Social recovery therapy for young people with emerging severe mental illness: the Prodigy RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 10/104/501. The contractual start date was in September 2015. The draft report began editorial review in September 2019 and was accepted for publication in December 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

© Queen’s Printer and Controller of HMSO 2021. This work was produced by Fowler et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2021 Queen’s Printer and Controller of HMSO

Scientific background

It is now widely recognised that most socially disabling chronic and non-psychotic severe and complex mental health problems begin in adolescence, with 75% of all severe and chronic mental illnesses emerging between the ages of 15 and 25 years. 2,3 A series of retrospective studies have consistently shown that severe mental illness is often preceded by social decline, that this often becomes stable, and that such premorbid social disability is predictive of the long-term course of the disorder. 4 Between 3% and 5% of adolescents present with non-psychotic severe and complex mental health problems associated with social disability. 2 The young people at highest risk of long-term social disability present with emerging signs of social decline, in association with low-level psychotic symptoms and emotional and behavioural disorder often accompanied by substance misuse problems, and risk to self and others. 2,3

Despite poor outcomes and the cost of disorders leading to social decline, young people with complex needs frequently do not access treatment, and less than 25% of young people and their families who have such needs access to specialist mental health services. 5–8 More complex cases are found in areas of social disadvantage and among those who are not in employment or education. 1 The economic costs of not addressing this disability are very high. 9 Persistent mental health problems associated with social disability in young people do not resolve naturally and may persist across the life course, resulting in severe distress and social disability, and high costs to health and a range of social and other services. 2,3 Health economic modelling of lifelong costs in this area is emerging. A recent estimate suggests that mental health problems in childhood and adolescence can result in problems across marital satisfaction, self-esteem and quality of life, and can lead to a 28% reduction in economic activity at age 50 years or a £388,000 lifetime loss per person. 10 Young people who have a combination of severe and persistent mental health needs and who are socially disabled present with problems that have the highest lifelong burden.

Several recent reports have highlighted that there is a major gap in identifying and managing the mental health problems of young people with non-psychotic severe and complex mental health problems, particularly those at risk of social disability. 5,7,8,11–13 New approaches to detection and intervention are required to meet the needs of these young people. 1 There is a gap in the evidence base for these young people. 1 Several National Institute for Health and Care Excellence (NICE) guidelines have highlighted this issue, including those for social anxiety,14 depression,15 and detection of people at risk of psychosis, and the research recommendation deriving from the NICE guideline on psychosis and schizophrenia in children and young people. 13

Young people who have non-psychotic severe and complex mental health problems and who are socially disabled are complex. Thus, they tend not to be suitable for or respond to short-term evidence-based therapies for more discrete mental health problems, such as cognitive–behavioural therapy (CBT) for anxiety, depression and conduct disorder, which are available via the Improving Access to Psychological Therapies initiative. Moreover, although this group show clear evidence of social disability, they do not meet the criteria for first-episode psychosis (FEP) and so they are not suitable for Early Intervention in Psychosis (EIP) services, for which there is now considerable evidence of benefits on social functioning. 4,16,17

Our aim in the present project is to identify and target the group of young people who are socially disabled and have non-psychotic severe and complex mental health problems and are at risk of long-term severe mental illness to offer them a new psychological intervention specifically tailored to their needs.

The most systematic service provision is often outside mental health services in statutory and voluntary sector provision for young people who are not in education, employment or training (NEET). In these services, the focus is primarily on obtaining employment and, thus, the mental health problems that present barriers to activity, work, education and training may not be recognised. 18 However, the degree to which NEET status is associated with mental health problems is increasingly recognised. 18–24 Detection of this population therefore must focus on the screening of the mental health problems of young people who have links with NEET services and who are under primary mental health care, alongside seeking referrals from those referred to Child and Adolescent Mental Health Services (CAMHS) and adult services. This group may be detected in services for those in the at-risk mental states for psychosis (ARMS), where these services are present.

Current evidence for effective interventions to address social disability among young people in the early course of severe mental illness is very limited. 4 A series of studies have been undertaken that aimed to identify people at ultra-high risk (UHR) of poor long-term outcomes associated with severe mental illness, focusing predominantly on risk of psychosis. 25–28 The success of the UHR studies is that they have shown that it is possible to set up services to identify and treat cohorts of young people who can be identified as having ARMS using defined operational criteria and structured assessment tools. 29 Furthermore, these studies have consistently identified that those who are at the highest risk are young people who present with social decline as well as subthreshold psychotic symptoms. 30,31 However, the focus of these studies has been on prevention of episodes or symptoms of psychosis, not social disability. 1 Recent studies have shown that cohorts identified using these criteria may have more transient problems than previously thought and that only a subset go on to have long-term socially disabling mental health problems. 30,31 Several prominent UHR researchers are now highlighting an alternative strategy, which is to examine functional outcome in the UHR group. This study is consistent with this strategy.

