Criminogenic Risk and Mental Health: A Complicated Relationship

The relationship between criminogenic risk and mental illness in justice involved persons with mental illness (PMI) is complex and poorly understood. As previously noted, Reference Bartholomew and Morgan ¹ the general public is misinformed on the nature of this relationship, erroneously believing that mental illness causes violence and crime. This perception is compounded by news reports immediately speculating about mental illness in response to sensationalized criminal acts such as mass shootings, as well as in popular and social media. Of greater concern, however, is when clinicians, administrators, and policy makers are also misinformed. Criminal risk includes static (eg, age, gender) and dynamic (eg, antisocial attitudes, substance misuse) factors that place an individual at greater risk of involvement in crime. Criminogenic risk, on the other hand, refers to dynamic risk factors that directly contribute to criminal behavior and therefore need to be a focus of intervention (eg, need to reduce antisocial attitudes, need to reduce substance misuse). Known criminogenic risk factors provide treatment targets for professional service providers and decision and policy makers aiming to reduce crime.

Criminogenic risk is well understood in correctional settings and often guides the nature of rehabilitative services.Reference Bonta and Andrews ² Criminogenic risk is less understood in mental health circles, however, and even when working with justice involved PMI, mental health professionals often remain uninformed on the necessity of including criminogenic risk factors in mental health treatments. In fact, mental health professionals working with incarcerated PMI have historically emphasized mental health recovery (eg, reduce symptomatology, medication management) over, and at times to the exclusion of, criminogenic risk.Reference Bewley and Morgan ^{3 ,} Reference Morgan, Flora, Kroner, Mills, Varghese and Steffan ⁴ This is particularly problematic given evidence that targeting criminogenic risk reduces criminal involvement, including for PMI (see Reference Bonta and Andrews ⁵ for a thorough review). Notably, it is the first author’s experience that when consulting with mental health professionals in correctional and forensic settings, there remains a belief that PMI are involved in the criminal justice system due to lack of adequate mental health care, and that mental illness is the primary culprit when it comes to these individual’s criminal behavior. However, research data clearly refutes this position such that this naivety and lack of understanding of criminogenic risk is inexcusable.

Researchers have clearly demonstrated the link of criminal risk with criminal outcomes for PMI. In fact, a body of literature now supports the conclusion that mental illness is not the driving force behind PMI’s criminal justice involvement, but that criminogenic risk (dynamic risk factors commonly associated with criminal activity) is likely the primary cause of crime, similar to criminal justice populations that do not have mental illness (see for example ^6-8 ). In other words, it is now well understood that the relationship of mental illness to crime is weak and that other factors, including criminogenic factors, better account for crime.Reference Skeem, Winter, Kennealy, Louden, Tatar and Joseph ⁹ Most compelling in this line of investigation were the results of a meta-analysis showing that traditional criminogenic risk factors (eg, antisocial attitudes, antisocial associates, substance misuse) were better predictors of criminal justice involvement (ie, recidivism) than were traditional clinical (ie, mental health) factors for justice involved PMI.Reference Bonta, Law and Hanson ¹⁰ In fact, results demonstrated that the variables that best predicted recidivism for PMI were essentially the same as the variables that predicted recidivism for offenders without mental illness. Specifically, this body of research demonstrates that justice involved PMI exhibit criminal risk factors similarly to nonmentally ill criminal justice populations across a variety of criminal justice settings and populations, including prison inmates, Reference Morgan, Fisher, Duan, Mandracchia and Murray ^{11 –} Reference Wolff, Morgan, Shi, Fisher and Huening ¹³ young jail inmates, Reference Wilson, Farkas and Ishler ¹⁴ forensic psychiatric patientsReference Bolaños, Mitchell and Morgan ¹⁵ and justice involved PMI hospitalized in acute inpatient psychiatric units.Reference Bolaños, Mitchell and Morgan ^{15 ,} Reference Gross and Morgan ¹⁶ Nevertheless, the traditional approach of targeting psychiatric stabilization remains the predominant service model used for justice involved PMIReference Skeem, Manchak and Peterson ¹⁷ without concomitant services targeting their criminal risk or criminalness. Thus, it is not surprising that traditional psychiatric services have had limited impact on criminal justice outcomes.Reference Hodgins, Müller-Isberner and Freese ¹⁸ To be effective, services for justice involved PMI must address the co-occurring issues of mental illness and criminal risk.Reference Draine, Salzer, Culhane and Hadley ^{7 ,} Reference Morgan, Fisher, Duan, Mandracchia and Murray ^{11 ,} Reference Hodgins, Müller-Isberner and Freese ¹⁸

