Early Criminal Recidivism Among Mentally Disordered Offenders

Abstract

Criminal recidivism was studied during 2 years in a Swedish population–based cohort (N = 318) of mentally disordered male offenders who had undergone a pretrial forensic psychiatric investigation, been convicted in subsequent trials, and been sentenced to forensic psychiatric treatment (FPT; n = 152), prison (n = 116), or noncustodial sanctions (n = 50). Recidivism was analysed in relation to index sanctions, levels of supervision, diagnoses, and criminological factors. Significantly lower recidivism in the FPT group was related to lower crime rates during periods at conditional liberty in this group alone, and recidivism was significantly more common among offenders with at least one of the two diagnoses of substance abuse disorder and personality disorder than among those with psychotic or other mental disorders alone. Age at index crime and number of previous crimes emerged as significant predictors of recidivism. The results of this study suggest that the relapse rates depend as much on level of supervision as on individual characteristics.

Keywords

mental disorder criminal sanction prison forensic psychiatry

Studies on criminal recidivism in offenders with mental disorders vary widely in several core aspects. Mentally disordered criminal offenders may, for instance, be defined in various ways, ranging from offenders who at some time have been assigned a diagnosis of a mental disorder to offenders in whom a causal relationship is assumed between a severe mental disorder and a crime. Follow-up periods may be as short as approximately 1 year (Steadman et al., 1998) or extend over a decade or more (Buchanan, 1998; Warren, Mullen, Thomas, Ogloff, & Burgess, 2008). Of course, longer follow-up periods provide more data on recidivism, but relapse rates will also vary depending on whether the outcome measure is based on rate of recidivism, that is, the proportion of offenders who relapsed in crime, or on reconviction rate, that is, the proportion of offenders with reconvictions during the follow-up. In addition to these variations, one apparent limitation within this field is that research has almost exclusively been focused on recidivism in relation to individual characteristics, such as diagnoses and psychopathy, and not on outcome in relation to different settings with different levels of supervision during and after serving a sanction or undergoing treatment. Although this is likely to be an important factor involved in recidivism, only a limited number of studies have, to the best of our knowledge, reported on outcome during different sets of circumstances with different degrees of supervision. Bailey and Macculloch (1992) found significantly higher reconviction rates among patients discharged from special hospitals without conditions into the community compared with those who were discharged with conditions. Among personality-disordered offenders discharged from forensic hospitals, Davison, Jamieson, and Taylor (1999) reported a lower criminal relapse rate during continued treatment in general psychiatric hospitals and outpatient compulsory care than during nonconditional discharge to the community. Finally, a couple of studies from North America have found support for the effectiveness of conditional release programs because insanity acquittees on conditional release tend to show lower rates of reoffenses compared with those who are absolutely discharged (McGreevy, Steadman, Dvoskin, & Dollard, 1991; Wiederanders, 1992; Wiederanders, Broley, & Choate, 1997).

International studies, though working with different follow-up periods and cultural contexts as well as with different samples of mentally disordered offenders have presented a picture of violent recidivism among offenders with mental disorders that seems to vary between 6% (Maden, Scott, Burnett, Lewis, & Skapinakis, 2004) and 12% (Friendship, McClintock, Rutter, & Maden, 1999). In a Swedish study, the number of recidivists in violent crimes was even larger: the rate of recidivism within 2 years of discharge from sanctions was 26% in offenders with a personality disorder (PD) and 14% in those with schizophrenia (Grann, Belfrage, & Tengström, 2000).

Most studies on criminal recidivism in mentally disordered offenders have been performed in the special category of severely disordered patients in forensic psychiatric treatment (FPT; Jamieson &Taylor, 2002; Maden, Rutter, McClintock, Friendship, & Gunn, 1999). Some have focused on specific diagnostic groups with or without concomitant disorders in offenders admitted for a forensic psychiatric investigation (FPI), irrespective of the subsequent sanction (e.g., Grann et al., 2000).

Mental disorders are known to be more prevalent among prison inmates than in the general population (Hodgins, 1992; Rasmussen, Almvik, & Levander, 2001; Teplin, 1990). In a review of mental disorders in prison populations (Andersen, 2004), the prevalence of psychiatric diagnoses was 2% to 4% for psychotic disorders, 2% to 14% for depression, 6% to 30% for “neurotic disorders,” and 30% to 75% for substance abuse disorder (SAD). The frequency of PD, predominantly antisocial PD, was 46% to 88% in North American studies and 37% to 56% in European studies (Andersen, 2004). A systematic review from 12 Western countries showed an overall prevalence rate of 65% for PD in male prisoners (Fazel & Danesh, 2002), whereas the corresponding figure in a community sample was below 10% and showed a different distribution of the PD subtypes (Torgersen, Kringlen, & Cramer, 2001). A high criminal propensity in offenders with an early development of norm-breaking behaviours (corresponding to the trajectory from conduct disorder to antisocial PD) is well known, especially when associated with an early onset of polysubstance abuse (Kreuger, Markon, Patrick, Benning, & Kramer, 2007; Manuzza, Klein, & Moulton, 2008), referred to as “type II alcoholism” in the taxonomy established by Cloninger, Bohman, and Sigvardsson (1981).

Overall, PD and SAD are the most common diagnoses in violent offender populations, far more prevalent than major mental disorders such as affective disorders and schizophrenia. Violent criminality among offenders with major mental disorders has been related to concomitant SAD (Wallace et al., 1998), whereas the risk of criminal recidivism in offenders with psychotic disorders alone, without either PD or SAD, has been consistently reported as significantly lower (Fazel, Långström, Hjern, Grann, & Lichtenstein, 2009; Moran et al., 2003; Steadman et al., 1998; Walsh, Buchanan, & Fahly, 2001). In Sweden, Grann and Fazel (2004) have reported that during 1988 to 2000, 1.9% of the population over the age of criminal responsibility with at least one admission to inpatient treatment with a SAD diagnosis committed 25% of all violent crimes.

