scripture calls it a grave sin.

The verse takes on its full scope in the scene of the temptation in the desert (Mt 4:1–11; Lk 4:1–13), where Satan proposes to Jesus the exact scenario of a scientific study of prayer:

“Jesus said to him, ‘Again it is written, You shall not tempt the Lord your God.’” — Gospel according to St. Matthew 4:7^[108]

The diabolical proposal consists in asking Jesus to throw Himself from the pinnacle of the Temple in order to verify whether the angels will catch Him, as Psalm 91 promises^[109]. It is an experimental protocol: intervention (the leap), prediction (angelic intervention), empirical verification. Christ refuses by quoting Deuteronomy. The very scenario of a distant-intercession trial — verifying whether God intervenes on command for measurable medical outcomes — is described by Jesus as tempting God, a sin refused at the outset of His public ministry.

This does not mean that researchers conducting a study like STEP are driven by bad intent: subjective responsibility depends on intention, context, formation. But it does mean that the theological logic of the protocol — verifying whether God intervenes on command for measurable outcomes — reproduces the logic of the second temptation in the desert: putting God to the test in order to verify Him. This correlation with modern randomized trials is a contemporary reasoning proposed here on the basis of the doctrine set out: neither St. Thomas (q. 83, a. 2, which treats of Providence and prayer) nor St. Alphonsus pronounced on a modern clinical protocol.

The prayer of Gethsemane

The absolute counter-model is found a few hours before the Passion. In the garden of Gethsemane, Jesus prays to His Father in the most perfect way a human can pray — and His literal request is not granted:

“My Father, if it be possible, let this cup pass from me; nevertheless, not as I will, but as thou wilt.” — Gospel according to St. Matthew 26:39^[110]

If the measure of prayer's efficacy were “obtaining the content of the request,” then the most perfect prayer in history — Christ's at Gethsemane — would be classified by STEP as ineffective. Christ asked that the cup pass. It did not pass. The crucifixion took place. But that prayer accomplished the redemption of the world, the central event of history according to the Christian faith^[111]. The medical protocol of intercession trials has no means of measuring what was at stake that night.

The Catechism of the Catholic Church promulgated by John Paul II in 1992 devotes its entire fourth part (§§ 2558–2865) to prayer^[112]. Eight paragraphs (§§ 2734 to 2741) deal directly with the objection on which this article is built: why do so many prayers seem to go unanswered?

The difficulties of prayer (CCC §§ 2734–2741)

The text confronts the difficulty head-on, in a spirit with nothing defensive about it:

“Filial trust is tested – it proves itself – in tribulation (cf. Rom 5:3–5). The principal difficulty concerns the prayer of petition, for oneself or for others in intercession. Some even stop praying because they think their petition is not heard.” — Catechism of the Catholic Church, § 2734^[113]

The diagnosis of the following paragraph is even sharper — and touches directly on the presuppositions of clinical trials:

“One aspect of this is helpful to consider. When we praise God or give him thanks for his benefits in general, we are not particularly concerned whether or not our prayer is acceptable to him. On the other hand, we demand to see the results of our petitions. What is the image of God that motivates our prayer: an instrument to be used? or the Father of our Lord Jesus Christ?” — Catechism of the Catholic Church, § 2735^[114]

This questioning of the Catechism anticipates, fourteen years in advance, the objection built on STEP: if prayer produces no measurable results, then God does not exist or prayer is ineffective. The Catechism's answer is unambiguous: this reasoning treats God as a means to be used. It is not the Christian's attitude toward his Father.

“Ask according to His will” — 1 John 5:14

St. John's epistle explicitly conditions the promise of being heard:

“And this is the confidence which we have in him, that if we ask anything according to his will he hears us.” — First Letter of St. John 5:14^[115]

The biblical promise of being heard is therefore structurally conditioned on the conformity of the request to God's will^[116]. This does not mean one must guess that will before praying: we pray precisely in order to enter into that will. But it does mean that any “imperative formulation” — Lord, if You exist, reduce this patient's 30-day complication rate — falls outside the evangelical conditions of the promise.

