Sage Journals: Discover world-class research

Abstract

Objective:

From time to time misconceptions about violence risk assessment raise debate about the role mental health professionals play in managing aggression, with associated concerns about the utility of violence risk assessment. This paper will address some of the misconceptions about risk assessment in those with serious mental illness.

Methods:

The authors have expertise as clinicians and researchers in the field and based on their accumulated knowledge and discussion they have reviewed the literature to form their opinions.

Results:

This paper reflects the authors’ views.

Conclusion:

There is a modest yet statistical and clinically significant association between certain types of mental illness and violence. Debate about the appropriateness of clinician involvement in violence risk assessment is sometimes based on a misunderstanding about the central issues and the degree to which this problem can be effectively managed. The central purpose of risk assessment is the prevention rather than the prediction of violence. Violence risk assessment is a process of identifying patients who are at greater risk of violence in order to facilitate the timing and prioritisation of preventative interventions. Clinicians should base these risk assessments on empirical knowledge and consideration of case-specific factors to inform appropriate management interventions to reduce the identified risk.

Keywords

Violence risk assessment risk management mental illness

Introduction

Violence risk assessment is critical to contemporary general mental health. The primary goal of violence risk assessment is the prevention rather than the prediction of violence. To be effective, violence risk assessment must be linked to a process of better management of patients that raise concern, by managing risk factors. The important issue here is that it is not the violence that is being managed, but the risk factors for violence (which relate to propensity). While related concepts, propensity refers to tendency, and prediction to actual outcome.

Some have suggested that psychiatrists and others should not engage in violence risk assessment (Large et al., 2011; Ryan et al., 2010). This argument is based purely on a statistical perspective, focuses on the limits of prediction, and ignores clinical, legal, social, moral ethical and contextual imperatives. Experts have long accepted that prediction (that is, identifying which individuals will or will not be violent) cannot be achieved (Cocozza and Steadman, 1976; Monahan, 1981; Ogloff and Davis 2005). This is not new knowledge. But to argue against violence risk assessment because prediction is not possible is to miss the point; and most worryingly, potentially leads to the abandonment of violence risk management and thus prevention.

This paper will examine the role of mental health clinicians in violence risk assessment and management in the context of contemporary mental health systems in developed countries.

Methods

Due to the complexity of this topic, we did not conduct a systematic review, which would have included using key terms to identify relevant literature in recognised electronic databases. Instead, the arguments we present are a summary of accumulated knowledge gained by the authors, keeping up date with the literature and discussing amongst each other and with colleagues nationally and internationally over the years. The authors have expertise as clinicians and researchers in the field and based on their accumulated knowledge and discussion have reviewed the literature to form their opinions.

Results

The following is an account of the authors’ views.

Understanding the problem

It is inevitable (as with all medical decision making) that clinicians will from time to time be asked to form an opinion on the probability of an outcome. In this context they might express their opinion as a percentage or rate and thus make predictive statements. There is clear evidence that in doing so, using assessment approaches that are structured and have an empirical basis provides stronger predictive validity and are more reliable than unstructured clinical judgement alone (Ægisdóttir et al., 2006; Hanson and Morton-Bourgon, 2009; Mossman, 2000).

Mental health clinicians are regarded as experts in the assessment and management of disturbing behaviours derived from mental illness. The community has a long history of expecting mental health professionals to assist in this process and in deciding who requires detention and treatment under restrictive conditions. For example, Section 26(2)(b), Part IV, of the Mental Health Act, 1959 (Vic) [7 & 8 Eliz. 2, Ch. 72] provides that ‘An application for [involuntary] admission for a treatment may be made in respect of a patient on the grounds … that is necessary in the interests of the patient’s health or safety or for the protection of the other persons that the patient should be so detained’ (emphasis added). This is reflected in current Mental Health Acts, which afford psychiatrists power to involuntarily detain and treat people with serious mental illnesses. In addition, the judiciary and adjudicators (e.g. parole boards, mental health boards or tribunals) rely heavily on clinicians’ opinions in dispositional decisions that result in the detention and supervision of people against their will, or release.

Patients with mental illness could at one time be locked away for extended periods on the signature of a doctor. However legislation has curtailed clinicians’ powers. Modern notions of due process and natural justice have led to increased human rights protections in mental health legislation. Despite these safeguards, mental health clinicians are still perceived as broadening the traditional treating role to one of public protection. Clinicians should not lose sight of the possibility of arbitrary and capricious deprivation of liberties, which has occurred in the past.

The management of patients with mental health problems requires consideration of numerous outcomes including: relapse, abuse of substances, self-harm/suicide and psychosocial disadvantage. While not all patients are at risk of every one of these adverse outcomes, clinicians need to be aware of the possibility of such outcomes and when identified, take reasonable action to intervene by appropriate management. Violence can be regarded as a behavioural complication of serious mental illness in some sufferers and an outcome that requires assessment and management (Mullen, 2006). It is thus understandable that mental health professionals have an obligation to identify and manage risk of this outcome as well. Without first conducting a thorough assessment of the relevant risk factors, effective management strategies to assist with wellness, recovery, adaptation and prevention cannot be identified. Taking action only after an incident does a disservice to the patient and the broader community.

The ultimate expectation of any health service is the prevention of premature mortality and amelioration of morbidity. Mental health services are no different. In mental health, the most severe form being self-harm/aggression and suicide/homicide. Mental health professionals must develop skills and methods to assertively manage that possibility (Mullen and Ogloff, 2009).