Although it is clear that criminogenic risk places PMI at risk for criminal justice involvement, we still do not know the nature of this criminogenic-mental illness relationship. Importantly, then, we also do not know how this relationship impacts treatment needs, and of ultimate concern, what this relationship means in terms of individual and societal outcomes. Bartholomew & MorganReference Bartholomew and Morgan ¹ proposed that “mental illness and criminalness feed each other in a continuous loop” (p. 5) as depicted in Figure 1. If Figure 1 accurately reflects the relationship between mental illness and criminalness, there would be a complex directionality such that mental illness and criminalness are independent (untreated mental illness results in increased psychiatric recidivism and untreated criminalness results in increased criminal recidivism), but also multi-directionality such that untreated mental illness results in increased criminal risk, and vice versa. In this latter relationship, for example, a PMI experiencing decompensation in terms of mental health functioning (eg, increased symptomatology) will experience reciprocal decompensation in their cognitive and behavioral controls that allow them to manage their criminal risk.

Figure 1.

Directionality of mental illness and criminalness.

From the Risk-Need-Responsivity (RNR) model, the criminalness-mental illness relationship, as depicted in Figure 1, suggests that mental illness may be a responsivity factor for crime, but of equal importance is that criminalness may be a responsivity issue for mental health functioning. The nature of this relationship and the conceptualization of mental illness and criminalness as risk or responsivity factors will, in our opinion, guide the next wave of research. Such research will have practical implications beyond our understanding of this relation with the potential to significantly alter how we approach the issue of mental illness across the criminal justice landscape, and similarly, how we address criminal justice involvement across the mental health landscape. Even as we write this narrative discussing both criminal justice and mental health systems—structured as separate entities in the United States—we recognize that such research may help break the artificial silos that now exist, which put justice involved PMI in a criminal justice or mental health system vacuum.

Applying the Model of Comorbid Mental Illness and Criminalness in Mental Health Settings

Although criminalness is typically considered in the context of corrections, it is relevant in community mental health settings as well. For example, in a sample of 79 men and 63 women in an acute inpatient psychiatric unit, more than half (n = 74, 52.1%) endorsed previously being convicted of a crime (misdemeanor and/or felony).Reference Bolaños, Mitchell and Morgan ¹⁵ This finding is consistent with the rate of criminal justice involvement in a similar sample of psychiatric inpatients from another acute inpatient psychiatric unit where among 61 patients (32 male, 29 female), 55.7% (n = 34) reported they had been convicted of a crime.Reference Scanlon, Morgan and Mitchell ¹⁹ Compared to inpatient samples, the prevalence of prior convictions in outpatient mental health care samples appears to be a little less, with one study showing a prevalence of about 36%.Reference Theriot, Dulmus and Sowers ²⁰

Beyond the prevalence of criminal justice involvement in mental health samples, there is a need to further examine the nature of the mental illness–criminalness relation in community-based psychiatric patients. One important variable linked to criminalness is antisocial attitudes, which are attitudes or dispositions that violate the social norm, including criminal and defiant behavior. Antisocial attitudes or cognitions are considered a primary risk factor for predicting crime. In fact, antisocial cognition is one of the “Central Eight,”Reference Bonta and Andrews ⁵ which consists of the strongest empirically validated criminal risk factors. Importantly for purposes of this discussion, antisocial attitudes are prevalent in justice involved persons with and without mental illness alike.Reference Morgan, Fisher, Duan, Mandracchia and Murray ¹¹ Thus, this provides an important variable from which to explore the nature of the mental illness–criminalness relationship.

