Crisis stabilization units (CSUs) serve justice-involved populations experiencing ongoing mental health and substance use disorders. In Arkansas, where our research team is located, CSUs function as a diversion to jail for individuals who come into contact with law enforcement while experiencing mental health crises. CSU guests receive respite care, pharmacological intervention, psychotherapy, and referrals to needed community resources [Reference Barocas, Umar, Praseuth, Evans, Zaller and Zielinski1,Reference Zielinski, Praseuth and Gai2]. The population served by CSUs experience profound and unique challenges (e.g., resource deprivation, housing insecurity, and poverty) which are compounded by criminal justice involvement and mental and behavioral health symptoms [Reference Zielinski, Praseuth and Gai2]. Empirical research is needed to fully understand the barriers to care experienced by this population so that policy-makers, practitioners, and other stakeholders can leverage this information to increase access to needed services, identify potential intervention points, and improve overall quality of life in an underserved and underrepresented population. However, there are many challenges to longitudinal research with justice-involved populations such as those served by CSUs.
In this paper, we offer a series of recommendations to improve study engagement, standardization, retention rates, and overall data quality in longitudinal research aiming to track justice-involved populations. Our recommendations stem from lessons learned while conducting a longitudinal cohort study of people who are discharged from a CSU. Participants are enrolled during their CSU admission and then complete five follow-up assessments over a one-year follow-up period. Considering the challenges experienced by this population – and the substantial need for research that evaluates health outcomes and service needs for justice-involved persons – translational researchers would benefit from guidance about the skills, strategies, and research infrastructure necessary to successfully engage and retain justice-involved participants in longitudinal research.
Lessons Learned
Table 1 summarizes the challenges faced by our research team when conducting longitudinal data collection with individuals discharged from a CSU, the majority of whom were justice involved. The table also offers study design and training considerations for research with similar populations. It should be noted that the challenges observed in this population often intersect. For example, people experiencing substance use disorders may also experience increased admissions to residential treatment or hospitals and/or may be arrested or become incarcerated. Individually, these circumstances make it difficult to maintain contact and conduct follow-up assessments. When these challenges co-occur, they compound and create complex barriers to participant tracking and retention efforts. Thus, the recommendations outlined in Table 1 should be used concurrently and considered throughout all phases of research planning.
Table 1. Challenges and recommendations for longitudinal studies with justice-involved persons
Additionally, each participant and the unique challenges they may experience should be evaluated to determine which considerations will be most relevant. Strategies that work well for some participants may not work for others, even if they appear to be experiencing the same challenges. Documentation of all contact attempts and retention efforts is essential to identifying variables that contribute to reduced study engagement and can help study staff determine which strategies have successfully mitigated challenges in the past.
Conclusions
Conducting rigorous research with justice-involved populations requires forethought and flexibility from investigators. It is important to mirror the needed flexibility in the study protocol—particularly with regard to the methods and resources used to remain in contact with participants—and to adequately communicate the benefits and risks of enrolling in a longitudinal study within the informed consent document(s). The study from which these recommendations were formed was methodically planned out and rigorously designed; however, unprecedented challenges (e.g., the COVID-19 pandemic) required study staff to continually reevaluate study procedures and revise training manuals. This resulted in the development of briefing documents, training presentations, and manuals which ensured that all study staff received standardized training and were able to handle unexpected challenges. Table 2 describes these materials. We also purchased access to a web-based, HIPPA-compliant participant management and scheduling tool [3] which facilitated documentation of all participant contact attempts and history of study engagement.
Table 2. Training and procedural materials
As evidenced in Table 1, most challenges within the current study centered on maintaining regular contact with study participants and completing follow-up assessments. Though data collection is ongoing, retention rates have remained relatively stable—near 75% for most time-points. This is in part because study staff cultivated relationships with community organizations who also serve the study population. To date, study staff have worked with 45 organizations, including residential treatment facilities, assisted living facilities, homeless shelters, community outreach organizations, day centers, mental health clinics, and correctional settings to maintain contact with participants. The ability to successfully engage these agencies has been key to contacting participants and keeping them engaged with the study. These successes demonstrate the efficacy of the recommendations described in Table 1. We hope the lessons we learned in conducting health research with justice-involved individuals will be considered when designing future longitudinal research studies with similar populations.
Acknowledgments
The authors wish to thank the leadership, staff, and guests of the Pulaski County Regional Crisis Stabilization Unit and the Sebastian County Crisis Stabilization Unit for their engagement with our research team. Execution of this study and manuscript preparation were supported by a research grant from Arnold Ventures. Manuscript preparation was also supported by K23DA048162 (PI: Zielinski). The authors report no conflicts of interest.
Disclosures
The authors report no conflicts of interest.
Open access
