Introduction
Youth populations, which include adolescents and young adults, are particularly affected by mental health disorders. An estimated 75% of disorders begin before the age of 25 (Reference Kessler, Demler, Jin, Merikangas and Walters1). Traditional organization of mental health services separates child and adolescent services from adult services based on the legal age of majority (Reference Davis and Vander Stoep2). Problems with access, engagement and continuity have been acknowledged and service organisation may be inadequate for youths. In fact, while 18% of young people ages 15 to 24 suffer from a mental health disorder (Reference Bédard, Connolly and Gravel3), only a quarter of them seek mental health services (Reference Bergeron, Poirier, Fournier and Roberge4). Moreover, if treatment is delayed for too long, recovery can be seriously be compromised. For example persistent symptoms and poorer response to treatment have been observed (Reference Perkins, Gu, Boteva and Lieberman5, Reference Marshall, Lewis, Lockwood, Drake, Jones and Croudace6). The onset of a mental health disorder when youth must transfer from child and adolescent services to adult services is not without consequences. It can impact their coping skills and result in difficulties that go beyond clinical symptoms to affect various areas of their life trajectory, including their academic, professional or social integration.
To address the issue, programs intended simultaneously for both adolescents and young adults have been developed to meet their needs at every clinical stage of mental illness. While these services focus primarily on youth living with a psychotic disorder, there is a growing interest at both the clinical and organizational level to adapt these programs to other mental health disorders. The Douglas Mental Health University Institute, a specialised mental healthcare organisation, requested an HTA on the efficacy of such models and programs on youths with a mental health disorders or at risk of developing one. Mental health disorders of interest were mood, psychotic, anxiety or personality disorders. The HTA focus was on service organisation and not on specific interventions or therapy.
On this basis, a systematic review (SR) of systematic reviews was conducted to answer the following question: what is the efficacy of specialized models, programs and services targeting simultaneously adolescents and young adults with a mental health disorder or considered at risk of developing one?
Analytic framework
An analytical framework for early intervention services was developed by emergent design using the performance measures identified by Addington (Reference Addington, McKenzie, Addington, Patten, Smith and Adair7), the recovery dimensions developed by Whitley and Drake (Reference Whitley and Drake8), and the FAME (Feasibility, Appropriateness, Meaningfulness and Effectivenes) model developed by the Joanna Briggs Institute (9). This framework consists of four dimensions (Table 1): A) Accessibility; B) Engagement and continuity; D) Recovery; and D) Meaningfulness and acceptability.
Table 1. Analytical framework for early intervention services for youth with mental health disorders
Model developed by emergent design on the basis of models developed by Addington (Reference Addington, McKenzie, Addington, Patten, Smith and Adair7), Whitley and Drake (Reference Whitley and Drake8) and Johanna Briggs (9)
Methods
Method – Scientific data
A SR of SRs was conducted to assess the scientific evidence regarding the effectiveness of early intervention models, programs and services intended for youth living with a mental health disorder. A search of the literature published in English or French was carried out in ASSIA, MEDLINE, Embase and CINAHL databases by a professional information specialist. Three key concepts were used: mental health disorders, mental health services and systematic review. The literature search strategies were tested using relevant studies found in an exploratory search. Articles published between January 2000 and January 2017 were retained, since the majority of research on early intervention services was conducted during this period. The selection was made by inter-rater agreement (AL and GAT).
Inclusion criteria and the PICOTS elements were:
■ Population: Adolescents and young adults with a mental health disorder or at risk of developing one (mood, psychotic, anxiety or personality disorders)
■ Intervention: Model, program or specialized services whose access is not based on having reached the age of majority
■ Comparison: No restriction
■ Outcomes: Accessibility, continuity, recovery, acceptability, satisfaction, engagement, early intervention, autonomy or suitability
■ Time: No restriction
■ Setting: No restriction
Exclusion criteria, based on the decision makers’ needs, were as follows:
■ Any model, program or service intended for a specific group based on age
■ Studies evaluating transitional programs between child or adolescent services to adult services
■ Primary care
■ Clinical or support interventions that are program components
■ Autism spectrum disorders, intellectual disability, etc.
