Evidence base for early intervention in psychosis services in rural areas: A critical review

Early Intervention in Psychosis teams (EITs) are a growing entity internationally, yet they remain under‐researched given challenges facing their delivery. Model adaptations include stand‐alone services, a hub‐and‐spoke model with various bases and integrating specialist staff into existing mental health teams. The present critical review focuses on reviewing the evidence base for the delivery of EITs in rural areas, first pertaining to evidence for model adherence and second to clinically outcomes.

psychosis (Bird et al., 2010;Marshall & Rathbone, 2011). EITs offer prompt access to a range of psychological, social and medical interventions, including Cognitive Behavioural Therapy for Psychosis, family therapy, vocational training and support, and antipsychotic medication (Wyatt & Henter, 2001). They are established worldwide, including New Zealand, United States, throughout Europe and in China (eg, Nishida et al., 2016;Srihari et al., 2015). EITs have been established in rural areas across the world, including in the United Kingdom (Burbach, Grinter, & Bues, 2009), Australia (O'Kearney, Garland, Welch, Kanowski, & Fitzgerald, 2004), Greece (Mantas & Mavreas, 2012) and Canada (Cheng, Dewa, Langill, Fata, & Loong, 2014). Research has reported similarities in conceptualization and delivery of the EIT model across the world (McDaid, Park, Iemmi, Adelaja, & Knapp, 2016). Further research suggests that adherence to EIT model by consistent, timely delivery of the range of psychosocial and medical interventions, is linked to positive clinical outcomes (Fusar-Poli, McGorry, & Kane, 2017). What is less clear is how this translates to EITs in rural areas, where challenges to model adherence and timely access due to geographical locations exist (Thirthalli et al., 2017). The present review will explore this issue.
Relevant EIT service models will first be considered.

| Models of EITs in rural areas
Research in health care in rural and remote areas has recently implied that standard outpatient community models of care may not meet the needs of non-urban populations (Mitton, Dionne, Masucci, Wong, & Law, 2011). Rural areas lack a single agreed upon definition in the literature due to differences in what constitutes a rural area internationally, for example, an average rural area likely being smaller in the United Kingdom than an average rural area in Australia. However, Wakerman (2004) purports that an average definition of a rural area across settings are where the average population lives more than 60 km from the nearest Primary Care facility. Rural areas are there large, often sparse geographical locations which hold particular challenges for delivering community services, such as standard outpatient clinics not being accessible and an increased travel burden for clinicians to see patients (Thirthalli et al., 2017). In line with this, recent research has begun to explore adapted EIT models for different geographical areas and distinct population groups. There are currently three main models. The "stand-alone" model consists of one full, comprehensive team who work assertively with one geographical area, most often urban areas (eg, Petersen et al., 2005). The "hub-andspoke" model consists of a central "hub", or base, where managers, administrators and some members may be based, but has various "spokes" around a large geographical region where specific team members may work from, mainly covering larger rural areas (Bostock & Britt, 2014). EITs may therefore have staff established across a wider geographical region. The third model is "specialistwithin-generalist", in which EIT members are based within a general mental health team, such as a Community Mental Health Team (CMHT), and function as a specialist pathway provision (Behan, Masterson, & Clarke, 2017). Behan et al. (2017) reviewed studies directly comparing different models of EIT and found that the stand-alone EIT model may be more clinically effective and cost-effective, but may face significant practical barriers such as distance to service users which makes it less viable in rural areas. The review highlights that type of service model may be a key determinant of clinical outcomes, as rural areas may be better suited to models which adapt to the specific demands. What remains unclear is the current effectiveness and delivery of EITs in rural areas; a prior review will now be summarized.
1.2 | The existing evidence for rural EITs Welch and Welch (2007) conducted a prior review of the evidence for rural EITs, finding three studies (O'Kearney et al., 2004;Tee, Ehmann, & MacEwan, 2003;Welch & Garland, 2000). Two of these studies only described the implementation of rural EIT services in Australia (Welch & Garland, 2000) and Canada (Tee et al., 2003). Both studies described establishing local networks with existing services to facilitate the identification and early treatment of first-episode psychosis, though did not report any clinical outcomes. O'Kearney et al.'s (2004) study presented the evaluation of the EIT described in Tee and colleagues' (2003) study with regards to model adherence. It reported that model adherence was greatest where a diagnosis of psychosis was in place and they had been registered under the EIT programme. The review considered this early but limited evidence of the ability to adhere to EIT service protocols in a rural area, though concluded that rural EITs remained "underserved and underresearched" (Welch & Welch, 2007, p. 489). The review concluded with recommendations for further research to establish the clinical outcomes and required adaptations to rural EITs, such as increased partnerships with local services and different routes of access to care.
Further research has since been published reporting the clinical outcomes of rural EITs and evaluating adherence to service model. The present review will therefore aim to provide a summary of this literature and an appraisal of its quality. O'Kearney et al.'s (2004) study will be included in the present review to enable consideration of the evidence base as a whole, both in reporting their outcomes and appraising its quality.

