Collaborative care intervention for individuals with severe mental illness: the PARTNERS2 programme including complex intervention development and cluster RCT

Article history

The research reported in this issue of the journal was funded by PGfAR as award number NIHR200625. The contractual start date was in June 2019. The draft manuscript began editorial review in January 2022 and was accepted for publication in August 2023. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ manuscript 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 article.

Copyright statement

Copyright © 2024 Plappert et al. This work was produced by Plappert et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2024 Plappert et al.

Background

Bipolar, schizophrenia and other psychoses are the single largest cause of disability in the UK,1 and yet probably one-quarter to half of people receive no specialist mental health care. 2 The prevalence of bipolar, schizophrenia and other psychoses in England, defined as the number of people on the general practice Quality and Outcomes Framework (QOF) registers for severe mental illness, was 0.8% for QOF year 2011–23 and has since risen to 1%. 4 Such numbers have a considerable impact on the economy, with total service costs for people with a diagnosis of schizophrenia and bipolar estimated as £3.8 billion in 2007 and likely to rise to £6.3 billion by 2026. 5

People with a diagnosis of schizophrenia, bipolar or psychosis have a significantly reduced life expectancy compared with the general population. 6 Two-thirds of the mortality gap can be explained by physical disorders. 7 This is primarily due to a combination of lifestyle factors and medication side effects contributing to cardiovascular and respiratory risk. Additionally, the diagnosis of other significant illnesses may be delayed because of diagnostic overshadowing.

The NHS England policy The Five Year Forward View for Mental Health highlights the low level of primary care engagement with this group: ‘We should have fewer cases where people are unable to get physical care due to mental health problems … we need provision of mental health support in physical health care settings – especially primary care’ (p. 11; Contains public sector information licensed under the Open Government Licence v3.0). 8

Poor continuity of care and lack of information exchange between primary and secondary care also create barriers to effective support. The PARTNERS1 review of primary care records found that approximately 31% of people with a diagnosis of schizophrenia, bipolar or other psychosis in the UK were seen only in the primary care setting and that those seen in secondary care received only minimal support. 2 Primary care practitioners find it difficult to effectively support patients with serious mental illness (SMI), often lacking the necessary time and training to address these patients’ mental health needs. 9–11 Furthermore, access to health prevention and promotion activities in primary care is reduced for people with SMI. 12,13

Recent UK policy has promoted joined-up care, including the integration of primary and secondary services to provide better care for harder-to-reach groups. 14 The recent NHS Community Mental Health Transformation policy15 also aims to address this problem by ensuring that all those with a diagnosis of schizophrenia, bipolar or other psychosis and requiring care are well supported, ideally by collaboration between primary care, secondary services and third-sector organisations. This is also consistent with person-centred care (e.g. the Comprehensive Model of Personalised Care),16 which offers an integrated approach to health care for people with complex needs and provides proactive support for physical health conditions.

A collaborative care model, whereby a specialist healthcare professional works in primary care forging collaboration between primary and secondary care, is a potential approach to achieving better integration between separate parts of the health and social care system. 1 Collaborative care has been shown to be effective in improving mental, physical and social functioning across a range of mental health conditions. 2,17,18

Most of the collaborative care evidence for SMI has been developed and evaluated in the USA, where the nature of service user populations and of service use differs from the way we fund, structure and use NHS England. 19 Although there is considerable evidence of effectiveness, this largely relates to depression. 18

While those with severe symptoms of psychosis generally receive significant attention from services, it is clear that for those at lower risk care is much more haphazard. This is aggravated further by the increases in discharges from specialist services over recent years. There have been sustainable approaches to providing specialist mental health input for those with psychosis in primary care in only a few settings, such as East London. 20

Studies over time report that primary care practitioners find it difficult to effectively support this group of patients, often lacking the necessary time and training. 9–11 There is often poor continuity within primary care teams and poor co-ordination with secondary care.

In light of the above, this research programme aims to further understanding of the nature of current care, decide which outcomes are important, and then develop and test a collaborative, person-centred intervention to address deficits in care. It focuses on the estimated 70% of adults with a diagnosis of schizophrenia or bipolar who are currently seen and treated in primary care alone, or those currently seen in secondary care with lower levels of risk (operationalised as diagnostic clusters 11 and 12). Risk in this context refers to self-harm, self-neglect, suicide or harm to others.

How to navigate the PARTNERS2 research programme report

The National Institute for Health and Care Research (NIHR)-funded PARTNERS2 programme ran between 2014 and 2021, across two phases consisting of a total of six workstreams (WSs) (Figure 1), all supported by three Lived Experience Advisory Panels (LEAPs).

FIGURE 1.

The PARTNERS2 programme.

Phase 2: randomised trial, cost-effectiveness study and integrated process evaluation (workstream 6)

In this stage we conducted:

• a cluster randomised controlled trial (RCT)

• a cost-effectiveness study

• a mixed-methods process evaluation.

Public involvement underpinned all of the WS activity through the study LEAPs during the programme life cycle and the employment of service user researchers in the project team.

Alterations to the programme

The original programme proposed an external pilot trial of 300 participants from 60 general practitioner (GP) practices across three sites and did not include any effect sizes, power analysis or primary outcome. In 2016, a request was made to change the original design to carry out a definitive trial within the programme remit. The changes were agreed, converting the external pilot to an internal pilot trial with the following modifications:

1. agreement of a primary outcome and re-estimation of required sample size

2. reappraisal of sample size based io internal pilot

3. conversion to a definitive trial including stop–go rules set prior to trial onset [in conjunction with the Trial Steering Committee (TSC) and NIHR].

The trial arm of the programme [now split into two phases: development of feasibility stage (formative evaluation) and RCT (including with internal pilot)] is detailed in Table 1.

Early in the programme, the COS work was expanded and streamlined. Instead of creating a COS for SMI, after consultation with service users, carers and practitioners a decision was made that both a COS for bipolar and a COS for schizophrenia were needed. This was because two outcome sets for bipolar and schizophrenia emerged during the preliminary qualitative work. The decision was made that the final COS would relate to bipolar only rather than to schizophrenia or to SMI generally (the PARTNERS2trial population). This was because of the larger number of available data on and greater research and advisory team expertise relating to bipolar. Within the resources of the PARTNERS2 programme, it was possible to create only one COS using robust methods and including user involvement. A COS for bipolar was selected and created.

