Coordinated Specialty Care Discharge, Transition, and Step-Down Policies, Practices, and Concerns: Staff and Client Perspectives

First-episode psychosis (FEP) is a serious mental health condition with historically poor long-term outcomes, particularly for young people with nonaffective psychosis (1, 2). Fortunately, the development of specialized services for early intervention in psychosis (EIP)—widely referred to as coordinated specialty care (CSC) in the United States (3, 4) has substantially improved 2- to 3-year postonset outcomes (5–9). However, both premature disengagement (10) and longer-term postdischarge erosion of the advantages of EIP/CSC participation relative to status quo services remain significant concerns (8, 11–14). In particular, concerns about postdischarge outcomes and uncertainty regarding optimal length of EIP/CSC services have prompted a series of recent “extension” trials (15–17) as well as international dialogue about the long-term impact of variables such as duration of untreated psychosis and the quality and intensity of EIP/CSC program components focused on functional and vocational recovery in addition to symptom remission (18–21).

To date, studies of EIP/CSC client discharge outcomes and trajectories have been limited in number and predominantly conducted in the context of programs located outside the United States. This is particularly true with respect to clinicians’ and clients’ perspectives. In fact, to the best of our knowledge, we are aware of only three qualitative studies—two studies based in the United Kingdom and focused on clients’ transition experiences (22, 23) and one study of staff perspectives on discharge (24). In the United States, the heterogeneity of state and regional funding and support for mental health services (25), fragmentation of health care more broadly (26, 27), and diversity of EIP/CSC models, programs, and standards (3, 28) all raise questions regarding the circumstances and consequences of program duration and discharge practices that international studies can only partially address.

The analyses reported in this article focused on the subjective views of two key stakeholder groups: CSC program staff and current and former CSC clients. Primary research goals for staff interviews were to better understand the national landscape regarding discharge and transition, the range of transition strategies that diverse CSC programs have developed, and concerns regarding transition. Primary research goals for client interviews were to explore how recipients describe their communication with CSC clinicians about discharge and their concerns and perceived readiness for discharge.

Methods

Design, Sampling, and Procedure

The research described in this article is part of a larger, longitudinal, mixed-methods study evaluating the implementation and outcomes of CSC services that are supported at least in part by Mental Health Block Grant (MHBG) 10% set aside funds (29). Additional details regarding study methodology and design are provided in an online supplement to this article.

Study sites.

The study includes 36 sites implementing services with varying levels of fidelity to the CSC model, selected from among more than 200 CSC programs across the United States. Sites were selected in collaboration with federal sponsors and were chosen to ensure diversity in geographic distribution, specific program model (OnTrack, EASA, NAVIGATE, etc.), and urbanicity.

Recruitment and interview procedures.

Data presented here were collected during two visits to each site, conducted between January and June 2018 and between January and May 2019. Site visits included semistructured interviews with CSC team members and administrators and between one to four CSC clients per site. Clients were selected by local program staff without imposition of inclusion or exclusion criteria, and demographic data collection was limited to age, gender, race-ethnicity, and length of time in CSC services. All interviews were conducted by experienced mental health services researchers, including several authors (TD, PG, AR, and DS).

Interview protocols were developed by the MHBG evaluation team, incorporating feedback and suggestions from governmental sponsors, expert advisors, and partners. Questions covered a range of areas related to CSC program implementation, including explicit questions regarding transition practices—namely, Does the program have a step-down program or any services for clients after they have participated in the program? Has the program had any clients “graduate” or leave the program over the past year, and if so, did they leave because they were doing well, because they hit some type of eligibility limit, or some other reason? Was there anything that the program did to help with the transition? Does the program have any contact with clients who have left the program?

For clients, the two primary questions asked were, Is there a set length of time that the participant will be enrolled in the program? and What does the client think it will be like when he or she is no longer part of the CSC program? If participants were nearing the end of their time in the program, they were also asked how they felt about the idea of moving on (e.g., whether they felt ready) and whether anyone was helping them connect with services.

Sample characteristics.

For providers, individual demographic characteristics were not collected. A total of 338 CSC team members and administrators participated in interviews across the 36 sites and two time points. At each site, between three and 11 staff and administrators participated in each interview, with participants potentially including care managers or case managers, nurses, psychiatrists, supported employment and education specialists, clinicians, support staff, peer specialists, and individuals in a range of administrative roles. For the discharge and transition sections of the interviews included in this report, discussions and comments from a total of 108 staff were actively coded (i.e., in many interviews, additional staff were present but did not actively weigh in or speak about these focus issues, and thus they did not provide material that could be coded).

