Establishing the research priorities of paediatric emergency medicine clinicians in the UK and Ireland

Objective Paediatric Emergency Research in the UK and Ireland (PERUKI) is a collaborative clinical studies group established in August 2012. It consists of a network of 43 centres from England, Ireland, Northern Ireland, Scotland and Wales, and aims to improve the emergency care of children through the performance of robust collaborative multicentre research within emergency departments. A study was conducted regarding the research priorities of PERUKI, to establish the research agenda for paediatric emergency medicine in the UK and Ireland. Methods A two-stage modified Delphi survey was conducted of PERUKI members via an online survey platform. Stage 1 allowed each member to submit up to 12 individual questions that they identified as priorities for future research. In stage 2, the shortlisted questions were each rated on a seven-point Likert scale of relative importance. Participants Members of PERUKI, including clinical specialists, academics, trainees and research nurses. Results Stage 1 surveys were submitted by 46/91 PERUKI members (51%). A total of 249 research questions were generated and, following the removal of duplicate questions and shortlisting, 60 questions were carried forward for stage 2 ranking. Stage 2 survey responses were submitted by 58/95 members (61%). For the 60 research questions that were rated, the mean score of ‘relative degree of importance’ was 4.70 (range 3.36–5.62, SD 0.55). After ranking, the top 10 research priorities included questions on biomarkers for serious bacterial illness, major trauma, intravenous bronchodilators for asthma and decision rules for fever with petechiae, head injury and atraumatic limp. Conclusions Research priorities of PERUKI members have been identified. By sharing these results with clinicians, academics and funding bodies, future research efforts can be focused to the areas of greatest need.


INTRODUCTION
The volume of children attending emergency departments (EDs) with presentations encompassing the full spectrum of illnesses and injuries should create an ideal environment in which to perform research. Such opportunities must not be wasted, particularly given that parents/carers generally have a positive attitude to the potential recruitment of their children into clinical trials. 1 However, research in paediatric emergency medicine (PEM) brings with it the challenges of both the impediments of the ED clinical environment (activity, unpredictability, noise, time-critical patient management) and the limitations of research involving children. 2 Owing to the complexities of organising collaborative work, there has been infrequent use of the significant numbers of paediatric attendances to EDs within the UK and Ireland to conduct research that can provide answers to important clinical questions. As a result, the majority of PEM studies within these nations have traditionally been performed in a small number of institutions, often single centre in nature.
In contrast, other countries have had more success in conducting multicentre PEM research, largely thanks to the formation of PEM research networks. These include the Pediatric Emergency Medicine Collaborative Research Committee and the Pediatric Emergency Care Applied Research Network (PEMCRC and PECARN, USA), Paediatric Emergency Research Canada (PERC), Paediatric Research in Emergency Departments International Collaborative (PREDICT, Australia/ New Zealand) and Research in European Paediatric Emergency Medicine (REPEM). The existence of these networks led to the creation of Pediatric Emergency Research Networks (PERN), a research initiative formed with the vision of answering globally relevant PEM research questions. 3 Paediatric Emergency Research in the UK and Ireland (PERUKI) was established in August 2012 as a collaborative clinical studies group among individuals passionate about providing high-quality PEM research. 4 From its inception, many lessons were learned from the principles adhered to by other PEM research networks, with the aims of overcoming the inherent challenges of conducting Soon after PERUKI was established, it was recognised that there was an urgent need to determine the research priorities of its members in order to set the research agenda for PEM in the UK and Ireland. A prioritisation exercise was therefore conducted using a modified Delphi technique. This paper describes the methodology of that process and reports the research priorities that were identified.