Systematic reviews of CBT for psychosis, including NICE guidelines, have consistently shown a moderate effect size on improvements in social disability where this has been assessed as a secondary outcome. 32 This has been confirmed in the recent review of the NICE guidelines for schizophrenia. 33 However, these studies have predominantly been carried out among chronic participants, not young people. 1 The feasibility of using CBT with young people who are at UHR of long-term poor outcomes has been shown in the recently completed, multicentre Early Detection and Intervention Evaluation for people at high-risk of psychosis 2 (EDIE-2) trial,34 which has shown reductions in the severity of psychotic symptoms. However, the focus of the therapy in EDIE-234 was symptom reduction,28 and this approach neither targeted nor had a significant benefit on social disability. EDIE-234 clearly demonstrated the ability of collaborating sites to recruit young people at high risk and successfully retain them in research and therapy. However, as described above, the group recruited in EDIE-234 were heterogeneous in terms of social disability. The present trial builds on EDIE-234 by focusing on a group that has a more homogeneous set of social disability problems, defined by low-activity levels, and targeting this group with a multisystemic intervention that specifically aims to address social disability.

Better outcomes on social disability and hopelessness can be obtained from a more targeted intervention specifically focused on improving social disability among those who have low functioning. We have developed a multisystemic form of CBT that targets social disability. 35,36 A successful Medical Research Council trial37 was carried out with a group of young people who had established chronic and severe social disability problems up to 8 years after a first episode of psychosis. This trial demonstrated gains in structured activity and hope, as well as reductions in symptoms,37 with evident long-term maintenance of gains in structured activity. 38 Clear indications of health economic benefits were demonstrated. 39 However, the trial was small and there was a high level of uncertainty associated with these estimates. A further, larger National Institute for Health Research (NIHR)-funded trial demonstrated gains in structured activity and reductions in symptoms for EIP service users who were experiencing their first episode of psychosis. 40

The intervention used in this study has been refined from experience in previous studies to apply to socially withdrawn young people with non-psychotic severe and complex mental health problems. 35,36 The intervention is considered developmentally appropriate for young people owing to evident gains for young people accessing EIP services. 40 Moreover, social recovery therapy (SRT) is a collaborative approach, led by the young person’s personally valued and meaningful goals. 36 This individualised goal focus means that SRT is in keeping with the adolescent and emerging adulthood stages of development, in which young people are beginning to individuate and separate from the family unit and to explore and develop their sense of self-identity. 41,42 Furthermore, SRT is in keeping with the youth perspective on social recovery, in which biographical disruption from mental health problems gives way to new meanings, identities and social connections. 43 To our knowledge, this trial was the first to specifically address both social disability and mental health problems among a high-risk population of young people presenting with social disability and non-psychotic severe and complex mental health problems.

Aims and objective

We aimed to undertake a definitive randomised trial to determine the clinical effectiveness and cost-effectiveness of SRT plus enhanced standard care (ESC) compared with ESC alone in young people who present with social withdrawal and non-psychotic severe and complex mental health problems, and who are at risk of long-term social disability and mental illness.

The primary hypothesis was:

• In young people who are socially disabled and have non-psychotic severe and complex mental health problems, SRT plus ESC is superior to ESC alone in improving social recovery [as measured by hours in constructive economic activity assessed on the Time Use Survey (TUS)] over a 15-month follow-up period.

The secondary hypotheses were:

• SRT plus ESC is superior to ESC alone in terms of cost-effectiveness.

• SRT plus ESC is superior to ESC alone in terms of effects on mental health symptoms [attenuated psychotic symptoms (APS) and emotional disturbance].

Design

This was a pragmatic, multicentre, single-blind, superiority RCT with ascertainment of the clinical effectiveness and cost-effectiveness of SRT delivered over a 9-month period plus ESC compared with ESC alone in young people (aged 16–25 years) with non-psychotic severe and complex mental health problems and showing early signs of persistent social disability. Primary and secondary outcomes were evaluated at 15 months post randomisation (i.e. 6 months after the end of intervention or control) and limited assessment of longer-term outcomes was evaluated at 24 months post randomisation.

Participants

Analytic plan

The analytic plan was detailed in the Statistical Analysis Plan (version 0.2, 4 July 2018) and approved by the Data Monitoring and Ethics Committee (DMEC). The primary analysis compared SRT plus ESC with ESC alone on time use at 15 months post randomisation. The primary analysis was on the intention-to-treat (ITT) principle (i.e. all participants were followed up for data collection irrespective of adherence to treatment and were analysed according to group allocation rather than intervention received). We also completed a per-protocol (PP) analysis of primary and secondary outcomes.

All hypothesis testing was at the two-sided 5% statistical significance level. CIs for parameter estimates were at the corresponding 95% level. Analyses were conducted by the trial statistician blinded to group identity (i.e. ‘subgroup’ blind). 1 Assuming a normal distribution (potentially of transformed values) for time use, a general linear model was constructed. This model included all stratification variables [i.e. recruiting site (as a random factor), age (16–19 years, 20–25 years), severity of social disability (withdrawn or extremely withdrawn) and meeting symptomatic criteria for an ARMS or not]. Time use at baseline was also included as a covariate. Logical memory at baseline was also included as a prognostic variable for long-term social recovery along with verbal fluency (total score). Treatment arm was included as a fixed effect. Estimation of model parameters was on the basis of ‘type III’ (or ‘adjusted’) least squares. The residuals from this model were examined to assess the normal distribution and homoscedastic assumptions. Transformations of the outcome were considered in the case that this assumption did not appear appropriate.

