A clinician’s guide to probabilistic suicide risk prediction tools: cautions and pitfalls

Journal article : Review

Summary: There are a growing number of new tools designed to predict suicide risk. One, OxSATS, developed in Oxford (UK) using Swedish data, produces a probabilistic risk of suicide in people who have self-harmed. It is accompanied by a web-based calculator, and states that it can ‘accurately predict 12-month risk of suicide’. It represents a departure from longstanding research arguing that risk prediction provides insufficient information to be clinically useful. We analyse the use of OxSATS from a clinician’s perspective using eight illustrative vignettes. For each, we use the OxSATS online tool to calculate the 12-month risk of suicide and consider how clinicians might interpret or act on the results. We highlight several potential harms to patients arising from the tool’s use. In our discussion, we explore broader limitations of OxSATS and similar tools, some of which are insidious. These tools can shift resources towards perceived higher-risk patients, often older men, diverting attention away from prevention, younger women and even the treatment of mental illness. Their reductionist approach misunderstands the complexity and stochastic nature of suicide. Tools tend to be disliked by patients and can subvert a clinician’s role away from helping patients, towards mitigating perceived risk. We conclude that tools such as OxSATS should be treated with significant caution and require careful scrutiny before being considered for clinical use. At present, psychosocial assessments and understanding patients’ narratives remain at the heart of good care for suicidal patients.

Authors: Cockburn, A; Large, M

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Cambridge University Press
• Journal: The British Journal of Psychiatry
• Pages: 1-5
• Publication date: 2025-11-05
• Acceptance date: 2025-09-22
• DOI: 10.1192/bjp.2025.10458
• EISSN: 1472-1465
• ISSN: 0007-1250
• Language: English
• Keywords: Suicide; risk prediction; self-harm; prognostic/prediction modelling; risk assessment
• Subtype: Review