Although mental disorders and SAD have been found to affect the rate of criminal recidivism, an extensive meta-analysis, including totally 64 samples found that previous criminality was the most important predictor of criminal recidivism whether or not the offenders suffered from a mental disorder. Thus, the major correlates of crimes, that is, previous criminal behaviour and personal demographics (gender and age) did not differ between mentally disordered offenders and nonmentally disordered offenders (Bonta, Law, & Hanson, 1998). The superior predictive accuracy of earlier criminal behaviour over psychopathic personality traits was also established in an analysis of six different forensic/correctional samples assessed with the four facets of the psychopathy checklist revised (Walters, Knight, Grann, & Dahle, 2008).

To describe criminal recidivism during different forms of sanctions and levels of supervision during the first 2 years after the index sentence, we have followed a population-based cohort of criminal offenders referred to pretrial forensic investigation in Sweden, diagnosed with at least one mental disorder, and sentenced to FPT, prison, or noncustodial sanctions.

Aims of the Study

The aim of this study was to assess the following hypotheses:

Hypothesis 1: Offenders sentenced to FPT have significantly lower recidivism rates than those sentenced to prison or other sanctions.

Hypothesis 2: Offenders with either SADs or PDs, or both, are significantly more prone to criminal recidivism than offenders without these two diagnoses.

Hypothesis 3: Criminological factors contribute significantly as predictors of early recidivism.

Ethical Considerations

The study was approved by the Research Ethics Committee of the University of Gothenburg in accordance with the law regulating access to registers of delicate information for researchers (the Personal Data Act).

Method

This study is based on individuals who have been subjected to a FPI, but the fact that some aspects of the Swedish legislation and praxis within this field differ from the norm in most other countries warrants a brief background description.

Crime Rates and FPIs in Sweden

In Sweden, the crime rates for any type of crime and for violent crimes are approximately in the midrange in comparisons across the European countries (Council of the European Union, 1999). European crime rates are generally lower than those in the United States (Federal Bureau of Investigation, 2002). However, comparisons of criminality between countries are uncertain due to differing legislations, definitions of crimes, and registration of reported and prosecuted offenses (Segessenmann, New Zealand, 2002; Swedish National Council for Crime Prevention, 2001). According to official Swedish statistics on violent crimes, 150 convictions during 2009 concerned murder or manslaughter and 9,250 concerned assault and aggravated assault (Swedish National Council for Crime Prevention, 2009). In general, each year, 550 individuals prosecuted for crimes, primarily crimes of violence, undergo a FPI ordered by the court before sentencing (data from the National Board of Forensic Medicine, Annual reports, 2001).

Legislative Background

The Swedish legislation differs from legislations in most other countries in that offenders with mental disorders are never declared not fit to stand trial or not guilty by reason of insanity. They are regarded as accountable for their crimes and given a sanction. However, the type of sanction is dependent on whether or not the offender is found to suffer from a “severe mental disorder” (Swedish Penal Code, chap. 3, section 3). This is a medicolegal concept that refers to type of mental disorder with a distorted interpretation of reality, that is, psychotic states, and to degree of disorder with severe loss of control with serious mental, psychological, and social consequences. Considering degree, an equivalent to the concept of severe mental disorder can be exemplified by depression with pronounced risk of suicide, severe PDs with impulsivity, and serious symptoms of organic or obsessive-compulsive disorders. Thus, a sanction to FPT depends not only on type of diagnosis but also on severity of symptoms.

Offenders who suffered from a severe mental disorder when the crime was committed must not be sentenced to prison (Swedish Penal Code, chap. 30, section 6). Instead he or she is sentenced to FPT if the crime under prosecution is severe enough to warrant a prison sentence and that a court-ordered, standardized, pretrial FPI has ascertained that the offender suffered from such a disorder at the time of the crime(s).

A FPI is requested by the court when, for example, the life history and current behaviour of the offender or the nature and circumstances of the crime throw doubt on the mental health of the person under prosecution (Swedish Code of Justice Procedure, chap. 40, section 1). A prerequisite for a FPI is that the offender has confessed or is tied to the crime by irrefutable evidence. The FPIs are carried out at specialized units run by a state authority (the National Board of Forensic Medicine) and conducted by teams, including a psychiatrist, a psychologist, a social investigator, and ward staff, all with access to information about the social, criminal, and medical histories registered by the health and social services, the prison and probation service, and reported by key collateral informants. All secrecy regulations are legally overruled for FPI purposes. Offenders are normally in custody at the FPI units during up to 4 weeks of examinations, tests, interviews, and observation. The expert opinion of the team is submitted to the court in the form of a written “forensic psychiatric court report,” including a statement whether a “severe mental disorder” was at hand at the time of the crime.

The offenders with mental disorders that do not meet the criteria required to be regarded as “severe” will be sentenced to prison, and those with a severe mental disorder will be sentenced to FPT and transferred to special hospitals or wards. An offender who suffered from a severe mental disorder at the time of the crime but has recovered by the time of the FPI may be sentenced to a noncustodial sanction, such as probation. A noncustodial sanction might also be issued when mitigating circumstances motivate a reduction of punishment for crimes that otherwise would carry sanctions to prison (Swedish Penal Code, chap. 29, section 3).

Conditions during sanctions

A sanction to FPT differs from the other forms of sanction, not least in regulations regarding conditional and absolute discharge from the institution, which are dependent on medical status and court approval. The regulation of periods under supervision after conditional release from prison and under probation follows the same rules issued by the same authority (the National Prison and Probation Services).