Secondary analyses were conducted in an analogous fashion through a generalised linear model, with an appropriate link and error term depending on the nature of the outcome of interest (e.g. a logistic regression model for binary outcomes). Stratification variables were again included in the model and, where available, the outcome variable at baseline was also included. Treatment group was included as a fixed factor.

The analytic plan included a consideration of the moderation of treatment effect with respect to the following baseline characteristics:

• ARMS29 status

• low or very low functioning according to structured activity hours per week at baseline46,47 Logical Memory I (Logical Memory Scaled Total Score61)

• COWAT62 total score.

This was to be achieved by using an interaction term (between the putative moderator and treatment arm) to formally assess differences in treatment effect. The sample size calculation did not include reference to hypothesis testing of interaction effects and this analysis was considered exploratory.

Internal pilot statement

The internal pilot (NIHR reference 10/104/501) was funded in March 2012 and began recruitment in January 2013. The aims of the pilot were to (1) assess recruitment rate, quality of data collection and follow-up, (2) provide a final check on procedures in the protocol and (3) conduct a qualitative substudy to inform the objectives, using a qualitative, service user perspective. The following stop–go criteria were used to determine if it was appropriate for the pilot to be considered an internal pilot and progress to the definitive trial: (1) no necessity for substantive changes to the protocol, (2) recruitment of 80% of the planned total in the first 12 months, (3) retention of participants within the study with baseline and outcome assessments completed for > 80% of participants for the secondary and other outcomes and mediators, and for 90% of participants for the primary outcome, and (4) satisfactory delivery of competent and adherent therapy to > 80% of the treatment group.

The stop–go criteria were all satisfied. No changes to the trial protocol were necessary. The pilot recruitment was on time and to target. Primary and secondary outcome retention was excellent, achieving 92% at both 9- and 15-month assessment points. Primary and secondary outcome retention at the later-added 24-month assessment point necessitated a new consent procedure and a completion rate of 69% was achieved. Of those randomised to SRT (n = 47), 41 (87%) were deemed to have received competent and adherent therapy. The pilot DMEC and Trial Steering Committee (TSC) reviewed trial data in 2014 and supported confirmation of the pilot as an internal pilot. These two committees and the PPI representatives (the PAT) reached consensus that no changes to the trial protocol were required. The sponsor also performed a full Good Clinical Practice audit of the trial and concluded that no changes to trial procedures were needed. Additional funding was sought (NIHR reference 10/104/51) for the extension and was awarded in May 2015, with recruitment beginning in September 2015.

Process evaluation

The PRODIGY process evaluation was an evaluation of both the research and the intervention processes undertaken as three substudies, two involving patient participants and one involving SRT therapists, during the initial internal pilot phase. The patient process evaluation was conducted by a subresearch team led by a user researcher (RB) and an independent qualitative researcher (CN), neither of whom were involved in the development of SRT. Both parts of the patient process evaluation (i.e. research and intervention process substudies) were investigated using a qualitative methodology with qualitative interview data collection (see Report Supplementary Material 3) and thematic analytic methods. 69 These substudies were published as project outputs. 70,71 The final part of the process evaluation was conducted by a separate subresearch team during the non-pilot extension phase and focused exclusively on SRT therapist experiences of therapy delivery.

Participant flow

Participant flow through the trial is depicted in Figure 1.

FIGURE 1.

The Consolidated Standards of Reporting Trials (CONSORT) flow diagram.

Intervention results

Main trial results

Serious adverse events

All serious adverse events (SAEs) and adverse events (AEs) were categorised and recorded from randomisation until completion of the final follow-up assessment. Definitions used for delineating events are provided in Appendix 4. All events were reported to the chief investigator, sponsor, local trial site, Clinical Trials Unit, and trial oversight committees (i.e. DMEC and TSC). All SAEs were reviewed by a clinician independent of the trial for possible relatedness. SAEs would have been reported to the NHS REC, as appropriate (i.e. deemed possibly related to the trial). Zero events were considered possibly related to the trial and, thus, none was reported to the REC. The frequency and nature of SAEs and AEs are reported in Tables 29–31.

Frequencies of AEs and SAEs are shown in Table 29. There were 167 reportable AEs in the trial, 83 of which were considered serious. All events were considered unrelated to the trial. It is noted that events are clustered around individuals (i.e. individuals who frequently report more than one event); therefore, the number of events exceeds the number of individuals reporting them. Moreover, an excess of both SEs and AEs in the intervention arm is noted. This excess can be explained according to the reporting mechanism (see Table 30). The number of AEs reported to RAs at assessment points was the same or very similar in both groups, but a subset of individuals in the intervention arm reported AEs to therapists during the intervention period, a mechanism that was not available to participants in the ESC-alone arm.

