How and how well have older people been engaged in healthcare intervention design, development or delivery using co-methodologies: A scoping review with narrative summary

Co-methodological working is gaining increasing traction in healthcare, but studies with older people have been slower to develop. Our aim was to investigate how and how well older people have been engaged in healthcare intervention design, development or delivery using co-methodologies. We conducted a systematic search of four electronic databases to identify international literature published between 2009 and November 2019. We included peer-reviewed empirical research of any design. Three authors screened papers. Our review is reported in accordance with the Joanna Briggs Institute manual for scoping reviews, we have referred to the preferred reporting


| INTRODUC TI ON
Co-methodologies in healthcare are increasingly considered to be a 'good thing'. However, the language of 'co' working is not fully defined and remains a fundamentally contested concept (Flinders et al., 2016). The terms co-design, co-production, co-creation, participatory research or participatory design are increasingly used, sometimes interchangeably by researchers and research funders (National Institute for Health Research INVOLVE (2020)).
For some, co-methods are synonymous with Patient and Public Involvement and Engagement (PPIE), whilst for others they represent a much more considered approach in which lay people and professionals work as equals at every stage in the research process (Co:Create, 2020; National Institute for Health Research INVOLVE, 2020).
Over the years, efforts have been made to devise methods of assessing forms of participation, some models express these hierarchically based on different criteria. The most frequently cited are the ladder of participation (Arnstein, 1969) and the typology of participation (Pretty, 1995). While a good starting point, these tools have been widely critiqued. Some argue that Arnstein (1969) over-emphasises power and, the lack of acknowledgement of different forms of knowledge and expertise limit the value of the typology (Tritter & McCallum, 2006). Cornwall (2008) states that both models retain a certain vagueness about the meaning of participation and suggest more attention needs to be paid to who is participating, in what and for whose benefit.
Some authors suggest non-linear approaches such as the Capire Engagement Triangle (CAPIRE, 2020) and the Spinning Wheel of Participation (Davidson, 1998). The Capire Engagement Triangle identifies desired outcomes of engagement based on the overarching objectives of informing decisions, building capacity and strengthening relationships. The non-linear nature of the tool means that it describes meeting these different objectives, however, it also assumes that participants need to be 'educated' and have their knowledge 'increased' or behaviour 'changed' (specified in the objective of building capacity). The tool does not include 'empowerment' and/or acquiring an 'equal partnership', as one of its objectives, thus like other models, the Engagement Triangle tool also appears to take a 'top-down' and a 'tokenistic' approach. The 'Spinning Wheel of Participation' is intended to define and encourage levels of citizen participation for community planning (Davidson, 1998). The spinning wheel is made up of four key themes, information, consultation, participation and empowerment and requires considerable commitment from community members. Aner (2016) outlines four increments of participation level when engaging with older people; (a) right to be informed, (b) right to be heard, (c) right to codetermine and (d) right to self-organization. This paper spans participation from three perspectives; older people's own interests, policy and legislation, and research and practice. The authors identify essential elements of effective participation including working together and creating opportunities to be engaged at different stages of research. Teoh et al. (2018) report two key challenges in co-methods work, firstly the potential for hierarchy within the team and secondly power dynamics.
In seminal work, Reed (2008) offered eight points of best practice in stakeholder participation. A decade later, Reed et al. (2018) categorised stakeholder and public engagement in the Wheel of Participation as: (a) top-down one-way communication and/or consultation, (b) top-down deliberation, and/or co-production, (c) bottom-up one-way communication and/or consultation and (d) bottom-up deliberation, and/or co-production. These authors suggest that the approach taken explains variation of outcomes. Hurlbert and Gupta (2015) offer a split ladder of participation which they describe as a diagnostic, evaluation and strategic tool. Both Reed et al. (2018) and Hurlbert and Gupta (2015) offer methodologically robust approaches, however, their work is complex to apply in real world settings.
The value and challenge of achieving 'good' co-working is well rehearsed in the literature (Ramirez, 2020). Advice on co-working is becoming increasingly available, for example 'Co:Create' offer a Co-production Matrix for self-evaluation purposes (Co:Create, 2020) and Ramirez (2020) a 'meditation' on meaningful participation. Good practice guidance is summarised in several documents (Flinders et al., 2016; National Institute for Health Research INVOLVE, 2020;Shimmin et al., 2017). Synthesis of these varied sources suggests common themes that contribute to 'good' co-working which include: engaging all stakeholders from an early stage, valuing the perspectives of all participants, avoiding existing stereotypes and managing power relations and inter-personal interactions.
Further value of using co-methods in healthcare research include bringing together researchers, healthcare staff and end users earlier in the research process, facilitating genuine innovation and improving performance (Jackson & Greenhalgh, 2015). This approach is congruent with current policy, particularly regarding the increasing emphasis on pragmatic, 'real world' approaches where, for complex