The lengths of time to be served inside a locked institution, on conditional leave, and on probation until absolute release are specified in the prison sentence but during FPT, the corresponding timeframes are continuously determined on the basis of clinical assessments and reported to an administrative court, which makes the final decisions. Leaves of absence during inpatient treatment may be permitted as steps in programs for training and adaptation to community living and periods of conditional liberty are interspersed with institutional periods during FPT.

Another difference is the lower threshold for readmittance to hospital versus prison. Conditionally released FPT patients who do not comply with the compulsory regulations may be readmitted to institutional care, whereas revocation of conditional release from prison or probation is enforced only in cases of very serious lack of cooperation, in which case the offenders may be taken back into custody for no more than a fortnight.

Study Population

The study cohort consisted of male offenders referred to pretrial FPI during 1988 to 1995, given at least one diagnosis of mental disorder, living in the city of Gothenburg or the adjacent counties (as previously described in Lund & Forsman, 2005), and were possible to follow-up in detail for 2 years after the index sentence. Women were not considered for inclusion because they represent less than 10% of FPI cases and most probably have gender-specific developmental trajectories of mental disorders and criminal behaviours. The offenders were investigated by means of information in the FPI court reports, official registers, and medical files. The follow-up period started when the index sentences gained legal force and covered the next 2 years, that is, the period when offenders tend to be most apt to relapse in criminality (Bailey & Macculloch, 1992; Monahan, 2001; Yoshikawa et al., 2007). The observation period included different degrees of supervision, as the offenders could be transferred between institutional care, conditional liberty, and absolute liberty or back to the institution again after relapse crimes and new sanctions.

From a total of 349 offenders, 31 were excluded from the follow-up: 7 did not meet criteria for any psychiatric diagnosis, data were missing in another 7, 2 had left the country, and 15 had died during the follow-up period. Finally included were, thus, 318 mentally disordered criminal offenders divided into the following three groups based on type of index sanction: the FPT group (n = 152), the prison group (n = 116), and the noncustodial group (n = 50).

Mean age in the whole study group was 33 years (median 31 years, interquartile range 25-40 years). In total, 23% (73/318) were born outside the Nordic countries, and 82% (267/318) depended on the social services or the health and welfare system for financial support. The majority, 80% (254/318), lived without a regular partner. Previous contacts with the psychiatric health service were common but varied widely in number of visits, admissions, continuity of treatment, and diagnoses.

A total of 86% of the cohort (274/318) were convicted of totally 416 violent index crimes, such as homicide (n = 38), assaults (n = 155), unlawful threats or deprivation of liberty (n = 129), arson (n = 26), robbery (n = 26), rape of an adult (n = 25), and sexual child abuse corresponding to current definitions of rape (n = 17). In total, 14% (44/318) were convicted of nonviolent index crimes, most often substance-related offenses, theft, and fraud, but half of the offenders with nonviolent index crimes had previous convictions for violent crimes, giving a total of 93% (295/318) with at least one violent crime on record. Among the 274 offenders with violent index crimes, 183 had also committed nonviolent crimes.

Data Collection

The forensic psychiatric court reports at the central and local archives of the National Board of Forensic Medicine were scrutinized for information collected during the FPI. Data on FPT during the study period were obtained from the hospital discharge register of the National Board of Health and Welfare and from hospital files. The Central Criminal Records of the National Police Board and the National Council of Crime Prevention provided dates of crimes and convictions, dates of legal force of the index sentence, and periods of sanctions. Data on relapse criminality, including date and specification of crimes and number of convictions, were retrieved from the national registers of crimes and sanctions.

Procedure

Characteristics of the studied individuals were divided into the following: (a) criminal history variables (before the index crime), (b) index variables (circumstances registered in the FPI court reports and in the court sentences), and (c) follow-up variables (outcome in terms of relapse criminality in relation to level of supervision during and after sanctions).

Criminal history variables

The criminal history variables included age at the first registered crime and the total number of convictions for violent crimes and for all crimes before the index sentence. Classified as violent crimes were homicide, assault, threat, robbery, arson, rape, and child abuse corresponding to current definitions of rape.

Index variables

The index variables included age at the index crime, number and percentage of offenders with violent index crimes, psychiatric diagnoses, and homelessness defined as having no permanent address, that is, no fixed abode, and moving around between temporary shelters for at least 1 year before the FPI.

Psychiatric diagnoses were set in accordance with the diagnostic practices at the time of the FPI, which means that the International Classification of Diseases (9th revision; ICD-9; World Health Organization, 1976) alone was used until 1992, after which time diagnoses were assigned in accordance with the Diagnostic and Statistical Manual of Mental Disorders (3rd ed., revised; DSM-III-R; American Psychiatric Association, 1987) as well as the ICD-9. For this study, the DSM-III-R diagnoses obtained from the FPI court reports were used to classify all individuals into the following four main diagnostic categories: (a) psychotic disorders (codes 295, 297, 296, and 298); (b) SADs, including directly related psychiatric complications (codes 291, 292, 303, 304, and 305); (c) PDs (code 301); and (d) “other mental disorders” (OMDs), including a few cases each of mild forms of organic mental disorders and mental retardation, anxiety, depressive reactions, and other maladaptive psychiatric reactions (codes 290, 294, 300, 309, 310, 311, 314, 315, and 317). The DSM-III-R diagnoses assigned at index were reconsidered in relation to the corresponding DSM-IV diagnoses for the present study (American Psychiatric Association, 1994), which showed that none of the diagnostic groups would have lost or gained offenders if the DSM-IV had been used instead of the DSM-III-R. All offenders were also categorized into two mutually exclusive groups: those with diagnoses of SAD, PD, or both (combined or not combined with other diagnoses) and those with neither SAD nor PD diagnoses.

