Assessing and managing the risk of our patients being violent towards others now occupies a prominent position in virtually all forms of mental health practice, but it remains a contentious area. At the highest level researchers, psychometricians, and statisticians argue about almost every aspect, even whether anything useful can be said about individual outcomes rather than group indicators. At the next level an industry flourishes of selling training, and risk assessment tinstruments, to those who then appear as experts in a wide range of mental health and criminal justice contexts. On the ground, almost everyone in mental health is drawn into filling out purpose-designed forms and complying with protocols, most of little or no demonstrated validity. This chapter is intended to make clinicians aware of both the possibilities and limitations of existing approaches to the assessments of risk. Given that there is no reason for mental health professionals to evaluate risk without gaining information to manage it, this chapter will also address the management of risk for aggression and violence.

A critical analysis will be attempted of how risk has come to be constructed in our society and how this is impacting on mental health and criminal justice. When an approach is adopted which attempts to reveal the foundations and historical evolution of a widely accepted social construct, like risk, there is a danger that it will appear overly sceptical or even mocking. It is important to emphasize at the outset that:

1 The assessment of the probability of patients behaving in ways damaging to others and the management of that risk is a legitimate clinical activity.

2 That attributions of levels of risk to a patient occurs in a social and cultural context and is inescapably a construct.

The discourses around the dangerousness of the mentally ill have gradually been replaced by those of risk. This change is usually presented as a product of the progress of knowledge and the improved conceptualization of that knowledge.^(1,2,3,4) The language of danger transmuted into the language of risk also emphasizes the probabilistic nature of risk assessment. We are not now and probably never will be in a position to be able to determine with certainty who will or will not engage in a violent act. Relying on a range of empirically supported risk factors, though, we can makes a reasoned determination of the extent to which those we are assessing share factors that have been found in others to relate to an increased level of risk. Risk embodies the interaction of a range of factors, which are not necessarily dangerous in themselves, such as age, gender, marital status, ethnicity, employment status and, of course, mental disorder.⁽⁵⁾ Risk factors can be any variables which are statistically associated with a future violent episode or event. There is no assumption of causality linking the predictor to the predicted.

Risk assessment came relatively late to the mental health field. Not until ‘harm to others’ or ‘undue risk’ became criteria for involuntary hospitalization and forensic detention did ‘dangerousness’ of patients assume the spotlight. The focus on risk first surfaced in Western Societies in the 1970s in the context of concerns about damage being inflicted on individuals by the actions, or inactions, of corporate and governmental agencies. These concerns fed the emergence of widely based environmental movements as well as an escalating number of class actions and individually driven litigations. Under the banner of risk a new blaming system emerged of which Douglas⁽⁶⁾ writes ‘we are … almost ready to treat every death as chargeable to someone’s account, every accident as caused by someone’s criminal negligence, every sickness a threatened prosecution. Whose fault? is the first question … then what damages? what compensation? what restitution …’ (pp. 15-16). One response to this culture of blame has been the emergence of what O’Malley⁽⁷⁾ refers to as a new prudentialism in which individuals, professionals and corporations, increasingly held responsibility for the impact of their actions on others, resort to risk management strategies in which risk is assessed, managed, insured against, and where possible removed.

Psychiatrists and psychologists are among those who have become caught up in the ‘culture of blame’. Any damaging or distressing occurrence which is experienced by, or caused by, someone who is, or has been, a patient of the mental health services, is transformed into a preventable tragedy for which professionals are
to be held responsible. Rose⁽⁸⁾ suggests the new imperatives of risk assessment and risk management operate to establish mechanisms to control mental health professionals which through standards, audits and enquiries not only regulate professionals but hold them personally responsible for unwanted outcomes. Douglas (6) argues ‘probability analysis arrives at politics in the form of a word ‘risk’ … the word gets its connection with probability squeezed out of it and put to the same primitive political uses as any term for ‘danger” (p. 48). Risk assessment and risk management are concepts which have the potential to shift blame towards clinicians who have failed to follow procedure and away from managers who fulfilled their responsibilities by ensuring correct protocols were in place, irrespective of the possibility of the realistic application of such protocols. The language of risk can also shift the focus from politicians who determine resources and establish systems of care to those who fail to identify and manage risk in the individual case. Perhaps most importantly the increasing centrality of risk assessment potentially creates a vision of the mentally disordered as primarily embodiments of varying degrees of risk and the mental health services as agents in controlling and obviating the supposed danger to the community.

