Across multiple countries, programs serving individuals with early psychosis have been found effective (
1). Some countries (e.g., United Kingdom) have adopted top-down standardized models, whereas in the United States, several states have allowed for a bottom-up approach. In 2004, California passed the Mental Health Services Act (MHSA), which established specific mental health services funding, including Prevention and Early Intervention services. Funds are distributed at the county level with autonomy in how they are allocated. Many counties used these resources to develop early psychosis programs. Rather than implementing one treatment model across the state, California counties are permitted to adopt different evidence-based early psychosis care models, with the ability to modify some program details to appropriately address local needs. In 2014, state mental health block grant funds were allocated across the United States for early psychosis services, leading to the development of early psychosis programs using a variety of evidence-based treatment models that were executed at a state or local level. Although California has served as a precursor to the national expansion of early psychosis programs in the United States, its county-driven mental health system led to implementation of diverse programs with little top-down coordination.
For U.S. programs, evidence-based treatment components include broad community-based outreach with rapid referral to reduce duration of untreated psychosis (
2), comprehensive assessment to determine eligibility, and team-based coordinated specialty care (CSC) (
3). Treatment includes case management, ongoing psychiatric or medical assessments and treatment, client and family psychoeducation and psychotherapy, educational and vocational support, and relapse prevention. Most programs provide services to individuals who recently developed a psychotic disorder or those at clinical high risk of psychosis to reduce the likelihood of developing full psychosis.
Early Psychosis Programming in California
Though founded on existing evidence-based treatment models, California counties have discretion in how they implement their individual early psychosis programs. This approach allows counties to tailor services to the needs of the local population and the resources available (
4). Although such customization may be practical for individual programs, this lack of consistency could dilute the measurable impact of these programs on client outcomes (
5–
10). A similar issue exists at a national level, given that individual states or local jurisdictions have chosen to implement early psychosis programming out of a variety of potential models (
4). CSC is effective in improving outcomes in early psychosis (
11,
12). However, it is not clear which particular components are key to improving outcomes. Additionally, although recent studies suggest that it is feasible to implement CSC in clinical practice (
13), it is unclear how effective the intervention is when delivered in this setting, as opposed to within the more structured environment of a clinical trial. Evaluating the effectiveness of CSC across a range of existing heterogeneous community programs for early psychosis represents an important step toward determining the effectiveness of the model in standard clinical practice. Exploring the impact of program-level differences across these services may help to identify which particular components of care are key to improving specific outcomes. However, before an evaluation can be conducted, it is critical to understand the composition of the programs that may be included, their data collection practices, and the nature of the heterogeneity between programs.
This article provides a descriptive summary of California’s early psychosis programs, including the composition of program services, funding sources, data collection practices, inclusion criteria, and use of data collection system (e.g., electronic health records). Given recent interest in harmonized data collection and coordination for early psychosis programs nationally (
14), this represents a necessary first step in developing an evaluation approach for the state’s complex landscape of early psychosis programs.
Methods
Design
From May to October 2016, active early psychosis programs were identified through a multiphase process that included review of mental health and county program Web sites, MHSA plans, Substance Abuse and Mental Health Services Administration (SAMHSA) mental health block grant applications, and stakeholder feedback. The MHSA coordinator in each county was contacted to confirm the existence of an early psychosis program and identify a program or county representative.
In October 2016, early psychosis program representatives were contacted via e-mail with a project overview and participation request, followed by a link to the California Early Psychosis Assessment Survey (CEPAS). In counties with more than one program, representatives were asked to complete separate surveys to capture the nuances between programs. If the representative failed to respond after 2 weeks, three courtesy calls were administered and additional reminder e-mails were sent to encourage survey completion. Once the surveys were completed, representatives were contacted to clarify vague responses, resolve discrepancies in the data, or resubmit missing data through May 2017. This evaluation was reviewed and approved by the University of California, Davis Institutional Review Board.
