An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Benjamin Druss, MD, MPH, Cathy A. Lally, MSPH, Jianheng Li, MPH, Stephanie Tapscott, MSW, and Elizabeth R. Walker, PhD, MPH, MAT.
Author Information and AffiliationsPatients with mental disorders are among the highest users of emergency department (ED) services, with visits characterized by long lengths of stay, intensive treatment, and elevated rates of hospitalizations. After discharge, fewer than half of these patients successfully transition to outpatient care, with high rates of readmission to the ED. Programs that allow patients with mental disorders to successfully engage with outpatient care hold the potential not only to improve quality and outcomes of care for people with mental disorders, but also to free capacity within EDs to provide care for other urgent needs.
Care management programs can facilitate care engagement in people with mental illness after ED discharge. However, it has been difficult to disseminate these models more broadly. One reason for this challenge has been a shortage of mental health professionals who can serve in these roles. There is now an opportunity to improve care and fill these gaps with a new provider group, certified peer specialists—persons with a history of mental illness offering services to other individuals with mental illness. This study examines the potential benefits and trade-offs between certified peer specialists and professionals (nurses, social workers, and counselors) in managing care transitions for patients with mental disorders after ED discharge. We also aimed to study other patient and community factors that predict successful treatment engagement.
The original study aims were to (1) compare rates of treatment engagement achieved by certified peer care managers vs those achieved by professional care managers; (2) compare clinical outcomes between the 2 groups; and (3) examine treatment heterogeneity between the 2 groups, with a focus on whether patients presenting with different types of barriers to engagement demonstrate differential benefits across provider types. Due to accrual challenges and low follow-up rates, a decision was made to end the study early, to rely on administrative data for exploratory study outcomes, to modify the qualitative aim, and to reduce the target enrollment for the study from 1000 to 290. The study aims were modified as follows: Aim 1, use quantitative data from the randomized study sample to assess factors including care manager group (peer vs professional) predicting patient engagement after a mental health ED visit; and aim 2, use qualitative interviews with patients and providers to understand stakeholder perspectives on patient engagement after a mental health ED visit.
Eligible participants across 8 sites in South Carolina who were scheduled to be discharged from an ED visit to a participating community mental health center (CMHC) were randomly assigned at the individual level to either a professional care manager (nurse, social worker, or counselor) or a peer care manager. All care managers delivered an adapted version of a 1-year, manualized, evidence-based care management program for patients with mental illnesses. The program has been demonstrated to improve the quality and outcomes of care. This intervention supports patient engagement and care coordination. To assess factors that predicted patient engagement after the ED visit, including care manager group (peer vs professional), we used bivariate and multivariate hierarchical linear models to examine patient, provider, and community predictors of engagement in care. The primary study outcome was at least 1 outpatient clinic visit for a mental health problem in the 30 days after ED discharge. Secondary outcomes were the proportions of outpatient visits attended, ED readmissions, and all-cause inpatient admissions in the 6 months after ED discharge.
For the qualitative aim, purposive sampling was used for patients and care managers. Randomized trial participants who had completed a 6- or 12-month follow-up survey were invited to take part in an in-depth interview. All active care managers were invited to participate in an interview. A total of 45 semistructured interviews with patients (n = 30 interviews) and care managers (n = 15 interviews) assessed barriers and facilitators to engagement with outpatient services at the patient, provider, and health care services levels as well as the use of strategies to promote engagement in treatment. Thematic analysis was used to identify themes at each of the 3 levels.
A total of 326 participants were randomly assigned to either peer specialists or professionals; 316 participants had data available for analysis. There was a significant difference in the rates of 30-day follow-up between the 2 intervention groups, with participants assigned to professionals being significantly more likely to have successful transitions to outpatient care than those assigned to peers (55% vs 43%; P = .03). Peers had greater rates of turnover and wider variability in rates of outpatient follow-up than did professional care managers.
In multivariable models, a range of demographic (age, gender), clinical (mental health diagnosis and comorbid substance use), and geographic (rurality, distance from the participating CMHC) factors were associated with different measures of treatment engagement. The qualitative aim found several determinants of successful transitions from the ED to outpatient care, including transportation challenges, patients' openness to receiving care, financial insecurity, and severity of mental health symptoms.
By identifying patient-, provider-, and system-level barriers, this study highlights the importance of providing services to support the transition to outpatient mental health care after discharge from the ED. On average, professional care managers had higher rates of 3-day outpatient follow-up than did peer care managers; in addition, peer care managers had higher rates of job turnover and variable performance across individual peer care managers. Additional research is needed to help identify barriers and develop strategies for optimizing care transition interventions across provider disciplines.
The early study termination precluded the opportunity to collect and analyze data for patient-reported outcomes and to qualitatively examine the components of the intervention. Other factors, including turnover of peer care managers and variability in the local clinic scope-of-practice rules, also limited our ability to draw definitive conclusions about the differential effectiveness of peers vs professionals in facilitating treatment engagement.
Emergency departments (EDs) serve a critical role as a safety net for uninsured and medically disenfranchised individuals. The growth of population-based models of care under the Patient Protection and Affordable Care Act has provided an impetus to reduce unnecesssary ED use and improve engagement with outpatient care.1,2 As such, there is an urgent need to improve care transitions for patients with mental disorders seen in EDs. Patients with mental disorders are among the highest users of ED services,3-10 with visits characterized by long lengths of stay in the ED, intensive treatment, and elevated rates of hospitalizations.11 After discharge to the commuity, fewer than half of these patients successfully transition to outpatient care, with high rates of readmission to the ED.12,13 Timely clinic follow-up after discharge from ED visits is a common performance metric representing a successful initial transition of care.14
Programs that allow patients with mental disorders to successfully transition to outpatient care hold the potential not only to improve quality and outcomes of care for mental disorders, but also to free capacity within EDs to provide care for other urgent needs.15 Care management, which provides patients with education and logistical support to overcome barriers to care, appears to be one such evidence-based method to improve treatment engagement after ED discharge for patients with mental disorders.
