Overselling risk assessment

I need to congratulate Roychowdhury & Adshead ¹ on a thought-provoking critique. Their arguments struck a chord in exposing the flaws in risk assessment tools and their unjust application in preventative detention; however, I was disappointed that they did not go further. All of these tools, structured clinical judgement included, apply population-derived data to individuals, thus painting them with the behaviour of their peers. The central flaw of risk assessment lies in presuming causality from association. The premise in these tools that symptom severity invariably correlates with risk is demonstrably fallacious, as any psychiatrist could counter-cite cases where treating the mental illness improves functional ability in patients who choose pro-criminal lifestyles.

The second problem, as previously highlighted by Szmukler, ² is their inherent determinism by casting the subject (participant) as a hapless automaton. Society is rightly critical of the boorish youth who binge drinks and gets into fights, yet exculpates the capacitous non-adherent person with schizophrenia - and holds their psychiatrist vicariously liable for their violence.

Risk assessment attempts to sanitise an unpalatable fact that violence is part of the human condition, which exists independently of mental illness. Milgram ³ and Zimbardo ⁴ infamously illustrated this. Nonetheless, even when convicted, the offender without a mental disorder rarely faces the sanction of possible indefinite detention. Indeed, it was implicit in the debate around dangerous and severe personality disorder and the 2007 revisions to the Mental Health Act that psychiatry could be manipulated into preventatively detaining risky individuals in society without the bothersome need for a trial. ⁵

The truth is that risk assessment has become an industry. Those devising the next ‘marginally-better-than-chance’ tool can live off the proceeds of the copyright, training seminars and subsequent release of version 2.0. It is also politically expedient in reverse-engineering a scapegoat and providing glib platitudes that ‘lessons are learnt’, and ‘something is done’ in a world increasingly tilting at the reality of rare unpleasant events.

I believe that expectation regarding the prescience of risk assessment has far outstripped the reality of what it can achieve. The evidence base for risk assessment, by the authors' own conclusion, would not support its use as a diagnostic instrument; yet in clinical practice it is insidiously taking over as a priority. Criminal justice operates on the principle that it is better to let ten guilty men go free than convict one innocent. If the original question was one of ethics, surely for an exception to be made for those with a mental illness is frankly discriminatory.

Furthermore, the question around the ethical principle of beneficence remains unanswered: if risk assessment is a priority activity, what is the evidence that it improves clinical outcomes over and above quality standard care? I cannot offer an alternative other than to lament the fact that the Richardson Committee's report in 1999 on transforming mental health legislation from risk- to capacity-based was never realised. We need to refocus this debate clinically by emphasising ‘needs assessment’ over ‘risk assessment’. Risks are unavoidable; but good-quality evidence-based care should not be usurped by the latest fashionable risk assessment tool.