C. S. Lewis's category mistake

The philosopher C. S. Lewis, in his essay The Efficacy of Prayer and his posthumous book Letters to Malcolm, formulated in contemporary terms what the Catholic tradition had always said^[117]. His thesis holds in one distinction: request versus constraint.

“Religious people don't talk about the ‘results’ of prayer; they talk of its being ‘answered’ or ‘heard’. (…) A suitor wants his suit to be heard as well as granted.” — C. S. Lewis, Letters to Malcolm, letter 10 (X), 1964^[118]

Lewis applies here, without naming it, what the philosopher Gilbert Ryle called the category mistake: confusing a relational act with a measurable physical effect^[119]. Prayer belongs to the category of free relational acts between persons — like trust, forgiveness, love. None of these acts is measurable by a double-blind randomized trial. That does not mean they do not exist: it means the measuring instrument is the wrong one.

If the Catholic Church refused on principle any scientific verification of phenomena linked to prayer, the skeptical objection would carry weight. Yet the Church does precisely the opposite: at Lourdes, the scene of the 1858 apparitions, it instituted — from 1883 onward, a quarter of a century later — a Medical Bureau open to physicians of all faiths to examine reputedly miraculous cures. This institution is exactly the counter-model of STEP-style trials: it examines proven individual cases, not group statistics on standardized protocols.

The Lourdes Medical Bureau: 1883, Pius X 1905

The Bureau of Medical Observations was created in 1883 by physicians of the Lourdes sanctuary^[120]. In 1905, the Holy See confirmed to the Bishop of Tarbes his right to use the procedures of the sanctuary's Medical Bureau to study reported cures^[121]. In 1947, the Bureau became the “Lourdes Medical Bureau” and Bishop Théas, Bishop of Tarbes and Lourdes, created a National Medical Committee; in 1954, an International Medical Committee (CMIL) succeeded it to provide external scientific expertise^[122].

Today, the Medical Bureau welcomes physicians, nurses, physiotherapists and pharmacists of all religious faiths or none^[123]. The Bureau was headed until 2026 by Dr. Alessandro de Franciscis, an Italian pediatrician, the first non-Frenchman in the post (2009–2026); Dr. Giada Monami, a physician at the civil hospital of Venice appointed in February 2026, succeeded him in mid-April 2026^[124].

Seven criteria (modern presentation after Lambertini): instantaneous, complete, lasting, inexplicable cure

For a cure to be recognized as “medically inexplicable” — the preliminary step before any ecclesial recognition as a miracle — it must satisfy seven criteria, in the modern presentation usual at Lourdes, inspired by the codification of Cardinal Prospero Lambertini (the future Benedict XIV) in his treatise De servorum Dei beatificatione et beatorum canonizatione^[125]:

Seventy-two recognized cures, thousands documented

Out of more than 7,000 reported cure files at the Medical Bureau since 1858, according to the Lourdes sanctuary^[126], only 72 cures have been medically certified as inexplicable and then formally recognized as miracles — not by the Bureau, which rules only on medical inexplicability, but by the miraculé's diocesan bishop at the end of a demanding canonical process^[127]. The last two: the 71st, that of British soldier John Traynor (healed at Lourdes in 1923, miracle declared by the Archbishop of Liverpool on 8 December 2024); and the 72nd, that of Italian Antonietta Raco (healed of primary lateral sclerosis in 2009, announced by rector Michel Daubanes then proclaimed by the Bishop of Tursi-Lagonegro on 16 April 2025)^[128].

“Lourdes is the only pilgrimage site in the world, of all religions combined, to have entrusted the validation of its miracles to the medical domain.” — Lourdes Medical Bureau, official communication, Lourdes sanctuary^[129]

Why Lourdes works where STEP fails

The contrast with STEP-style trials is total and instructive:

What works at Lourdes is not a statistical measurement of God: it is a rigorous medical investigation of individual cases, conducted by physicians of all faiths, following criteria codified three centuries ago. It is the inverse model of the STEP trial — and it is precisely the one the Church retained for discerning cures, the Bureau ruling on medical inexplicability and the bishop on the miracle^[130].

