The prevalence of mental illness in the criminal justice system has increased dramatically (
1,
2). In response, mental health courts (MHCs) were created. MHCs are a type of court diversion program, designed to decrease recidivism and increase engagement in community treatment among offenders with mental illness (
3,
4).
A defining feature of MHCs is the provision of mental health treatment in lieu of penalizing options to reduce recidivism (
5–
7). Although research has consistently found that MHC participants have more positive criminal justice outcomes (for example, fewer arrests or jail days) compared with similar individuals not in MHCs (
7–
11), it is unclear what leads to these successful outcomes. One assumption is that MHCs reduce recidivism through provision of outpatient mental health services, such as medication management, individual and group therapy, and intensive case management (
5,
7,
12). However, despite the importance placed on the assumed link between community treatment and recidivism, research on this relationship is rare (
13,
14).
To our knowledge, only three MHC studies have examined the trajectory from community treatment to recidivism. First, Herinckx and colleagues (
15) found that in one MHC, participants who were rearrested were more than twice as likely as those who were not rearrested to have had an inpatient psychiatric admission in the year before MHC entry. Second, Steadman and colleagues (
7) found that the absence of treatment in the six months before MHC entry was strongly associated with more arrests and jail days after entry. Treatment was defined as outpatient treatment, case management, and medication services. Third, Keator and colleagues (
16) found no impact of community treatment (that is, medication services, crisis intervention, inpatient psychiatric services, individual or group therapy, case management, and other supportive services) on the number of arrests or jail days among MHC participants. The latter two studies included a dichotomous treatment variable (yes-no). However, a number of treatment-related variables (for example, treatment type and compliance) may have an impact on future crime. Moreover, previous studies have often lumped together treatment received in the community, jail, and prison, when the goal of MHCs is to increase community-based treatment.
Outside the MHC context, research has examined the role of treatment factors (for example, compliance, perceived voluntariness, and motivation) on criminal justice outcomes. In general, negative relationships have been found, such that increased treatment leads to decreased recidivism. For example, a meta-analysis of 114 studies found that treatment attrition among offenders led to increased recidivism (
17). Furthermore, a study of drug courts, which are problem-solving courts like MHCs, found that community treatment and multisystemic therapy led to decreases in substance use, criminal behavior, and psychiatric symptoms (
18).
Treatment motivation and perceptions have also been examined in relation to rearrests. One meta-analysis of 129 studies concluded that voluntary treatment significantly reduced recidivism, whereas mandated treatment did not (
19). However, within mandated treatment programs, perceptions of treatment as less voluntary have been found to be associated with higher rates of recidivism (
20). Some studies have found that outcomes for persons under legal pressure are better than outcomes for those not under legal pressure (
21). For example, Junginger and colleagues (
22) found that offenders with severe mental illness under involuntary outpatient commitment had fewer arrests than offenders not under mandatory treatment and that the association between mandatory treatment and reduced arrest was mediated by an increase in mental health treatment. Further, Levenson and Macgowan (
23) found that participants with higher levels of engagement (for example, attendance) were more likely to progress in group therapy and that prisoners with higher treatment motivation were more likely to receive aftercare treatment, which, in turn, significantly reduced recidivism (
24). Finally, one study that included approximately 800 drug court participants concluded that self-reported motivation positively predicted future program completion, which was negatively associated with future recidivism (
25).
This study compared treatment outcomes between two study groups (MHC and treatment as usual), examining the impact of community treatment on recidivism at the time of court entry or arrest and six months later between and within groups. Treatment examined in this study refers to mental health and substance abuse treatment services and not to services targeting criminogenic factors. Three hypotheses were tested. First, MHC participants will show increases in treatment compliance, utilization, motivation, and perceived voluntariness during the postenrollment six-month period. In contrast, the treatment-as-usual group will show decreases in or remain the same across these factors over the same period. Second, when analyses control for between-group differences, community treatment outcomes (for example, increased compliance) over time will be better for MHC participants than for treatment-as-usual participants. Third, improvements in treatment-related variables factors will be associated with fewer arrests for both groups, although the effect of community treatment will be larger for the MHC group than for the treatment-as-usual group.