Understanding the type of population the person falls into enables awareness of the propensity for violence and the consequent management of the patient according to their individual needs. Risk assessment is the process of identifying empirical and clinically derived risk factors. Risk management is a process of ameliorating this propensity through multidisciplinary intervention, addressing the identified risk factors that moderate the interaction between mental illness and violent behaviour (Mullen, 2006).

There is an empirically established relationship between serious mental illness and violence based on studies of prisoners, mentally ill offenders, mentally ill people who do not offend and the general population (Mullen et al., 2000; Swanson et al., 1990, 2006a; Wallace et al., 2004). This relationship is mostly studied in those with schizophrenia and psychosis not otherwise specified. Douglas et al. (2009) and Fazel et al. (2009) in comprehensive meta-analyses investigated the relationship between psychosis and violence. Results revealed a modest yet statistically significant and clinically important relationship even when controlling for moderating variables. While psychosis alone is a statistically significant risk factor for violence, the risk becomes greater when a patient has co-morbid substance misuse (Douglas et al., 2009; Fazel et al., 2009; Wallace et al., 2004) and/or an antisocial personality disorder (Tengström et al., 2004). Many other, less statistically significant, factors also have a moderating influence on the relationship between psychosis and violence (Douglas et al., 2009).

Violence by those with mental illness has a significant impact on the victim, the victim’s family, the perpetrator’s family, carers, staff and the patient. Violence adds to the stigma that those with mental illness are to be feared. The publicity that follows a violent offence committed by a patient with mental illness, in particular where there is a coroner’s finding that psychiatric care was inadequate, does damage to the public image of patients with mental illness and public confidence in mental health services (Hall, 2011). Violence also causes significant difficulties for the patients themselves. Most patients dislike acting in an aggressive manner, often request assistance to refrain from such behaviour, and often suffer considerable guilt and shame afterwards. Their aggression erodes relationships, and results in restrictions on liberty including confinement in restrictive, anti-therapeutic environments such as prison.

In addition there are important clinical decisions that require careful consideration of the potential of the risk posed. For example, there are circumstances where confidentiality can be breached to protect third parties. Professional codes of ethics allow (and when children are involved, expect) clinicians to disclose confidential information under certain circumstances, such as serious concern about risk of violence (Kaempf et al., 2009) to a named person.

Dealing with uncommon events

Commentary suggesting that violence risk assessment is inappropriate because the prevalence (base rate) of violence is too low (Large et al., 2011) requires explanation. The argument that one cannot accurately predict who will or will not be violent is an old one, emphasised years ago by Monahan (1981), based on his review of studies using unstructured clinical judgement, in the absence of adequate research. Monahan’s findings led to the development of better assessment procedures.

There is no one fixed rate of violence in the mentally ill. Reported rates of violence vary according to the severity and nature of violence measured, how information on violent behaviour is sourced, the population studied and the duration of follow-up.

Swanson et al. (2006a) examined a wide spectrum of patients with schizophrenia, and found a 6-month prevalence of 3.6% for serious violence. Douglas et al. (1999) found that 10% of patients involuntarily hospitalised in general psychiatric hospitals discharged to the community were either arrested or convicted for a violent offence (serious violence) during a follow-up period of approximately 20 months (1999). On the other hand, base rates for ‘any violence’ over a period of 1 year in the general community has been reported to be between 25% and 30% for general psychiatric patients and higher for specific populations of people with mental illnesses and offence histories (Douglas et al., 1999; Hodgins et al., 2007; Nicholls et al., 2004; Steadman et al., 1998; Swanson et al., 2006a; Walsh et al., 2001; Wootton et al., 2008). Steadman et al. (2000) found that more than one-third (35.7%) of 10,000 involuntarily committed patients released to the community engaged in some form of violence within 20 months of discharge.

In psychiatric hospitals the base rate of violence is much higher, with low severity aggressive behaviour being remarkably common (Daffern and Howells, 2002). Assessing risk of aggression in this context is critical, and risk-related decision making is central to decisions about leave, escorting requirements, the need for a patient to be moved to a high-dependency unit and even the need for medication (Ogloff and Daffern, 2006).

Rates of violence increase, to a point, with the presence of additional risk factors. For example, in the Wallace et al. (2004) study, the rate of violent offending in those with schizophrenia alone was 8%, but rose to 26% in those with schizophrenia and a co-morbid substance use disorder. In a study by Wootton et al. (2008), 22% of general psychiatric patients discharged and followed up for 2 years committed an assault but 37% of those who manifested other risk factors (younger age and male, assault in the previous 2 years and substance abuse in the previous 1 year) committed an assault within 2 years. In the MacArthur Foundation study (Steadman et al., 1998) the 1-year aggregate prevalence of violence was 17.9% for patients with a major mental disorder discharged from general psychiatric hospitals and 31.1% with a co-morbid substance abuse diagnosis were violent; substance abuse contributed significantly to violence in this population.

