Can prayer be tested in a clinical trial?

Two questions at the bedside

A patient awaiting surgery may ask someone to pray. A researcher may ask whether patients assigned to receive organized intercessory prayer have fewer complications. These questions meet at a hospital bed, but they are not identical.

The patient may be seeking God's help, the presence of a friend, courage, or a way to speak fear aloud. A trial must specify an intervention, a comparison, an outcome, and a period of observation. “Does prayer work?” is too broad to tell researchers what to count.

A trial needs an outcome people can count consistently: complications after surgery, for example, recorded over a specified period. Its protocol says who will participate, what the intervention involves and how the groups will be compared. NHLBI's explanation of clinical trials

The details become especially consequential when the intervention is prayer. Praying alone, being visited by a chaplain and being prayed for by strangers are different experiences. A study has to choose what it will investigate.

What STEP found

Published in 2006, the Study of the Therapeutic Effects of Intercessory Prayer, or STEP, randomized 1,802 coronary bypass patients at six hospitals into three groups. Two were uncertain whether they would receive the study's prayer; one was told it would. Organized prayer lasted fourteen days. The main outcome was any complication within thirty days. Benson and colleagues' original study

The reported main-outcome rates were about 52% among uncertain patients assigned prayer and 51% among uncertain patients not assigned it: no statistically significant difference. Among patients certain of receiving prayer, the rate was about 59%, significantly higher than the 52% in the uncertain prayer group. Mortality did not differ significantly. The authors lacked a clear explanation for the excess complications and allowed that it could be a chance finding. Family and private prayer were not prohibited.

These findings did not demonstrate a benefit from the assigned intervention. They also do not establish that prayer generally harms people. The comparison involving certainty concerns a particular disclosure and intervention in a particular clinical setting; it does not identify a universal effect or its mechanism.

In STEP, the strangers were members of three Christian prayer groups, two Catholic and one Protestant. They received lists of patients’ first names and last initials, without contact with the patients or their families. Prayer began the night before surgery, and participants added a shared intention for successful surgery and recovery without complications to their usual prayers. The clinicians recording outcomes and the independent medical-record auditors did not know the patients’ assignments. STEP methods

A result needs the right comparison

The first comparison asks what happened when organized prayer was added without telling patients whether they received it. The second asks what happened when receiving that prayer was accompanied by certainty about receiving it. Neither control group was required to go without family or private prayer. The trial compared additions to patients’ lives, rather than creating a group from which all prayer had been removed.

Random assignment is useful because a comparison between people who independently choose to pray and those who do not can mix many differences together. Severity of illness, social support, habits, and expectations could differ. If the groups have different outcomes, deciding which difference mattered becomes difficult. A trial tries to create a cleaner comparison, though it can never turn a person's entire life into a controlled laboratory object.

A 2009 Cochrane review assessed ten trials involving 7,646 participants. Its overall analysis found no clear effect of intercessory prayer on death or general clinical state. The review described limitations in trial design and reporting. The included trials differed substantially: participants had conditions ranging from heart disease to leukemia, and one trial assigned prayers years after the hospital admissions it examined. That diversity and the quality of the underlying methods complicate a single pooled answer. The review is from 2009; it should be read as a historical synthesis rather than a current survey of every study. The full Cochrane review

An absence of a statistically significant difference does not prove that the true effect is exactly zero. Equally, uncertainty does not supply positive evidence of a benefit. The justified claim is bounded by the estimates, their uncertainty, the study population, and the intervention actually evaluated.

What Catholic prayer claims

Catholic prayer includes petition, but it is not defined as a technique guaranteeing a chosen clinical outcome. The Catechism's discussion of filial trust asks whether prayer treats God as an instrument or approaches him as Father. It includes the transformation of the person praying within its account of prayer's fruitfulness. Catechism, paragraphs 2734–2741 Teresa of Ávila’s account of prayer as friendship gives that understanding a lived history, including distraction and perseverance.

The tradition explicitly prays for healing. It also explicitly teaches that even intense prayer does not obtain the cure of every illness. The duty to care for sick people accompanies the duty to pray for them. Catechism, paragraphs 1508–1509

A Catholic therefore cannot reasonably turn a poor medical outcome into a score of the patient's faith. A person who dies after being prayed for has not thereby failed a spiritual test. Nor does recovery authorize certainty about why that individual recovered.

There is also a responsibility on the religious side of the argument. If someone makes a concrete claim that a specified prayer practice reliably reduces a measurable complication, that claim is open to investigation. Calling prayer relational does not excuse ignoring evidence about a promised result. A religious account of prayer still has to answer for the health promises made in its name.

Keeping the person in view

Could any result settle whether God exists? A clinical comparison would still require interpretation. An apparent benefit could invite questions about bias, chance, replication, and possible mechanisms. A null result would leave open the distinction between an intervention lacking a measured effect and the much larger claim that no God can hear prayer.

The distinction becomes practical when someone offers prayer to a sick friend. Promising a better surgical outcome goes beyond these findings. Offering to listen, visit and pray makes a different commitment, one that can be kept while the medical outcome remains uncertain.

For the person who asks, “Will you pray for me?”, a truthful answer needn’t contain a prediction. It can be a promise of attention and companionship. That leaves room to pray earnestly for healing without making the patient responsible if healing never comes.