To address these hypotheses, data collected in the MacArthur MHC study (
7,
26) were analyzed. The MacArthur MHC study was a prospective, longitudinal, quasi-experimental four-site study that was the first to include multiple sites and experimental (MHC) and comparison groups.
Methods
Participants
Participants in the MacArthur MHC study came from four courts: San Francisco County (N=143) and Santa Clara County (N=241) in California, Hennepin County in Minnesota (N=186), and Marion County in Indiana (N=171). The MHC sample included participants who enrolled in the program during the study period (2005–2008). The treatment-as-usual sample included recently arrested individuals with mental health problems. To be eligible for the treatment-as-usual sample, the person could not have been referred to or rejected from one of the MHCs. In this study, only participants who completed both the pre- and postenrollment interviews were included: MHC (N=357) and treatment as usual (N=384).
Measures
Recidivism.
Rearrest (excluding warrants and violations) was measured as the number of times the person was arrested in the postenrollment six-month period. Arrest data were obtained from the Federal Bureau of Investigation (the National Incident-Based Reporting System). Most participants had one or no arrests in the postenrollment six-month window (88%–91% depending on the sample). Overall, 479 participants (65% of both samples) were not arrested in the postenrollment period.
Treatment compliance.
Treatment compliance was assessed with two self-report measures: appointment and medication compliance. Appointment compliance was measured on a scale from 1 to 5 (1, avoided keeping appointment; 5, never missed an appointment). Medication compliance was measured on a scale from 0 to 5 (0, never took medication; 5, never missed taking medication). Both were measured for the pre- and postenrollment periods.
Treatment perceptions.
Treatment perceptions included perceived voluntariness and motivation. Perceived voluntariness was measured by the MacArthur Perceived Coercion Scale (MPCS), which has been adapted for use in MHCs (
27,
28). The MPCS includes eight statements (for example, “I felt free to do what I wanted about going to treatment”) measured on a scale from 1 to 5 (1, strongly disagree; 5, strongly agree), with higher scores indicating more choice about treatment. Treatment motivation was measured with the Treatment Motivation Questionnaire (TMQ) (
29), which asks about reasons for entering treatment and feelings about current treatment and has been used with involuntary treatment recipients (
29,
30). The TMQ includes two series of statements with the following opening stems: “I came for treatment because [3 statements]”; and “If I remain in treatment it will probably be because [3 statements].” Also included are four questions about how true each statement is for the participant (for example, “I came to treatment now because I was under pressure to come”). Participants were asked to rate all statements from 1, not at all true, to 7, very true. To create one treatment motivation score, answers to the ten statements were summed.
Treatment use.
Self-reported treatment use included mental health and substance abuse services received in the community. Jail and prison treatment were excluded. Participants were asked how many nights (if an overnight setting, such as a hospital) or times (if a nonovernight setting, such as outpatient treatment) they received services in the pre- and postenrollment periods. Participants also reported the types of services received, including individual counseling, group therapy, couples-marital-family therapy, intensive case management, other case management, peer support or self-help services, day treatment, short and specific classes, and medication management.
Days in the community.
In both interviews, participants were asked to report residence information (description and number of days). All days that reflected time in the community (that is, not incarcerated or hospitalized) were summed. Because MHCs divert offenders from incarceration, days in the community for the MHC and treatment-as-usual groups were expected—and found—to differ (142.81 and 95.22 days, respectively). To account for different amounts of treatment as a result of between-group differences in days in the community, treatment use was weighted by multiplying by the percentage of days in the community in the six-month period after entry.