To illustrate the above point, we consider the assessment of violence risk in a general psychiatric population using the Historical, Clinical Risk, Management-20 (HCR-20), a tool to guide the structured professional judgement (SPJ) approach (discussed below) to violence risk assessment (Webster et al., 1997). The tool has been subject to more than 130 studies, finding good reliability (e.g. inter-rater reliability = 0.80–0.85) and moderate to strong predictive validity (e.g. area under the curve = 0.65–0.89) (Douglas and Reeves, 2010). A study in Canada examined the accuracy of the HCR-20 in involuntary psychiatric patients discharged to the community from the hospital (Douglas et al., 1999; Nicholls et al., 2004). The rate of ‘any violence’ (‘any violence’ included acts of physical violence (any attacks on persons), non-physical violence (serious threats to harm, verbal attacks), and criminal violence (criminal charges or convictions for contact offences or robbery)) within 20 months of discharge to the community from general psychiatric inpatient services was approximately 38%. Patients who scored above 25 (out of a potential 40) on the HCR-20 at discharge fell into a population of patients of whom about 75% engaged in ‘any violence’ within approximately 20 months. Patients who scored below 15 fell into a population of patients of whom about 10% engaged in ‘any violence’.

These prevalence rates are not insubstantial and on this basis it would be fair to say that patients with schizophrenia (and to a lesser but not inconsequential extent patients with other serious mental illnesses) are at an increased risk of aggression particularly within certain settings. Furthermore, higher scores on some risk assessment instruments should cause clinicians to pause, carefully consider their care planning and put in place risk management strategies to avert violence and other aggressive behaviours.

While the HCR-20 is an SPJ tool it can, for research purposes, produce numerical scores by totalling the 20 items, producing total scores in the range 0–40. This ‘actuarial’ approach is not advocated in clinical practice (nor do the authors of the HCR-20 advocate this approach), because it is a ‘tick box’ approach and ignores other clinical factors.

In practice, however, it is not researchers or commentators who determine what is or is not an acceptable threshold required for an intervention. Rather, it is the clinician, having regard to clinical, personal and social context of the particular patient that defines the intervention. For example, if a patient was assessed (on an actuarial or SPJ assessment) to cause concern for aggression and was incorporating their mother, with whom they lived, into their delusional system and making threats to harm her (that is, on clinical assessment), the clinician might regard more frequent home visits and/or temporary placement in alternative accommodation as appropriate. If, however, the same person was assessed to be of concern for aggression (on an actuarial or SPJ assessment), was not living with their mother, or incorporating her into a delusional system or making threats (clinical assessment), the clinician might consider less restrictive alternatives. Arguments against risk assessment based on rates of violence found in empirical studies always overlook individual contexts in which clinical decision making occurs (Douglas and Skeem, 2005).

Further, empirically reported base rates of violence are likely underestimates of the true base rate for a number of reasons: family members are commonly victims and less likely to report violence; people with mental illness are often diverted into mental health services rather than the criminal justice system; self-report is probably limited by a need to avoid sanction or achieve social acceptability; there are problems with recollection; and clinical documentation is often inaccurate with regard to violent incidents. Empirically reported rates of violence are also concerned with people released to the community and followed up by mental health services, thus reflecting rates of violence despite management of the patient. One can only be concerned what the empirical rates would be if mental health services did not actively manage violence in those deemed to be at risk. Most studies are of managed patients.

Finally, lower base rates of ‘serious violence’ (as opposed to ‘any violence’) should not cause particular concern to scholars and clinicians involved in violence risk assessment. While it has long been accepted that events with especially low base rates, such as homicide, will be nearly impossible to predict (Szmukler, 2001), most forms of violence have much higher base rates, but are also harmful and low severity aggression often precedes serious violence. The base rates of ‘any violence’ in specific populations (such as those involuntarily detained in civil psychiatric hospitals, and more so in those with a high loading of risk factors) are significantly higher than the general population noted above.

The ultimate goal is to manage specific populations of patients with an increased propensity for ‘any violence’ (those with risk factors). This approach is more likely to capture those at risk for more serious and sometimes rare violent events such as homicide. Not just ‘serious violence’, but ‘any violence’ engaged in by patients requires the best clinical attention. This certainly does not mean that these patients always require involuntary admission, as suggested in recent publications (Large et al., 2011; Ryan et al., 2010). In most cases what they require is adequate psychiatric assessment, treatment and care to address clinical needs.

Risk assessment in the mental health context

The process of violence risk assessment is similar to other diagnostic and prognostic processes in other fields of medicine. That is, risk factors are identified through clinical enquiry. Consider the task of determining the risk of a person suffering a cardiac event; this is often based on the presence or absence of risk factors such as hypertension, smoking, prior cardiac events, family history and high cholesterol. The clinician integrates objective information with subjective perception about whether or not a patient falls into an at-risk population, and decides on the intervention and course of action/treatment. Risk assessment and management is a fundamental medical process.

Simply allocating patients to a category of ‘low, medium or high’ risk, within the complexity of real-world risk management decisions is of course insufficient, as it would be in the assessment of a cardiac event. A more nuanced approach is now recognised in sophisticated critiques of the field of violence risk assessment (Mossman, 2006). Contemporary approaches to risk assessment and risk management require clinicians to formulate and understand the risk. Risk formulation is a description of the potential nature of violence, the patterns (escalation, de-escalation and persistence), potential victims, underlying motives, antecedents, and perpetuating and protective factors. Some suggest also reporting on the most likely scenarios to arise if the behaviour is to recur (Mullen and Ogloff, 2009). This provides others with a far better understanding of the problem rather than a statement that the person is ‘low, medium or high’ risk. These categories are relatively meaningless in clinical situations (unless there is a shared understanding of the definition of these terms and they are attached to a particular intervention protocol).