During the formative evaluation phase, flexible approaches for recruitment were tested. However, Health Research Authority guidance changes meant that these processes were no longer permitted in the second phase due to confidentiality concerns. Additional changes in one of the host NHS trusts (Lancashire Care Foundation Trust) causing financial and time pressures meant that the site had to withdraw during the set-up phase. The programme was allowed to continue along with a change in the co-ordinating Clinical Trials Unit (CTU) (from Birmingham to Plymouth, in line with RB taking on a co-chief investigator role). At this stage it was also agreed that the health economics modelling and stated preference survey planned for phase 1 would not proceed, with resources reallocated to the trial. Alternative sites were recruited [Somerset Partnership Trust and Livewell Southwest Trust (Plymouth)].

Following slower-than-hoped recruitment, a re-estimation of sample size (as per protocol for internal pilot) and changes in power, requirements decided following input from the Programme Steering Committee and the funder meant that we were able to stop recruitment in January 2020, completing the trial within the funding envelope. Table 1 details the trial development, including changes.

After recruitment closed and during intervention delivery and follow-up, the COVID-19 pandemic and UK national lockdown necessitated significant programme changes. After consultation with the Programme Steering Committee and the funder, the research team adapted methods for intervention delivery, follow-up data collection and the ongoing process evaluation (Table 2). The changes were tested for an 8-week period. Once they were shown to be successful, and after risk assessments were conducted for both service users and practitioners, the decision was made to continue remote delivery, where possible, both of the intervention and of follow-up and process evaluation data collection, until the end of the trial.

Phase 1: developmental and preparatory studies

Phase 1 included research to further understand the context (WS1), trial science work (WSs 2, 4 and 5) and intervention development (WS3).

Assessment of local care pathways and current services

Update of Cochrane review of collaborative care

Initial programme theory development

The PARTNERS2 programme theory for the final intervention model was developed over three phases:32

1. drafting of an initial model and development into a prototype

2. refinement of the prototype model through formative evaluation into the trial intervention model

3. further refinement of the intervention model through the RCT with parallel process evaluation.

The process of developing the initial PARTNERS2 model has been published in Gwernan-Jones et al. (2019),32 which includes a supplementary file of the explanatory statements making up the intervention.

Stakeholder involvement events included regular consultation meetings with LEAP members over the full period of the research project (2014–21); two consultation meetings with LEAP members, researchers (including service user researchers), practitioners and policy-makers during the development of the initial model (November 2014 and January 2015); and a stakeholder meeting (October 2016) during the formative evaluation attended by researchers, practitioners delivering the partners service and LEAP members. Stakeholder input fed into model development by contextualising, shaping and providing feedback on the practicality and relevance of proposed aspects of the intervention.

An intervention was proposed in the original research programme funding application based on a recent systematic review of collaborative care for psychosis21 and the model of chronic care33 framed by concepts of personal recovery. 34 Using this proposal as a foundation, and drawing on a realist approach,35 a directory of 453 explanatory statements was iteratively created that articulated a rationale for why, how and for whom it was perceived the different aspects of the intervention would work. This directory of explanatory statements drew on a number of data sources:

• literature on collaborative care for mental health, and personal recovery literature

• 11 telephone interviews with key leaders in collaborative care and personal recovery to explore their perceptions of best practice

• six focus groups with 33 participants living with psychosis (13 women and 20 men) about their experiences of care.

Through cycles of discussion and consensus building, researchers and stakeholders debated the proposed approaches to delivering the intervention, and their decisions guided consolidation of 106 explanatory statements and graphic and written representation of the programme theory in the form of a prototype model. Finally, the prototype model was interrogated internally (by comparing the explanatory statements with the practitioner manual) and externally (by comparing the explanatory statements with National Institute for Health and Care Excellence (NICE) guidelines), and then through debate across the research team, refined to establish consistency and rigour within the model and across the supporting documents. A manual for practitioners (care partners and supervisors) and guides for service users and their friends and family were developed.

Formative evaluation of the PARTNERS service

Description of the PARTNERS2 intervention

The intervention is complex, involving change at three levels, namely institutional level (secondary care trusts/CMHTs and primary care), practitioner level (care partners, supervisors, other primary and secondary care staff, third-sector and community organisational staff) and service user level (service users and friends and family, where there is consent). The 14 components are described in Table 3, and the relationships between contexts, resources, mechanisms (reasoning and reactions) and intermediate and intervention outcomes are shown in Figure 2. The manual is available as a supplementary file.

FIGURE 2.

Depiction of the PARTNERS2 initial model.

Development of a set of outcomes for a randomised controlled trial to evaluate the effectiveness of the PARTNERS2 intervention (workstream 2)

Development of recruitment processes (workstream 4)

Setting up the randomised controlled trial (workstream 5)

Transfer of both the intervention and the trial procedures to the RCT was not straightforward because of a combination of NHS pressures and the implementation of more stringent national research governance measures.

We had assumed that secondary care partners in the feasibility study would continue their participation into the RCT stage. However, changing NHS financial landscapes, staffing resource pressures, and GP (clusters) recruitment delays led to two trusts (Devon and then Lancashire) withdrawing. Consequently, we lost care partners who had been trained over a period of 2 years, including in the feasibility phase. Replacement sites to identify and train care partners were approached based on access considerations for the research team, and maintaining sociogeographic demographics (e.g. an urban/rural mix). We required substantially greater numbers of GP practices for the RCT. Although our approach mirrored the feasibility stage, it had not been possible to test the time and staff resource required or the local variation in response when recruiting a larger sample.

We intended to identify potential participants for the trial using the process trialled in the feasibility stage. However, local, changing, interpretations of research and information governance requirements meant that it was deemed inappropriate for the research team to view patient records prior to patient consent, even with permission from the practice and ensuring that no data left the practice. Therefore, primary/secondary care staff were required to identify potential participants. This is a time-consuming task, the reallocation of which led to a delay in participant recruitment. The onerous nature of this task also adversely affected practice recruitment. There were differences across regions in whether CRN-funded NHS research staff could and should support hard-pressed practice staff to carry out this screening of patient records.

We also sought ethics permission to send an invitation letter, followed by an accompanying telephone call, to potential participants who had not returned an expression of interest. The follow-up telephone call had to be undertaken by a clinician because of the governance changes noted above. Not all sites had the resource to provide this call; later in the programme, permission was obtained for NHS research staff to make these calls, which was associated with a boost in recruitment.

