Variations in stakeholders' priorities and views on randomisation and funding decisions in out‐of‐hospital cardiac arrest: An exploratory study

Abstract Background and aims Prehospital critical care for out‐of‐hospital cardiac arrest (OHCA) is a complex and largely unproven intervention. During research to examine this intervention, we noted significant differences in stakeholders' views about research, randomisation, and the funding of prehospital critical care for OHCA. We aimed to answer the following questions: What are stakeholders' priorities for prehospital research? What are stakeholders' views on randomisation of prehospital critical care? How do stakeholders consider allocation of resources in prehospital care? Methods We undertook an explanatory qualitative framework analysis of interviews and focus group with 5 key stakeholder groups: patients and public, air ambulance charities, ambulance service commissioners, prehospital researchers, and prehospital critical care providers. Results We undertook 3 focus group discussions with a total of 23 participants and 8 interviews with a total of 9 participants. Despite sharing a common appreciation of the concepts of scientific enquiry, fairness, and beneficence, the 5 relevant stakeholder groups displayed divergent views of research and funding strategies regarding the intervention of prehospital critical care for the condition of OHCA. The reasons for this divergence could largely be explained through the different personal experiences and situational contexts of each stakeholder group. Many aspects of the strategies suggested by the stakeholder groups only partially aligned with principles of traditional evidence‐based medicine, but were held with strong conviction. Discussion Analysis of the views of 5 stakeholder groups regarding research and the funding of prehospital critical care for OHCA revealed shared values but a variety of different strategies to achieve these. This knowledge can help researchers in similar fields in the planning and presentation of their research, to maximise impact on decision making.


| INTRODUCTION
Out-of-hospital cardiac arrest (OHCA) is defined as the sudden cessation of cardiac activity, leading to collapse and absence of signs of life, outside of the hospital setting. 1 It is frequently caused by underlying ischaemic heart disease and considered to be 1 of the main causes of mortality worldwide. 2 In the UK, the incidence of OHCA is estimated at over 28 000 per year, with a survival rate of less than 10%. 3 The current standard treatment entails advanced life support (ALS), consisting of lung ventilation with oxygen, CPR, defibrillation, and administration of intravenous medication. 4 In an effort to improve outcomes, some regions in the UK dispatch prehospital critical care teams to OHCAs, in addition to the standard of ALS paramedic care. 5 Prehospital critical care can be described as a bundle of interventions beyond the remit of standard emergency medical service (EMS) treatment, delivered by a group of specialist prehospital health care providers. 6 The mechanisms by which prehospital critical care might improve outcomes in OHCA are the experience of the providers, advanced interventions (such as prehospital anaesthesia), or the ability to transfer patients over greater distances to cardiac arrest centres, using equipment and interventions not available to ALS paramedics. 6 However, there is currently no clear evidence that prehospital critical care improves survival following OHCA when compared to ALS care. 5 Prehospital critical care is funded through a complex and variable combination of charity support (particularly if associated with an air ambulance) and National Health Service (NHS) ambulance service funding. 7,8 Researching potential benefits from prehospital critical care for OHCA is important to guide further funding, but is challenging because of the complexity of prehospital critical care itself and the limitations of undertaking prehospital research in OHCA. The Medical Research Council recommends researchers evaluating complex interventions to "involve stakeholders in the choice of question and design of the research to ensure relevance". 9 Furthermore, the document encourages researchers to "always consider randomisation, because it is the most robust method of preventing […] selection bias". 9 During the planning process for a research project to investigate prehospital critical care for OHCA, we discussed whether this should be a randomised or observational study design, with an OHCA patient and public involvement (PPI) group, other researchers, and clinical colleagues working in EMS.
The different stakeholder groups disagreed strongly about the ethical acceptability of randomising the intervention of prehospital critical care for OHCA, the information that is required to direct health care funding, and even the need to research the question at all. Given the importance of undertaking stakeholder-relevant research, we decided to formally investigate this wide range of relevant stakeholder views. Using prehospital critical care for OHCA as an example of a complex intervention, this qualitative research aims to answer the following questions: • What are stakeholders' priorities for prehospital research?
• What are stakeholders' views on randomisation of prehospital critical care, and what are the underlying principles?
• How do stakeholders consider allocation of resources in prehospital care?
We hope that an awareness and understanding of the differences in stakeholders' views can improve the dialogue between stakeholders and help in future planning of research of complex interventions, particularly in prehospital care. The following paragraph provides a short overview of relevant factors of the intervention and condition discussed in this research.

