Response

Daray and associates report a timely systematic review and meta-analysis, ‘Prediction of fatal and non-fatal suicide attempts by the Columbia Suicide Severity Rating Scale (C-SSRS)’.^{Reference Daray, Grendas, Neupane, Chiapella, Fisher and Gould1} The meta-analysis of 27 studies found that C-SSRS defined suicidal behaviour and suicide attempts are associated with future non-fatal suicide attempts, with odds of about three, a statistical effect size that is similar in magnitude to the association between male gender and suicide in most western countries. Their results raise two issues for clinicians interpreting this paper.

First, contrary to the suggestion of the title, the authors acknowledge in the text that they did not identify sufficient studies to examine the association with suicide mortality. The lack of evidence for the C-SSRS as a predictor of suicide is surprising given that over two dozen published studies report on non-fatal outcomes. This low level of reporting might reflect the low base rate of suicide over clinically meaningful time frames, suggests the possibility of publication bias against C-SSRS studies finding a weak association with suicide and should mute calls for the wider adoption of the C-SSRS.

Second, the authors hold that their results with respect to non-fatal self-harm support the use of the C-SSRS. However, the odds ratios achieved by the C-SSRS are modest and similar prognostic information can be obtained by simply talking with the patient about their experiences of suicidality.^{Reference Large, Corderoy and McHugh2,Reference Ribeiro, Franklin, Fox, Bentley, Kleiman and Chang3}

The C-SSRS and many other scales can identify a group of patients with a modestly increased probability of future suicide above the expected base rate, but do not predict suicide in the ordinary sense of the word.^{Reference Belsher, Smolenski, Pruitt, Bush, Beech and Workman4–Reference Large6} While some suicides can be prevented by thorough and sympathetic care, the use of suicide risk assessment scales is best thought of as semi-ritualised safety behaviour and response to, and cause of, clinician and institutional anxiety about rare, tragic and ultimately unpredictable events.