CEPAS
The CEPAS is a structured online survey designed to gather information about early psychosis program characteristics and the nature of the data collected (see Appendix 1, which is available as an
online supplement to this article). The First-Episode Psychosis Services Fidelity Scale (FEPS-FS 1.0) (
10,
15,
16), a standardized measure of fidelity to program best practices (see Supplemental Methods in the
online supplement), was integrated into the CEPAS to assess both the presence and the absence of treatment model components. Additionally, the CEPAS includes multiple-choice and open-ended questions on the following areas: client age, diagnoses served, outcomes data collection methods, program funding sources, program outreach methods and family involvement, program treatment components, pharmacotherapy options offered, administrative program components (e.g., staff-to-client ratio, types of staff employed), use of Early Psychosis Clinical Services PhenX toolkit measures (
17), challenges or barriers to program implementation, and opinions on each component of early psychosis care described within the FEP-FS.
Prior to CEPAS distribution, local stakeholders including early psychosis program managers, MHSA staff, and clients with lived experience reviewed the scale to confirm that items were understandable and captured the necessary data. Results of the presence or absence of treatment components are reported here, whereas program ratings of the importance of these components are reported elsewhere (
18). FEPS-FS items were scored in a present or absent manner for 23 of the 30 items assessed. The exceptions were items 21–27, in which a FEPS-FS score of 3 or higher was scored as endorsing the item (see Table 1 in the
online supplement for a description of the endorsement criteria for each item).
Results
Across 58 California counties, 24 (41%) reported having at least one active program for treatment of early psychosis, with five counties reporting multiple programs. Twelve counties (21%) reported having programs in development, and 22 counties (38%) reported no early psychosis program. Of the 30 active programs identified (
Table 1), 28 programs (93%) provided complete data on the CEPAS, one county provided partial data, and one county did not provide data. The final analysis includes the 29 programs that provided complete or partial data on the CEPAS (
Figure 1 in the
online supplement).
Client Population Characteristics
Program-level details are reported in
Table 1. Of the 29 programs, 22 (76%) serve both clients with first-episode psychosis (FEP) and clients at clinical high risk of psychosis, five (17%) serve FEP clients only, and two (7%) serve CHR clients only. Twenty-five programs (86%) reported serving clients diagnosed as having a
DSM-IV schizophrenia spectrum disorder, whereas 86% serve those having any psychotic spectrum disorder. Twenty-one programs (72%) serve clients diagnosed as having mood disorders with psychotic features (e.g., major depressive disorder with psychotic features, bipolar disorder with psychotic features), and six programs (21%) serve clients with a diagnosis of mood disorders without psychotic features. One program reported that it provides services to clients diagnosed as having mood disorders only if they meet criteria for bipolar disorder I, with or without psychotic features.
The most common reason for ineligibility for services was intellectual disability (N=22 programs, 76%), followed by a diagnosis of a substance-induced psychotic disorder (N=19, 66%). Eighteen programs (62%) excluded individuals if they are not county residents, and 13 programs (45%) exclude individuals because of substance dependence. Almost all programs provide services to uninsured clients (N=25, 86%) or undocumented clients (N=23, 79%). Twenty-two programs (76%) provide services to privately insured clients, whereas only two programs (7%) do not serve any of these types of clients.
Characteristics of Program Services and Model Elements
Twenty-eight programs provided data regarding number of eligibility evaluations, with a median of seven individuals per program receiving evaluations per month (mean±SD=11.45±11.54, interquartile range [IQR]=4–15, range 0–40), yielding approximately 84 clients per program per year. All 29 programs provided data regarding number of individuals engaged in ongoing treatment services per month, with a median of 35 individuals per program receiving services per month (mean=50±62.10, IQR=18–50, range 2–300). Sixteen programs (55%) reported the target duration of services was up to 2 years. Five programs (17%) reported a target treatment duration of 1 year or less, four programs (14%) reported a target duration of 3 years, and one program reported a target duration of up to 4 years (3%). Three programs (10%) reported treatment was available indefinitely based on need.
