Kleiman and Lui ^{Reference Kleiman and Liu1} have conducted a respectable study that suggests that people in the USA who attend church frequently are less likely to die by suicide than other people. Although the finding is interesting, it is not surprising. It is congruent with Durkheim’s 19th-century theory about the role of anomie.

Chris Cook, ^{Reference Cook2} in a moderately worded editorial, suggests that this finding has implications for British clinical practice. In my opinion, this is profoundly misguided.

First, the finding is specific to the USA, a country with exceptionally high rates of religious involvement, where church attendance and social respectability are intimately linked. The social meaning of church attendance is completely different in the UK. Although I guess that a UK study would be likely to yield similar findings, scientific rigour demands that this cannot be assumed.

Second, Cook says that the finding merits discussion with patients at risk of suicide. It is far from clear what he means by this. I doubt if he means to imply that psychiatrists should explain to patients abstracted epidemiological factors that might affect their actuarial risk of suicide.

It is always important to understand the social and emotional supports that tend to protect patients from taking their own lives. This is a matter of proper assessment. However, there are no grounds for psychiatrists to advocate church attendance to individuals who consult them. Kleiman and Lui have identified a demographic factor that appears to be protective. They have not evaluated an intervention. Even if they had, in the UK setting proselytising of religion by medical practitioners is a serious breach of professional boundaries.

It is difficult to identify the line between evangelisation and ostensibly more benign types of religious intervention (for example, suggesting that churchgoers might attend more frequently), which illustrates why boundaries need to be clear rather than blurred. It is hard to understand how a discussion of churchgoing as part of a psychiatric intervention could avoid promotion of a particular religious viewpoint. With regard to protection of patients, Cook cites the College Position Statement ^{Reference Cook3} that he wrote: ‘much is properly left to the judgement of the psychiatrist’. Everything we have learnt about boundary violations over the past 20 years tells us that this is an unreliable way of protecting patients, which is why some of us strongly disagree with the College Position Statement.

Fortunately, Chris Cook and I are not simply trapped in a cycle of disagreement. ^{Reference Poole and Cook4} With colleagues from Bangor and Durham, we have been developing research to explore the boundary issues over religion and spirituality. Until that work is completed, and possibly thereafter, it is important to be clear that there is a serious difference of opinion over bringing religion into the clinical setting. This is determined by factors other than religious faith, or the lack of it.