Follow-up variables

The follow-up variables included number of recidivists in violent crimes and in all types of crimes and number of such relapse crimes per offender. The follow-up period was classified into three distinct categories depending on the level of supervision: (a) time inside institution (hospital or prison), (b) time at conditional liberty (keeping up FPT contacts while living in the community, conditional release from prison, or probation), and (c) time at absolute liberty (after discharge from the completed sanction). In each case, all time at each level of supervision was summed up as the offender was followed from one level of supervision to another and back again if reconvicted and having to serve a new sanction during the observation period. The following information was analysed: number of offenders during each specific period, average number of days spent in each period, and number of crimes committed during each period. The incidence rates of the crimes for each such period were calculated. Because periods in custody before starting to serve a prison sentence will automatically reduce the time in prison with the corresponding number of days, four offenders in the prison group actually served no prison time. Eight offenders with index sanctions to FPT had relapsed and were serving new sentences to prison and probation when committing another 21 new crimes, including seven acts of violence. These crimes and periods in new sanctions were omitted from the analyses of incidence rates of crimes for the three different follow-up periods, presented in Table 1, to avoid complicated patterns of mixed sanctions.

Table 1.

Comparisons of Criminological, Index, and Follow-Up Data Between the Three Sanction Groups

	All offenders (N = 318)	FPT group (n = 152)	Prison group (n = 116)	Noncustodial group (n = 50)	p
Criminal history
Age at first registered crime (years)^a	25 (21) 16-29	26 (22)16-31	23 (20) 16-28	21 (21) 16-31	.19
Number of previous convictions^a
Violent crimes	1.3 (1) 0-2	1.1 (0.5) 0-1	1.5 (1) 0-2.8	1.1 (0.5) 0-1	.45
All crimes	4.0 (2) 0-6	3.7 (2) 0-5.8	4.9 (3) 1-7	2.9 (1) 0-3.8	.11
Index data
Number of homeless offenders	164 (52%)	80 (53%)	58 (50%)	26 (52%)	.89
Age at index crime (years)^a	33 (31) 25-40	34 (32) 26-42	33 (32) 25-39	32 (28) 21-41	.21
Number of offenders with violent index crimes	274 (86%)	137 (90%)	97 (84%)	40 (80%)	.12
Number of offenders with diagnoses of
Psychotic disorders	89 (28%)	85 (56%)	3 (3%)	1 (2%)	<.001
SADs	147 (46%)	60 (39%)	64 (55%)	23 (46%)	.038
PDs	195 (61%)	83 (55%)	81 (70%)	31 (62%)	.040
OMDs	98 (31%)	47 (31%)	37 (32%)	14 (28%)	.88
SAD and/or PD	251 (79%)	103 (68%)	104 (90%)	44 (88%)	<.001
Follow-up data
Entire 2-year period
Number of recidivists
Violent crimes	54 (17%)	15 (10%)	25 (22%)	14 (28%)	.003
All crimes	114 (36%)	37 (24%)	52 (45%)	25 (50%)	<.001
Number of relapse crimes/offender
Violent crimes	0.54	0.25	0.75	0.92	.002
All crimes	2.54	1.17	3.64	4.16	<.001
Total time inside institutions		(n = 152)	(n = 112)	(n = 13)
Mean number of days		369	366	123
Number of relapse crimes
Violent crimes		7	7	0
All crimes		16	37	5
Incidence rates of crimes in offenders at risk
Violent crimes		0.03	0.04		.65
All crimes		0.11	0.27	0.37	.55
Total time at conditional liberty		(n = 113)	(n = 92)	(n = 45)
Mean number of days		262	345	638
Number of relapse crimes
Violent crimes		6	75	37
All crimes		40	333	172
Incidence rates of crimes in offenders under risk
Violent crimes		0.10	0.69	0.53	<.001
All crimes		0.48	3.79	2.97	<.001
Total time at absolute liberty		(n = 73)	(n = 56)	(n = 10)
Mean number of days		314	213	483
Number of relapse crimes
Violent crimes		18	5	9
All crimes		101	52	31
Incidence rates of crimes in offenders under risk
Violent crimes		0.36	0.19	0.90	.003
All crimes		2.22	2.48	3.75	.16

Note: FPT = forensic psychiatric treatment; PD = personality disorder; OMD = other mental disorders.

Mean (median) interquartile range.

The variables described above were compared between the three index sanction groups (the FPT, the prison, and the noncustodial groups). The criminal history and index variables were described for the four main diagnostic groups (i.e., psychotic disorders, SAD, PD, and OMD) and compared between the two groups of those with SAD, PD, or both and those without any of these diagnoses. The criminal history and index variables were finally compared between recidivists and nonrecidivists and between recidivists in violent crimes and all other offenders.

Statistical Analyses

For comparisons between the three index sanction groups (FTP, prison, and noncustodial) dichotomous variables were analysed by chi-square test and continuous variables by Kruskal–Wallis Test. Comparisons between two groups were analysed by Fisher’s exact test for dichotomous variables and Mann–Whitney’s U-test for continuous variables. A logistic stepwise regression analysis was performed to find the strongest predictive model for criminal recidivism and violent criminal recidivism. All tests were two-tailed, and the significance level was set at 5%. After the Bonferroni–Holm correction formula was applied to avoid errors due to multiple comparisons, the level of significant difference was corrected to 0.5%. All analyses were performed in the SPSS 17.0 software (further references omitted).

Results

Criminal History and Index Data

Few baseline differences were noted between mentally disordered offenders sentenced to FPT, prison, or noncustodial sanctions (Table 1). They were of similar age at their first registered crimes and did not differ significantly in numbers of convictions for either violent or all types of crime. The three groups were also similar in current age, type of index crime, and extent of homelessness and had committed similarly high frequencies of violent index crimes. As expected, the main difference between the three groups was their patterns of mental disorders, as psychotic disorders were almost entirely seen among those sentenced to FPT, whereas SAD and PD were present across all groups but more common among offenders sentenced to prison (90%) than to FPT (68%).

