Higgins et al (2004) begin with a reasonable summary of the views of the ‘for’ and ‘against’ lobbies in the great UK risk assessment debate, however, the authors are too polite to point out just how bizarre this argument appears in more enlightened parts of the world. The evidence on this question, from both forensic and general psychiatry, is unequivocal; the best assessment of violence risk in an individual patient is provided by structured clinical judgment (Reference Monahan, Steadman and SilverMonahan et al, 2001). We should be using standardised data collection to improve, but not to replace, clinical formulations.

Those clinicians opposed to standardisation cling to the fantasy that violence risk assessments are an optional extra for psychiatrists. Again, the evidence is clear. Since the epidemiologists have demonstrated a small but significant association between schizophrenia and violence (e.g. Reference Swanson, Holzer and GanjuSwanson et al, 1990), any attempt to claim that violence has nothing to do with us serves only to cause further damage to the profession's public image. If we want to be like other doctors, we need to accept that much medical effort goes into managing the risks of complications of a disease, rather than the symptoms of the disease itself. Hypertension is the classic example, with no symptoms but plenty of treatments, all aiming to reduce the risk of complications such as strokes and myocardial infarctions. Psychiatry's misfortune has been to choose a disease whose complications affect a third party, but the principle is the same.

Opponents of standardisation are also happy to sign forms to detain patients under the Mental Health Act 1983 ‘for the protection of other persons’. They must base this decision on a risk assessment, explicit or not. The honest position is to acknowledge that we all do violence risk assessment, and the important question is how well we do it.

Certainly, some Mental Health Act assessments are straightforward and require no sophisticated analysis, but there is a strong case for a more systematic approach in marginal cases. Psychiatry has much to be modest about in its current, unstructured assessment, given the over-representation of ethnic minorities among detained patients. It may be that all our decisions are unbiased and based on the best evidence, but the profession would not harm its public standing by using some objective, operational data to support all the gut feelings, instincts, intuitions and rules of thumb.

Also, standardisation opens up the possibility that we can begin to talk to each other about risk in a meaningful way. Measurement is the first step in any scientific inquiry, without which further investigation is impossible. Standardised measures of violence risk could justify their existence simply by opening up the area to scientific study, even if they contribute nothing else.

For those unimpressed by the science, we have the politics. British psychiatry has got itself into a mess over violence. Public confidence in the profession is at a low level, with the tabloids fighting to outdo each other in psychiatric horror stories. Government confidence is also low, and it has never been so difficult to get Home Office consent to the conditional discharge of restricted patients. The College is right to be concerned about stigma, but it is arguable that the profession is now more stigmatised than the patients.

Like the rest of life, none of this is fair. Both the government and the public overestimate the risk of violence by psychiatric patients. Tabloid editors overstate the risk in order to sell newspapers. But, crying foul and complaining about stigma will not win back the public confidence in mental health services that our patients need.

The experience of forensic psychiatry is instructive in this regard. When services appear not to be managing risk, as in the pre-Fallon Ashworth personality disorder unit (Reference Fallon, Bluglass and EdwardsFallon et al, 1999), the politicians come steaming in with inquiries, reports and reforms. Few people within the service approve of this intervention. They point out that the problems were in a tiny part of a system that had a good security record overall, but nobody takes any notice. Protest is futile once trust has been lost. Only when services show they are responding to popular concern and taking the risks seriously, can the politicians afford to relax and move on to the next headline issue.

The lesson for general psychiatry is that, once the public and politicians have made violence a major issue, services need to be seen to be taking it seriously. Structured risk assessment is not the whole answer, but it sends the right message. Certainly, blanket opposition to structured violence risk assessment is political or public relations suicide, and it invites outsiders to impose solutions upon us.

