Individual Community-Based Treatment of Offenders With Mental Illness

Deinstitutionalization and Mental Illness

This overrepresentation and increase in SMI cases within the criminal justice system may be associated with the deinstitutionalization of mentally ill individuals from state/province-based psychiatric facilities. The 1960s, 1970s, and early 1980s marked a period of mentally ill individuals being released from state hospitals, asylums, and other mental health institutions into community-based mental health care systems (Searight & Handal, 1986-1987). The community was unable to cope with the influx of mentally disordered individuals into society due to lack of preparedness, inadequate financial support, and an unstructured community-based mental health care system. Of course, it is also possible that with an increased sensitivity to the issue of SMI, better detection of such groups may also have contributed to increased incidence observed in more recent studies. However, even a cursory review of the literature reveals that SMI is a very significant concern within correctional environments.

Constantine et al. (2010) examined arrest trajectories among 3,769 offenders suffering from SMI over a 4-year period. Results indicated that arrest rates among offenders with SMI were unusually high as compared with arrest rates of offenders in general. The authors also found that these arrest patterns were established in adolescence and persisted over time, suggesting that successful interventions must address offender needs early in their criminal careers.

Douglas, Guy, and Hart (2009) provided a comprehensive meta-analysis related to whether psychosis represents a risk factor for violence. A total of 885 effect sizes (odds ratios) were calculated or estimated from 204 studies. Results indicated that psychosis was significantly associated with a 49% to 68% increase in the odds of violence. The average effect size for psychosis was similar to that obtained by others for well-known risk factors of recidivism (e.g., history of violence).

In addition to increased arrest rates and failure on supervision, current research has found that SMI is also linked to treatment program attrition. A recent meta-analysis by Olver, Stockdale, and Wormith (2011) examined offenders with different types of SMI and their treatment outcome. The authors found that offenders suffering from SMI were less likely to complete treatment programs. In turn, program attrition predicted recidivism, as offenders who did not complete programming were significantly more likely to reoffend, suggesting that SMI indirectly influences recidivism by way of impeding the ability of the offender to complete programming.

Forensic Assertive Community Treatment (FACT) and Forensic Intensive Case Management (FICM) Programs

Recent research into treatment for offenders presenting with SMI include FACT and FICM interventions. These models are extended versions of the original Assertive Community Treatment (ACT) Model, a service delivery model designed to provide mental health treatment to consumers based on their individual needs by professionals from a variety of disciplines, combining treatment, rehabilitation, and support services (Morrissey, Meyer, & Cuddeback, 2007). The ACT team determines the individual’s needs and provides treatment for as long as needed, as opposed to having a structured time frame like other treatment models. ACT was intended to provide community-based treatment that would offer the elements of hospital treatment, including 24/7 crisis response and support, and intensive psychopharmacological treatment, all while aiding in the improvement of community living skills with the aim of reducing hospitalization and improving symptoms (Morrissey et al., 2007). In their review of the evidence base for ACT, Morrissey et al. (2007) found that although ACT programs were successful in reducing rates of psychiatric hospitalization and improving housing stability, they were not successful in reducing rates of arrest or incarceration, leading to the conclusion that specific interventions are needed to target criminal behavior (see also Skeem et al., 2011, for a discussion).

FICM programs were developed in response to the concern over costly FACT programs. FICM uses case managers with individual caseloads and does not use a contained, multidisciplinary team for each offender. These programs also do not directly provide psychiatric treatment, but instead help offenders obtain services. Morrissey and colleagues reviewed the limited research on FICM programs, and found similar results to that of their review on FACT programs. Their review indicated that there were inconsistent results in the few studies on FICM and a weak evidence base for producing positive results at a reduced cost (Morrissey et al., 2007).