Such categorical statements are more relevant in legal situations where experts are asked to provide an opinion on risk of recidivism in categorical terms. In those circumstances the proportionality of the risk is legally determined and often complex. For example, the proportional level of risk required to further detain a serious sex offender after expiry of a term of incarceration has been determined to be not higher than 50%. (The Court of Appeal in Tillman v Attorney General for New South Wales [2007] NSWA 327, referred to the term ‘likely’ in s 17 (3) of the Crimes (Serious Sex Offenders) Act 2006 to mean ‘a degree of probability at the upper end of the scale, but not necessarily exceeding 50 per cent’.)

Nevertheless, it is worth noting that there is evidence that legal adjudicators who receive information in a descriptive form that demonstrates an analysis of risk factors and provides a risk management plan are more likely to release patients found ‘not guilty by reason of mental illness’ than if they received information that was only categorical or predictive form (Dolores and Redding, 2009). This supports the view that proper risk assessment reduces inappropriate detention rather than aggravating it. This also provides a degree of reassurance to clinicians that the literature supports the assumption that management recommendations will be acknowledged and potentially incorporated into dispositional options by legal decision makers.

The argument against violence risk assessment is sometimes based on the concern that an unacceptable number of people will have their rights impinged, because an unacceptable number of patients who would not have been violent will be involuntarily detained (Buchanan, 2008; Buchanan and Leese, 2006; Large et al., 2011). Arguing against risk assessment based on the potential consequence of involuntary admission is to confound the issue. If the option of involuntary admission based on risk was removed from the statute, the need for the assessment of violence risk in mentally ill patients would remain, and probably be of even greater significance given that aggressive behaviour is a complication of severe mental illness and a potential mechanism to manage it in some cases, removed.

The predictive utility of risk assessment instruments is sometimes (but rarely) measured by the number of people needed to be detained to prevent one violent episode (NND) (Buchanan, 2008; Buchanan and Leese, 2006). This is an unhelpful way of expressing utility if it is read as implying that the purpose of risk assessment is to identify those requiring involuntary admission. It also improperly implies that involuntary admission is the only intervention in cases where risk of violence is identified. Obviously, this is not the case. Given that the purpose of assessment is to ‘assess’, violence risk assessment is not and should not be considered an intervention or treatment. The intervention or treatment is the risk management. It would be preferable to consider the number of patients needed to assess (NNA) to correctly identify one patient who goes on to be violent.

Whether or not psychiatry should have a role in the removal of a person’s civil rights is a valid and worthy argument. While the practice of psychiatry occurs in the context of a ‘coercive backdrop’ of Mental Health Acts (MHAs), the presence of a MHA is a legislative fact that psychiatrists are expected to have regard to. Until there is legislative change, if there is an expectation of psychiatrists to involuntarily detain, then decisions should be made utilising the best available evidence, even if violence risk assessment has limitations. Utilising a non-empirical and unstructured clinical approach will result in even more people being erroneously detained because these types of judgements are the most inaccurate risk assessment method, and unstructured risk assessments typically result in an overestimation of risk.

Some have regarded violence risk assessment in psychiatry as a ‘new’ process, becoming the focus of mental health practice, seeking to bring the present into the future and making it calculable, making clinicians agents of control and in the process contributing to social exclusion and discrimination of patients (Rose, 1998). As stated already, risk assessment is not ‘new’ to medicine; it is integral to daily medical practice in diagnosis and prognostication.

It is worth noting that to a large extent forensic mental health services are a repository for patients who have been failed by the mental health system, which sometimes does not ‘see’ the risk and so does not manage it, thus adding to the stigma that those with mental illness are to be feared. Nevertheless, it is important that strengths as well as risk factors are elicited. Patients need to be seen as more than the sum of risk factors.

Approaches to assessment

Violence is a complex and multiply determined behaviour. No risk assessment approach is perfect, and clinicians should be aware of the limitations and benefits of the different approaches.

A common argument against violence risk assessment that the authors experience in clinical practice is that it is too rigorous and requires a specialist approach. A proper risk assessment requires little more than a comprehensive psychiatric assessment in most cases. The problem is not that general clinicians are untrained in risk assessment or that it is too rigorous. The problem is that they are often insufficiently focussed on the application of information they have already obtained in a usual comprehensive interview, to the evaluation of violence risk. There is little in a tool such as the HCR-20 that does not form part of such an evaluation. Almost all factors associated with violence risk are embraced in a multidisciplinary approach to the treatment of patients.

Beginning in the 1970s, systematic evaluations appeared of the predictive validity of mental health professionals’ assessments of ‘dangerousness’ (as it was then known). These were ‘first-generation’ approaches. Cocozza and Steadman (1976) found that there were no differences in subsequent recidivism or violent recidivism among 257 forensic patients assessed by psychiatrists and designated as ‘dangerous’ (of whom 49% offended and 14% offended violently) or ‘non-dangerous’ (of whom 54% offended and 16% offended violently). They concluded there was ‘clear and convincing evidence’ (p. 1084) of mental health professionals’ inability to accurately predict violence. Studies such as this have uniformly denounced ‘unstructured clinical judgement’ as an acceptable method of risk assessment (Heilbrun et al., 1999).

Monahan (1981) identified errors with the unstructured clinical approach including lack of specificity in defining the outcome, ignoring statistical base rates of violence, relying upon illusory correlations (i.e. variables misperceived to have a relationship with violence), and failing to incorporate environmental or contextual information into assessments.

The ‘second-generation’ (actuarial) approach to violence risk assessment focussed on the identification of empirically determined variables associated with violence. They provide objective, formal mathematical weighting of risk variables to arrive at a decision on risk level (Grove and Meehl, 1996). Actuarial tools are developed solely on the statistical relationship between a range of predictive variables (‘risk factors’) and the likelihood of violence. Aside from the clinical judgement needed to obtain the information for each predictor, the final assessment of risk is purely mechanical (Quinsey et al., 2005).

