prediction of dangerousness in mentally disordered offenders in germany

16
Prediction of dangerousness in mentally disordered offenders in Germany Dieter Seifert a, * , Karen Jahn b , Stefanie Bolten b , Markus Wirtz c a Senior Research Scientist, Institute of Forensic Psychiatry, University/Polytechnic of Essen, Rhineland Clinics Essen, Virchowstrasse 174, 45147 Essen, Germany b Research Associate, Institute of Forensic Psychiatry, University/Polytechnic of Essen, Rhineland Clinics Essen, Essen, Germany c Research Associate, Institute of Psychology III, Westphalian Wilhelms University of Mu ¨nster, Mu ¨nster, Germany 1. Introduction In Germany, offenders who committed crimes while being not or not fully responsible due to a mental disorder are placed in forensic hospitals according to x 63 of the German penal code. Moreover, a negative legal prognosis is mandatory. In the last two decades the number of patients placed in forensic institutions has increased from about 2500 to about 3800. At present there are approximately 60,000 convicts imprisoned. This article deals with the question according to which criteria therapists in forensic hospitals assess the dangerousness of patients about to be released. Up to 188 patients are now included in this research. The assessment of recidivism risk is being carried out with a prognostic questionnaire specifically developed for this purpose. Subjects released from forensic hospitals are monitored over a period of 5 years in order to test the questionnaire with regard to its prognostic validity. 1.1. The problem of criminal prediction concerning forensic patients Serious crimes, such as homicide or sexual offenses, committed by relapsed forensic patients are rare events. Nonetheless, the focus of current forensic research and the German public is still on the issue of assessing the risk of recidivism among this group (e.g., Leygraf, 2000; Seifert & Leygraf, 1997a). International research and literature on the subject show the extensive demand for practicable lists of criteria. The Dangerous Behavior Rating Scale (DBRS, Menzies & Webster 1995, among others), the Violence Risk Appraised Guide 0160-2527/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII:S0160-2527(01)00096-6 * Corresponding author. Tel.: +49-201-7227-103, +49-201-7227-101; fax: +49-202-7227-105. E-mail address: [email protected] (D. Seifert). International Journal of Law and Psychiatry 25 (2002) 51–66

Upload: dieter-seifert

Post on 16-Sep-2016

216 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Prediction of dangerousness in mentally disordered offenders in Germany

Prediction of dangerousness in mentally disordered

offenders in Germany

Dieter Seiferta,*, Karen Jahnb, Stefanie Boltenb, Markus Wirtzc

aSenior Research Scientist, Institute of Forensic Psychiatry, University/Polytechnic of Essen,

Rhineland Clinics Essen, Virchowstrasse 174, 45147 Essen, GermanybResearch Associate, Institute of Forensic Psychiatry, University/Polytechnic of Essen,

Rhineland Clinics Essen, Essen, GermanycResearch Associate, Institute of Psychology III, Westphalian Wilhelms University of Munster,

Munster, Germany

1. Introduction

In Germany, offenders who committed crimes while being not or not fully responsible due

to a mental disorder are placed in forensic hospitals according to x 63 of the German penal

code. Moreover, a negative legal prognosis is mandatory. In the last two decades the number

of patients placed in forensic institutions has increased from about 2500 to about 3800. At

present there are approximately 60,000 convicts imprisoned.

This article deals with the question according to which criteria therapists in forensic

hospitals assess the dangerousness of patients about to be released. Up to 188 patients are

now included in this research. The assessment of recidivism risk is being carried out with a

prognostic questionnaire specifically developed for this purpose. Subjects released from

forensic hospitals are monitored over a period of 5 years in order to test the questionnaire with

regard to its prognostic validity.

1.1. The problem of criminal prediction concerning forensic patients

Serious crimes, such as homicide or sexual offenses, committed by relapsed forensic

patients are rare events. Nonetheless, the focus of current forensic research and the German

public is still on the issue of assessing the risk of recidivism among this group (e.g., Leygraf,

2000; Seifert & Leygraf, 1997a). International research and literature on the subject show the

extensive demand for practicable lists of criteria. The Dangerous Behavior Rating Scale

(DBRS, Menzies & Webster 1995, among others), the Violence Risk Appraised Guide

0160-2527/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved.

PII: S0160 -2527 (01 )00096 -6

* Corresponding author. Tel.: +49-201-7227-103, +49-201-7227-101; fax: +49-202-7227-105.

E-mail address: [email protected] (D. Seifert).

International Journal of Law and Psychiatry

25 (2002) 51–66

Page 2: Prediction of dangerousness in mentally disordered offenders in Germany

(VRAG, Harris, Rice, & Quinsey, 1993), the Psychopathy Checklist (PCL-R, Hare, 1991)

and the HCR-20 (Quinsey, Rice, & Harris, 1995; Webster, Douglas, Eaves, & Hart, 1997)

have been developed in recent years. These questionnaires are mainly based on historical case

data (e.g., age when first offense was committed, previous delinquency). Their validity has

been adequately proven (Menzies et al., 1994). Particularly the predominance of statistically

predictive items has provoked criticism (e.g., Grubin, 1997). There is still a lack of clinically

dynamic criteria, i.e., predictive criteria alterable by treatment.

