Training in health coaching skills for health professionals who work with people with progressive neurological conditions: A realist evaluation

Abstract Background Supporting people to self‐manage their long‐term conditions is a UK policy priority. Health coaching is one approach health professionals can use to provide such support. There has been little research done on how to train clinicians in health coaching or how to target training to settings where it may be most effective. Objective To develop theories to describe how training health professionals in health coaching works, for whom and in what circumstances, with a focus on those working with people with progressive neurological conditions. Design Realist evaluation using mixed methods (participant observation, pre‐ and post‐training questionnaires, and telephone interviews with participants and trainers). Realist data analysis used to develop and refine theories. Intervention Two 1‐day face‐to‐face training sessions in health coaching with 11 weeks between first and second days. Setting and participants Twenty health‐care professionals who work with people with neurological conditions in the UK, two training facilitators. Results Four theories were developed using context‐mechanism‐outcome configurations to describe how training triggers critical reflection; builds knowledge, skills and confidence; how participants evaluate the relevance of the training; and their experiences of implementing the training. Some participants reported a major shift in practice, and others implemented the training in more limited ways. Discussion Fully embracing the role of coach is difficult for health professionals working in positions and settings where their clinical expertise appears most highly valued. Conclusions Training should address the practicality of using coaching approaches within existing roles, while organizations should consider their role in facilitating implementation.

effectiveness of health coaching interventions is unclear, partly because of the variability in existing studies in mode of delivery, duration, intensity, characteristics of coaches and those being coached, which hinders conventional systematic reviews. [21][22][23][24] Research approaches that are designed to account for and understand this complexity are required to better understand health coaching interventions.
Training health professionals in coaching skills is the first essential stage of any health coaching intervention. There is currently a 'dearth of research', 162 pp., in relation to the training of health coaches. 25 While there is general agreement that training should last at least 2 days, and involve opportunities to practise coaching and experience being coached, 26,27 previous evaluations have found that responses to training and subsequent implementation vary widely. 17 To inform the design and tailoring of future programmes, and to assist commissioners in making decisions about what types of clinicians, working in which settings, the training might be most influential, we aimed to explore how staff working in the UK NHS with people with PNCs responded to 2 days of training in health coaching skills.
The two questions we aimed to address were as follows:  28,29 This evaluation builds on the earlier findings (examining a range of possible self-management support approaches) and explores their relevance to a specific health coaching intervention.

| ME THODS
We undertook a realist evaluation of a health coaching course delivered over 2 days (11 weeks apart) to 20 UK health professionals who work with people with neurological conditions. Realist evaluation is a theory-driven approach, sensitive to complexity, which focuses not on the average effectiveness of interventions, but on explaining the reasons why interventions work differently in different settings. 30 As such, it was well suited to addressing the aims of the research, and likely to produce policy-relevant findings. Realist researchers focus on building theories about causation. Interventions are understood to offer a certain set of resources, which are then introduced into a unique context. Contextual influences may include characteristics of individuals (e.g. level of enthusiasm), wider organizations (e.g. financial incentives) and other influences such as the history of the setting (similar interventions implemented in the past). Features of the context influence how individuals reason about the new intervention, and this reasoning process can then lead to or prevent outcomes of interest occurring. Theories which describe causation are presented using context-mechanism-outcome configurations which outline how the intervention mechanisms (a combination of the resources provided and the reasoning triggered), influenced by context, act to generate outcomes. 30,31 K E Y W O R D S continuing professional development, health coaching, person-centred care, realist evaluation, self-management Realist researchers recognize that our understanding of causation is only ever partial, and work towards a better understanding through an iterative process of theory generation, testing and refinement. 30 The evaluation reported in the current paper aimed to further test and refine the theories developed in our earlier realist review. 28 The theories presented in our review are briefly outlined in Table 4 mapped against the theories developed in the evaluation stage reported here.

| Recruitment
The training course was advertised through professional networks, emailing lists and at two national conferences. Expressions of interest were sought, including brief information about potential participants' professional roles and patient groups. The relatively small number of applicants formed a reasonably mixed sample, so further purposive sampling was not pursued. There were 44 expressions of interest, 38 training spaces were offered and 21 of these offers were accepted. Six were not offered a space because the course had reached capacity. One participant was unable to complete training (their pre-training data were excluded from analysis).