One last clarification is needed. The failure of distant-intercession trials says nothing about the other field — far more solid — of research on personal prayer and religious commitment as health factors. On that ground, the literature is massive and largely convergent.

Harold Koenig and the Handbook of Religion and Health

The Handbook of Religion and Health (Oxford University Press, 2nd edition, 2012), coordinated by Harold G. Koenig, psychiatrist and director of the Center for Spirituality, Theology and Health at Duke University, reviews more than 3,000 studies on the links between religion, spirituality and health^[131]. The most replicated associations — correlations, not demonstrated causalities:

• people regularly attending a service show, at equal age, lower mortality than non-practitioners (association observed in the cohort studies reviewed);
• personal prayer and practice are associated with less depression, anxiety and addiction (association);
• religious commitment is correlated with better post-surgical recovery (correlation);
• several studies observe slightly better immunity and greater longevity among practicing believers (association).

Koenig is professor of psychiatry at Duke and directs the Center for Spirituality, Theology and Health; the Handbook is a quantitative synthesis of nearly 3,000 studies^[132]. He rigorously distinguishes two objects:

Why this literature is compatible with the Catholic faith

The mechanisms Koenig identifies to explain the positive effect of personal prayer are psychobiological: stress reduction, meaning given to illness, community social support, lifestyle hygiene, neuroendocrine regulation^[133]. None of these mechanisms supposes direct supernatural intervention.

The Catholic theologian welcomes this without difficulty: God created man as a body-soul unity; grace does not destroy nature, it elevates it (gratia non tollit naturam sed perficit, says St. Thomas)^[134]. That personal prayer also acts through natural psychobiological channels is neither surprising nor degrading: it confirms that the human being was created open to God.

Practical conclusion: nothing in science invites us to stop praying for ourselves or our loved ones. Science invites at most recognizing that a randomized trial on the prayer of unknown strangers at a distance is an exotic epistemological object, whose equivocal results touch ordinary Christian practice not at all.

No randomized trial can replace the voice of the saints on what Christian prayer is. A few words from the Church's greatest spiritual figures put the scientific debate back into perspective.

Saint John Mary Vianney: “prayer lifts the soul”

The holy Curé of Ars (1786–1859), patron of priests, spent hours daily on his knees in his church and recognized in prayer the motor of all spiritual life^[135]:

“Prayer disengages our soul from matter, and elevates it on high, even as the fire inflates balloons and causes them to ascend.” — Saint John Mary Vianney, Thoughts of the Curé of Ars^[136]

No STEP protocol measures this elevation of the soul: which does not mean it does not take place, but that the measuring tool is fundamentally unsuited.

Saint Teresa of Avila: prayer as friendship

Saint Teresa of Avila (1515–1582), Doctor of the Church and reformer of Carmel, defined in the Book of Her Life (ch. 8, written 1562–1565) what Catholic mental prayer is^[137]:

“Mental prayer in my opinion is nothing else than an intimate sharing between friends; it means taking time frequently to be alone with Him who we know loves us.” — Saint Teresa of Avila, Book of Her Life, chapter 8, 1565^[138]

The image is decisive: prayer is a commerce of friendship. What randomized trial claims to measure the efficacy of a friendship double-blind?

Saint Thérèse of Lisieux: “a simple glance toward Heaven”

Little Saint Thérèse of Lisieux (1873–1897), Doctor of the Church, formulates in her Manuscript C the best-known definition of prayer in modern spirituality:

“For me, prayer is a surge of the heart; it is a simple look turned toward heaven, it is a cry of recognition and of love, embracing both trial and joy; it is something great, supernatural, which expands my soul and unites me to Jesus.” — Saint Thérèse of Lisieux, Autobiographical Manuscripts, C, 25 r°–v°^[139]

This definition became paragraph 2558 of the Catechism of the Catholic Church^[140]. It says the essential: prayer is not a technique, nor a measurable force, nor a protocol — it is a surge, a glance, a cry.