What is known about this topic
• Co-methodological working has become increasingly valued in healthcare • It is used less with older people What this paper adds • Co-methods frameworks, of which there are many, imply achievement of each included element is necessary.
• Engagement at every stage of a project may be unrealistic if studies are prolonged and the level of commitment beyond what any older person can offer.
• Reporting what co-methods work, when and for whom, would enhance future co-methodological research and practice.
interventions to be effective, they must be adapted to local contexts (Jackson & Greenhalgh, 2015). Increasing recognition of this value provides the opportunity to explore the extent to which different studies have achieved co-methodological working.
With a worldwide ageing population (World Health Organisation, 2020a), research must better address the needs of older people. In this paper we focus on how and how well older people have been engaged in healthcare intervention design, development or delivery using co-methodologies. For the purpose of the paper, we have defined older as participants described by authors as older, elder or senior. However, we acknowledge that definitions of older persons vary widely. Most developed countries use the chronological age of 65 years as a definition of 'elderly' or older person (World Health Organisation, 2020b). This age is also an implicit marker of old age by the United Nations (2020). Perceptions of old age are influenced by culture and also the age of the person offering the definition. For example, people in early adulthood tend to categorise any age from 50 years upwards as old (Aged Care Guide, 2020).
Globally there is a steady rise in the number of centenarians (Robine & Cubaynes, 2017) and many people may experience living in older age over a long time span during which their health and wellbeing will inevitably change.
Research partnerships with older people are increasing, but the ethos of co-methods have been slower to develop with this group than with others (Blair & Minkler, 2009;Littlechild et al., 2015) and older people are at risk of systemic exclusion (Shimmin et al., 2017). Ability and desire to engage in co-methodological working will vary between individuals but may also be influenced by age; for example the 'older old', those aged over 80 years are more likely to be living with sensory and cognitive impairment (Wayne & Johnsrude, 2015). As the silent generation, they may need more support to express their views and needs (McIntosh-Elkins et al., 2007).
At present there is a dearth of literature that fully illuminates process, practice and benefits of co-methodological working with older people in healthcare. Given the current trend towards co-methodological working and the relative lack of guidance on approaches with older people (Littlechild et al., 2015), this review aims to investigate how and how well older people have been engaged in healthcare intervention design, development or delivery using co-methodologies. We have chosen to focus on a wide breadth of healthcare including interventions in primary care, secondary care and the community setting, however to provide a degree of focus we excluded non-direct health interventions, for example public health interventions such as those addressing the environment, housing or social isolation. Mapping, synthesis and identification of gems of good practice in available literature will be of value for clinicians, managers and policy makers who are planning such endeavours. This new knowledge will help to inform future co-working and benefit the older population.
For clarity, we have used co-methods (C-M)/co-methodological working (C-MW) as umbrella terms, but have referred to specific approaches as reported in each included paper. The protocol for this review is registered at the Centre for Open Science, Open Science Framework .

| Objectives
To conduct a scoping review of empirical research reporting use of co-methodological working with older people in relation to design, development or delivery of healthcare interventions. The review questions were: 1. How have older people been engaged in co-methodological working in relation to healthcare interventions?
2. How well have older people been engaged in co-methodological working in relation to healthcare interventions?

| Design
Our review is reported in accordance with the Joanna Briggs Institute (JBI) manual for scoping reviews, as our aim was to provide a broad overview of our topic (Peters et al., 2020). For completeness we have referred to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Moher et al., 2009). How well older people have been engaged in C-MW is considered in relation to the Co:Create Co-production Matrix. We offer a narrative summary of included papers and identify gems of good practice, by which we mean small extracts to which we were drawn and which offered a key to understanding what matters in C-MW with older people and which illuminate researcher's commitment to C-MW and may inform future planning.