Actuarial methods provide a systematic, transparent, objective approach. They are, however, limited by a range of factors. They rely on static/historical factors, provide little information pertaining to dynamic (changeable) current and circumstantial factors and thus neglect factors amenable to treatment. Actuarial methods are insensitive to change over time and have problems with generalisability (Davis and Ogloff, 2008; Mullen and Ogloff, 2009). They do not allow for consideration of important aspects specific to the individual case. They are largely comprised of ‘risk markers’ associated with but not necessarily causally related to violence (Mullen and Ogloff, 2009).

Notwithstanding this, numerous studies show that even simple actuarial methods consistently outperform ‘unstructured clinical judgements’ in a variety of tasks (Dawes et al., 1989), including violence risk assessments (Ægisdóttir et al., 2006; Gardner et al., 1996; Hanson and Morton-Bourgon, 2009; Mossman, 1994; Olver et al., 2011). The amenability of actuarial tools to scrutiny provides grounds for arguments about their limitations.

The ‘third-generation’ structured professional judgement (SPJ) approach relies on a combination of static risk factors, dynamic risk factors (addressing the fluctuating nature of risk) and case-specific factors (individualised and contextual), anchored by the items in the instrument. Rather than providing mathematical probability estimates (as with actuarial approaches), the SPJ approach enables a more in-depth and individualised understanding of the person’s propensity for violence, and informs a more specific and individualised formulation and treatment plan.

The SPJ approach goes beyond simply summing items to generate a total score, as with pure actuarial instruments. ‘Structured’ means that the opinion is not statistical, but is still informed by empirical research, unlike ‘unstructured’ assessment. The SPJ approach allows for the consideration of the personal circumstances of the patient, the nature of the victims, underlying motivations, prior patterns of aggression, context, prior experience of the patient and other case-specific risk factors relevant to the individual (de Vogel et al., 2004). An SPJ tool provides a structured guide to enable clinical decision making and risk management anchored around empirically derived risk factors.

There is evidence that a structured approach to clinical decision making can be as accurate, and in some cases more accurate, than a purely actuarial approach. Guy (2008) found four studies that tested the predictive validity of the HCR-20 completed in two ways. The first approach produced ‘numeric’ scores totalling the 20 risk factors on the HCR-20 (the actuarial approach), and the second, ‘summary’ scores where the HCR-20 was totalled and clinicians were able to modify the overall level of risk using their clinical judgement after considering all of the information available (the SPJ approach), arriving at ratings of high, moderate or low risk. Guy found that in all of the studies multivariate analyses demonstrated that the ‘summary’ ratings (the SPJ approach) added incrementally to the simple numeric use of the instrument (see also Douglas et al., 2003).

Using a structured instrument as a guide avoids many of the clinical biases identified by Monahan (1981). SPJ schemes encourage specification of the criterion, minimise the deleterious effects of making illusory correlations, and in some cases, encourage consideration of contextual features, thereby enabling flexible use.

Managing risk in mental health

There is argument made that, because there is no evidence that violence risk assessment reduces harm, clinicians should probably abandon the practice (Large et al., 2011). In general medical practice, finding that mere observation of a fever does not change the outcome of the infection, and thus coming to the conclusion that clinicians should not take a temperature would be regarded as incorrect. Taking a temperature is risk assessment for a fever, a fever is a risk factor for infection, and an antibiotic is risk management of an adverse outcome (possibly mortality). Violence risk assessment is not an intervention and conducting an assessment of violence risk does not change the outcome. Simply conducting a risk assessment without a plan to manage risk factors would attract severe criticism in most other areas of medicine.

Structured risk assessment leads to risk management strategies that mediate between the risk assessment and aggression (Belfrage et al., 2012). The question therefore is not, ‘Does violence risk assessment reduce harm?’ but rather ‘Does management of risk factors associated with violence (identified by risk assessment) reduce harm?’

There has been limited investigation into how services use risk assessment to influence management of potentially violent individuals. The research that does exist, however, is promising. Abderhalden and colleagues (2008) conducted a cluster randomised controlled trial of the implementation of structured risk assessment in acute psychiatric admission wards. ‘Treatment’ units were those that introduced a standardised risk assessment following admission with a mandatory implementation of prevention in high-risk patients. The ‘control’ units were ‘business as usual’. The results showed that the incidence of aggression decreased substantially in the treatment units, with little change in the control units. Two other studies found that the risk factors from the HCR-20 validly predict risk level and that intense management of moderate- and high-risk patients reduces violent outcome (Dernevik et al., 2002; Pedersen et al., 2012; see also Torrey et al., 2008).

There is evidence that antipsychotic medication reduces aggression in some patients with serious mental illness (Arango and Bernardo, 2005; Chengappa et al., 2002). Clozapine has superior effects on aggression in treatment-resistant patients with schizophrenia (Dalal et al., 1999; Krakowski et al., 2006; Volavka and Citrome, 2008). But mere prescription of antipsychotics and a focus on symptom management is not enough to reduce violence in the longer term. However, increased frequency of clinical contact reduces the occurrence of aggression (Monahan et al., 2001). While it is accepted that assertive community treatment (ACT) is clinically advantageous and cost effective in reducing re-hospitalisation and time in hospital (because it focuses on symptom reduction), neither standard care nor ACT has been shown to be effective in reducing violence in the community over a period of 2 years (Walsh et al., 2001). In the absence of longer-term treatments that focus on psychosocial factors associated with increased risk of violence, the effect of antipsychotics on violence diminishes with time (Bobes et al., 2009; Swanson et al., 2006a, 2008). Antipsychotic medications reduce violence in those without a history of conduct problems but less so in those with a history of conduct disorder (Swanson et al., 2008).

