Preparing for Mental Health Act reform: Pilot study of co-produced implementation strategies for Advance Choice Documents

Background Advance Decision Making (ADM) is strongly supported by stakeholders but implementation remains challenging. In England and Wales, implementation strategies are urgently required to prepare for the introduction of mental health ‘Advance Choice Documents’ (ACDs) as part of Mental Health Act reforms. We report on a pilot project which aimed to co-produce and evaluate implementation strategies for ACDs with those who experience fluctuating mental capacity in the context of bipolar. Methods A co-produced prototype ACD template was piloted in ‘Plan, Do, Study Act’ (PDSA) cycles. Implementation strategies were co-produced with participants and mapped onto the Expert Recommendations for Implementing Change (ERIC) framework. Strategies were evaluated during thematically analysed qualitative interviews. Results We piloted the template with 17 service users during 5 successive PDSA cycles and conducted 75 in depth interviews with stakeholders. Key strategies identified as accessible, appropriate and feasible were: interactive assistance from an independent ‘supporter’, a structured template and active offers of involvement to service users and informal carers. Conclusions Mental health professionals and organisations must prepare for increased expectations around mental health ADM. We recommend further pilot projects and the establishment of ‘ACD workshops’. Resource is essential to fund independent ‘supporters’, training, network building and embedding ADM in clinical pathways.


Introduction
Mental health advance decision making (ADM) is about supporting service users who live with severe mental illness (SMI) to make use of their expertise and state preferences for treatment in any future mental health crises. ADM is particularly salient for those with fluctuating decision-making capacity for treatment (DMC-T) and are at risk of experiencing compulsory treatment. A paradigm example is the experience of people who live with Bipolar Affective Disorder (bipolar). When well, individuals are very able to engage in decisions about their treatment. However, during relapses, particularly manic phases, DMC-T may be rapidly lost then re-gained during the recovery period (Owen et al., 2008). Research suggests that there is both an ethical and practical imperative for statutory ADM (Owen et al., 2019). Ethical reasons include respect for service user interest (Hindley et al., 2019;Swanson et al., 2006a), autonomy and reducing inequalities in decision making (Council of Europe: Enabling citizens to plan for incapacity) . Practical reasons include potential therapeutic benefit (Stephenson et al., 2020b) and reduction in the need for compulsory treatment (de Jong et al., 2016;Molyneaux et al., 2019). Multiple jurisdictions (including USA, India, Australia, Netherlands and Scotland) have now introduced tailored, statutory support for ADM. However, in jurisdictions where outcomes have been monitored uptake of mental health ADM has remained low (Hindley et al., 2019;Swanson et al., 2006a Mental Welfare Commission for Scotland: Advance statements in Scotland).
Research in this field has already identified multi-level barriers to implementation which need to be overcome. Shields et al. (Shields et al., 2014) conducted a systematic review of barriers and results included: concerns about legal provisions and liability, resource implications, lack of knowledge amongst health professionals and service users, concerns about clinical feasibility including mental capacity, lack of support for service users and concerns that documents would not be accessed or applied. The scale of these barriers are unsurprising given that according to Medical Research Council guidance ADM can be understood as a complex intervention; it requires multiple components and various behaviours at several points in time from multiple groups (Skivington et al., 2021). Some implementation strategies have been suggested around engaging and supporting stakeholders and addressing practical issues (Stephenson et al., 2020b;Zelle et al., 2015). But so far, these strategies have not been co-produced with stakeholders, piloted or held within a comprehensive implementation science framework which would facilitate wider evaluation and use.
In England, the setting for this project, grappling with ADM implementation is an urgent issue as government has committed to introducing ADM documents in the form of 'Advance Choice Documents' (ACDs) (Reforming the Mental Health Act: Department of Health and Social Care). At present the only legal provision for ADM is under the Mental Capacity Act 2005 (MCA) for 'Advance Statements' and 'Advance Decisions to Refuse Treatment'. These provisions do allow service users to state treatment requests and refusals for mental health treatments, but they do not have formal legal status if the person is detained under the Mental Health Act 1983 (MHA). In contrast, according to a recent White Paper (Reforming the Mental Health Act Government Response) ACDs would retain some legal weight if the person were detained under the MHA. These reforms are likely to impact a significant proportion of service users given that in 2019-2020 alone 50,893 new detentions under the MHA were recorded (Mental Health Act statistics annual figures 2019-2020). However, in order to ensure that these reforms achieve their intended goals anticipatory research focussed on implementation is urgently required.
Therefore the aims of this study were to: 1. Co-produce, describe and pilot implementation strategies for ACDs with key stakeholders 2. Evaluate the acceptability, adoption, appropriateness and feasibility of these implementation strategies 3. Recommend implementation strategies for future pilots and wider roll out