| Study selection
We conducted a systematic search of the electronic databases The Cumulative Index to Nursing and Allied Health Literature (CINAHL),

MEDLINE, PsycINFO and Applied Social Sciences Index and
Abstracts (ASSIA). Databases were selected to capture key health literature. A limit of 10 years was applied (2009 -November 2019) as C-MW is a rapidly evolving field. To aid the development of key search terms (see Table 1) we used the PICo (Population, Intervention and Context) framework (Moule et al., 2016). Boolean terms including truncation were applied but MeSH terms were not used to avoid inadvertent exclusions. Additionally we searched for existing reviews in Prospero and the Cochrane Database of Systematic Reviews. We TA B L E 1 Key search terms older or elder* or senior AND co design or co production or co creation or 'participatory research' or 'participatory design'. completed forward and backward citation searching of included papers. We did not engage with the grey literature as our focus was solely on peer reviewed work.

| Study eligibility
Inclusion criteria included international peer-reviewed empirical research studies published in English language. We were interested in health-related C-MW with older people. Further details of eligibility are summarised in Table 2.

| Search outcomes
We examined a total of 2,093 titles with 1,886 being excluded as not relevant to the aim of our review. Subsequently two authors (JD, FC) independently reviewed the remaining 207 full texts. Of these 159 were excluded. We resolved disagreements by discussion and when required by consulting with a third author (MS). Reasons for exclusion were; no evidence of C-MW, not older person specific and descriptive papers. No literature reviews of any type were identified.
Forty-eight papers are included in our review. Figure 1 outlines the process of study selection.

| Quality appraisal
Our review focused on quality of C-MW rather than quality of the research per se. For this reason we did not assess methodological quality as this may have excluded papers that included information pertinent to our review. As discussed above, we assessed quality of C-MW against the Co-create Matrix. We are mindful that the Matrix is intended as a self-assessment tool rather than as a quality measure for papers reporting C-M studies, nevertheless it offers a framework that we consider most congruent with current thinking about C-M, particularly in the context of healthcare. The eight Co:Create descriptors can be assessed at 'we're not doing this and we've not started to think about it yet' which we scored at zero, Bronze (score 1), Silver (score 2) and Gold (score 3). The minimum score possible was zero and maximum 24. These descriptors are the closest fit with our area of interest and are presented in Table 3. Quality appraisal was conducted separately by two authors (FC, RP) with any discrepancies being resolved in discussion with a third author (JD). Where appropriate we extracted gems.

| Data extraction
The papers (n = 48) were divided between authors and each extracted data independently. We then discussed a sample of papers (n = 18) selected arbitrarily, and any discrepancies were identified and resolved as a team. Adjustments to data extraction from other papers were made if required to ensure consistency. To address our review question, we designed a bespoke data extraction sheet with a focus on extracting data directly relevant to co-methodological working with older people. Specifically we wanted to (a) map the type of interventions being co-created (aim of study), (b) identify focus of the paper, whether reporting on the intervention development per se or offering a process evaluation (methodology), (c) report details regarding C-MW (co-approach, aim of co-approach, summary of co-procedures) and (d) provide a total Co:Create score.
Data extraction is summarised in Table 4.

| Data synthesis
The included papers (n = 48) were heterogeneous in nature with a mix of qualitative and mixed methodologies and process evaluations reporting C-MW from differing perspectives. All papers reported on C-M to a greater or lesser degree. We conducted a narrative synthesis through textual description to identify commonalities and gems to progress understanding of how best to engage older people in C-MW. Through this we offer coherence in a diverse body of evidence (Campbell et al., 2018). In presenting our results we firstly offer a summary of study characteristics. We then focus on our review objectives to assess of how and how well older people have been engaged in C-MW. Finally, we synthesise learning from reported studies and point towards future research and practice possibilities.