Phase 2

Phase 2 consisted of a RCT (initially an internal pilot trial), a health economics cost-effectiveness analysis and a mixed-methods parallel process evaluation.

Cluster randomised controlled trial

Background

The aim of the definitive trial was to assess the effectiveness and cost-effectiveness of the developed primary-care-based collaborative model (PARTNERS) for people with a diagnosis of schizophrenia, bipolar or other psychoses on improving QoL. The protocol paper for the trial has been published as Plappert et al. 46 A paper detailing the trial results has been accepted for publication (Byng et al. ; Figure 3). 47

FIGURE 3.

CONSORT diagram. SD, standard deviation.

Cost-effectiveness study

Delivery and fidelity analysis

Results

A care partner was in place for at least 70% of the intervention period for 75% (15/20) of the intervention practices, and the majority of intervention group participants (91%) had at least one care partner interaction of any type. Care partners reported discussing goals with 101 participants (87%).

Of the 116 participants in the intervention group, fidelity questionnaire data were available for 81 (71%). The distribution of responses (frequencies and per cent) to individual items is reported in Table 11 (see Appendix 2). Figure 4, from which ‘not applicable’ responses were excluded, shows the proportion of those receiving the intervention (and who considered the question applicable to them) who recognised if they had received key components. Over half of the respondents to the fidelity instrument reported that 19 out of 26 items definitely happened. Overall, the data suggested that fidelity to the PARTNERS model was more likely to have occurred with interpersonal practices (e.g. listening and understanding, 89% of those who considered this question applicable to them, n = 70/79) than activities related to the co-ordination of care (e.g. linked to GP, 53% of those who considered this question applicable to them, n = 17/32).

FIGURE 4.

Participant responses to the fidelity questionnaire. Figure excludes ‘not applicable’ responses. Questions 1–14: 0 = didn’t happen, 1 = possibly happened, 2 = definitely happened; questions 15–26: 0 = no, 1 = to some extent, 2 = yes.

Future work will investigate the psychometric properties of the PARTNERS2 Collaborative Care Fidelity instrument, including divergent (or discriminant) validity, item homogeneity and data reduction analyses.

Realist process evaluation

Reflections on what was and was not successful in the programme

Key successes of the programme included iterative development of a theory-based intervention valued by individuals; an in-depth quantitative description of standard care; a COS for bipolar; an analysis of our partnership with service user researchers and the LEAPs; methodological innovations for complex intervention evaluation; and the delivery of a cluster RCT in adverse conditions. We report by WS and then more general reflections to provide a narrative over the life of the programme.

Limitations

The limitations in phase 1 are covered above. Here we focus on the strengths and limitations of the trial, the process evaluation and the cost-effectiveness analysis.

The strengths of the trial included the real-life setting, with high proportions of recruited individuals from deprived rural, urban and coastal areas of England. We did not recruit as ethnically diverse a sample as we had hoped, with figures lower than those we found in WS1.

The cluster RCT design was considered necessary to avoid contamination within practices, but it resulted in an uneven randomisation of individuals because there was more variation in practice size (cluster size) than had been anticipated. The delivery of the intervention was suboptimal due to several care partner changes in employment and sickness affecting each site. We did not aim to quantify any differences in service user outcomes between care partners, but we did show clear differences in delivery between care partners in the case studies. Our process evaluation captured a wide range of data, and although we would have liked more data from interactions between the care partner and service user or the care partner and supervisor, and it is possible to get only a small window into intervention delivery over 2 years, we consider that the analysis revealed clear examples of potential benefit (and underlying mechanisms) in spite of the challenges of implementation, which were also revealed in detail.

The nature of psychosocial interventions and informed consent meant that blinding of participants was impossible. Although most of the primary trial outcome measures were carried out through self-report completion of questionnaires, some individuals required support in completing these, and, in a proportion of these for logistical reasons, the researchers were unblinded. There was also a problem in collecting two of the planned measures. First, the Brief-INSPIRE measure, which aimed to examine experiences of recovery-orientated care, was not analysed inferentially due to significant between-site differences in collection rates and both the numbers and patterns of missing data at both baseline and follow-up. Second, the physical healthcare process data were not completed due to reductions in permitted researcher presence in surgeries during COVID-19 pandemic conditions. This also limited the economic evaluation.

Despite outcome measure selection being based on an analysis of the intervention theory and predicted stages of change, it is recognised that health and QoL outcomes are often impacted downstream. It is notable that with person-centred interventions, changes may be highly idiosyncratic or individualised and these changes, however important to the individual, may be missed by standard outcome measures. Even though the delivery of PARTNERS was compromised, the process evaluation provided evidence that key elements were routinely delivered and that for some people meaningful benefits were evident. This adds weight to the possibility that, currently, trials of person-centred approaches are limited by our inability to measure initial unpredictable benefits that could lead to later changes in health status or QoL.

Conclusions for whole programme

The 7-year PARTNERS2 programme incorporated several substudies within two phases. Highlights of new knowledge gained are recorded here. During the first phase, a range of studies were carried out. The detailed analysis of care across three sites in England demonstrated significant variation in the proportion and extent of care. The Cochrane review of collaborative care provided limited evidence indicating that collaborative care may be more effective than standard care. However, our confidence in these findings is extremely limited because of the low certainty of evidence. Furthermore, the review pointed to a lack of detailed process evaluations of collaborative care.

The COS for bipolar was developed through engagement of stakeholders examining literature and a stepped Delphi approach. This provides a comprehensive set of 11 outcome domains, including symptom recovery, connectedness, patient experience, self-monitoring and the use of coercion.

The intervention development’s realist synthesis process led to a detailed programme theory for the PARTNERS2 intervention. This is laid out in an intervention manual (available via the website https://www.plymouth.ac.uk/research/primarycare/mental-health/partners2) and supported by the range of other documents co-produced with our LEAP. At the heart of the intervention is a coaching-based approach to working with individuals with SMI in a primary care setting. This approach helps individuals identify important personalised goals in the areas of physical health care, social outcomes or psychological problems, and supports achievement of these. This is in line with current policies but significantly different from the longstanding Care Programme Approach, which was used until 2022. A substantive implementation package involving training and top-up training, as well as supervision (in addition to the manual support), was developed to support delivery of the PARTNERS approach.

Phase 2 involved the main trial, including an analysis of quantitative outcomes, a realist process evaluation and a cost-effectiveness analysis. No significant differences were found using standard outcome measures for assessing QoL, mental health well-being, personal recovery or time use. The net costs and health benefits (QALY) were uncertain, with no indication of statistically significant differences between usual care and PARTNERS.