Assessing dangerousness used to be the almost exclusive province of the forensic mental health professional.^(9,10) It was a marginal activity based on arcane knowledge and assumed wisdom that only experience could provide.^(11,12) Risk assessment and management in contrast have become central to current mental health practice in almost all its guises. It has become among the most important activities defining professional competence. Understanding the cultural, legal, and political roots of the increasing hegemony exerted by the rhetoric of risk over psychiatric practice may demystify, but does not free the professional from the imperatives of operating effectively in this new environment. It seems so obvious, as to be self evident, that mental health professionals are expected to consider the probability that their patient will act in a destructive manner and to act to prevent such harms. But the self evident is often the ideological unconsidered. It is not obvious that a mental health professional’s primary responsibility is to the wider community rather than their patient. It is not obvious that it is possible to effectively predict such risks as they apply in the individual case. It is not at all obvious how we should act in the face of a prediction of risk, and it is certainly not obvious that such concerns should be a major determinant of our approach to patients.

Words are rarely innocent. Risk is not the same as probability, for risk implies a degree of danger. Even those deemed to be at ‘low risk’ still appear to present some degree of danger. No one, it seems, is considered risk free. Risk management is not the same as harm minimization, for it promises a prevention of unwanted outcomes. Psychiatry deals with disorders which have both substantial morbidities and mortalities. Good management may reduce but cannot, in our present state of knowledge, prevent all such morbidity and mortality. Furthermore, reducing morbidity and mortality long term may only be possible at the price of accepting an increased probability of mortality in the short term. Suicide in prison, for example, can be prevented by isolating and observing vulnerable inmates in transparent plastic bubbles, bereft of features from which suspension is possible, or by the simple expedient of chaining them hand and foot to a bed (both strategies are in use today). If the only good is preventing self harm such draconian measures acquire currency irrespective of the psychological damage and abuse of basic human dignity involved. Moreover, as the aim has become removing all risk of suicide in prison, more and more vulnerable prisoners are being subjected to such restrictions and in many jurisdictions ‘witch hunts’ now regularly follow every death across jurisdictions. We cannot force the genie of risk back into the lamp. Mental health professionals will continue to be made publicly accountable. Professional self regulation is being replaced by statutory regulation and the ravages of civil litigation.⁽¹³⁾ As Rose⁽⁸⁾ notes we will be forced to play a central role ‘in the strategy of reducing risk and minimizing harm under threat of sanction and within the disciplines imposed by a plethora of practices of blame’ (p. 18).

It is no longer possible for mental health professionals to distance themselves from the process of risk assessment. Throughout the 1970s and 1980s mental health professionals were almost of one voice in proclaiming both their inability to predict dangerousness and the basic pacifivity of the mentally disordered.⁽¹⁴⁾ Despite this public stance dangerousness criteria came to dominate civil commitment, with courts simply ignoring arguments that the prediction of dangerousness was beyond the ken of psychiatrists and psychologists.⁽¹⁵⁾ In addition liability based on failures to predict dangerousness was established in landmark cases including those arising out of Poddar’s killing of Tarasoff⁽¹⁶⁾ and Hinkley’s attempt to assassinate President Reagan.⁽¹⁷⁾ Last, but not least, the criminal courts increasingly suborned mental health professionals into predicting dangerousness in the pursuit of such sentences as preventive detention and death.^(15,18) In the US and the UK the emergence of the language of risk at the end of the 1980’s was in part a recognition of the centrality assumed by predicting and managing the potential for violence perceived by the public to reside in the mentally disordered. In the UK the public inquiry into the killing of Jonathan Zito by the psychiatrically disordered Christopher Clunis set a pattern for future homicide enquires.⁽¹⁹⁾ Failures of risk assessment and management by individual practitioners, together with inadequacies of communication and service provision, were identified as major contributors to ‘avoidable’ killings.⁽¹⁹⁾ Such developments set risk assessment at the very centre of the mental health agenda.

Given the challenges of risk assessment, what is the current state of knowledge and can it be of assistance to clinicians? The limited research available on ‘dangerousness prediction’ conducted in the 1970s showed that psychiatrists and psychologists had unacceptably low levels of accuracy in predicting which patients would go on to be violent in the future.⁽²⁰⁾ It was found, perhaps not surprisingly, that psychiatrists, psychologists, and release decision-makers tended to make conservative decisions that suggested that people were at risk for dangerousness or violence when, in fact, they were not. Similar findings have been obtained over time. For example, Belfrage⁽²¹⁾ found that clinicians found 90 per cent of a group of 640 offenders sentenced to psychiatric treatment in Sweden to be at ‘risk of severe criminality;’ when, in fact, only 50 per cent went on to commit any kind of crime.