Inclusion Exclusion
Design: • Peer-reviewed empirical research of any design • Relevant to health care in its broadest sense (primary care, secondary care, independent sector, community settings), focus on healthcare interventions • Explicit use of co-methodologies (co-design, co-production and cocreation or participatory research, participatory design) Population: Participants older people as defined by the authors (older, elder, senior) Limiters: English language, 2009-2019 Design: • Non-peer-reviewed articles, editorials and discursive (opinion) papers, protocols, theses, grey literature • Not healthcare related, for example interventions to improve the environment, housing or circumstances such as social isolation • Focus on engagement, involvement or consultation of service users without specific reference to 'co-' or 'participatory' methodologies Population: People not defined as older by authors Limiters: Non-English language, not 2009-2019

| Study characteristics
The majority of studies were conducted in the UK (n = 13) and the USA (n = 12). Other countries included the Netherlands and Canada (n = 5 each), Australia and Sweden (n = 3 each), one each from Slovenia, Denmark, Thailand, Ireland and China. Two papers report on dual country studies involving Netherlands/Hungary (n = 1) and Netherlands/Italy (n = 1). In many cases, it was not clear in what type of venue the work was conducted, but most were sited within local communities. Methodologically we have categorised papers as broadly qualitative (n = 32), mixed methods (n = 3) or process evaluation (n = 13) according to the major focus of the paper. In reality, this represents a continuum as many authors report on a combination of primary research with a greater or lesser degree of reflection on C-MW. External funding is reported in most papers (n = 37), this ranges from modest scholarships to large national grants.
Interventions reported ranged from equipment design, for example mobility aids (Boerema et al., 2016) and bottle openabillty (Flinn et al., 2013), production of culturally appropriate information for patients (e.g. Parker et al., 2012), development of on-line resources, for example telemedicine and telehealth solutions (Duh et al., 2016) and an eHealth carer needs assessment tool . Whilst some projects had a micro focus, for example Baur and Abma (2012) worked to improve meals in one residential home, others undertook work to intended to have a wider sphere of influence through health promotion interventions (e.g. Bone et al., 2013;Schensul et al., 2009).
Types of participant varied across studies with 15 involving only researchers and older people. Detail about the constitution and expertise of the research teams in these studies is sparse but some are clearly multi-disciplinary, for example Sandlund et al. (2016) includes researchers with expertise in physiotherapy, informatics and knowledge engineering. The remainder of the studies (n = 33) engaged broader teams including informal caregivers, members of the wider community, influential stakeholders (e.g. elder services providers, advocates, senior housing managers), subject experts (e.g. educators, sociologists, nutritional scientists, graphic or industrial designers, software engineers), clinical experts (e.g. nurses, medical staff, physiotherapists). Nguyen et al. (2019) and Span et al. (2018) engaged with different combinations of participants for each stage of the process. Where reported, older person participant numbers ranged from seven (Baur & Abma, 2012) to 386 (James et al., 2015).
The age of older participants is detailed in some studies, but many offer general terms such as 'older adults', 'older people' or 'mid-life or older members'. Where specified participants were aged 55 years or older.

| Methodological quality
The assessment details of co-methodological quality for each of the included studies are presented in Supplementary File 1  Older adults (n = 45) and researchers.
Researchers compared differences in participatory processes.

Qualitative
Multi-phase codesign process. Using the co-design process in order to offer older patients with cancer a usable website.  Five workshops over 4 months with reflection at end and circulation of notes and agreed actions.

TA B L E 4 (Continued)
overall score offered in Table 4. No studies scored zero 'we're not doing this and we've not started to think about it yet' for any descriptor, as all had at least implicitly considered every descriptor at some level. Figure 2 presents a summary of scores across papers by descriptor. We expected that process evaluation papers would achieve higher scores in view of the focus, but this was not the case.
Scores ranged from 8 in a mixed-methods paper by Smith-Jackson et al. (2010) to 21 in a qualitative report by Schensul et al. (2009). The descriptors with the lowest scores overall were for 'resourced' with papers scoring 1 (bronze) (n = 23) and 'sustainable' scoring 1 (n = 24).

