The assessment of the probability of mentally disordered offenders committing violent and criminal acts continues to be subject to extensive research. If such research is to be applicable and ethically acceptable, it must be grounded in clinically-based studies that link risk assessment to management strategies capable of ameliorating that risk. The profession of forensic psychiatry should not restrict itself to being an entrepreneurial risk assessment industry without ties to the traditional roles of medicine: to treat and to care. Thus, treatment and rehabilitation need to occupy a central position in forensic mental health. A scientific evaluation of our present practice will form a valid base for risk prediction that is sensitive to the dynamics of forensic practice. These studies should utilise both qualitative and quantitative methodologies, and focus on the questions of whether or not, to what extent and when forensic psychiatric rehabilitation alters the individual's level of risk. There is a dearth of literature addressing these issues, which leaves clinical practice without a solid scientific base for its actions. Variables derived from risk research presented hitherto are largely static and disregard and potential for rehabilitation to change the mental state and behaviour of an offender. In losing sight of this central task of forensic mental health they seriously diminish it as a discipline. This paper discusses some essential features of a forensic psychiatric rehabilitation system in an endeavour to create a conceptual framework for research and practice. We suggest that this research paradigm will be supportive of, and conducive to, ongoing developments and advances in rehabilitative practice in forensic mental health.

THE EVOLUTION OF RISK RESEARCH

Several studies conducted during the 1990s have shown that there is a modestly increased risk in violent behaviour among people with major mental disorders when compared with the general population (Reference Swanson, Holzer and GanjuSwanson et al, 1990; Reference Link, Andrews and CullenLink et al, 1992; Reference MonahanMonahan, 1992). One question that arises from such research is in which subgroups of patients and in which contexts does violence by people with mental illnesses most frequently occur. The development of actuarial risk assessment tools has coincided with this second generation of research, and much faith is invested in these instruments from which probability predictions are made. In our minds, this is a hazardous tack on which to embark if not properly safeguarded. These tools are largely based on static and historical factors that cannot hope to be targeted or changed by rehabilitative endeavours.

One risk is that some patients, falsely associated with a high actuarial risk score (false positives), will be locked up because of ‘guilt by statistical association’. The equally unacceptable risk is that those patients with a falsely reassuring low actuarial risk score (false negatives) may persuade judges or review panels that they are ready to be released despite clinical concerns. Finally, the correctly identified group with a high probability of future violence is in danger of having the label ‘dangerous’ attached. This risks their being treated as if they have an immutable quality of viciousness rather than possessing a range of properties and predispositions, many open to modifications.

There is evidence, although disputed, that mental health professionals' predictions of violence are substantially better than chance (Reference MossmanMossman, 1994). It appears that the addition of actuarial risk assessment tools to clinical judgement alone improves the accuracy of determining a patient's risk of recidivism (Reference BorumBorum, 1996; Reference Gardner, Lidz and MulveyGardner et al, 1996; Reference Bonta, Law and HansonBonta et al, 1998). However, the currently used predictor variables of recidivism are similar in different offender groups, irrespective of any contribution from mental illness, and are often regarded as largely immutable. This does not obviate the task of psychiatric rehabilitation, nor render the treating team jail wardens awaiting the expiry of a sentence to indicate that community placement is appropriate. That forensic psychiatric populations compare favourably with matched criminal populations with respect to recidivism is not surprising, given the vastly disparate nature of their treatment and rehabilitation. Better questions to pose then may be what is there to treat, and what is it about forensic psychiatric rehabilitation that works?

FUTURE RISK RESEARCH

Delineating dynamic risk variables (or treatment factors) is the first step towards investigating the effectiveness of forensic psychiatric rehabilitation. However, such pursuits must avoid overemphasising the reliability of these predictor variables without due consideration of their validity. That is, although a dynamic risk variable translates into a potential focus for treatment, even if its reduction reliably predicts a reduced risk of recidivism, at this point we have not ascertained the validity of this construct in rehabilitation — hence the importance of further clinically-based research.

The need for more and better scientific evaluations of treatment programmes is therefore overdue in the evolution of risk research, because successful interventions to reduce risk will need to be founded on both reliable and valid assumptions. There is an inherent risk in the clinical setting that in viewing the notion of risk assessment purely within a quantitative paradigm based on static and historical factors we will potentially dilute the effectiveness of other rehabilitative endeavours that aim to change the mental state and behaviour of an offender.