Among a cross-sectional sample of 61 inpatients mentioned above, scores on the Psychological Inventory of Criminal Thinking StylesReference Walters ²¹ (PICTS) General Criminal Thinking scale, an overall self-report measure of criminal thinking, was significantly associated with a number of the self-reported Millon Clinical Multiaxial Inventory-III Reference Millon, Millon and Davis ²² (MCMI) scales of psychiatric syndromes, including the clusters of clinical syndromes scale (Anxiety, Somatoform, Bipolar, Dysthymic, Alcohol, Drug, and Post-Traumatic Stress disorder scales; n = 61, β = 0.513, p < 0.001, R ² _adj = 0.251), severe clinical syndromes scale (Thought Disorder, Major Depression, and Delusional Disorder scales; adjusting for sex, n = 61, β = 0.442, p < 0.001, R ² _adj = 0.208), Borderline scale (adjusting for sex, n = 60, β = 0.477, p < 0.001, R ² _adj = 0.244), Antisocial scale (adjusting for relationship status, n = 49, β = 0.732, p < 0.001, R ² _adj = 0.517), and an average of the combined raw MCMI scales which was used as a holistic measure of mental health symptomology (n = 61, β = 0.604, p < 0.001, R ² _adj = 0.354).Reference Scanlon, Morgan and Mitchell ¹⁹ Furthermore, the General Criminal Thinking scale scores accounted for 20.8 to 51.7% of the variance in these MCMI scales. This suggests that the relation between criminal thinking and mental health symptomology is an appreciable one for justice-involved PMI, and that criminal thinking may be contributing to a variety of psychiatric symptoms, thereby interfering with mental health treatment. It should be noted, however, that due to the cross-sectional nature of this data, it is not feasible to make causal or temporal conclusions regarding the relationship of these constructs.

Although criminal thinking, as measured by the PICTS, was a significant predictor of symptoms associated with mental illness, criminal risk was not statistically significantly associated with self-reported length of psychiatric hospitalization (a behavioral indicator of the patients’ mental health functioning) in a subsample of this clinical population. Specifically, neither the PICTS General Criminal Thinking scale (n = 19, β = 0.071, p = 0.773, R ² _adj = –0.054), nor the Criminal Sentiments Scale–ModifiedReference Shields and Simourd ²³ (CSSM) Total Score (n = 16, β = 0.068 p = 0.803, R ² _adj = –0.066) were statistically significantly associated with current length of hospitalization in the psychiatric facility. In other words, antisocial cognitions were related to psychiatric symptoms, but not length of hospital stay. This finding suggests that criminalness is a responsivity factor for mental illness given the obtained relationship with psychiatric symptomatology, but not for psychiatric recovery as criminalness was not predictive of length of hospitalization. Given these are correlational findings from cross-sectional data, more research is needed to explore the nature of these relationships, including temporality and causation.

We further examined the mental illness–criminalness relationship by studying the relationship between psychiatric functioning and institutional misconduct. Specifically, we assessed the extent to which severe psychiatric symptomology (ie, mania and thought disorders), criminal attitudes, and the interaction between the two were related to both violent and nonviolent disciplinary infractions among an incarcerated population (n = 265). Among violent infractions, main effects of gender (β = 2.49, p = < 0.0001), criminal attitudes (β = 0.02, p = <0.0001), symptoms of mania (β = 0.05, p = 0.001), and thought disorder (β = –0.04, p = 0.01) were noted. A model that included all possible interactions between mania, thought disorder, and criminal attitudes was also examined. None of the interactions were significant; however, all previously mentioned main effects remained significantly related to violent misconduct. When looking at nonviolent infractions, gender (β = 1.52, p < 0.0001) and criminal attitudes (β = 0.01, p = 0.04) were statistically significant contributors to the model; however, the practical significance of criminal attitudes was negligible. Neither mania (p = 0.70) nor thought disorder (p = 0.37) were significantly related to nonviolent infractions.Reference Van Horn, Morgan and Grabowski ²⁴ Results from this study again suggest that criminalness and mental illness play a role in disruptive behavior. Furthermore, given that mental illness and criminal attitudes were not related to nonviolent infractions, it is possible that as the behavior becomes more antisocial, both mental illness and criminalness factors should be addressed.Reference Van Horn, Morgan and Grabowski ²⁴ As this was a retrospective study, conclusions regarding the causal nature of these relationships cannot be drawn; however, the results support the hypothesis that criminalness is a responsivity factor when addressing mental health concerns.