Study design
Implementation experts draw an important distinction between an intervention, 'the thing', and implementation strategies, 'the stuff that helps you do the thing' (Curran, 2020). For clarity, in this study the intervention was formal mental health ADM which relies on the existing legal framework of the Mental Capacity Act 2005 (MCA) provision for Advance Statements and Advance Decisions to Refuse Treatment. We could not use actual ACDs as the intervention as the reforms have not yet been passed but shaped the template to mirror the shape of anticipated reforms as closely as possible. Participants were made aware of the current limitations of the legal framework. As the intervention has already been established the key outstanding questions for service users, their family members, clinicians and mental health organisations are around which implementation strategies could help utilise ACDs.
To address our first aim service user participants along with a supporting family member/friend and a treating health care professional were invited to take part in creating a prototype ACD using a collaborative template called a 'Crisis PACk' (available in a data repository). This template was based on a document co-produced with stakeholders during a previous study (Stephenson et al., 2020b) and adapted to the needs of the local setting. Of note, the template also offers service users the option to 'self-bind' i.e. to request compulsory treatment in advance even if they anticipate that when they are unwell and the treatment is needed they are likely to refuse it. This option was included in response to evidence that it may have particular salience for people with bipolar Gergel & Owen, 2015;Stephenson et al., 2020a) Participants created their documents in successive 'Plan, Do, Study Act' cycles according to 'Model for Improvement' methodology (NHS Online library of Quality Service Improvement and Redesign tools; Langley et al., 2009). This methodology was selected as it is supported by the site of implementation and allows for rapid feedback and involvement from stakeholders. Throughout the process stakeholders were asked to generate ideas for implementation strategies, including developing the template itself. This information could then be applied for further refinement and feedback in the next PDSA cycle. Informal feedback was collected from all participants throughout the process with prompts in all contacts e.g. administrative phone calls and emails with an open invitation for participants to email the project lead.
To address our second aim, in depth qualitative interviews were conducted with participants from all stakeholder groups before and after they had completed their documents to explore their experiences. Interviews were undertaken by a clinical researcher (L.A.S, female, white) who has had over 10 years clinical experience working in south east London and training in psychiatry and psychotherapy. The participants were not known to the researcher prior to the project but relationships were established with participants over the study course during repeat interviews and contacts. The participant information sheet contained information about the researcher's roles and aims for the project. Throughout the project topic guide domains were updated to reflect new implementation strategies and ask for feedback on them. A measurement plan was created to govern the collection of quantitative data about the process of making the document.
The refined implementation strategies were described according to criteria outlined by Proctor et al. (Proctor et al., 2011) and mapped onto the Expert Recommendations for Implementing Change (ERIC) framework .