In a prospective study Verhagen (1993) (cf. Leuw, 1995) showed that the patient’s

assessing his own capabilities in a rather unrealistic way and the clinical staff’s extreme

caution towards the patient (‘‘team assessment’’) suggest a negative legal prognosis.

Historical data do not contribute significantly to improving recidivism prediction. These

results qualify the assumption made at the end of the 1970s, that doctors and psychologists

are not better suited to making risk predictions than other professional groups (Montandan

& Harding, 1984; Quinsey & Ambtman, 1979).

2. Study

This study’s design is prospective. The questionnaire is filled out immediately before the

patient is released from the forensic hospital.

The prognostic questionnaire consists of three parts:

A: sociostatistical and historical items (60 items)

B: psychological testing and biological items (e.g., Neurological Soft Signs, see Schroder

et al. (1992, 1993), intelligence tests, cCT and EEG)

C: clinical items (133 items).

This clinical questionnaire was developed based on predictive characteristics in literature

(e.g., Rasch, 1999) and the prediction index items used in almost all German penal

institutions (Leygraf & Nowara, 1992). The reliability analysis of the questionnaire (Weber,

1996) was satisfactory from the statistical point of view and in line with the results of other

observer rating questionnaires used in general psychiatry (kappa according to Cohen: 0.62).

A few items were added—primarily to distinguish between different sexual deviant develop-

ments and the situation at release. This questionnaire is included in the current study as Part C

(for an overview, see Table 1). It is completed by local psychologists or doctors who,

according to the prestudy examination, have been in charge of the subject’s therapy for more

than 6 months. The sociostatistical, biological, and psychological testing variables are

primarily collected from case files. Subsequent offenses after release are recorded from

excerpts of the central federal data bank (Bundeszentralregister). Apart from the seriousness

of each offense—compared to the offense originally leading to placement—other constel-

lation factors (work and family situation, outpatient follow-up care, living conditions, etc.)

based on probation officers’ reports are considered in the analysis. This method corresponds

with current developments where calls have been made for, e.g., new studies in contextual

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6652

Page 3: Prediction of dangerousness in mentally disordered offenders in Germany

conditions concerning recidivism (Reed, 1997). The prognostic questionnaire consists of

three parts.

The individual items are phrased as questions that therapists have to assess on a four-point

rating scale (example given in Fig. 1).

The majority of questions relates to the patient’s well-being during the last 3 months before

release. Only the 16 questions in Item Group 7 (development/progress) provide an assessment

of progress throughout the period of placement (see Fig. 2 for examples).

For the ‘‘forensic opinion survey’’ therapists use a 10-cm analogue scale to record their

judgment in regard of the probability of a patients relapse or withdrawal of the conditional

release—ranging from extremely low (0) to extremely high (100) in relation to:

1. recidivism with a minor offense

2. recidivism with a serious offense

3. recidivism by violation of court orders

3. Results

3.1. Patients’ characteristics

A total of 188 subjects from 23 forensic hospitals/departments (total of approximately

2000 forensic patients) have so far been included in the study. Thirty-seven subjects are

Table 1

Overview on item groups in the clinical questionnaire (Part C)

1. Current clinical symptoms 13 items

2. Social behavior during placement 18 items

3. Personality stress factors 12 items

4. Adaptive behavior in daily life at present 16 items

5. Emotion/motivation of the patient 24 items

6. Functional and control aspects 20 items

7. Development/course throughout placement 16 items

8. Situation on release (social circumstances on release) 14 items

‘‘Forensic opinion survey’’ (risk of recidivism assessed with a visual analogue scale)

Fig. 1. Item from the clinical questionnaire (Part C).

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 53

Page 4: Prediction of dangerousness in mentally disordered offenders in Germany

presently living in halfway housing. This means that some of the results relate to a smaller

sample scope. The sociodemographic data correspond, to a greater degree, to those of

criminological at-risk groups: three-fourths (76.1%) come from the two lowest socio-

economic status (SES) levels. Over two-thirds are unmarried (71.3%) and nearly half are

high school dropouts (50.5%). More than two-thirds are unemployed (71.7%). The average

age is 35.6 years (median 33.5 years). Four-fifths (81.9%) have been placed in a forensic

hospital for the first time, 6% are placed because of a withdrawal after primary suspension,

the remaining have been placed previously.

The distribution of diagnosis resembles that in previous German release or cross-

sectional studies (Dessecker, 1997; Jockusch & Keller, 1993; Seifert & Leygraf,

1997b). Differences are shown exclusively in the following groups: The percentage

of schizophrenic patients is disproportionately high (45,2%), whereas patients with a

personality disorder (31.9%) are clearly underrepresented (detailed data in Seifert et al.,

2001). In addition, a growing addiction problem has been noticed since the 1980s.

This is explicitly listed as the first or second diagnosis for 37.2% of the subjects (see

Fig. 3).