■ Documentary search in at least three database in the SR
■ A selection diagram based on inclusion and exclusion criteria for the SR
We developed an extraction grid based on the analysis framework. Data extraction was performed by a first reviewer (AL) and then validated by a second (SB). Quality control was conducted through inter-rater agreement (AL and GAT) with the ROBIS tool (Reference Whiting, Savović, Higgins, Caldwell, Reeves and Shea10).
Method – Experiential and contextual data
An advisory committee of experts (researchers, clinicians, managers and users) was established to provide contextual and experiential data to complete scientific data. Seven group meetings were held with the advisory committee and six individual interviews were conducted. In addition, an exploratory research was performed to obtain contextual data on the organization of mental health services for youth.
Data analysis and synthesis
The data was analyzed and synthesized in narrative form. The level of evidence for effectiveness was assessed by using the scientific data convergence method developed by the National Autism Center (11), and adapted by the Institut national d'excellence en santé et services sociaux (12) and Beauchamp et al. (Reference Beauchamp, Drapeau and Dionne13). For quality assurance purposes, the study was reviewed by two independent reviewers.
Results
1,841 references were found in the four databases, for a total of 1,054 articles, after removing duplicates. Following selection based on titles and abstracts, 36 articles were read. After the final selection, five articles on SR were retained (Figure 1). Overall, the risk of bias of these studies is estimated at 20%, which we considered satisfactory. Not all primary studies selected in these SR provided relevant scientific data to the evaluation question. Data from 52 primary studies pertaining specifically to the evaluation question was extracted and analysed. Given that the selected SRs have different research questions and with no overlapping analytical framework categories, we assumed that results from the primary studies have not been reported in more than one selected SR. Lloyd Evans et al. (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14) studied access, Cooper et al. focused on meaningfulness and acceptability (Reference Cooper, Evans and Pybis17), Randall et al. (Reference Randall, Vokey, Loewen, Martens, Brownell and Katz15) analysed clinical recovery while Bond et al. (Reference Bond, Drake and Luciano16) evaluated functional recovery.
Figure 1. PRISMA selection diagram.
A synthesis of results from primary studies, following the analysis framework (Table 1), is presented in Table 2 to evaluate convergence of scientific data. The name of the model, program or service and the performance measures are identified. The ratio between the number of primary studies reporting a positive result (in support of the model, program or service) on the total number of primary studies is presented for studies with comparison group and studies with no comparison group. The number of primary studies reporting a positive and significant result (p < 0.05) is also presented.
Table 2. Synthesis of primary study results
Impact of programs on access
The impact of implementing early intervention on access has been the subject of very few SRs. One SR, by Lloyd-Evans et al. (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14), pertaining to the impact of early intervention for psychosis (EIP) on the duration of untreated psychosis (DUP) was identified. The DUP is the time between the onset of psychosis and the start of treatment. It is a commonly used indicator in mental health. Scientific data from five out of the 11 primary studies selected in this SR, all with a retrospective group, was extracted and analysed. Three programs are evaluated: a Norwegian specialized detection and treatment (TIPS), Prevention and Early Intervention Program for Psychosis (PEPP) in Canada and the Early Psychosis Prevention and Intervention Centre (EPPIC) in Australia. TIPS included the development of EIP in conjunction with a very large scale multi-focus campaign including newspaper, television, radio and cinema ads, combined with school visits with teachers, counselors and pupils as well as seminars for health professionals during four years. PEPP included a large scale awareness campaign, contacts with general practitioners and visits with school counsellors during two years. EPPIC focused on improved access and included a clear point of referral with rapid intervention response. Small scale community-based networking and education initiatives are also mentioned (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14). In the case of TIPS, the observed DUP was much lower compared to traditional services (Md 5 vs 26 weeks, p < 0.05) (Reference Johannessen, McGlashan, Larsen, Horneland, Joa and Mardal18, Reference Melle, Larsen and Haahr19)). For EPPIC, however, a higher DUP was observed compared to traditional services (Md 52 days vs 30 days, p < 0.05) (Reference McGorry, Edwards, Mihalopoulos, Harrigan and Jackson20) and PEPP (Md 24 vs 22 weeks, n.s., (Reference Malla, Norman, Scholten, Manchanda and McLean21)). The number of references in the region for first episode psychosis did not increase following the implementation of TIPS and PEPP. Scientific evidence has not been established regarding the impact of a change of service configuration on DUP. The results appear to be more promising when services are deployed in conjunction with a large scale multi-focus campaign to encourage help seeking by young people.