The most frequently adopted CSC model was Maine’s Portland Identification and Early Referral (PIER) model (N=6, 17%) (
19,
20), followed by the Felton Institute Prevention and Recovery in Early Psychosis model (N=5, 17%); the University of California, Davis, Early Diagnosis and Preventative Treatment model (N=5, 17%); the Recovery After an Initial Schizophrenia Episode model (N=2, 7%); and the Oregon-based Early Assessment and Support Alliance model (N=2, 7%) (
4). Eight programs reported using other models that include various CSC components. For example, Los Angeles reported using the University of California, Los Angeles, Center for the Assessment and Prevention of Prodromal States model; Contra Costa County reported using the PIER model with some adaptations; and Madera County reported using a “peer supportive service” within a full-service partnership to support linkage to medications and therapy. Two programs (7%) were “uncertain” about their model.
Twenty-eight of 29 programs provided sufficient data to evaluate the number of FEPS-FS components offered (
Table 2). The most commonly reported components of early psychosis programs included explicit admission criteria, targeted outreach and education across community for referrals, assignment of a case manager to each client, individualized treatment plans, and client and family involvement in initial assessment. Twenty-two programs reported providing at least half of the FEPS-FS components of evidence-based FEP care. These data and programs’ reported CSC models suggest that many California early psychosis programs are providing a reasonable level of evidence-based care, although fidelity levels required for good outcomes is unclear.
Program Funding Sources
Twenty-eight programs reported funding data, with 15 programs (54%) receiving MHSA funding. Twelve programs (43%) reported receiving Medi-Cal or Early and Periodic Screening, Diagnosis and Treatment funding for children under age 21, 10 (36%) receive SAMHSA mental health block grant funding, six (21%) receive funds from private insurance, six (21%) receive self-pay funds, five (18%) receive research grants funding, five (18%) receive philanthropic funding, and one program reported county-specific funding for early bipolar disorder treatment. Based on responses from 22 programs, 14 programs (64%) reported that they are reimbursed per unit of service, four (18%) programs reported reimbursement from the SAMHSA mental health block grant, one program (5%) reported monthly reimbursement, and one program did not provide data.
Methods for Collection of Outcome Data
Programs were asked to report the types of data they collect and collection time points (
Table 3). Of the 28 programs that provided data, 18 programs (64%) reported collecting data on at least five relevant outcome domains at both intake and at least one follow-up point. All 28 programs reported collecting basic demographic data at intake. The most commonly reported types of information collected at both intake and follow-up were substance use information (21 programs, 75%), risk assessment data (19 programs, 68%), psychosocial data (18 programs, 64%), medication data (17 programs, 61%), hospitalization data (16 programs, 57%), and emergency room or crisis services use (15 programs, 54%). Notably, only four programs collected data at intake and follow-up on a maximum of 15 of the 20 domains assessed, with only 9 domains in common. This suggests a significant lack of overlap between programs in the longitudinal outcome data collection.
Of the 29 programs, 17 (59%) reported using a mix of paper and electronic records, five programs (17%) reported using a paper-only system, and seven programs (24%) reported using a solely electronic system. Eight programs (28%) began prior to electronic record implementation, with 2 to 4 years of early records remaining in a paper-only format.
Discussion
This report provides a descriptive summary of California early psychosis programs funded through a variety of entities, the individuals served and services provided, the types of data they collect, and the data collection systems they use. In terms of populations served, 22 of the 29 programs included provide care for both FEP and CHR clients, five programs serve FEP only, and two serve CHR only. A variety of funding streams, from federal, state, and donor sources, is used to support services. The majority of programs serve individuals between the ages of 12 and 25 years, include clients who have experienced psychosis for up to 24 months, and provide services for up to 2 years. Twenty-two programs reported providing at least half of the FEPS-FS components of evidence-based FEP care. Of the 28 programs that provided sufficient data, 18 programs collect data on five or more relevant outcome domains at intake and follow-up, although which outcomes were collected and when varied substantially between programs. Finally, 24 programs have at least some data stored in an electronic format.