Offenses leading to placement. The low percentage of sexual offenses (13.3%) among the

crimes leading to placement is particularly striking. In Dessecker’s (1997) release study this

type of offense is represented substantially stronger (27.3% respectively 19.6%) as the cause

of placement. A comparatively high amount of crimes against the person (homicide and

physical injury) was committed by more than two thirds (70%) of patients with a

schizophrenic psychosis. In Dessecker’s survey, however, the proportion of homicidal crimes

was considerably lower (14.3% respectively 17.4%). Two patients committed homicidal

offenses with a sexual motivation. Seventy percent of the patients have been convicted

previously. It has been reported that every ninth patient currently placed in forensic hospitals

committed crimes that usually are less serious by definition (e.g., theft and threatening

behavior) than the original offense.

The duration of placement (n = 151) ranges from 1 to 33 years (median 4.6 years;

Dessecker, 1997: 4.3 years). Patients who committed homicidal crimes stayed approximately

twice as long as the remaining (median: 86.2 vs. 44.4 months). Dessecker (1997, p. 121) finds

‘‘no indication for crime-specific influences.’’ There is no correlation between duration of

placement and a particular kind of disorder.

Fig. 2. Item from the clinical questionnaire (Part C), Item Group 7: Development/Course.

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6654

Page 5: Prediction of dangerousness in mentally disordered offenders in Germany

3.2. Clinical prediction characteristics

The results of a first evaluation of the clinical questionnaire (Part C) are represented as

follows. The question is, according to which clinical characteristics therapists in forensic

institutions conclude their assessment of risk. Two central questions will be answered in

the overall study:

1. Which characteristics are the therapist’s predictions about the recidivism risk at

the time of release based on? This is of particular interest because faced with a

forthcoming release the court primarily counts on the judgment of the therapist

in charge.

2. According to which characteristics may a patient’s actual relapse be predicted? Of

particular interest is whether the risk assessed at the time of release allows a

reliable prognosis.

This analysis only deals with the first question. Results of the predictive quality of

therapists’ assessments cannot be announced until follow-up data have been collected.

Fig. 3. Distribution of offences leading to placement—prediction study 2000 versus Dessecker 1997*.

*Dessecker compared data of two inquiry dates (1980 vs. 1986) in his 1997 published study.

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 55

Page 6: Prediction of dangerousness in mentally disordered offenders in Germany

4. Methodology

The prediction of therapists’ judgments are limited to four out of eight content spheres (see

Fig. 1: Item Groups 1, 2, 3, and 7). This is because according to clinical practical experience

and the current developments in research these four item groups play the most important role

concerning legal predictive assessment of the relevant group. However, reducing the number

of predictive variables may result in increasing the meaningfulness of the statistical tests

(reducing the probability of a errors).

To be able to guarantee the reliability of prediction variables, homogenous item groups

within the content spheres are identified by means of factor analysis. If these subgroups prove

to be internally consistent (Cronbach’s a), the relevant item results will be summed in scale

results. At the scale level, the evaluation now refers to only a few fundamentally more reliable

characteristics, so that the power of statistical tests has been increased and the threat of wrong

decisions has been reduced.

Thus identified scales are correlated with the therapist’s assessment of the recidivism

probability particularly with regard to serious offenses (2. question of the ‘‘forensic opinion

survey’’). In case of highly reliable factors, an examination at the level of single items follows.

There is one difficulty, however, about examining the reliability of the dependent variable,

i.e., the assessment of recidivism risk on the visual analogue scale: It is only one single person

who carries out the assessment. In forensic hospitals, there usually are few persons with a

high knowledge level about a patient. Thus, the person who possesses the most information

answers the ‘‘forensic opinion survey’’ and gives evidence at court. Furthermore, it has to be

considered that this person comes to a decision only after a detailed examination of the

patient’s situation. This decision is the result of a long-lasting cognitive process in contrast to

an ad hoc assessment. Accordingly the ‘‘forensic opinion survey’’ currently appears to be the

only useful and at the same time practicable method to measure assessments of a patient’s

dangerousness. Furthermore, it constitutes the most valid judgment (construct validity).

Since, under these factual circumstances, no comparative judgments with a similarly high

validity can be obtained, an empirical examination of the reliability is not possible.

4.1. Assessed risk of recidivism

For the total study population (N = 188) recidivism with a minor offense or based on

violation of a court order is regarded as more probable than recidivism with a serious offense,

F(2,370) = 30.22*** (see Fig. 4).

4.1.1. Factor analysis for four item groups and correlation analysis at

the individual item level (clinical questions, Part C)

Based on the evaluations for the four item groups (see Table 1) with factor and individual

item analysis, statements can be made about how therapists assess their patients regarding the

clinical factors or individual items if a relatively high or low risk of recidivism, particularly in

respect of serious offenses (2. question of the ‘‘forensic opinion survey’’), is expected

(summary of results see Table 2).