Scores of 3 (gold) were most frequent for the descriptors 'inclusive'
(n = 11) and 'iterative' (=10). We treat total scores with caution, as many papers report on high quality co-methodological working in discrete areas, for example in intervention design. This perhaps reflects the relative immaturity and specific challenges of C-MW with older people with or without associated health conditions.

| How were older people engaged in comethodological working?
Precise description and underpinning theory of the C-M used is absent in many papers. Broadly methods comprised participa- Our data extraction included the specific aim of the C-MW. In many cases this was presented as a variation to the aim of the study itself or in vague language for example, to work in a participatory and responsive manner (Clarke et al., 2009). Others were more precise, for example to engage wheel chair users, care providers and clinicians in an iterative design and development process (Giesbrecht et al., 2014).
The most frequently reported type of engagement was through variously described communication such as interviews, focus groups, facilitated discussions, community consultation and workshops. Some older people worked practically alongside researchers to design and/or test 'products' for example a re-ablement system (Bond et al., 2015), a peer-education guide (Clarke et al., 2009), development of a home based dizziness intervention (Grönvall & Kyng, 2013) and a digital life-story book (Subramaniam & Woods, 2016).

| How well were older people engaged in comethodological working?
We report here according to descriptors of the Co:Creation Matrix.
In each case a brief overview of the descriptor is provided as the context for our assessment.

| Holistic
Best practice guidance suggests that C-MW should take place at every stage of a project including planning, delivery, evaluation F I G U R E 2 Summary of quality appraisal scores across papers by descriptor and governance. In virtually every included paper the original study idea was generated before the C-M team was constituted.
To some extent this is inevitable in that (a) until researchers know the question they are addressing they are not able to seek the older people with the required knowledge and experience and (b) all research activity has to be resourced and until funding is secured it is not always possible to progress. Although many funders now require PPIE in applications, this is often quite separate from the research that follows. In the United Kingdom (UK) there is a move towards C-MW to identify important research questions for specific conditions through priority setting partnerships (PSP; James Lind Alliance, 2020).
Early stage older person engagement is evident in a PSP to prioritise the research agenda in improving the oral health of older people (Brocklehurst et al., 2015). However this PSP followed the prescribed design and although older people were involved at each stage up to question generation it is not clear what part, if any, they played in disseminating the work and influencing research to address prioritised questions. Two studies engaged older people at the design stage, to improve cancer screening (Bone et al., 2013) and to develop a culturally appropriate diabetes intervention (Wang-Letzkus et al., 2012). Other studies report C-MW at multiple stages including for example, design workshops and committee meetings , group meeting with site meetings and intervention refinement processes (van Velsen et al., 2015) and designing interview tools, participant recruitment, interviewing, data analysis and dissemination of findings (Ellins & Glasby, 2016). The most comprehensively reported element in most papers is the 'doing' element of C-MW through communication and design and/or testing of 'products' as discussed above.
Few studies explicitly report co-data analysis, in most instances there is a sense that researchers 'take' the data, analyse it and, in some cases, present it back to the group. Dissemination activity appears to occur post project in most studies. Study governance is not explicitly addressed in any included paper.

| Resourced
Excellent C-MW requires sufficient resource. Although financial support is crucial, perhaps more important is the need for adequate time for the project and the ability to adjust timescales and process as the project progresses to ensure that C-MW is achieved. The majority of studies report external funding, but the value of this to C-MW per se is not specified. Reimbursement for participation is recorded by Giroux et al. (2019) and adjustments to timescales by Ralston et al. (2017) and Revenäs et al. (2018). Only three studies explicitly report resource in terms of extra time being used to adapt to enable full engagement of a 'weak group' (Grönvall & Kyng, 2013), people with dementia (Revenäs et al., 2018) and the whole team (Ellins & Glasby, 2016). Researchers invested time and thought into project planning prior to older person engagement, however reports suggest that the C-MW was delivered at a pre-ordained time and using methods fixed by the researchers in advance, but this does not imply that the C-MW was without benefit.

| Transparent
Older people should understand why they are involved in C-MW, their remit in the project and should be able to reframe the work as it progresses. Reporting of older people's understanding of their role suggested that in most cases they understood the purpose of the research rather than appreciating the purpose of CM-W in and of itself. This is perhaps reflective of some uncertainty for the research team about why they were using C-Ms. Other than that it is a 'good thing', few papers clearly articulated the aims and added value of this approach.
Detail about how older people were prepared for C-MW or their understanding of overall aims, limitations, expectations and commitment is not often reported. Exceptions are Baur and Abma (2012) who conducted an early stage informational dialogue group with older residents who, having had time to think, were able to opt into C-MW. Preparation for C-MW was provided by Ellins and Glasby (2016) in a series of five training sessions. Other teams offered training for specific elements, for example training on interview techniques (Jacobs, 2010), focus groups in preparation for the participatory design session (Lucero et al., 2014), three sessions to prepare co-researchers (Tanner, 2012) and a training session on photo-voice (Yankeelov et al., 2015). In reality, it is likely that many older participants received a research participant information sheet, as required by most research ethics committees, and this is simply not reported. It does however suggest that the C-MW approach is well established before the older people join the team.