Participants and procedures
The project took place between January 2020 and November 2021. We targeted English speaking service users with a primary diagnosis of bipolar documented by a health professional. Records were reviewed by a clinical researcher (L.A.S) to ensure that the diagnosis was consistent with their recorded presentations. Participants had to have experienced at least 1 detention under the MHA, be interested in making an ADM document and have capacity to make an ADM document. Participants were all in active contact with secondary mental health services in South London and Maudsley NHS Foundation Trust (SLaM) for treatment of bipolar. SLaM is a busy, large, urban mental health trust offering a range of services to a diverse population in south-east London. Participants were recruited from a variety of settings across the trust: inpatient wards, Home Treatment Teams and Community Mental Health Teams. Participants self-referred to the project or a treating clinician suggested it. Once participants had been referred to the project they were contacted by phone or email to discuss what the project would involve and sent written materials with further details.
Once the service user participant had been recruited, we asked permission to interview a health professional and a carer/family member/friend/advocate who would be involved in making their document. We set out to recruit 12 service user participants, 12 family/friend participants and 12 health professional participants to make ADM documents and participate in the qualitative interviews. This sample size was chosen as sufficient to achieve qualitative data saturation whilst being manageable.

Data collection.
Interviews were held online/by telephone (depending on participant preference and access to technology) due to pandemic restrictions. Participants were informed interviews would last approximately 1 hour but length was guided by the participants and according to their level of comfort and time restrictions (e.g. many clinicians were time limited due to providing front line services during pandemic). Interviews were semi-structured using a pre-prepared topic guide. Question domains were informed by a previous survey (Hindley et al., 2019) and focus group study (Stephenson et al., 2020b). The topic guide was discussed amongst the research group and Service User Advisory Group. It was adapted throughout the project to ensure the content was driven by participant experience and to probe emerging themes. Interviews were conducted with all participants before and after a document had been made. Arrangements were made to follow participants up and conduct a final interview after any crisis and/or at the end of the study period. The results of the final time point interviews will be reported in a forth coming publication. Following the interview participants were advised that they could contact the researcher at any point if they wished to make additional points or change their expressed views. There was a team consensus that data saturation was achieved within the sample of interviews.

Consent
The process of consent was impacted by the pandemic which prevented face-to-face contact. All participants in the qualitative interview study were sent the participant information sheet and consent form in advance by email or post according to their preference. A remote contact was then arranged by a clinical researcher to read through the information sheet and consent form point by point. Where handwritten consent was not possible participants were asked to confirm consent via email from their personal account. Consent was also confirmed verbally before and after each qualitative interview. For those participants who did not take part in the qualitative interview research study but still wished to make an ADM document written consent was not required as this part of the project fell under the governance of the mental health trust Quality Improvement team. This did not require formal Research Ethics Committee approval as ADM is already supported in the MCA. It is not a novel intervention and should already be part of standard clinical care for bipolar as recommended in NICE guidelines (National Institute for Health and Care Excellence, 2014). These participants were given written and verbal information about the project and their consent to take part confirmed verbally by a member of the project team and/or their treating clinician at each contact.

Ethics
Ethical approval for the qualitative interview research study was granted by Camberwell and St Giles ethics committee (REC reference 19/LO/1142). Initial approval was granted on 03.09.2019 with an amendment approved on 04.08.2020 to allow the research to continue under pandemic restrictions. The part of the project which involved the Quality Improvement process (i.e. making the ADM document) was approved and supported by the SLAM Quality Improvement team. This process included project review, training and support from a Quality Improvement mentor.

Analysis
Interviews were audio recorded, transcribed and uploaded to a data management programme (NVivo Release 1.5 (4577). Data was thematically analysed following Braun and Clark (Braun & Clarke, 2006) during a process which included regular, reflective interdisciplinary team meetings. The team included experts in psychiatry (GO, LR, L.A.S), philosophy (TG), lived experience (TG) and law (ARK). This diverse interdisciplinary environment supported a reflective, rounded process of data analysis. One researcher (L.A.S) read all the transcripts with two other researchers (GO and TG) read a sample. Initial ideas were noted and then discussed to generate an initial coding tree. L.A.S then coded all transcripts with GO and TG coding a sample, discrepancies were discussed and a final coding tree agreed. Proposals for emerging themes were discussed at team meetings then checked against the data for coherence. The overarching thematic analysis of participants experience and methodology will be presented in a forthcoming publication. In the present publication we report on specific feedback around implementation strategies. Standard descriptive statistics were used to summarise and analyse quantitative measures.