Analytic Strategy
We first conducted paired t tests to compare pre- and postenrollment differences in treatment utilization and perceptions within each group. Then we ran a fixed-effects regression model to determine the impact of the intervention (MHC) on arrests in the postenrollment period. Fixed-effects regression, the ideal method to control for selection bias between study groups and times (
31,
32), handles issues of omitted-variables bias and provides an estimation of the long‐term effect between groups (
33). Omitted variables were non–time-varying variables, including demographic characteristics (race-ethnicity, gender, and age) and criminogenic factors (family history, childhood experience, and age at first arrest) (
32). This model was particularly useful to measure the impact of treatment, after controlling for time effects; variables not included in the analysis; and selection bias. As mentioned, 65% of both samples had no new arrests. Because an ordinary Poisson model cannot handle excessive zeros (
31,
34), we implemented a negative binomial model, which is designed to handle counter variables with many zeros (
31). The negative binomial model defined unobserved heterogeneity with the purpose of distinguishing between individuals who were not arrested and those who were arrested.
Results
As shown in
Table 1, compared with the treatment-as-usual sample, the MHC sample had significantly smaller proportions of Hispanics and persons with depression and a significantly larger proportion of persons with schizophrenia.
Table 2 presents data on the type of treatment received during the pre- and postenrollment periods. For mental health services in the treatment-as-usual group, significant increases were noted over time for receipt of short and specific classes and medication management. For substance abuse services in the treatment-as-usual group, the use of peer support or self-help services and specific classes also significantly increased. In the MHC group, significant increases were noted for all types of mental health services except couples-marital-family therapy. For substance abuse services in the MHC group, use of five of the nine types of services increased significantly.
Treatment-Related Variables and Arrests
During the postenrollment period in the treatment-as-usual group, significant reductions were seen in the number of arrests and the number of days in the community, and significant increases were seen in appointment compliance and use of mental health and substance abuse services (
Table 3). For the MHC group, the number of arrests and perceptions of treatment voluntariness significantly decreased, and significant increases were seen in medication compliance and use of mental health and substance abuse services. In the postenrollment period, both groups reported similar amounts of mental health services (treatment as usual, 61.26 times or nights; MHC, 60.35 times or nights).
Fixed-effects model
Table 4 shows time and group differences between the pre- and postenrollment periods for each treatment variable. Appointment compliance, but not medication compliance, meaningfully increased; however, an interaction by sample group was not found. For treatment perceptions, the MHC group had increased treatment motivation over time but decreased perceptions of voluntariness, compared with the treatment-as-usual group. On the whole, the entire study sample received more mental health services but fewer substance abuse services in the postenrollment period. However, when the time and group interaction term was included, MHC participants received more mental health services and substance abuse services over time compared with treatment-as-usual participants.
Table 5 presents data on the impact of community treatment on arrest for the two groups, after the analysis controlled for days in the community. For the treatment-as-usual group, none of the treatment variables were associated with rearrests. For the MHC group, however, increased medication compliance and increased use of mental health services were associated with a significant decrease in arrests postenrollment.
Discussion
We examined relationships between community treatment and criminal recidivism among MHC participants and offenders processed through the traditional court system. We expected the MHC sample, but not the treatment-as-usual sample, to experience significant increases in treatment compliance, motivation, perceived voluntariness, and utilization over time. Furthermore, we expected the MHC sample to demonstrate more robust improvements in treatment outcomes over time, compared with the treatment-as-usual sample. We found that both groups improved in the postenrollment period, such that their treatment compliance and use increased and arrests decreased. After the analysis controlled for a priori time and group differences, only appointment compliance significantly increased over time, but a between-group difference was not found.
In the six months after MHC entry, MHC participants were found to perceive their treatment as significantly less voluntary, whereas perceptions of voluntariness among treatment-as-usual participants did not change significantly over time. Similarly, in the fixed-effects analysis, the treatment-as-usual group had higher levels of perceived treatment voluntariness, compared with the MHC group. This finding is not surprising, because MHCs mandate treatment (
14). Some previous research has found that higher levels of perceived coercion lead to better treatment retention over time (
35). Mandatory treatment has been found to lead to significant reductions in recidivism, compared with treatment that is not mandatory (
22,
36).