Recidivism Compared Between Sanction Groups

Relapse crimes during the follow-up were more than more than twice as common in the prison as in the FPT group. In the former, the proportion of recidivists was double that in the FPT group concerning violent crimes and almost double concerning all types of crime (Table 1). To control for age, the data were reanalysed for offenders 45 years or younger with virtually unchanged results (data not shown but available from the author on request). Totally 171 violent relapse crimes were registered: 38 in the FPT group, 87 in the prison group, and 46 in the noncustodial group. The violent recidivism included aggravated assaults and unlawful threats, violence and threats against an officer, rape, and robbery. The total number of all relapse crimes was 808: 178 in the FPT group, 422 in the prison group, and 208 in the noncustodial group.

The incidence rates of crimes per person-year during time inside institution did not differ between the FPT and prison groups. During conditional liberty,the incidence rates in all types of crimes and in violent crimes were lower in the FPT group than in the prison and noncustodial groups (each p< .001; Table1).

In the group of the 251 offenders whose mental disorders included one or both of the diagnoses of SAD and PD, comparisons between the three sanction groups showed a pattern of recidivism that closely resembled that in the entire cohort. Totally, during the follow-up, more crimes of all types were generated by the prison group (4.05 crimes per offender, on average) and in the noncustodial group (4.68 crimes) than in the FPT group (1.50 crimes, p = .001). More violent crimes were also committed by the prison group (0.84 crimes) and the noncustodial group (1.05 crimes) than in the FPT group (0.36 crimes, p = .008). The same pattern of recidivism as in the entire study population was seen during the three periods with different degrees of supervision, that is lower crime rates in the FPT than in the prison and non-custodial groups during time at conditional liberty, and the highest crime rates in the FPT group occurring first after completion of sanction.

Recidivism Compared Between Diagnostic Groups

The offenders in the four main diagnostics groups as well as those with/without SAD and/or PD are described in Table 2. Considerable diagnostic overlaps were seen between the four diagnostic groups as 54% of all offenders met criteria for more than one of the diagnostic definitions. In each diagnostic group, the proportion of offenders with diagnoses from at least one of the other diagnostic groups ranged between 60% and 82%. Totally 146/318 offenders (45%) had only one diagnosis. Among the remaining 172 offenders, 133/318 (42%) had two diagnoses, whereas three diagnoses were seen in 39/318 (12%). This overlap precluded testing of statistical differences between the four diagnostic groups but not between the offenders meeting criteria for SAD and/or PD (n = 251) and those without either SAD or PD (n = 67; Table 2). Compared with the group without SAD or PD, the offenders with SAD and/or PD had index crimes of comparable severity but more pronounced criminal behaviours in every other respect. Their onset of criminality was earlier, they had more previous convictions for any type of crime although they were considerably younger at index, and they had significantly higher recidivism scores (Table 2).

Table 2.

Characteristics of the Four Main Diagnostic Groups and Comparisons Between the Two Groups of Offenders With and Without SAD and/or PD

	Psychotic disorders (n = 89)	SAD (n = 147)	PD (n = 195)	OMD (n = 98)	Either SAD or PD or both (n = 251)	Neither SAD nor PD (n = 67)	p
Criminal history
Age at first registered crime (years)^a	25 (21) 16-30	22 (18) 15-24	23 (19) 16-27	29 (27) 18-34	23 (20 )16-27	31 (28) 19-37	<.001
Previous convictions^a
Violent crimes	1.0 (1) 0-1	1.8 (1) 0-3	1.5 (1) 0-2	1.0 (0) 0-1	1.4 (1) 0-2	0.7 (0) 0-1	.006
All crimes	3.7 (2) 0.5-5.5	5.7 (4) 1-8	4.6 (2) 1-7	3.0 (1) 0-5	4.5 (2) 1-7	2.2 (1) 0-3	<.001
Index data
Number of homeless offenders	50 (56%)	95 (65%)	105 (54%)	41 (42%)	135 (54%)	29 (43%)	.16
Age at index crime (years)^a	33 (31) 26-41	33 (32) 25-40	32 (30) 24-38	36 (35) 28-42	32 (31) 25-39	37 (33) 27-44	.040
Number of offenders with violent index crimes	75 (84%)	127 (86%)	172 (88%)	88 (90%)	218 (87%)	56 (84%)	.61
Index sanctions
FPT	85 (96%)	60 (41%)	83 (43%)	47 (48%)	103 (41%)	49 (73%)	<.001
Prison	3 (3%)	64 (43%)	81 (41%)	37 (38%)	104 (41%)	12 (18%)
Noncustodial sanctions	1 (1%)	23 (16%)	31 (16%)	14 (14%)	44 (18%)	6 (9%)
Follow-up data
Recidivists
Violent crimes	8 (9%)	31 (21%)	44 (23%)	8 (8%)	53 (21%)	1 (2%)	<.001
All crimes	25 (28%)	65 (44%)	79 (41%)	27 (28%)	104 (41%)	10 (15%)	<.001
Number of relapse crimes/offender
Violent crimes	0.26	0.67	0.76	0.28	0.68	0.01	<.001
All crimes	1.26	2.93	3.12	1.56	3.11	0.42	<.001

Note: PD = personality disorder; OMD = other mental disorders; FPT = forensic psychiatric treatment.

Mean (median) interquartile range.

In total, 67 offenders (21%) with a psychotic disorder or an OMD not combined with SAD or PD generated only 28 (3.5%) of the 808 registered relapse crimes. Only one of the totally 171 violent relapse crimes (0.6%) committed in the entire cohort was carried out by one of the 36 offenders (11%) with a psychotic disorder alone (Table 3).

Table 3.

