Team profile by Maria Tackett, Nicole Schramm-Sapyta, Sanskriti Purohit, Foxx Hart, Maya Pandey, Zoe Svec, Madeline Brown, Irene Biju, Jordan Hamelsky and Will Lieber

Through the last two semesters, our team has worked on a number of projects related to the interests of Durham County identified through ongoing conversations with the Durham Justice Services Department, Stepping Up Initiative and Crisis Intervention Team. 

We have a special interest in supporting people with substance use and/or mental health disorders who also experience incarceration in the Durham County Detention Facility. To that end, we analyze a de-identified, merged dataset from Duke Health and the Durham County Detention Facility of people who have experienced incarceration in Durham County between 2014-2020.

In striving to understand associations between healthcare utilization and interactions with the justice system in Durham County, we divided our work this year into four subteams, each of which we describe in detail below: 

1. Engaging with focus groups with justice-involved community members
2. Analyzing the effects of the 2019 cash bail reform policy on those with diagnosed severe mental illness and substance use disorders
3. Examining costs to Duke Health incurred by this population
4. Conducting a meta-analysis of the use of medication assisted treatment for opioid use disorder in carceral settings

This important work would not have been possible without the guidance and support of our stakeholders in Durham County. We are extremely grateful for their partnership this year, and we look forward to continued collaboration!

Focus Groups with Justice-Involved Community Members

Introduction

From our group’s past quantitative research, we know that people with co-occurring serious mental illness (SMI) and substance use disorder (SUD) are most likely to be trapped in a cycle of recidivism and using emergency medical services for non-emergent health concerns, most often related to SMI or SUD diagnoses.

Durham County provides services through the Justice Services Department (JSD) that are aimed at assisting this population while incarcerated in the Durham County Detention Facility (DCDF) or during their re-entry period. However, more research is needed to understand how accessible and effective these services are.

To address this question, we are facilitating focus group discussions with people who have been incarcerated and have used the JSD’s services in the past five years. In this way, we will be better able to understand the perceived accessibility and efficacy of services from the perspective of individuals with lived experience in the Durham County Detention Facility.

Methods

In the spring of 2024, we initiated this study by convening two focus groups consisting of eight and five participants, respectively. Focus group participants were recruited by community partners at the Recovery Community of Durham.

With participant privacy and confidentiality in mind, these focus group conversations were recorded for qualitative analysis. Themes from each focus group were identified and compared. We anticipate performing 4-5 more focus groups in the next few months to complete this data collection.

Results

Preliminary results indicate that while participants think existing programs are very helpful, programs should focus more on reforming deeply ingrained thought patterns and advertising the resources offered.

Next Steps

This project is still ongoing. Our next steps include completing additional focus groups and conducting a rigorous thematic analysis once all data have been collected. Results will be reported in multiple formats, including an academic journal submission and a lay-language report for both participants and county stakeholders.

Cash Bail, Pretrial Release and Re-Bookings: An Analysis of 2019 Policy Reform

Introduction

The cash bail system has garnered criticism in recent years due to mounting concerns over racial and socioeconomic disparities within the criminal justice system. Critics argue that the system perpetuates inequalities by criminalizing poverty and disrupting social networks, ultimately fueling recidivism. In response, jurisdictions like Durham, North Carolina, have implemented reforms, de-emphasizing cash bonds in favor of alternative pretrial release methods. 

This subteam delved into the consequences of Durham County’s 2019 policy changes, focusing on its effects on people with serious mental illness (SMI) and substance use disorder (SUD) diagnoses. By providing empirical insights, our study seeks to contribute to more targeted and effective reforms in the criminal justice and health systems, particularly concerning individuals grappling with mental health and substance use issues.

Methods

We examined these issues in a two-step process. First, we tested whether the cash bail policy change was effectively implemented, testing whether there was an increase in Release on one’s own Recognizance (ROR) after implementation compared to before. Second, we tested whether the increase in ROR was associated with an increase in re-bookings for those released. At each stage, we specifically examined subpopulations with SMI and SUD. We used logistic regression models to address both questions.

Results and Conclusions

The odds of receiving ROR increased by 63% after the policy change. Therefore, we conclude that judges in Durham did indeed increase the use of ROR following the 2019 policy change.

With regard to who specifically received ROR, there was no significant difference in the odds of receiving ROR between people with SMI, SUD, co-occurring, or no diagnoses across periods. Thus, ROR was not given more or less preferentially to people with SMI, SUD, or co-occurring diagnosis after the policy change.

After the policy change, rebooking was 35% more likely. However, this difference was not driven by individuals who were ROR. Instead, we observed that the increase in re-bookings after the policy change was driven by the absence of a seasonal decrease in bookings in the winter. Specifically, our period of observation was September 1-March 1 of 2018-2019 (before period) and 2019-2020 (after period). In the winter of 2019-2020, after the policy change, the typical seasonal “dip” in arrests was not observed, in contrast to patterns observed in previous winters.

