Assessing the Potential Efficacy of an Intervention for Incarcerated People With Mental Illness

This study reports results from an RCT that was the second phase of a two-part study. This RCT included a small-scale pilot study of a newly adapted intervention. The original study design planned for four intervention cycles. However, two cycles were halted before the experimental treatment was successfully initiated, the first time because of administrative issues within the prison system and the second time because of COVID-19. Consequently, an intent-to-treat analysis was used for the two successfully initiated intervention cycles, which took place between January 2019 and December 2020. The study was approved by the institutional review board of the University of North Carolina at Chapel Hill. Written informed consent was obtained from all participants.

The experimental intervention had two components: T4C, version 3 (9)—an existing evidence-based, criminogenically focused intervention—and the TSDA (8). To support our goal of assessing the potential efficacy of the newly adapted intervention, study team members with at least a master’s degree in social work or an equivalent degree delivered the intervention. T4C is a highly structured, 25-session, group-based cognitive-behavioral therapy program, delivered in a closed format to 10–12 people up to twice a week across 14 weeks. This intervention comprises three interrelated modules: social skills, cognitive restructuring activities, and a problem-solving method for managing interpersonal conflict. The TSDA was used in conjunction with T4C to compensate for the neurocognitive and social impairments that impeded participants’ engagement with the intervention content (which remained unchanged).

The control condition included standard prison treatment and programming. In this study, standard prison mental health services included crisis services focused on addressing suicidal thoughts and behaviors and other mental health symptoms, psychiatric evaluations, consultations, medication management, and individual counseling to help stabilize mental illness so that a person could function appropriately in the prison environment. Study participants in both the experimental and control groups were eligible to receive standard prison treatment and programming. Prison staff made all decisions regarding individuals’ eligibility and the allocation of the services and programming offered by the prison.

Forty-seven people participated in the two intervention cycles, which took place in a men’s prison facility in the southeastern United States. At the beginning of each recruitment cycle, correctional staff provided the research team with a list of all potentially eligible participants at the facility, and study staff completed all recruitment activities. A description of recruitment efforts for this study has been reported (12).

Study inclusion criteria required that individuals be ages 18 years or older and have a mental illness (schizophrenia, schizoaffective disorder, or other psychotic disorder; bipolar disorder; or major depression), moderate to high criminogenic risk, and ≥1 year remaining in their prison sentence. Exclusion criteria included intellectual or developmental disability, restrictions that precluded involvement in group gathering spaces, and participation in T4C during the previous 6 months.

Study data were collected through four face-to-face interviews (screening, baseline, and 3 and 6 months after baseline) and a review of administrative records at 9 months after baseline. Study eligibility was determined through a screening interview, during which sociodemographic information was collected via participant self-report. An envelope method was used to randomly allocate eligible participants immediately following study enrollment. This random assignment method comprised consecutively numbered sealed envelopes with security linings to conceal group assignment from study staff and participants. Each envelope was assigned a study condition according to a randomization table. Study staff distributed envelopes in sequential order, noting participants’ study IDs on the outside of each envelope. The contents of the envelopes were then reconciled with the randomization table to ensure the integrity of the randomization process. All study interviews were conducted by research staff, who had a master’s degree in social work or an equivalent degree and were trained in the administration of all study measures by the study’s principal investigator (A.B.W.), a licensed mental health professional.

Mental health diagnoses were assessed by using the Mini-International Neuropsychiatric Interview (MINI) (13), a short, structured diagnostic interview that was developed and standardized on the basis of the Composite International Diagnostic Interview. The version of the MINI administered at study screening was informed by the DSM-5 and ICD-10.

Criminogenic risk level was assessed by using the Level of Service/Case Management Inventory during the screening interview. This tool is one of the world’s most frequently used and reviewed correctional assessment instruments.

Level of aggression was measured at baseline and at 3 months (i.e., treatment completion) by using the 12-item Buss-Perry Aggression Questionnaire–Short Form. Items on this questionnaire are scored on a 6-point Likert scale, with possible total scores ranging from 12 to 72 and higher scores reflecting more aggression (14).

A dichotomous variable was created by using prison administrative data to indicate receipt of at least one behavioral infraction for disciplinary reasons during the 6-month posttreatment period (i.e., between 3 and 9 months after baseline). Prison administrative data were used to calculate the cumulative number of days each participant spent in administrative segregation during the 6 months after completion of treatment.

Treatment targets were measured at baseline and 3 months (i.e., treatment completion). Interpersonal problem solving was measured via the 52-item Social Problem-Solving Inventory–Revised: Long (SPSI-R:L). Responses are recorded with a 5-point Likert scale. Raw scores for the entire scale were computed with the SPSI-R:L raw score derivation sheet. Total SPSI-R:L raw scores range from 0 to 20, with higher scores corresponding to more positive social problem-solving skills (15).