The delivery of the intervention was suboptimal, with significant absences of care partners for key periods, inconsistent supervision and a significant proportion of individuals choosing not to engage in the intervention; the last included when service users’ care passed to a new care partner, with engagement diminishing for some as a result.

The in-depth process evaluation also showed the extent to which several practitioners were able to practise the nuanced aspects of coaching and person-centred care. However, it also revealed the difficulty some individuals had responding to this and identifying and working on goals. Additionally, it showed, when care partners were available and individuals engaged, that for some individuals support led by goals generated short-term positive but not longer-term effects. We found a generalised mechanism, with some participants identifying how the care partner acted as a ‘professional friend’, offering support and encouragement through coaching. This aspect of the intervention effect was not covered in the measures used in the trial.

Recommendations for future research

PARTNERS2 implications for research build on the commentary above, outlining what went well and less well in this programme. In phase 1, the in-depth study quantifying care received by individuals with SMI demonstrated the lack of integration of electronic health records. It is inefficient to carry out case-by-case data extraction in the way we did to understand patterns of care; this points to the need for a new approach to routinely keeping track of care across and within systems. At a population level, we should be able to describe care received by individuals in different care clusters, including whether they are getting mental and/or physical health care input, and from which team. Support systems to organise the significant data already captured into a more usable form for such population-based monitoring would provide the basis for quantitatively monitoring care received in real time and context. This could supplement more qualitative assessment, either across large numbers of individuals using surveys or in-depth interviews for smaller numbers. We need research into how to reliably measure how much care groups of people sharing characteristics such as a diagnosis, poverty or ethnicity are receiving. Such a population-based approach is arguably more important than the more commonly used team- or service-based approaches, which miss out those not managed in secondary mental health services. This is in keeping with the move to population health management in Integrated Care Systems.

It is interesting to note that the trial itself recruited twice as many people with bipolar as with schizophrenia. We might not have been able to clearly communicate the ideas behind PARTNERS to people with schizophrenia and other psychoses and sufficiently engage them so that they became interested enough in our proposition to meet us and consent to take part. While we involved service user researchers in developing protocols and personalising information booklets with messages from service users and carers, and went through several iterations of trying to make our approach more flexible, we think that the research community needs to rethink current ethics and governance standards/requirements, which may, through their concerns about preventing harm, inadvertently be contributing to the non-involvement of people with schizophrenia and other psychoses in research processes.

Realist methods for developing interventions are still relatively unusual, and our approach appeared to be capable of generating an intervention that was acceptable to practitioners and service users. Relatedly, the realist process evaluation alongside the trial showed the value of both depth and breadth of data from different sources, including interviews from a range of stakeholders as well as observation and video. We are aware, however, that trying to capture the process of engaging individuals in a complex intervention over a period of months across several regions is a challenge and, inevitably, biased in terms of both selection of participants and availability of data. The research process relies, from a capacity point of view, almost entirely on paid researchers who are distant from events, whereas it is the practitioners and service users who are present and potentially able to record what is happening in detail. New techniques are therefore required for process evaluation work to engage practitioners and patients in the research process, taking into account the former’s capacity and the latter’s busyness. There is potential for this to cross over with reflective practice and service user involvement and with co-production for services more generally. A potential hybrid between the formative and process evaluation techniques used to evaluate complex interventions and methods used in quality improvement could be developed for use, for example, in the implementation of the CMHF policy, where there are many uncertainties about how to proceed.

Lastly, given the weaknesses in intervention implementation, there is a question of whether a further RCT should be carried out to determine whether person-centred coaching approaches to collaborative care are helpful or not. We are cautious about suggesting further trials both because the current NHS context is not always supportive and because of doubts about whether a single primary outcome measure can capture the multiple possible outcomes of potential importance in person-centred care. An alternative is to shift the emphasis to a combination of learning from innovation and implementation, and the development of mixed-methods approaches where the quantitative measurement of ‘reach’ to poorly served populations, experience of care and cost of intervention delivery is combined with qualitative approaches. The latter could include both high-volume, low-intensity data and in-depth observational approaches. Randomisation could still be used but perhaps with ‘reach’ and equitable access or patient experience as the key outcomes rather than health or QoL. Indeed, our measure of intervention fidelity illustrated that the aspects of care more related to care partner action were more clearly recognised as having been delivered. Those aspects that were more related to liaison with local services and within practice were more likely not to have been delivered. Fidelity to the PARTNERS model was more likely to have occurred with interpersonal practices (e.g. listening and understanding, 89%, n = 70/79) than with liaison activities (e.g. linked to GP, 53%, n = 17/32).

Implications for practice and lessons learnt

The PARTNERS2 programme provides evidence in relation to four aspects of care for individuals with schizophrenia, bipolar and other psychoses: what kind of interactions are important to service users and what practitioners might undertake, how practitioners can be supported in this care, how care is organised, and how structures and processes might support such system-based changes.

The PARTNERS intervention is based on sets of practices, those involving direct contact with service users and those more related to liaison and co-ordination. It is possible that 16 components were too many. The coaching approach tested in PARTNERS2, while not shown to have a direct impact on composite and mental health outcomes, was often experienced as positive (by both care partners and service users) when engagement was achieved. While a minority of individuals did not wish to continue with the care that was offered, most of those responding to the fidelity questionnaire and process evaluation interviews suggested that PARTNERS care was valuable. The process evaluation suggested that rather than the active goal-oriented practice of coaching being key, some individuals most valued the presence and interactions of practitioners as ‘professional friends’: someone alongside them whom they could trust and who could really understand them. It is interesting to consider whether or not such relationships needed to be ongoing. The process evaluation also showed that a wide variety of practices and short-term outcomes resulted from the motivational coaching approach. Some individuals made significant gains towards social goals or goals more related to psychological well-being, and few engaged in more proactive care of physical conditions. There was evidence that for some people getting to know, trust and have ongoing contact with a practitioner was important, while for others doing some work and then coming back at a later point was also experienced as helpful. This is important for those developing services for people with SMI as part of the CMHF transformation programme, where systems may be incentivised to provide time-limited interventions. Perhaps most importantly, the qualitative results showed that having a practitioner really listening and providing support with what the individual cared about most, rather than focusing on a model of fixing deficits, was generally appreciated.