Reasons given for the errors made in risk prediction include clinicians’ confusion and lack of knowledge of valid risk markers
and risk factors. In addition, as with other areas of decision-making, even if clinicians do have a reasonable understanding of risk factors, it is difficult to systematically consider them and to put together a risk appraisal in a systematic way. Advances in risk assessment have included the identification of an expanded range of predictor variables relevant to violence. Most important among these are those variables that are subject to change (i.e., they can change over time and they can be influenced by treatment or other intervention). Generally speaking, risk assessment variables can be classed as ‘static’ (i.e., those that cannot be changed) and ‘dynamic’ (i.e., those that can change over time). Actuarial risk schemes, which will be discussed later in this chapter, are based upon variables that were measured from the past. These historic variables generally could not change over time. For example, if one began being violent as a young person, that fact will not change over time. Dynamic variables, in contrast, are subject to change over time, sometimes rapidly. These variables include such things as state of mind, situational factors, attitudes, plans, support, etc. Effective risk assessment must take into account both static and dynamic variables; however, risk management generally requires an understanding of the dynamic risk variables. Contemporary approaches to risk assessment and management take into account both static and dynamic variables, thus considering an individual’s past, present, and future risk factors that might affect the likelihood of him or her becoming violent.

There has been considerable progress since Monahan⁽²⁰⁾ first reported that psychologists and psychiatrists were essentially unable to predict risk to any acceptable extent. Current research shows that risk assessment approaches provide a level of accuracy that now far exceeds chance.⁽²²⁾

There are two legitimate perspectives on risk assessment:-

1 The clinician whose work involves considering patients’ levels of risk as part of a process whose purpose is primarily to improve the management of patients.

2 The forensic evaluator for whom risk assessment is a tool to improve the reliability of opinions provided to courts and tribunals charged with making decisions about an individual.

Forensic mental health professionals in the US and Canada tend to see their role primarily as evaluator. It is not uncommon for forensic psychiatrists and psychologists to primarily conduct assessments for the courts as their means of employment. This reflects the generally accepted separation of the assessment and treatment roles. In stark contrast in the UK and Northern Europe the clinician perspective dominates even for forensic specialists. Relatively few forensic clinicians would operate as court evaluators alone, though some like their US colleagues would try and separate the roles.

This chapter is written for mental health professionals in a wide range of contexts, not for forensic specialists. As a result the emphasis on this chapter will be on the clinical perspective. North American readers should, however, keep in mind that this perspective is not shared by many specialists in the forensic field there who would consider it not just problematic but ethically questionable to mix the assessment and treatment roles.

Boundaries need to be drawn around when, where and for what purpose, mental health professionals can ethically engage in assessing the probability of an individual committing violent or criminal acts. There are somewhat different constraints operating on clinicians than for those carrying out evaluations for the purpose of preparing reports for decision-making bodies like courts and tribunals.

The ethical and practical constraints on a clinician assessing risk entirely in service of effective treatment include:

1 Ensuring the assessment serves the interests of the patient in terms of improving management and protecting them from acts which will damage their interests. The seriously mentally ill when they become violent all too often target those who support and care for them thus destroying the relationships critical to their own social survival. In addition, criminal and violent acts of the sort usually associated with the mental disordered rarely, if ever, brings anything but increased problems for the patient. Thus, reducing the likelihood that the patient will engage in criminal and violent acts serves not only the public interest but also the patient’s interests.

2 Mental health variables (which include psychological variables and personality traits) are a prominent feature of the individual’s clinical picture and are also of potential relevance to the probability of future damaging behaviours.

3 Avoiding providing greater emphasis to risk than is necessary given the totality of the patients needs and vulnerabilities.

4 The assessment should wherever possible connect potential risk to those factors whose amelioration will reduce that risk. Ultimately a health professional can only justify engagement in risk evaluation if they lead to better outcomes for the evaluated as well as the community. Risk assessment finds its ultimate justification in risk management.

5 Avoiding using risk and risk assessment to disqualify patients from access to the treatments they require. Increased risk requires increased therapeutic enthusiasm, not rejection.⁽²³⁾

6 That any concerns raised by a risk assessment are shared with the patient and the proposed management strategies explained. Even unwelcome restrictions are resented less if the reason for their imposition is explained.

7 Retaining an awareness of the limitations of the predictive power of risk assessments and the need to ensure proportionality between the risk actually apprehended and any imposed remedy.

8 Ensuring a level of professional competence adequate to the task.

9 Making use, where possible, of the skills and knowledge of the multidisciplinary team in the assessment.

Those obligations on those engaging in risk assessment as part of an evaluation for a decision-making body are even more onerous and should include:-

1 The patient consents to the examination in the knowledge of the nature of the assessment, the purposes to which it may be proffered, and the limitations on confidentiality that may apply.

2 A reasonable body of empirical evidence exists to guide the risk assessment including, where possible, empirically validated structured risk assessment measures.

3 The risk assessment is conducted in consideration of the legal parameters governing the decision-making body (e.g., criteria to be considered for change of orders or release for forensic psychiatric patients) while realizing that the legal questions to be answered never parallel the clinical or evaluative results mental health professionals can reasonably provide (e.g., there is no clinical parallel to legal criteria such as ‘undue risk’).