METHODS
The work consisted of an initial survey in which all PERUKI members were invited to submit research questions. This bank of questions was then refined and resultant questions were distributed to the membership for priority ranking. This modified Delphi process has been used by other groups of clinicians for the purpose of establishing research priorities. 6 Classic Delphi methodology involves two or more survey rounds, with each participant receiving the replies of each participant, before being asked to reconsider their own answers, with the aim of achieving consensus. 7 The modified Delphi process used in this study relied on a small group to extract specific submissions from the initial survey round.
Stage 1A: survey to identify important research questions An online survey was composed using Bristol Online Surveys. This survey consisted of a single question: "Thinking about your clinical practice in the field of paediatric emergency medicine, what are the most important research questions which need addressing?" Each respondent could enter up to 12 research questions and, for each, the respondent was required to categorise the question from a pre-populated list of common topic areas. This approach had two purposes: (i) it aided with subsequent results analysis; and (ii) by providing a list of themes (ranging from analgesia and sedation, to gastroenterology, to education and training), this stimulated research ideas. Respondents were asked, where possible, to submit questions in standard 'PICO' ( population, intervention, control, outcome) format, and an example question was provided that illustrated this structure.
This survey was open for 3 weeks. At study commencement, 87 individuals were registered as members of the PERUKI network, with a further 4 clinicians registering during the period that the study was open, and these new members were also invited to undertake the survey (total distribution=91 members). Reminders and progress updates were sent by email 1 and 2 weeks after opening the online survey, and 24 h prior to its closing. Respondents were given the option to either respond anonymously or provide their email address within their submission. This ensured that (i) an open environment was created for sharing of ideas and (ii) further clarification could be sought where necessary for any proposed research questions.

Stage 1B: refinement of research questions
A list of all unique research questions was compiled, grouped according to different topic areas. These questions were reviewed, discussed and refined by the PERUKI Executive Committee, generating a shortlist of research questions for ranking in the second stage. A question was considered eligible if it had not already been answered within the scientific literature and if the question leant itself to multicentre research within the ED setting. Some questions, felt to be ineligible owing to an established evidence base, were identified for the future development of review articles to address knowledge gaps. In some cases, similar or related research questions were merged into a single research question that was carried forward to the second stage survey.

Stage 2A: ranking of research questions
The second stage survey was conducted using the same online survey platform. This survey was sent via email to all PERUKI members, which by this time had increased to a total of 95 individuals, irrespective of whether they had submitted responses to the stage 1 survey. Within the stage 2 survey, members were asked to review each of the shortlisted research questions and rate each of them on a seven-point Likert scale ('not a priority' to 'essential priority'), based on the importance of the question to their own clinical practice in PEM. The survey remained open for a period of 3 weeks. Reminders and progress updates were sent by email after 1 and 2 weeks of the survey opening, and 24 h prior to the survey closing.

Stage 2B: analysis and prioritisation
The questions considered in stage 2 were then ranked according to the total priority score.

Stage 1 survey
Completed surveys were submitted by 46/91 members (51% response rate). A total of 249 research questions were generated, representing a mean of 5.4 questions (median 5 questions, range 1-12 questions, SD 3.23) per respondent. Following the removal of duplicate questions, 206 unique research questions were available for further refinement. These included questions from 22 topic areas-the topics that yielded the most questions were analgesia and sedation, service planning, minor injuries (including minor head injuries), respiratory problems and major trauma (table 1).
43% (88/206) were deemed eligible questions to be carried forward to the second stage. There was some overlap between questions. These were merged and reworded to produce a final shortlist of 60 research questions. Table 1 shows the number of research questions, by topic, generated in stage 1A and subsequently shortlisted for ranking. Of the remainder, some were excluded because they had already been answered in the literature (42/206, 20%), for example, questions about the effectiveness of removable wrist splints for immobilisation of distal forearm buckle fractures. 8 Ten topics were identified for the future development of review articles, when there was an apparent knowledge gap among multiple survey responders, or where the proposed question did not lend itself to primary research. Other proposed research questions (76/206, 37%) were not felt to be amenable for study in the environment of EDs, or else were felt to be more applicable to research by other specialty areas. In some cases (eg, research relating to education theory), it was acknowledged that PERUKI alone might not be able to answer the proposed questions, but in future a working group within PERUKI with a particular interest could collaborate with external groups with specialist knowledge.

Stage 2 survey
Stage 2 surveys were submitted by 58/95 PERUKI members (61% response rate). For the 60 research questions that were rated, the mean score of 'relative degree of importance' was 4.70 (range 3.36-5.62, SD 0.55). Table 2 lists the top 20 ranked by total priority score.
The full list of 60 research priorities is available (see online supplementary appendix 1).
Respondents were also given the opportunity to provide free text comments about the research questions and the survey. Some PERUKI members remarked that some of the research questions were less applicable to certain geographical regions or to the patient population of individual EDs.

