Religious service attendance and mortality

Back in 2001, Leonard Leibovici presented an RCT which found a correlation between remote, retroactive intercessionary prayer and length of stay in hospital. The patients in question had bloodstream infections in Israel during the period 1990–6; the intervention involved saying ‘a short prayer for the well being and full recovery of the group as a whole’ in the year 2000 in the USA, long after recovery or otherwise actually took place. The study also found a correlation between the intervention and duration of fever. The author concludes:

“No mechanism known today can account for the effects of remote, retroactive intercessory prayer said for a group of patients with a blood-stream infection. However, the significant results and the flawless design prove that an effect was achieved.” (Leibovici, 2001, Effects of remote, retroactive intercessory prayer on outcomes in patients with bloodstream infection: randomised controlled trial. British Medical Journal, 323:1451.)


As I explain in this paper, the claim that causation is established here is very much in line with present-day EBM, which focusses on RCT evidence, but not in line with EBM+, because what evidence there is points to the absence of a mechanism of action. I’d suggest that in this example it is reasonable to remain sceptical about the causal claim. If so, this example counts in favour of EBM+ and against present-day EBM.


More recently, Li, Stampfer, Williams & VanderWeele reported a correlation between religious service attendance and mortality amongst nurses surveyed in a large longitudinal study (2016: Association of religious service attendance with mortality among women, JAMA Internal Medicine 176(6):777-785). They used rather complex marginal structural modelling techniques to suggest that attending religious services reduces mortality by about a half. Tyler WanderVeele also wrote an article for The Spectator on the subject called, “People who go to church live longer. Here’s why“.


From the EBM+ perspective, the question arises whether there is some mechanism of action, or whether the correlation is spurious – i.e., attributable to chance, confounding, bias or modelling assumptions. The authors of the paper investigate certain possible mediating variables  – e.g., depressive symptoms, smoking, optimism, and social integration – but find that these variables fail to explain anywhere near the full extent of the correlation.


What should one make of this? From the EBM+ point of view, the fact that evidence points against the existence of a mechanism, or combination of mechanisms, that explain the full extent of the correlation, casts some doubt on the causal import of this particular study.


One might think that doubt should be cast only on the proportion of the effect size that is unexplained by credible mediating variables. If so, then one would still be able to infer that religious service attendance has a significant effect on mortality. However, the modelling methods that led to claims about mediating variables were the same methods that posited such a dramatic net effect size. So, doubts about the effect size must to some extent spread to the analysis of mediating variables. In sum, there are grounds for caution even with regard to that part of the effect size that is explicable.


Jon Williamson

(With thanks to Jan Vandenbroucke for pointing out this latest paper.)

Leave a Reply

Your email address will not be published. Required fields are marked *