Number of Recidivists and Crimes in Each of the Four Main Diagnostic Groups With and Without Comorbid Diagnoses

	Psychoses (n = 89)		SAD (n = 147)		PD (n = 195)		OMD (n = 98)
	Without comorbidity	With comorbidity	Without comorbidity	With comorbidity	Without comorbidity	With comorbidity	Without comorbidity	With comorbidity
Number of offenders	36	53	27	120	55	140	28	70
All crimes
Number of recidivist	7 (19%)	18 (34%)	12 (44%)	53 (44%)	25 (45%)	54 (39%)	3 (11%)	24 (34%)
Number of crimes	13	99	121	309	258	351	15	138
Number of crimes/offender	0.36	1.87	4.48	2.58	4.69	2.51	0.54	1.97
Number of crimes/recidivist	1.86	5.50	10.08	5.83	10.32	6.50	5.00	5.75
Violent crimes
Number of recidivists	1 (3%)	7 (13%)	5 (19%)	26 (22%)	16 (29%)	28 (20%)		8 (11%)
Number of crimes	1	22	12	86	50	99		27
Number of crimes/offender	0.03	0.42	0.44	0.72	0.91	0.71		0.39
Number of crimes/recidivist	1.00	3.14	2.40	3.31	3.13	3.54		3.38

Note: PD = personality disorder; OMD = other mental disorders.

Comparisons Between Recidivists and Nonrecidivists

Recidivists and nonrecidivists in any crime or specifically in violent crimes are described and statistically compared in Table 4. The recidivists were younger when starting their criminal careers and had more previous sentences (both for any type of crime and for violent crimes). Homelessness was more common among recidivists than among nonrecidivists, as were SAD and/or PD diagnoses.

Table 4.

Comparisons Between Nonrecidivists and Recidivists and Between Violent Recidivists and All Other Offenders

	Nonrecidivists (n = 204)	Recidivists (n = 114)	p	Nonrecidivists + nonviolent recidivists (n = 264)	Violent recidivists (n = 54)	p
Criminal history
Age at first registered crime (years)^a	27 (24) 17-33	21 (18) 15-23	<.001	26 (21) 16-31	21 (19) 15-22	.003
Number of previous convictions^a
Violent crimes	1.0 (0) 0-1	1.8 (1) 0-3	<.001	1.0 (0) 0-1	2.4 (1.5) 0-4	<.001
All crimes	2.9 (1) 0-4	6.1 (4.5) 1-9	<.001	3.6 (1) 0-5	6.0 (4) 1-9	.001
Index data
Number of homeless offenders	93 (46%)	71(62%)	.006	130 (49%)	34 (63%)	.10
Age at index crime (years)^a	34 (32) 26-42	31(30) 24-38	.020	37 (32) 25-41	31 (29) 24-37	.10
Number of offenders with violent index crimes	190 (93%)	84 (74%)	<.001	228 (86%)	46 (85%)	.96

Mean (median) interquartile range.

The variables demonstrating statistical differences between recidivists and nonrecidivists (age at the first registered crime, previous sentences for any crime and for violent crimes, homelessness, and age at the index crime) were entered in a multiple, stepwise logistic regression model. The best model explaining recidivism in any crime was the total number of previous sentences (odds ratio [OR] = 1.15, 95% confidence interval [CI] = [1.09, 1.21], p < .001) and age at the index crime (OR = 0.96, 95% CI = [0.93, 0.98], p = .001). Repeating the analysis for violent recidivism, previous sentences for violent crimes (OR = 1.35, 95% CI = [1.17, 1.55], p < .001) together with age at the index crime (OR = 0.97, 95% CI = [0.93, -0.99], p = .038) were identified as significant predictors. This logistic regression analysis enabled predictive models for recidivism to be derived. In the case of recidivism into any type of crime, the regression model managed to correctly predict 72% of those who actually relapsed, whereas the model for violent recidivism correctly predicted 62% of those who relapsed into violent crimes during follow-up.

Discussion

Criminal recidivism during the 2-year follow-up was significantly less common in the FPT group than in the other two sanction groups. During time at conditional liberty, the incidence rates for violent crimes as well as for any crime were significantly higher in the prison and noncustodial groups than in the FPT group. However, the incidence rates in the FPT group increased after absolute discharge, reaching a level similar to those in the other sanctions. This relapse pattern supports earlier findings pointing in the direction of higher relapse rates among mentally disordered offenders who are discharged into unconditional liberty compared with those who are discharged into mandatory conditional release programs (Bailey & Macculloch, 1992; Wiederanders, 1992) and suggests that the relapse rate is not as dependent on individual characteristics as on level of supervision and contact with supporting/supervising services.

The lower recidivism in the FPT group during the sanction period was also in agreement with the findings of Wiederanders et al. (1997) who concluded that the risk of criminal recidivism was reduced by the risk of having conditional liberty revoked. The lower threshold for readmittance to a closed hospital ward for an undetermined period contrasts with the predetermined periods of prison and probation sanctions, where the consequence of failure to fulfill the obligations associated with conditional liberty will rarely be more than a short period in custody. Further support for the significance of supervision is the finding that most crimes in the FPT group took place after absolute discharge. Davison et al. (1999), who compared criminal recidivism after transfer from special hospitals to less secure mental hospitals, with compulsory outpatient treatment, or with voluntary care after absolute discharge, suggested a relationship between lower recidivism and compulsory treatment.

The lower recidivism in the FPT group before discharge may also be related to the effects of pharmacological treatment, reduction of substance abuse, and psychosocial support efforts in the FPT group. During the 2-year follow-up, recidivism seemed to be postponed until after absolute discharge, which was not the case in the prison and noncustodial groups. This was also observed in a study on violent behaviour in offenders with severe mental illness who were either committed to involuntary outpatient treatment on a regular basis or absolutely discharged, where the former were significantly less violent as a result of improved compliance to medication and reduced substance abuse (Swanson et al., 2000).