Eighty-two clients (45% female, N=37) at 34 sites responded to questions about transitions. Clients were Caucasian (57%, N=47), African American (36%, N=29), Asian American (5%, N=4), American Indian/Alaska Native (1%, N=1), and individuals identifying as two or more races (1%, N=1), with 21% (N=17) reporting Latino or Latina ethnicity. The mean±SD age of client respondents was 22.5±4.46 (range 14–33). The length of time enrolled in the CSC program at the time of the interview ranged from 1 month to 4 years, and six participants were either close to discharge or had already been discharged. Two of the discharged clients, although they had transitioned from initial CSC services, were actively enrolled in an early psychosis–specific step-down continuation program at the time of the interview.

Data Analysis

Using interviews across both data collection time points, we extracted all material containing responses to protocol questions concerning transition and discharge as well as additional discharge-relevant material arising in the context of other protocol topics. To verify that all text relevant to transition was captured, systematic key term searches were also conducted—i.e., searches for key terms such as step-down, transition, graduation, discharge, and other related words—and the surrounding narrative was reviewed in order to verify that no relevant text was missed. Extracted sections were then coded by using systematic content analysis (30, 31), with the research goal of characterizing all information relevant to transition and discharge.

A priori codes were generated on the basis of both interview protocol questions and the team’s understanding of discharge-related CSC issues (e.g., use of discharge planning tools). Additional (emergent) codes were generated through the comprehensive coding of approximately a third of participant transcripts (until code saturation was reached). In some cases, uncertainties in the interviews prompted additional triangulation work. For example, ambiguous responses to questions regarding state or regional discharge or length-of-service policy prompted the research team to review existing administrative policy (including manuals and policy documents where available) or to clarify policy with appropriate administrative leadership.

Following a multiphase, iterative code development process, the combination of a priori and emergent codes was piloted on additional transcripts prior to finalization by the coding team (NJ and BG). Because we wished to ensure code reliability, 15% of transcripts were blind double-coded; code disagreement was minor (less than 10%), primarily consisting of missed codes rather than disagreement about correct application of existing codes. The qualitative software program Atlas.ti (32) was used for all coding and initial analysis, and codes were exported to quantitative software (Stata [33] for clinician data and SPSS [34] for client data) for the computation of frequencies.

For staff interview analyses, we sought to provide both code frequencies, where appropriate, and narrative reporting intended to characterize the nature and range of participant perspectives. Wherever thematic frequencies are provided, a denominator is included, with the denominator varying relative to the number of interviews in which material relevant to the code in question was present. For client interviews, frequency reporting includes both overall frequencies as well as further breakdowns by length of time in CSC services at the time of the index interview. This breakdown is reported because of the very different implications of, for example, perceived lack of understanding of discharge policies after only 2 weeks in CSC versus 2.5 years. For these purposes, clients were grouped into four service length categories: enrolled for 0–12 months, enrolled for 13–23 months, enrolled for ≥24 months, and at discharge or already discharged.

Results

Staff Themes

Codes developed for the staff interviews were grouped into three higher-level domains, reported as subsections below: policies and practices related to discharge and transition, discharge strategies, and clinician experiences and concerns.

Policies and practices related to discharge, transition, and step-down.

Table 1 summarizes policies regarding planned or maximum program length for all 36 CSC sites. A 2-year target, with flexibility to extend services, was the modal response. Variations in flexibility were considerable, with some programs able to extend services for a few weeks or months, whereas others were able to extend in fixed 6-month increments. Several sites noted greater flexibility when the program was below census. Three of the 36 programs were described as abandoning an initially rigid 2-year limit and adding flexibility in response to clients’ extended service needs (“It's not 24 months anymore. It was. We were limited at first. But now since the research is starting to show that people are benefiting from 3 to 5 years of service, we're keeping people longer.”). Conversely, three programs that initially allowed services beyond 2 years subsequently established a 2-year limit to more strongly emphasize a transitional focus.

Only one of the 36 programs was described as having a time-limited effort to assess postdischarge outcomes through informal qualitative interviews with two client cohorts. Staff at one additional site mentioned quasi-systematic follow-up to ensure that clients had successfully engaged with other services (“We check in [with discharged clients] every once in a while to see how things are going.”) Although staff of other programs described occasional informal contact with selected discharged clients—seeing clients at agencywide events, for example, or at CSC program–hosted reunions—these interactions provided, at best, a very partial window into postdischarge trajectories.