| Rationale for the present review
The present critical review aims to expand a prior review of the evidence base for effective delivery of EIT in rural areas (Welch & Welch, 2007) by including more recently published studies and utilizing a quality appraisal tool to consider the quality of the evidence (Critical Skills Appraisal Program, 2018a, 2018b. The present review will include studies for EITs in rural areas pertaining to both clinical outcomes and outcomes for services' adherence of the EIT model, as research highlights that this is an important aspect of effectiveness (Behan et al., 2017). It will focus on global studies given the international nature of this PIPKIN issue, though the implications of the diversity of the populations involved will be considered. The aim of the review is to answer the following questions: 1. What is the evidence base for the clinical effectiveness of EIT in rural or remote areas? 2. What is the evidence base regarding the adherence to the EIT service model in rural and remote areas?
A literature search will be undertaken, followed by a critical appraisal of the relevant studies. The implications will then be discussed.

| METHOD
A narrative review was conducted with the aim of capturing the available evidence of rural EITs and appraising their quality. The author conducted the search and reporting, the implications of which will be considered.

| Search terms
In line with expanding Welch and Welch's (2007) original review, their search terms were utilized for publications from 2004 to present. This included variants of the clinical term "early intervention in psychosis" (including first-episode psychosis; prodrome psychosis; first-identification psychosis, and so on), geographical terms (rural; sparse population; remote; regional) and evidence terms (quantitative study; clinical outcomes; and so on). Dominant associated terms for "early intervention" and "psychosis" were used and altered per each database searched to ensure any alternate language was included. The definition of rural varies across countries which poses challenges for conceptualization, such as a rural area likely being considerably more remote in Australia than in the United Kingdom.
Research also has no standardized definition of a rural area and measurements vary (Wakerman, 2004). For the purposes of gathering the available evidence of rural EITs to offer a summary and quality appraisal, as in Welch and Welch's (2007) original review, rural was considered self-defined in line with the aim of capturing all available global evidence, though the issue of defining rural remains a limitation. Search terms were updated as the literature was explored, and definitions between papers were cross-checked for major discrepancies.

| Search method
The search was run in the databases PsycInfo, AMED, BNI, CINAHL, EMBASE, HBE, HMIC, Medline, PubMed, and Ovid. The reference lists of key reviews and recent publications from Early Intervention in Psychiatry were also scanned.

| Inclusion criteria
Studies published between 2004 to present were included. O'Kearney et al. ' (2004) paper from Welch and Welch's (2007) review was included to enable oversight of the available evidence sourced and to allow its quality to be appraised. The review adopted a hierarchy of evidence where randomized controlled trials (RCTs) are the highest form of evidence, followed by cohort studies and case controls (Murad, Asi, Alsawas, & Alahdab, 2016). All types of research design were included. Papers included were any study using any research design pertaining to evaluation of an EIT within a rural area. Mixed urban/rural area papers were included where the rural aspect was noted, such as comparing a rural area to another directly or implementing an alternative service model to serve the fact that the majority of the area was rural.

| Exclusion criteria
Papers not written in English were excluded due to the constraints of time and resource. Poster presentations, grey literature and unpublished transcripts were not included. Papers simply describing the implementation or key issues without specifically reporting either clinical outcomes or measurement of adherence to service model were also excluded. Studies addressing a partly rural area but without separating the specific issues regarding the rural area were excluded, for example papers exploring a mixed urban/rural area where the rural part was not predominant nor explored directly by the research.