| METHODS
This qualitative research used a pragmatic mixture of stakeholder focus groups and interviews, followed by framework analysis to address the research questions listed in the introduction. The focus is on providing useful and applicable information for all stakeholders involved in prehospital care research. We initially planned for data from all stakeholder groups to be collected through focus groups.
However, because of the geographic dispersion and limited availability of some of the participants, for 2 out of the 5 stakeholder groups (air ambulance charity staff and prehospital researchers), this had to be changed to interviews. The research team monitored for signs of bias which might be introduced through this mixture of data collection methods, and the conduct of the interviews closely mirrored that of the focus groups.

| Research paradigm
The researcher team consisted of 3 members. JVVF and JB work as emergency medicine and prehospital physicians, and have published largely quantitative research with an underlying positivist epistemology. 10,11 JBR is a research fellow who has focused on qualitative research in a variety of health care settings 12 and has previously worked with JB on a qualitative study in prehospital care. 13 This exploration of stakeholders' views of prehospital critical care for OHCA is nested within a quantitative analysis of the effects of this complex intervention on survival after OHCA. The limitations and challenges of research in prehospital care and particularly OHCA require a certain degree of flexibility to the methods used. 14 We, therefore, adopted a pragmatic research paradigm, which is reflected in the mixed-methods structure of the overarching project. 15,16 Research paradigms can be seen as the link between the aim and the methods of a research project, and described by their ontology, epistemology, and methodology. 17 To the research presented here, under the pragmatic paradigm, reality is considered to be something to be negotiated and agreed on (ontology). 18 The pragmatic paradigm does not prescribe a particular epistemology or methodology but considers approaches valid if they work and provide useful results. 18 With this focus on useful results in mind, we chose the framework approach for data analysis, which falls under the wider method of thematic analysis. 19

| Participant selection
We identified stakeholder groups for a hypothetical randomised controlled trial of prehospital critical care for OHCA, based on the authors' experience in previous relevant research and in preparing the fellowship application for this project. 10,15 The groups varied significantly in their background, availability to attend focus groups or interviews, and their geographic distribution.
The recruitment process was, therefore, tailored to each group to achieve effective recruitment. A description of the recruitment method for each group can be found in Appendix A. For each group, we aimed for 4 to 10 participants in the focus groups and 4 to 6 participants in interviews. To allow stakeholders to comfortably express their opinions and to fully explore each stakeholder group's view, we chose to undertake data collection in homogenous groups, rather than mixing participants from different stakeholder groups in the same focus group. This decision was based on the strength of opinions and emotive reactions of stakeholders encountered by the authors during the preparation phase.

| Ethics and consent
The study was reviewed and approved by the Sheffield National Research Ethics Service Committee, York and Humber on July 29, 2016, reference number 16/YH/0300. All participants were given written patient information sheets, and written consent was obtained prior to the focus groups/interviews.

| Conduct of the focus groups/interviews
All selected stakeholder groups were known to have previous experience in and/or an understanding of prehospital research. Prior to each focus group/ interview, participants were given a short presentation on the overarching research project and the issues outlined in the introduction section (see Appendix B). Participants were asked if they required any further information prior to starting. This was only requested by the PPI group who considered a clear understanding of the difference between the 2 potential interventions (ALS and prehospital critical care) to be essential for the discussion.
If confusion or misunderstandings arose, particularly to the difference between randomised controlled and observational research, this was explained during the focus group/interview, where needed.
The duration of the focus group discussions was 45 to 90 minutes; the interviews lasted 30 to 60 minutes. All focus group discussions and interviews were undertaken in a semistructured fashion. 20 The same question guide was used for focus group discussions and interviews and was constructed to explore the 3 key questions underpinning this research, as outlined in the introduction section follow-up questions were increasingly directed as required. Only minimal refinement of the question guide was required over the course of the research, as well as minor adjustments to accommodate each stakeholder groups' distinct background. All focus groups and interviews were undertaken by JVVF and audio recorded. In addition, JVVF took brief field notes during the interviews/discussions. Because of budget limitations and the logistics of undertaking interviews in various locations, we were unable to provide a second person to facilitate the focus groups/interviews. To minimise potential bias from this limitation, JBR debriefed JVVF after each focus group and after completion of interviews for each stakeholder group. The debriefs also included monitoring for any effects of the data collection method. Key differences between the 2 methods are that focus groups allow for interaction between participants, while interviews ensure that each participant can fully express their opinions. 16,20 All recordings were transcribed using a professional transcription service.