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6656

Page 7: Prediction of dangerousness in mentally disordered offenders in Germany

In another analysis step diagnosis groups are separated. Because of the required sample

size, the following three groups are being chosen: schizophrenic psychosis (N = 84),

personality disorder with and without insufficient talent (N = 58) and subjects with an

intellectual handicap or a cerebral organic disorder (N= 46).

4.1.1.1. Current clinical symptoms (Item Group 1, 13 items). Two factors can be extracted

from the 13 individual items in which only Factor 1.1 (psychotic symptoms) shows a

sufficiently high reliability (a = .63). This factor does not correlate significantly with

assessments concerning serious offenses. However, a significant correlation with the

second question of the ‘‘forensic opinion survey’’ can be shown for the subgroup of

schizophrenic patients (r= .234*)1. For this group of subjects therapists assess a high

probability of recidivism with a serious offense if the patient psychopathologically

unchanged shows distinct positive and/or negative symptoms (Table 3).

Fig. 4. Analogue scales for the three questions of the ‘‘forensic opinion survey.’’

1 Significances *P< .05 (two-sided); **P< .01 (two-sided); ***P< .001 (two-sided).

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 57

Page 8: Prediction of dangerousness in mentally disordered offenders in Germany

The analysis at the level of individual items shows that the subgroup of subjects with a

sexually deviant development correlates significantly with the second question of the

‘‘forensic opinion survey’’ (r= .375*): The more ego-syntonic the subject experiences his

or her sexual disorder the worse the therapist’s prediction will turn out.

Table 2

Correlation between individual items and the second question of the ‘‘forensic opinion survey’’

Individual items

Pearson correlation

(r)aSignificance level

( p) Factor

Current symptoms

Personality disorder � .225 < .01

Sexually deviant development .197 < .01

Syntonic/dystonic sexual development � .375 < .05

Addictive potential—alcohol � .165 < .05

Social behavior

General team assessment .424 < .001 2.1: conspicuous social

Dealings of team cautious/tense � .318 < .001 behavior

Quarrelsomeness towards fellow patients � .157 < .05

Quarrelsomeness towards team � .150 < .05

Social contact with members of the team .297 < .001 2.2: ego strength in social

Inconsistent social behavior � .217 < .01 contact

Social contact with fellow patients .180 < .05

Social contact outside of the hospital .163 < .05

Relationship disorder � .178 < .05

Courage of one’s convictions .269 < .001

Personality stressors

Hostility � .343 < .001 3.1: violent attitude

Unpredictability � .316 < .001

Sadistic tendencies � .224 < .01

Aggressiveness � .199 < .01

Lack of emotion � .202 < .01 3.2: schizoid structure

Mistrust � .185 < .05

Inability to bind � .171 < .05

Development/Course

Addiction control improved? � .379 < .001

Reflexivity improved? � .289 < .001

More realistic future planning? � .262 < .001

Basic psychological disorder improved? � .233 < .001

Subculture identification decreased? � .298 < .01

Ability to assert improved? � .235 < .01

Attitude towards offense more self-critical? � .206 < .01

Control of aggression increased? � .200 < .05

Perception of disorder improved? � .153 < .05

Time structuring possible? � .160 < .05a Differing negative/positive correlations result from the polarity of the items.

#

#

#

##

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6658

Page 9: Prediction of dangerousness in mentally disordered offenders in Germany

4.1.1.2. Social behavior during placement (Item Group 2, 18 items). Two factors with a

very high reliability can be extracted from this item group (each a = .8), which at the

same time correlate highly significantly with the second question of the ‘‘forensic

opinion survey’’. Factor 2.1 (conspicuous social behavior; r= .358**) includes assess-

ments referring to the social, particularly aggressive, contact to other persons during

placement and, furthermore, the care-unit’s general assessment (Table 4): Accordingly,

therapists presume a negative legal prognosis if the patient’s social behavior during

placement is primarily characterized by ‘‘quarrelsomeness,’’ ‘‘manipulative scenes,’’ less

featured ‘‘social competence,’’ and the team is ‘‘particularly cautious, observant, and

tense’’ in dealing with the patient.

Factor 2.2 (schizoid structure; r= .309**) describes the social and relationship pattern

towards fellow patients and members of the team (Table 4). If a patient withdraws,

avoids—if possible—close emotional relations to other people or shows to be only

superficial and changeable in establishing contact, a high recidivism risk is presumed.

The analysis of variance (ANOVA) shows that with regard to Factor 2.1 the different

diagnosis groups differ highly significantly, F(2,151) = 11.7***: In comparison to the

other two diagnosis groups, a significantly higher ‘‘conspicuous social behavior’’ has

been found in the group of schizophrenic patients. Accordingly, Factor 2.1 is of deciding

importance. It shows a high internal consistency that correlates highly significantly with

the second question of the ‘‘forensic opinion survey.’’ Furthermore, this factor discrim-

inates (highly significantly) between the three diagnosis groups and includes both of the

individual items, that show the highest correlation with an assessment of a serious

recidivism offense (‘‘care unit’s general assessment’’ [r= .424***] as well as the ‘‘care

unit’s dealing with the subject’’ [r = .318***]).