The majority of offenders had diagnoses of SAD and/or PD, but this proportion was significantly smaller in the FPT than in the other groups. Excluding all offenders without SAD and/or PD from the analyses did not change the differences in recidivism between the sanction groups or during the different levels of supervision. Thus, the smaller recidivism in the FPT group was not caused by a larger proportion of offenders without SAD and/or PD in this group.

Recidivism was considerably more common in the offenders with SAD and/or PD than in those without these diagnoses. The offenders with SAD and/or PD also had more previous convictions than the other offenders. The finding that they had started their criminal careers at an earlier age than those without any of these diagnoses is consistent with the identification of early-onset conduct disorders as a core problem in the antisocial PD construct.

Offenders with a psychotic disorder without a diagnosis of SAD or PD showed an extremely low relapse rate into criminality, with only one violent relapse crime during the study period. This is contrary to public opinion but in line with an increasing amount of research results, for example, Fazel and coworkers (Fazel, Gulati, Linsell, Geddes, & Grann, 2009; Fazel, Långström, et al., 2009), who in large register-based samples have found that violent criminality among psychotic patients tends to be mediated by comorbid diagnoses of substance use. It thus seems, not least in view of our results, important to shift the focus from the patients with psychotic disorders to those with SAD and/or PD and to develop conditional release programs that meet the needs of this later group.

The total number of previous convictions together with age at index crime was significant predictors of criminal recidivism, supporting the notion that criminological factors are the strongest predictors of relapse criminality. The predictive value of number of previous convictions seemed to be type-specific inthe sense that number of previous convictions for violent crimes predicted violent recidivism, and the total number of previous convictions predicted recidivism in any type of crime. Results for criminal recidivism and violent recidivism similar to those found in this study have been reported in a meta-analysis by Bonta et al. (1998). Homelessness also emerged as more common among recidivists than among nonrecidivists. It is therefore appropriate to conclude that even if early criminological factors are identified as strong predictors, complex patterns of factors contribute to the emergence of a crime, including the current social context and interpersonal interactions.

Limitations

An obvious limitation to the study of outcome after various sanctions is the fact that the law does not permit randomization to different sanctions. The influence of the concept of “severe mental disorder” on criminal recidivism, which in this study distinguished the FPT group from the prison group at the index sentence, must be considered. There might thus be a difference in criminal propensity between the sanction groups, at least between the FPT and the prison groups. The absence of statistical differences in criminal history variables between the FPT and the prison groups at baseline does not mean that the categories are similar in these respects. Another limitation consists of the length of the follow-up period, as a longer observation time might lead to different patterns of recidivism, including more offenders who have completed their sanctions and spent more time at large in the community. Finally, it is also important to consider that recidivistic criminal acts are consequences of a complexity of psychosocial circumstances and interpersonal interactions rather than mere expressions of clinical factors.

Practical Implications

The results of this study indicate that it would be possible to reduce the institutional period of sanctions with legislation that allows for compulsory treatment measures with supervision and high-quality health and supportive measures in the community. It appears especially important to develop this type of programs for the relapse-prone group of patients with SAD and/or PD as they seem to relapse to a lesser extent under conditional release than when they are absolutely discharged. However, the balance between public and patient interest is a challenge for the legal system, having to consider both the legal security of the individual and public protection.

Footnotes

Acknowledgements

The biostatistician Gunnar Ekeroth is gratefully acknowledged for expert statistical advice and Agneta Brimse, BA, is acknowledged for invaluable language assistance. Research time was generously provided by the National Board of Forensic Medicine (T.N.) and the Västra Götaland Region (C.L.).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research and/or authorship of this article.

References

American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Andersen

H. S.

(2004). Mental health in prison populations. A review—with special emphasis on a study of Danish prisoners on remand. Acta Psychiatrica Scandinavica Supplement, 424, 5-59.

Bailey

Macculloch

(1992). Patterns of reconviction in patients discharged directly to the community from a special hospital: Implications for aftercare. Journal of Forensic Psychiatry, 3, 445-461.

Bonta

Law

Hanson

(1998). The prediction of criminal and violent recidivism among mentally disordered offenders: A meta-analysis. Psychological Bulletin, 123, 123-142.

Buchanan

(1998). Criminal conviction after discharge from special (high security) hospital. Incidence in the first ten years. British Journal of Psychiatry, 172, 472-476.

Cloninger

C. R.

Bohman

Sigvardsson

(1981). Inheritance of alcohol abuse: Cross-fostering analysis of adopted men. Archives of General Psychiatry, 38, 861-868.

Council of Europe. (1999). European Sourcebook of Crime and Criminal Justice Statistics. Strasbourg: European Committeee on Crime Problems (CDPC).

Davison

Jamieson

Taylor

P. J.

(1999). Route of discharge for special (high-security) hospital patients with personality disorder. Relationship with reconviction. British Journal of Psychiatry, 175, 224-227.

10.

Fazel

Danesh

(2002). Serious mental disorders in 23000 prisoners: A systematic review of 62 surveys. Lancet, 359, 545-550.

11.

Fazel

Gulati

Linsell

Geddes

J. R.

Grann

(2009). Schizophrenia and violence: Systematic review and meta-analysis. PLoS Medicine, 6, e1000120. doi:10.1371/journal.pmed.100012010.1371/journal.pmed.1000120

12.

Fazel

Långström

Hjern

Grann

Lichtenstein

(2009). Schizophrenia, substance abuse, and violent crime. Journal of the American Medical Association, 301, 2016-2023.

13.

Friendship

McClintock

Rutter

Maden

(1999). Reoffending: Patients discharged from a regional secure unit. Criminal Behaviour Mental Health, 9, 226-236.

14.