Characteristics
Overall, 36 service user participants were referred to the project. Of this group 23 (64%) accepted the initial offer of making a document and in total 17 people (47%) drafted documents. Reasons for not moving onto make a document are outlined in Table 1. Of the 17 service users making documents 14 were eligible and wished to take part in the qualitative interview study, this involved completing a demographics questionnaire. The demographics of these participants are outlined in Table 2-Table 4. Of those who drafted documents but did not Table 1. Reasons for declining to make a Crisis Pack.

Reason n
No longer in secondary care services 1 Unable to contact 7 Team members not supportive 3 Too unwell 3 Other commitments 1 Concerns stress of process may trigger relapse 2 Reason not stated 2  Prefer not to say 2 complete them 1 was not completed due to pandemic interruption, 1 required more time due to a health professional absence, 1 preferred to not include health professionals in the process and 1 required more time as the service user experienced relapse. No participants expressed a wish to drop out of the process completely.

Outcomes
The outcomes and process measures of the intervention strategies employed to support all participants in making their ADM document are detailed in Table 5.
Co-producing and evaluating implementation strategies Implementation strategies were generated and refined throughout the study period in 5 successive PDSA cycles. These PDSA cycles are summarised in Table 6, and the final implementation strategies are summarised in Table 7. This table contains the strategies used by the research team to facilitate organisational engagement as well as co-produced strategies generated and evaluated by project participants. The table maps the strategies co-produced by the project participants and the interdisciplinary research team onto the ERIC  arranged into concept mapped clusters for easier references (Waltz et al., 2015). Below, the refined implementation strategies are described according to their ERIC cluster.
Evaluative and iterative strategies. Informal feedback and solution ideas were collected from the participants and collated in field notes throughout the project. At the end of each cycle a meeting was held with a working group to review progress, feedback and discuss any necessary adaptations. Evaluation took place throughout the process and formally in final in depth interviews.

Provide interactive assistance.
Making mental health ADM is a relational, bespoke activity that required time and investment from all parties. Health professionals reflected that this was different from 'business as usual' and should not be simply rolled into normal processes e.g. Care Planning Approach (CPA) meetings. Some expressed anxiety about the process of completing a Crisis PACk as it was unfamiliar -particularly navigating the legal meaning of the document. The role of a 'supporter' was clarified during project. A service user advisory group (SUAG) were consulted about the appropriate language and stance for someone in this role. They advised that language of support was preferred to reflect the power dynamics and ensure that the process respected the primacy of service user ownership and expertise. Participants suggested supporters could be any mental health professional, a peer supporter or advocate with suitable training but expressed concerns that a person in this role would require commitment, advanced listening, and interpersonal skills to ensure the service user did feel truly supported while engaged in a potentially triggering task. There was a preference, amongst all participants, that, although they highly valued involvement from their treating team, the supporter should be someone independent i.e. not the service user's care coordinator/usual psychiatrist, to manage the power differentials involved.
The supporter provided a 1:1 drafting session with the service user prior to meeting with health professionals which focussed on eliciting the service user preferences and navigating the administrative aspects of completing the document. This drafting support session was seen as having an important role in managing emotional distress from previous traumatic experiences in mental health services, overcoming ambivalence and navigating power imbalance. Service users reported that it meant they felt prepared for meeting with a health professional as they were clear in their own minds about what they wanted. If wished the supporter would chair meetings with treating health professionals/family members and disseminate the document. The boundary for the supporter role was around providing any clinical advice on the content of the document. For health professionals the supporter would provide materials in advance of meeting and offer a coaching session. Most often this took the form of an email with a sample agenda and a brief video call in advance of meeting with the service user. At baseline several professionals expressed concern about the resource and clinical time that making a Crisis PACk would require. However, after being part of the process most felt it was a valuable exercise and worth the upfront investment in the collaborative effort to make the document.