Table 3

Factor extracted from Item Group 1: current clinical situation

Factor 1.1: psychotic symptoms

� Hospitalism/residual symptoms� Productive-psychotic symptoms� Gravity of disease/disorder

Table 4

Factor extracted from Item Group 2: social behavior during placement

Factor 2.1: conspicuous social behavior Factor 2.2: ‘‘ego strength in social contact’’

� Quarrelsomeness towards fellow patients � Social contact with team� Quarrelsomeness towards team � Social contact with fellow patients� Manipulative scenes � Social contact outside the hospital� Quarrelsomeness towards family/friends � Relationship disorder� Team’s dealing with the patient � Stability and constancy of social behavior� Sort of influence on fellow patients � Reclusiveness� General assessment by team

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 59

Page 10: Prediction of dangerousness in mentally disordered offenders in Germany

4.1.1.3. Personality stressors (Item Group 3, 12 items). With these items therapists have

to present a cross section of the patient’s personality (‘‘patient’s basic attitude’’). A factor

analysis results in the following two factors (internal consistency: a = .78 and .71) of

which each correlates significantly with the assessment of a serious recidivism offense

(Table 5). Factor 3.1 (violent attitude; r = .249**) discriminates highly significantly

between the three diagnosis groups, in which particularly the diagnosis groups schizo-

phrenia and personality disorder can be differentiated distinctly: Therapists describe the

basic attitude of discharged schizophrenic patients as being significantly less ‘‘hostile,

aggressive, and unpredictable’’ than that of patients with a personality disorder.

4.1.1.4. Development/course of clinical patients’ characteristics (Item Group 7, 16

items). These 16 items give information about the subject’s development and the course

of the therapy during the whole period of placement. The items can be combined in a scale of

very high internal consistency. Because of the strong homogeneity, all effects at the scale

level appear to resemble those at the individual item level. According to therapists’ assess-

ments altogether seven items are of particular importance with regard to the question of

recidivism. Criteria consulted to assess a positive legal prognosis are shown in Table 6.

4.2. Construct validity

All of the above-mentioned factors (with high internal consistency) show a highly

significant intercorrelation. This constitutes one aspect of construct validity. The

Table 5

Factor extracted from Item Group 3: personality stressors

Factor 3.1: aggressive/violent attitude Factor 3.2: schizoid structure

� Aggressiveness � Inhibitedness� Ambition for power � Mistrust� Externalization tendencies � Inability to connect� Sadistic tendencies � Lack of emotion� Unpredictability � Depressiveness� Hostility

Table 6

Significant correlation between the course items and the second question of the ‘‘forensic opinion survey’’

Pearson correlation (r) Significance level ( p)

Control of drug consumption improved .379 < 0.001

No subculture identification (anymore) .298 < 0.001

Reflexivity improved .289 < 0.001

Future planning appears to be realistic .262 < 0.001

Ability to assert improved .235 < 0.01

Basic psychological disorder improved .233 < 0.01

Assessment of offense more self-critical .206 < 0.01

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6660

Page 11: Prediction of dangerousness in mentally disordered offenders in Germany

correlation between the data from the prognostic questionnaire and the external

criterion ‘‘recidivistic versus not recidivistic’’ will essentially indicate the construct

validity throughout this study. The usefulness of the prognostic questionnaire will thus

be proven.

4.2.1. Results for the diagnosis groups schizophrenia and personality disorder

4.2.1.1. Schizophrenic patients (N = 84). The three diagnosis groups differ highly

significant in terms of the second question of the ‘‘forensic opinion survey.’’ In

comparison to the group of patients with a personality disorder (mean 18.3) the

probability of a serious subsequent offense after placement is assessed to be significantly

lower, F(2,173) = 6.1**, for the group of schizophrenic patients (mean 11.3) and patients

with a cerebral organic disorder or an intellectual handicap (mean 11.4). The assessment

of recidivism risk is primarily based on the items and factors given in Table 7.

Altogether, there are three factors and three course criteria (Item Group 8) that correlate

significantly with the second question of the ‘‘forensic opinion survey.’’ Furthermore,

five individual items can be found that correlate highly significantly with the assessment

of recidivism risk: Accordingly, a schizophrenic patient’s unpredictability (r= .435***),

aggressiveness (r= .391***), tendency to conversion symptoms (r= .348**), and exter-

nalization tendencies (r= .305**), and the fact that the care unit is particularly cautious,

tense, and observant in everyday dealing with the patient (r= .341**), suggest a rather

negative legal prognosis.

4.2.1.2. Personality disorder (N= 58). For this group there are four factors and four

course criteria therapists correspondingly use to judge the legal prognosis (Table 8).