A discrepancy remained (see Table 30) in that more individuals in the SRT plus ESC arm than in the ESC-alone arm reported SAEs to researchers. However, it is also notable that the proportion of participants who completed all follow-ups was greater in the SRT plus ESC arm; therefore, events reported to researchers were further delineated with respect to follow-up point to allow SAEs to be calculated as proportions of completers. A smaller proportion of participants in the SRT plus ESC arm reported a lower number of SAEs to researchers at their 9-month assessment than participants in the ESC arm. At 15 and 24 months, a slightly higher percentage of SRT plus ESC participants reported SAEs to the researcher than participants in the ESC arm. In summary, apparent differences in the number of AEs appear to be due to the reporting mechanism and differences in the rates of missing data between the groups.

As shown in Table 31, the nature of AEs was similar in arms and was as expected in this vulnerable population group. The most commonly reported AEs and SAEs centred around suicidal ideation, self-harm and overdose behaviours, and suicide attempts.

Abstract

Background

Methods

Analysis

Three analyses were undertaken, where the following approaches were adopted in all analyses. In each, a within-trial analysis was conducted in which costs and benefits were estimated over a 24-month period, with costs incurred in the second year discounted at 3.5%. 85 A within-trial analysis, rather than a decision model, was undertaken as we are not aware of any previous studies that have compared SRT with ESC. The ITT approach was also followed, in which participants were analysed within the group to which they were allocated, regardless of, for example, the number of therapy sessions received. To estimate the mean incremental cost and incremental effect (QALY gain) associated with the intervention, compared with that for the control group, seemingly unrelated regression87 was undertaken to allow for any correlation between costs and effects. Baseline demographic variables [age (16–19 years, 20–25 years) and sex (male/female)] were included as covariates, along with the total baseline health professional and hospital admission cost for the overall NHS and PSS cost, and the baseline EQ-5D scores for the total QALY score. In the absence of dominance (where higher costs and lower benefits were associated with a particular intervention), the incremental cost and incremental effect would be used to estimate the ICER (mean incremental cost/mean QALY gain). 85 It was also assumed that an estimated ICER below the cost-effectiveness threshold (λ) value of £20,000 per QALY85 would suggest that an intervention constituted value for money. In addition, to estimate the level of uncertainty associated with the decision about whether or not the intervention was cost-effective, the bootstrap technique88 (with 5000 replications) was used to estimate the probability that each intervention was cost-effective according to the CEAC. 89 In particular, the probability of SRT being cost-effective was specifically estimated at the λ of £20,000 per QALY.

The first (base-case) analysis undertaken was based on overall NHS and PSS cost, and the total QALY score, where MI90 was used to estimate any missing data and enable all participants to be included. The MI model included costs (health professional and hospital admission costs at baseline, and at 9, 15 and 24 months, as well as total intervention costs) and outcomes (EQ-5D scores at baseline, and at 9, 15 and 24 months) and demographic variables [age, sex and social functioning (low functioning/very low functioning)]. EQ-5D scores were included, rather than individual dimension scores, as missing EQ-5D data were generally for the whole questionnaire. Health professional and hospital admission costs were, however, separated to allow for the possibility of different levels of missing data for these two variables. Two further complete-case91 sensitivity analyses92 were also conducted. In the first sensitivity analysis (SA1), participants were included only if they had complete cost and effect data at each time point. This enabled a comparison with the results of the base-case analysis, to assess whether or not results differed for participants who did not have missing data values imputed. A second sensitivity analysis (SA2) was also conducted, for a similar rationale, where only total intervention costs were included, as the level of missing data for this variable was anticipated to be lower.

Results

Discussion

The aim of this study was to undertake a definitive randomised trial to determine the clinical effectiveness and cost-effectiveness of SRT compared with ESC in young people who present with social withdrawal, and non-psychotic severe and complex mental health problems, and who are at risk of long-term social disability and mental illness. The primary hypothesis was that, in young people who are socially disabled and have non-psychotic severe and complex mental health problems, SRT will be superior to ESC in improving social recovery (as measured by hours in structured activity assessed on the TUS) over a 15-month follow-up period. There was no evidence of superiority of SRT over ESC at 15 months, nor at 9 or 24 months, with respect to time spent in structured activity. There was no evidence for the superiority of SRT over ESC in terms of cost-effectiveness or effects on secondary outcomes of mental health symptoms (APS and emotional disturbance). One secondary outcome, drug use, showed significant superiority in the SRT plus ESC arm compared with the ESC-alone arm at 24 months. The most appropriate summary of the results of this trial is that there is no evidence of superiority of SRT over ESC for young people presenting with social withdrawal and non-psychotic severe and complex mental health problems.

On some dimensions there appeared to be mean differences favouring ESC over SRT plus ESC. However, the differences on the primary outcome and the large majority of secondary outcomes did not meet the level for conventional significance, apart from social phobia and some subscales of negative symptoms at 15 months. At 24 months, mean differences favoured ESC over SRT plus ESC on structured activity. Missingness of data was consistently higher in the ESC-alone group than in the SRT plus ESC group, and the bias and total amount of missingness of data increased over time. At 24 months, > 30% of data were missing and there was a bias resulting from the fact that the amount of missingness of data in the ESC-alone group was twice that in the SRT plus ESC group. It is plausible that differential missingness of data could bias results in favour of ESC, particularly at the later assessment stages. We can be clear that there is no superiority of SRT, but, despite trends favouring ESC alone, we should be more cautious in concluding that ESC alone was superior, as most of the results did not reach conventional statistical significance levels and there was a clear bias because of unbalanced levels of missingness of data.

