Effects of remote, retroactive intercessory prayer on outcomes in patients with bloodstream infection: randomized controlled trial. Published by Leibovici in the British Medical Journal BMJ in 2001. BMJ currently has an impact factor of 17.215.
The title of this article sounds kind of strange, so let’s see if I have got this right:
Retroactive: backwards through time, in this case praying for the health of people in the past.
Intercessory prayer: praying to your deity of choice on behalf of somebody else’s health, hoping that said deity will intervene in response to the prayer and speed up recovery of the sick person. Intercessory prayer is actually fairly well studied: there are 94 articles listed in Pubmed (8 of which are responses to the Leibovici paper); however most of them are published in fringe low impact journals. A Cochrane review did not recommend further clinical trials.
Bloodstream infections: Bacteria that spread into the blood of a patient
Randomized controlled study: “A study design that randomly assigns participants into an experimental group or a control group”
The author seemed to have studied the effects of remote prayer on blood stream infections backwards through time.
Let’s now have a look at the abstract: Typical for medical abstracts it is divided into objective, study design, results and conclusions. It confirms what the title promises, the author researched the effects of remote prayer backwards through time. The patients were hospitalized between 1990 and 1996, while the praying took place in 2000. The author measured mortality and length of stay in the hospital as his variables and concluded that the intervention group fared better on both counts. He therefore concludes that retrograde remote prayer works and should be considered as a useful tool for medical doctors.
Let’s examine how he reached this conclusion:
Introduction: here the author sets up the bold hypothesis that time might not be linear which he supports with this highly theoretical paper.
Methods: 3393 patients diagnosed with bloodstream infections in an intensive coronary care unit were randomly distributed into two groups. A list with the first names of one group was given to a (holy?) person who said a prayer for them. Then three variables, death, duration of hospital stay and duration of fever were compared. The null hypothesis, prayer works backwards through time was tested using a chi2 test or a Wilcoxon rank-sum test.
And now the most interesting part, the results. They are rather limited, the main findings are summarized in this table.
There is no significant difference in the duration of fever between the treatment groups, however the duration of hospitalization differs significantly between the two groups. The maximum stay in the control group was almost twice as long as the intervention group. While there are very few differences in the lower quartile the upper quartile differs (significantly?) between the two treatment groups. In conclusion, prayer for people backwards through time works.
The author states that the trial design is “flawless”, because of the “perfect blinding to patients and medical staff of allocation of patient allocation to the two groupss and even the existence of the trial”. He did not offer an explanation or hypothesis for this astonishing result.
Of course there are many problems with this paper, not in the least being the lack of informed consent of the participants and permission of an oversight committee at the hospital such as an IRB or ethical review committee. It appears that most of the significance of this study can be ascribed to one outlier in the control group, whose stay in the hospital was extended. However, without access to the raw data it is hard to prove this. The fact that the median does not differ between the two treatment groups is another hint, i.e. that the results might look very different when the outlier is removed.
This paper takes a ridiculous hypothesis and runs with it. It shows that to do good science we should not just discard a hypothesis, because it sounds too far fetched, but do our best to prove or disprove it. Leibovici did not do this. The experiment wasn’t very well designed, no matter what he claims and additional key information, such as what god was chosen to intervene following receipt of prayers are missing from the paper. The statistical tests the author chose to perform are not commonly used to test for differences in treatment groups during clinical trials, perhaps a T-test would have been more appropriate.
A better-designed experiment might have yielded a different result. However, it is important as scientists, to not just say “that’s stupid, so it can’t be true”, but to prove from the data presented that the conclusions reached are incorrect or point out flaws in the experimental design.
Naturally quite a few papers were published in response to this study, e.g. “You cannae break the laws of physics, Captain.”
This paper also sparked an interesting discussion on the BMJ website and is often cited by people, who believe in intercessory prayer and don’t have a scientific background. This shows that even with the best of intentions to write a light hearted scientific paper with a comedic edge can have a very real and sad backlash.
In conclusion, I like this paper, mainly because in the end they show a picture of a reindeer for some reason.