Violence risk: Is clinical judgment enough?
Dear Dr. Mossman:
Multiple studies support the reliability and validity of actuarial measures—such as the Historical, Clinical, and Risk Management (HCR-20) risk assessment scheme—to assess violence risk, whereas physicians’ clinical judgment is highly variable. Should clinicians use actuarial measures to assess a patient’s risk of violence? Could it be considered negligent not to use actuarial measures?—Submitted by “Dr. S”
In the 30 years since the Tarasoff decision—which held that psychiatrists have a duty to protect individuals who are being threatened with bodily harm by a patient1—assessing patients’ risk of future violence has become an accepted part of mental health practice.2 Dr. S has asked 2 sophisticated questions about risk assessment. The short answer is that although so-called “actuarial” techniques for assessing risk are valuable, psychiatrists who do not use them are not practicing negligently. To explain why, this article discusses:
- the difference between “clinical” and “actuarial” judgment
- the HCR-20’s strengths and weaknesses
- actuarial measures and negligence.
Do you have a question about possible liability?
- Submit your malpractice-related questions to Dr. Mossman at firstname.lastname@example.org.
- Include your name, address, and practice location. If your question is chosen for publication, your name can be withheld by request.
- All readers who submit questions will be included in quarterly drawings for a $50 gift certificate for Professional Risk Management Services, Inc’s online marketplace of risk management publications and resources (www.prms.com).
Clinical vs actuarial judgment
In the 1970s and 1980s, mental health professionals believed they could not accurately predict violence.3 We now know this is not correct. Since the 1990s, when researchers adopted better methods for gauging the accuracy of risk assessments,4-6 research has shown that mental health clinicians can assess dangerousness with clearly-better-than-chance accuracy, whether the assessment covers just the next few days, several months, or years.4
Over the same period, psychologists recognized that when it comes to making predictions, clinical judgment—making predictions by putting together information in one’s head—often is inferior to using simple formulae derived from empirically demonstrated relationships between data and outcome.7 This approach—“actuarial” judgment—is how insurance companies use data to calculate risk.
By the late 1990s, psychologists had developed actuarial risk assessment instruments (ARAIs)8 that could accurately rank the likelihood of various forms of violence. Table 1 lists some well-known ARAIs and the populations for which they were designed. In clinical practice, psychiatrists usually focus on risk posed by psychiatric patients. The HCR-209 was designed to help evaluate this type of risk.
Examples of actuarial risk assessment instruments (ARAIs)
Violence in psychiatric populations, such as formerly hospitalized patients
Classification Of Violence Risk (COVR)
Violence by civil psychiatric patients following discharge into the community
Violence Risk Assessment Guide (VRAG)
Violent recidivism by formerly incarcerated offenders
Recidivism by sex offenders
HCR-20’s pros and cons
The HCR-20 has 20 items:
- 10 concerning the patient’s history
- 5 related to clinical factors
- 5 that deal with risk management (Table 2).
To evaluate a patient’s risk of violence, you score each item 0, 1, or 2, depending on how closely the patient matches the described characteristic. For example, when scoring item C3 (active symptoms of major mental illness), a patient gets 0 for “no active symptoms,” 1 for “possible/less serious active symptoms,” or 2 for “definite/serious active symptoms.” An individual can receive a total HCR-20 score of 0 to 40. The higher the score, the higher the likelihood of violence in the coming months.
To use the HCR-20 as an exercise of true actuarial judgment, you would base your opinion of a patient’s risk of violence solely on the HCR-20 score, without regard for other patient factors. However, the HCR-20’s developers think this approach “may be unreasonable, unethical, and illegal.”9 One reason is that the HCR-20 omits obvious signs of potential violence, such as a clearly stated threat with unambiguous intent to act.
The HCR-20’s designers hope clinicians will use this instrument to “structure” clinical judgments about dangerousness. The HCR-20 reminds clinicians to identify and evaluate known risk factors for violence. Clinicians can then address those factors to better manage their patients.
For example, if a patient is doing well in the hospital (and has a low score on HCR-20 clinical items), a psychiatrist might assume the patient will cause few problems after discharge. But if the risk management items generate a high score, the psychiatrist should realize that these factors raise the patient’s violence risk and may require additional intervention—perhaps a different type of community placement or special effort to help the patient follow up with out-patient treatment.
Items from the Historical, Clinical, and Risk Management (HCR-20)
Risk management items
H1 Previous violence
C1 Lack of insight
R1 Plans lack feasibility
H2 Young age at first incident
C2 Negative attitudes
R2 Exposure to destabilizers
H3 Relationship instability
C3 Active symptoms of major mental illness
R3 Lack of personal support
H4 Employment problems
R4 Noncompliance with remediation attempts
H5 Substance use problems
C5 Unresponsive to treatment
H6 Major mental illness
H8 Early maladjustment
H9 Personality disorder
H10 Prior supervision failure
Score each item 0, 1, or 2, depending on how closely the patient matches the described characteristic. For example, when scoring item C3 (active symptoms of major mental illness), a patient gets 0 for “no active symptoms,” 1 for “possible/less serious active symptoms,” or 2 for “definite/serious active symptoms.” An individual can receive a total HCR-20 score of 0 to 40. The higher the score, the higher likelihood of violence in the coming months.