A 2015 Agency for Healthcare Research and Quality (AHRQ) evidence synthesis provided a comprehensive review of strategies for linking inpatients to aftercare and reducing psychiatric readmissions.16 The review found that care management strategies that follow patients over time from acute settings to outpatient care are a particularly promising approach to improve patient engagement and reduce readmissions. Across multiple studies, care management was found to reduce the time until first outpatient follow-up visit and to increase the continuity of outpatient care.17-25 Successful programs combined needs assessments during the acute crisis with ongoing care management to facilitate engagement with outpatient care.16
However, despite the potential of care management programs to address transitions of care in people with mental illness, it has been difficult to disseminate these models more broadly.26 One reason for this challenge in dissemination has been a shortage of mental health professionals who can fill these roles, including a shortfall of mental health care nurses, who frequently serve in care manager roles.27 This problem has become worse in recent years as the psychiatric nursing workforce has aged.28 Similar shortages are seen for mental health social workers.29
There is now an opportunity to improve care and fill these gaps with a growing group of mental health providers, certified peer specialists—persons with a history of mental illness offering services to other individuals with mental disorders.30 Beginning in the 1990s, certified peer specialists began to be employed to provide peer support in community mental health settings.31 Certified peer specialists help other patients improve outcomes by modeling successful behaviors and coping strategies and by guiding patients to use community resources more effectively. Nationwide, there are approximately 25 000 peer support specialists, and 41 state Medicaid programs recognize and reimburse mental health services delivered by certified peer specialists, helping ensure the financial sustainability of programs using these providers.32 Certified peer specialists thus hold real promise in providing these services. Mental health organizations must therefore decide if and when they can augment or replace professional providers with peers to facilitate effective transitions of care.
Several reviews,33,34 including a 2013 Cochrane Systematic Review,35 have examined the evidence comparing peer specialists with traditional mental health professionals, such as nurses and social workers. Across 11 randomized trials35 and an additional 9 quasi-experimental studies,34 care delivered by peer specialists was as good as or better than care delivered by professional staff in similar roles for outcomes including quality of life and mental health symptoms.34,35 Peer specialists provided more time face to face with clients,33 and patients treated by peer specialists had greater improvements in recovery-oriented outcomes and social function,34 suggesting that peers may offer advantages over traditional professionals for these functional domains.
The AHRQ evidence review16 found only 1 pilot study examining the effectiveness of certified peer specialists in reducing readmissions of inpatients with mental disorders.36 The study recruited and randomly assigned 74 psychiatric inpatients to either a peer mentor or usual care, and reassessed outcomes at 9 months. Participants with a peer mentor had significantly fewer mean rehospitalizations (0.89 vs 1.53, respectively; P = .03) and mean hospital days (10.08 vs 19.08, respectively; P = .03). The authors noted that the study was limited by its small sample size, single site, and incomplete service use data outside the study hospital.36 In contrast to our proposed project, this pilot study by Sledge et al also focused on psychiatric patients discharged from inpatient status rather than ED users. Nonetheless, the AHRQ review identified this as “an important strategy” that merited further systematic study.16
Addressing this workforce comparative effectiveness question first requires a better understanding of the clinical, social, and contextual factors that facilitate or hinder effective follow-up after mental health visits to the ED. Studies using US Department of Veterans Affairs (VA) data37 and Medicaid claims data38,39 have found that sociodemographic factors, such as age and race; clinical factors, including mental health diagnosis; and county-level factors, including rurality, are predictors of follow-up after ED visit in those populations. However, these findings may not be generalizable across a wider range of payers than across Medicaid and the VA. Furthermore, claims data provide limited information about the full range of barriers to follow-up after discharge for mental health ED visits. Mixed methods that combine these findings with qualitative data may offer complementary information for understanding these contextual issues.40-43
The existing literature suggests 3 important areas for research: first, the potential effectiveness of care management delivered by mental health professionals in improving treatment engagement; second, the ability of certified peer specialists to effectively deliver a range of services for patients with mental disorders; and third, the contextual factors pertaining to engagement that are critical for planning and implementing these interventions. However, there have been no studies comparing certified peer specialists with professionals in interventions to increase engagement with treatment and reduce readmissions after ED visits. Moreover, few data exist about the contextual factors that can be used to help identify, develop, and target peer-led interventions. The study's findings may help patients, providers, and policy makers understand the benefits and trade-offs of using these 2 types of providers to facilitate care transitions.
This study randomly assigned patients seen in EDs for mental health problems to receive an adapted version of an evidence-based care management intervention44,45 delivered by either peer specialists or professionals (nurses, social workers, or counselors). The study examined factors associated with treatment engagement, including provider type, after discharge from the ED. The goals of the study were to better understand predictors of treatment engagement, including the potential benefits of care management, after discharge from the ED. The original study aims were to (1) compare rates of treatment engagement achieved by certified peer care managers vs those achieved by professional care managers; (2) compare clinical outcomes between the 2 groups; and (3) examine treatment heterogeneity between the 2 groups, with a focus on whether patients presenting with different types of barriers to engagement demonstrate differential benefits across provider types.
Due to accrual challenges and low follow-up rates, a decision was made in conjunction with PCORI to end the study early, rely on administrative data for exploratory study outcomes, modify the qualitative aim, and reduce the target enrollment for the study from 1000 to 290 participants. More detail about these changes is provided in the “Changes to Original Protocol” section.
There are 2 modified study aims:
The study engaged a diverse group of stakeholders throughout the study process.46 A national advisory board, chaired by the executive director of a consumer-run organization, included representatives from national provider, payer, and patient advocacy groups. The advisory board, which met annually, assisted in the overall planning and implementation of the project. For example, the advisory board provided feedback that was incorporated into the care management program, such as the importance of early connection between the care manager and patient, ideally while the patient is still in the ED. Guidance from the advisory board helped (1) optimize the “warm handoff” to care managers while that patient was in the ED or soon after, (2) increase involvement with patients' families and support systems, and (3) improve connections between community mental health centers (CMHCs) and ED staff.
A state implementation team was composed of 1 patient, 1 certified peer specialist, 1 member of a patient advocacy organization, and 2 state policy makers from the South Carolina Department of Health and Human Services and Department of Mental Health. The implementation team members were recruited via study team contacts in the South Carolina Department of Mental Health and recommendations from the advisory board. The implementation team, which met quarterly, helped oversee the day-to-day functioning of the project, troubleshoot challenges, and consider how the findings would be scalable within the state. Meetings were set up to promote discussion between team members, and feedback from all team members was considered equally. The implementation team reviewed the care management program materials; their feedback contributed to refinements in the protocols for recruitment and enrollment in the EDs as well as in the care manager manual.
Advisory board and implementation team members were compensated for time spent planning and attending meetings.
We also conducted developmental interviews with clinicians and focus groups with patients as part of the process for planning the study intervention and workflow.46 Interviews with 7 clinicians from EDs and CMHCs provided valuable information about the practices in each care setting. The main themes from the interviews were the importance of being nonjudgmental and supportive of patients as well as providing validation for patients' concerns. We conducted 2 focus groups with 12 patients at 2 different CMHCs. Information from the qualitative data was used to refine the care manager manual, training, and materials. For example, patients emphasized the need for care managers to show compassion and treat each patient as an individual. Therefore, the wording in the care manager manual was revised to ensure that empathy and support were conveyed. Feedback from the clinicians was used to refine recruitment scripts and protocols, as well as study materials given to patients in the ED. The study team worked closely with recruitment sites for education, training, and project refreshers. Site visits were held to update leadership and staff on project progress, encourage continued site recruitment efforts, and assist in the facilitation of referrals given site procedures and preferences. The team made at least 1 in-person visit and monthly phone contact with each active site to better understand specific procedures and preferences impacting the referral process.