The Integrated Risk–Need–Responsivity (RNR-I) Model

Recently our team has developed an alternative model for the treatment of offenders with SMI (Abracen & Looman, in press; Looman & Abracen, 2013a). One of the core assumptions of this model is that both issues associated with criminogenic need (see Andrews & Bonta, 1998, 2010, for a discussion) and SMI need to be incorporated into the contemporary management of groups of moderate- and high-risk offenders. We have also argued that, given the frequent histories of physical, emotional, and sexual abuse among such groups of offenders, issues associated with what has been called complex trauma (see Courtois & Ford, 2009) need to be included in any comprehensive model of offender treatment.

A number of studies conducted by our team on groups of high-risk, high-need sexual offenders attest to the utility of such comprehensive (though time-limited) approaches (see Abracen & Looman, 2004, in press; Abracen et al., 2011; Abracen, Looman, & Langton, 2008; Looman & Abracen, 2013a,; Looman, Abracen & Di Fazio, 2014). At the Regional Treatment Centre Sex Offender Treatment Program (RTCSOTP), we have incorporated both individual- and group-based treatments with a focus on both criminogenic need (e.g., criminal personality, associates, and attitudes) and mental health issues (e.g., management of negative emotionality and symptom management).

Although similar to the Risk–Need–Responsivity (RNR) model espoused by Andrews and Bonta (1998, 2010), the RNR-I model provides a more integrated approach to the management of high-risk, high-need offender populations and suggests ways in which SMI/complex trauma interact with criminogenic needs. The RNR-I model argues that treatment must be contextualized for individual offenders. We have included a basic outline of the model in the appendix.

In community settings, it may not be feasible to offer a full-time program including both individual- and group-based components. The aim of the present study is to determine whether individual treatment is related to significant reductions in recidivism. The approach adopted by staff in the Central District (Ontario) psychology department is in keeping with the tenants of the RNR-I model. It was these staff and contract psychologists, under the authority of the first author, who provided individual treatment to offenders included in the present investigation. The aim of the present investigation was to extend our research on the RNR-I model with a group of high-risk, high-need offenders in a community setting. It was hypothesized that those offenders receiving higher dosages of treatment (defined as more individual therapy sessions) would recidivate at significantly lower rates in comparison with offenders living in the same Community Correctional Centre (CCC) but who received no treatment or who were only assessed.

Method

Results

Logistic regression analyses were used to investigate the relationship of the three treatment levels on recidivism while controlling for actuarially assessed risk (GSIR). Regression analysis was used to provide odds ratios to illustrate the difference among treatment levels and recidivism. This sample of offenders was pre-selected to be at high risk based on their conditions of residency at a CCC on release to the community. This was illustrated by the mean GSIR score of the sample at −6.66 (SD = 7.63), which corresponds to the second highest category of risk on the GSIR: moderate/high risk. First, our measure accounting for actuarially assessed risk (GSIR) was found to be a significant predictor of recidivism (p < .01). GSIR accounted for almost 15% of the variance and predicted with meaningful accuracy. These results are illustrated in Table 1.

OPEN IN VIEWER

Table 1.

Results of Analysis on GSIR as a Significant Predictor of Recidivism.

	B	SE	Significance	Exp(B)	Classification Accuracy (%)	Nagelkerke’s R2
GSIR	−0.098	.029	.001	0.907	63.5	.148
Constant	−0.294	.272	.279	0.745

Note. GSIR = General Statistical Information on Recidivism.

Logistic regression analysis of the psychological treatment data revealed highly significant results. In a model containing GSIR, having had moderate amounts of psychological treatment was associated with an offender being 7.7 times less likely to reoffend compared with having had no psychological treatment at all (p < .01). Even more significant were the findings of having had a higher dosage of psychological treatment, which was associated with an offender being 11.6 times less likely to reoffend, compared with having had no psychological treatment at all (p < .001). This model had significant predictive power and accounted for 39.5% of the variance, as demonstrated below in Table 2.

OPEN IN VIEWER

Table 2.

