On Tuesday, April 9, 2024, Mark Ward, Sinn Féin spokesperson on mental health, used Dáil question time to enquire about an updated suicide prevention strategy. In response, Mary Butler, Minister of State at the Department of Health, referred to Connecting for Life: Ireland’s National Strategy to Reduce Suicide 2015-2024, which focuses on the prevention of suicidal behaviour and addresses a broad range of risk and protective factors. The strategy had been due for renewal in 2020, but was extended for four years by the Cabinet. Ms Butler observed that the strategy’s implementation under the auspices of the National Office for Suicide Prevention (NOSP) is currently being evaluated to inform the successor strategy which the Department of Health has now commenced work on.

In the midst of all this, it is not unusual for practising frontline psychiatrists to be criticised by lawyers in coroners’ courts for not preventing individual suicides. Indeed, the term ‘never event’ was popular around 10 years ago, originally coined by Ken Kizer in 2001. Aside from confusing an adverb with an adjective, the unfortunate term ignores the reality that suicide is sadly almost impossible to predict in an individual. 

At various times, clinicians have tried to devise valid screening questionnaires that distinguish those who will die from suicide from those who will not. Most major psychiatric illnesses carry an eventual suicide rate of around 10%, but this is a long-term risk that exists over a lifetime. I encounter perhaps one suicide per year among my hundreds of patients; day to day, it is a relatively rare event. Asking a psychiatrist to predict which of their outpatients will die by suicide this week is like asking a cardiologist to say which of their thousand or so diabetic, hypertensive, smoking outpatients will have a fatal heart attack in the same period. 

The problem lies with the statistics. The predictive value of screening tools depends largely on how rare or common an event is. The rarer an event is, the lower the positive predictive value of a test and the higher the negative predictive value. As such, for a rare event like suicide, it is far easier to use a screening tool to predict the large number of people who will not die by suicide than the small number of people who will.

The research bears this out. In a famous prospective study involving 4,800 inpatients of a Veterans Administration hospital, Pokorny used a wide range of instruments and measures previously thought to predict suicide. The screening tools predicted around 1,200 suicides that did not occur. Meanwhile, of the 63 suicides that did actually occur, only half were predicted by the tools. As such, 19 suicides were incorrectly predicted for every suicide that actually occurred, while half of those suicides were not predicted at all. As the author put it: “Each trial missed many cases and identified far too many false positive cases to be workable”. A more-recent meta-analysis by Large et al involving over 20 studies of suicide among psychiatric inpatients found that, although there were risk factors significantly associated with suicide, the rarity of suicide as an outcome meant that less than 2% of patients deemed by clinicians to be ‘high risk’ actually went on to die by suicide.

The cost of overpredicting suicide in individuals includes unnecessarily restrictive care and even civil rights infringements, along with a lesser share of scarce staffing and resources for everyone else in need of treatment. In the end, suicide prevention is the business of society, not just psychiatry. Time to update the strategy.