| Inclusive
C-MW is intended to engage all members, with activities being designed to be accessible and a range of viewpoints represented. In some cases, for example (Allen et al., 2016;Baur & Abma, 2012;Flinn et al., 2013) there is marked demographic homogeneity in older persons thus limiting the opportunities for including wide ranging views. Likewise, in a minority of cases, the 'usual suspects' have been engaged through existing channels such as the Alzheimer's Society research network volunteers (Rapaport et al., 2018) limiting diversity and breadth of understandings and experience. Several studies report multi-disciplinary, lay-professional-practitioner-researcher C-MW, for example Bond et al. (2015) included alongside older people and researchers, developers, direct users and industry re-ablement representatives. Similarly service users, third sector partners and professionals worked together on a PSP (Brocklehurst et al., 2015). In some cases, entirely appropriately, projects included only small numbers of older people and researchers or older people from specific groups. For example, Tanner (2012) worked intensively with three older persons with dementia to prepare and support them in interviewing peers. Likewise, Grönvall and Kyng (2013) worked in triads with older persons in their own home and a specialist physiotherapist to maximise opportunity for the design and development of a home-based, technology-assisted, dizziness prevention intervention that would be acceptable for, and used by, older people. The aim for cultural competence guided the recruitment of older persons in creating a talking book for Vietnamese elders with dementia (Goeman et al., 2016) and in exploring palliative and hospice care with Native Americans (Isaacson, 2018). Inclusivity is clearly concerned with getting the right people for the study, rather than aiming for large numbers or unwarranted diversity.

| Iterative
Ideally for example a prototype web-based falls prevention system (Lucero et al., 2014) and (b) researcher led feedback from one session used to introduce the next. Papers we assessed at the highest level went beyond feedback alone to report how group responses were integrated into the next stage of the project; for example changing direction of the project (Baur & Abma, 2012), amending plans (Clarke et al., 2009) and reconsidering interventions (Ellins & Glasby, 2016;Wang-Letzkus et al., 2012).

| Positive
All members should be valued, heard, engaged, committed and empowered. Building relationships is key to positive C-MW experiences but reporting is limited. Exceptions include, for example, one instance of emphasis on getting to know each other and feeling more comfortable in order to share experiences (Baur & Abma, 2012), two reports of development and maintenance of relationships across the team (Span et al., 2018;Tanner, 2012) and one description of using the culturally sensitive 'talking circle' method to promote communication (Isaacson, 2018). Details of how older persons' views were elicited are well documented. However, in many instances there was a sense of views being 'taken' by researchers and used to progress the project with limited input from older participants. Nuanced evaluation from older people about their experiences of C-MW was notably absent, indeed many authors did not reflect on this at all.
Some suggested that feedback was broadly positive (Brocklehurst et al., 2015;Bulsara et al., 2016;Clarke et al., 2009;Tanner, 2012). Bone et al. (2013) imply that older people were able to articulate the specific value of C-MW in creating a new navigation system.
Older participants designing a Parkinson's disease eHealth intervention suggested how future C-MW may be enhanced (Revenäs et al., 2018) and in the case of Schensul et al. (2009)

