 |
 |
|
|
|
Risk assessment is a method of calculating the relative value and likelihood of different possible outcomes (1). In this article, we review the need to perform risk assessments in current clinical practice. We then discuss the relative benefits and disadvantages of some risk assessment tools with which clinicians should be familiar. Using a case scenario and analysis to illustrate, we describe how risk assessment in practice becomes an exercise in risk management (2). Finally, we discuss a model to assist clinicians to act after assessing risk, so as to safeguard the community while balancing patient rights. Psychiatrists should be familiar with several key factors relating to risk. Norko and Baranoski review the extensive literature on the relation between violence and mental health issues and conclude that substance abuse, both alone and in combination with other mental health factors, is a risk factor for violence (3). In addition, as well as mental illness, certain other sociodemographic environmental factors have been found to contribute to violence. The likelihood of causing harm to others is a primary criterion for certification in all Canadian jurisdictions (4). Although the statistical relation between mental disorder and future violence is highly complex, the assessment of risk is a daily part of many facets of psychiatry, because clinicians deal with potentially violent patients every day. Risk assessment is necessary in emergency psychiatry, in inpatient psychiatry, and in the treatment of mentally abnormal offenders. Harrison describes the climate of expectation that exists in regard to standards of care, together with the obligations of mental health professionals to protect both their patients and the public from untoward incidents (5): all mental health workers working in emergency rooms must be trained and familiar with risk assessment and risk management. Discharge planning from an inpatient unit requires similar skills and procedures. Outpatient and community practitioners must also be familiar with these skills, since it is not uncommon for concerns about violence to arise. In addition to attending to their duty to patients and the public, the potential for litigation in the event of a negative outcome should increase psychiatrists’ mindfulness of risk issues. In the UK, the National Health Service has suggested that best practice should include a documented risk assessment before any patient is discharged to the community (6). Subsequent effective risk management strategies should be delineated in the management plan agreed to by all parties. Harrison notes that this duty goes beyond gaining the patient’s consent, implying that all reasonable steps must be taken to contact the patient and intervene to manage risk, regardless of the patient’s wishes (5). Reed (6) and Harris (7) emphasize the need for training of mental health workers and for mandated recording of the risk management strategy in each case. Advantages and Disadvantages of Risk Assessment MethodsWe have come a long way since Monahan’s 1981 statement that the best clinical research indicated psychiatrists and psychologists accurately predicted violent behaviour in not more than one-third of patients (8). In 1997 he was a little less pessimistic, stating that predictions of violence appear to have a modest, better-than-chance level of accuracy (9). A great deal of debate has focused on the use of clinical assessment schema to improve accuracy. Typically, these instruments are constructed with a retrospective research design attempting to look for factors associated with recidivism in offender populations. These factors are then noted and weighted to produce a score that is calculated to give a probability of future violence. The tool is then scored, and its accuracy is tested against the original cohort (10). The first instruments developed tend to rely exclusively on historical factors, which are thought to have the advantage of being static and unchangeable. These instruments are generally transparent and give an impression of reliability and objectivity. Conversely, clinical judgment relies on dynamic or changing factors, often recording an individual’s temporary states. Clinical factors are therefore criticized as being opaque, random, and subjective. Another issue that bears examination is whether these risk assessment instruments more accurately predict short-term violence or long-term violence. It seems to make intuitive sense that they would better predict short-term dangerousness than long-term dangerousness. However, by reexamining clinicians’ predictions, Mossman demonstrated that they are as accurate in predicting long-term as short- and medium-term clinical violence (11). It has been claimed that actuarial tools are so superior to clinical judgment that they should be used in isolation and that adjunctive clinical assessments may in fact be detrimental (12). However, this debate is rendered moot in that most tools include many items that rely on such clinical skills