| Data analysis
The field notes, recordings, and transcripts were analysed using a framework approach. 19 JVVF undertook the analysis, with support from JBR who reviewed the findings regularly. In addition, we received independent feedback on data analysis from a lay person outside the study team who was a member of the transcription service. Analysis followed a 5-step approach and was undertaken using N-Vivo software (version 11). The 5 steps were as follows: 1. Coding. We reviewed all transcripts multiple times. We used a mixture of predefined codes (deductive element), based on our previous experiences, and combined these with an open coding strategy (inductive element) to include possible unexpected but important themes.
2. Construction of a thematic framework. All codes were reviewed and arranged according to the 3 predetermined topics (research priorities, attitudes towards randomisation, and funding strategies). Within each of the 3 topics, codes were grouped into themes and subthemes which emerged during the analysis, creating an initial framework.
3. Indexing. The framework created in step 2 was systematically applied to all transcripts, while paying particular attention to any data that might not fit the framework.

4.
Charting. Data supporting the themes and subthemes was condensed and rearranged within the framework to facilitate analysis. For each topic, this was done first by case (stakeholder group), then by theme.

5.
Mapping and interpretation. We mapped the range and nature of themes as well as their interactions and relationships. We searched for underlying structures and explanations for the findings of the framework.
To data saturation, there is only limited data to base an accurate estimation on, as some of the stakeholder groups have, to our knowledge, never been researched. Given the anticipated homogeneity of views within each stakeholder group, we anticipated that views could be explored sufficiently within 1 focus group or 4 interviews for each stakeholder group. JVVF and JBR assessed whether the discussions were exhausted and/or views fully explored after each focus group/ 4 stakeholder group interviews. If we considered further focus groups or interviews with a given stakeholder group to be of potential benefit, the protocol allowed for a further round of focus groups and/or another 4 interviews per stakeholder group. The decision on whether to extend data collection in this fashion was based on a consensus between JVVF and JBR, rather than predetermined criteria.

| Data presentation
In keeping with the research questions, we will present the results according to the 3 main topics.
• Topic 1: priorities influencing prehospital research • Topic 2: randomisation of prehospital critical care for OHCA • Topic 3: funding decision making After demonstrating the contrasts between the stakeholder groups' views, the discussion section will aim to identify common underlying values through the application of concepts identified in the literature (fairness, beneficence, and scientific enquiry).

| RESULTS
In total, 23 people participated in 3 focus group discussions and 9 people participated in 8 interviews (1 interview with air ambulance charity representatives included 2 members of the same charity). See Table 1 for an overview of the demographics of the participants, according to stakeholder group.

| Main themes
• Broad support for research to improve prehospital care • Differences in stakeholders' strategies to improve prehospital care As hypothesised, views regarding the priorities influencing prehospital research were similar within each stakeholder groups but differed significantly between groups. Figure 1 provides an overview of the importance of the main priorities for each stakeholder group based on the frequency that the topic occurred, the strength of the opinions expressed (as judged by the research team during data analysis), and whether participants discussed the topic spontaneously or after prompting.
The main theme which emerged from the discussion of priorities influencing research was the consistent emphasis by all stakeholders on their support for prehospital research with the aim of improving prehospital care. While all groups agreed on the importance of improvements in prehospital care, each stakeholder group differed in their approach to this priority. Table 2 illustrates how strategies to improve prehospital care through research are determined by the   Table 3 for representative quotes from all stakeholder groups.