| Equal
All C-M participants should be equal, power imbalances should be addressed and solutions to problems should be address collaboratively. Around two thirds of included studies were researcher led and five did not explicitly state leadership (Boerema et al., 2016;Clarke et al., 2009;Holliday et al., 2015;Morrison & Dearden, 2013;Wang-Letzkus et al., 2012). Exceptions included studies that claimed to be 'stakeholder-led' (Bond et al., 2015) or guided and advised by a committee or project team (Bone et al., 2013;James, et al., 2015;Nguyen et al., 2019;Ralston et al., 2017). Three teams adopted a model of shared control (Hwang et al., 2015;Sukwatjanee et al., 2011;Wang et al., 2014). There was one example of participants leading the agenda and priorities although the overall project was led by researchers (Yankeelov et al., 2015). There are different forms of leadership, and democratic approaches with equal weight given to opinions of older people may be most apposite. The greatest need is to ensure that older person contributions are explicitly valued and form part of the project as a whole. Equality is not necessarily desirable or achievable and it may be that equity is a more realistic aim. There is a need to consider power within the team; this is only explicitly addressed in three papers. Morrison and Dearden (2013) report actively avoiding power differentials through the use of mechanisms such as shared language and engagement with artefacts. Dominant and silent participants were managed using skilful facilitation by Revenäs et al. (2018) and this approach was used by Zeitz et al. (2011) to mitigate tensions between clinicians and older people when the latter thought that they were merely being used to 'rubber-stamp' existing decisions.

| Sustainable
Impact of CM-W on sustainability of the project should be clear to all and evidenced, with older people able to articulate the value of the endeavour. In general papers reported one-off C-M projects.
While many papers suggest good project outcomes, for example the availability of a new website, patient information leaflet or aid, few document experiences of older participants or indicate if or how the group may work together in the future and how the outputs from the project will be initiated and sustained. There are exceptions with older people reporting 'very rewarding' (Allen et al., 2016) and'empowering' and'positive' experiences (Bulsara et al., 2016). Carers and professional participants reported co-researchers with dementia had enjoyed the process of interviewing and suggested it had increased their self-esteem and social skills (Tanner, 2012). Older people engaged in a World Café expressed a desire for their falls prevention work to continue (Bulsara et al., 2016). Similarly, older people designing a Parkinson's eHealth intervention envisaged possibilities for further co-design (Revenäs et al., 2018). In two cases co-working groups stayed together beyond the life of the project.
Both were community focused, Clarke et al. (2009)  Intention of future C-MW is implied by authors who reflect on lessons learnt from the C-M project. However, discussion is often superficial and largely based on the opinions of authors rather than data, and focuses on outputs rather than C-MW processes.
Three key lessons are identified. First, retention can be problematic with older people becoming tired (Clarke et al., 2009;Grönvall & Kyng, 2013), finding it hard to focus for long (Duh et al., 2016) or fully engage particularly for people with cognitive impairment (Span et al., 2018). To address this, authors suggest the need to be flexible (James et al., 2015), allow plenty of time (Jacobs, 2010;Nguyen et al., 2019;Span et al., 2018;Wang et al., 2014;Zeitz et al., 2011), allow time for small talk (Span et al., 2018) and consider the unique support needs of each person (Jewitt et al., 2016) including physical, sensory or cognitive impairment (Boerema et al., 2016). Second, communication and trusting, facilitative relationships were considered key (Tanner, 2012). To address this, approaches included the use of artefacts. For example, three representational artefacts: emotion maps, stories, and tracing paper were used by Morrison and Dearden (2013) to help to facilitate interactions between lay participants and health professionals. A web-based research platform offered an opportunity to view study progress as not all participants were able to attend every session (James et al., 2015).
Cue cards were valuable memory prompts and building on people's own experiences and adequate briefing/debriefing in supported full, respectful and meaningful participation (Tanner, 2012). Finally, empowerment, equity and power relations were considered to be necessary both for the process (Baur & Abma, 2012;Jacobs, 2010;Morrison & Dearden, 2013;Wang-Letzkus et al., 2012) and outcome (Buckley et al., 2018;Flinn et al., 2013). Cultural awareness and language proficiency were considered necessary (Isaacson, 2018) along with a non-linear, reflexive process with space to exchange ideas and experiences (e.g. Baur & Abma, 2012;James et al., 2015;Morrison & Dearden, 2013). Additionally, more time than expected may be needed and timeframes should be flexible (Wang et al., 2014). When under time pressure there was a tendency to slip back into more traditional researcher/participant relationships (Jacobs, 2010 (Hewitt et al., 2013) was considered essential. Tanner (2012) was the only author who noted the importance of maintaining contact with the team during fallow periods of the process.

