Abstract
See Review by Large et al., 2011, 45(8): 619–628; See also Editorial by Mulder 2011, 45(8): 605–607
The paper by Large et al. (2011) and the accompanying editorial by Mulder (2011) in the August 2011 issue of the journal were far too pessimistic and could be interpreted as inferring that we are powerless to intervene.
It appears that each generation needs to rediscover that the assessment of risk with the aim of predicting suicide is not possible in an individual, but that does not mean that it is of no value.
The limited utility of the prediction of suicide in an individual, Mulder’s first ‘myth’, was noted over 50 years ago by Rosen (1954), and two decades later, Farberow and MacKinnon (1975) reaffirmed that ‘the level of prediction of suicide is still too minimal to permit individual clinical application’. Pokorny (1983) and Goldney and Spence (1987) in an early editorial of this journal reached a similar conclusion.
However, would an oncologist or cardiologist state that it was a ‘myth’ of prediction if one was able to predict that 3% of a group of patients would die in the next year? That is one in 33. Is an odds ratio of 3.94 for high-risk patients being more likely to die than other discharged patients not important?
I suspect the average person, and probably the average coroner, would be interested in these data.
Mulder’s second ‘myth’ that ‘people in high risk groups are likely to die by suicide’ is clearly not a ‘myth’, as demonstrated by the work of Large et al. (2011). His third and fourth ‘myths’ appear to be written tongue in cheek, as he concedes in both cases that there may be evidence to the contrary.
Yes, we must acknowledge the limitations in the prediction of suicide in an individual. However, on the basis of aggregate data, including that of Large et al. (2011) who have demonstrated that 3% of a high-risk group will die by suicide in the ensuing year, we have consistent evidence of areas of focus for intervention. This is potentially of importance, as long-term studies have demonstrated the enduring nature of suicidality in some persons (Goldney et al., 1991), and neurobiological research has shown that reduction in serotonin transporter binding sites in the prefrontal cortex may be part of a predisposing diathesis to suicide (Arango et al., 2002).
It could also be argued that prediction research does not really address the issue that actually confronts us in practice: we do not usually make a decision on whether a person will die by suicide in the next 12 months, but rather, on the basis of whether suicide appears imminent. In this regard it appears prudent to err on the side of caution, an approach that appears to be ignored at times in busy emergency departments.
It is important that clinicians do not become complacent in their assessment and management of those who are suicidal. In this regard it is pertinent to recall the observations from two large studies of suicide following contact with psychiatric services (Appleby et al., 2006; Burgess et al., 2000). Thus, about 20% of those suicides were considered preventable, but for inadequate assessment and management of depression and other disorders, poor staff–patient relationships and inadequate continuity of care, particularly in the transition between hospital and the community.
An alternative view of the results of Large et al. (2011) is that as we are able to predict on the basis of aggregate data, they can be used to emphasize that all those who are suicidal warrant the best available assessment, including that of risk, which informs us of potential areas of intervention. Indeed, Mulder (2011) is correct in emphasizing that risk management should be part of, rather than the focus of patient care.