4 The assessment is based on a careful analysis of the relevant characteristics of the particular individual which in all but exceptional circumstances have been obtained in part by a direct examination of the individual.

5 The risks are expressed in terms of probabilities (not attributions of dangerousness) with clear admissions of the fallibility and potential variability in the prediction. The problem should be acknowledged of employing risk factors derived from studies on populations from different cultures and contexts. After all nobody would use risk data from Los Angeles to calculate the car insurance premium for a driver in Dublin or Oslo without considerable caution.

6 Account is taken not just of the probability of damaging behaviour but the nature and severity of such conduct. Proportionality needs to be maintained between what is predicted and the response. It is all too easy to employ methods which establish increased risks of a wide range of unwanted behaviours only to find them used to justify draconian and punitive responses which would only be acceptable in the face of the imminence of serious violence.

7 The confidence and certainty with which any prediction is formulated to take account of the implications for the person being assessed. Risk predictions may be offered in terms of probabilities but they will almost always be used to justify all or nothing decisions.

8 That personal and professional integrity is strictly maintained. This is no simple matter when the evaluator is either in the pay of the patient, or of those whose interests are not necessarily those of the patient.

The potential conflicts generated when acting for a patient as both a professional evaluator and a clinician are so considerable that some experts argue that such a dual role is inherently unethical. They argue in preparing reports for decision-making bodies the interaction ceases to be that of health professional and patient, and becomes entirely that of expert, or ‘forensisist’, and evaluated. The expert’s obligation is then not to the evaluated but to their own professional competence and the rules governing the process the report will serve (e.g., criminal court, family court, mental health tribunal).

In our view even in an encounter between a health professional and a person purely for the purposes of an evaluation for a court there remains obligations to the person as patient [patient from the Latin for to suffer, who in this instance suffers the intrusions of an examiner who cannot for the evaluated entirely cast off the guise of physician or healer]. The solution is to learn to live with the contradictions and accept the dialectic between responsibilities to the patient and obligations to the agencies of society. The result of accepting the duality of the role inherent in assessing a patient’s risk of harming others does exclude participation in a process that could increase the risk to the patient of fatal (e.g., death penalty evaluations) or serious harm (e.g., sexual predator laws whose sole purpose is justify a process of prolonging incarceration beyond the expiry of a sentence). It does however legitimate having the dual role of clinician and evaluator in some circumstances. In fact those who totally eschew ever taking on such a dual role, we believe, are at risk of deluding themselves that they can caste off the mantle of clinician for the patient, and become a socially neutral, objective, observer and reporter.

The dialectic between the demands of a professional responsible to the health of your patient and the demands of professional integrity and honesty owed courts and tribunals is almost always possible to resolve. To fail to accept engagement with the conflicts which in reality usually exist for evaluators and clinicians reporting to external authorities is in our opinion an act of self deception in which you become an agent either of patient, or authority, and no longer an autonomous responsible professional.

Some situations make nonsense of the above considerations, notably death penalty hearings. One of us (J.O) has extensive experience of working with those on death row, the other (P.M) only a comparatively slight acquaintance. We are both of the opinion, however, that it is impossible to honestly discharge your responsibilities as clinician, as evaluator, or as decent human being, in such circumstances.

There are five basic approaches to evaluating the risk of violence:-

1 Probability models based on established risk factors. The risk factors can be derived actuarially from studies of particular populations (e.g. Violence Risk Appraisal Guide (VRAG) and Static 99) or rationally ascertained from the risk literature (e.g. Historical Clinical Risk 20 (HCR-20)).

2 Clinical experience based on recognizing previously encountered (personally or in the literature) patterns associated with future violence. The clinical approach is largely relevant to the avoidance of obvious errors like discharging morbidly jealous men who are threatening to kill their partner.

3 A mixture of 1 and 2 where the risk assessment instrument is employed to guide the appraisal of risk factors and clinical judgment is applied to balance idiographic information with the nomothetic variables as in the structured professional judgment approach of which the HCR-20 is the prime example.⁽²²⁾

4 The strictly idiographic approach which employs individual profiles of violent offenders to detect those on a similar pathway to attack. The idiographic approach is employed to evaluate the risks of rare events, such as attempts to assassinate a head of state, and has little application in general mental health.

5 A plethora of local risk assessment tools have sprung up. Sometimes it seems as if every psychiatric service, probation/parole service, prison, and security consultant have their own unique sheet of questions which are supposed to establish the future probability of whatever particular piece of nastiness currently concerns the organization. These ad hoc parochial risk assessment protocols have no evidentiary basis or psychometric integrity (even if they incorporate aspects of other properly constituted instruments). It is far better to validate existing
empirically supported measures for use in a particular setting and with a particular population.