| Summary of search
The below flowchart (Diagram 1) outlines the process of the literature search. A total of 1071 unique publications were returned on the first search. The author screened the titles and abstracts of all papers, resulting in 53 papers being identified as relevant. All 53 relevant papers were read in full with nine papers being included using the above criterion. A summary of the final included papers can be found in Table 1.

| Critical appraisal tool
The critical appraisal tools provided by the Critical Skills Appraisal Program (2018a, 2018b) were used to guide the critical appraisal of the studies, specifically the case control and cohort study tools as per the design used in each individual study.

| RESULTS
The literature searches described produced a total of nine papers which met the inclusion criteria. The studies contained a mixture of naturalistic cohort studies and case control studies comparing different service models. The highest level of evidence presented according to the hierarchy of evidence criteria were levels 2 (cohort studies) and 3 (case control studies). This reflected the need for naturalistic studies of established services and direct comparisons of existing services.
Five studies focused on evaluating clinical outcomes of rural EIT services, whereas four evaluated interventions aimed at improving adherence to the EIT model. The studies will therefore be grouped in line with the research questions; clinical effectiveness of EIT in rural areas (n = 5; see Table 1); adherence to the EIT model in rural areas (n = 4; see Table 2).

| Studies evaluating user outcomes of the EIT model in rural areas
Full details of the individual studies can be found in Table 1. Two studies used retrospective case data to compare service models (Dodgson et al., 2008;Fowler et al., 2009); one was a cross-sectional case control study comparing two EIT service models (Cheng et al., 2014); and two were naturalistic service evaluations (Burbach et al., 2009;Mantas & Mavreas, 2012). Comparisons between studies are not possible due to different outcome measures being used, differences in geographical, population and health service factors and lack of reported specificity of what interventions are provided. However, four studies demonstrated positive clinical outcomes for hub-and-spoke EIT models. Two United Kingdom-based evaluation studies demonstrated positive outcomes for use of hub-and-spoke EIT (Burbach et al., 2009) with one demonstrating favourable outcomes over a standard community mental health team (Dodgson et al., 2008). An evaluation study in Greece further demonstrated positive outcomes for a hub-and-spoke model (Mantas & Mavreas, 2012). It is worth noting that all three studies used different outcome measures; the United Kingdom-based studies relied on hospital admission data whereas Mantas and Mavreas (2012)   Evaluating non-EIT service n = 9 Mixed urban/rural without addressing rural n = 1 D I A G R A M 1 A flow chart outlining the screening procedure for the included and excluded studies used in the present review (n = 45) Found a reduction in PANSS from initial referral (mean = 87.6, SD = 8.9) to 6-month follow-up (mean = 64.7, SD = 11.9) and an improvement in GAF (initial mean = 33.9, SD = 6.5; 6-month followup mean = 68.5, SD = 6.1). Relapse rate was defined in both studies (Burbach et al., 2009;Mantas & Mavreas, 2012) as increase in psychotic symptoms requiring hospital admission following a period of remission, though neither study specified the time frame for remission. All three studies described successful implementation of the hub-and-spoke model in their rural areas, commenting it provides managerial and geographical flexibility to enable early access to care. Professional networks with existing services was a commonly reported facilitator of successful implementation. Cheng et al.' (2014) study in Canada provided a direct comparison of a stand-alone and hub-and-spoke EIT both in rural areas using case control design. They reported that the stand-alone EIT had 70% of its population admitted to hospital whereas the hub-and-spoke had 31.8% admitted to hospital. They found higher reported general functioning using the Multnomah Community Ability Scale in the hub-andspoke model, though this was not statistically significant. Though describing differences, the paper was unable to explain why they existed.
Lastly, Fowler et al.' (2009) used historical case control design to compare a stand-alone EIT to a partial EIT and a community mental health team, finding that the stand-alone EIT demonstrated better outcomes than the partial EIT on a range of measures. However, their study lacked stringent methods of comparison as each group had different data points and differed significantly on demographic information.
In summary, four studies demonstrated positive clinical outcomes for hub-and-spoke EIT models in rural areas, two of which were direct comparisons to other service models. A stand-alone EIT also outperformed a partial EIT and community mental health team, suggesting that both hub-and-spoke and stand-alone EITs may be efficacious in rural areas and some limited evidence to suggest that huband-spoke may outperform stand-alone EIT in a Canadian setting.