The other key component of clinical care is liaison, and while we were not able to directly observe liaison, we did learn how the key partnerships required with the GPs and nurses in primary care teams, and with a large variety of voluntary sector organisations working with both, was challenging for care partners. Support to do this through supervision, but also organisational relationship building with primary care and the voluntary sector, is vital to ensure a supportive context for collaborative work.

Supporting practitioners to shift from a focus on risk and/or fixing deficits is a challenge that needs to be addressed in many aspects of community-based care where individuals with complex needs such as frailty, dementia or homelessness are also being asked to take a different approach. The training we provided for PARTNERS2, in particular the ongoing top-up training and support provided by the research team, was generally valued as a way of helping individuals see what needed to change. Even this relatively basic combination of initial and top-up training is often absent in the NHS during system change. However, embedding supervision in the PARTNERS model was difficult, and more work is needed to address why supervision proved so hard to deliver.

The detailed tape-assisted recall work also showed that practitioners had a tendency to feel that they provided person-centred coaching approaches when they had not shifted from more paternalistic stances. The tape-assisted recall interviews can be seen as a way of supporting practitioners to understand what was happening in their interactions and has parallels with techniques used in some training programmes, such as observation of practice or the use of videos and subsequent reflective practice. While these were time-consuming to perform for the research team, and not all practitioners consented to participate, where achieved this is a thorough method of validating good practice and helping practitioners see where improvements can be made.

Supervision delivery was weak due to pressures on supervisors to manage teams or large caseloads, or because of illness, meaning the research team often had to step in. When supervision occurred, and particularly when cases were discussed in depth and supportive challenge was provided, supervision was perceived as helpful overall. We conclude, therefore, that PARTNERS2 probably demonstrates the need for a package of support for practitioners, which might include written manuals, training, top-up training, clinical discussions as a team, peer support and supervision. This challenge for shifting practice and developing a supportive culture focused on learning to change is particularly important during the implementation of the CMHF, which requires new roles and changes in thinking and practice.

The two key organisational shifts in PARTNERS care are a shift towards more proactive care for those receiving only general practice mental health care, and a shift from secondary care to primary-care-based care with a specialist input. While the trial did not show measured benefits, there was no evidence of deterioration in terms of the primary outcomes and the safety variables. In particular, there was no indication of an increase in crisis for those receiving the intervention. Therefore, although the trial results do not show that current policy would lead to benefits, they do provide evidence that it is unlikely to be unsafe to shift individuals with SMI who are progressing their recovery well to primary-care-based care with specialist support. The results from phase 1 demonstrated very great variation in the extent of current treatment as usual, with some people receiving large amounts of care and others very little secondary care. Similarly, in the trial, there were differences with engagement, pointing to a need for both flexibility and assertive engagement (rather than the current practice of discharge during disengagement). Such practices also potentially necessitate a shift in culture (shared thinking, language and behaviour) within services. Our study did not address this directly; it is likely that such changes can be bottom up through shifts in practitioner thinking but can also be supported by thoughtful leadership activities.

Our final reflection as a team is how important the way we have worked together has been for the study. We have people in varied roles as care partners, LEAP members, research assistants, co-applicants, CTU staff, administrative support; everyone has been vital. Learning together and delivering this complex intervention as a team has been crucial, and we remain committed to this work (with PARTNERS3 currently ongoing) so we can continue to test these ideas and achieve better outcomes for people with schizophrenia, bipolar and other psychoses across England.

Acknowledgements

We would like to thank all service users and practitioners who participated in the PARTNERS2 process evaluation and randomised controlled trial. We had a large LEAP whose members contributed to this research project and are not authors of the report, but we would like to acknowledge the contribution of each member: Dawn Allen, Gail Benton, Lindsey Cree, Shaimaa El Naggar, Stewart Hendry, Beverly Jones, Hameed Khan, Dougie Mclellan, Mary Mancini, Manoj Mistry, Mary Nettle, Juliet Rawlins, Angela Sanders, Deb Smith and Diane Wright.

We acknowledge the work of the late Professor Helen Lester, who led the original application for the research, and whose ideas and ethos permeated the programme.

Contributions of authors

Humera Plappert (https://orcid.org/0000-0002-0959-2394): Department of Social Work and Social Care Programme Manager, University of Birmingham. Programme manager, programme design and management, intervention development, trial science development, randomised controlled trial and report writing.

Richard Byng (https://orcid.org/0000-0001-7411-9467): GP and Professor in Primary Care Research, University of Plymouth. Co-chief investigator, original conception, programme design and management, intervention development and formative evaluation, trial science development, randomised controlled trial, process evaluation and report writing.

Siobhan Theresa Reilly (https://orcid.org/0000-0003-4372-4415): Director – Applied Dementia Studies, University of Bradford. Site lead, original conception, programme design, Cochrane review, trial science development, randomised controlled trial, process evaluation and report writing.

Charley Hobson-Merrett (https://orcid.org/0000-0002-2990-2871): Research Fellow, University of Plymouth. Programme design, trial science development, randomised controlled trial, process evaluation and report writing.

Jon Allard (https://orcid.org/0000-0003-3778-0015): Honorary University Fellow, University of Plymouth. Intervention development, trial science development and report writing.

Elina Baker (https://orcid.org/0000-0001-9064-5897): Consultant Clinical Psychologist, Livewell Southwest. Intervention development and formative evaluation, and report writing.

Nicky Britten (https://orcid.org/0000-0002-7533-414X): Professor of Applied Health Care Research, University of Exeter. Original conception, lead qualitative analysis, programme design, intervention development and formative evaluation, process evaluation and report writing.

Melanie Calvert (https://orcid.org/0000-0002-1856-837X): Professor of Outcomes Methodology, University of Birmingham. Original conception, core outcome set lead, trial science development and report writing.

Michael Clark (https://orcid.org/0000-0003-4964-5005): Associate Professorial Research Fellow and Research Programme Manager, NIHR SSCR Care Policy and Evaluation Centre (CPEC), The London School of Economics and Political Science. Original conception, programme design, intervention development, randomised controlled trial, process evaluation and report writing.

Siobhan Creanor (https://orcid.org/0000-0002-7373-8263): Director of Exeter CTU; Professor of Medical Statistics & Clinical Trials, University of Exeter. Trial science development, randomised controlled trial, statistical analysis and report writing.

Linda Davies (https://orcid.org/0000-0001-8801-3559): Professor of Health Economics, University of Manchester. Original conception, trial science development, randomised controlled trial, health economic evaluation and report writing.