Finally, both before and after the policy change, individuals with co-occurring SMI and SUD were twice as likely to be re-booked as people with no diagnoses. This conclusion aligns with our previous research that people with co-occurring diagnoses remain a particularly high-need population within Durham County.

Next steps

We are preparing to submit these findings to the American Journal of Criminal Justice.

Describing Our Familiar Neighbors: Cost and Demographics

Introduction

As a preliminary step to begin exploring better services for neighbors with serious mental illness, substance use disorder and justice involvement, this project sought to explore the health care costs incurred by these individuals. We were motivated by research conducted throughout the early 2000s suggesting that the “housing first” model may be a more cost-effective and health-promoting way to support people who frequent both the health and justice systems. Previous research has shown that providing housing reduces emergency department (ED) and other service utilization.

In pursuit of this goal, we first sought to identify demographic and diagnosis patterns across cost categories.

Methods

First, we engaged in an extensive data cleaning and preparation process. We defined subgroups of patients with serious mental illness, substance use disorder, co-occurring disorders, frequent ED use (“familiar neighbors”) and type of insurance. We calculated “lifetime” costs over the 7 years of observation using administrative cost data. We then sorted patients into a range of “lifetime” cost categories from lowest percentiles (0-25) to highest percentiles (98+) based on total cost accrued during emergency department visits between 2014 and 2020. We then analyzed the relationship between insurance coverage, diagnoses and cost percentiles using a chi-squared residual analysis.

Results and Conclusions

Those with familiar neighbor status in the Emergency Department (having 5 or more visits to the ED in a calendar year) represent 18% of the justice-involved population. Sixty-nine percent of the visits to Duke Health by the familiar neighbors are covered by public insurance (Medicaid or Medicare), while 22% of the visits by this population are uninsured. The remainder are covered by private insurance.

The remaining 82% of the justice-involved population (the non-familiar neighbors) have a different mix of insurance coverage. Among non-familiar neighbors, 50% of Duke Health visits are covered by public insurance, while 26% of the visits are uninsured, and the remaining 24% are covered by private insurance.

Examining the patients after sorting into ED cost categories, we observed (as expected) that higher cost categories are associated with more emergency room visits. Next, we found that individuals with co-occurring diagnoses were overrepresented in the highest cost percentiles. In contrast, individuals with no SMI or SUD diagnoses were overrepresented in the lowest cost percentiles. (See Figure 1.)

In addition, the likelihood of becoming a chronic user of the emergency department (those with two or more 365-day periods with 5 or more ED visits) increased with increasing severity of diagnosis.

Next steps

Having spent a great deal of effort to clean and prepare this dataset, we are now open to suggestions as to appropriate next steps in using it to find better ways to support the population with SMI and SUD who also experience incarceration in Durham.

Interesting areas to explore might include:

• Effects of age, race, neighborhood or other demographic characteristics on costs incurred
• Examination of costs incurred in emergency department vs. outpatient and inpatient services
• Temporal progression of service utilization, i.e., how often do people move from frequent ED use to outpatient clinic use?
• With soon-to-be-acquired data from the Justice Services Department, how effective are release services at ensuring continuity of outpatient care vs. return to ED usage?

These questions and others relate closely to research on familiar neighbors being carried out by Michele Easter, who will soon submit a paper describing primary care utilization by the jail-involved population.

The policy landscape in North Carolina continues to evolve as we do this work, in ways that stand to benefit this population. For example, in addition to recent Medicaid expansion, health opportunity pilots (1115 waivers) can now support non-medical interventions for social determinants of health (such as homelessness or food insecurity). The V88 register now allows hospitals to be paid to coordinate housing before release of patients from hospitals. Finally, beginning in October 2024, North Carolina will expand the 1115 waiver to include support for transitions from incarceration back to the community.

We remain hopeful and committed to exploring avenues that increase both cost-effectiveness and wellness for this population.

Meta-Analysis of Medication Assisted Treatment for Opioid Use Disorder

Introduction

Medication Assisted Treatment (MAT) for opioid use disorder was implemented on-site in the Durham County Detention Facility in 2019 through the SMART program. Therefore, this project reviewed published research examining the efficacy of medication-assisted treatments (MAT) for opioid use disorder when initiated in carceral settings. We focused on studies in which the efficacy of MAT to prevent relapse to drug use was assessed through urinalysis six months after release.