Participants in both arms of the trial made large gains over time from baseline on all measures. There were large effect size gains in structured and constructive economic activity of > 10 hours per week in both arms. This is more than double the 4 hours that constitutes a clinically meaningful difference. There was a > 50% improvement in the rate of participants meeting diagnostic criteria for depression, panic, agoraphobia and social phobia in both groups, and large effect gains in self-reported assessments of depression, social anxiety, hopelessness and schizotypal symptoms of paranoia and anomalous experiences. There were also marked reductions in alcohol and drug use disorders.

The ESC provided in this trial was designed and intended to maximise the delivery and availability of combinations of existing evidence-based interventions to this group of young people with non-psychotic severe and complex mental health problems. Close examination showed the ESC delivered in the trial to be highly active, involving comprehensive individualised packages of care with combinations of medication, evidence-based psychological therapy (including symptom-focused CBT) and social care (including specific employment support). Most participants in the ESC condition received case management support, a majority (in both arms) were taking antidepressant medication, around half had psychological therapy over the course of the trial, and around one-third had comprehensive employment support. The packages of care delivered outside the trial were similar in both arms, but slightly higher rates of receiving psychological therapy and packages of NHS treatment were recorded in the ESC-only group. What the trial appears to indicate is that the presence of active packages of youth mental health care, including primary care support, antidepressant medication and, where indicated, standard psychological therapy and employment and youth support, is sufficient to make relatively large effect size gains in activity and in mental health symptoms, and that the adjunctive provision of specific intensive SRT adds little to such gains. The results are possibly akin to those of the IMPACT trial93 on adolescent depression, which showed relative equivalence of the effectiveness of case management in comparison with different types of psychological therapy (psychoanalytical psychotherapy and CBT).

The results are surprising given the evidence base that gave rise to this trial. Previous research has tended to find that patients with severe and complex problems struggle to access and may be relatively unresponsive to existing interventions. 5–8 Furthermore, follow-up studies have suggested that the longer-term outcomes are poor, with social withdrawal in association with non-psychotic severe and complex mental health problems and APS predictive of poor outcomes. 2,3 In contrast, in this trial both the ESC-alone group and the SRT plus ESC groups achieved large effect size gains in structured and constructive economic activity, as assessed by time use, and in mental health symptoms, particularly in hopelessness, anxiety, paranoia and APS, over time and for up to 2 years. It is important to note that ESC in this trial was not treatment as usual. It was an active and comprehensive intensive treatment control condition comprising a combination of evidence-based interventions. As identified in our qualitative participant process evaluations,70,71 participation in the ESC arm of the trial and in the trial procedures appeared to be an enhanced intervention experience that was beneficial to participants. Future research should perhaps focus on how routine services can maximise the implementation of the optimal combination of existing evidence-based interventions for this often-neglected group of participants.

Strengths and limitations

This trial had many strengths, most notably that it is a large trial involving the recruitment of a very withdrawn population of young people with severe mental illness across diverse regions of the UK and recruited from a wide array of services. This is a difficult-to-research group for whom any consistent evidence of outcomes is difficult to obtain because of problems with non-engagement and retention. Recruitment and retention of this sample is in and of itself an achievement, and the information on outcomes for the cohort provides novel data on a group that is of increasing interest to policy-makers. The further strengths of this study were good internal and external validity for the trial on the primary outcome and on the main secondary outcomes. The study was conducted with a high degree of rigour and retention to the primary outcome was highly satisfactory. All researchers involved in the study received regular supervision and careful routine checks on inter-rater reliability throughout. All outcomes were reliable and completed blind to intervention allocation. SRT was delivered rigorously and thoroughly, with good adherence to the therapy model by a group of highly committed therapists, and was received well by participants.

A limitation of the study was that it was compromised by the level and pattern of missing data in the secondary outcomes, particularly at the 24-month assessment. However, it should be highlighted again that the primary outcome was originally designed to be assessed at 15 months and the 24-month assessment was brought in at a later stage as a specific standalone follow-up study. Although there were missing data, this was not regarded as being to a degree that compromised the use of ITT analyses as specified in the statistical analysis plan for the main study on primary and secondary outcomes at 15 months. The characteristics of the target group in this study, being by definition a difficult-to-engage and extremely withdrawn sample, represent a challenge to researchers, especially where longer-term follow-up assessments are reliant on face-to-face assessments. Notably, in the SUPEREDEN3 trial, which compared SRT plus EIP services with EIP services alone, there was also differential missingness of data with greater retention in the arm receiving SRT. 40 In the present trial, additional measures were used to improve retention across trial arms, including the provision of participant reimbursement at all assessment points, mid-point telephone contact and provision of a newsletter to participants between follow-up points; yet such measures did not appear to result in equivalent retention for those individuals not receiving SRT. Potentially, future studies in this area might maximise the rate of follow-up by focusing on hard proxy variables of engagement in services, which may be derived from health-care records rather than from face-to-face assessments, especially as engagement is itself an important outcome in this population.