Rebecca Denyer (https://orcid.org/0000-0002-9984-2704): Research Assistant, University of Exeter. Data collection, randomised controlled trial data collection, process evaluation and report writing.

Julia Frost (https://orcid.org/0000-0002-3503-5911): Senior Lecturer Health Services Research, University of Exeter. Randomised controlled trial, lead qualitative analysis, process evaluation and report writing.

Linda Gask (https://orcid.org/0000-0002-7918-3967): Emerita Professor of Primary Care Psychiatry, University of Manchester. Original conception, programme design, intervention development and formative evaluation, intervention training lead, process evaluation and report writing.

Bliss Gibbons (https://orcid.org/0000-0002-1310-6838): Research Associate, University of Birmingham. Intervention development and formative evaluation, trial science development, randomised controlled trial and report writing.

John Gibson (https://orcid.org/0000-0003-1731-4869): Senior Service User Researcher, The McPin Foundation. PPI co-ordination, intervention development and formative evaluation, trial science development, randomised controlled trial data collection, process evaluation and report writing.

Laura Gill (https://orcid.org/0000-0001-6224-5693): Research Group Co-ordinator, University of Plymouth. Intervention development and formative evaluation, trial science development and report writing.

Ruth Gwernan-Jones (https://orcid.org/0000-0002-0223-6581): Complex Intervention Senior Research Fellow, University of Exeter. Programme design, intervention development and formative evaluation, process evaluation and report writing.

Joanne Hosking (https://orcid.org/0000-0001-7090-0205): Senior Research Fellow & BMBS Lead for Statistics & Numeracy, University of Plymouth. Trial science development, randomised controlled trial, statistical analysis and report writing.

Peter Huxley (https://orcid.org/0000-0001-5152-8030): Chair in Mental Health Research, Bangor University. Original conception, programme design, intervention development, trial science development, randomised controlled trial, process evaluation and report writing.

Alison Jeffery (https://orcid.org/0000-0001-7539-5957): Clinical Trials Manager, University of Plymouth. Trial management, randomised controlled trial, process evaluation and report writing.

Benjamin Jones (https://orcid.org/0000-0001-7374-1107): Postdoctoral Research Fellow in Medical Statistics, University of Exeter. Randomised controlled trial, statistical analysis, process evaluation and report writing.

Tom Keeley (https://orcid.org/0000-0003-1584-9093): Director, Patient Centred Outcomes (Value Evidence and Outcomes) at GSK. Core outcome set development, trial science development and report writing.

Richard Laugharne (https://orcid.org/0000-0001-9916-343X): Consultant Psychiatrist at Cornwall Partnership NHS Foundation Trust. Site lead, supervisor, randomised controlled trial and report writing.

Steven Marwaha (https://orcid.org/0000-0002-0303-9942): Clinical Professorial Fellow, Institute for Mental Health, University of Birmingham. Site lead, trial science development, randomised controlled trial and report writing.

Claire Planner (https://orcid.org/0000-0003-0938-0444): Research Associate, University of Manchester. Original conception, intervention development and formative evaluation, trial science development and report writing.

Tim Rawcliffe (https://orcid.org/0000-0001-7765-8052): Service User Researcher, Lancashire Care NHS Trust. Intervention development, trial science development and report writing.

Ameeta Retzer (https://orcid.org/0000-0002-4156-8386): Research Fellow, University of Birmingham. Intervention development and formative evaluation, trial science development and report writing.

Debra Richards (https://orcid.org/0000-0002-2368-6522): Research Assistant, University of Plymouth. Randomised controlled trial data collection, process evaluation and report writing.

Ruth Sayers: Senior Peer Researcher, The McPin Foundation. PPI lead, intervention development and formative evaluation, trial science development and report writing.

Lynsey Williams (https://orcid.org/0000-0002-2037-2093): Research Assistant, University of Plymouth. Trial science development, randomised controlled trial data collection, process evaluation and report writing.

Vanessa Pinfold (https://orcid.org/0000-0003-3007-8805): Research Director, The McPin Foundation. PPI lead, original conception, programme design and management, intervention development and formative evaluation, trial science development, randomised controlled trial, process evaluation and report writing.

Maximillian Birchwood (https://orcid.org/0000-0002-7476-0171): Professor of Youth Mental Health, University of Warwick. Co-chief investigator, original conception, programme design and management, intervention development and formative evaluation, trial science development, randomised controlled trial and report writing.

Disclosure of interests

Full disclosure of interests: Completed ICMJE forms for all authors, including all related interests, are available in the toolkit on the NIHR Journals Library report publication page at https://doi.org/10.3310/YAET7368.

Primary conflicts of interest: Richard Byng received additional support from the National Institute for Health and Care Research (NIHR) Applied Research Collaboration South West Peninsula. Maximillian Birchwood received additional support from the NIHR Applied Research Collaboration East Midlands and is partly supported by NIHR Applied Research Collaboration West Midlands. Richard Byng, Julia Frost, Vanessa Pinfold and Maximillian Birchwood received an NIHR Policy Research Programme grant to examine implementation of PARTNERS. Richard Byng is a member of the Health Technology Assessment (HTA) Prioritisation Committee A. Michael Clark was a member of HTA Funding Teleconference, NIHR CRRSU Funding Board, HTA Prioritisation Committee B Methods Group, HTA General Committee, and COVID-19 Reviewing. Siobhan Creanor was a member of Clinical Trials Units funded by NIHR.

Patient data statement

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 is important that there are safeguards to make sure that they are 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

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Ethics statement

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All procedures involving human subjects/patients were approved by the West Midlands-Edgbaston Research Ethics Committee (reference number 14/WM/0052). Local NHS approvals were obtained before the start of recruitment in each region from 27 February 2018. We had an independent Data Monitoring Committee and a Trial Steering Committee, as agreed with the funder. The trial is registered (ISRCTN95702682).

Information Governance Statement

University of Plymouth is committed to handling all personal information in line with the UK Data Protection Act (2018) and the General Data Protection Regulation (EU GDPR) 2016/679. Under Data Protection legislation (UoP) is the Data Processor; the many general practices and NHS Trusts involved in the study are the Data Controllers, and we process personal data in accordance with their instructions. You can find out more about how we handle personal data, including how to exercise your individual rights here: https://www.plymouth.ac.uk/students-and-family/governance/information-governance/information-security. The contact details for UoP’s Data Protection Officer is dpo@plymouth.ac.uk.

