Abstract
There are a number of myths surrounding suicide risk assessment. The first is that certain risk factors can predict the likelihood of suicide in individual patients. The second is that people in high risk groups are likely to die by suicide. The third is that focusing resources on these high risk groups reduces the number of suicides. The fourth is that addressing and treating risk factors will result in a lower suicide rate.
These myths have led to the assumption that suicide could be reduced if increased risk assessments were more rigorously applied and appropriate treatment initiated. This assumption has led to expectations that clinicians routinely perform risk assessments on all patients. Such expectations have led patients, their relatives, their advocates and the coronial services to the belief that suicide is frequently the result of inadequate risk assessments within mental health services.
Large et al.'s paper [1] challenges these myths and expectations. They select a group of individuals with relatively high risk of suicide; patients discharged from a psychiatric hospital. They then perform a meta-analysis of controlled studies of suicide within a year of discharge. Their results are clear and emphatic. With regard to the myth that risk factors can predict the risk of suicide in individual patients they report ‘No factor, or combination of factors, was strongly associated with suicide in the year after discharge.’ This result is consistent with every study that has examined predictive risk factors in suicide. Beautrais [2] for example, after following up a cohort of individuals who had made a serious suicide attempt reported that only a very limited prediction of suicide was possible.
The second myth that people in high risk groups are likely to die by suicide is also refuted. Large et al. [1] report that 97% of their high risk sample did not commit suicide. Again this is consistent with all similar research. Harriss et al. [3] examined a population of individuals who had overdosed and reported that they had a 97% overall survival rate. Even those with the three most important risk factors (alcohol, separated/widowed/divorced, age 55 plus) had a 95% survival rate. High risk groups are very unlikely to die by suicide and only slightly more likely to do so than low risk groups.
The third myth is that focusing on high risk groups reduces the number of suicides. It might do but not by very much. The problem, as Large et al. elegantly demonstrate, is that the majority of suicides occur in low risk groups. This finding arises because although low risk groups contain individuals at lower risk, they also contain many more members than high risk groups. While Large et al. report 60% of suicides are in the low risk group this figure is even higher in non-selected samples. According to the National Confidential Inquiry into Suicide and Homicide in People with Mental Illness, 86% of suicides are in the low risk groups [4,5]. This implies that even if we could accurately predict all high risk patients and had effective treatments to prevent them all committing suicide we would only reduce the suicide rate by 14%.
This leads on to the fourth myth, that treatment of risk factors has an impact on survival. Very few psychiatric interventions have been shown to reduce the incidence of suicide; perhaps clozapine in schizophrenia, lithium in bipolar disorder and possibly structured psychotherapies in borderline personality disorder [5]. Large et al. 1 report that patients in receipt of less psychiatric follow up were less likely to commit suicide. While, as they point out, a plausible reason is that patients provided with more follow up were, as a group, more likely to commit suicide, the finding is hardly reassuring. More disturbingly, Kessler et al. [6] compared two general population cohorts for suicidal behaviours. They reported that despite a dramatic increase in treatment, no significant decrease occurred in suicidal thoughts, plans, gestures or attempts in the USA during the 1990s. A recent systematic review in young people reported no differences between treatment and control groups in reducing suicide risk except for one small study [7].
It seems sensible to conclude, as indeed Large et al. [1] do, that risk categorization is of very limited value in predicting which patients will commit suicide after discharge. It is worth emphasizing that no study has prospectively examined suicide among patients already categorized as high or low risk. As the authors point out a prospective high risk model is likely to perform even more poorly than the results from their meta-analysis, since different risk factors were significant in different studies.
However, Large et al. [1] go further, pointing out that preoccupation with risk assessment may be more than a waste of time: it may do harm. They note that ‘erroneous beliefs about which patient characteristics are associated with post-discharge suicide could lead to needlessly more restrictive treatment for some patients wrongly labelled “high risk” and the misallocation of health care resources away from patients wrongly labelled “low risk”.’
The ‘risks of risk assessment’, as Undrill [5] titled a recent article are only beginning to be discussed among clinicians. Three of the authors of the Large et al. paper wrote a recent article in Australasian Psychiatry outlining why clinical decisions in psychiatry should not be based on risk assessment [8]. They argue that risk assessments, particularly for rare and serious events, have little predictive value and are likely to expose some patients to adverse consequences. They suggest using capacity in preference to risk categorization as a base for involuntary treatment.
Undrill [5] goes further, positing that risk assessment is often an organizational attempt to tame anxiety rather than to improve patient care. He distinguishes primary and secondary risks arguing that the latter is clinicians becoming increasingly preoccupied with managing their own risk rather than the patient's risk. Patients may therefore be detained not for treatment needs but because not detaining them produces intolerable anxiety in the staff involved in the assessment. The influence of secondary risk management may have a corrosive effect on the doctor–patient relationship. Doctors construct patients as a source of threat to their professional standing and may not always act in their patients’ interests [5].
Therefore our current preoccupation with risk assessment has the potential to harm patients, clinicians and the organizations in which they work. It has created a mythology with no evidence to support it, a sense of unease among clinicians and a culture of blame when things go wrong. What can be done to change this?
First, we need to acknowledge the uncertainty of our knowledge base and that risk assessment is problematic and may carry its own risk. We need to re-educate the public that we are poor at predicting rare events like suicide and that mental illness models are neither necessary nor sufficient to explain suicidal behaviour.
Second, the best way to manage organizational risks is to improve core tasks of patient care. Risk should primarily, but not exclusively, be seen in terms of risk faced by the patient. The psychiatric interventions associated with reducing the risk of suicide such as clozapine, lithium and structured cognitive behaviour therapy are given because they are effective treatments, not for the purpose of reducing risk.
Third, organizations should take steps to protect their staff from the necessity of secondary risk management. Risk assessments should be used where appropriate and be a consensual process with the patient and clinician striving towards a realistic conceptualization of the risk then deciding how best to manage it. Detailed and complex risk assessments should be curtailed since they may amplify rather than reduce anxiety [5]. Accumulating more and more risk information of marginal utility is of little value in helping patients. Risk assessments of suicide are vulnerable professional decisions in the face of uncertainty. What is certain is that we will make both false negative and false positive judgements. While risk assessments are here to stay they should not form the basis of clinical decision making. The aim is to provide optimal care according to the psychiatric needs of each patient, regardless of the perceived risk of suicidal behaviours.
In summary, the current preoccupation with risk assessment is at variance with the evidence that such assessments are accurate. Because uncertainty is impossible to eliminate and an adverse outcome may carry hazards, clinicians often carry out risk assessments to minimize their anxiety in a rationally selfish way [5]. This leads to unintended negative consequences for patients, clinicians and health services. The challenge for clinicians is to return risk management to be a part, rather than focus of patient care.