Participants were recruited across 8 sites, comprising 7 CMHCs each paired with 1 ED and an eighth CMHC paired with 2 smaller EDs, and randomly assigned to either a peer care manager or a professional care manager. All care managers delivered an adapted version of a manualized, evidence-based care management program for patients with mental illnesses. The program has been demonstrated in randomized controlled trials to improve the quality and outcomes of care.44,45 This intervention supports patient engagement with health and mental health care as well as care coordination. For the revised study, bivariate and multivariate models examined patient, provider (including peer vs professional care manager), and community predictors of engagement in care.
The revised study employed a convergent mixed-methods design, whereby the quantitative and qualitative strands were designed and implemented separately and the findings then considered together. For aims 1 and 2, the methods for sampling, data collection, and analysis are described separately in the sections below. The results from the 2 types of data were compared and interpreted together. The convergent design allowed for corroboration of findings across data-collection methods.47
The 8 sites represented a mix of suburban, urban, and rural locations (Table 1). A total of 6 sites had turnover of peer care managers during the study period, and 2 had turnover of professional care managers.
Site Characteristics.
ED staff referred patients admitted to the ED for a primary diagnosis of a mental disorder to be assessed for potential inclusion in the study. Via videoconferencing equipment or telephone, a trained research interviewer gathered information from patients to assess them for potential study eligibility before discharge from the ED.
The following broad inclusion criteria were chosen to maximize the generalizability of the study findings:
The exclusion criteria consisted of cognitive impairment based on a score ≤3 on the Callahan Six-Item Screener, a validated instrument developed for clinical research48 (which could impede capacity to provide informed consent as well as effective participation in the intervention); not being able to speak English; and being admitted to the hospital from the ED.
Participants were randomly assigned, using a computer-generated algorithm and concealment of allocation techniques to minimize assignment bias, to either the professional or peer care manager from the participating CMHC. Randomization was stratified by study site. Initially, 1000 participants were expected to be enrolled, but due to recruitment and accrual challenges, the target number of participants was reduced to 290. Recruitment exceeded this modified target goal, with a total of 326 participants who were randomly assigned. See the “Changes to Original Protocol” section for more details.
The study compared the same care manager intervention implemented by 2 different provider groups: peers and professionals. We used the Coordination, Access, Referral and Evaluation (CARE) intervention, a manualized care management program to improve follow-up and treatment engagement for patients after admission to the ED. It is built on the Primary Care Access, Referral and Evaluation (PCARE) intervention, which was designed to improve treatment engagement and outcomes for patients in outpatient mental health settings.49,50 The primary enhancement to the PCARE intervention to develop the CARE intervention for this project was to add an initial treatment linkage visit during the ED visit. Professional and peer care managers each delivered the same intervention; the goal of the study was to understand differential benefits of treatment by different provider types. The intervention lasted 1 year.
The intervention took a highly patient-centered approach to care management. Shared decision-making between care managers and patients was the organizing principle used to achieve balance between patient choice and concordance with evidence-based standards of care.51,52 Shared decision-making has been defined as
a mechanism to decrease the informational and power asymmetry between doctors and patients by increasing patients' information, a sense of autonomy and/or treatment decisions that affect their well-being.51
It involves (1) explaining the need to consider alternatives as a team (team talk); (2) describing alternatives in more detail (option talk) using decision tools; and (3) helping patients explore and form their personal preferences (decision talk).53
The intervention combined a traditional medical model of care management with a recovery-based approach. Whereas clinical definitions of recovery frame it in terms of symptom remission, recovery as defined by patients with mental illnesses reflects
a deeply personal, unique process of changing one's attitudes, values, feelings, goals, skills and roles. It is a way of living a satisfying, hopeful, and contributing life even with the limitations caused by illness. Recovery involves the development of new meaning and purpose in one's life beyond the diagnosis of mental illness.54
Recovery has become a central organizing theme in current efforts to create a more patient-centered mental health system.55 The CARE intervention framed health goals and the use of health care services in the broader context of recovery.
After a patient was enrolled in the study, the study staff provided the patient's contact information to the care manager from the assigned CMHC. The care manager contacted the patient by phone either while they were in the ED or within 48 hours of discharge and established rapport with the patient. The care manager confirmed contact information for the patient, including telephone and cell phone numbers,56 address, and contact information for friends/relatives. The care manager also ascertained the reason for the ED visit and set an initial appointment to follow up clinically at the partner CMHC within 2 weeks.
The purpose of the initial visit at the CMHC was to take a history of the patient and begin the process of identifying and addressing barriers to engagement in care. The care manager then reviewed the roles of the patient and care manager, emphasizing the importance of the collaborative relationship between them.
The care manager examined the reasons for the ED visit, including whether it was urgent or routine or was expected or unplanned, as well as how the patient perceived the usefulness of the visit. Next, the care manager and patient collaboratively identified attitudinal and logistical barriers to care and developed a plan to address one of the barriers. Attitudinal barriers include (1) believing that there is no need for services; (2) believing that treatments for mental disorders are not effective; (3) believing that the problem will get better on its own; (4) believing that the problem is not serious enough to need treatment; and (5) not wanting to be labeled “mentally ill.” Logistical barriers include (1) not being aware of an available place to obtain care; (2) not having transportation to a clinic; (3) feeling it is too inconvenient to obtain treatment; (4) being unable to make an appointment; and (5) feeling it is too expensive to make an appointment. After the patient identified their primary barrier, the patient and care manager collaboratively reviewed a process to develop a goal and make actionable steps to address the barrier.
After the initial visit, the care manager monitored patients' engagement in care and helped address attitudinal and logistical barriers to care. In follow-up meetings, the care manager assessed patients' progress on their goal to overcome the barriers to care identified in the initial session. Logistical barriers may have been addressed by assisting patients in applying for health benefits, providing transportation vouchers, and assisting patients in finding local childcare options. The care manager and patient reassessed the goal at each session and reviewed the goal-setting process to make updates as needed. The CARE program was structured to include 5 monthly follow-up meetings (6 total in-person meetings, including the initial visit) between the care manager and participant, with a phone call between visits. After 6 months, the care managers made monthly follow-up calls to check in with participants about the progress toward their goals.
