Abstract
To the Editor
Draper raises issues following our recent meta-analysis of factors associated with suicide in the year after discharge from psychiatric hospitals (Draper, 2012; Large et al., 2011a). First, he suggests that suicide risk assessment “personalized to the individual” is less likely to suffer from the shortcomings in prediction we found in our meta-analysis. Second, he asserts that a suicide risk assessment is likely to be more valid over the short-term than over a period of up to one year. Neither assertion can be supported.
First, there is no evidence that individualised clinical risk assessments are more accurate than assessments based on objective risk factors. In fact to the extent that either form of assessment is likely to validly categorise individuals by likelihood of future suicide, it is clear that “checklist approaches” are more valid (Meehl, 1986). This should not be surprising since clinical judgments have a range of well-established weaknesses including the failure to consider known risk factors, an inability to consider co-variation between risk factors, underutilisation of base rate data, and a range of cognitive biases including confirmatory bias supporting an initial hypothesis (Ziskin, 1991).
Second, there is no evidence that suicide risk assessments are more valid over a short period. In fact there is almost no evidence of the validity of short-term suicide prediction at all (Simon, 2006). We could not find a sufficient number of studies of short-term suicide risk factors to consider in a meta-analysis. However, even if such studies were available it is very unlikely that short term suicide risk assessment would have greater validity than suicide risk assessment for the longer term, since that base rate of suicide in the short term will be even lower than it is in the longer term and it is the low base rate that undermines all serious attempts to categorise people by their likelihood of future suicide (Large et al., 2011b).
None of this is to suggest that we disagree with Draper’s plea for careful and empathic clinical assessment of the patient and for good communication between the family and treating team. A full understanding a patient’s circumstances, attitudes and psychiatric disorder is essential for quality psychiatric care, and optimal psychiatric treatment for all patients is likely to prevent some suicides. In contrast, categorising patients to be at low risk can pose a barrier to good care while a high risk categorisation inevitably leads to unnecessarily restrictive and counter-therapeutic approaches.