| Ethics
Ethical approval was obtained from Cardiff University School of Medicine Research Ethics Committee (SMREC 17/66). Participants were made aware of the research project in the initial information they received about the training. Participation in the research was not a mandatory component of the training, but all participants agreed to be involved and provided written informed consent.
Participants were given regular opportunities to decline to participate in parts of the evaluation in order to minimize any perceived coercion. Interview transcripts and field notes were anonymized for confidentiality. While the responses of other participants were discussed in broad terms during the realist interviews, anonymity was strictly preserved. Funding for the training was provided by an education grant from Novartis Pharmaceuticals. Novartis had no input into the training content, provision, evaluation or reporting.

| The intervention
A more detailed description of the training intervention compiled using the Template for Intervention Description and Replication (TIDieR) checklist 32 can be seen in Appendix 1. In brief, training was provided during two 1-day sessions in early 2018 by an external training provider with extensive experience of working with NHS staff. Usually, the second day is delivered after a 2-to 4-week gap to allow participants time to try the techniques in their clinical practice, but due to severe adverse weather the second training day was postponed resulting in an 11-week gap. Two trainers with clinical backgrounds facilitated the highly interactive course which they conceived and developed and have been running for several years. The approach combines executive coaching skills, behaviour change skills and clinical communication skills. Training included short presentations, coaching demonstrations, discussions with other participants and working in pairs to practise coaching and being coached using a variety of techniques. The training covered a range of specific coaching techniques as well as introducing other behaviour change strategies and topics relating to self-management support (see Table 1).

| Data sources
Realist evaluations typically use a mixed-methods approach, recognizing that different types of data can provide insights into different elements of context, mechanisms and outcomes. 40 Table 2 summarizes the data sources and the rationale for each chosen approach.

| Observations
FD, a practising clinician (general practitioner) and researcher, acted as a participant observer. Field notes were taken during both days

| Interviews
FD conducted all interviews and was already known to participants from the training. A realist approach to interviewing was used in which the theories under development were the main focus of the discussion. 30,42 Theories developed from our earlier review 28 were used to develop a topic guide. A teacher-learner style was used, especially in the later interviews when descriptions of the researcher's theories in development were 'taught' to participants, with the researcher seeking to 'learn' how these fitted with individuals' experiences. 30,42 The focus of the interviews therefore changed depending on the stage of theory development that had been reached, and the particular theories to which individual participants were expected to be able to contribute data. Both trainers were also interviewed. The interviews were audio-recorded and transcribed verbatim.

| Questionnaires
Bespoke questionnaires were designed and delivered at three time points: immediately pre-training, immediately post-training (paper- responses to be tracked over time.

| Data analysis
A triangulation approach was used with the qualitative and quantitative data being analysed concurrently by FD, (who has expertise and training in qualitative and realist methods). 44,45 The quantitative All qualitative data were imported into NVivo 11. An initial set of codes was generated using the theories from our review 28 and supplemented by additional codes developed from a reflective journal kept by FD during the interview stage. Further codes were inductively developed during the initial coding of the qualitative data from the questionnaires and the first five interviews. 46 After initial 'first pass' coding of all questionnaire and initial interview data, the coding framework was further refined, with some codes merged. An audit trail of all decisions was maintained. 'If-Then' statements were generated after exploring the coded data, initially at a lower level, close to the data, before being grouped together into related topic areas. 47 Four topic areas were developed which were used to generate higher-level theories, described using context-mechanism-outcome configurations. The qualitative data were reviewed alongside the quantitative data at this stage, with a focus on exploring the reported outcomes of the training.

| Enhancing rigour
This article was prepared with reference to the RAMESES II publication guidelines for realist evaluation. 48 The research was theory-driven, based on the findings of our earlier literature review. 28 Triangulation of data sources and data collection methods provided a more comprehensive understanding of how the training worked, and allowed convergence of the results to be identified. 49,50 The learner-teacher approach to interviewing meant that participants were able to refute and refine the researcher's theories. 30,42 Attention was also paid to the role of the researcher as a health professional, and the way in which this may have influenced the data collected and the interpretations made. For example, being observed by a colleague may have influenced participants' behaviour towards supporting the intervention. While being a clinical researcher facilitated rapport building due to shared understandings, it also increased the risk of making assumptions or missing the obvious. Regular meetings between all authors were held throughout the study and emerging findings discussed. Data extracts are presented with the results below to allow the reader to judge the inferences made.