Collectively if these findings are supported by further research, they suggest that criminalness is a vital, albeit overlooked factor in the treatment of PMI in community mental health settings—many of whom have a history of criminal justice involvement. Furthermore, these findings highlight that criminogenic risk is not only important for criminal risk outcomes, but that it is also entangled in mental health treatment, making psychiatric symptoms more problematic (ie, criminalness is a responsivity issue for PMI). In other words, the presence of criminalness (a broader concept than criminogenic risk) can negatively impact not only an individual’s psychiatric symptoms, but also motivation, commitment, and sustained efforts toward recovery. For example, a PMI high in antisocial attitudes may find self-medication for symptom management more convenient with more pleasurable psychoactive effects (and fewer negative side-effects) than working through the complexities and bureaucracy of the community mental health system to receive their psychotropic medications. Therefore, in order to improve outcomes in justice involved persons with mental illness, it is essential to attend to criminogenic risk not only in criminal justice settings (eg, corrections), but also in both forensic and nonforensic psychiatric settings. Given the conclusion that mental illness and criminalness have an interactive and reciprocal relationship, these results and the resulting model ( Figure 1) have significant treatment implications; however, it is first relevant to discuss the assessment of criminalness broadly, but also specifically as it pertains to psychiatric patients.

Using Risk Assessments to Inform Psychiatric Treatment Planning

Risk assessments serve multiple purposes when providing services to criminal justice involved persons, with the most common being to inform security and release decisions. Initially, risk classification was based on clinical judgments (eg, “gut feelings”) that were primarily informed by unstructured interviews with the justice involved individual. Overall, these judgments of risk lacked accuracy, and better methods were needed.Reference Dawes, Faust and Meehl ^{25 ,} Reference Monahan ²⁶ The second generation of risk assessments improved upon its predecessor by including structured assessments; however, they were not theoretically informed and mainly assessed for risk factors that were static and unchangeable (eg, prior criminal history, type of offense).Reference Andrews, Bonta and Wormith ²⁷ Although this generation of assessments improved upon the first generation, they were only marginally useful in predicting future criminal behavior given their focus on unchangeable risks.Reference Bonta and Andrews ⁵ The third generation of risk assessments continued to implement structured assessments but expanded to include dynamic risk factors that could be addressed in an attempt to improve risk estimates. Finally, the fourth generation of risk assessment has attempted to address the difficulty of translating the information gleaned from these assessments to appropriate use for classification and treatment by incorporating elements of case management.Reference Bonta and Andrews ⁵

Although the use of risk assessment instruments has increased exponentially over the last decade, there remains considerable debate regarding how to most effectively communicate their results.Reference Mills ²⁸ One of the more contentious topics centers around whether actuarial (ie, statistical method of estimating risk using risk factors with known probabilities) or structured professional judgment (SPJ) instruments (ie, incorporates actuarial information but estimate of risk is at the discretion of the assessor) yield more accurate predictions; however, an exploration of the current issues surrounding violence risk assessment called attention to two key misunderstandings in this debate.Reference Mills ²⁸ First, the assessment of risk is a process and the administration of one actuarial instrument cannot be considered a stand-alone risk assessment.Reference Mills ²⁸ Second, people, not instruments, make decisions regarding risk and a risk assessment that does not address all known risk factors, whether through actuarial or SPJ methods, is neither comprehensive nor sufficient.Reference Mills ²⁸

There are other notable issues regarding the communication of risk assessment results. Some risk assessment results are communicated in “categories” of risk (ie, high, moderate, low); however, studies have shown that clinicians’ interpretation of these terms vary widely.Reference Mills ²⁸ Other risk assessment results are communicated in terms of an individual’s likelihood of reoffending. Studies have shown that individuals, including highly educated professionals and clinicians, tend to neglect base rate information when making individual predictions whether the information is presented categorically or probabilistically.Reference Mills and Kroner ^{29 ,} Reference Walters, Kroner, DeMatteo and Locklair ³⁰ Furthermore, some research has found that inclusion of risk relevant information in narrative form can lead clinicians to overestimate risk, despite this information having been included in the estimated probability of re-offense.Reference Hilton, Harris, Rawson and Beach ³¹ Collectively then, the accuracy of risk assessments means little if the results are not communicated clearly, interpreted accurately, Reference Mills ²⁸ and effectively used to inform treatment for justice involved PMI in both criminal justice and mental health settings.