Altogether, five individual items, which are also included in the factors, correlate highly

significantly with the assessment of recidivism: In this group, therapists accordingly

consider hostility, that has been noticed during everyday life within the hospital

(r = .489***), lack of emotion (r = .385**), and unpredictability to be negative prognosis

Table 7

Factors/course items relevant for the diagnosis group schizophrenia

Schizophrenia Pearson correlation (r) Significance level ( p)

Factor

Factor 3.1: violent attitude .346 < .01

Factor 2.1: conspicuous social behavior .331 < .01

Factor 1.1: psychotic symptoms .219 < .05

Course and development criteria

Control of aggression increased? .310 < .05

Realistic future planning? .304 < .01

Levelheadedness (reflexivity) increased? � .234 < .05

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 61

Page 12: Prediction of dangerousness in mentally disordered offenders in Germany

criteria. Further risk characteristics are a lack of social contact to the team (r= .369**)

and a rather (general) negative team assessment (r= .371).

5. Discussion

The complexity of assessing dangerousness is an international problem, as clearly verified

in a WHO study (Harding & Adserballe, 1983, 1984). Research results on the validity of

clinical characteristics, which enable therapists to make statements about the legal prognosis

of forensic patients, are rarities. Although this conclusion has been drawn in literature

repeatedly, in practice the question of release from a forensic institution is primarily judged

on whether a patient’s development observed during placement has made such good progress

that, according to law, ‘‘it is not to be expected, that the placed person will commit any further

illegal acts after release.’’ After all, there is a large degree of correspondence in this

proceeding and the clinical methods of prognosis. What this clinical assessment is based

on in particular has neither been recorded nor been verified empirically. Because of these

methodological uncertainties, there are continuous complaints about a reduction of criteria

used in forensic prediction practice (Rasch, 1999; Weber, 1996).

In short, the following results can be recorded from the interim evaluation of our study,

which is still in progress.

5.1. Current release practice in Germany

In comparison to previous research the number of releases from forensic hospitals in

Germany has fallen significantly (Dessecker, 1997; Jockusch & Keller, 1993). This

conservative release practice primarily refers to patients who, by diagnosis, are classified

as having personality disorders and/or those convicted of a sexual offense (Fig. 3). An

explanation for this may be that forensic psychiatry in Germany has become a political issue

over the past 5 years. Consequently, there is a noticeable trend in Germany towards

Table 8

Factors/course items relevant for the diagnosis group personality disorder

Personality disorder Pearson correlation (r) Significance level ( p)

Factor

Factor 2.2: ego strength in social contact .432 < .01

Factor 2.1: conspicuous social behavior .334 < .05

Factor 3.1: violent attitude .290 < .05

Factor 3.2: schizoid structure .245 < .05

Course and development criteria

Control of drug consumption improved? .572 < .001

Basic psychological disorder improved? .356 < .01

Levelheadedness (reflexivity) increased? � .157 < .05

Improved ability to establish social relationships? � .150 < .05

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6662

Page 13: Prediction of dangerousness in mentally disordered offenders in Germany

overpredicting the risk posed by mentally ill offenders in an attempt to create a greater

certainty in the decision-making progress. This problem had been widely discussed in the

1970s in the USA (Monahan, 1978; Steadman, 1983).

5.2. Clinical criteria used by therapists to predict dangerousness

To start with, a basic problem has to be mentioned: In an attempt to investigate valid

clinical predictive items, apart from the familiar methodical problems (low basis probability

of violent delinquency, failure to identify so-called ‘‘false positives,’’ prediction period),

further difficulties turned up:

� Historical case-detail data are very suitable for operationalized recordings and, hence,

dominate the current predictive measures whereas this is not the case for clinical

dynamic criteria.� Even a predictive tool including main clinical parameters cannot be used

indiscriminately for all patients. The group of placed patients is too heterogeneous in

terms of disorder pattern, offense leading to placement, socialization, etc. Ultimately an

individual prognosis is required, in which particular items play a more important role

than others. For instance, questions about the patient’s addictive potential are only

relevant in terms of prediction, if an addiction actually exists and is causally related to

the patient’s delinquency.

Based on these reflections the most important aspects according to which therapists

achieve a rather positive or negative assessment of their patients’ dangerousness are

summarized for the total study group as follows:

� On the whole, therapists use clinical criteria. This corresponds to the results of a Dutch

prospective study (see Leuw, 1995; Verhagen, 1993). A recently published Swedish

(retrospective) research including 40 subjects (Strand, Belfrage, Fransson, & Levander,

1999) has also shown that particularly the clinical items of the predictive score HCR-20

have a highly valid meaningfulness. In the therapists’ opinion the greatest predictive

value is attributed to questions about ‘‘social behavior’’ (Item Group 2), as reflected

statistically by the quantity of significant individual items, but also with regard to the—

factor-analytically identified—latent dimensions. Consequently, social skills during

placement (on the ward, during leave, in the residential group) are considered the best

qualification for a crime-free life after having completed the forensic therapy. This does

not only mean ‘‘adaptive behavior in institutional life’’ (known from and criticized by

literature), but rather describes the patient’s continual social competence and ability to

establish relationships.� The item ‘‘team assessment’’ achieves the highest correlation among all 133 individual

items: If the care team’s general assessment is positive, the therapist will also certify a

favorable legal prognosis. If, however, the team is particularly ‘‘cautious, observant, and

tense’’ in dealing with the patient, a high risk of recidivism is assumed. This finding is