| Fairness and equity
Fairness was referred to frequently by stakeholder groups during the discussion of research priorities and funding decisions, but the interpretation of this concept differed between groups. The PPI group and charity representatives focused strongly on the provision of optimised prehospital care to patients with OHCA or critical illness, respectively. They essentially argued for vertical equity, whereby patients with the greatest need receive the most services. 21 Out-ofhospital cardiac arrest is an unpredictable, and often chaotic and dramatic event, with a 90% mortality rate and potential for psychological distress for survivors and witnesses. 22,23 Providing a higher level of care for OHCA than for less severe prehospital conditions could be considered a fair approach, when fairness is viewed under the principle of vertical equity. 24 A limitation to this argument is that vertical Ambulance service commissioners "I would go back to the question, though why would anybody be investing that level of money into [randomising] a critical care team in that way, on a hypothesis which doesn't seem to have much evidence behind it to even get that point?" Prehospital researchers "Well whilst [patient-level randomisation] is not unacceptable to me, I think the best way to do it is probably at a cluster level." Prehospital providers "Basically we're happy to randomise things that we think don't work. Aren't we. And we've got belief that we potentially do [improve outcomes]." PPI group "So funding is an emotional decision.
[…]Which will always be, regardless of whether you have a randomised set of data or observational." Air ambulance charities "What we require in terms of evidence is probably a lot less because we're going to be able to take that view of, well, common sense […]" Ambulance service commissioners "That you would look at the strength of evidence but you have to weigh that up against everything else, ie, the cost and what're you going to compromise in terms of other services." equity seeks to balance health resource distribution where outcomes are unnecessarily unfair, rather inevitably unequal. 25 The question in this context, therefore, becomes one of "is a 90% mortality rate from OHCA a consequence of insufficient (and unfair) resource distribution?

Prehospital researchers
Or is it an inevitable aspect of OHCA, despite adequate treatment?" The 2 stakeholder groups in this study which argued the latter were the ambulance service commissioners and prehospital researchers.

| Limitations
For logistic reasons, we used a combination of interviews and focus group discussions, and this may have influenced the results. However, the research team actively monitored for any signs of missing participant voices because of the research design, and were satisfied that the exact method of information gathering did not seem to influence the results significantly. While we would have liked to use purposive sampling strategies for all participants, this was only possible for the charity, commissioner, and researcher stakeholder groups. We did not provide participants with a detailed explanation of the counterfactual framework underpinning randomised controlled trials and causal inference. It is possible that such a description would have reduced some of the resistance that participants expressed towards randomisation. However, a full appreciation of these theories is difficult to achieve within a reasonable timeframe and would have altered the perspective of stakeholder groups participating in this research, thus limiting generalisability. Furthermore, the discussions around randomisation were largely driven by emotive or practical considerations, rather than theory. To generalisability, the study's results are based on a relatively small number of participants and we cannot fully exclude the possibility that we did not capture the full extent of views for each stakeholder group. The group of prehospital providers all worked within the same organisation, and might not be representative of other providers working under different circumstances. The PPI group's view was influenced by their personal experience, and it is, therefore, unlikely that they fully represent the general population. The question of representativeness in PPI is a largely unresolved issue. 28 Our view is that the PPI group in our study is limited in its representation of the general population, but it, nevertheless, represents a major stakeholder in OHCA. Finally, a degree of subjectivity to the data collection process, as well as analysis and presentation of data will have inevitably shaped the results of this research. We, therefore, described the underlying research paradigm in the methods section and placed our findings within the context of existing research in the discussion section. In summary, this exploratory research is by no means an exhaustive representation of all potential stakeholders' views but focuses on important mechanism which determined contrary views of 5 key stakeholder groups.

| CONCLUSION
Analysis of the views of 5 stakeholder groups regarding research and the funding of prehospital critical care for OHCA revealed shared values, but a variety of different strategies to achieve these. The results of this exploratory research can help researchers in similar fields in the planning and presentation of their research, to maximise benefits of stakeholder engagement on decision-making.