Discharge and transition strategies.

Although staff at only two programs described early psychosis–specific postdischarge services, sixteen were able to refer clients to within-agency programs of varying intensity. In six of these 16 cases, programs were also able to facilitate ongoing treatment from one or more members of the client’s CSC team. For example, one staff member said, “A lot of our clients once they discharge from the program continue to see the exact same prescriber and therapist. . . . It just wouldn’t be as part of the [CSC] program anymore.” Another staff member noted, “When we say ‘graduate,’ a lot them will continue to see the [same] psychiatrist for medication, but that might no longer be a part of the program because they will only be receiving one service.”

Although continuity predominantly centered on therapists or psychiatrists, staff at one program described the ability and capacity to provide indefinite supported education and employment services for clients who had graduated or been discharged. Of the 16 programs that were able to refer clients to within-agency programs, all nevertheless described discharging clients to outside services at least some of the time. In some cases, this was based on client needs or the availability of appropriate within-agency services; for example, one agency could place transition-age youths internally but not clients who were age 25 or older.

CSC staff described a range of strategies and tools designed to ease the transition. Three notable examples are described here. First, some programs used specific structured or semistructured tools or checklists, completed with clients prior to discharge. Some of these procedures included advance crisis planning or advance directives and collaborative development of a plan for accessing diverse postdischarge services across service domains (housing, medications, and employment). Second, some CSC staff described in-person linkage and hand-off prior to discharge—i.e., CSC clinicians would accompany clients to initial meetings with prospective postdischarge clinicians or programs, ensuring client satisfaction with new services prior to finalizing their discharge. Third, a number of teams described a systematic step-down process prior to discharge, typically described as spanning 4 to 6 months, in which CSC services were gradually tapered in order to facilitate increasing autonomy and self-reliance as well as adjustment to a lower level of clinical support. For example, a program might slowly reduce the frequency of therapy or case management or taper off certain program components completely. Virtually all CSC teams remarked on the highly heterogeneous nature of FEP populations. Transition-relevant aspects of heterogeneity discussed included the need for diverse postdischarge services—for example, services designed to support both Medicaid and privately insured clients, as well as a mix of residential, higher-intensity, and lower-intensity services. Many staff noted, for example, that some clients might need and want only sporadic medication management after discharge, whereas others would require therapeutically intensive residential or day programs. In addition, many clinicians emphasized variations in available family support and social capital as having an impact on discharge planning and client prognosis. Heterogeneous responses to treatment affected service duration. Providers at multiple sites described a subset of clients with transient or highly treatment-responsive symptoms who could be effectively discharged after as little as 6 months, whereas others, even after 3 or 4 years, continued to experience debilitating symptoms or functional limitations that required intensive ongoing care.

Clinician discharge experiences and concerns.

Most program staff expressed at least some concerns regarding postdischarge services and client trajectories. These concerns were place specific and involved strengths and weaknesses of programming in their region. Staff from several programs provided anecdotes of clients who had been doing very well when they left the program only to relapse. Staff at one program described having concerns with clients’ well-being and access to services after conducting an informal effort to contact several clients after discharge:

Table 2 summarizes other specific areas of concern, often based on cases in which the program had been alerted to postdischarge outcomes for selected clients.

Although concerns of various kinds were common, with many teams expressing an overall preference for a service period significantly longer than 2 years, others emphasized the importance of keeping CSC programs transitional in order to prevent potentially harmful “engulfment” in a patient role. Three sites reduced their service length for this reason. Other staff, speaking in this vein, emphasized differences between ongoing long-term medication management (which might be desirable or necessary in the long-term) versus therapy. For example, one staff member noted, “I don't think therapy should be lifelong. It should be one of your goals, but to set you in that direction. With this program, [CSC], let's make sure you're well set in dealing with your symptoms. . . . Set a foundation for those goals and let you go on your way.”

Client Themes

Client themes were grouped into four high-level domains: client-provider discharge planning, client discharge awareness, perceived readiness for discharge, and concerns and feelings about discharge. Table 3 summarizes higher-level themes within each domain as well as lower-level codes under each theme, including example quotations. Table 4 provides the number of interviews in which each domain was mentioned and the frequency with which specific codes were discussed, with a further breakdown by length of time spent in CSC services at the time of the index interview.