Antipsychotic medication is likely to be more effective in reducing violence in patients where there is a more direct relationship between symptoms and aggression; but as chronic illness exerts its effect on psychosocial functioning over time, psychosocial circumstances likely begin to mediate and increase the risk of violence over and above illness symptoms. Other factors such as personality and conduct disorder, disadvantaged social circumstances, substance abuse, poor social and familial supports, and employment problems (Bonta et al., 1998) make a larger contribution to violence in the longer term than symptoms of serious mental illness. Thus, risk management of propensity for violence requires consideration of a wide variety of risk factors and the use of a range of interventions, not just the prescription of medication.

It is notable that many forensic patients, once discharged from a forensic hospital and followed up by community forensic mental health services, commonly have a more stable lifestyle and less offending in the community than before their acquisition of forensic status and involvement in more comprehensive management (Hodgins et al., 2007; Ong et al., 2009). This is probably attributable to long-term intensive care with a focus on risk factors that include but extend beyond the management of mental illness symptoms alone. Programmes that are multidisciplinary, have small patient to staff ratio, address criminogenic factors, incorporate legal mechanisms to maintain adherence, as well as residential programmes and substance abuse rehabilitation have been shown to reduce violent behaviour (Cusack et al., 2010; Gilbert et al., 2010; Swanson et al., 2001, 2006b).

Conclusions

In the face of the available evidence, violence is often a complication of mental illness that affects a disproportionate number of mentally ill people. The community is unlikely to absolve clinicians of playing a role in managing this public health problem, no matter what the statutes say (Maden, 2007). Reducing morbidity and mortality due to illness is the primary goal of most medical specialties, mental health included. To better manage the propensity for violence by those suffering serious mental illness, clinicians need to understand the problem and become better at identifying that propensity. The primary purpose of risk assessment is to identify risk and protective factors associated with violence in order to manage the risk factors and build on the protective factors. Risk assessment should not be an exercise with the end goal of categorising the person, although this is often an unavoidable outcome, particularly in some legal contexts (where the limits should be explained). Risk assessment is a precursor to treatment and management aimed at preventing harm.

Violence risk assessment is also not a task that is in the exclusive domain of forensic mental health professionals; it is a clinical process with which all mental health professionals should gain familiarity. General mental health services are on the frontline and are regularly confronted with patients who have a propensity for violence; and many carers who cannot manage them are frightened and worry those in their care will harm themselves or others. It is an unavoidable reality of daily general mental health practice that violence risk is an issue that needs to be addressed. To do so, general mental health practitioners would benefit from developing expertise and employing a structured, evidence-based approach to assessment, treatment and management. To avoid utilising a structured approach or reverting to an unstructured approach because ‘risk assessment is an inexact science’ is arguably unacceptable.

Footnotes

Acknowledgements

The authors are grateful to Professor Jay Singh who offered comments on an earlier version of the article, and Professor Dan Howard SC for the useful comments on the final draft.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Declaration of interest

The authors report no conflict of interest. The authors alone are responsible for the content and writing of this paper.

References

Abderhalden

Needham

Dassen

. (2008) Structured risk assessment and violence in acute psychiatric wards: randomised controlled trial. British Journal of Psychiatry 193: 44–50.

Ægisdóttir

White

Spengler

. (2006) The meta-analysis of clinical judgment project: Fifty-six years of accumulated research on clinical versus statistical prediction. The Counseling Psychologist 34: 341–382.

Arango

Bernardo

(2005) The effect of quetiapine on aggression and hostility in patients with schizophrenia. Human Psychopharmacology: Clinical and Experimental 20: 237–241.

Belfrage

Strand

Storey

. (2012) Assessment and management of risk for intimate partner violence by police officers using the Spousal Assault Risk Assessment Guide. Law and Human Behavior 36: 60–67.

Bobes

Fillat

Arango

(2009) Violence among schizophrenia out-patients compliant with medication: prevalence and associated factors. Acta Psychiatrica Scandinavica 119: 218–225.

Bonta

Law

Hanson

(1998) The prediction of criminal and violent recidivism among mentally disordered offenders: a meta-analysis. Psychological Bulletin 123: 123–142.

Buchanan

(2008) Risk of violence by psychiatric patients: Beyond the ‘actuarial versus clinical’ assessment debate. Psychiatric Services 59: 184–190.

Buchanan

Leese

(2006) Quantifying the contributions of three types of information to the prediction of criminal conviction using the receiver operating characteristic. British Journal of Psychiatry 188: 472–478.

Chengappa

Vasile

Levine

. (2002) Clozapine: its impact on aggressive behavior among patients in a state psychiatric hospital. Schizophrenia Research 53: 1–6.

10.

Cocozza

Steadman

(1976) The failure of psychiatric predictions of dangerousness: Clear and convincing evidence. Rutgers Law Review 29: 1048–1101.

11.

Cusack

Morrissey

Cuddeback

. (2010) Criminal justice involvement, behavioral health service use, and costs of forensic assertive community treatment: a randomized trial. Community Mental Health Journal 46: 356–363.

12.