Adapt and tailor to context.
A key adaptation occurred at the start of the pilot project in response to the pandemic and the new urgency to implement ADM this generated within the mental health trust. We had planned to pilot the PACT template which was co-produced as part of a previous research project (Stephenson et al., 2020b). This had a relatively specialist focus around people with a diagnosis of bipolar who wished to self-bind to treatment. However, the mental health trust advised that they wished to design an ADM template in response to anticipated MHA reform, to reduce inpatient bed use and address concerns about physical health ADM if a service user were to contract covid whilst in hospital. This template would need to have a wider reach than the PACT template. To facilitate this adaptation a working group was formed comprising senior SLAM clinicians and IT specialists plus members of the research team which comprised clinical academics (LS, GO, LR) working within the trust and a lawyer (ARK). The working group also had links with a service user advisory group and experts with lived experience within the Recovery College. The PACT was adapted to the 'Crisis PACk'. This group met after every PDSA cycle completed, in total 4 times over 12 months. The template was reviewed and updated with a synthesis of participant and working group suggestions.

Develop stakeholder interrelationships.
During the project it became clear that a clinical champion role was needed to act as a coordinator for ADM and point of contact for people interested in making a document. As interest in ADM increased the advisory and networking side of this role grew to incorporate consultation for other interested academic researchers and clinical bodies. Coalitions were built with: relevant health trust committees (e.g. Trust executive board, Mental Health Law committee, older adult ADM group), third sector organisations (e.g. Recovery College, Bipolar UK), professional bodies (e.g. Royal College of Psychiatrists), governmental bodies (e.g. Department of Health and Social Care), and other research groups (e.g. Adstac group exploring ADM with people from black backgrounds).
Train and educate stakeholders. The Crisis PACk template and guidance was co-produced within the interdisciplinary research team, service user advisors and participants. The guidance was distributed to participants and members of frontline teams that were visited to deliver educational settings. Teams that worked with service users likely to benefit from making Crisis PACks were contacted and offered a training session at a time of their convenience in person or online. This was well received when packaged as a training session about current and future legal provision for ADM.
Key learning from participants was around simplifying the template and guidance. One set of guidance was produced for service users and another for professionals and it was located within the template inside the relevant box for ease of use. The process of making the document was also enormously simplified. The initial guidance contained 7 steps in a narrative format for making the document. After receiving participant feedback about this and asking for solutions this was altered to a visual aid (available on Crisis PACk template within extended data with 3 steps: draft, discuss, disseminate).
Although the materials and training were appreciated, we found that participants valued in person and tailored advice far more highly than referring to guidance documents. The general thoughts around the template were that it was clear, fit for purpose and not a barrier but that support was still required to complete it. The education and training sessions had an important role in awareness raising but we found that the 'on the spot' coaching was far more effective in increasing people's confidence to complete ADM documents. It seemed that, if participants had educational materials alone, they would be unlikely to make the ADM documents.