Overall, client reports of team communication regarding program time limits and discharge planning were quite varied, as were feelings regarding readiness for discharge and associated concerns. Even among those already served for 2 or more years or approaching discharge, a substantial minority reported fears and discomfort—for example, feeling unprepared or distressed—and concerns about the implications of losing valued supports and structure. One participant, for example, compared current CSC staffing to projected future services: “Any doctors [in standard treatment] . . . they're more, I don't know, they see their patient. Here they have a lot of resources like occupational therapy and all that. And there it [will be] more just like talking to the doctor for 20 minutes and then going home.” Both of the clients who were interviewed after discharge were currently being served in a specialized early psychosis extension program, and both planned to remain in services for several more years (after having already received 2 or more years of initial CSC).

Although loss of program structure clearly undergirded many clients’ discomfort, attachment to particular clinicians—and the distressing prospect of losing these relationships—were frequently mentioned: “I told them that I didn't like the fact that I wasn't going to be able to reach out to them anymore after the program. . . . I didn't like the fact that after my time here, like after the 2-year mark, I can't reach out to [team members] because of confidentiality reasons. . . . I don’t like that.” Many clients also appeared to view discharge from CSC as entailing the permanent loss of access to close therapeutic relationships, whether because of cost, access, or standards of care.

Almost half (43%) of the 30 clients who were explicitly asked about provider communications regarding future discharge reported that they were unaware of discharge policies and had not discussed discharge with providers, representing ten of the 36 evaluation sites. Among these clients, a quarter had already been served for more than a year and 38% for 2 years or more. In nearly all cases, providers working in the programs in which these clients were served described early and proactive discussion about discharge.

Other instances of discordant perspectives between providers and clients were also present. Clinicians, for example, noted instances in which clients might feel prepared for discharge but in fact were not or, conversely, feel unprepared when, from the providers’ point of view, they were ready. One of the providers involved in the cohort follow-up effort described above noted, “[Some former clients] said they were fine, and now I'm seeing them in the home and they are not fine. . . . They're saying, ‘Everything's great, I'm great, I'm doing wonderful. I just started a business.’ And then their parents are like, ‘Yeah, they made that business up. It's not a real thing’”

Discussion

Main Findings

These findings underscore the tremendous variability of discharge practices and policies across the study sample, including a wide range of program lengths (between 2 and 5 years), and an array of discharge strategies and internal capacity for continuity of care. Only two of the 36 sampled programs provided targeted postdischarge extension or step-down services, and only one was described as having regular, clinically oriented follow-up to ensure (or assess) the success of the transition. Concerns about discharge and the accessibility and quality of postdischarge services were widespread among clinicians, and a significant proportion of clients expressed apprehension or sadness about discharge, including clients already served for more than 2 years or at the point of discharge. Data also highlight the heterogeneity of client needs and experiences, with implications for ongoing improvement and refinement of CSC and post-CSC services.

Clinical Implications and Future Directions

The overarching goals of early detection and early intervention include a fundamental redirection of the historically poor prognosis of individuals with early psychotic disorders (35–37) and the transformation of initial systems of care (38). Follow-up research (8, 10–14) and associated step-down and extension trials (15–17) have raised important concerns about the durability of the initial advantages of CSC over conventional treatment and the best strategies for sustaining improved long-term outcomes. Although many practice guidelines and performance-monitoring efforts include at least some guidance regarding transition (39–42)—in some cases providing genuinely substantive advice (39)—empirical substantiation of best practices remains scarce (43). Our findings underscore the absence of a unified national strategy for CSC transition in the United States and, presumably, other countries with distributed CSC services. Our findings also highlight the role of transition and postdischarge services and the importance of incorporating explicit service quality metrics related to discharge into future follow-up studies (43).

Overall, more research on transition and discharge is essential, ideally including comparative effectiveness studies designed to discern which of the many strategies discussed in our interviews (and the broader literature [22–24]) should guide clinical policy, including the development or refinement of fidelity tools. Ideally, in this context, practice-oriented research initiatives would also make an explicit commitment to unpacking the implications of national, regional, and client-level heterogeneity and how best to craft and finance models that preserve the initial benefits of CSC programs while also titrating supports to foster independence. Key considerations include local non-CSC service availability and financing and, at the client level, age (youth versus adult), degree of disability, and relative socioeconomic disadvantage.

