1. Assessing the Predictive Validity of the HCR-20V3 in
Gauging Civil Psychiatric Patients’ Short-term
Violence Risk
Meghan Banks, B.S.
Fordham University 2016
2. Background
Assessment and prevention of violence are critical in psychiatric care,
management, and treatment.
Clinicians have to make decisions about their patients’ violence risk,
especially when presented with behavioral emergencies that may
require hospitalization.
Clinicians’ unaided judgment in assessing violence risk results in a
greater likelihood of false positives.
Four commonly used violence screening measures:
1. Violence-Screening Checklist (VSC)
2. Brøset Violence Checklist (BVC)
3. Dynamic Appraisal of Situational Aggression (DASA)
4. Historical-Clinical Risk Management Version 3 (HCR-20V3)
3. Present Study
The present study assessed the predictive validity of the HCR-20V3
Summary Risk Ratings (SRRs) in gauging civil psychiatric patients’
violence risk:
1. Addressed whether the HCR-20V3 Case Prioritization, Severe Physical
Harm, and Imminent Violence risk ratings would each be associated with
aggression frequency and severity among civil psychiatric patients during a
3 month follow-up period.
2. Addressed whether Case Prioritization ratings would predict aggression
occurrence, frequency, and severity.
3. Addressed whether Severe Physical Harm risk ratings would predict
aggression severity.
4. Addressed whether Imminent Violence risk ratings would predict aggression
frequency and severity.
4. Methods
63 civil psychiatric patients admitted to an urban public hospital between
February and December of 2013.
Aggression: 1 = present and 0 = not present.
Aggression frequency: number of aggressive acts committed.
Aggression Severity: 1 = minimal (e.g., verbal), 2 = moderate (threat with
weapon), and 3 = severe (life threatening).
HCR-20V3 risk ratings were completed within 2 to 3 weeks after hospital
admission based on medical record information and brief interviews with
patients’ treatment teams.
1 = low risk, 2 = moderate risk, 3 = high risk.
Data regarding aggressive incidents was extracted from the hospital’s
database.
5. Results
Table 1
Cross-Tabulation of Aggression Occurrence by Case Prioritization
Case Prioritization Yes No X2
Low 6 4 1.14
Moderate 20 10
High 12 11
Total 38 25
• There was a weak, positive, and non-significant correlation between
aggression and Case Prioritization rating, rs (61) = .10 p = .46.
• Aggression occurrence did not significantly differ by low, moderate, and
high risk patients on Case Prioritization, X2 (2, N = 63) = 1.14, p = .56.
6. Results
There were no significant differences in aggression frequency and severity
between low, moderate, and high risk patients on Case Prioritization, F (2,
62) = 0.57, p = .57 and F (2, 62) = 0.71, p = .49, respectively.
There was not a significant difference in aggression severity between low,
moderate, and high risk patients with regard to engaging in severe physical
harm, F (2, 62) = 1.18, p = .32.
There was not a significant difference in aggression frequency between
low, moderate, and high risk patients in engaging in imminent violence, F
(2, 62) = 0.47, p = .63.
There was a significant difference in aggression severity between low,
moderate, and high risk patients in engaging in imminent violence, F (2, 62)
= 3.47, p = .03.
Positive significant association between aggression severity and risk of
engaging in imminent violence, rs = .25, p .03.
7. Results
Case Prioritization and
Severe Physical Harm
ratings had weak and
non-significant predictive
validity.
Imminent Violence risk
ratings had moderate
predictive validity that
approached
significance, AUC = .63,
p = .08.
8. Results
Table 2
Aggression Frequency and Severity by Case Prioritization
Case Prioritization Frequency Severity
Low 10 (M = 0.40, SD =
0.52)
10 (M = 0.90, SD =
1.20)
Moderate 30 (M = 1.27, SD =
1.05)
30 (M = 0.63, SD =
0.93)
High 23 (M = 1.00, SD =
1.41)
23 (M = 0.96, SD =
1.07)
9. Results
Table 2
Means and Standard Deviations of Aggression Severity by Severe
Physical Harm
Severe Physical Harm Severity
Low 33 (M = 0.67, SD = 0.99)
Moderate 25 (M = 0.84, SD = 1.07)
High 5 (M = 1.40, SD = 0.89)
11. Discussion
The HCR-20V3 demonstrated limited predictive validity in
gauging civil psychiatric patients’ violence risk.
Although Imminent Violence SRR demonstrated moderate
predictive validity with regard to severity of violence, it was
still weakly associated with aggression.
Although more than half of the sample committed at least
one aggressive incident, severe aggression was not
common.
12. Limitations
Study’s definition of “violence”.
Did not compare the predictive validity of other validated
violence screening measures.
Sample size and limited statistical power.
Unable to look at potentially important variables (i.e.,
diagnostic category).
Lack of variability in diagnosis.
Allocation of primary and aggressive interventions to high
risk patients.
13. References
Howe, J., Rosenfeld, B., Foellmi, M., Stern, S., & Rotter, M.
(2015). Application of the HCR-20 version 3 in civil psychiatric
patients. Criminal Justice and Behavior, 43(3), 398-412.
Editor's Notes
Talk about why there is a need
No explicit standard of violence screning measures
Given that the HCR-20V3 has been used in community settings and hospitals prior to discharge, the current study….
Severe Physical Harm= if a patient engages in aggression that leads to severe physical harm, does he/she have a high level of aggression severity?
Imminent Violence= if a patient engages in imminent violence, will they commit they greatest number of aggressive acts and have the highest level of aggression severity?
Consisted of 47 male (73%) and 17 female (27%) patients
Age ranged from 18 to 70 years old (M = 37.97, SD = 13.15).
All diagnosed with schizophrenia or schizoaffective disorder (86%, n = 55) or mood disorder (14%, n = 9).
Identified as Black (69%), White (14%), Asian (2%), mixed race/ethnicity (5%), or did not have race/ethnicity coded into charts (11%).
1. Only a handful of moderately severe aggressive acts, none of which were life threatening.