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 63

Page 14: Prediction of dangerousness in mentally disordered offenders in Germany

in agreement with the Dutch study by Verhagen (1993). The parameter clearly shows

the possibility to operate with clinical items. ‘‘Team assessment’’ constitutes an

intuition, a rather instinctive feeling. Which detailed reflections have been the reason for

this assessment cannot be put into concrete terms at the moment. The ‘‘intuitive’’ feeling

is based on the knowledge that results from the care unit’s long-term observations and

experiences in daily dealing with this patient and consequently is the result of a

cognitive process.� Concerning recidivism assessment, therapists are guided by the personality profile

that has emerged during the period of placement. An aggressive attitude observed in

daily contact and/or a schizoid structure are regarded as criteria for a rather negative

legal prognosis.� Only in cases of items referring to ‘‘current situation’’ the often complained about lack

of criteria can be confirmed. With regard to recidivism risk, a (significant) importance of

the factor ‘‘psychotic symptoms’’ can only be attributed relating to the group of

schizophrenic patients.� The legal prognosis of patients with a personality disorder is basically (significant)

worse in comparison to other diagnosis groups. At the same time correspondingly used

prognosis criteria can be found more often for this group than for schizophrenic patients.

The markedness of violent attitude (Factor 3.1) and conspicuous social behavior (Factor

2.1) are decisive criteria, which furthermore significantly distinguish between these two

diagnosis groups. The main tenor is that therapists classify patients with schizophrenia

as less conspicuous on both factors than patients with personality disorder.

6. Prospect

This research shows that therapists in forensic hospitals do consult clinical criteria to assess

their patients’ legal prognosis. Thus, the permanently complained about (at least in Germany)

‘‘reduction of criteria’’ does not reflect the current situation. This result is of importance for

the forensic experience: The quality of prognostic reports can be improved by using this

questionnaire as a checklist, which for reasons of close relation to therapy is helpful in

decision making. Even while working on the individual items, clinical and case history

aspects are touched on and are therefore not simply forgotten, as proven in retrospective

studies on recidivists (Pierschke, 1998). Furthermore, the extracted criteria can be considered

in legal predictive reflections and discussed in relation to individual cases. The results

reported in our study are exclusively based on therapists’ assessments. Whether these criteria

actually have predictive relevance will not be certain until the 2- to 5-year follow-up has been

completed. At that time, historical data will also be included. It will be less important to work

in a dichotomous way, i.e., to figure out whether the historical or the clinical data are the

decisive prediction criteria. Nor is a single item such as ‘‘age at first offense’’ or a clinical

variable such as ‘‘tolerance of frustration’’ expected to be the decisive characteristic. It is

more likely that a certain pattern of predictive characteristics proves to be of importance in

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6664

Page 15: Prediction of dangerousness in mentally disordered offenders in Germany

assessing legal diagnosis for the affected group of patients (e.g., with a particular diagnosis

and/or offense group). Besides, further analysis using probation officers’ reports may give

information about relevant interactive factors in the development of recidivistic offenses—

comparable to the reflections about the validity of criminalistic models for legal prognostic

assessments (Krauss, Sales, Becker, & Figueredo, 2000). In this context, a distinction must be

made, whether a recidivism was potentially predictable or at the same time could have been

avoided (Munro & Rumgay, 2000).

Acknowledgments

Financial support is provided by the German Research Association (DFG).

References

Dessecker, A. (1997). Straftater und Psychiatrie. Eine empirische Untersuchung zur Praxis der Maßregel gemaß x63 StGB im Vergleich mit der Maßregel gemaß x 64 StGB und sanktionslosen Verfahren. Wiesbaden, Germany:

Kriminologische Zentralstelle.

Grubin, D. (1997). Predictors of risk in serious sex offenders. British Journal of Psychiatry, 170 (32), 17–21.

Harding, T. W., & Adserballe, H. (1983). Assessment of risk in serious sex offenders. British Journal of Psy-

chiatry, 6, 391–398.

Harding, T. W., & Adserballe, H. (1984). Assessment of dangerousness: Observations in six countries. A summary

of results from a WHO coordinated study. International Journal of Law and Psychiatry, 6, 391–398.

Hare, R. D. (1991). Manual for the Hare Psychopathy Checklist—Revised. Toronto: Multi-Health Systems.

Harris, G. T., Rice, M. E., & Quinsey, M. L. (1993). Violent recidivism of mentally disordered offenders: the

development of a statistical prediction instrument. Criminal Justice and Behavior, 20, 315–335.

Jockusch, U., & Keller, F. (1993). Differenzierung und Spezialisierung einer Abteilung des Maßregelvollzuges in

der psychiatrischen Klinik—mogliche Auswirkungen auf Behandlungsdauer und strafrechtliche Ruckfallig-

keit. Psychiatrische Praxis, 20, 12–15.

Krauss, D. A., Sales, B. D., Becker, J. V., & Figueredo, A. J. (2000). Beyond prediction to explanation in risk

assessment research. International Journal of Law and Psychiatry, 23 (2), 91–112.