Methods

With assistance from Duke librarians, we performed a thorough search of the PubMed Database. Our initial search of PubMed yielded 130 potentially relevant papers. Careful screening resulted in 3 papers which were included in our final analysis. These three studies were the only ones to meet our strict criteria of initiating MAT in a carceral setting and evaluating relapse to opioid use disorder at six months post-release using urinalysis. Data from these three studies was combined and analyzed using mixed-effect regression models.

Results

Across the three papers we examined, any use of MAT for opioid use disorder decreased the odds of relapse six months from release, an effect that was particularly seen with Methadone.

Compared to controls, patients across all three studies who received MAT were 2.67 times less likely to relapse. Those who specifically received methadone were 4.13 times less likely to relapse compared to controls.

Conclusions

Based on the results of this meta-analysis, we conclude that MAT, and the use of Methadone in particular, is an effective intervention for reducing the odds of relapse to opioid use disorder at 6 months. For Durham County, this study affirms the value of programs such as the SMART program. Durham County should continue to offer MAT in this carceral setting and should continue to serve as a successful example for other counties in North Carolina.

Conclusion

It is our pleasure to share the progress made by the “Mental Health and the Justice System in Durham County” team from Duke University. In the 2023-2024 academic year:

• We engaged individuals with lived experience through focus groups to evaluate their perceptions of existing programs administered through the Justice Services Department.
• We analyzed the 2019 cash bail policy, and among other findings, observed that those with co-occurring SMI and SUD were the most likely to be rebooked both before and after the policy change.
• We found that those with co-occurring SMI and SUD diagnoses were overrepresented among the highest cost patients at Duke Health.
• Finally, through a meta-analysis of published literature describing MAT implemented in carceral settings for opioid use disorder, we found that MAT, and particularly Methadone, decreases odds of relapse at six months post-release. 

We are incredibly grateful for your support, continued partnership, and commitment to the city and county of Durham. We welcome feedback on these analyses and ideas or requests for future studies.

Project Contributors

Dr. Nicole Schramm-Sapyta, Ph.D. is an Associate Professor of the Practice and Associate Director of the Duke Institute for Brain Sciences. She attended North Carolina State University before earning her Ph.D. in pharmacology from Vanderbilt University. Dr. Schramm-Sapyta is a co-lead of the Mental Health and the Justice System in Durham County team and contributed to all four projects.

Dr. Maria Tackett, Ph.D. is an Assistant Professor of the Practice in the Department of Statistical Science at Duke University. She attended University of Tennessee, Knoxville before earning a Ph.D. in statistics from the University of Virginia. Dr. Tackett is a co-lead of the Mental Health and the Justice System in Durham County team and contributed to all four projects.

Dr. Michele Easter, Ph.D. is an Assistant Professor in the Department of Psychiatry and Behavioral Sciences, Division of Child, Family, and Community Psychiatry. She attended Oberlin College, followed by UNC-Chapel Hill (M.A. Religious Studies) before earning her Ph.D. in Sociology from UNC-Chapel Hill. Dr. Easter is a contributor to the team and provides advice and guidance to all projects.

Sanskriti Purohit, M.S. graduated from Delhi University with a degree in Mathematics. She graduated in Spring 2024 from Duke with a master’s degree in Statistical Science. Sanskriti was the graduate manager of the team and contributed to all four projects.

Foxx Hart was a graduating senior at Duke University studying Global Culture & Theory. Foxx led the “Meta-Analysis of Medication Assisted Treatment for Opioid Use Disorder” subteam.

Maya Pandey was a graduating senior at Duke University studying Public Policy and Economics. Maya led the “Cash Bail Policy Analysis” subteam.

Zoe Svec was a graduating senior at Duke University studying Statistics. Zoe co-led the “Focus Groups” subteam.

Madeline Brown, B.S. was a third-year medical student at Duke University School of Medicine and a student in the Duke Master's in Public Policy program. Maddie co-leads the “Focus Groups” subteam in 2023-2024 and will lead the remaining focus groups and analyses to complete the project.

Irene Biju was a sophomore at Duke University studying Computer Science and Public Policy. Irene co-led the “Cost Data Analysis” subteam and will return to the team in her junior year.

Jordan Hamelsky was a junior at Duke University studying Statistics. Jordan aided all four subteams with coding and technical support. Jordan will return to the team next year to perform an analysis of neighborhood-related characteristics as they relate to re-arrest and health care usage.

Will Lieber was a junior at Duke University studying Health & Incarceration. Will co-led the “Cost Data Analysis” subteam and will return to the team in his senior year.

Mental Health and the Justice System in Durham County: Striving to Understand Associations Between Healthcare Utilization and Interactions with the Justice System in Durham County

Poster by Maria Tackett, Nicole Schramm-Sapyta, Irene Biju, Madeline Brown, Maya Pandey, Zoe Svec, Sanskriti Purohit, Jordan Hamelsky, Foxx Hart and William Lieber