DYNAMIC RISK VARIABLES

Relatively static risk factors such as personal demographics and personality characteristics are uncommon targets for rehabilitation, yet they form the core of all risk assessment tools currently in use. It is predictable that a patient who has a history of persistent antisocial behaviour and falls ill with a major mental disorder will be an ongoing risk to the public even after the mental disorder is controlled. In risk rehabilitation we must focus on risk factors that potentially can be changed.

Risk assessment can be linked to at least four dynamic features of the individual patient and their treatment setting, each with varying resistance to change, and necessitating a different therapeutic task:

1. (a) the disorder itself, where active psychotic symptoms can create dangerous situations, especially if the person feels threatened, controlled or is commanded to harm others (Reference Link, Andrews and CullenLink et al, 1992; Reference Wessely, Buchanan and ReedWessely et al, 1993; Junginger, Reference Junginger1995, Reference Junginger1996);

2. (b) family problems and poor sociocultural circumstances, including the extent and quality of interactions with family, peers and society;

3. (c) substance misuse, which can result in impairments of judgement, empathy, temper and impulsiveness, both on the basis of acute intoxication and the sequelae of long-term misuse (Reference LindqvistLindqvist, 1986; Reference Lindqvist and AllebeckLindqvist & Allebeck, 1989; Reference Swanson, Holzer and GanjuSwanson et al, 1990);

4. (d) anti-therapeutic system dynamics creating stigmatised, expelled, underrated, incarcerated or misunderstood mental patients, with reciprocal friction and aggression.

Many forensic patients have multiple and complex diagnoses and are subjected to treatment that varies widely in quality and efficacy. This may be one reason for the dearth of literature on treatment evaluation. What we are to analyse is almost as difficult as capturing the wind. What is evident, however, is that any assessment that looks only at the patient and neglects to analyse the forensic system within which the individual lives will reveal only part of the answer to what will constitute optimal treatment for that individual. In the past, there has been insufficient analysis of those factors external to the patient and of their influence on the prognosis. It is, therefore, essential to focus therapeutic rehabilitative endeavours on factors within both the person and the rehabilitative environment in which they live.

FORENSIC PSYCHIATRIC REHABILITATION

Rehabilitation is a process where the outcome is the result not only of the sum of individual contributions within the treatment system, but also of the interacting effects of these various efforts. At best these contributions will be synergistic and complementary; at worst, the effect may be negative and associated with a detrimental outcome. These individual contributions or factors are innumerable but we believe that some items preferentially determine the outcome and form the core of a therapeutic culture and process of rehabilitation. We will not address in any detail the specific treatment of positive and negative symptoms, depression, aggression, criminality, life skill deficits and substance misuse because these have been reviewed extensively elsewhere (e.g. Reference Rice and HarrisRice & Harris, 1997). In our view, they are also to an extent subservient to the importance of a therapeutic culture.

RESEARCH ON TREATMENT

Any research aimed at analysing the effects of forensic psychiatric rehabilitation will be hampered by:

1. (a) the complexity of forensic treatment systems;

2. (b) the problems of constructing randomised controlled study designs in respect of patients and treatment systems;

3. (c) the difficulties in defining and operationalising concepts important in the process of recovery;

4. (d) the problems associated with having low base rate outcome measures;

5. (e) incarceration effects, and the effects of ageing;

6. (f) the inability of current risk assessment tools to assist reliably in rehabilitative work.

Measuring the post-discharge effect of forensic rehabilitation is fraught with difficulty, whether one works within a quantitative or a qualitative paradigm. However, the description and analysis of the treatment system requires recognition as an important variable in the scientific analysis of outcomes in this area of research. Gottfredson (Reference Gottfredson1984) proposed programme development evaluation as a way of measuring the quality and internal integrity of interventions, and this approach has been usefully applied to forensic treatment settings (Reference Rice, Harris, Quinsey and WeisstubRice et al, 1990).

Another way to circumvent the temptation to launch reductionistic studies that focus on one easily-measured item at a time would be to organise a number of separate forensic psychiatry services into a joint collaboration, creating a situation that resembles a natural experiment. The legal and professional practices vary within and between countries and states, whereas in our experience patient characteristics and the nature of offences do not know national borders. With this design, pre-admission data, treatment factors and post-discharge information could be collected and configured uniformly at several sites simultaneously, creating the possibility for comparative research. Such a collaborative approach would allow both qualitative and quantitative research endeavours, leading to the development of more valid and reliable measures of prognostic significance in rehabilitation with these populations. Hopefully, this will capture some core elements of good forensic psychiatric practice, which will enable us to throw away some bad practices and begin to create a more solid scientific, evidence-based practice in forensic psychiatric rehabilitation.

Clinical Implications and Limitations