Religion and the risk of suicide: longitudinal study of over 1 million people

Dermot O'Reilly, Michael Rosato

Research output: Contribution to journalArticlepeer-review

20 Citations (Scopus)
598 Downloads (Pure)


Background: Durkheim’s seminal historical study demonstrated that religious affiliation reduces suicide risk, but it is unclear whether this protective effect persists in modern, more secular societies.

Aims: To examine suicide risk according to Christian religious affiliation and by inference to examine underlying mechanisms for suicide risk. If church attendance is important, risk should be lowest for Roman Catholics and highest for those with no religion; if religiosity is important, then ‘conservative’ Christians should fare best.

Method: A 9-year study followed 1 106 104 people aged 16–74 years at the 2001 UK census, using Cox proportional hazards models adjusted for census-based cohort attributes.

Results: In fully adjusted models analysing 1119 cases of suicide, Roman Catholics, Protestants and those professing no religion recorded similar risks. The risk associated with conservative Christians was lower than that for Catholics (HR = 0.71, 95% CI 0.52–0.97).

Conclusions: The relationship between religious affiliation and suicide established by Durkheim may not pertain in societies where suicide rates are highest at younger ages. Risks are similar for those with and without a religious affiliation, and Catholics (who traditionally are characterised by higher levels of church attendance) do not demonstrate lower risk of suicide. However, religious affiliation is a poor measure of religiosity, except for a small group of conservative Christians, although their lower risk of suicide may be attributable to factors such as lower risk behaviour and alcohol consumption.
Original languageEnglish
Pages (from-to)466-470
Number of pages5
JournalThe British Journal of Psychiatry
Issue number6
Publication statusPublished - Jun 2015


Dive into the research topics of 'Religion and the risk of suicide: longitudinal study of over 1 million people'. Together they form a unique fingerprint.

Cite this