At each site, the research team trained a certified peer specialist and a professional working for a local CMHC. Licensed professionals (nurses, social workers, or counselors) or certified peer specialists working at 8 state-funded CMHCs served as care managers for the current study. Professional care managers had at least 12 months of experience in working with patients with mental illnesses. Peer specialists had a minimum of a high school education, a history of a mental illness, were self-described as “in recovery,” were certified by the state as care providers, and had reliable transportation to and from the CMHC and ED. All certified peer specialists were trained in a curriculum that supports identifying and pursuing goals for recovery, developing and documenting recovery-focused treatment plans, and supporting linkages with community-based services. Peers learned to help other individuals with mental health conditions to facilitate mental health dialogues, explore mental health choices and options, identify and work with a clinician, and obtain access to community health supports. This training, which is required for state certification, is administered through a state nonprofit agency, which also helps link graduates with available job positions.57
Each care manager had an initial 2-hour training session on how to deliver the CARE program from the research team. At the end of the training, care managers completed an evaluation, including role play, to ensure that they were able to deliver the program. Fidelity to the program was assessed by examining the care managers' session notes. Feedback on fidelity was compiled and shared in quarterly supervision meetings. These meetings also allowed care managers to share successes and challenges in working with the study participants, think collaboratively about ways to address challenges, and provide suggestions for the program (eg, patient engagement, comments on the online submission form for documenting patient encounters). Separate meetings were held for peers and professionals to prevent contamination between the study groups. Starting in August 2018, we sent quarterly newsletters to the care managers to provide program updates, education, and support.
The study experienced many challenges in meeting recruitment targets, with rates of accrual regularly running below 50% of the projected milestones. The most important impediment to recruitment was that due to the large number and geographic dispersion of sites, recruitment was conducted remotely via videoconferencing or telephone. Because this strategy relied on referrals from local ED staff, we could screen only a relatively small portion of this potentially eligible group for study inclusion. Several additional challenges also hampered recruitment efforts. Difficulty in hiring and retaining care managers, particularly in rural areas, necessitated taking several sites offline temporarily. Patients who did not live in the catchment area were not eligible to receive services at the CMHC and therefore could not participate in the study.
Follow-up was hampered by challenges in reaching participants remotely after discharge from the ED. Although we collected extensive contact information at baseline, many participants did not answer calls or return messages later on. When reaching out for follow-up surveys, the research team contacted participants via their preferred method of contact (ie, telephone, mail, email). Participants received mailed letters highlighting their scheduled upcoming research interview appointment and an appointment reminder telephone call. Participants also received gift card incentives for their time and effort spent in completing interviews. However, despite these measures, many participants were difficult to reach for their follow-up interviews.
Many efforts were made to mitigate these challenges, including outreach to ED staff and leadership at the sites, extension of the recruitment period, and exchange of a low-volume site for a high-volume site. However, recruitment challenges persisted, and ultimately, the monthly targets were not deemed achievable. A decision was made to end the project early and to focus on analyzing the existing administrative data. The following modifications were made to the study protocol:
The modified protocol made it possible to test the originally proposed primary hypothesis in the smaller sample (ie, the relationship between provider group and 30-day follow-up after discharge) but with the ability to detect a larger effect size (16% vs the 9% used in the original protocol) than was projected in the original proposal. The approved modified protocol removed the self-report measures, added a set of exploratory quantitative analyses, and modified the qualitative analyses to focus on the barriers to treatment engagement. The modified study aims are detailed below.
Baseline data were collected from participants at the time of enrollment. These were linked to the South Carolina Office of Revenue and Fiscal Affairs (RFA) integrated data warehouse. The RFA data warehouse pulls client-specific data from an array of health and human services facilities, agencies, and organizations and makes possible the integration of data from disparate sources at the client level by means of an internally assigned unique tracking number. State legislation requiring reporting of both private-sector and public-sector client-level data ensures that the data are comprehensive and complete. The current data warehouse includes information from private-sector facilities (eg, hospitals, EDs) and state agency systems (eg, Medicaid, mental health centers, substance abuse, criminal justice).
The study team provided the RFA with identifying data for those participants who signed informed consents to access their data. These identifiers were used by the RFA staff to link to the data stored with the RFA integrated data warehouse. The linked files were transferred to the RFA using a secure file transfer protocol (SFTP) site and were encrypted; they were password protected before transfer. After data linkage, research files were generated using the identifiers provided to the RFA, transferred to the study team using the same SFTP site, encrypted, and password protected. The RFA used a matching algorithm based on date of birth, gender, and Social Security number to match and generate an encrypted ID number linking the RFA data to the study participants' records.
Data adequacy was assessed by reviewing the records to ensure that the participant had an administrative record at the ED on the date of study enrollment ±1 day.
The Area Health Resource File58 was linked to the research file using the participants' home address zip code provided at study enrollment and used to identify county-level community predictors of treatment engagement, including home address in a rural vs urban county; county-level percentage poverty; and county mental health provider shortage area.
At least 1 outpatient visit to any provider at the CMHC (including but not limited to the care manager) for a mental health problem in the 30 days after discharge from the ED. This is the one of the core performance indicators from the National Center for Quality Assessment (NCQA)'s Healthcare Effectiveness Data and Information Set (HEDIS) performance measure suite.59 The Follow-Up After Emergency Department Visit for Mental Illness measure assesses the percentage of ED visits for the patient with a principal diagnosis of mental illness who had a follow-up visit for a principal mental illness diagnosis within 30 days of the ED visit. The specifications followed the NCQA's measure guidelines.60
Adherence and compliance with scheduled outpatient visits, calculated as the total number of visits divided by the number of visits scheduled in the 6 months after enrollment
ED readmissions, measured as all-cause ED readmissions in the 6 months after study enrollment
All-cause inpatient admissions in the 6 months after enrollment
Peer vs professional care manager: all participants were randomly assigned to either a peer specialist or a professional (nurse, social worker, or counselor) employed by the CMHC
Other exploratory independent variables included in the models were as follows:
Demographic—age at the time of enrollment, self-reported race/ethnicity, and insurance status from the RFA integrated data warehouse
Specific primary mental health diagnosis (psychosis, bipolar disorder, depression, other diagnoses) from either the integrated data warehouse or reported to the study team at the time of enrollment
Co-occurring mental health diagnosis (psychosis, bipolar disorder, depression, other diagnoses) from either the integrated data warehouse or reported to the study team at the time of enrollment
Any comorbid substance use diagnosis from the integrated data warehouse
Distance from home address to nearest CMHC from the intake interview, calculated using ArcGIS, a software package for mapping geographic information system (GIS) information61
Home address in a rural vs urban county from the Area Health Resource File58
County-level percentage poverty from the Area Health Resource File58
County mental health provider shortage area from the Area Health Resource File58 based on the ratio of mental health providers, which includes psychiatrists, clinical psychologists, clinical social workers, psychiatric nurse specialists, and marriage and family therapists, to the population of the county17
The original study sample size was targeted to achieve 80% power to detect a 9% difference in the primary study outcome (30-day outpatient follow-up) between the 2 care manager groups with a statistical significance level of P < .05. This estimate was based on previous estimates of care management interventions relative to usual care found in other ED-based outreach programs.13
For the modified protocol, we calculated that with a projected sample size of 290, it would be possible to detect an effect size of 16% for bivariate models and 17.5% for multivariate models, with an α of .8 and P value threshold of .05, using data from the state's integrated data warehouse.