The best example of such externally imposed measures is the draft mental health bill, essentially the product of a non-dialogue in which the profession talked long and loud about patient autonomy and mental capacity, when the people who pay for the service wanted to hear about risk management. Another good example is the proliferation of risk assessment forms described in the paper. National Health Service managers, who tend not to prosper without some grasp of political issues, realised that services had to have a response to violence risk that went slightly beyond: ‘Trust me, I’m a doctor’. So they created forms to fill in.

The impetus behind the forms was that they were, literally, better than nothing. The problem is that they are not much better. The authors note great variation in content; there may or may not be research to justify the content of the forms; and clinicians often lack confidence in them, so many consultants do not attend the training, in the 50% of trusts that provide any training. The variation between trusts is also a problem, as it negates the advantage of improving communication about risk by using the same measures. The point of standardised assessments is that they are standardised.

The findings also suggest that risk assessment in these trusts was separate from the process of care and treatment, rather than integrated into it. The UK debate has focused on whether structured assessment provides a better measure of risk, but an important additional function of standardised measures is to guide treatment, so that risk is reduced. This is unlikely to happen when the assessment is disconnected from the rest of the care package.

So what is the ideal instrument for use in general psychiatry? It needs to be fairly quick and easy to complete, and it needs to be flexible so that it can adapt to a vast range of risk, from high to virtually zero. The structured risk assessment should also integrate with clinical procedures and, particularly, with the care programme approach (CPA). The aim should be seamless progression from risk assessment to clinical management.

While this seems like a difficult question, there is a simple answer in the Historical/Clinical Risk-20 (HCR-20; Reference Webster, Douglas and EavesWebster et al, 1997). It begins with ten historical variables that provide a good summary of actuarial risk. In effect, these static background factors help to establish the size of the stakes involved in a particular case. The five clinical variables concern present state and they introduce a dynamic (in the sense of changeable, rather than psychodynamic) aspect. They are able to reflect changes in risk consequent on changes in the patient's clinical state, so they indicate the extent to which our interventions affect the risk. The five risk management variables look to the future and are concerned with treatments, attitudes, compliance and destabilising factors. They assume that the question of risk cannot be divorced from the question of what treatment is to be provided, and where the patient will be located.

All of the variables were selected by reviewing the literature on associations with violence, so they make intuitive, clinical sense. The emphasis throughout is on relative rather than absolute measures of risk. It is acceptable to add up the scores on the different items but it is not mandatory. Qualitative use is encouraged, and the final stage of the HCR-20 involves the team in setting out likely scenarios of violence, estimating their probability, identifying factors likely to increase or decrease the risk, then formulating a management plan to reduce those risks. This approach fits easily into the CPA, as the focus on past, present and future supports proactive management of a case. The HCR-20 measures changes in risk with treatment, so it can be repeated for subsequent CPA meetings in order to show what progress is being made. It is also ideal for completion by a clinical team rather than an individual.

Training is necessary, usually for 3 days, and includes the Psychopathy Checklist-Screening Version (PCL-SV; Reference Hart, Cox and HareHart et al, 1995), which is one of the historical items on the HCR-20. This is not an insignificant commitment, but all staff are now meant to be trained in risk assessment, and the PCL-SV component provides the bonus of training in standardised assessment of personality disorder. Anyway, if trusts are prepared to provide training in their own, untested assessments of risk, they ought to be willing to train staff in an instrument that is widely used and approved in several countries.

Once staff are trained, the time taken to complete structured assessments is often overstated. The time-consuming part is the collection of relevant information, particularly historical records, but clinical teams should have most of this information already. With the information to hand, it is easy to fill in the forms. The requisite team discussion ought to be happening at CPA meetings as a matter of course.

So, for once, no extra research is necessary; we can just get on with it. In fact, we would probably be doing it already, if clinicians had not been sidetracked down the blind alley of a debate about whether or not to assess risk. Whilst we were arguing, the managers got on with it and we are left to struggle with the consequences. It is time for clinicians to take back the initiative and to take the lead in introducing scientifically based, clinically meaningful risk assessment.