Attitudes were measured via the 46-item part B of the Measures of Criminal Attitudes and Associates instrument, which assesses attitudes and beliefs associated with criminal behavior. Responses are recorded in an agree versus disagree format (agree=1 and disagree=0, except for seven items that are reverse-coded). Scores range from 0 to 46, with higher scores indicating more supportive attitudes about crime (16).

Impulsivity was measured via the 30-item Barratt Impulsiveness Scale. Items are scored on a 4-point Likert scale, with scores ranging from 30 to 120. Higher scores indicate greater impulsiveness (17).

All analyses were conducted in SAS, version 9.4. First, chi-square tests were used to assess differences in behavioral infractions between study conditions (experimental vs. control) during the 6-month posttreatment period, and then nonparametric Wilcoxon rank sum tests were used to assess differences in median number of days spent in administrative segregation. Next, differences in treatment targets and study outcomes at treatment completion were examined with linear mixed-effects models via “PROC MIXED” in SAS. These models were used to examine change from baseline to postintervention in estimated mean scores within and between the experimental and control groups. Each model included fixed effects for group, time, and a group × time interaction term, with a random intercept specified for each participant. The group × time interaction term represented the treatment effect.

The 47 men participating in the study had a mean±SD age of 41.6±9.9 years and a mean length of incarceration of 12.4±7.7 years. Thirteen participants identified as Black or African American (28%), 17 identified as White (36%), and 17 (36%) identified as more than one race, other, or unknown. Approximately one in 10 participants also identified as Hispanic. In terms of psychiatric diagnoses, 13 (28%) participants had a diagnosis of schizophrenia, schizoaffective disorder, or other psychotic disorder; 15 (32%) had bipolar disorder; and 19 (40%) had major depression. Finally, 37 (79%) participants had a “high” or “very high” criminogenic risk score, and 28 (60%) were incarcerated for a violent offense.

Table 1 shows the results of the linear mixed-effects models at treatment completion. The first section of the table demonstrates that the changes in estimated mean score for participants in the experimental group were all statistically significant and showed a trend in the desired direction. The third section of the table shows the treatment effects, which can be interpreted as the difference in the average change for each measure between the experimental and control conditions from baseline to 3 months. Whereas participants in the experimental condition (N=24) experienced a statistically significant decline in impulsivity (7.61 units) from baseline to 3 months, participants in the control group (N=23) had a nonsignificant average decline of 0.51 units. The associated treatment effect of −7.10 units was statistically significant. Additional analyses of the two remaining study outcomes, administrative segregation and behavioral infractions, during the 6 months following treatment completion found no statistically significant differences by study condition.

The results demonstrated that people with mental illness can engage in and benefit from existing evidence-based criminogenically focused interventions, which offer the potential to produce positive impacts in a number of domains (e.g., impulsivity). These findings underscore the need for continued and accelerated research in this area through large-scale randomized explanatory trials.

Of note, the participants in the experimental group experienced statistically significant changes in all three of the intervention’s key treatment targets, which represent dynamic aspects of criminogenic risk factors. These findings demonstrated that these treatment targets can be effectively influenced among people with mental illness. In addition, because prior research has found that reduced impulsivity can lead to substantial reduction in criminal legal recidivism (2), our findings provide a promising avenue for future research.

We found no differences in the number of behavioral infractions or days in administrative segregation between the experimental and control groups, despite prior research demonstrating that interventions such as T4C can positively affect these outcomes. These findings suggest the need for further research on these outcomes among incarcerated people with mental illness, with an emphasis on how they relate to criminal legal outcomes after release.

The goal of this small-scale study was to explore the effects of an adapted intervention on treatment targets and outcomes to determine whether larger-scale studies are warranted. Given the preliminary nature of this research, several limitations deserve consideration. First, although two cycles of the intervention were successfully completed, the original research design was not implemented as intended. Second, the use of treatment as usual as the control condition calls for caution in applying the findings to other treatment settings. Third, this research could not assess whether the T4C was more effective with the TSDA or alone. In addition, instances of behavioral infractions and time in administrative segregation may have reflected factors beyond the participants’ control, and the connection between these outcomes and community-based recidivism requires further research. The delivery of the intervention by study staff also may have affected the outcomes. Given the additional therapeutic support provided by the TSDA and the generally limited availability of mental health professionals in prisons, future research on this intervention will require implementation strategies to support its sustainability and scalability in prisons and other criminal legal settings. Finally, the criminogenically focused intervention chosen for this study, like all interventions of its class, was developed for use in correctional settings. Further research is needed to determine the best way to increase the reach and impact of similar interventions in community-based mental health settings.

To address the entrenchment of people with mental illness in the criminal legal system, policy makers must expand the continuum of services available to this population. This continuum must include interventions that directly target the risk factors most closely associated with criminal legal system involvement. The results of this study showed that, with support, people with mental illness and moderate-to-high levels of criminogenic risk factors can benefit from existing criminogenically focused interventions. Informed by decades of science, these existing interventions provide an evidence-based approach to addressing the criminogenic risk factors found among justice-involved people with mental illness. This research has provided support for the potential efficacy of a short-term, self-contained intervention in augmenting traditional mental health services.