The data analysis for this study was conducted using SAS version 9.4 (SAS Institute). Bivariate and hierarchical regression models examined the intent-to-treat effect of peer vs professional care managers on 30-day follow-up after admission to the ED. A separate hierarchical regression model examined the association between the patient-, provider-, and community-level predictors of 30-day follow-up. Regression analyses also examined the effects of patient- and system-level characteristics on the secondary outcome measures: proportion of completed ED visits, repeat ED visit within 6 months, and inpatient admission within 6 months.
We employed a 2-level hierarchical Poisson regression model with robust error variances to estimate the relative risk (RR)62 for binary outcomes (30-day outpatient follow-up after discharge from the ED, repeat ED visits, inpatient admission) and a 2-level hierarchical linear regression model for continuous outcomes (proportion of completed visits). The hierarchical modified Poisson regression model is appropriate for studying data with group structure and a binary response variable given that an odds ratio (OR) estimated by logistic regression tends to overestimate RR when the incidence of an outcome of interest is relatively common in the study population. The binary response variable is 30-day outpatient follow-up, and the group structure is site of randomization. The site of randomization is included as a random effect to measure effects at the provider level while accounting for additional variability at the patient level.
The general model used was:
Ln(rate) is the log rate that the outcome variable equals 1 instead of 0 (ie, the probability that patient i from CMHC j received a 30-day outpatient follow-up visit). B00 is the fixed intercept, whereas u0j is the deviation of the CMHC-specific intercept from the fixed intercept.
Multiple imputation with fully conditional specification was used to impute missing data. Estimates from 5 imputed data sets were combined to generate valid statistical inferences.
In-depth interviews were conducted with patients and care managers to gain a deeper understanding of the barriers to and facilitators of engaging in outpatient treatment after being discharged from the ED. Qualitative interviews are an appropriate method for gaining a deeper understanding of individuals' experiences and the context in which those experiences occur.63 The qualitative aim was guided by the social ecological model because patients' engagement in care is impacted by factors at the patient, provider, and health care systems levels. We also used a thematic analytic approach for identifying patterns of barriers and facilitators at each of the 3 levels.64
The rescoping of the study led to a change in focus for the analysis. Originally, a main purpose of the qualitative interviews was to examine the experiences of patients and care managers with the care management program. After rescoping, the focus of the qualitative aim shifted to barriers to and facilitators of transitions of care from the ED and engagement with outpatient mental health care.
Qualitative data were collected in 2 rounds, February-August 2018 and January-May 2019, to include participants who had recently completed a 6- or 12-month follow-up survey. During the first window of data collection, 39 patients completed a survey, and 31 patients (79.5%) indicated interest in a qualitative interview; we completed 15 interviews (48.4%). During the second round of data collection, 66 patients completed a survey, 53 patients (80.3%) indicated interest in an interview and were contacted, and 15 interviews were completed (28.3%)
We completed interviews with 7 of 14 active care managers in 2018 (50%) and 8 of 21 active care managers (38%) in 2019, for a total of 15 care manager interviews. Two care managers who were active in 2018 and 2019 completed an interview in each round of data collection, resulting in 15 interviews with 13 unique participants. A total of 46% of interviews were conducted with certified peer specialists and 54% with mental health professionals (therapists, counselors, social workers).
Approximately 16 to 24 interviews are necessary to reach thematic saturation63; therefore, we chose our sample size to have enough interviews across the 2 rounds of data collection and 2 groups of participants to ensure thematic saturation.
Purposive sampling was used to choose patient participants for the interviews. Participants were eligible for the qualitative interview if they were enrolled in the parent study and completed a follow-up survey, regardless of the extent of their engagement with their care manager. At the 6-month and 12-month follow-up surveys with research staff, patients were asked if they were willing to take part in a qualitative interview. Research staff followed up with interested individuals by phone to complete the informed consent process for the qualitative part of the study, in which they explained the interview and answered any questions. Patients provided informed consent either by signing and returning a paper copy by mail or verbally after the research staff read the consent form. Once informed consent was received, the staff member scheduled a time to conduct the interview by phone. Patients received a $50 money order as compensation for their time.
During each round of data collection, purposive sampling was used to recruit care managers for interviews. All active care managers who had completed their training were eligible to participate. They did not need to currently have program participants on their caseload. All active care managers were emailed an invitation to participate in an interview. Care managers were emailed a consent form, which they signed and returned to the research team by secure mail or fax. On completion of the interview, care managers also received a $50 money order.
We developed 2 semistructured interview guides—1 for patients and 1 for care managers—that were informed by information collected via stakeholder interviews and patient focus groups during the formative phase of this study (see the Appendix).46 The patient interview guides included questions about their reasons for going to the ED, connection to and attendance at the CMHC, barriers to and facilitators of attending their appointments, and interactions with mental health providers. The care manager interview guide included questions about care managers' assessment of their patients' barriers to and facilitators of transitioning from the ED and engaging in care at the CMHC as well as questions about interactions with patients. After the first round of interviews, we added questions about engagement with care to capture additional detail on patients' views about their involvement in treatment and care managers' perspectives on their patients' engagement.
Interviews were conducted by Elizabeth Walker, PhD, who is an expert in qualitative methods, and 3 additional researchers (2 students working toward their MPH degree, and a study staff member with previous qualitative research experience). Dr Walker provided training to the 3 researchers before the start of qualitative interviewing. We conducted interviews by phone; interviews lasted from 15 to 60 minutes (average of 30 minutes for patients and 45 minutes for care managers). After each interview, field notes were written to capture the main topics discussed by patients and care managers, as well as any contextual information. All interviews were digitally recorded and then transcribed verbatim by a professional transcriptionist. We deidentified the transcripts and reviewed them for accuracy. The qualitative software package MaxQDA (VERBI GmbH) was used for data management and analysis. All data were stored on a secure network drive that was accessible only to the study team.
We used thematic analysis to guide our analysis of the interview data.65 We developed an initial codebook to be applied to both patient participant and care manager interviews, using topics from the interview guides and categories of barriers to care from the literature.66 Examples of these deductive codes include logistic barriers, such as transportation and financial difficulties, and attitudinal barriers, including the degree to which patients saw a need for treatment. As we reviewed field notes and transcripts, we added inductive codes, including severity of symptoms, family responsibilities, and other competing priorities. For each round of data collection, the study team met to receive training on the codebook and coding, led by Dr Walker. The team coded the transcripts by hand and discussed the codebook and application of the codes. All transcripts were double coded, with Dr Walker coding all interviews and the 4 members of the research team dividing the interviews. To examine the reliability of the coding, Dr Walker compared the coding between the 2 coders. Although a formal interrater reliability statistic was not computed, application of the codebook was similar across coders. When discrepancies occurred, Dr Walker made updates based on the description of how to apply the code in the codebook.