Engage stakeholders in ADM process: service users, family/friends, health professional.
Service users reported finding the collaborative process of making the document to be a therapeutic activity. Service users were given a choice about who to involve. Usually they chose one family member/friend and one health professional from their treating team. The supporter offered to coordinate meetings with the people the service user wished to include. Every effort was made to overcome barriers to engagement including meeting times and mediums (e.g. online/phone etc). Service users highly valued the opportunity to have health professionals and family members/friends involved in making their Crisis PACk. Family member/friend participants expressed views that it was also beneficial for them to be involved in the process to reduce the stress of future crises and empower them to help manage these more confidently and in line with service user wishes.
Participants emphasised the importance of an active, and repeated, offer of support with ADM made at the right time for them. It would not be sufficient to simply wait for service users to ask to make a document. Learning from the project suggested the following points in the clinical pathway would be appropriate to make offers of engaging with ADM: initial signpost on discharge from an inpatient admission, initial signpost during Home Treatment Team/community crisis care, active offer on intake to CMHT, active offer at point of discharge from CMHT as a tool to summarise learning from spell with team. Participants were unified in thinking that although everyone with experience of detention should be offered the opportunity to make an ADM document only people who have accepted their diagnosis and who have had previous experience of managing a crisis might feel ready to take on this task.
An important topic emerging in the interviews was the experience of service users from Black backgrounds within mental health systems. Multiple service users, family/friends and professional participants identified this group as more likely to experience discrimination, micro-aggressions and trauma during mental health treatment. They felt that this made it even more important for service users from this group to make ADM documents yet they faced additional barriers in the form of the increased emotional load required to make the document as well as lower trust in services. When asked to generate potential solutions to these difficulties participants again emphasised the importance of active offers of culturally sensitive support with ADM to all (to avoid risk of selecting those from demographics who might be assumed as more likely to engage with ADM) and suggested community engagement e.g. liaising with religious groups and third sector organisations to raise awareness and increase acceptability.
Concerns/unintended consequences. Participants used the interviews to express their views about the importance of implementation efforts for ADM and the potential unintended consequences if ACD implementation were not taken seriously or appropriately resourced. Key concerns were that it could become a 'tick box' exercise, that was rushed and that future professionals would not access or take documents seriously. Participants felt that the consequences of this could be reduced trust and disengagement. Service users were most commonly offered the opportunity to make a document by CMHTs rather than inpatient settings or crisis teams. Our experience on the project was that people were usually still too unwell to make Crisis PACks whilst on an inpatient ward. Health professionals most commonly identified the need for making an ADM document rather than service users themselves or family members/friends. The extra time required for making the document was on average a 38 minute session to draft the document (provided by a supporter) and a 78 minute session (including service user, treating health professional, family member/friend and supporter) to discuss and complete the document. Both of these contacts could be completed satisfactorily remotely during dedicated online meetings.
Key strategies adopted by participants and identified as accessible, appropriate and feasible were: providing interactive assistance in the form of an independent, skilled 'supporter', training and education for stakeholders including provision of a structured ADM template and guidance and engagement of stakeholders with active offers of involvement to service users and family members/friends.

International context
These findings build on a developing body of international literature which has confirmed an aspiration/implementation gap for mental health ADM and outlined barriers to implementation. Hypothesised implementation strategies to address these barriers have included: facilitation of ADM documents (Ruchlewska et al., 2014;Swanson et al., 2006b), appointing clinical champions (Zelle et al., 2015), stakeholder outreach meetings (Lenagh-Glue et al., 2021), training and IT support (Stephenson et al., 2020b;Zelle et al., 2015). This study adds to this literature in that it: offers a deeper exploration of the reasons why support is so crucial for service users in managing trauma and power differentials, addresses key issues in producing materials and processes for ADM, generates and evaluates multiple co-produced implementation strategies, clarifies stakeholders and organises these co-produced implementation strategies within a recognised framework ready for future studies.
The findings of this study are consistent with the clearest message from the literature so far: that service users require support in creating ADM documents (Swanson et al., 2006b) (Ruchlewska et al., 2009). In the international literature on ADM several models which are comparable to the Crisis PACk prototype and utilise support have been developed. In the US facilitated ADM was successfully trialled. This model involved offering in person support to write an ADM document and go through all the official procedures to register. Facilitators were research assistants trained by a psychologist and facilitation sessions lasted 120 minutes. Participants who received this support were significantly more likely to go on to complete an ADM document (Swanson et al., 2006b).
In Europe and the UK the Joint Crisis Plan (JCP) has been trialled. This involves two sessions; an independent facilitator meets with the service user and their care coordinator to prepare for a second planning meeting with a psychiatrist and family member (if desired). This is an informal document with no ability to self-bind. The multi centre UK trial did not show any significant benefit: the study authors concluded this was largely due to lack of clinical buy in and merging the process for creating a JCP with a Care Programme Approach (CPA) meeting which serves more professional purposes (Thornicroft et al., 2013). The European trial explored different options for facilitators and concluded that a trained peer supporter could achieve better results than clinicians in support for drafting ADM documents (Ruchlewska et al., 2014). A recent randomised controlled trial in France also concluded that ADM documents completed by service users with SMI supported by peer workers resulted in significantly fewer compulsory admissions (Tinland et al., 2022). Interestingly, the New Zealand experience suggested that more service users were able to complete their document with a clinician, the key factor being trust rather than nature of the facilitator's previous experience (Lenagh-Glue et al., 2021).
Looking at the Crisis PACk model in this context it is most similar to the Joint Crisis Plan in that it involves a supported two-step process. A distinguishing characteristic of the Crisis PACk model is that the emphasis is on the document ownership belonging primarily to the service user. The initial drafting session is 1:1 with the supporter (without a care coordinator). This offers the maximum chance that the service user can freely express their opinions and receive emotional support and preparation for meeting with a health professional. The 'discussion' stage similarly involves a healthcare professional and a family member which can be facilitated by the supporter who, significantly, is not a member of the treating team. Also, the Crisis Pack is a formal document making explicit use of the available legal framework and it supports self-binding. A unique part of the process of making a Crisis PACk compared to other similar models is the use of technology. This was necessary during the pandemic and this pilot has demonstrated the feasibility of completing these documents remotely which may increase time efficiency and accessibility for many service users. However, caution should be taken to ensure that non digital approaches remain viable for those without tech access.