DISCUSSION
This is the first study to establish the research priorities of PEM clinicians in the UK and Ireland. The top 10 research priorities include questions on biomarkers for serious bacterial illness, major trauma, intravenous bronchodilators for asthma and decision rules for fever with petechiae, head injury and atraumatic limp. The results likely reflect clinical dilemmas or areas in which evidence is lacking, which PERUKI members face in their day-to-day practice. They are deemed as priority areas by respondents either due to the frequency with which they occur or due to the severity of disease burden.
The study was performed soon after PERUKI was established for several important reasons, including the observation that it had been key for other research networks, such as PECARN, to perform this type of exercise, so establishing and publicising their research focus within their national setting. 9 It is interesting to note that there are many similarities between the priorities identified by the PERUKI and PECARN networks. For example, respiratory illnesses/asthma, clinical prediction rules for highstakes/low-likelihood diseases, and treatment of infectious diseases were identified as top 10 priorities by both networks.
Setting the research agenda for PEM in the UK and Ireland was also important for PERUKI to support enthusiastic researchers and assist them in conducting studies rather than them branching out in multiple directions with little coordination or not having a long-term strategy, a phenomenon that has the potential to occur following the establishment of any research network or collaborative. From the outset, PERUKI has aimed to give clarity to its projects and to the specific research questions within themes, such that clinicians and researchers could select appropriate studies that could be supported through its network of EDs. This support could be provided at a number of different stages, as outlined in table 3.
As a new network, PERUKI is working to identify its capacity for the number and the complexity of studies that will be feasible. Early work, including surveys and observational studies, 10-13 has stress-tested the network to a degree and PERUKI has successfully collaborated with other international PEM networks. In the future, each member site is likely to encounter its own resource and capacity issues, and the overall capacity of PERUKI is not yet understood. By focusing on the high-priority research questions identified in this study, the network can ensure that the most important work is addressed before it reaches limits of capacity. In addition, it is hoped that by sharing and publishing the research priorities of PERUKI, this will lead to collaboration with other networks. There is considerable enthusiasm for research within the PERUKI network, and a wide range of research questions have been proposed for future studies. Some key themes were evident from the final list of priorities. Unsurprisingly, clinical conditions that present frequently to EDs, such as febrile illness, asthma and major/minor trauma, were well represented in the final results list. Several research questions related to a perceived requirement to validate decision rules that are in common practice, such as the CHALICE rule for head CT scans, as contained within the National Institute for Health and Care Excellence (NICE) head injury guideline. 14 There was recognition that the creation of decision rules for other presenting complaints, such as atraumatic limp and petechial rash, would assist clinicians and reduce practice variation. There was also an appreciation that there are several themes that require considerable research, but that the first step should be to establish current practices within EDs in the UK and Ireland.
Some of these research questions can be addressed for relatively little funding, generating clinically important results, and the prospect of early translation to practice and improvement in clinical care. In contrast, other studies will require a long-term approach and access to competitively sought funding streams.
Workstreams have been established for clinical topics in which several research questions have been prioritised, so that future research projects are conducted with an underlying strategy in a logical and coordinated manner. Within these workstreams, knowledge gaps can be filled by research using various study types including evidence synthesis, reviews, retrospective and prospective observational work and, ultimately, interventional work. Current PERUKI workstreams include the Childhood Acute Asthma Research Program (CAARP), 10 11 a working group appraising and developing prehospital paediatric trauma triage tools, 12 13 and a further workstream establishing best practice in paediatric procedural sedation throughout the UK and Ireland.
A major strength of the prioritisation exercise was that the PERUKI members who responded to the surveys represent a variety of health practitioner roles and department types and sizes, and the vast majority of membership is clinical rather than academic.