A further limitation of this study was that many of the secondary outcomes have wide CIs (an indication of low statistical power), suggesting low precision or uncertainty in the estimation of treatment effects, which is probably an unavoidable consequence of undertaking research with what is a complex and heterogeneous clinical group with differing symptomatology. Furthermore, the study was, by necessity, single blind, and thus may be affected by the fact that participant were aware of their allocation and whether or not they were receiving the intervention of interest.

In line with good practice recommendations for cost-effectiveness analyses,94 we concentrated on expected large cost drivers (e.g. health professional visits). A potential limitation is, however, that certain aspects of health care (e.g. medications) were not costed. Similarly, any differences in the loss in productivity, as well as family and carer costs, were also not costed, although, given that the intervention was not estimated to be more effective than standard care, we would not expect there to be a difference in these costs between arms. Thus, the study result would have been expected to be the same, even if a wider cost perspective were adopted.

A further potential limitation is that our analysis is based on evidence from just one trial. However, if an economic model were to be created based on these trial data, then one would not expect the long-term or extrapolated results, for example, to differ from those presented here, as the intervention was not estimated to be effective.

Clinical implications

The assumption that underpinned this study was that young people who are withdrawn and have non-psychotic severe and complex health problems represent a group who are unresponsive to existing interventions and, therefore, may require a dedicated intensive psychosocial intervention. The findings of the study are important in clearly demonstrating that, contrary to prediction, this group can respond well and make clinical and social recovery if a comprehensive range of existing interventions are made available. What is needed is an enhancement of standard care to ensure the provision of the optimal combination of currently available evidence-based interventions. This includes a combination of case management and support with appropriate medication, specific symptom-focused psychological therapy, and employment and youth support. Service user feedback suggests that delivering such services in the context of hope and messages of recovery is key. 95–97 Similarly, both participants and SRT therapists in the present process evaluation emphasised the importance of hopefulness and motivation for this patient group (Briony Gee, Norfolk and Suffolk NHS Trust, 2021, personal communication). 98 The take-away message is not to do nothing; instead, it should be to highlight the need to assist services and frontline clinicians to enhance care delivery for this population and ensure that optimal care pathways are delivered for those most at need. Such care pathways must also recognise the dynamic identity exploration occurring in the those aged 14–25 years41,42 and the need for developmentally appropriate interventions for this group.

The key clinical and research implication of this study is to examine the optimal ways that services can be provided in such cases. Although the evidence here is that young people with non-psychotic severe and complex mental health problems can and do respond if the right combination of evidence-based interventions is made available, all too often such interventions are not available in routine services. The factors associated with effective delivery and implementation of the optimal combinations of interventions to this often-neglected group of young people are complex. One issue is resources and it should be noted that the present trial was conducted in services that already had a specific focus on youth mental health and the provision of interventions to young people with complex needs. In these services, an optimal combination of evidence-based interventions was available and was delivered to participants in the trial. Not all services in the NHS, let alone those in less well-resourced health-care settings, have such resources available. The present trial provides support for the provision of comprehensive youth mental health services that can offer a full range of medical, psychological and psychosocial interventions. Moreover, even where resources are available, implementation and delivery of care can be complex, as, young people may need help to be detected, to engage with services and to continue with interventions. Sometimes these young people can fall between the gap of services; for example, some of our participants had previous experience of being regarded as not having sufficiently severe symptoms to meet the criteria for EIP services, but, at the same time, their symptoms were deemed to be too severe or they were considered incapable of engaging with services for treatments to address common mental health problems. Others fall into gaps between child and adult services. This study suggests that good outcomes are possible by systematic provision of currently available evidence-based interventions. The results of this study are encouraging to all clinicians to do what is possible with the range of existing interventions to deliver optimal packages of care. If delivered in a spirit of hope for realistic recovery, good outcomes can be expected.

The qualitative research suggested that the trial procedures themselves provided a focus for optimal detection and delivery of interventions, including ESC. Possible ways to replicate such procedures to organise and implement detection and delivery of care to high-risk groups that may not otherwise engage in routine practice might be possible in more dedicated youth mental health services. These services offer open-door youth-oriented engagement and assertive case management to ensure that combinations of existing evidence-based interventions are delivered to those who need them. The qualitative research also revealed that hopelessness can potentially lead to non-engagement and lack of implementation of interventions, both within the young person and by staff who treat them, who can become overwhelmed by the complexity of clinical presentations. The results of this study, again, provide hope in suggesting that, despite initially highly complex and very severe presentations and extreme withdrawal over a period of 15 months to 2 years, many young people with complex and severe mental health and social problems improve significantly. We cannot be absolutely certain that it was the systematic detection and delivery of the right combination of existing evidence-based interventions that facilitated this; it is possible that some participants made natural recoveries. Given the chronicity, severity and complexity of problems at baseline, it seems probable that the enhancement of standard care and the systematic provision of evidence-based interventions according to need in both arms played a role in promoting recovery.