Department of Health and Social Care disclaimer

This publication presents independent research commissioned by the National Institute for Health and Care 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, MRC, NIHR Coordinating Centre, the PGfAR programme or the Department of Health and Social Care. This monograph was published based on current knowledge at the time and date of publication. NIHR is committed to being inclusive and will continually monitor best practice and guidance in relation to terminology and language to ensure that we remain relevant to our stakeholders.

Publications

Keeley T, Khan H, Pinfold V, Williamson P, Mathers J, Davies L, et al. Core outcome sets for use in effectiveness trials involving people with bipolar and schizophrenia in a community-based setting (PARTNERS2): study protocol for the development of two core outcome sets. Trials 2015;16:47. http://doi.org/10.1186/s13063-015-0553-0

Baker E, Gwernan-Jones R, Britten N, Cox M, McCabe C, Retzer A, et al. Refining a model of collaborative care for people with a diagnosis of bipolar, schizophrenia or other psychoses in England: a qualitative formative evaluation. BMC Psychiatry 2019;19:7. https://doi.org/10.1186/s12888-018-1997-z

Baker E, Gwernan-Jones R, Britten N, McCabe C, Gill L, Byng R, Gask L. Using interpersonal process recall to understand empowerment processes in a collaborative care intervention for people with a diagnosis of psychosis. Psychosis 2019;11:350–61. http://doi.org/10.1080/17522439.2019.1640274

Gwernan-Jones R, Britten N, Allard J, Baker E, Gill L, Lloyd H, et al. A worked example of initial theory-building: PARTNERS2 collaborative care for people who have experienced psychosis in England. Evaluation 2020;26:6–26. https://doi.org/10.1177%2F1356389019850199

Retzer A, Sayers R, Pinfold V, Gibson J, Keeley T, Taylor G, et al. Development of a core outcome set for use in community-based bipolar trials – a qualitative study and modified Delphi. PLOS ONE 2020;15:e0240518. http://doi.org/10.1371/journal.pone.0240518

The PARTNERS2 writing collective. Exploring patient and public involvement (PPI) and co-production approaches in mental health research: learning from the PARTNERS2 research programme. Res Involv Engagem 2020;6:56. https://doi.org/10.1186/s40900-020-00224-3

Plappert H, Hobson-Merrett C, Gibbons B, Baker E, Bevan S, Clark M, et al. Evaluation of a primary care-based collaborative care model (PARTNERS2) for people with diagnoses of schizophrenia, bipolar, or other psychoses: study protocol for a cluster randomised controlled trial. BJGP Open 2021;5:BJGPO.2021.0033. https://doi.org/10.3399/BJGPO.2021.0033

Reilly S, McCabe C, Marchevsky N, Green M, Davies L, Ives N, Plappert H, et al. Status of primary and secondary mental healthcare of people with severe mental illness: an epidemiological study from the UK PARTNERS2 programme. Br J Psychiatry 2021;7:e53. https://doi.org/10.1192/bjo.2021.10

Byng R, Creanor S, Jones B, Hosking J, Plappert H, Bevan S, et al. The effectiveness of a primary care-based collaborative care model to improve quality of life in people with severe mental illness: PARTNERS2 cluster randomised controlled trial. Br J Psychiatry 2023;222:246–56. https://doi.org/10.1192/bjp.2023.28

Disclaimers

This article presents independent research funded by the National Institute for Health and Care 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, CCF, PGfAR 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, the PGfAR programme or the Department of Health and Social Care.

Methods

The economic evaluation of the PARTNERS2 intervention was a trial-based analysis conducted as part of the RCT reported in Cluster randomised controlled trial in the main report.

Approach

The aim of the economic evaluation was to estimate the cost-effectiveness of the PARTNERS2 primary-care-based model of collaborative care (intervention) compared with usual practice (comparator) from the cost perspective of the NHS and social care (costs). The time horizon was the scheduled follow-up of 10 months (9 months for the final participants recruited) ± 1 month.

The measure of health benefit for the primary economic analysis was QALY estimated from the EQ-5D-5L and EQ-5D-5L Crosswalk Index Value Calculator as recommended by NICE. 44–46 QALYs were estimated as:

where U = utility and t = time at assessment. The time between assessments is the time from baseline to primary end-point follow-up (i = 1).

The economic analysis used patient-level service use data to estimate costs using published national unit costs47,48 for the year 2019–20. The primary outcome of the economic analysis was the ICER (which combines service use costs and health benefit):

The methods for cost and QALY estimation, handling of missing data and descriptive summaries, as well as the primary and sensitivity analyses, were defined and added to the statistical analysis plan for the trial prior to data transfer to the health economist.

Cost estimation

The PARTNERS2 intervention was developed to use existing resources by placing an experienced care partner from secondary care in the GP practice. Accordingly, it was assumed that the care partner represents a transfer of resources/costs from secondary to primary care rather than an additional cost of implementing the intervention. Additionally, it was expected that the level of contacts with the care partner would be captured in the primary care service use data collection and associated costs.

Participant-reported service use at follow-up was collected for a 3-month, rather than 10-month, recall period. This was because of concerns about the burden to participants of recalling service use over a longer period and the need to balance complete service use data with incomplete recall, potentially large numbers of inconsistent or missing data, and limited resources for data collection. Additional data collection from an audit of case notes in secondary mental health and GP practices was planned to cover the period from baseline to the end of scheduled follow-up.

The collection of service use data from case notes across the multiple care providers involved in providing services (primary, secondary and community health care and social care) is complex, highly resource intensive and costly in the absence of integrated record linkage for individual service users. Accordingly, the case note audit focused on (1) key high-cost psychiatric secondary and crisis care services that could be expected to differ if the intervention was effective but may not be used within the 3-month recall period at the end of scheduled follow-up and (2) GP, practice nurse and other GP practice consultations (mental and physical health). However, the latter was not feasible within the trial budget and the impact of COVID-19-related constraints on access and researcher time. Data were collected from a sample of 31 participants. A comparison of the cost per day for these participants showed no differences between the data collected from the primary care audit and those reported by participants.

For the primary analysis, regression was used to predict a cost per day for all service use for participants with complete 3-month service use data collected at the follow-up interview, for the pooled (masked to treatment allocation) data. A generalised linear model (gamma log) was used, adjusting for baseline covariates (categories of services used, mental health medications, MANSA score, locality, practice size and GP practice). The full cost from baseline to end of follow-up was passively estimated following multiple imputation of missing cost data (predicted participant-reported cost per day plus the cost per day of psychiatric secondary care, multiplied by the number of days’ follow-up from baseline).

