August 20, 2013

expert reaction to new study on predicting suicide

Findings from a small sample of men in Molecular Psychiatry showed blood biomarkers that may be associated with suicidal states, potentially providing clues about the biological mechanisms that underlie suicidal thinking and behaviour. 

Professor Keith Hawton, Director, Centre for Suicide Research, University of Oxford, said:

“Identifying predictors of suicide is a very difficult task in that predictive power of any factors that have been examined is low in every case, be they biological, social, psychiatric or psychological.  The main issue is in terms of positive predictors i.e. the extent to which an individual with a risk factor is actually going to die by suicide.  I know of know of no well-powered study where the extent of positive for any factor has been high.  Even after attempted suicide (the best predictor), the risk of suicide is only 0.5-3.0 in the following year (the highest risk time).

The same issues apply to potential biological risk factors.  Also, there is a big difference between finding differences between groups (as in this study) compared with risk in actual individuals, the latter being the real test of predictors.

The size of the suicide study is extremely small (N=9) and even the size of the replication studies for history of suicide attempts is relatively small.  Therefore I would say that the findings are of interest and may point the way to some future research based on large samples, but no more than that.  I would certainly hope that the media don’t pick this up as a test for suicide, as has happened in the past regarding some brain findings.”

Professor Matthew Hotopf, Professor of General Hospital Psychiatry, Institute of Psychiatry (IOP), King’s College London, said:

“These small sample sizes and biomarker studies – like any other – need external replication, but more than others they need this because the potential list of biomarkers is so huge – much bigger than the number of participants.

“It’s one thing to find a biomarker which might be associated at a statistical level with suicide/suicidal behaviour. It’s quite another to use it to make any kind of prediction which has clinical utility.  To do this, one needs very large effect sizes and specificity.

“So  these findings may attract media attention, but they are very much preliminary and my money would be on (a) failed replication and even if replication was successful (b) lack of predictive power to be a useful clinical tool.”

‘Discovery and validation of blood biomarkers for suicidality’ by H Le-Niculescu et al published in Molecular Psychiatry on Tuesday 20 August 2013.