We wrote analytic memos to summarize the main information of each code and subcode and wrote about the relationships between codes. From these memos, we began to identify themes related to engaging in care at the patient, provider, and health care systems levels. We created matrices that summarized the information at the 3 levels, stratified by patient and care manager groups. Using the analytic memos and matrices, we examined the degree of convergence of themes across the patient and care manager groups and finalized the themes at each level.
We examined the findings from aims 1 and 2 to identify areas of convergence or divergence; this information informed the interpretation of the data described in the “Discussion” section. In particular, the qualitative data provided additional context about barriers to engagement that emerged in the quantitative results.
A total of 630 individuals were identified for potential study eligibility. Of these, 304 individuals were excluded, and 326 individuals provided informed consent and were randomly assigned. After participants provided written informed consent and were randomly assigned, signed consent forms were faxed by ED staff to the study team. Ten informed consent forms were not returned by ED staff to the study team; therefore, these participants could not be included in either the intervention or data analysis. Figure 1 documents the study flow.
CONSORT Flow Diagram.
In total, data from 316 randomly assigned participants were available to calculate baseline characteristics (Table 2a). The mean participant age was 35 years. In total, 54% were male, 53% were White, and 55% were uninsured. The most common diagnoses were major depression, psychotic illnesses, and bipolar disorder. Fewer than 25% had an outpatient mental health visit in the 6 months before study enrollment. Additionally, more than half lived in an area with a mental health provider shortage. Table 2b, which presents sample characteristics across sites, demonstrates variability in study recruitment yield and demographic characteristics across the study sites.
Study Participant Characteristics Overall and by Type.
Participant Demographic Characteristics by Study Site.
In bivariate models, participants assigned to professional care managers were significantly more likely to have an outpatient follow-up visit than were those assigned to peer care managers (55% vs 43%, respectively; P = .03; Table 3), representing a 1.28 increased likelihood of follow-up (RR, 1.28; P = .03). There was no significant difference between the care manager study groups for any secondary outcome measures, including ED readmission, inpatient admission, proportion of outpatient visits attended, or number of outpatient visits.
Bivariate Outcomes.
To better understand provider-level differences in performance, we examined the data by site, each of which had 1 peer care manager and 1 professional care manager (Figure 2). Peer care managers had a wider range of performance as measured by 30-day outpatient follow-up, whereas professional peer managers were more consistent. Although both groups had high performers, the peer specialist group also had several care managers with follow-up rates that were substantially below the mean. All 3 of the sites with the lowest-performing peers had peer turnover during the study period (Table 1 and Figure 2).
Site-Level Outcomes.
In a multivariable model (Table 4), several patient factors, including age >50 years (RR, 1.37; P = .005), female gender (RR, 1.15; P = .02), lack of insurance (RR, 1.17; P = .03), absence of a co-occurring substance use disorder (RR, 0.78; P = .03), and living in a nonmetropolitan area (RR, 1.65; P = .002), were associated with a greater likelihood of outpatient follow-up at 30 days (Table 5).
Multivariate Predictors of 30-Day Follow-up.
Multivariate Secondary Outcomes.
For the multivariable analyses for each of the secondary outcomes (Table 5), patients with a comorbid substance use disorder had a significantly lower proportion of outpatient visits attended (P < .01). Having a psychotic disorder was associated with a greater likelihood of a repeat ED visit (RR, 1.45; P < .01) and inpatient admission (RR, 2.5; P = .02), whereas having a diagnosis of bipolar disorder was associated with an increased risk of a repeat ED visit (RR, 1.44; P = .03). Greater distance from the patient's home to the nearest CMHC predicted a 1.38 times increase in likelihood of ED readmission (P = .01) and a 1.87 times increase in likelihood of inpatient hospitalization (P = .01).
The final sample for the qualitative part of the study comprised 30 patients (15 in each round of data collection) and 15 interviews with 13 care managers (7 interviews in 2018 and 8 interviews in 2019; 2 care managers took part in both rounds of data collection). The majority of patient participants identified as male (60%), and nearly all self-identified as Black (50%) or White (47%), with 7% reporting Hispanic/Latino ethnicity. Participants' average age was 38 years (range, 20-63 years). These demographic characteristics were similar to those of the full study population. The primary reasons patients were admitted to the ED were suicidal ideation or attempt, severe depressive episode or anxiety, or psychotic episode. A majority of patients (63%) attended an appointment at the CMHC within 30 days of ED discharge, which is a higher percentage than the full sample (49%). The qualitative sample included a smaller proportion of people who were uninsured (43%) than in the full sample (55%). Patient participants came from 7 of the 8 CMHC sites.
The care manager sample included 46% certified peer specialists and 54% mental health professionals (therapists, counselors, social workers). The majority of care managers identified as female (92%). Care managers represented 7 of the 8 CMHC sites (with the missing site being different from that of the patients). The participation rate of care managers was 50% (n = 7 of 14) in the first round of data collection and 38% (n = 8 of 21) in the second round.
Overall, patients and care managers reported similar barriers to and facilitators of transitions in care from the ED to the CMHC and subsequent engagement in care at the CMHC. Peer and professional care managers also described similar barriers to and facilitators of engagement in care, except where noted below. Patients often experienced several, often long-standing barriers to engaging with their care managers and other providers.
We found 2 main themes related to engagement in care at the patient level: (1) the degree of openness to receiving care and (2) the logistical challenges patients experienced in getting to the CMHC.
Both patients and care managers discussed patients' degree of openness to care as a factor in engaging with their care managers. Patients' receptivity to care depended on their experiences with stigma, seeing a need for treatment, and their readiness to address their mental disorder. Several patients did not seek mental health care before being admitted to the ED because they had not yet received a diagnosis, they thought they could handle the symptoms on their own, or they felt uncomfortable reaching out for help. One patient said, “I think I just kind of delayed it because … I thought I was at a point that I could kind of handle things on my own, but around that time, it was a lot stressful.” Patients and care managers mentioned the role of stigma, including from family members, as a reason that patients avoid seeking treatment. Care managers discussed that patients sometimes were not ready to “make a choice” to receive care. Patients who were open to treatment described the benefits of attending appointments at the CMHC. One patient noted:
I try to make sure that I go to every appointment, because it's something that really helps me…. And so I kind of make it almost like a religious thing to go to my appointments, because it helps me that much.
Several logistical barriers—including transportation, financial insecurity, severity of mental health symptoms, and other needs—presented challenges to patients' engagement in care. Patients and care managers most frequently mentioned lack of transportation as a primary barrier to transitions and engagement in care. One patient commented, “I wanted to meet [with my provider] more, it was just I didn't have the transportation.” Patients experienced difficulty accessing public transportation, particularly in rural areas, and in securing reliable rides from friends and family members; they also had trouble paying for gas or fares for bus or cab rides. Patients with Medicaid are eligible to sign up for rides from a Medicaid van that picks them up and brings them to the CMHC; however, the number of rides is capped. Patients said that lining up nonemergency medical transportation requires advance planning to contact the provider and schedule a ride to the clinic.