as interviewing and diagnosis. For instance, the Violence Risk Appraisal Guide (VRAG, 13) and the Historical/Clinical/Risk Management 20-item (HCR-20, 14) rely, among other items, on the Psychopathy Checklist-Revised (PCL-R), which requires a clinical interview and a diagnosis of a clinical condition (15). The actuarial tools rely on clinical assessment to obtain data, and modern clinical assessments include both clinical and actuarial information. The HCR-20 takes it name from 3 scales—Historical, Clinical, and Risk Management—and from its 20 items. It is a structured guide through various risk factors for violent behaviour and is designed mainly for individuals with personality disorders and (or) major mental illnesses. It includes 10 items that consider historical factors, 5 that consider current clinical variables, and 5 that stress future risk management issues. Some HCR-20 items are familiar to clinicians; they address insight, negative attitudes, impulsivity, response to treatment, and active symptoms of a major mental illness. Risk management items address feasibility of plans, exposure to destabilizers, social supports, treatment compliance, and stress. This instrument has been validated by independent researchers on different populations (16). Another new scheme for assessing violence risk, the Iterative Classification Tree (ICT, 17), used a development sample of civil (that is, nonforensic) psychiatric patients. The authors specifically attempted to increase the clinical utility of this tool by adopting risk factors commonly available in hospital records or routinely assessed in clinical practice. Subjects can be accurately delineated as low risk or high risk, although the authors acknowledge that some cases may fall between these thresholds. Several other tests (15) and actuarial tools (10) are available. The PCL-R requires a 2-hour to 3-hour interview and chart review. It comes with a manual containing supportive psychometric data, and the assessor is encouraged to attend a 2-day training course. Although designed as a descriptive tool, it has been found to have good predictive properties in various populations (15). However, it has limited use in practical clinical situations, and it is best reserved for forensic practice. The Sex Offender Risk Assessment Guide and VRAG (12,13) used development samples based on a limited and primarily forensic population. Both rely heavily on the PCL-R as the strongest predictive factor, and both purport to predict long-term recidivism. Taking these factors into consideration, these instruments, too, have little daily clinical use. All instruments must have an appropriate validation population (18), that is, a validation group similar to the patient group. Further, instruments should be established and well-validated. Reid argues that instruments should be used only as part of a risk assessment to help clinicians formulate their thoughts and ensure a comprehensive assessment (19)—analogous to Hanson’s notion of a guided clinical assessment (20), wherein he warns against overreliance on actuarial instruments and checklists. Any assessment should use as much information as possible and include clinical and static factors, as well as collateral information. Reed emphasizes that, in practice, risk assessment remains a clinical skill, albeit a difficult skill of uncertain validity (6). In practical terms, he outlines 4 elements of a comprehensive risk assessment: 1. Ensure that relevant information about past and present is available. 2. Define future circumstances likely to increase risk and what changes might reduce this risk. 3. Seek expert help when necessary. 4. Record the risk management strategy and decide when it needs to be reviewed. Reed also describes the use of supervision registers, such as those developed and introduced in the UK to identify potentially violent or vulnerable patients, to target services and prioritize high-risk patients in the community. Acting on Risk Assessment: The Duty to ProtectThe relative likelihood of violence or harm to a third party has become a key issue in psychiatric practice. The fiduciary duty of psychiatrists to their patients may have to give way to a duty to protect and thus breach confidentiality when third parties are at risk (21). The duty to protect applies not only to patients: it extends through patients to third parties (22). Indeed, the boundaries of these duties have been extended to settings outside our traditional domain—the hospital—into the community and also to various third parties who may include unspecified persons or groups of persons (23). Tarasoff I (24) was the seminal case articulating the duty to warn; it led to Tarasoff II (25), which differentiated duty to warn and duty to protect. This concept has evolved in Canadian law and was clearly enunciated by the Supreme Court decision in Smith v Jones (26). In this case, Mr Jones was charged with the aggravated sexual assault of a prostitute. When assessed by a psychiatrist retained for purposes of preparing