| RE SULTS
Twenty participants completed the two training days. The professional backgrounds of the participants are shown in Table 3.
Complete data (completed interview and questionnaire) were available from 95% of participants at the immediate post-training stage and from 40% of participants at the follow-up stage (with a further 40% providing partial data). Nineteen participants were female. Sixty percent had worked in neurology for more than 10 years. Nine participants attended the training alone and 11 with someone else from their organization. Two participants worked with people with non-progressive neurological conditions.
They were invited to attend as several other members of their team were also attending, and it was felt that training a large team together could help to develop the theory about the importance of team support. Half of the participants worked with people with a single neurological condition, while half worked with people with a range of different neurological conditions. One participant was employed by a third-sector organization, and the remainder all had NHS roles.
We produced four refined theories describing how the training works, for whom and in what circumstances which are presented below. The content of each theory is first briefly described, and the context-mechanism-outcome configuration generated is then presented, followed by some of the evidence used to develop this theory.
Box 2 explains the labels used within the theory statements. Table 4 shows how the theories build upon and relate to the earlier work.

| Theory 1: critical reflection
This theory describes how training led participants to critically reflect on their current approach and their need to change their practice. The data suggested there was an interaction between different elements of context (individual and organizational factors), but it was not possible to identify the relative influence of the different elements with the data available.

| Theory 2: Knowledge, skills and confidence
This theory describes how the training process builds participants' knowledge, skills and confidence. The questionnaire data showed 90%-95% of participants' self-reported understanding of health coaching techniques and confidence in using them improved immediately post-training (see Table 5).
However, not all participants maintained these improvements when  Participants who were less experienced in their current roles reported finding it easier to prioritize the 'medical' aspects of the consultation. Some described that reverting to information provision was easier than trying the coaching approach.. Some more experienced practitioners who were confident in their roles appeared to find it easier to accept a coaching role, which emphasized their medical expertise less.
P7: I think if you're not so confident, or you're, then you feel that you've got to solve it [the patient's problem], or sort it and actually you haven't (follow-up interview) The development of confidence was closely linked to perceptions about how the approach could be used, and challenges related particularly to the PNC setting.

| Theory 3: Relevance to setting
This theory describes how the participants evaluate the relevance of the training to their own clinical setting. While the importance of evidence was highlighted in the earlier review, 28 the participants rarely brought up the evidence base for health coaching. When exploring the influence of evidence, 'evidence-based practice' was recognized as a gold standard, but personal evidence was often cited as more influential. Research evidence took on increased importance if participants needed to justify new practices to colleagues. As well as gathering personal evidence for the effectiveness, participants also made judgements about perceived usefulness for their patient group and fit within their existing routines. For those who worked in roles where medical or technical tasks were often the focus of the consultation, it was harder to be sure how to integrate coaching and 'role conflict' could occur when the coaching approach was not seen to fit with other tasks they were expected to complete. 51 P10: If someone is presenting with pain or spasticity and swallowing issues that they don't know what needs to be done, or what medication needs to be prescribed … but when it's more about talking to them about physiotherapy and exercise and lifestyle changes, that I think is where the coaching will come in a little bit more (initial interview) Therapists who were familiar with goal setting and challenging their patients, and who frequently discussed lifestyle changes with their patients, appeared to see most easily how coaching aligned with their existing roles. Some raised concerns that the training had inadequately prepared them to deal with more challenging scenarios, such as using coaching with people with mental health problems or cognitive impairment.

TA B L E 5 Quantitative results summary
The trainers' use of coaching techniques when queries were raised appeared to leave some participants with unanswered questions.
Other trainees were already confident in working with these patient groups, and this seemed to help them feel confident to try coaching.
Participants suggested it was helpful for teams working together to have a shared understanding of coaching, and it could be difficult to use a coaching style when working directly with colleagues who used an alternative more traditional approach.
P14: what I found difficult was being in a clinic with a physio, because I do joint clinics with a physio and they didn't, others hadn't always been on the training so that was quite hard (follow-up interview) Many participants were expected to provide in-service training to colleagues when they completed the training course, and this had raised awareness and encouraged participants to revisit the learning resources provided. Some reported using existing resources within their teams (e.g. access to clinical psychologist support) to continue to build their skills. It was also helpful if supporting self-management was identified as a local priority. As shown in the quote from Participant 1 above (Theory 1), the need to prioritize the completion of mandatory assessment forms which did not easily accommodate a coaching approach was cited as a barrier to implementation. For participants to experience success, they needed to try out using a coaching approach. The interview data indicated that while some motivated individuals made changes soon after training, others found disrupting their usual routines more challenging. Somewhat unexpectedly, even those who were low in confidence in their own skills often chose to try out coaching for the first time with patients who they were most struggling to support effectively. These scenarios seemed to offer a low-risk way to trial the new approach as participants had already accepted that their current way of working was ineffective. It was in these scenarios that many most valued the coaching techniques.