Other common financial challenges included being unemployed, not having insurance, and not being able to cover the cost of treatment. Several patients were not aware of payment assistance or sliding scale programs at the CMHC that could help cover the costs of appointments. One patient noted that lack of insurance and knowledge about payment assistance at the CMHC prevented her from attending an appointment after being discharged from the ED: “I didn't have no money during that time…. I didn't know until afterwards I could pay a dollar or two and they still would see me.”
Additionally, patients and care managers reported that patients' symptoms from their mental disorder or other physical conditions limited their ability to attend appointments. One patient said, “When you live with depression and … sometimes when you wake up in the morning … life tells you, you know, I'm not going to walk out the door today.” However, care managers explained that when patients attended appointments, they were more likely to remain engaged when they saw a reduction in symptom severity. Finally, other priorities, such as family obligations, unreliable childcare, inflexible work schedules, and insecure housing, precluded patients from engaging in care.
The 2 themes at the provider level centered on (1) the connection between the patient and the provider and (2) establishing and maintaining contact.
Building rapport, fostering comfort, and showing compassion were key ways that care managers could facilitate patient engagement. Early establishment of this connection through the warm handoff phone call aided in transitions of care. Patients said that the care managers made them feel that they were not alone and that there was someone who understood what they were going through. Once at the CMHC, the relationship with the care manager was a key facilitator for patients remaining engaged in treatment. One patient said:
I just felt secure. I felt like she was really there to help me and not hurt me, and keep going over those coping skills and coming up with different ones that I could do, especially when I have my anxiety attacks.
Similarly, care managers described the importance of building rapport and being compassionate with their patients, from the ED to the clinic.
Despite the importance of the connection between patients and care managers, the care managers reported having difficulty maintaining contact with patients. When attempting to connect with patients in the ED or soon after, care managers often experienced difficulty reaching them. One care manager said:
Most of the people I call, the phone's cut off, or the voicemail's not set up, and then once I do get a hold of [the patient], they'll make appointments and then they don't show up.
If patients came for an initial appointment after discharge from the ED, many would not return for subsequent visits. One provider noted with some frustration, “I just couldn't connect with them, yeah, calling at different times, and I just couldn't connect.” Additionally, care managers' ability to reach out to patients was constrained by their high caseloads and the time spent with other nonstudy patients. Mental health professionals described caseloads of 100 to 150 patients, whereas peer specialists reported smaller caseloads and more flexibility. Several care managers said that they made time in their schedules to follow up with study patients.
The main themes for health care system–level factors that influence patient engagement were (1) coordination between the ED and CMHC, (2) scheduling of appointments at the CMHC, and (3) available health care resources.
Care managers stated that coordination and communication between the ED and CMHC is necessary to facilitate transitions of care. However, they noted that this communication does not always happen. For example, care managers described not always knowing that a patient was in the ED or said that patients might show up at the CMHC after ED discharge without prior notification from the ED. Some care managers also serve as mental health liaisons in the ED; they help make the process run more smoothly by making appointments and communicating with the providers at the clinic. One patient who talked to a mental health liaison in the ED said that when they got home, “I felt hopeful. I mean, I was still in a dark place mentally, but I felt like I was being connected with people who could really help me.” Most patients received appointments at the CMHC before leaving the ED. Care managers noted that the clinics try to get patients in quickly, making appointments that range from the same day of discharge from the ED to within 3 days. Some CMHCs have walk-in appointments for first-time patients or patients coming from the ED.
Once a patient is enrolled in the CMHC, both patients and care managers said that subsequent appointments were generally easy to set up. As noted by one patient, “Oh, all I have to do is call and within a day or so I'll have an appointment.” Care managers said, however, that it can be a challenge for patients to navigate the number of appointments they have scheduled. Care managers described the balance between connecting patients coming from the ED with needed treatment and resources—which may involve seeing a psychiatrist, therapist, peer support specialist, and case manager—and the potential of overwhelming patients.
Finally, patients were not always aware of the resources available in the health care system that might alleviate some of their barriers to care. Care managers reported that their CMHCs have sliding scales and, often, medication assistance programs to help patients who do not have insurance and experience challenges paying for treatment. To increase awareness, one care manager said:
So we're pretty much stepping up our ante now and making sure that everybody knows what we offer, because there are services, they just don't take advantage of them just yet.
Patients and care managers described the multiple and long-standing challenges that patients often faced in trying to engage in care. As noted by a clinician, “They're usually in overwhelming situations. Their goal is mostly just to make it day to day.” Barriers tended to cluster together. For example, transportation issues, lack of insurance, and financial difficulties commonly converged for some patients. Another cluster of barriers included the difficulties of juggling work schedules and limited childcare. Several patients mentioned that poor physical health overlaid and compounded other barriers to attending appointments.
The study sought to better understand the factors predicting treatment engagement, including the relative benefits of having a peer vs professional care manager. We found several notable results. First, there was a statistically significant difference in rates of 30-day follow-up between peers and professionals, with participants assigned to professional care managers being significantly more likely to have successful transitions to outpatient care than those assigned to peer specialists (55% vs 43%; P = .03). Second, in multivariable models, a range of demographic (age, gender), clinical (mental health diagnosis and comorbid substance use), and geographic (rurality, distance from the participating CMHC) factors were associated with different measures of treatment engagement. Finally, the qualitative aim found many determinants of transitions at the patient, provider, and systems levels, including transportation challenges, patients' openness to receiving care, financial insecurity, and severity of mental health symptoms.
We discuss each of these findings in sequence.
Despite the reduced sample size relative to our original projections, we found a statistically significant advantage for the professional care manager group relative to the peer care manager group in 30-day outpatient follow-up. This result differs from those in previous published studies, which found that peers can serve in these roles as effectively as or more effectively than professionals. Several factors could account for this difference.
First, the training time to become a certified peer specialist is typically much briefer and more varied than the training time for registered nurses or licensed social workers. This greater consistency in the training of nurses and social workers may allow for more uniformity of service delivery. Peer specialists rely on their own lived experience of mental health conditions and working within the mental health system; this experience will, by definition, be more variable than will formal training provided to professionals who must meet professional standards for certification. This could confer advantages for patient-reported outcomes such as mental health recovery but may not translate into administrative outcomes, such as follow-up after discharge from the ED, that were captured in this study.
Second, peer care managers on average had greater rates of turnover across the study sites than did professionals. It is notable that all 3 sites with the lowest rates of outpatient follow-up in the peer care manager group had turnover of peer managers during the study period. Previous research has suggested that low salaries and lack of recognition can contribute to difficulties in recruiting and retaining peer specialists in community mental health settings.67 The findings from the current study suggest that addressing high rates of turnover in the peer workforce could be important for optimizing peer specialists' potential to improve engagement in treatment.