Impact of programs on engagement and continuity
No SR dealing with the impact of programs on engagement and continuity was identified.
Impact of programs on recovery
Clinical recovery
Randall et al. (Reference Randall, Vokey, Loewen, Martens, Brownell and Katz15) evaluated the effect of EIP on the use of hospital services in a SR. The EIP interventions retained were based on components of assertive community treatment (ACT), which includes a multidisciplinary team. Scientific data from all the primary studies selected from this SR (15/15) was extracted and analysed. All were cohort studies with a comparison group. Among the 15 studies, three randomized trials were found (Reference Killackey, Jackson and McGorry22–Reference Garety, Craig, Dunn, Fornells-Ambrojo, Colbert and Rahaman24). The programs studied were OPUS in Denmark (Reference Bertelsen, Jeppesen, Petersen, Thorup, Øhlenschlæger and Quach25); Lambeth Early Onset (LEO) in England (Reference Craig, Garety, Power, Rahaman, Colbert and Fornells-Ambrojo26), Programma 2000 in Italy (Reference Cocchi, Mapelli, Meneghelli and Preti27), Parachute in Sweden (Reference Cullberg, Levander, Holmqvist, Mattsson and Wieselgren28) PEPP in Canada (Reference Goldberg, Norman and Hoch29), PEP in Australia (Reference Petrakis, Penno, Oxley, Bloom and Castle30), Early Treatment and Home-based Outreach Service (ETHOS) in England (Reference Singh, Grange, Vijaykrishnan, Francis, White and Fisher31) and a program from Norway (Reference Grawe, Falloon, Widen and Skogvoll32). Randall et al. (Reference Randall, Vokey, Loewen, Martens, Brownell and Katz15) observed a significant decrease in the risk of being hospitalized at least once during the follow-up period for EIP participants in a meta-analysis of 13 primary studies (OR = 0.33; 95% IC = 0.18-0.63; p < 0.01). Randall et al. (Reference Randall, Vokey, Loewen, Martens, Brownell and Katz15) also observed a significant decrease in the number of days hospitalised in a meta-analysis of 11 primary studies (SMD = −0.38; 95% IC = −0.53—0.24). Sensitivity analyses support a significant relationship in favour of EIP. However, no significant difference between EIP and traditional services was reported in the three randomized trials (Reference Killackey, Jackson and McGorry22–Reference Garety, Craig, Dunn, Fornells-Ambrojo, Colbert and Rahaman24). The level of scientific evidence is established as promising.
Functional recovery: employment and education
Bond et al. (Reference Bond, Drake and Luciano16) evaluated the impact of EIP programs on employment and enrolment in education. In this SR, results from youths aged 14 to 35 were analysed, with no distinction between adolescents and young adults. Scientific data from 20 out of the 28 primary studies selected by Bond et al. (Reference Bond, Drake and Luciano16) was extracted and analysed. EIP with supported employment was also evaluated. Higher employment rates were observed in four primary studies comparing EIP with no identified vocational service component to traditional services (Reference Eack, Hogarty, Greenwald, Hogarty and Keshavan33–Reference Penn, Uzenoff, Perkins, Mueser, Hamer and Waldheter36). Bond et al. (Reference Bond, Drake and Luciano16) observed that the impact of EIP is further increased when programs included supported employment, with seven out of eight primary studies identified using the individualized supported employment (IPS) model. The authors reported an employment rate (OR = 3.66 (1.93-6.93); p < 0.0001) and an increase in employment rate (OR = 4.97 (1.53-16.22); p < 0.008) significantly greater in EIP with supported employment compared to early intervention services without supported employment in the meta-analysis of four primary studies (Reference Killackey, Jackson and McGorry22, Reference Killackey, Allott, Cotton, Chinnery, Sun and Collins23, Reference Major, Hinton, Flint, Chalmers-Brown, McLoughlin and Johnson37, Reference Nuechterlein, Subotnik, Ventura, Turner, Gitlin and Gretchen-Doorly38). The level of scientific evidence is established as promising for employment rates.