Statistical Results of Psychological Treatment Data in a Model Containing GSIR.

	B	SE	Significance	Exp(B)	Classification Accuracy (%)	Nagelkerke’s R2
GSIR	−0.091	.033	.006	0.913
No treatment					74.8	.395
Moderate treatment	−2.039	.587	.001	0.130
High treatment	−2.455	.605	.000	0.086
Constant	0.641	.351	.068	1.898

Note. GSIR = General Statistical Information on Recidivism.

Discussion

As noted above, there is mounting evidence that rates of SMI among offenders housed in correctional environments have been rising for at least the last two decades, and that this trend has been associated with the closing of provincial or state-based psychiatric facilities in the community. A previous study by our team (Abracen et al., 2014) indicated that there were very high rates of mental illness among offenders housed at the Keele CCC in the Ontario region. As well, we have previously demonstrated that, across all CCCs operated by CSC, there have been substantial increases in the rates of mental illness from at least 1998 onward (Abracen et al., 2012).

Recently, we have demonstrated that for a group of high-risk sexual offenders presenting with either SMI and/or cognitive impairment, individual therapy may be as effective at reducing rates of recidivism as was a full treatment program (consisting of both individual and group therapies; Looman et al. (2014). This raises the interesting possibility that for certain groups of high-risk, high-need populations, individual therapy may be more cost-effective than group-based treatment.

In keeping with the findings of Looman et al. (2014), the present data highlight the utility of individual therapy at meeting the needs of a group of high-risk, high-need offenders presenting with SMI. The vast majority of the clients treated presented as comorbid for a variety of Axis I and Axis II disorders. It is interesting that a seemingly arbitrary criterion (number of individual therapy sessions attended) with no consideration of such responsivity issues as motivation, whether the client dropped out of therapy or the therapy was terminated, should produce such striking findings as those observed in the current study.

The main finding of the present investigation is that the amount of individual therapy that clients received was directly linked to outcome as measured by officially recorded recidivism. Clients who received more than 20 hr of individual therapy with psychological staff were approximately 12 times less likely to recidivate. It should be emphasized that this finding applied after actuarial assessment of risk was statistically accounted for as measured by the GSIR.

Although these data need to be replicated with a larger sample, the results of the present study highlight the potential utility of the RNR-I model at least with groups of moderate- to high-risk offenders. Both staff and contract psychologists who delivered individual therapy for the present sample were familiar with the elements described in this model, and identified treatment targets listed in the RNR-I when addressing the treatment concerns of the clients. There were also frequent consultations with the first author regarding the choice of treatment targets and the direction of therapy with individual clients.

It should also be emphasized that clients living at the Keele CCC attended regular meetings with parole officers. Their behavior was closely monitored and clients were aware that violations of their release conditions could result in re-incarceration. It is undoubtedly true that the close monitoring of clients living at the CCC contributed to the outcome among offenders attending treatment. Perhaps one advantage of attending treatment was that clients could gain perspective regarding the restrictions placed on their freedom, as the vast majority of clients presented as resistant to this type of intervention.

It might be argued that we should also have accounted for prior psychiatric diagnosis among the treated and comparison samples. Nonetheless, as we had argued in our earlier investigation (Abracen et al., 2014), we believe that virtually all the individuals housed at the Keele CCC would meet the diagnostic criteria as listed in Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association, 2000) for antisocial personality disorder (APD) and that the vast majority would meet criteria for at least one substance abuse/dependence condition. However, in that study a very conservative methodology was adopted for the coding of psychiatric diagnoses and, as a result, some clients who likely met criteria for APD were not coded as having the disorder. If coders could not find reference to an actual diagnosis made in a psychological or psychiatric report authored by a registered mental health professional for the 5-year period prior to being housed at the Keele CCC, the client was not coded as having a disorder. This certainly yielded a lower rate of recorded mental illness than would actually have been the case if all offenders had been interviewed for the investigation. As this study was retrospective in nature, it was not possible to conduct such assessment interviews on all subjects included in the study.

