Dynamic risk assessment HCR-20 & SAPROF Michiel de Vries Robbé, Vivienne de Vogel & Ellen van den Broek Van der Hoeven Kliniek, The Netherlands 12 th Annual Conference IAFMHS, Miami 2012
Jun 23, 2015
Dynamic risk assessmentHCR-20 & SAPROF
Michiel de Vries Robbé, Vivienne de Vogel & Ellen van den Broek
Van der Hoeven Kliniek, The Netherlands
12th Annual Conference IAFMHS, Miami 2012
Presentation outline
• Dynamic risk assessment HCR-20 & SAPROF• File study results• Clinical results• The individual case
Van der Hoeven KliniekUtrecht, The Netherlands
• Forensic psychiatric hospital: 286 beds– Mostly TBS order: involuntary treatment– 50/50 personality / psychotic disorders– Holistic approach, emphasis on CBT & relapse prevention– Therapeutic community: taking responsibility– Rehabilitation: ‘transmural phase’
• Risk assessment in consensus– HCR-20 & SAPROF (+SVR-20, FAM)– Repeated regularly to inform treatment
Risk & Protection
Risk factors
Protective factors
HCR-20 & SAPROF
Historical factorsH1 Previous violenceH2 Young age at first violenceH3 Relationship instabilityH4 Employment problemsH5 Substance use problemsH6 Major mental illnessH7 Psychopathy (PCL-R)H8 Early maladjustmentH9 Personality disorderH10 Prior supervision failure
Clinical factorsC1 Lack of insightC2 Negative attitudesC3 Active symptoms of major mental illnessC4 ImpulsivityC5 Unresponsive to treatment
Risk Management factorsR1 Plans lacks feasibilityR2 Exposure to destabilizersR3 Lack of personal supportR4 Noncompliance with remediation attemptsR5 Stress
Risk factorsInternal factors1 Intelligence2 Secure attachment in childhood3 Empathy4 Coping5 Self-control
Motivational factors6 Work7 Leisure activities8 Financial management9 Motivation for treatment10 Attitudes towards authority11 Life goals12 Medication
External factors13 Social network14 Intimate relationship15 Professional care16 Living circumstances17 Supervision
Protective factors
Research with the HCR-20 & SAPROFVan der Hoeven Kliniek, The Netherlands
1. Retrospective file study- N = 188 violent + sexual ♂- 108 also pre-treatment rating- Treatment length 5.7 years- Outcome: Reconvictions for violence- Follow-up in community after discharge
- 1 year- 3 year- M = 11 year
De Vries Robbé, De Vogel & Douglas, in preparation
Predictive validity violent recidivismRetrospective file study Violent+Sexual (N=188)
AUC 1 year follow-up
14 recidivist
AUC 3 years follow-up 34 recidivists
AUC 11 years follow-up (M)
68 recidivists
SAPROF (total)
.85*
.75*
.73*
HCR-20 (total) .84* .73* .64*
HCR-SAPROF (total)
.87* .76* .70*
FPJ no violence 5-pt .83* .71* .67*
FRJ all violence 5-pt .84* .72* .68*
N = 188, * p < .01
HCR-SAPROF > HCR-20:χ² (1, N = 188) = 13.4, p < .001 (11 year)
De Vries Robbé, De Vogel & Douglas, in preparation
Logistic regression: sign. incremental predictive validity SAPROF over HCR-20
Changes during treatmentRetrospective study (n = 108)
0
5
10
15
20
25
30
Pre-treatment Post-treatment
Historical
Clinical
Risk managementTotal HCR-20
0
2
4
6
8
10
12
14
Pre-treatment Post-treatment
Internal
Motivational
ExternalTotal SAPROF
HCR-20 SAPROF
HCR-20 total: t (107) = -11.70, d > 0.84, p < .001 SAPROF total: t (107) = 15.63, d > 1.74, p < .001
Treatment changesRecidivists (n = 33) vs. Non-recidivists (n = 75)
HCR-SAPROF Change Recidivists vs. Non-recidivists:t (106) = -4.11,d > 0.85, p < .001
0
2
4
6
8
10
12
14
16
Recidivists Non-recidivists
HCR-20 ChangeSAPROF ChangeHCR-SAPROF Change
HCR-20 and SAPROF Change between pre- and post-treatment ratings
Treatment progress & recidivism
The more progress onprotective & risk factorsduring treatment..
Treatment
Treatment
Community
Community
..the less likelyviolent recidivism
Start mr. X
Start mr. Y
End mr. X
End mr. Y
0102030405060708090
100
1 year 3 year 11 year
Lowprotection
Moderateprotection
0
10
20
30
40
50
60
70
80
90
100
1 year 3 year 11 year
LowprotectionModerateprotectionHighprotection
Moderate risk High risk
Differentiation of risk groupsFinal Protection Judgment�Low
�Moderate
�High
Final Risk Judgment�Low
�Moderate
�High
Logistic regression at all f-u: sign. incremental predictive validity FPJ over FRJ
Research with the HCR-20 & SAPROFVan der Hoeven Kliniek
2. Prospective clinical study• 879 assessments on 325 offenders ♂/♀ (max 1 RA / stage)
- intramural (233)- supervised leaves (159)- unsupervised leaves (141)- transmural first (162)- transmural last (113)- discharge (72)
• Outcome for 315 assessments:- Violent incidents during treatment (12 months)
De Vries Robbé, De Vogel, Wever & Douglas, in preparation
Changes duringtreatmentN = HCR-20 879/ SAPROF 452risk assessments
0
5
10
15
20
25
30
Intramural Supervisedleaves
Unsupervisedleaves
Transmuralfirst
Transmurallast
Discharge
HCR-20SAPROFHCR-SAPROF
Violence risk
Treatment progress
Tot
al s
core
233 159 141 162 113 72
HCR-SAPROF total Intramural vs. Discharge: t (123) = 8.95, d > 1.68, p < .001
Predictive validity violent incidentsn = 315 Clinical risk assessments
0
5
10
15
20
25
30
35
Intramural Supervisedleaves
Unsupervisedleaves
Transmural
HCR-20SAPROF
HCR-SAPROF
Violence risk
29% 15% 7% 3%
Violent incident rate
Total sample 34 incidents
SAPROF (total)
.77*
HCR-20 (total) .79*
HCR-SAPROF (total) .81*
FPJ no violence .70*
FRJ all violence .76*
n = 315, * p < .001
The individual case
0
5
10
15
20
25
30
Intramural Supervised Leaves Unsupervisedleaves
Transmural Discharge
Treatment progress
Tot
al s
core
Changes in scores: theoreticalHCR-SAPROF scores of different risk assessments during treatment
Changes in scores: actualN=1 HCR-SAPROF scores
0
5
10
15
20
25
30
Transmural theoretical Transmural actual
RA 1RA 2RA 3RA 4RA 5RA 6RA 7
Treatment progress
Tot
al s
core
Relapse Relapse
Dynamics of risk assessmentHCR-20 & SAPROF
• The dynamic factors HCR-20 & SAPROF show clinically valuable changes in risk and protective factors
• The more change, the less recidivism
• Individual change in clinical practice is less smooth than the group average suggests
Thank you!Michiel de Vries Robbé /
Vivienne de Vogel / Ellen van den Broek
Van der Hoeven KliniekP.O. Box 174, 3500 AD Utrecht
[email protected]@deforensischezorgspecialisten.nl