a defence, Mr Jones admitted to fantasies of raping and killing prostitutes. Dr Smith commenced an action in civil court to breach confidentiality and entitle him to declare this information. The case was expedited to the Supreme Court of Canada, which ruled that protecting the public can overrule even solicitor–client privilege, the highest privilege in the land. The Court made clear that, if solicitor–client privilege has a public safety exception, this applies to all types of privilege, presumably including doctor–patient confidentiality. The Court declined to define the exact steps to be taken to prevent harm but did suggest that they might include notifying a potential victim, the police, or a crown attorney, depending on the circumstances. Bearing this in mind, the Canadian Psychiatric Association’s Standing Committee on Professional Standards and Practices developed a position paper (27) that includes the following recommendations reflecting the Supreme Court of Canada decision in Smith v Jones: 1. As part of the informed consent process, patients need to be warned of limits to confidentiality. 2. A duty to protect (warn, or inform) exists
Chaimowitz and Glancy have described a process to undertake when any patient is discharged into the community (22). This process applies not only to forensic practice but also to discharge from inpatient care, day-patient care, and the emergency department. The authors have suggested an algorithm, evolved from others (28,29), to guide clinicians dealing with threats. Following a threat, a clinical interview should include a risk assessment. If the patient is considered dangerous, a period of negotiation should take place. The patient may be persuaded to enter into hospital or take certain steps that resolve the situation. If the patient agrees to be admitted to hospital or the clinician’s assessment deems the patient certifiable under the relevant mental health act, hospitalization is effected. This hospitalization works to protect an intended victim while at the same time maintaining relative confidentiality. At the time of decertification or discharge, the issue of the duty to warn and protect needs to be revisited. Alternatively, if negotiation produces no resolution and an identifiable victim or group of victims can be ascertained, the standard of care probably includes warning the victim or group of victims and (or) contacting the police. Case Scenario: Applying Theory to PracticeJohn, aged 19 years, entered his first year of university and wanted to be a physicist. He began to sleep less and became obsessed, day and night, with writing his “thesis.” He was irritable and dogmatic. His girlfriend felt that she could no longer cope with him and terminated the relationship. John was angry and assaulted her. He continued to contact her against her wishes, claiming he had “Godlike powers” and “could possess her and take away life.” His parents took him to the emergency department for assessment, and he was certified on an order for psychiatric assessment. He had no history of violence and no previous criminal record, nor was there a history of personal or family substance abuse. He admitted to experiencing violent fantasies about his girlfriend while with her. He was diagnosed with schizoaffective disorder. Case Analysis John was detained under the provincial mental health act. He was started on treatment with lithium carbonate and risperidone. It took some time to titrate the dosages of these medications, but he began gradually to respond to treatment over a period of approximately 4 weeks. His psychiatrist put a warning on his chart that, if he discharged himself against medical advice, he should be reassessed for certifiability. If the assessing doctor did not consider that he was certifiable under the relevant mental health act, his psychiatrist recommended that his previous girlfriend and the police should be notified. However, owing to pressure from his parents, and with their support, he agreed to stay. After some time, he seemed to respond well to treatment. His mood returned to normal, and his grandiose delusions subsided. He was still somewhat preoccupied with his “theory” but had turned his mind to practical matters, such as continuing his university education. A Risk Management Plan Arranging for his discharge, a multidisciplinary team met with John and his parents to devise a risk management plan. Prior to this meeting, his psychiatrist performed a risk assessment using a guided clinical approach; that is, he undertook a clinical assessment to elicit the pertinent factors, in conjunction with the ICT (17) and HCR-20 (14). The ICT suggested that John was at high risk owing to the presence of mania, recent violent fantasies, being an involuntary patient, and recent serious violence. The lack of personal or family substance abuse, together with his good family background, obviated the risk somewhat. Conversely, the HCR-20 found him to be at low risk for violence at the time of discharge, although it allows for increasing