A variety of measures are available for the prediction of risk. The Level of Service Inventory-Revised (LSI-R) is the most commonly used risk assessment worldwide with over one million official administrations in the year 2010 alone.Reference Wormith ³² A summary of the meta-analytic studies of the LSI-R indicated that its average predictive criterion validity for predicting general recidivism is 0.36, and 0.25 for predicting violent recidivism.Reference Andrews, Bonta and Wormith ²⁷ Other research has indicated LSI-R scores are also predictive of halfway house success, parole success, self-reported criminal activity (as opposed to new arrests or convictions), and institutional misconduct.Reference Andrews, Bonta and Wormith ³³ The LSI-R and its fourth-generation revision, the Level of Service/Case Management Inventory (LS/CMI), have been shown to be predictive of reoffending in a wide range of justice involved persons, including those with severe mental illnessReference Ferguson, Ogloff and Thomson ³⁴ and drug offenses.Reference Kelly and Welsh ³⁵

A large-scale meta-analysis of other commonly used risk assessments found overall moderate predictive validity for violent recidivism for the Historical, Clinical and Risk Management Violence Risk Assessment Scheme (HCR-20; r = 0.25), the Psychopathy Checklist-Revised (PCL-R; r = 0.24), and the Violence Risk Appraisal Guide (VRAG; r = 0.27).Reference Campbell, French and Gendreau ³⁶ Some differences among these measures were noted in terms of predicting institutional violence, with the HCR-20 evidencing the strongest predictive validity (r = 0.31) and the PCL-R evidencing the weakest (r = 0.15).Reference Campbell, French and Gendreau ³⁶ An examination of violence prediction in a sample of psychiatrically institutionalized justice involved persons also found that the HCR-20 and PCL-R were significant predictors of both verbal and physical aggression.Reference Gray, Hill, McGleish, Timmons, MacCulloch and Snowden ³⁷ Interestingly, findings also indicated the Brief Psychiatric Rating Scale evidenced a strong relationship with physical and verbal aggression in this population.Reference Gray, Hill, McGleish, Timmons, MacCulloch and Snowden ³⁷

Clearly, risk assessments can provide invaluable information pertaining to security and release decisions; however, information regarding risk should also be used to guide treatment.Reference Andrews and Dowden ³⁸ Measures that include an examination of dynamic risk factors can inform treatment planning by helping clinicians understand important issues to address in treatment that will ultimately reduce risk and subsequently improve public safety.

The LS/CMIReference Andrews, Bonta and Wormith ³³ is an excellent example of a theoretically driven risk assessment instrument that can function both as a tool to inform decision-makers regarding risk of re-offense, and a tool to ensure that the intensity and targets of treatment are appropriate. Research has continually shown that the individuals who need the most intensive services are those with a higher probability of reoffending.Reference Hilton, Harris, Rawson and Beach ^{31 ,} Reference Andrews and Dowden ^{38 ,} Reference Lowenkamp, Latessa and Holsinger ³⁹ Even more importantly, research has also shown that providing intensive services to low-risk individuals can increase their risk of re-offending.Reference Andrews and Dowden ³⁸ The LS/CMI functions as a risk assessment tool by examining an individual’s risk on each of the Central Eight risk factors and providing a total score that is associated with a risk category (ie, low, moderate, high). It functions as a treatment planning tool when clinicians use the information learned about criminogenic risk to prioritize treatment goals. For example, an individual with high scores on the antisocial attitudes and employment subscales would best be served by cognitive behavioral therapy to address criminal thinking, as well as vocational training and assistance with job placement.