| Studies evaluating adherence to the EIT model in rural areas
Full details of the studies can be found in Table 2. Two studies used cross-sectional survey designs to evaluate the implementation of different aspects of the EIT model and factors that may be associated with implementation (Catts et al., 2010;Durbin et al., 2016), whereas two studies utilized a pre-and post-training audit design to evaluate the impact of training on adherence to aspects of the EIT model (Bedard et al., 2016;O'Kearney et al., 2004). Direct comparisons with these studies are not possible due to lack of standardized measures, different training programmes and different care settings across Australia and Canada. However, the studies reported low rates of between 16%-50% for adherence to EIT service model based on their individual measures (Durbin et al., 2016;O'Kearney et al., 2004).
Training programmes showed limited and variable impact in improving adherence across teams (Bedard et al., 2016;O'Kearney et al., 2004), with authors suggesting that practical barriers such as lack of funding, managerial support and infrastructure for supporting and evaluating services may be contributing factors.
Rural EITs report less funding that urban services (Catts et al., 2010;Durbin et al., 2016) and less access to resources, such as Psychiatry time and physical health appointments (Bedard et al., 2016). Catts et al. (2010)  focused training for EIT service delivery (Bedard et al., 2016;Catts et al., 2010;Durbin et al., 2016). Studies all used different self-report measures and different training programmes, alongside having different geographical, cultural and care settings, and so direct comparisons are difficult.

| DISCUSSION
The presented studies demonstrate mixed results for the evidence base for EIT services in rural areas. With regards to clinical outcomes, the studies' findings were largely in line with prior research on EIT services generally and specifically urban and inner-city areas, finding improvements in psychotic symptoms, general functioning, relapse rates and reductions in hospital admissions (Behan et al., 2017). Four studies demonstrated positive clinical outcomes for a hub-and-spoke model in rural areas and described its successful implementation,

| Limitations of review
The present review is limited for being a narrative review conducted by a lone author. It is acknowledged that this results in bias in study selection and reporting is a risk to validity. The review only included papers written in English, which presents a potential bias regarding the culturally-bound nature of the findings. Additionally, though the evidence base has increased substantially since the prior Welch and Welch (2007) (Behan et al., 2017). Clinically, the papers report that the huband-spoke model enables effective rural outreach, stable managerial support located in the hub, and opportunities regarding engagement with community venues (eg, voluntary job opportunities in farming communities; engagement with local schools or youth hubs). Relapse rates and clinical outcomes compare favourably to CMHT and partial EIT services. Second, a predictor of the clinical adherence to the EIT model may be having some form of registration under a governing programme or body, so local initiatives can support this in rural areas using training, on-going supervision and managerial support.

| Research implications
Research into rural EIT services has grown since the last review (Welch & Welch, 2007), though it remains limited in both design and scope. Further research could begin to explore whether elements of the hub-and-spoke model address the specific challenges facing rural service provision, the feasibility and efficacy of specific interventions including CBTp, and focus specifically on how able rural EIT services are to adhere to evidence-based service provision. Regarding specific interventions, other fields of rural mental health care include telemedicine approaches using video conferencing to deliver specific interventions, which research has begun to explore in early psychosis (Stain et al., 2011). Further research is required here into the acceptability and outcomes of such techniques in rural EIT populations.

| Service and political implications
Rural EIT services are a growing entity internationally, yet it seems they are under-resourced and under-researched. Service providers can follow the successful co-ordination of the implementation and evaluation of services, such as efforts made in Ontario (Durbin et al., 2016); providing training for EIT services shows some efficacy for improving adherence to the model, and managerial support and supervision facilitates engagement with local mental health and community networks. Politically, this growing field highlights a tension in providing robust, evidence-based services with enough flexible adaptations to a population with different needs than present knowledge wholly accounts for. More efforts regarding targeted research exploring the needs of this population and what works would be beneficial.

| CONCLUSIONS
The present review updated a prior review exploring the state of the evidence base for EIT services in rural areas (Welch & Welch, 2007).
Although the evidence base has grown from three to nine studies, it remains limited in both design and scope. Emerging findings indicate that rural EIT services are as effective as urban services using some adaptations to aid implementation, including using hub-and-spoke models. Research from service providers highlights that rural services may be under-funded, and training has some limited impact on improving adherence to the EIT model. Future research and political efforts should continue to evaluate the specifics of the needs of this population and how a flexible use of the EIT model can be effective.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.