| Theory 4: Experiences of implementation
P6: for patients that I see that tend to throw up barriers to everything they want to do, I feel like I have a tool that I can work with them and apply it, and that's been really helpful, it's nice to know that I've got that if I need it (follow-up interview) When participants tried a coaching approach and observed how people responded, this could trigger significant reflection about the deficits in their previous approach. Those participants who experienced this type of transformative learning 52 described a move towards seeing patient engagement as a more co-constructed process, 53 and they started to understand the influence of their own behaviour.
P14: we do get the same patients sometimes coming through and I think sometimes we think 'it's them', and I do, that's my kind of shifting thought now is -is it because it's them? or are we actually giving them any responsibility over their health? (initial interview) The shift in awareness about the influence of their own approach on subsequent engagement appeared important in motivating continued use of health coaching skills and techniques following training.
While the follow-up questionnaire data were incomplete (65% response rate), they indicated that the benefits of the training were not maintained for all (see Table 5). Decreases in the ratings for confidence (46%), perceived usefulness (39%), perceived ease of use (39%) and motivation to apply the techniques (69%) were reported when compared with their immediate post-training rating (see Table 5). This may reflect how challenging some participants found it to apply coaching techniques and may also have been influenced by a lack of on-going support post-training.
Most participants were highly autonomous practitioners who had long appointment slots and could control how their work was organized to a certain extent. Even in these circumstances, which the earlier literature review 28   Other health coaching intervention studies found that some professionals reported already using a biopsychosocial approach and felt coaching aligned well with their role, while others who conceived their roles to be about providing professional advice and who wished to do what they felt was best for their patients found a coaching approach more challenging to integrate. 16,18,54 The way in which services are organized and audited provides clear messages about the value of professional expertise. The tendency to prioritize 'medical or technical tasks' seen in this study may reflect the lack of routine measurement of person-centred care, with work that is audited seen as the highest priority. 55 Prioritization of person-centred approaches by organizations influences how individuals prioritize these activities. 56 Organizations have an important role in promoting person-centred approaches as ways of completing routine work, rather than extra activities required in addition to other clinical tasks. 57 When coaching is seen as a way to manage demand and to work more successfully when people appear 'stuck', then individuals may be more receptive to integrating the approach. 17

| An overall programme theory
Perceived patient-level barriers to promoting a self-management approach identified here and elsewhere included co-existing physical and mental health problems 16,58 and wider social context (such as a lack of social or economic resources). 16,58,59 This realist evaluation used a theory-driven approach to test and refine a set of explanatory theories developed from the wider self-management support literature in the setting of a health coaching training intervention. 28 Key training mechanisms and the contexts in which they are facilitated or inhibited have been described.
While multiple data sources were used, increasing the trustworthiness of the findings, 49 61 We therefore recognize that the theories presented remain partial, and in line with the realist approach, new evidence could lead to further theory development. Further objective assessment of professional behaviour change and of subsequent patient-level outcomes is needed to further develop the theories proposed. Research to clearly define the desired outcomes of integrating coaching into routine care from the perspectives of a range of stakeholders could help inform future evaluations.

| CON CLUS I ON S AND RECOMMENDATIONS
Specific training in using health coaching techniques to make consultations more person-centred was highly valued by participants.

ACK N OWLED G EM ENTS
We thank all of the participants who took time to participate including the training providers for their openness to the evaluation process which was part of a PhD programme at Cardiff University.
Thanks to Deborah East for supporting the organization of the training days. Thanks also to Chris Burton and Patricia Wilson for their helpful comments on the research.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare. FD originally conceived the study, collected and analysed the data and prepared the first draft of the manuscript. AB, AE, CW and FW provided advice on the study design and data analysis and revised the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared. Materials: Describe any physical or informational materials used in the intervention, including those provided to participants or used in intervention delivery or in training of intervention providers. Provide information on where the materials can be accessed (eg online appendix, URL)

O RCI D
Topics and techniques covered are outlined in Table 1. All trainees were provided with a 123-page resource guide (which included space for notes). The booklet included all of the slides presented by the trainers during the two workshops (and some extra slides that were not discussed during the training days) Participants were encouraged to write in the resource guides If the intervention was planned to be personalized, titrated or adapted, then describe what, why, when and how Intervention encouraged participant interaction. Group discussion sessions were shaped by the issues raised by the participants and felt to be most relevant to them