Another theoretically grounded, comprehensive risk assessment instrument that was designed to address treatment needs is the Violence Risk Scale (VRS).Reference Wong and Gordon ⁴⁰ The VRS assess 6 static and 20 dynamic risk factors and includes an assessment of the individual’s treatment readiness on each dynamic risk factor, as indicated by their stage of change (for an in-depth discussion, see Reference Reference Prochaska and DiClemente41). The inclusion of treatment readiness allows the clinician to choose appropriate interventions for an individual’s readiness to change, and when used multiple times during treatment also allows the VRS to function as a measure of treatment progress.Reference Wong and Gordon ⁴⁰ The VRS evidences good predictive validity, with AUC values ranging from 0.73 (p < 0.001) for any re-offense at one-year follow-up, to 0.74 (p < 0.001) for any re-offense at three-year follow-up.Reference Wong and Gordon ⁴⁰

Although several risk assessments can be used to inform treatment planning to address criminal risk factors, there is no widely available assessment that address both criminogenic and psychiatric risk simultaneously. Currently, treatment providers must use a battery of assessments to identify the co-occurring needs of justice-involved PMI. The Services Matching InstrumentReference Morgan, Kroner, Mills and Olafsson ⁴² (SMI) is being developed to meet this need. The SMI is intended as an integrated, theoretically grounded treatment planning tool designed to identify empirically identified risk factors for both criminal (eg, criminal history, criminal associates) and psychiatric recidivism (eg, psychiatric symptomology, social support). A preliminary examination of the psychometric properties of the SMI suggests the measure evidences good test re-test reliability (r = 0.78 to 0.91), concurrent validity (r = 0.41 to 0.67), and internal consistency (α = 0.77 to 0.94), Reference Olafsson, Morgan and Kroner ⁴³ with additional evaluation of the measures psychometric properties in progress.

The assessment of risk has evolved significantly in the last several decades and can be an invaluable tool in improving public safety; however, risk assessment instruments are under-utilized in terms of improving criminal rehabilitation. Unfortunately, these instruments fail to capture how mental illness may affect the presentation of criminogenic risk factors, and vice versa. More work is needed to develop measures that integrate these needs for the growing population of justice-involved PMI, and to examine the efficacy and accuracy of how risk is interpreted by clinicians and communicated to decision-makers. In addition, more research is needed to better understand the interaction of criminalness and severity of mental illness symptomology such that treatment needs can be more fully conceptualized and prioritized.

Treating Criminalness and Mental Illness

The lead author has written about treating justice involved PMI and the necessity of targeting mental illness and criminalness as co-occurring problems;Reference Morgan, Flora, Kroner, Mills, Varghese and Steffan ^{4 ,} Reference Morgan, Fisher, Duan, Mandracchia and Murray ^{11 ,} Reference Morgan, Kroner, Mills, Batastini, Weiner and Otto ⁴⁴ however, these discussions have not fully captured the complexity of the criminalness–mental illness relationship. It is easy to think of criminalness and mental illness as univariate constructs with both impacting unique outcomes of interest (criminalness impacting criminal recidivism and mental illness impacting psychiatric recidivism) and we often develop treatments accordingly. For example, recent efforts to modify Cognitive Behavioral Therapy programs aimed at reducing criminal risk (eg, Moral Reconation Therapy Reference Little and Robinson ⁴⁵ ) to also address issues of mental illness, Reference Rotter and Carr ⁴⁶ are done in this vein. But simply adding mental health-focused treatment strategies or interventions to programs designed to treat criminalness, and vice versa, does not account for the complexity of the criminalness–mental illness relationship and risks missing the treatment mark.

It is commonly understood in health and behavioral health circles that the more problem areas introduced into an individual’s life (eg, mental illness, criminalness, substance abuse, physical disability), the more problematic (and typically negative) the outcomes.Reference Goodell, Druss, Walker and Mat ⁴⁷ There are many examples of this in educational, medical, psychiatric, and criminal justice literatures, but one example highlights this concern specifically for justice involved PMI.

Research examining barriers to employment has demonstrated that a criminal record Reference Holzer, Raphael and Stoll ^{48 –} Reference Varghese, Hardin, Bauer and Morgan ⁵⁰ and mental illness ^51-53 are significant barriers to employment. Stigma is a serious concern when it comes to employment for those involved in the justice system (see Reference Reference Varghese, Hardin, Bauer and Morgan50 for a thorough review); however, most concerning for justice involved PMI is that these barriers are compounded when both mental illness and criminal history are present. In other words, employers appear least likely to hire individuals with both a criminal record and mental illness, compared to those with just a criminal record or PMI who are not justice involvedReference Batastini, Bolanos and Morgan ^{54 ,} Reference Graffam, Shinkfield and Hardcastle ⁵⁵ ; multiple areas of concern result in poorer outcomes.