Leuw, E. (1995). Recidive na ontslag uit tbs. Arnhem, Netherlands: Gouda Quint.

Leygraf, N. (2000). Die Begutachtung der Prognose im Maßregelvollzug. In: U. Venzlaff, & K. Foerster (Eds.),

Psychiatrische Begutachtung (3rd ed., pp. 349–358). Stuttgart: Fischer.

Leygraf, N., & Nowara, S. (1992). Prognosegutachten. Klinisch-psychiatrische und psychologische Beurteilungs-

moglichkeiten der Kriminalprognose. Forensia-Jahrbuch, 3, 43–53.

Menzies, R., & Webster, C. D. (1995). Construction and validation of risk assessment in a six-year follow-up of

forensic patients: a tridimensional analysis. Journal of Consulting and Clinical Psychology, 63 (5), 766–778.

Menzies, R. J., Webster, C. D., McMain, S., Staley, S., & Scaglione, R. (1994). The dimensions of dangerousness

revisited: assessing forensic predictions about violence. Law and Human Behavior, 18, 1–28.

Monahan, J. (1978). Prediction research and the emergency commitment of dangerous mentally ill persons: a

reconsideration. American Journal of Psychiatry, 135, 198–201.

Montandan, C., & Harding, T. (1984). The reliability of dangerous assessments. A decision making exercise.

British Journal of Psychiatry, 144, 149–155.

Munro, E., & Rumgay, J. (2000). Role of risk assessment in reducing homicides by people with mental illness.

British Journal of Psychiatry, 176, 116–120.

Pierschke, R. (1998). Totungsdelikte nach—scheinbar—gunstiger Legalprognose. In: H. L. Krober, & K. P. Dahle

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–66 65

Page 16: Prediction of dangerousness in mentally disordered offenders in Germany

(Eds.), Sexualstraftaten und Gewaltdelinquenz. Verlauf—Behandlung—Opferschutz (pp. 185–188). Heidel-

berg: Kriminalistik.

Quinsey, V. L., & Ambtman, R. (1979). Variables affecting psychiatrists’ and teachers’ assessments of the danger-

ousness of mentally ill offenders. Journal of Consulting and Clinical Psychology, 477, 353–362.

Quinsey, V. L., Rice, M. E., & Harris, G. T. (1995). Actuarial prediction of sexual recidivism. Journal of Inter-

personal Violence, 10 (1), 85–105.

Rasch, W. (1999). Forensische Psychiatrie (2nd ed.). Stuttgart: Kohlhammer.

Reed, J. (1997). Risk assessment and clinical risk management: the lessons from recent inquiries. British Journal

of Psychiatry, 170 (32), 4–7.

Schroder, J., Niethammer, R., Geider, F.-J., Reitz, C., Blinkert, M., Jauss, M., & Sauer, H. (1992). Neurological

soft signs in schizophrenia. Schizophrenia Research, 6, 25–30.

Schroder, J., Richter, P., Geider, F.-J., Niethammer, R., Blinkert, M., Reitz, C., & Sauer, H. (1993). Diskrete

motorische und sensorische Storungen (neurological soft signs) in Akutverlauf endogener Psychosen. Zeits-

chrift fur Klinische Psychologie, Psychopathologie und Psychotherapie, 41, 190–206.

Seifert, D., Bolten, S., & Jahn, K. (2001). Wer wird momentan aus dem Maßregelvollzug (x 63 StGB) entlassen?

Erste Ergebnisse einer prospektiven Studie zur Gefahrlichkeitsprognose bei psychisch kranken Straftatern.

Seifert, D., & Leygraf, N. (1997a). Straftaten wahrend und nach einer Behandlung im Maßregelvollzug gemaß x63 StGB. Deutsche Richterzeitung, 8, 338–345.

Seifert, D., & Leygraf, N. (1997b). Die Entwicklung des psychiatrischen Maßregelvollzugs (x 63 StGB) in

Nordrhein-Westfalen. Psychiatrische Praxis, 24, 237–244.

Steadman, H. J. (1983). Predicting dangerousness among the mentally ill: art, magic and science. International

Journal of Law and Psychiatry, 6, 381–390.

Strand, S., Belfrage, H., Fransson, G., & Levander, S. (1999). Clinical and risk management factors in risk

prediction of mentally disordered offenders—more important than historical data. Legal and Criminological

Psychology, 4, 67–76.

Verhagen, T. (1993). Paper presented at the 19th International Congress on Law and Mental Health. Lisbon, June

13–17, 1993.

Weber, F. (1996). Gefahrlichkeitsprognose im Maßregelvollzug. Pfaffenweiler: Centaurus.

Webster, C. D., Douglas, K. S., Eaves, D., & Hart, S. D. (1997). The HCR-20 Scheme. The assessment of danger-

ousness and risk. Vancouver: Simon Fraser University, Burnaby Mental Health, Law and Policy Institute.

D. Seifert et al. / International Journal of Law and Psychiatry 25 (2002) 51–6666