In conclusion, clinicians should be supported in their attempts to manage more complex clients with clear hope that systematic delivery of the right combination of existing interventions based on needs can be effective. Support and training to give clinicians confidence to manage such cases may also be warranted. The structure offered by the recruitment and assessment processes in this trial possibly provides systematic ways in which routine services can identify cases and offer hopefulness through a structured assessment of needs and feedback to clinicians.

Further research

Conclusions

The rationale for the trial emphasised the need for detection and intervention as it had been commonly highlighted that cases of young people with non-psychotic severe and complex mental health problems tended to be neglected by services. Even though these young people often seek treatment, they were identified as falling between gaps in services and being regarded as difficult to engage and manage. This study did not find evidence for the superiority of SRT when delivered as an adjunctive to enhanced treatment as usual. This study showed that it is possible to successfully detect, recruit and retain young people with severe and complex mental health difficulties and social withdrawal into a psychological treatment trial of this type. The participants recruited demonstrated very severe withdrawal, with a mean of 11 hours per week activity compared with a population mean of > 60 hours per week for young people in the same age group;47 this was accompanied by very severe levels of depression, social anxiety, hopelessness and the presence of APS.

The study shows that it is possible for standard services to deliver enhanced packages of care to this group if resources are available and suggests that these packages can be effective. The results provide support for and are consistent with lobbying for generic youth mental health services that seek to engage young people. The results of the present study show that, in terms of interventions, offering the range of existing services available within the NHS (case management, medication, symptom-focused CBT, and employment and youth-focused support) is sufficient to obtain a clinically important effect and that a more focused, intensive and specific psychological intervention, such as SRT, may not be needed. SRT was also not estimated to be cost-effective. Potentially, it may be only young people with psychosis who need a more intensive specialist intervention and for whom interventions such as SRT show a specific effect in enhancing standard care. 40 The clinical implications of the study are therefore significant in emphasising the importance of mental health services to identify, engage and treat the types of non-psychotic severe and complex mental health problems targeted in this trial using their existing repertoire of interventions, with the confidence of obtaining good outcomes even if the initial presentations can appear somewhat intransigent, and indeed complex and overwhelming.

Contributions of authors

Professor David Fowler (https://orcid.org/0000-0001-5806-2659) planned the funding applications for the internal pilot and extension phase of the trial. Professor Fowler led the design of the trial, and the development of the protocol and statistical analysis plan. Professor Fowler led the trial as the Chief Investigator with Professor Paul French. Professor Fowler wrote the final report with Dr Clio Berry.

Dr Clio Berry (https://orcid.org/0000-0003-1164-9836) made a substantial contribution to the development of the trial protocol and statistical analysis plan. Dr Berry managed the trial as the trial manager (extension phase): managing the delivery of the trial, ensuring the implementation of the protocol and associated amendments, supervising and training research staff, supervising and co-ordinating recruitment, and managed the trial data, including data entry and quality appraisal, with support from NCTU. Dr Berry prepared the final data set for analysis. Dr Berry made substantial contributions to the design, delivery and analysis of the process evaluation involving research therapists. Dr Berry wrote the final report with Professor Fowler.

Dr Joanne Hodgekins (https://orcid.org/0000-0003-4124-854X) made substantial contributions to the applications for funding, design of the trial and development of the protocol. Dr Hodgekins led the trial in the East Anglia site as Principal Investigator with Dr Jon Wilson. Dr Hodgekins provided expert training and supervision for all trial staff in the assessment of time use. Dr Hodgekins made substantial contributions to the design, delivery and analysis of the process evaluations involving young people. Dr Hodgekins provided trial therapy, supervised research and therapy staff, and contributed to trial management oversight. Dr Hodgekins made substantial contributions to and critically read the final report.

Professor Robin Banerjee (https://orcid.org/0000-0002-4994-3611) made contributions to the funding applications. Professor Banerjee contributed to the design and development of the trial protocol. Professor Banerjee made substantial contributions to the development and delivery of protocol amendments. Professor Banerjee contributed to trial management oversight. Professor Banerjee made contributions to and critically read the final report.

Dr Garry Barton (https://orcid.org/0000-0001-9040-011X) was the health economist. Dr Barton made contributions to the funding applications. Dr Barton made substantial contribution to the development of the trial protocol and statistical analysis plan. Dr Barton conducted the health economic analysis and made substantial contributions to and critically read the final report.

Dr Rory Byrne was the lead for PPI. Dr Byrne made contributions to the funding applications. Dr Byrne made substantial contributions to the design of the trial and development of the protocol. Dr Byrne led the PPI activity with the PAT. Dr Byrne made substantial contributions to the design, delivery and analysis of the process evaluations involving young people. Dr Byrne made substantial contributions to the design of patient-facing materials used in the delivery of the trial. Dr Byrne contributed to trial management oversight. Dr Byrne critically read the final report.