Given these findings across a range of disciplines, it is not surprising to find suggestions for treating co-occurring disorders in a unified treatment protocol (see for example Reference Allen, McHugh, Barlow and Barlow ⁵⁶ ) or a combined/interdisciplinary approach (see Reference Reference Aaronson, Katzman, Gorman, Nathan and Gorman57). It was in this vein that Changing Lives and Changing Outcomes Reference Morgan, Kroner and Mills ⁵⁸ (CLCO Footnote ¹ ) was developed. CLCO is a comprehensive program systematically developed to meet the unique treatment needs of individuals with co-occurring mental health and criminogenic concerns. The aim of this program is not to cure mental illness or criminalness, but rather to maximize adaptive behaviors to optimize functioning while reducing psychiatric relapse and criminal reoffending. Although CLCO is rooted in cognitive-behavioral theory, with a significant psychoeducational component outlined in a comprehensive and structured treatment manual, the uniqueness of the program is in the integration of best mental health practices and best correctional rehabilitation practices throughout the therapeutic modules. In other words, consistent with Figure 1, the aim of CLCO is to treat not only the unique features of mental illness and the unique criminogenic needs of justice involved PMI, but to also treat these problems as integrated, with complex ties to both criminal justice and mental health outcomes of interest. For example, one of the CLCO modules entitled “Mental Illness and Criminalness Awareness” is not limited to understanding mental illness. Rather, it also incorporates awareness of one’s criminal proclivity and how problems in one area (ie, mental illness or criminalness) can increase problems in the other domain. Similarly, the Medication Adherence module not only emphasizes the role of psychotropic medication in recovery from mental illness, but also as a responsible life choice that is consistent with achieving one’s prosocial life goals (ie, a better choice in regard to criminal justice outcomes than, for example, “self-medicating” with illicit substances).

We use CLCO not as an example of the best treatment program available for justice involved PMI, but rather as an example of how treatments—regardless of one’s treatment of choice—need to integrate treatments of mental health and criminogenic risk in treatment work with justice involved PMI. As the research summarized in this paper demonstrates, this means going beyond merely adding a mental health focus to an already effective correctional program (eg, Thinking for a Change, Moral Reconation Therapy). Instead, the field should aim to develop and incorporate treatments that are meaningfully integrated, to better meet the need of, address, and account for the complex, interwoven relation of mental illness and criminalness in correctional and psychiatric settings alike.

Although these treatment recommendations historically have stronger ties in criminal justice settings, these recommendations are also reflective of changes needed in forensic and nonforensic psychiatric settings. In forensic settings, for example, treatments for competency restoration and insanity should not be limited to psychiatric recovery. This may appear less relevant in the legal context of competency to stand trial, where the emphasis is typically on abatement of symptoms that impact one’s ability to factually and rationally understand proceedings or to assist in one’s defense. Although criminal risk is not predictive of competency restoration outcomes, Reference Colwell and Gianesini ⁵⁹ as Gowensmith and colleaguesReference Gowensmith, Frost, Speelman and Therson ⁶⁰ alluded, failure to integrate a holistic perspective that includes an individual’s criminogenic risk presents a missed opportunity. Although competency may be restored with a univariate focus on mental health functioning, sustained competency, further psychiatric recovery efforts, and reduced criminal and psychiatric recidivism outcomes are possibly missed. In nonforensic settings, it is common to encounter a significant number of patients with justice involvement (ie, approximately 50% of patients) such that psychiatric recovery (see for example, Illness Management and Recovery Reference Gingerich and Mueser ⁶¹ ) needs to also account for the criminalness–mental illness relationship. At the very least, it appears that criminalness is a responsivity factor for mental illness. Therefore, given the impact of criminogenic risk factors on psychiatric symptomatology, recovery efforts that do not account for criminogenic risk will fall short of treatment goals.