Nothing in between: a multi-faith response to the paper on religion and suicide

As a Liberal Jewish psychiatry registrar and a moderately observant Hindu psychiatry senior house officer, we read with great interest Kleiman & Liu’s fascinating paper on the relationship between religious service attendance and suicide risk.¹ We were also pleased to note that the paper has already generated sufficient interest to give rise to a fascinating editorial by Cook² as well as helpful correspondence between the authors and Professor Nebhinani.³ This seems to attest to the importance of this topic, and we hope that our additional reflections on the methodology of Kleiman & Liu’s study can be part of an evolving dialogue around the interactions between mental health and religion/spirituality.

First, while there are many advantages to the prospective study design, difficulties are produced when the outcome of interest (here, completed suicide) is a relatively rare one. Only 25 completed suicides occurred, and the absolute numbers occurring in the two groups (frequent v. less frequent service attenders) were not specified. Even a very small swing in the distribution of the suicides from one group to the other could significantly alter the apparent magnitude of the protective effect of service attendance.

Second, the absence of any intermediate data between baseline and the end of the study period makes it difficult to draw conclusions about the potential link between religious service attendance and suicide - hence the title of our letter, ‘Nothing in between’. During the follow-up period (12-18 years), much might have changed in people’s lives, behaviours and health. In particular, people’s level of religious observance (in the form of service attendance) might well have varied over the study period - as might their mental health. Moreover, there might well be interactions between these two variables. With only two data-points (baseline self-report and a dichotomous outcome of suicide/not-suicide), it is impossible to know people’s religiosity and mental health across the study period.

Third, this lack of intermediate data might stem from the fact that Kleiman & Liu’s study seems to have ‘piggy-backed’ onto a separate, pre-existing epidemiological survey,⁴ the primary objective of which was not the investigation of the relationship between religiosity and suicide. Convenient as it might have been to make use of pre-existing data, it might be that a study set up specifically to address the research question would offer richer information and allow greater extrapolation and clinical application.

Fourth, and also in terms of clinical applicability, we would question whether the focus on completed, as opposed to attempted, suicide is necessarily an advantage. As the authors observe, there do seem to be differences between the clinical profiles of those who complete, compared with attempt, suicide. However, the two are closely related, with previous suicide attempts a sufficiently well-recognised risk factor for suicide completion that it has been controlled for as a potential confounder in Kleiman & Liu’s study, even though it was not found to be ‘a significant predictor of death by suicide’. Moreover, in the clinical setting, suicide attempts are one of the primary risk events of concern, but the study does not provide information on how religious service attendance might relate to these - information which could be of considerable relevance for risk assessment.

Given the above reflections, as well as the study limitations identified by Kleinman & Liu, Cook and Nebhinani, we caution against an over-simplistic reading of the article’s headline finding. In our clinical experience, the relationship between a person’s religiosity and their risk of self-harm/suicide can vary considerably. We therefore urge that clinicians continue to conduct detailed explorations of each patient’s individual dynamic risk factors and not overly focus on particular population-level static risk factors.