the risk category, were he still experiencing grandiose delusions and violent fantasies. Under the risk management plan, John was to live with his parents and work temporarily with his father until he could register again for university in the next term. He was to be followed weekly by his psychiatrist. He would attend weekly for serum lithium treatment for the first 4 weeks and monthly thereafter. A social worker would visit the family weekly, assess John, and speak to the family, including his parents. It was noted that, if his serum lithium dropped, presumably because of noncompliance, he should be brought to the emergency department for assessment. He was not to contact his previous girlfriend, and his parents confiscated his cell phone. Risk Management and ReductionIn certain legal contexts, such as assessment for dangerous offender status in Canada or a sexually violent predator in the US (10), a court requires an expert opinion on whether the offender is likely to reoffend. The legal context is therefore concerned with behaviour stretching back for an indeterminate period and leading to conclusions about violence in the distant future. In clinical practice, however, risk assessment is more concerned with what, when, where, how, and why. Legal decision makers tend to overvalue the applicability of risk assessment instruments, whereas in clinical practice, the important issue is managing risk (2). In clinical practice, we need to know whether the patient is treated, what is the discharge plan, what is the system for early detection of problems, and whether this is enough. In forensically managing mentally unhealthy offenders, a graduated approach to treatment is generally taken. If, however, the likelihood of harm and the imminence and severity of harm are all high, an extremely conservative approach would be taken. Dvoskin therefore emphasizes that the essence of risk assessment is in fact a clinical duty of risk reduction (2). In other words, risk assessment identifies the treatable factors, thereby addressing and reducing risk. Risk reduction generally follows evidence-based practice and includes such elements as antipsychotic treatment, cognitive-behavioural treatment for sex offenders, anger management programs, and outpatient commitment. In clinical practice, risk reduction is generally based on the same principles. For outcome measures, we use measures of compliance or participation in treatment. This has been referred to as the demonstration model (2). Thorough assessment delineates problems and deficits that alone or in combination may contribute to the patient’s potential for violence. Treatment programs then focus on addressing these issues. If the program is successful, the patient is allowed to confront new responsibilities under gradually reduced supervision. This process continues until the patient is guided into the community, having demonstrated increased competence with increasingly more liberty. Most programs have developed risk management approaches that use predictive and actuarial tools. The risk management process is dynamic and iterative; it changes over time, depending on such factors as response to treatment, treatment compliance, and manipulation of environmental factors. It should be noted that actuarial tools in current use are not intended to measure change, although some authors have advised research in this area (2). Some of these tools, including the HCR-20 (14), the Level of Service Inventory-Revised (30), and the Sex Offender Needs Assessment Rating (31), do help to identify deficits for intervention planning. ConclusionsAlthough a debate continues about the utility of actuarial tools, compared with clinical prediction, in clinical practice the debate is superceded by the goals of risk management and risk reduction. We feel that this is a standard of care issue—that risk assessment should be incorporated into every clinical situation and that we should take into account any duty to warn and protect a third party. Risk assessment involves taking as full a history as possible, including all relevant information and some assessment of the probability of future violence. Checklists or actuarial tools may help clinicians if they are available and not too cumbersome. However, it is also important to evaluate the imminence and severity of the violence. Clinicians should consider future circumstances that are likely to present an increased risk and address what changes can be made to decrease those risks. Good documentation is essential, and consultation should be sought when necessary. Risk management plans need to be documented and reviewed regularly. Inherent in any risk management strategy is the definition and measurement of risk factors and deficits. To address these deficits, treatment programs based on best practice can then be followed. Participation and change should be measured. When return to the community is indicated, incremental community access and gradually increasing demands can help to demonstrate skill building and reduce the aforesaid deficits. As Reid points out, the challenge lies in whether a doctor reasonably recognizes and adequately deals with a particular level of danger (19). AcknowledgementWe are grateful to Professor Chris Webster for his helpful comments on an earlier draft of this paper. References1. Carson D. 1995 quoted in Bingley W. Assessing dangerousness: protecting the interests of patients. Br J Psychiatry 1997;170(Suppl 32):28–9. 