Daffern

Howells

(2002) Psychiatric inpatient aggression: A review of structural and functional assessment approaches. Aggression and Violent Behavior 7: 477–497.

13.

Dalal

Larkin

Leese

. (1999) Clozapine treatment of long-standing schizophrenia and serious violence: a two-year follow-up study of the first 50 patients treated with clozapine in Rampton high security hospital. Criminal Behaviour and Mental Health 9: 168–178.

14.

Davis

Ogloff

JRP

(2008) Key considerations and problems in assessing risk for violence. In: Canter

Zukauskiene

(eds) Psychology and Law: Bridging the Gap. Aldershot: Ashgate.

15.

Dawes

Faust

Meehl

(1989) Clinical versus actuarial judgment. Science 243: 1668–1674.

16.

De Vogel

De Ruiter

Van Beek

. (2004) Predictive validity of the SVR-20 and Static-99 in a Dutch sample of treated sex offenders. Law and Human Behavior 28: 235–251.

17.

Dernevik

Grann

Johansson

(2002) Violent behaviour in forensic psychiatric patients: Risk assessment and different risk-management levels using the HCR-20. Psychology, Crime & Law 8: 93–111.

18.

Dolores

Redding

(2009) The effects of different forms of risk communication on judicial decision making. International Journal of Forensic Mental Health 8: 142–146.

19.

Douglas

Guy

Hart

(2009) Psychosis as a risk factor for violence to others: a meta-analysis. Psychological Bulletin 135: 679–706.

20.

Douglas

Ogloff

Hart

(2003) Evaluation of a model of violence risk assessment among forensic psychiatric patients. Psychiatric Services 54: 1372–1379.

21.

Douglas

Ogloff

Nicholls

. (1999) Assessing risk for violence among psychiatric patients: the HCR-20 violence risk assessment scheme and the Psychopathy Checklist: Screening Version. Journal of Consulting and Clinical Psychology 67: 917–930.

22.

Douglas

Reeves

(2010) Historical-Clinical-Risk Management-20 (HCR-20) Violence Risk Assessment Scheme: Rationale, application, and empirical overview. In: Otto

Douglas

(eds) Handbook of Violence Risk Assessment. New York: Routledge.

23.

Douglas

Skeem

(2005) Violence risk assessment: Getting specific about being dynamic. Psychology, Public Policy, and Law 11: 347–383.

24.

Fazel

Gulati

Linsell

. (2009) Schizophrenia and violence: systematic review and meta-analysis. PLoS Med 6: e1000120.

25.

Gardner

Lidz

Mulvey

. (1996) Clinical versus actuarial predictions of violence in patients with mental illness. Journal of Consulting and Clinical Psychology 64: 602–609.

26.

Gilbert

Moser

Van Dorn

. (2010) Reductions in arrest under assisted outpatient treatment in New York. Psychiatric Services 61: 996–999.

27.

Grove

Meehl

(1996) Comparative efficiency of informal (subjective, impressionistic) and formal (mechanical, algorithmic) prediction procedures: The clinical–statistical controversy. Psychology, Public Policy, and Law 2: 293–323.

28.

Guy

(2008) Performance Indicators of the Structured Professional Judgment Approach for Assessing Risk for Violence to Others: A Meta-analytic Survey. Simon Fraser University, Burnaby BC, Canada.

29.

Hall

(2011) Waterlow found not guilty of murdering father and sister. Sydney Morning Herald. Available at: http://www.smh.com.au/nsw/waterlow-found-not-guilty-of-murdering-father-and-sister-20110419-1dmfd.html (accessed 1 August 2012).

30.

Hanson

Morton-Bourgon

(2009) The accuracy of recidivism risk assessments for sexual offenders: a meta-analysis of 118 prediction studies. Psychological Assessment 21: 1–21.

31.

Heilbrun

Ogloff

Picarello

(1999) Dangerous offender statutes in the United States and Canada. Implications for risk assessment. International Journal of Law and Psychiatry 22: 393–415.

32.

Hodgins

Müller-Isberner

Freese

. (2007) A comparison of general adult and forensic patients with schizophrenia living in the community. International Journal of Forensic Mental Health 6: 63–75.

33.

Kaempf

McSherry

Rothschild

. (2009) Confidentiality for Mental Health Professionals: A Guide to Ethical and Legal Principles. Sydney: Australian Academic Press.

34.

Krakowski

Czobor

Citrome

. (2006) Atypical antipsychotic agents in the treatment of patients with schizophrenia and schizoaffective disorder. Archives of General Psychiatry 63: 622–629.

35.

Large

Ryan

Singh

. (2011) The predictive value of risk categorization in schizophrenia. Harvard Review of Psychiatry 19: 25–33.

36.

Maden

(2007) Treating Violence: A Guide to Risk Management in Mental Health. Oxford: Oxford University Press.

37.

Monahan

(1981) Predicting Violent Behavior: An Assessment of Clinical Techniques. London: Sage Publications.

38.

Monahan

Steadman

Silver

. (2001) Rethinking Risk Assessment: The MacArthur Study of Mental Disorder and Violence. Oxford: Oxford University Press.

39.

Mossman

(1994) Assessing predictions of violence: Being accurate about accuracy. Journal of Consulting and Clinical Psychology 62: 783–792.

40.

Mossman

(2000) Evaluating violence risk ‘by the book’: a review of HCR-20: Assessing Risk for Violence, version 2 and The Manual for the Sexual Violence Risk-20. Behavioral Sciences and the Law 18: 781–789.

41.