Discrepancies between client and provider perceptions regarding communication and program policies related to discharge tentatively suggest that even in cases in which CSC programs are able to articulate well-developed policies and practices (in interviews), program clients may remain confused or unclear about what to expect. Future research and program development efforts would thus ideally include a strong focus on understanding clients’ subjective experience not only of discharge itself but also their advance knowledge of and preparation for discharge. Integration of participatory methods may be critical. Differences between provider and client perspectives—especially insofar as they may affect program-level decisions to extend care or move forward with discharge—also raise important questions with respect to shared decision making and client self-determination, which feature as key components of CSC programs (32). Who, we might ask, for example, should ultimately drive decisions regarding the timing of discharge?

A second important issue is the extent to which ongoing improvements to CSC programs—within a standard 2- to 3-year treatment window—could potentially affect long-term outcomes rather than (or in addition to) the extension of services or development of postdischarge step-downs. Although we lack follow-up outcomes data, potential approaches in the published literature include ramped-up supported education and employment services (18, 44) and CSC modifications designed to improve outcomes for high-risk subgroups, including youths and young adults experiencing or at risk of homelessness (45) and those identified early in CSC as likely to disengage from services (46). Conversely, we might argue that CSC represents a level of quality that should in fact characterize mental health services in general, as underscored by both staff and client concerns regarding the values and offerings of postdischarge services.

Finally, as evidenced by both clinician and client interviews, the tension between helpful structure and support and a potentially detrimental dependency on services can be challenging to navigate. For example, as we described, a large number of clients alluded to the value of therapeutic relationships and their desire to continue rather than move on from these relationships. Philosophical and ethical tensions surrounding the trade-offs and impact of long-term versus short-term, time-limited therapy have been a subject of debate and disagreement for many years (47–49), albeit with contemporary insurance policies strongly favoring (if not requiring) time-limited therapeutic interventions (50). Pronounced differences between clients in the level of ongoing disability or distress—and the time needed to lay the foundation for recovery—further compound the challenges of crafting policies and supports capable of meeting diverse short-term as well as longer-term needs. An alternative treatment model, such as is used in the Housing First model of supported housing (51, 52), might continuously titrate services to an individual’s wants and needs for an unlimited period rather than transitioning the client to alternative services. This would require staffing models different from those that are currently used, and these models would have to strive to help clients achieve as much independence from services as possible. Again, research concerning these issues would ideally be grounded in a careful consideration of multiple stakeholder preferences and perspectives (53–55) and oriented toward the investigation of mechanisms whereby interventions can facilitate long-term healing as well as short-term effectiveness (56–59).

Limitations

Major strengths of the analyses presented here include the number of programs and states included in the evaluation sample, and the projects’ large multistakeholder sample. Nevertheless, questions and discussion about discharge and transition constituted a relatively small piece of a comprehensive and multifaceted evaluation. In any given interview, transition-specific discussion may have spanned as little as 5–10 minutes, falling short of the depth of coverage that would have been enabled by a research project more exclusively focused on the topic of discharge and transition. Furthermore, depth of discussion varied considerably across interviews, and a significant proportion of interviews did not include discussion of important thematic domains. Although the number of client participants (N=82) was high overall, interviews included only one to four clients per site, who were selected by program staff, raising questions about potential selection bias. Similarly, the sites selected do not necessarily represent the much larger population of U.S. CSC services, and the overarching motivation for the study (federally sponsored evaluation of CSC services) may have biased provider narratives. Future efforts to unpack client experiences and priorities would ideally use careful sampling strategies to capture the range of experiences likely present even among current or former clients within the same program and would also ensure adequate representation of participants with varying levels of disability, premorbid accomplishments, socioeconomic backgrounds, and amount of family support.

The MHBG 10% Set-Aside Study and the data reported here also did not include family or caregiver perspectives. Given the critical role that families often play in supporting loved ones with psychosis (60, 61), future efforts would ideally fully integrate family perspectives and priorities with respect to discharge policies (and communication of those policies) as well as transition and postdischarge supports. Another critical but unrepresented group were clients who opted to withdraw or disengage from CSC services (62, 63). From a population health perspective, the ultimate success of CSC depends not only on sustaining the gains of those who opt to remain in CSC programs for their full duration but also on working to ensure the sustained well-being of those who disengage from CSC programs in their current form.

Conclusions

Internationally, concerns regarding the relative loss of initial treatment advantages of early intervention services are widespread. As our research documents, U.S. clinicians, both academic and community-based, share these concerns, as—in more indirect ways—do many CSC clients. Although programs have developed a range of strategies for supporting clients during and after discharge, implementation varies considerably, and research capable of guiding best practices is lacking. Moving forward, multistakeholder-informed intervention and applied services research on discharge, transition, and step-down should be prioritized.