Third, differences in scope of practice and supervision requirements across sites may have particularly impacted peer care managers' ability to link participants to outpatient services. For instance, at several sites, peer care managers were not able to directly schedule patients and were not allowed to accept appointments until the patient had an intake appointment with another licensed provider. This highlights a difference between comparative effectiveness workforce studies and comparative effectiveness studies of more uniform treatments, such as medications or surgical procedures. For workforce studies, financing, supervision requirements, and scopes of practice provide critical context for framing and interpreting comparative effectiveness questions. Future research should better describe these barriers and strategies for overcoming them to make optimal use of different provider types in facilitating transitions of care.
A range of patient-, provider-, and community-level factors predicted rates of treatment engagement, and the factors that predicted performance on each of these measures varied. The variability in which predictors are associated with these different outcome measures suggests that these constructs may be capturing distinct dimensions of engagement. Although timely follow-up after discharge from the ED is one of the most widely used behavioral health performance indicators,59,68,69 the current study's findings suggest several caveats in using this measure as the sole proxy for treatment engagement. It also suggests the value of complementing administrative measures of treatment engagement with patient-reported measures.
Several predictors of 30-day follow-up after an ED visit were consistent with the literature, whereas others ran against our expectations. Consistent with earlier studies,39 patients aged ≥50 years and female patients were more likely to follow up, suggesting that they may be easier to engage in mental health care than are younger and male patients. Those with substance use disorders were less likely to have 30-day follow-up, highlighting the challenges of engaging this group in care, and suggesting the importance of targeting engagement in care toward those with comorbid mental health and substance use disorder diagnoses.
Surprisingly, uninsured patients and those in nonmetropolitan areas had higher rates of 30-day follow-up than did those in the comparator groups. It is possible that this may reflect a lack of other community treatment options for these patients—for instance, those who are insured could follow up with private providers who were not captured in the all-payer data. Further research is needed to see whether these findings generalize to other settings.
Living a greater distance away from the patient's CMHC was the most consistent risk factor for problems in the secondary measures of treatment engagement. Geographic factors can represent substantial barriers to care, particularly in rural areas. The finding is consistent with reports from the qualitative interviews about transportation challenges playing a major role in obtaining outpatient services. Having severe psychiatric illnesses and comorbid substance use conditions was associated with lower performance on these treatment measures, which may reflect greater complexity or chronicity of the underlying illnesses.
The qualitative findings identified many barriers at the patient, provider, and system levels that present challenges for mental health care. In reference to the CARE program, both patients and care managers described care managers as helping patients address logistical challenges to obtaining care by scheduling appointments, facilitating transportation, and linking patients to social support services. However, care managers noted that clinic-wide, patients did not always know about the resources available at the CMHC. Therefore, efforts to inform all patients about resources may be important. The findings also pointed to the central role of the patient-provider relationship with care managers in the initial care transition and ongoing engagement in treatment. Patients reported having multiple, simultaneous challenges that had a synergistic adverse effect on their ability to engage in treatment. Care managers will typically need to engage challenges at multiple levels to achieve successful care transitions. At the health care systems level, the findings point to the importance of clear channels of communication between EDs and CMHCs.
Our experience in implementing this study can provide lessons for future studies implementing ED-based transition-of-care studies. First, recruitment in our study lagged throughout the project and varied across study sites. The high volume and competing demands at ED sites can make it difficult to rely on ED staff for study referrals. Buy-in from ED staff, ideally with ED staff at local sites as investigators on the study team, could have been helpful for optimizing the number and representativeness of participants. Having an in-person presence could also have been beneficial in tracking and retaining participants in the study over time.
Second, clinicians in community mental health settings have high caseloads and limited excess capacity, which required the study interventionists to take on new patients on top of their existing caseloads. Turnover may be high, particularly among peer providers. Careful consideration will be needed to balance providing a robust intervention with the desire to ensure generalizability to real-world settings and conditions.
Finally, successful transitions of care from the ED to outpatient treatment require ongoing coordination between multiple organizations, including the ED leadership and staff and outpatient clinics, which may be hampered both by a lack of communication channels and privacy regulations, such as HIPAA rules. They also involve social service agencies addressing housing and transportation barriers to care. Active, ongoing engagement is needed with all of these entities to troubleshoot challenges and ensure continuity of recruitment, study interventions, and data collection.
The study's findings should be understood in the context of several limitations. First, the study did not include a control group; thus, it was not possible to draw definitive conclusions about the overall effects of the care management intervention compared with usual care. Second, it was conducted in a single state (South Carolina), and differences in training, licensing, or scope of practice could raise issues in generalizing to other states or systems of care. Third, the sample size was lower than was originally projected due to challenges in accrual, with varied recruitment rates across the study sites. Although we did find a significant effect for the main study outcome, there is the possibility of an elevated false-discovery rate given the reduced statistical power.14 Therefore, these findings should be considered preliminary and warrant replication in future research. Fourth, low rates of follow-up contact made it infeasible to examine the patient-reported outcomes. Fifth, because sites each had 1 peer care manager and 1 professional care manager, and because of variability in how sites used their peer care managers, it is not possible to determine whether variability in keeping at least one 30-day follow-up clinic appointment (Figure 2) was driven by site- or provider-level differences. Finally, the sample for the qualitative study had a higher rate of keeping at least 1 CMHC visit appointment after ED discharge than did the full sample; therefore, our qualitative findings may not fully represent the experiences of patients who did not engage with the CMHCs.
The study findings and limitations highlight opportunities for future comparative effectiveness workforce studies on this topic. Perhaps most notably, future research should seek to better disentangle differences in quality or outcomes of care that can be attributed to differential training from differences due to varied scope of practice, turnover, and regulatory factors. Understanding and optimizing local scope-of-practice laws and improving staff retention practices will be important for effectively using peer care managers to facilitate transitions of care after ED visits for mental health problems.
By identifying patient-, provider-, and system-level barriers, this study highlights the importance of having support services provided in the transition process after ED discharge for mental health visits. On average, professional care managers had higher rates of follow-up;, peer care managers had higher rates of turnover and variability in performance across individual providers. Further research is needed to help identify and overcome barriers to optimizing care transition interventions across a spectrum of provider disciplines.
Research reported in this report was funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award IHS-1510-32431). Further information available at: https://www.pcori.org/research-results/2016/comparing-two-ways-help-patients-get-follow-care-after-mental-health-visit
Qualitative Interview Guides (PDF, 160K)
Druss B. Lally CA, Li J, Tapscott S, Walker ER. (2021). Comparing Two Ways to Help Patients Get Follow-up Care after a Mental Health Visit to the Emergency Room—The EPIC Study. Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/05.2021.IHS.151032431
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