Use of treatment increased for both groups over time. However, after the analysis weighted treatment use by days in the community and controlled for pre-enrollment differences, the MHC sample received significantly more community mental health and substance abuse services than the treatment-as-usual sample. This finding indicates that MHC participants received more treatment than the treatment-as-usual sample. An MHC goal is to increase community treatment, and this study found evidence in support of that goal. These results align with those of Cosden and colleagues (
37), who found that the MHC group received more treatment than the traditional court group.
Our third hypothesis was that increases in treatment use, treatment motivation, and perceptions of treatment as voluntary would be associated with a lower likelihood of arrest. This hypothesis was partially supported. We found that increases in medication compliance and mental health service use were associated with significant reductions in the likelihood of arrests in the MHC sample. But in the treatment-as-usual group, treatment-related variables were not associated with rearrests. Of note, both groups received a similar amount of mental health services during the postenrollment period, but treatment use was found to be associated with arrests only for the MHC sample. This finding suggests that treatment itself may not lead to meaningful criminal justice outcomes (for example, reduced arrests) but that treatment combined with court monitoring decreases arrests for offenders with mental illness. Furthermore, as indicated, there is evidence that mandated treatment leads to improved outcomes, such as reductions in arrests (
36,
38). The “black robe effect” (
39) of judges shuttling hard-to-treat offenders with mental illness into treatment may play a key role in reducing recidivism. Thus the lack of a significant treatment-arrest relationship for the treatment-as-usual group may be explained by the fact that treatment was a requirement for the MHC group but likely not a requirement for the treatment-as-usual group.
Few studies have examined the link between MHC treatment and future arrest. Steadman and colleagues (
7) found that MHC participants who had no treatment six months prior to program enrollment had higher recidivism rates 18 months after entering the program. In our study, the impact of treatment on recidivism was examined for both the MHC and the treatment-as-usual groups, which enabled us to differentiate effects. In the few previous studies that have examined predictors of recidivism, the findings are mixed regarding the link between treatment and future arrest. Keator and colleagues (
16) found that community treatment did not have an impact on the number of future arrests or jail days for MHC participants. Herinckx and colleagues (
15) found that MHC participants with higher recidivism rates were more likely to have received inpatient mental health services 12 months before MHC enrollment, but they found no association for receipt of inpatient services after MHC enrollment. Unlike past studies, our study also controlled for the number of days in the community, which is critical because the treatment-as-usual participants had significantly fewer days in the community and thus fewer opportunities for community treatment and arrest.
By testing the models developed, we were able to provide a clearer understanding of possible underlying treatment-related factors influencing recidivism. However, this study is one of the first empirical examinations of the impact of community treatment on recidivism in MHCs. Therefore, study results are only suggestive and in need of replication. Generalizations to all MHCs in the United States may not be warranted given that our study included four jurisdictions. In addition, although our study aimed to investigate relationships between community treatment and recidivism, other variables of importance, such as community environment, socioeconomic factors, and demographic factors, should be considered in future investigations. Similarly, we examined only mental health and substance use service use. Other forms of treatment (for example, treatment targeted at criminogenic factors) should also be studied. Our examination extended to only six months postenrollment, a period in which some of the MHC participants were active court clients. Future studies should examine relationships between treatment and recidivism after MHC participation has ended. Finally, we relied on self-report measures to test our hypotheses. Self-reports have known limitations, including errors related to memory and social desirability. We encourage future researchers to examine objective measures of treatment use in the community.
Conclusions
MHCs are alternative court systems that attempt to address the treatment needs of offenders with mental illness. MHCs adopt the idea that engagement in community treatment will lead to lower recidivism. A key finding of this study was the significant impact of mental health treatment on arrests for MHC participants only. To our knowledge, this is the first study to examine the trajectory from treatment (beyond simple receipt of treatment) to recidivism in MHCs. Increased attention to treatment factors in MHCs will play an important role not only in linking participants with mental health services but also in decreasing future arrests.
Acknowledgments
The authors thank Henry J. Steadman, Ph.D., and Policy Research Associates, as well as John Monahan, Ph.D., and the John D. and Catherine T. MacArthur Foundation Network on Community Mandated Treatment for generously supporting the original research.