Grann

Belfrage

Tengström

(2000). Actuarial assessment of risk for violence: Predictive validity of the VRAG and the historical part of the HCR-20. Criminal Justice and Behaviour, 27, 97-114.

15.

Grann

Fazel

(2004). Substance misuse and violent crimes: Swedish population study. British Medical Journal, 328, 1233-1234.

16.

Hodgins

(1992). Mental disorder, intellectual deficiency, and crime. Evidence from a birth cohort. Archives of General Psychiatry, 49, 476-483.

17.

Jamieson

Taylor

P. J.

(2002). Mental disorder and perceived threat to the public: People who do not return to hospital living. British Journal of Psychiatry, 161, 399-405.

18.

Kreuger

R. F.

Markon

K. E.

Patrick

C. J.

Benning

S. D.

Kramer

M. D.

(2007). Linking antisocial behavior, substance abuse, and personality: An integrative quantitative model of the adult externalizing spectrum. Journal of Abnormal Psychology, 116, 645-666.

19.

Lund

Forsman

(2005). Intended effects and actual outcome of the Forensic Mental Care Act of 1992: A study of 367 cases of forensic psychiatric investigation in Sweden. Nordic Journal of Psychiatry, 59, 381-387.

20.

Maden

Rutter

McClintock

Friendship

Gunn

(1999). Outcome of admission to a medium secure psychiatric unit. Short and long-term outcome. British Journal of Psychiatry, 175, 313-316.

21.

Maden

Scott

Burnett

Lewis

G. H.

Skapinakis

(2004). Offending in psychiatric patients after discharge from medium secure units: Prospective national cohort study. British Medical Journal, 328, 1534.

22.

Manuzza

Klein

R. G.

Moulton

J. L.

(2008). Lifetime criminality among boys with attention deficit hyperactivity disorder: A prospective follow-up study into adulthood using official arrest records. Psychiatry Research, 160, 237-246.

23.

McGreevy

M. A.

Steadman

H. J.

Dvoskin

J. A.

Dollard

(1991). New York state’s system of managing insanity acquittees in the community. Hospital and Community Psychiatry, 42, 512-517.

24.

Monahan

(2001). Rethinking risk assessment. The MacArthur study of mental disorder and violence. New York, NY: Oxford University Press.

25.

Moran

Walsh

Tyrer

Burns

Creed

Fahy

(2003). Impact of personality disorder in violence in psychoses. Report from the UK700 trial. British Journal of Psychiatry, 182, 129-134.

26.

Rättsmedicinalverket (National Board of Forensic Medicine). Årsredovisning 2000 (Annual report 2000). Stockholm: Norstedts.

27.

Rasmussen

Almvik

Levander

(2001). Attention deficit hyperactivity disorder reading disability and personality disorders in a prison population. Journal of the American Academy of Psychiatry and Law, 29, 186-193.

28.

Segessenmann

(2002). International comparisons of recorded violent crime rates for 2000. Wellington, New Zealand: Research and Evaluation Unit, Ministry of Justice.

29.

Steadman

H. J.

Mulvey

E. P.

Monahan

Robbins

P. C.

Appelbaum

P. S.

Grisso

. . . Silver

(1998). Violence by people discharged from acute psychiatric inpatient facilities and by others in the same neighborhoods. Archives of General Psychiatry, 55, 393-401.

30.

Swanson

J. W.

Schwarz

M. S.

Borum

Hiday

V. A.

Wagner

H. R.

Burns

B. J.

(2000). Involuntary out-patient commitment and reduction of violent behaviour in persons with severe mental illness. British Journal of Psychiatry, 176, 324-331.

31.

Swedish National Council for Crime Prevention. (2001). Developments in criminality in Sweden: The aspect of international criminality. Report 2008, 23, 448-451.

32.

Teplin

L. A.

(1990). The prevalence of severe mental disorder among mail urban jail detainees: Comparison with the epidemiological catchment area program. American Journal of Public Health, 80, 663-669.

33.

Torgersen

Kringlen

Cramer

(2001). The prevalence of personality disorders in a community sample. Archives of General Psychiatry, 58, 590-596.

34.

US Department of Justice. Federal Bureau of Investigation. (2002). Crime in the United States 2000. Uniform Crime Reports. Washington D.C.: U.S. Government Printing Office.

35.

Wallace

Mullen

Burgess

Palmer

Ruschena

Brown

(1998). Serious criminal offending and mental disorder. Case linkage study. British Journal of Psychiatry, 174, 477-484.

36.

Walsh

Buchanan

Fahly

(2001). Violence and schizophrenia: Examining the evidence. British Journal of Psychiatry, 180, 490-495.

37.

Walters

G. D.

Knight

R. A.

Grann

Dahle

K. P.

(2008). Incremental validity of the Psychopathy Checklist scores: Predicting release outcome in six samples. Journal of Abnormal Psychology, 117, 396-405.

38.

Warren

L. J.

Mullen

P. E.

Thomas

S. D. M.

Ogloff

J. R. P.

Burgess

P. M.

(2008). Threats to kill: A follow-up study. Psychological Medicine, 38, 599-605.

39.

Wiederanders

M. R.

(1992). Recidivism of disordered offenders who were conditionally vs. unconditionally released. Behavioral Sciences & the Law, 10, 141-148.

40.

Wiederanders

M. R.

Broley

D. L.

Choate

P. A.

(1997). Forensic conditional release programs and outcomes in three states. International Journal of Law and Psychiatry, 20, 249-257.

41.

World Health Organization. (1976). International Classification of Diseases (9th ed.). Geneva, Switzerland: Author.

42.

Yoshikawa

Taylor

P. J.

Yamagami

Okada

Ando

Taruya

. . . Kikuchi

(2007). Violent recidivism among mentally disordered offenders in Japan. Criminal Behaviour and Mental Health, 17, 137-151.