Dr Timothy Clarke made substantial contributions to the applications for funding, design of the trial and development of the protocol. Dr Clarke led the trial as the trial manager (internal pilot phase): managing the delivery of the trial, ensuring the implementation of the protocol and associated amendments, supervising and training research staff, supervising and co-ordinating recruitment, delivering therapy and supervising research therapists, and working with NCTU to set up the trial database. Dr Clarke made substantial contributions to the design, delivery and analysis of the process evaluations involving young people. Dr Clarke made substantial contributions to and critically read the final report.

Dr Rick Fraser made contributions to the funding applications, and the design and development of the trial protocol. Dr Fraser was the Principal Investigator for the Sussex site. Dr Fraser contributed to trial management oversight. Dr Fraser critically read the final report.

Dr Kelly Grant (https://orcid.org/0000-0001-5319-8127) conducted the statistical analysis with Professor Lee Shepstone. Dr Grant critically read the final report.

Professor Kathryn Greenwood (https://orcid.org/0000-0001-7899-8980) made contributions to the funding applications, and the design and development of the trial protocol. Professor Greenwood contributed to trial management oversight. Dr Greenwood critically read the final report.

Dr Caitlin Notley (https://orcid.org/0000-0003-0876-3304) was the qualitative research lead. Dr Notley made contributions to the funding applications. Dr Notley made substantial contribution to the development of the trial protocol. Dr Notley was the overall qualitative research lead for the process evaluations and made substantial contributions to the design, delivery and analysis of the process evaluations involving young people. Dr Grant critically read the final report.

Dr Sophie Parker (https://orcid.org/0000-0001-5596-7524) made substantial contributions to the applications for funding, design of the trial and development of the protocol. Dr Parker provided site co-ordination in Greater Manchester, including supervision of research staff and co-ordination of recruitment. Dr Parker provided expert supervision and training for all trial staff in ARMS and diagnostic interview assessment. Dr Parker contributed to trial management oversight. Dr Parker made substantial contributions to and critically read the final report.

Professor Lee Shepstone was the trial statistician. Professor Shepstone made contributions to the funding applications. Professor Shepstone made a substantial contribution to the development of the trial protocol and statistical analysis plan. Professor Shepstone led the statistical analysis, which was conducted with Dr Kelly Grant. Professor Shepstone made substantial contributions to and critically read the final report.

Dr Jon Wilson (https://orcid.org/0000-0002-5279-6237) made substantial contributions to the funding applications, design of the trial and development of the protocol. Dr Wilson was the Co-Principal Investigator for the East Anglia site with Dr Hodgekins. Dr Wilson contributed to trial management oversight.

Professor Paul French made leading contributions to the applications for funding, design of the trial and development of the protocol. Professor French led the trial as Co-Chief Investigator with Professor Fowler and was the Principal Investigator for the Greater Manchester site. Professor French provided training and supervision to research and therapy staff. Professor French made substantial contributions to and critically read the final report.

Publications

Data-sharing statement

We shall make data available to the scientific community with as few restrictions as feasible, while retaining exclusive use until the publication of major outputs. All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review. Further information regarding SRT and associated research is available at: www.socialrecoverytherapy.co.uk.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Analysis of secondary time use outcomes

Analysis of general psychopathology diagnoses continued

Analysis of negative symptoms

Per-protocol analysis of primary outcome

For the time use variables, we explored the analysis of the PP population as well as the ITT population. These results are presented in Table 53.

Time use linear models exploring interactions

When adding interaction terms (treatment arm × severity of social disability, or treatment arm × mental state risk) in turn to the linear models for the primary outcome measure (structured hours at 15 months), there is no evidence to suggest a difference between ESC-alone and ESC plus SRT arms, using either interaction term.

For the remainder of the time use outcome measures, when adding the interaction term (treatment arm × ARMS status), there is no evidence to suggest a difference between ESC alone and ESC plus SRT, for any of the time use variables. When adding the interaction term treatment arm × severity of social disability, Table 54 shows the relevant results (p < 0.1).

For the ITT population, there is moderate evidence at 24 months of a difference in structured activity hours and constructive activity hours (p = 0.032 and p = 0.048, respectively), both in favour of the ESC-alone arm. It is estimated that, at 24 months, hours of average structured acitivity is 36.3% higher in the ESC-alone arm than in the ESC plus SRT group, and hours of constructive activity is 37.7% higher in the ESC-alone group. These are slightly larger differences in averages than if not including this interaction term within the linear model.

For the PP population, there is moderate (p = 0.021) and weak (p = 0.051) evidence of a difference in structured and constructive activity hours, respectively, at 24 months in favour of the ESC-alone arm. It is estimated that, at 24 months, hours of structured activity is, on average, 44.8% higher in the ESC-alone group than in the ESC plus SRT group, and hours of constructive activity is, on average, 41.9% higher in the ESC-alone group. As with the ITT population, these average differences are slightly higher than if not including this interaction term within the model. There is also weak evidence (p = 0.070) of a difference between treatment groups of structured activity hours not including child care at 24 months, favouring the ESC-alone arm, with an average of 33.6% more hours. This outcome measure was not found to be significant when not using an interaction term within the model. It is to be noted that the CIs for these adjusted differences (for ITT and PP) include or are very close to zero, which may affect the significance of these results.

Analysis of missing data for primary outcome