Enrolment in education was evaluated in primary studies comparing EIP with supported employment with early intervention services with no supported employment. The SR by Bond et al. (Reference Bond, Drake and Luciano16) yielded mixed results: higher enrolment in education for EIP with supported employment was reported in two primary studies (35% vs 24% (Reference Killackey, Jackson and McGorry22), and 54% vs 41% (Reference Killackey, Allott, Cotton, Chinnery, Sun and Collins23)); while one study reported a lower enrolment rate (20% vs 24% (Reference Major, Hinton, Flint, Chalmers-Brown, McLoughlin and Johnson37)). No significant difference was reported in the meta-analysis of four primary studies (Reference Killackey, Jackson and McGorry22, Reference Killackey, Allott, Cotton, Chinnery, Sun and Collins23, Reference Major, Hinton, Flint, Chalmers-Brown, McLoughlin and Johnson37, Reference Nuechterlein, Subotnik, Ventura, Turner, Gitlin and Gretchen-Doorly38) conducted by Bond et al. (OR = 1.39 (0.86-2.24), p = 0.17 (Reference Bond, Drake and Luciano16)). Scientific evidence is not established for enrolment in education.
When enrolment in education and employment are added together to form a single functional recovery measure, the impact of EIP on functional recovery remains unclear. In two out of the three primary studies retained, EIP with no supported employment obtained positive results compared to traditional services (Reference Cullberg, Mattsson, Levander, Holmqvist, Tomsmark and Elingfors39, Reference Agius, Shah, Ramkisson, Murphy and Zaman40) but only one which reported a significant difference (Reference Bertelsen, Jeppesen, Petersen, Thorup, Øhlenschlæger and Quach25).
Impact of program on meaningfulness and acceptability
The meaningfulness and acceptability of the inter-agency collaborative model and the person-centred care model were evaluated in two SR (Reference Cooper, Evans and Pybis17, Reference Gondek, Edbrooke-Childs, Velikonja, Chapman, Saunders and Hayes41). The primary studies from these SR were qualitative and mostly focused on the identification of facilitating factors and constraining factors to the implementation of these models.
For the inter-agency collaborative model, scientific data from seven primary studies (Reference O'Herlihy, Oliva, Evans and Cooper42–Reference Spong, Waters, Dowd and Jackson48) out of the 33 primary studies selected by Cooper et al. in a SR. (Reference Cooper, Evans and Pybis17) was extracted and analysed. Based on Flexhaug's collaborative models of care (Reference Flexhaug, Noyes and Phillips49), the programs studied follow a co-location and collaborative approach. Programs were briefly described in the primary studies and the level of collaboration could not be further defined. The majority of primary studies identified by Cooper et al. (Reference Cooper, Evans and Pybis17) focused on the organizational perspective of professionals (6/7). The effectiveness of the Children & Young People's (CYPMH) collaborative model was evaluated in two primary studies. Hamilton et al. (Reference Hamilton, Begley and Culler44) reported improved satisfaction of professionals involved following the implementation of the collaborative model. O'Herlihy et al. (Reference O'Herlihy, Oliva, Evans and Cooper42) highlighted the importance and the relevance of inter-agency collaboration for users and their family. The most reported facilitating factors (in more than three primary studies) in the SR by Cooper et al. are joint training, good understanding, support from upper management, the use of a collaboration protocol and, last but not least, a model focused on youth and their family. The most reported constraining factors are inadequate resources, poor communication, lack of acknowledgement and respect, cultural differences and confidentiality issues.
For the person-centred care model, five primary studies were identified (Reference Idenfors, Kullgren and Renberg50–Reference Pelto-Piri, Engström and Engström54) among the twenty-three primary studies identified in a SR by Gondek et al. (Reference Gondek, Edbrooke-Childs, Velikonja, Chapman, Saunders and Hayes41). The most reported facilitating factors were good listening skills, respect and validation. The most reported constraining factors were lack of resources and confidentiality. This was also the case for the inter-agency collaborative model. Given the small number of SRs conducted on these models, the level of scientific evidence cannot be established.