The Eternal Father to Saint Catherine of Siena

In the 14th century, Saint Catherine of Siena (1347–1380), Doctor of the Church, received in the Book of Dialogues this word of the Father on prayer:

“I am constrained to do so by that inestimable love of Mine, by which I created them, and by the love, desire, and grief of My servants, since I am no despiser of their tears, and sweat, and humble prayers…” — The Eternal Father to Saint Catherine of Siena, Book of Dialogues, chapter IV, § 10^[141]

Jesus to Luisa Piccarreta on the humility of prayer

In the 20th century, the Italian mystic Luisa Piccarreta (1865–1947), Servant of God, received this complementary message:

“No grace leaves My Throne, except through humility. No request can be received by Me, if it does not bear the signature of humility. (…) No prayer is heard by My Ears nor moves My Heart to compassion, if it is not perfumed with humility.” — Jesus to Luisa Piccarreta, The Book of Heaven, volume 3, 12 January 1900^[142]

The condition for the efficacy of Christian prayer is never quantity or standardization — it is the humility of the one who prays. A moral quality that no randomized trial measures. The Divine Will — the third Fiat, that is, the gift of living in God's Will revealed to Luisa Piccarreta — gives this requirement its full scope: to pray is to conform one's will to the Father's, not to accumulate formulas.

The objection “science has proved that prayer does not work” systematically confuses three levels that must be distinguished to answer honestly. The following synthesis can serve as a clear answer to a skeptic of good faith.

Medical level

The available clinical trials have not demonstrated a stable, reproducible medical effect of distant intercessory prayer on hard endpoints (mortality, major complications, length of stay). But neither have they demonstrated the reverse. The Cochrane 2009 verdict is explicit: “The evidence does not support a recommendation either in favour or against the use of intercessory prayer.”^[143] Any sharper claim — “science has proved inefficacy” — is ideological extrapolation.

Methodological level

A distant-intercession trial protocol is structurally reductive:

A perfectly controlled trial is practically unfeasible: Sloan showed that more than 14,000 patients would have to be recruited to detect, under his assumptions, a small-sized effect, beyond the reach of any realistic study^[144]. And the philosopher Turner showed that even such a giant trial would not be conceptually valid: prayer cannot be studied “as just another drug”^[145].

Theological level

This is the decisive level. God is not an experimental variable. Christian prayer is not a doseable natural force, nor a drug dispensed on prescription, nor a reproducible causal mechanism. It is a free personal request addressed to a free Person, who answers according to His Wisdom and our real good, sometimes granting, sometimes deferring, sometimes refusing in order to give better^[146]. St. Thomas, Augustine, Garrigou-Lagrange, the Catechism, Lewis all say the same thing in different languages.

And above all, by the same contemporary reasoning applied to modern trials: attempting to verify whether God intervenes on command in a randomized protocol reproduces the second temptation in the desert, which Christ rejects by quoting Deuteronomy — “You shall not tempt the Lord your God”^[147]. The theological logic of the STEP trial is, despite its authors' possible good faith, structurally the logic of putting to the test.

The objection “studies have proved the inefficacy of prayer” therefore does not have the scope attributed to it. It proves nothing in the sense of strong data against prayer. It reveals, on the other hand, a metaphysical presupposition: that every reality — including the relationship with God — should be measurable by a randomized trial to deserve to be held true. This presupposition has a name: scientism^[148]. It is itself neither scientifically proved nor provable, and it collapses as soon as it is applied to itself.

The right question is not: “have studies proved the efficacy of prayer?” — science has no tool for that.

The right question is: “is Christian prayer, as the Gospel and tradition define it, a free relationship to a Father God, asking for the conformation of our will to His, in filial trust?” — to that, neither Byrd, nor Harris, nor STEP, nor Cochrane can answer. Only the spiritual life can answer. And it has answered for twenty centuries, in the silence of monastic cells, in the perseverance of tried families, in the amazement of the Lourdes Bureau's physicians before files they cannot explain.

“For me, prayer is a surge of the heart; it is a simple look turned toward heaven, it is a cry of recognition and of love, embracing both trial and joy; it is something great, supernatural, which expands my soul and unites me to Jesus.” — Saint Thérèse of Lisieux, Autobiographical Manuscripts, C, 25 r°–v°^[149]

• The problem of atheism and skepticism
• The miracles of Lourdes
• Discerning supernatural phenomena
• The proofs of God's existence
• The problem of evil and suffering
• Eucharistic miracles
• If God knows everything in advance, how can we have free will?