Strengths and Limitations
To our knowledge, this is the first report on the landscape of California early psychosis programs and provides a previously unrecorded insight into the similarities and differences between these programs and the types of data being collected throughout the state. This descriptive summary could inform large-scale evaluations and provides a clear methodology for gathering data across a wide array of programs at the state or national level. Because of the extensive follow-up procedure, the response rate to the CEPAS was exceptional, with 97% of active early psychosis programs providing data as well as clarifications or additional information as needed, significantly improving data reliability.
Regarding limitations, this descriptive assessment was based on survey data reported by staff associated with the early psychosis programs and counties; thus, the findings are contingent on the accuracy and completeness of the self-reported information. For some data (e.g., sources of funding), missing data precluded analysis or reporting. Importantly, the FEPS-FS data were not collected by an external evaluator as would be standard practice (
10,
15). Consequently, FEPS-FS program components are only reported at the group level because of the preliminary nature of this approach. Future evaluations examining the impact of fidelity on client outcomes should include a comprehensive evaluation of program treatment components to determine the actual type and amount of care received by program participants. California’s MHSA funding allows substantial flexibility in how funds can be used to support mental health services, which may not be available in other states. However, these study procedures could be replicated within or across other states to identify common program features and outcome data elements as a first step in developing a nationwide evaluation of early psychosis services, which is of growing interest at the federal level (
14).
Implications
With early psychosis programs expanding nationwide, states are increasingly looking to evaluate the impact of these programs. However, lack of consistency between programs may dilute the measurable impact of these programs on client outcomes (
5–
10). For example, 83% of California early psychosis programs serve individuals at clinical high risk of psychosis; the impact of CSC care among patients at clinical high risk has not been evaluated and the inclusion of these individuals in broad outcomes evaluation of early psychosis programming will affect findings. The reported variations in clinical populations, service structure, data collection practices, and outcomes collected between programs found in this study highlight the need to first accurately survey the programs under evaluation to determine what potential impact variations between sites may have. Additionally, this study also highlights the significant challenges of using retrospective data to evaluate program effectiveness.
This study identified large variations in maximum duration of psychosis used in the inclusion criteria. This is important, given that recent findings suggest that CSC is more effective than usual care only when treatment is initiated early (i.e., within 74 weeks) (
12). In addition, large variations in the length of treatment provided—ranging from 1 year to indefinite—were also noted. This is inconsistent with current national recommendations for treatment to be available for at least 2 years (
3) and evidence that treatments over an even longer period may be necessary for a subset of individuals to maintain long-term significant improvements (
21,
22). As a result, treatment of such relatively short duration may reduce both the short- and long-term effectiveness of early psychosis services.
Programs reported significant variations in the components of care delivered across the various services according to the FEPS-FS checklist. One site reported delivering 27 of the 30 FEPS-FS components assessed, whereas another program reported delivering only nine (
Figure 1). Although there is evidence to suggest that CSC is effective (
11,
12), it is still unclear which specific components of care affect client outcomes and whether variations in components offered affect treatment effectiveness. As a result, any large-scale evaluation of existing services requires careful examination of care components delivered by each service, both to aid interpretation of heterogeneity of treatment outcomes across services and to understand what components of the CSC may be key to improving outcomes.
Conclusions
This study indicates that there is considerable variability between early psychosis programs across California, including the components of care provided, inclusion criteria for service users, and data programs routinely collected. As a result, it is important to evaluate what impact these variations may have on treatment outcomes. In addition, this study highlights the significant challenges to conducting a retrospective statewide evaluation of early psychosis services, instead suggesting that prospective evaluation with synchronized data collection would be necessary for statewide or nationwide evaluation. Currently, multiple California counties are embarking on a collaborative effort to harmonize data collection across their early psychosis programs. Results of this project were used to identify the approaches and data elements that these programs already have in common, as well as areas in which additional standardization will be needed. Comprehensive fidelity evaluations of program components will enable evaluation of program-level differences on client outcomes. The collaboration hopes that evaluation results will inform the development and funding of future early psychosis programs across the United States and suggest the minimum standards necessary for programs to yield positive outcomes.
Acknowledgments
The authors acknowledge the support of the California counties that participated in this evaluation.