Source: Reprinted with permission from Webster CD, Douglas KS, Eaves D, Hart SD. HCR-20: assessing risk for violence, version 2. Burnaby, British Columbia, Canada: Simon Fraser University, Mental Health, Law, and Policy Institute; 1997
Is not using ARAIs negligent?
Some writers believe that using ARAIs should12 or may soon13 become the standard of care. Why, then, do psychiatrists seldom use ARAIs in their clinical work? Partly it is because clinicians rarely receive adequate training in assessing violence risk or the science supporting it. After a 5-hour training module featuring the HCR-20, psychiatry residents could better identify factors that affect violence risk, organize their reasoning, and come up with risk management strategies.2
Psychiatrists may have other reasons for not using ARAIs that make clinical sense. Although ARAIs can rank individuals’ violence risk, the probabilities of violence associated with each rank aren’t substantial enough to justify differences in management.14 Scientifically, it’s interesting to know that we can separate patients into groups with “low” (9%) and “high” (49%) risks of violence.15 But would you want to manage these patients differently? Most psychiatrists probably would not feel comfortable ignoring a 9% risk of violence.
Also, ARAIs typically focus on factors that influence violence risk over weeks, months, or years. But as Simon16 notes, clinicians often are asked to address “imminent” violence. No agreed-upon definition of imminence exists, but even if the meaning were clear, ARAIs “are insensitive to patients’ clinical changes that guide treatment interventions or gauge the impact of treatment.”16
To avoid negligence, psychiatrists need only “exercise the skill, knowledge, and care normally possessed and exercised by other members of their profession.”17 Psychiatrists seldom use ARAIs,12 so failing to use them cannot constitute malpractice. As Simon points out, a practicing psychiatrist’s role is to treat patients, not predict violence. He concludes, “at this time, the standard of care does not require the average or reasonable psychiatrist to use actuarial assessment instruments in the evaluation and treatment of potentially violent patients.”16
1. Tarasoff vs Regents of the University of California, 551 P. 2d 334 (Cal. 1976).
2. McNiel DE, Chamberlain JR, Weaver CM, et al. Impact of clinical training on violence risk assessment. Am J Psychiatry 2008;165:195-200.
3. Monahan J. The clinical prediction of violent behavior. Washington, DC: National Institute of Mental Health; 1981.
4. Mossman D. Assessing predictions of violence: being accurate about accuracy. J Consult Clin Psychol 1994;62:783-92.
5. Rice ME, Harris GT. Violent recidivism: assessing predictive validity. J Consult Clin Psychol 1995;63:737-48.
6. Gardner W, Lidz CW, Mulvey EP, Shaw EC. Clinical versus actuarial predictions of violence in patients with mental illness. J Consult Clin Psychol 1996;64:602-9.
7. Dawes RM, Faust D, Meehl PE. Clinical versus actuarial judgment. Science 1989;243:1668-74.
8. Hart SD, Michie C, Cooke DJ. Precision of actuarial risk assessment instruments: evaluating the ‘margins of error’ of group v. individual predictions of violence. Brit J Psychiatry 2007;190:60-5.
9. Webster CD, Douglas KS, Eaves D, Hart SD. HCR-20: assessing risk for violence, version 2. Burnaby, British Columbia: Simon Fraser University, Mental Health, Law, and Policy Institute; 1997.
10. Quinsey VL, Harris GT, Rice ME, Cormier CA. Violent offenders: appraising and managing risk. 2nd ed. Washington, DC: American Psychological Association; 2006.
11. Hanson RK, Morton-Bourgon KE. The accuracy of recidivism risk assessments for sexual offenders: a meta-analysis. Ottawa, Canada: Public Safety Canada; 2007. Available at: http://www.publicsafety.gc.ca/res/cor/rep/_fl/crp2007-01-en.pdf. Accessed April 21, 2008.
12. Swanson JW. Preventing the unpredicted: managing violence risk in mental health care. Psychiatr Serv 2008;59:191-3.
13. Lamberg L. New tools aid violence risk assessment. JAMA 2007;298(5):499-501.
14. Mossman D. Commentary: assessing the risk of violence—are “accurate” predictions useful? J Am Acad Psychiatry Law 2000;28:272-81.
15. Monahan J, Steadman HJ, Robbins PC, et al. An actuarial model of violence risk assessment for persons with mental disorders. Psychiatr Serv 2005;56:810-15.
16. Simon RI. The myth of “imminent” violence in psychiatry and the law. Univ Cincinnati L Rev 2006;75:631-43.
17. Dobbs DB. The law of torts. St. Paul, MN: West Group; 2000:269.