Limitations
A key limitation impacting generalisability is that we had a relatively small sample size and included only those with bipolar. However, this project relied on in-depth qualitative feedback around sensitive topics. All had experience of detention and compulsory treatment under MHA and reported traumatic experiences in this context. This suggests the sample is a 'hard to reach' population and representative of those whom ACDs are aimed at. Working consistently with the participants over time allowed the researcher to build relationships with them and supported them to be candid in their feedback. As such the sample size was appropriate and lays a foundation for later work scaling up these strategies. We focussed on service users with bipolar as the most straightforward case of fluctuating capacity to pilot the concept. Future work will be required with service users who have more diverse diagnoses. This paper focusses only on the process of making ADM documents rather than accessing them and applying them in crisis, a future publication will detail the outcomes experienced by participants during the follow up period. A key issue is likely to be around accessibility and IT infrastructure to facilitate this which we did not tackle in this stage of the project.

Recommendations for further research and ACD implementation
On 10/05/2022 the Queens Speech announced (for the second time) the intention that draft legislation to reform the MHA in England and Wales will be published. In the White Paper which preceded it, a commitment to introducing ACDs as part of these reforms had been made. We do not at the time of writing have the detail of precisely what shape any statutory reform to implement ACDs will take, but the key conclusion of our research is that, whatever shape it takes, NHS mental health trusts should start preparing for ACDs now. Without these preparations it is unlikely that ACDs will achieve their potential to increase service user autonomy, improve outcomes and reduce coercion. The preparations we recommend are set below.
Extending pilot projects. Pilots in other trusts are urgently required so that the implementation strategies proposed here can be scaled up and adapted. We suggest one model to pilot for wider roll out would be an ACD 'workshop'. This could comprise a multi-disciplinary team made up of a psychiatrist, systemic psychologist, and ACD supporters. This workshop could offer a hub of expertise around facilitating collaborative ACDs, trouble shooting, training and outreach to community groups and third sector organisations.

Resource for supporters.
Resources should be made available for independent 'ACD supporters' to be trained and work with a range of service users and clinical teams in each health trust. Supporters could come from diverse backgrounds e.g. professional/lived/advocacy/chaplaincy expertise/third sector.