Table 3
Examples of the support that a research network, such as PERUKI, may provide 1 Developing and refining the research question 2 Developing and/or reviewing the study protocol 3 Ensuring that study protocols are practicable and achievable for delivery, with due consideration of both the paediatric population and the ED environment 4 Assisting with grant applications and providing endorsement for studies 5 Delivering the research. This may be through the full PERUKI network of 43 member sites, or a subgroup of our EDs, selected according to the needs of the study, the target population and/or the research interests of the site leads 6 Assisting with the writeup of papers 7 Dissemination of the final results 8 Ensuring that there is translation of high-quality evidence into practice change throughout the PERUKI sites and beyond ED, emergency department; PERUKI, Paediatric Emergency Research in the UK and Ireland. In children with possible major trauma, which predictor variables identify serious injury requiring direct transport to a major trauma centre? 5.60 3 In children with septic shock, does aggressive fluid management, as opposed to judicious fluid management, improve mortality? 'That is, a response to FEAST study results in a UK population'

5.57
4 In children with acute severe asthma requiring intravenous therapy, is salbutamol, aminophylline, magnesium or a combination of these superior in safety, and clinical and cost effectiveness? 5.53

5=
In paediatric major trauma patients with major haemorrhage, does intravenous tranexamic acid compared with no treatment reduce mortality and morbidity? 5.50

5=
In children with c-spine injury, does currently available guidance provide satisfactory performance accuracy in identifying significant injuries? 5.50 7 In children with atraumatic limp (or possible orthopaedic sepsis), what is the best clinical decision rule for observation/investigation/management? 5.34 8 In PERUKI membership is non-restrictive, and any individual with enthusiasm for PEM research is entitled to join the network, regardless of their previous level of research experience. They therefore form a select, but representative, sample of the full range of PEM practitioners within the UK and Ireland. The chosen methodology allowed an initial phase of unrestricted idea formation before a second phase of review and scoring using a Likert scale. There were some limitations to the study. The survey was restricted to its own members who, by definition, have a selfdeclared interest in research. This subgroup of practitioners might be subject to bias because of their own research agenda, and it is possible that a process that was open to all PEM clinicians within the UK and Ireland might have generated different results. Furthermore, additional weight would have been conferred to the prioritisation results if other stakeholders, particularly parents/ carers, had been consulted at some point in the study.
The 51% response rate to the stage 1 survey was less than had been hoped for, although this response is generally accepted as adequate. The majority of the study was conducted using an online platform rather than face-to-face. However, this method allowed for the generation of a large number of questions from a broad range of professional types across the geographical areas that PERUKI incorporates. Finally, the refinement of the questions was conducted by a small group of individuals, who are subject to potential bias. It is also possible that this group might have missed subtle nuances of some of the stage 1 questions. However, efforts were taken to ensure that, where clarity was required, further detail was obtained from the person who submitted the question.
By distributing and sharing the results of this prioritisation exercise, it is hoped that they will spark interest and encourage any clinician to develop a research proposal and use the PERUKI infrastructure to develop a collaborative study. Trainees have been included in this information sharing in order to help inform, focus and motivate them to become involved in PEM research from a relatively early stage in their careers. When applying for research funding, the presence of a research question on the prioritisation list can be used to add weight to the submission, as such questions clearly demonstrate importance to clinicians working in PEM.
The results are also being shared directly with funding bodies, many of whom are actively seeking studies that are identified as priorities within a specific clinical field, particularly when clinicians and researchers within the field have identified their enthusiasm for participating in those studies designed to answer these research questions.

CONCLUSION
The research priorities of PEM clinicians in the UK and Ireland have been identified. These results serve as a catalyst for future collaboration in the development of the studies of greatest impact for PEM in our setting. As the identified research questions encompass several themes and are underpinned by varying levels of existing evidence, they require allocation to separate distinct workstreams and funding programmes. It is acknowledged that these results have been developed by just one stakeholder group and, therefore, should not be viewed as the sole research agenda for PEM, with no consideration for alternative research questions, such as those recommendations from the NICE guideline development groups. 15 Nevertheless, it is the hope that, by sharing this prioritisation exercise with clinicians, academics and funding bodies, there will be a tangible benefit in the quantity and quality of multicentre PEM research in the UK and Ireland, with a clear focus on studies that will bring about significant improvement in the emergency care of children.