2. Dvoskin JA, Heilbrun K. Risk assessment and release decision-making: toward resolving the great debate. J Am Acad Psychiatry Law 2001;29:6–10. 3. Norko M, Baranoski MV. The state of contemporary risk assessment research. Can J Psychiatry 2005;50:18–26. 4. Gray JC, O’Reilly R. Clinically significant differences among Canadian Mental Health Acts. Can J Psychiatry 2001;41:315–21. 5. Harrison G. Risk assessment in a climate of litigation. Br J Psychiatry 1997;170:37–9. 6. Reed J. Risk assessment and clinical risk management: the lessons from recent enquiries. Br J Psychiatry 1997;170(Suppl 32):4–7. 7. Harris M. Training trainers in risk assessment. Br J Psychiatry 1997;170(Suppl 32):35–46. 8. Monahan J. The clinical prediction of violent behaviour. Rockville (MD): National Institute of Mental Health; 1981. 9. Monahan J. Clinical and actual prediction of violence. In: Faigman D, Kay D, Sacks M, Sanders J, editors. Modern scientific evidence: the law and science of expert testimony. Volume 1. St Paul (MN): West Publishing; 1997. p 300–18. 10. Glancy G, Regehr C. The predicative validity of measurement tools for assessing dangerous in sex offenders. In: Roberts A, Yeager K, editors. Handbook of psychiatric based research. New York: Oxford University Press; 2002. 11. Mossman D. Commentary: assessing the risk of violence—are “accurate” predictions useful? J Am Acad Psychiatry Law 2000;28:272–81. 12. Quinsey V, Khanna A, Malcolm P. A retrospective evaluation of the regional treatment sex offender treatment program. Journal of Interpersonal Violence 1998;13:621–4. 13. Quinsey V, Harris G, Rice M, Cormier C. Violent offenders: appraising and managing risk. Washington (DC): American Psychological Association; 1998. 14. Webster C, Douglas KS, Eaves E, Hart SD. The HCR-20: assessing risk for violence. Version 2. Burnaby (BC): Simon Fraser University; 1997. 15. Hare R. Hare Psychopathy Checklist-Revised (PCL-R). Technical manual. 2nd ed. Toronto (ON): Multihealth Systems; 2003. 16. Douglas KS, Ogloff RP, Nicholls TL, Grant I. Assessing risk for violence among psychiatric patients: The HCR-20 risk assessment scheme and the psychopathy checklist: screening version. J Consult Clin Psychol 1991;61:917–30. 17. Monahan J, Steadman HJ, Appelbaum PS, Robbins PC, Mulvey EP, Silver E, and others. Developing a clinically useful actuarial tool for assessing violence risk. Br J Psychiatry 2000;176:312–9. 18. Rogers R. The uncritical acceptance of risk assessment in forensic practice. Law and Human Behaviour 2000;24:595–605. 19. Reid WH. Risk assessment prevention and foreseeability. Journal of Psychiatric Practice 2003;9:984–6. 20. Hanson RK. What do we know about sex offender assessment? Psychol Public Policy Law 1993;4:50–72. 21. Glancy GD, Regehr C, Bryant AG. Confidentiality in crisis: part I—the duty to inform. Can J Psychiatry 1998;43:1001–5. 22. Chaimowitz GA, Glancy GD, Blackburn J. The duty to warn and protect—impact on practice. Can J Psychiatry 2000;45:899–904. 23. Wenden v Trikha, Royal Alexandra Hospital and Yaltho (1993), 14 CCLT (2d) 225 (Alta. CA). 24. Tarasoff v Regents of the University of California, 188 Cal Rptr 129, 529 P2d 533, 1974. 25. Tarasoff v Regents of the University of California (1976), 17 Cal (3d) 425, 551 P2d 334. 26. O’Shaughnessy R, Glancy G, Bradford J. Smith v Jones: Supreme Court of Canada: confidentiality and privilege suffer another blow. J Am Acad Psychiatry Law 1999;27:614–20. 27. Chaimowitz G, Glancy G. The duty to protect. Ottawa: Canadian Psychiatric Association; 2002. Position Paper nr 2002-42. Available: www.cpa-apc.org/Publications/Position_Papers/duty.asp 28. Mills MJ. The so-called duty to warn: the psychotherapy duty to protect third parties from patients: violent acts. Behav Sci Law 1984;2:237–57. 29. Felthous AR. The clinician’s duty to protect third parties. Psychiatr Clin North Am 1999;22:49–59. 30. Andrews D, Bonta J. Level of Service Inventory-Revised, Toronto (ON): Multihealth Systems; 1995. 31. Hanson KR and Harris A. A structured approach to evaluating change among sexual offenders. Sexual Abuse: A Journal of Research and Treatment 2001;13:105–22. Author(s)Manuscript received and accepted November 2004. 1. Assistant Professor, Department of Psychiatry, University of Toronto, Toronto, Ontario; Clinical Assistant Professor, Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario. 2. Assistant Professor, Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario. Address for correspondence: Dr GD Glancy, 302 The East Mall, Suite 400, Etobicoke, ON M9B 6C7 e-mail: graham.glancy@utoronto.ca
1 | 2
|
||||||