Missing data

The cost and health benefit data were analysed by treatment allocated and include data for all participants whether or not they completed planned care. However, missing data are inevitable from loss to follow-up or missing observations. Single imputation was used for missing baseline measures of cost, utility and clinical indicators and indicators for missing demographic data. 49

Multiple imputation (predictive mean matching and sequential chained equations)50 from available data was used to generate 10 sets of estimates for costs and QALYs for the pooled data set. A literature review and a regression analysis of pooled baseline data (masked to treatment allocation) were used to identify key baseline and follow-up variables associated with costs and QALYs to include in the imputation models.

Primary analysis

The EQ-5D-5L and ICECAP-A data were summarised as proportions of participants reporting each level (n/N, %, 95% CI). Costs and summary health benefit measures were summarised descriptively (mean, standard deviation, 95% CI) for the intervention and comparator for the complete case and imputed data at baseline and the primary end point.

Regression analysis was used to estimate the net costs and QALYs of the intervention, adjusting for key covariates. These were identified from published literature and supplemented with an analysis of pooled (masked) baseline data. The trial cluster (GP practice) and stratification variables (locality and practice size) were included. The regression-based estimates of costs and outcomes were bootstrapped to replicate 10,000 pairs of net cost and QALY outcomes of the intervention. These were used to generate estimates of the net costs and QALYs for the PARTNERS2 intervention, the ICER, and the probability that the intervention is cost-effective for the primary and sensitivity analyses, as well as the cost-effectiveness plane and cost-effectiveness acceptability curve for the primary analysis. ICERs estimate the net cost per QALY gained by an intervention and raise the question of whether that cost is worth paying. To address this, the ICERs are compared with how much decision-makers may be willing to pay for an additional QALY. However, the UK has no universally agreed cost-effectiveness threshold. Reflecting this lack of consensus, the monetary value of simulated QALYs were estimated across the range of £0 to £30,000 WTPTs. To estimate the likelihood that the intervention is cost-effective for the primary analysis, a WTPT of £15,000 (the mid-point of the £0 to £30,000 range) was used.

Prespecified sensitivity analyses explored whether alternative measures or analyses could change the conclusions of the economic analysis. These included (1) using the ICECAP-A41 as an alternative measure to derive a QALY; (2) alternative methods of estimating 10-month costs from the 3-month data; and (3) complete-case analysis.

Additional exploratory analyses (not included in the statistical analysis plan) used the primary clinical outcome (change in MANSA score from baseline to follow-up) and hours of paid employment per week as alternative measures of health benefit with which to estimate an ICER. As these are not preference-based measures of health benefit, there is no defined range of willingness-to-pay values to gain an additional unit of outcome. Accordingly, the probability of cost-effectiveness was not estimated.

Stata SE version 16 (StataCorp LP, College Station, TX, USA) was used for data management, costing and all analyses.

Results

The demographic data are shown in the paper reporting the detailed clinical trial results. Tables 5–8 report unadjusted EQ-5D-5L, ICECAP-A, QALYs and costs as well as number of participants contributing to the estimates, using the available data. Tables 1 and 2 report the frequency of participants with no problems on the measures used to estimate QALYs for the primary (EQ-5D-5L) and sensitivity analysis (ICECAP-A). Tables 5 and 6 also report the average QALYs.

Tables 7 and 8 report the observed and predicted average participant-reported costs for the 3 months prior to the baseline and follow-up assessments that were used to estimate full costs for the primary and sensitivity analyses. The average mental health secondary care costs from the case note audit for the index mental health service are also included in Table 7. Table 9 reports the multiple imputed costs and QALYs for the full follow-up period. Overall, the 95% CIs for all the data are wide and overlap between the usual care and PARTNERS2 groups, indicating uncertainty in the estimates.

Table 10 reports the bootstrapped net costs and QALYs and the probability that the PARTNERS2 intervention is cost-effective. Overall, the 95% CIs for the net costs and QALYs are wide and cross zero, indicating a high level of variance and uncertainty. Figure 5 illustrates the wide distribution of net cost/QALY pairs over the four quadrants for the primary analysis.

FIGURE 5.

Distribution of net cost–QALY pairs on a cost-effectiveness plane: primary analysis.

At the prespecified WTPT of £15,000 to gain one additional QALY, the probability that the PARTNERS2 intervention is cost-effective is < 50% for the primary and all sensitivity analyses. Figure 6 shows that the probability PARTNERS2 is cost-effective changes as the willingness to pay to gain one QALY increases. If decision-makers are prepared to pay £4000 to gain one additional QALY, then the probability that PARTNERS2 is cost-effective increases to just over 50% for the primary analysis, but not for the sensitivity analyses.

FIGURE 6.

Cost-effectiveness acceptability curve: primary analysis.

The bootstrapped net change in MANSA score (from baseline to follow-up) for participants in the intervention group was 0.01 (standard error 0.13, 95% CI −0.24 to 0.27). The bootstrapped net hours of paid work per week for participants in the intervention group showed a reduction of −1.52 hours per week (standard error 1.05, 95% CI −3.57 to 0.53).

PARTNERS2 service user fidelity questionnaire

PARTNERS2 service user fidelity questionnaire version 1.0, 8 January 2020

Participant ID             Date _______________

Your experiences working with your PARTNERS2 care partner

In this form we ask you some questions about what your care partner …… did when they were working with you.

Whatever you tell us is confidential; we will not pass this onto them. We know that not all of the things mentioned will have been done for everyone. We would like you to be honest about your PARTNERS2 experience so that we can improve the service for people in the future.

In your own words please can you tell us about your experience of working with your care partner ……, what has been helpful or not? ………………………………

Results of responses for those receiving the intervention in the PARTNERS2 trial

Contact sheets to document extent of delivery

Contact sheets to document extent of delivery of the PARTNERS intervention were developed during pilot work and adapted for use in the trial. The care partners were trained and supported to complete the contact sheets during the PARTNERS2 trial. These were also included in the PARTNERS manual version 1.8.

Please complete the table below for all contacts with the service user (please note: not all rows need to be completed at every contact).

Communication/liaison about this service user

Please complete after any other service user-related communication (except supervision – which will be recorded on a separate sheet)