Skeem and her colleagues have argued (e.g., Louden & Skeem, 2013; Skeem et al., 2011) that mental health treatment does not reduce the risk of recidivism. This view is in keeping with Andrews and Bonta’s RNR model discussed above. As evidence for this perspective, Skeem and her colleagues discuss the available research regarding such programs as FACT and FICM approaches. As noted above, there is limited evidence in favor of such programs. Other reviews (e.g., Looman & Abracen, 2013a) have essentially arrived at the same conclusion and have argued that approaches that integrate both forensic and mental health care are more likely to meet with success. Nonetheless, few models have been proposed that integrate these approaches into a comprehensive model of care, and none, to the best of our knowledge, have been demonstrated to significantly reduce rates of recidivism.

We argue that issues related to mental illness generally, and complex trauma specifically (see Courtois & Ford, 2009, for a discussion), need to be incorporated into models of contemporary treatment. The “I” in the RNR-I model stands for an integrated approach to treatment. This model has been developed by our team over many years and has been incorporated into our work in the community as well as the institutional programs that we have run with groups of high-risk offenders. We have previously demonstrated the efficacy of these approaches with reference to our institutionally based treatment programs (see Abracen & Looman, 2004; Abracen et al., 2008, for reviews; Looman et al., 2014).

Although the model was developed for work with sexual offenders, we believe that the model would be equally efficacious with other groups of offenders. In fact, only a minority of offenders included in the present sample had a history of sexual offending. The present sample included 25 sex offenders, 19 of whom completed some community treatment. For offenders without a history of sexual offending, certain aspects of the model would not require clinical attention (e.g., deviant arousal); nonetheless, many elements incorporated into the model would be equally applicable to groups of violent non-sexual offenders. We believe that this model provides a concise illustration of factors that need to be assessed when working with groups of moderate- or high-risk offenders. It is designed to be heuristic and focuses on matters that have been found to be associated with recidivism in the empirical literature. Furthermore, we believe that psychometrically sound instruments have been developed to assess each of the domains included in the model.

The outline and organization of the model is based on earlier work published by Beech and Ward (2004), though the elements included in the model differ from their approach. We, nonetheless, agree that their model, with a focus on both proximal and distal factors, is quite relevant in practice. We would add, however, that the various factors included in the model may act synergistically. The way in which various risk factors may work to increase risk when present at the same time has not received much attention in the literature, and we believe that this needs to be addressed in greater detail. For example, it has been our experience that many of the clients who we treat present with issues related to complex trauma. These experiences (e.g., repeated physical/emotional abuse) are likely related to the development of substance abuse and relationship difficulties. We believe that comprehensive approaches to treatment must address both presenting difficulties and at least having treatment staff aware of the role that complex trauma may have played in the development of these other conditions. In short, we are arguing for a trauma-informed perspective when working with such groups of clients. From this perspective, issues associated with the therapeutic alliance and the pace of treatment become very germane clinical matters. Many of these clients will require attention by mental health professionals well versed in matters related to complex trauma and the therapeutic alliance on one hand, and concrete approaches associated with working with specific forensic matters (e.g., criminal thinking and associates) on the other.

We are also of the opinion that for groups of high-risk, high-need offenders, a period of inpatient/institutional placement may be required. This applies both to initial institutional placement and to when such groups are first released to the community. With reference to community placement, an institution such as a CCC allows not only for the gradual transition to independent living but also for close supervision of such groups of clients. It may be that such models as FACT and FICM have produced mixed results as they do not provide for both a heuristic model that treatment professionals can use in working with such groups of clients and specifically incorporating policy regarding a gradual transition to independent living. We believe that the RNR-I combined with the case management approach advocated above may result in demonstrably lower recidivism rates among high-risk, high-need groups of offenders.