Discussion
The results on recovery through EIP programs are promising. Indeed, young people in EIP programs were hospitalized less often. Furthermore, the advisory committee has observed other positive effects of EIP programs on clinical and functional recovery in their practice and experience. Randall et al. (Reference Randall, Vokey, Loewen, Martens, Brownell and Katz15) suggest that variation in the intensity of EIP may have decreased hospitalisation by improving youth engagement in treatment and continuity of care in the community. An increase in the availability of treatment in the community could also have helped reduce the need for hospital inpatient services. In regards to functional recovery, Bond et al. (Reference Bond, Drake and Luciano16) reported an increase in employment rates with EIP programs. According to the authors, this effect is stronger when EIP includes the IPS model of supported employment. However, EIP does not seem to have an impact on enrolment in education. Education, an important social indicator of health, marks an important step in the life trajectory of young people. Additional research is needed to determine how services could better support youth in their educational pursuits. Unfortunately, no SR was found for specific mental health disorders other than psychosis. The SR by Cooper et al. (Reference Cooper, Evans and Pybis17) and by Gondek et al. (Reference Gondek, Edbrooke-Childs, Velikonja, Chapman, Saunders and Hayes41) cover mental health disorders in general. Experts from the advisory committee recommend looking into whether organisational components of EIP models could be adapted to other mental health disorders and other settings such as community services, as early intervention has generated positive results for psychosis.
While the impact of EIP programs on recovery is promising, their impact on access has yet to be determined. Effectiveness of early intervention can only be considered partial if it does not reach all youth with a mental health disorder or at risk of developing one. According to Lloyd-Evans et al. (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14), EIP alone is not sufficient to decrease DUP. Observations by one expert on the advisory committee converge with these findings. The authors attribute the success of TIPS, compared to EPPIC and PEPP, to the intensity of their multi-focus campaign. According to Lloyd-Evans et al. (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14), the campaign focused on promoting help seeking behaviors and changing attitudes about psychosis. This type of campaign should be implemented jointly with the development of early intervention services. The SR by Lloyd-Evans et al. (Reference Lloyd-Evans, Crosby, Stockton, Pilling, Hobbs and Hinton14) included a first generation of primary studies on the development of early intervention services. Since publication of this SR in 2011, new models using a concerted community-based approach have emerged; these include Headspace in Australia and ACCESS (Adolescent /young adult Connections to Community-driven Early Strengths-based and Stigma-free services) in Canada.
Furthermore, the advisory committee recommends forging partnerships with the community and developing services for at-risk youth, more specifically through an at-risk youth evaluation service. The current context whereby services are being reorganized and integrated provides an interesting opportunity to improve mental health services for young people. These services would be anchored with specialized services such as early intervention, in order to improve access and treatment throughout the clinical stages of illness in a continuum of care. Given the non-linear progression of mental health disorders, the experts recommend that admission into services for at-risk youth be conducted regardless of diagnosis or whether or not the client has reached the age of majority. Instead of considering these programs as stand-alone services, perhaps we should consider them as entities part of a continuum of care and treat mental illness at every stage of its evolution. The facilitating and constraining factors identified by Cooper et al. (Reference Cooper, Evans and Pybis17) for the collaborative model should be taken into account when implementing partnerships.
While there are many primary studies dealing with the dimensions of engagement and continuity, no SRs have been identified for programs targeting simultaneously adolescents and young adults. The challenges are no less significant, and the facilitating and constraining factors identified for the person-centred care model (Reference Gondek, Edbrooke-Childs, Velikonja, Chapman, Saunders and Hayes41) should be considered. Experts have highlighted the importance of developing youth-friendly services.
Conclusion
Promising results were found in terms of recovery with EIP. Organisational components from such programs could be developed for youth exhibiting other mental health disorders, and, given the absence of evidence on such services, these models should continue to be evaluated through research. An integrated approach anchored within a continuum of care must be adopted to improve access to specialized services. The development of partnerships with primary care, community resources, social services and schools as well as the development of services for at-risk youth, regardless of diagnosis or age, are recommended to ensure that all young people exhibiting signs of psychological distress may obtain adequate services. Such service organisation would be more adapted to the maturation process of young people and the evolution of mental health disorders. Further research on the organisation and the integration of primary care and community services in the continuum of care is necessary. A multi-focus campaign should also be launched to advertise those services and encourage help-seeking by young people. Service development should build on the experience with EIP services.
Acknowledgements
The Institut national d'excellence en santé et en services sociaux of Quebec (Canada), for providing funding for this project.