Resource for training and education for stakeholders.
Key professional groups (e.g. psychiatrists/social workers/care coordinators/psychologists/mental health nurses) should have mandatory training on making and using ACDs. This should include training around supporting service users to overcome common barriers (e.g. cultural sensitivity, digital poverty, literacy issues, trauma) and managing concerns around mental capacity assessment when making ACDs and applying in crisis.
Resource for engagement/outreach. Health trusts should embed making active offers for support making ACDs at the following points within standard care pathways: initial signpost on discharge from an inpatient admission, initial signpost during Home Treatment Team/community crisis care, active offer on intake to CMHT, active offer at point of discharge from CMHT. These offers should include outreach to family members/friends as per service user preference.
Building stakeholder interrelationships. ACD champions should be appointed to collate expertise and facilitate intra-and inter-agency learning in organisations involved in responding to mental health crises. These organisations include: professional bodies for psychiatrists, nursing, social workers, psychologists, paramedics, emergency physicians and police plus mental health trusts and third sector organisations. Champions should be encouraged to form a national network to build expertise and establish best practice as ACDs are rolled out.
Evaluative and iterative strategies. ACD uptake and use must be monitored to identify inequalities in take up and areas where more support is required. We recommend the following key areas to measure: • Number ACDs offered • Number of ACDs created • Demographics of those creating ACDs (matched to local population) • Number of ACDs accessed in crisis • Number of ACDs overridden

Data availability
Underlying data Due to risk of de-anonymisation, the raw underlying qualitative data has not been made freely available. If researchers or referees would like access to the data for re-analysis, please contact the corresponding author (L.A.S.) by email at lucy.a.stephenson@kcl.ac.uk. Applicants who wish to access the data will need to demonstrate that they have a position at a recognised academic institution and secure data storage facilities. Partial data will be made available in response to specific research questions. In accordance with our study protocol data sharing will be proportionate, task-orientated and will occur subject to strict understandings about confidentiality. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Colin McKay
Centre for Mental Health and Capacity Law, Edinburgh Napier University, Edinburgh, UK The article discusses the use of improvement methodology to improve the uptake and quality of Advance Choice Documents in relation to mental health care for people with bipolar disorder. It is clearly set out with tables showing how the documents and the process for making them evolved through the improvement process. The study design is appropriate to support the practical, real world implementation of an agreed policy goal. It builds on existing studies and can be used to inform, as recommended, future pilot work in preparation for reforms to the Mental Health Act in England and Wales. The source data is only available on request because of anonymisation, but this should not affect the value of the research. This is a useful and timely piece of research. As well as MHA reform in England and Wales, the topic has wider international relevance given the imperative in Article 12 of the UN Convention on the Rights of Persons with Disabilities for 'support in exercising legal capacity', which the UN Committee has confirmed includes Advance Statements (General Comment Number 1): 'All persons with disabilities have the right to engage in advance planning and should be given the opportunity to do so on an equal basis with others.' In the English context, the usefulness of this work is enhanced by the fact that the UK Government originally proposed a standard format of Advance Choice Document but now proposes a more flexible approach and that 'it should be led by what the service user feels is most important to facilitating their recovery' Reforming the Mental Health Act: government response. The research highlights that commitment from and support for staff may be as important as supporting the maker of the document. Further work may need to address how time and resources can be found for this in pressured clinical settings. The education level of participants seemed relatively high (9 out of 14 university educated). Further work might usefully consider whether different support may be needed for people with different educational backgrounds -the Mental Welfare Commission report on Advance Statements in Scotland found that people from more deprived areas were more susceptible to their Advance Statement being overridden. Another possible limitation is that the work focused on supporting people who were already committed to making an advance statement. It is less clear whether this would address the barriers experienced by those who did not choose to make a document -more than half the original cohort. Further research/pilot work may be needed to identify ways to make ADM more attractive to wider group. The research suggests that the process works best for those who are accepting of their diagnosis. It would be interesting to follow up on how to support people who have a more fundamental difference of perspective from professionals -who in some ways would be people who might seem most in need of an opportunity to make advance choice The research used templates originally based on the Mental Capacity Act, and the design of the packs were changed to cover physical conditions. This suggests that MHA treatment is not the only driver of advance choice making, and future work could consider how to bring together the two different legal frameworks in one advance planning process. Previous research has found barriers to advance directives in the MCA for people with bipolar disorder -see I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.