Principal sources

• F. Galton, “Statistical Inquiries into the Efficacy of Prayer”, Fortnightly Review, vol. 12, August 1872, p. 125–135.
• R. C. Byrd, “Positive therapeutic effects of intercessory prayer in a coronary care unit population”, Southern Medical Journal, vol. 81, no. 7, July 1988, p. 826–829.
• W. S. Harris et al., “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, vol. 159, no. 19, 25 October 1999, p. 2273–2278.
• K. Y. Cha, D. P. Wirth, R. A. Lobo, “Does prayer influence the success of in vitro fertilization-embryo transfer? Report of a masked, randomized trial”, Journal of Reproductive Medicine, vol. 46, no. 9, September 2001, p. 781–787.
• J. M. Aviles et al., “Intercessory prayer and cardiovascular disease progression in a coronary care unit population”, Mayo Clinic Proceedings, vol. 76, no. 12, December 2001, p. 1192–1198.
• M. W. Krucoff et al., “Music, imagery, touch, and prayer as adjuncts to interventional cardiac care: the Monitoring and Actualisation of Noetic Trainings (MANTRA) II randomised study”, The Lancet, vol. 366, no. 9481, 16 July 2005, p. 211–217.
• H. Benson et al., “Study of the Therapeutic Effects of Intercessory Prayer (STEP) in cardiac bypass patients”, American Heart Journal, vol. 151, no. 4, April 2006, p. 934–942.
• K. S. Masters, G. I. Spielmans, J. T. Goodson, “Are there demonstrable effects of distant intercessory prayer? A meta-analytic review”, Annals of Behavioral Medicine, vol. 32, no. 1, August 2006, p. 21–26.
• D. R. Hodge, “A Systematic Review of the Empirical Literature on Intercessory Prayer”, Research on Social Work Practice, vol. 17, no. 2, March 2007, p. 174–187.
• L. Roberts, I. Ahmed, A. Davison, “Intercessory prayer for the alleviation of ill health”, Cochrane Database of Systematic Reviews, no. 2, April 2009, CD000368.
• R. P. Sloan, Blind Faith: The Unholy Alliance of Religion and Medicine, St. Martin's Press, New York, 2006, ISBN 978-0-312-34881-6.
• D. D. Turner, “Just another drug? A philosophical assessment of randomised controlled studies on intercessory prayer”, Journal of Medical Ethics, vol. 32, no. 8, August 2006, p. 487–490.
• H. G. Koenig, D. E. King, V. B. Carson, Handbook of Religion and Health, 2nd ed., Oxford University Press, 2012, ISBN 978-0-19-533595-8.
• J. C. Polkinghorne, Science and Providence: God's Interaction with the World, SPCK, London, 1989, ISBN 978-1-93203-192-8.
• St. Thomas Aquinas, Summa Theologiae, IIa-IIae, q. 83 “De oratione”, articles 1–17.
• St. Augustine, Letter 130 to Proba, Patrologia Latina 33, col. 494–507.
• St. Alphonsus Liguori, The Great Means of Prayer (Del gran mezzo della preghiera), 1759.
• R. Garrigou-Lagrange, Providence (French original La Providence et la confiance en Dieu, Desclée de Brouwer, Paris, 1932).
• Catechism of the Catholic Church, §§ 2558–2865 (“Prayer in Christian Life”), 1992.
• C. S. Lewis, The Efficacy of Prayer, The Atlantic Monthly, January 1959; reprinted in The World's Last Night and Other Essays, Harcourt, 1960.
• C. S. Lewis, Letters to Malcolm: Chiefly on Prayer, Geoffrey Bles, London, 1964.
• Trent Horn, “Why Science Can NEVER Disprove Prayer”, The Counsel of Trent no. 1017, Catholic Answers, 7 April 2025.
• Archidiacre, L'inefficacité de la prière, preuve de l'inexistence de Dieu ?, YouTube, 31 March 2020.