| D ISCUSS I ON
We identified 48 papers reporting on C-MW with older people in health care intervention design, development or delivery research.
Our aim to report how older people were engaged was met as most papers clearly reported stages of the project and to some extent the practicalities of older person involvement. In summary, older people were rarely engaged in planning or disseminating work. Their contributions were almost entirely concerned with providing data and to a lesser extent data analysis. Older people largely provided information predominantly in interviews, focus group and workshops and in design and development of a range of interventions including training, eHealth, web and hard copy information and equipment.
Heterogeneity of intervention was notable indicating a somewhat ad hoc approach to engaging older people in C-MW, seemingly often driven by local enthusiasts rather than being an approach embedded in policy.
Assessment of how well older people were engaged in the process proved more challenging. A systematic review of participatory action research (PAR) in gerontology (Blair & Minkler, 2009) offers an analysis of 13 exemplars reporting 10 studies presented under pre-defined PAR principles. The authors provide grounded suggestions on how effective engagement may be achieved. These include respect for life experiences, building two-way trust, and sufficient preparation. These authors highlight the need to accept that research will produce exemplary rather than generalisable results and the need for personal investment and tolerance of delays. Littlechild et al., 2015) suggests the need for broad inclusion to ensure representation of hard to reach groups (e.g. people living with dementia or from the black and minority ethnic communities).
Broad-based reviews of C-MW with other age groups, but still within the context of healthcare are scarce. The exception is an investigation to identify the concept of healthcare co-production and discuss its effects and implications. In contrast to our review, Palumbo (2016) concludes from included papers (n = 65), co-production in healthcare is problematic predominantly due to health care practitioner 'hostility' and patient 'unwillingness' to engage. If co-production is to improve this author suggests a need for greater inter-disciplinary working, more effective lay-practitioner communication and greater use of information technology. This difference in findings may be attributed, at least in part, to our clinically focused lens as opposed to the management perspective of Palumbo (2016).
In essence, existing research on C-MW with older people in healthcare concurs with our findings. It suggests practitioners and researchers who are committed to, but not always yet highly skilled in the practice of C-MW. Although C-MW with older people is an emerging field it may be that it is more advanced than is recognised given Palumbo's (2016) findings of hostility and unwillingness to embrace C-MW. These reviews highlight areas for development whilst our review extends knowledge to offer practical directions for future C-MW working with older people in healthcare.
In all areas of research there has been a proliferation of checklists and reporting guidelines. Most closely allied to this review is the GRIPP 2 checklist to report patient and public involvement in research (Staniszewska et al., 2017). We have no doubt of the value if transparent reporting however, we urge caution against employing checklists as 'how to' tools. Checklists can imply rigidity that is counterintuitive to the non-linear and dynamic nature of C-MW.
The use of any ladder, wheel or matrix of C-MW potentially implies that all elements are of equal importance and that to be successful a project should score as highly as possible in each. Notably, despite some high quality examples of C-MW, none of our included papers reached the maximum score. We argue that notions of 'best' and 'scores' are an oxymoron in C-MW and that aiming to achieve highly in all elements may stifle C-MW, particularly in the real world of health care where resources are finite and we may be working with people with varying degrees of functional and cognitive challenges.
What is more important is that researchers move towards C-MW where appropriate. We argue that engagement at every stage of a Our review, in which we have mapped, synthesised, identified gems of good practice and offered pragmatic directions for improving CM-W in healthcare with older people is of value to clinicians, managers and policy makers who are planning such endeavours. This new knowledge will help to inform future co-working and benefit the older population.

| CON CLUS ION
Our review demonstrates the diverse use and reporting of C-MW with older people. We found a lack of clarity about whether aims from participation had been met. Establishing how older people were engaged in C-M was largely straightforward. How well this was done was more challenging, however we have identified gems of good practice. The Co:Create Co-Production Matrix was the best fit for evaluating papers, however it is not intended as a measure per se. In essence, we argue that notions of 'best' and 'scores'

CO N FLI C T S O F I NTE R E S T
The authors declare no potential conflict of interests with respect to the review, authorship and/or publication of this article. All authors read, contributed to and approved the final manuscript. All authors confirm that there are no issues related to journal policies. The authors declare that they have no competing interests. All authors confirm that the content of the manuscript has not been published, or submitted for publication elsewhere.