Mossman

(2006) Critique of pure risk assessment or, Kant meets Tarasoff. University of Cincinnati Law Review 75: 523–609.

42.

Mullen

(2006) Schizophrenia and violence: from correlations to preventive strategies. Advances in Psychiatric Treatment 12: 239–248.

43.

Mullen

Burgess

Wallace

. (2000) Community care and criminal offending in schizophrenia. Lancet 355: 614–617.

44.

Mullen

Ogloff

JRP

(2009) Assessing and managing the risk of violence towards others. In: Gelder

Andreasen

Lopez-Ibor

. (eds) New Oxford Textbook of Psychiatry (2nd edn). Oxford: Oxford University Press.

45.

Nicholls

Ogloff

Douglas

(2004) Assessing risk for violence among male and female civil psychiatric patients: the HCR-20, PCL:SV, and VSC. Behavioral Sciences & the Law 22: 127–158.

46.

Ogloff

Daffern

(2006) The dynamic appraisal of situational aggression: an instrument to assess risk for imminent aggression in psychiatric inpatients. Behavioral Sciences & the Law 24: 799–813.

47.

Ogloff

JRP

Davis

(2005) Assessing risk for violence in the Australian context. In: Chappell

Wilson

(eds) Issues in Australian Crime and Criminal Justice. Chatswood: Lexis Nexis Butterworths, pp.301–338.

48.

Olver

Stockdale

Wormith

(2011) A meta-analysis of predictors of offender treatment attrition and its relationship to recidivism. Journal of Consulting and Clinical Psychology 79: 6–21.

49.

Ong

Carroll

Reid

. (2009) Community outcomes of mentally disordered homicide offenders in Victoria. Australian and New Zealand Journal of Psychiatry 43: 775–780.

50.

Pedersen

Ramussen

Elsass

(2012) HCR-20 violence risk assessments as a guide for treating and managing violence risk in a forensic psychiatric setting. Psychology, Crime and Law 18: 733–743.

51.

Quinsey

Harris

Rice

. (2005) Violent Offenders: Appraising and Managing Risk (2nd edn). Washington, DC: American Psychological Association.

52.

Rose

(1998) Governing risky individuals: The role of psychiatry in the new regimes of control. Psychiatry, Psychology and Law 5: 177–195.

53.

Ryan

Nielssen

Paton

. (2010) Clinical decisions in psychiatry should not be based on risk assessment. Australasian Psychiatry 18: 398–403.

54.

Steadman

Mulvey

Monahan

. (1998) Violence by people discharged from acute psychiatric inpatient facilities and by others in the same neighborhoods. Archives of General Psychiatry 55: 393–401.

55.

Steadman

Silver

Monahan

. (2000) A classification tree approach to the development of actuarial violence risk assessment tools. Law and Human Behavior 24: 83–100.

56.

Swanson

Borum

Swartz

. (2001) Can involuntary outpatient commitment reduce arrests among persons with severe mental illness? Criminal Justice and Behavior 28: 156–189.

57.

Swanson

Holzer

3rd Ganju

. (1990) Violence and psychiatric disorder in the community: evidence from the Epidemiologic Catchment Area surveys. Hospital & Community Psychiatry 41: 761–770.

58.

Swanson

Swartz

Van Dorn

. (2006a) A national study of violent behavior in persons with schizophrenia. Archives of General Psychiatry 63: 490–499.

59.

Swanson

Swartz

Van Dorn

. (2008) Comparison of antipsychotic medication effects on reducing violence in people with schizophrenia. The British Journal of Psychiatry 193: 37–43.

60.

Swanson

Van Dorn

Monahan

. (2006b) Violence and leveraged community treatment for persons with mental disorders. The American Journal of Psychiatry 163: 1404–1411.

61.

Szmukler

(2001) Violence risk prediction in practice. The British Journal of Psychiatry 178: 84–85.

62.

Tengström

Hodgins

Grann

. (2004) Schizophrenia and criminal offending. Criminal Justice and Behavior 31: 367–391.

63.

Torrey

Stanley

Monahan

. (2008) The MacArthur Violence Risk Assessment Study revisited: two views ten years after its initial publication. Psychiatric Services 59: 147–152.

64.

Volavka

Citrome

(2008) Heterogeneity of violence in schizophrenia and implications for long-term treatment. International Journal of Clinical Practice 62: 1237–1245.

65.

Wallace

Mullen

Burgess

(2004) Criminal offending in schizophrenia over a 25-year period marked by deinstitutionalization and increasing prevalence of comorbid substance use disorders. The American Journal of Psychiatry 161: 716–727.

66.

Walsh

Gilvarry

Samele

. (2001) Reducing violence in severe mental illness: randomised controlled trial of intensive case management compared with standard care. British Medical Journal 323: 1093–1096.

67.

Webster

Douglas

Eaves

. (1997) HCR-20: Assessing Risk for Violence (2nd edn). Vancouver: Mental Health, Law and Policy Institute, Simon Fraser University.

68.

Wootton

Buchanan

Leese

. (2008) Violence in psychosis: estimating the predictive validity of readily accessible clinical information in a community sample. Schizophrenia Research 101: 176–184.

Managing aggression and violence: The clinician’s role in contemporary mental health care

Abstract

Objective:

Methods:

Results:

Conclusion:

Keywords

Introduction

Methods

Results

Understanding the problem

Dealing with uncommon events

Risk assessment in the mental health context

Approaches to assessment

Managing risk in mental health

Conclusions

Footnotes

Acknowledgements

Funding

Declaration of interest

References