Suicide risk assessment: where are we now?

Patients who present in psychological crisis or after a suicide attempt are more than 50 times more likely than the general population to die by suicide in the following year.1,2 They require careful assessment and management. Numerous publications suggest, and some health department policies insist, that such patients should undergo a “comprehensive suicide risk assessment” in addition to a standard clinical assessment.3,4 Although most guidelines warn that suicide risk assessment does not substitute for clinical judgement, almost all also include long lists of patient characteristics that are claimed to enable identification of those patients at high risk of suicide. These lists are often fashioned into ad-hoc scales that purportedly stratify patients into categories of low, medium and high risk. However, it is simply not possible to predict suicide in an individual patient, and any attempt to subdivide patients into high-risk and low-risk categories is at best unhelpful and at worst will prevent provision of useful and needed psychiatric care. Here, we explain why this is so and outline what we believe clinicians should do instead.

Although patients presenting after suicide attempts or in crisis have a greatly increased probability of eventual suicide compared with the general community, fewer than one in 200 will actually die by suicide in the next 6 months.1 Suicide risk assessment is usually suggested as a way of determining who among these presenting patients should be offered different or more intensive treatment. However, the low base rate of completed suicide, combined with the lack of defining characteristics of people who proceed to suicide, makes this task impossible. Any useful further risk assessment would need to rely on a risk factor or a combination of risk factors that are consistently present, can be reliably assessed, and are very strongly overrepresented in the few people who do eventually suicide and yet rare or absent in those who do not. There is no evidence that such a risk factor or combination of risk factors exists. In 1983, the author of a large, well conducted prospective study of suicide lamented that “we do not possess any item of information or any combination of items that permit us to identify to a useful degree the particular persons who will commit suicide”.5 Thirty years later, nothing has changed: a similarly large and methodologically sound prospective study of psychiatric inpatients found that only 12% of suicides occurred among high-risk patients, and that fewer than one in 400 high-risk patients suicided.6,7

The list of generally accepted suicide risk factors is long and includes male sex, old age, divorce, medical illness, substance use, past suicide attempts and guilt. The most often cited risk factor for future suicide is expressed suicidal ideation.3,4 However, among psychiatric patients, expressed suicidal ideation does not discriminate between suicide completers and non-completers.8 Presumably this is because thoughts of self-destruction are very common in this group and because many people intent on suicide will not disclose that intent. Other intuitively appealing candidate characteristics include depressed mood and hopelessness. These, too, are almost universally found in the risk factors nominated in suicide risk assessment guidelines.3,4 However, not only are they very common in patients presenting with psychological problems, but most people with depressed mood will also feel hopeless, and vice versa; therefore, the combination of these risk factors offers little additional discriminative ability.

In recent decades, an enormous amount of research has gone into finding the magic combination of risk factors that acts as a sensitive and specific discriminator.5,6,9,10 Almost every study of completed suicide is retrospective, examining dozens, sometimes hundreds, of characteristics recorded in the medical records of both patients who suicided and controls, searching for factors or combinations of factors that were present more often in the patients who died.11,12 Such studies are likely to yield chance findings that appear to be significant, and they could identify some potentially reproducible associations with suicide. However, meta-analysis finds contradictory results between studies, leaving very few candidate risk factors among all the studies examined.11-13 Moreover, even a combination of reproducible risk factors still fails to provide a useful way of identifying high-risk patients. We examined studies investigating risk factors for completed suicide within a year of inpatient psychiatric treatment. Even making assumptions that optimise the power of suicide risk assessment, we estimated that while some patients could be categorised as “ultra-high-risk”, as few as 3% of these would go on to suicide in the year after discharge. Meanwhile, about 60% of people who did eventually suicide during the same period would have been categorised as being at lower risk.11

In original research and in meta-analyses, no generally accepted risk factors individually or in combination provide any useful way of further categorising patients who present in crisis. We understandably want to identify those who will later kill themselves — but this is simply not possible.

Here and now, clinicians should not be conducting “comprehensive suicide risk assessments”. They should be conducting comprehensive clinical assessments of each patient’s situation and needs. Suicidal ideation, for example, is not useful as an indicator of the likelihood of future suicide, but it is an invaluable sign of a person’s inner despair. So-called risk factors like depression, hopelessness and substance misuse are far too common and non-specific to be useful indicators of suicide risk, but they can be seen as symptoms of mental illness and important pieces of a revealed inner life that, taken together, give us a better understanding of the individual person and guide us towards what needs attention and discussion.

This is not to say that we should abandon patient safety. Bolstering family or social support, restricting access to lethal methods of suicide or making a crisis plan are steps that should be taken for all patients in psychological crisis — not just those whom we imagine might be more likely to suicide than others.

We cannot prevent tragedy by trying to identify those few souls who will be consumed by it. We must instead gather a comprehensive picture of each individual patient, and use this to tailor optimal management for the patients and families needing our care.