Background

Conducting clinical research involving critically ill children and young people is vital because paediatric critical care (PCC) is a high-cost, resource-intensive environment with a sparse evidence base [1]. Much of clinical practice within PCC is not supported by high-quality evidence and practice surrounding even commonly used interventions can vary widely between units and individual intensivists [2]. With recognition that research is a vital component of a high quality service [3], there is an aspiration that research should be the standard of care for all critically ill children and their families [4].

A key element to overcoming these challenges has been the establishment of national and international research networks to promote a collaborative approach to the successful design and conduct of Randomized Controlled Trials (RCTs) in PCC [5]. The Pediatric Critical Care Society (PCCS) is a professional, multi-disciplinary, membership organisation, representing the interests of those delivering pediatric critical care in the UK and Republic of Ireland [6]. The Society has developed UK quality standards which highlight the importance of research in service provision and established a research study group (PCCS-SG) [7]. PCCS-SG aims to improve the care of critically ill children and young people through the conduct of high-quality, multi-centre research studies within the UK and the Republic of Ireland (RoI) [5]. This coordinated approach is vital to develop methodologically robust trials to inform evidence-based guidelines within the PCC setting [8]. Currently there are a number of multi-centre RCTs in progress, aiming to recruit large number of patients. Oversight from PCCS-SG has meant these have been planned to run sequentially to avoid competition or there has been inter-trial collaboration to minimise burden associated with recruitment and consent. Research prioritisation is therefore an important function of the PCCS-SG [5], to ensure that the perspectives of healthcare professionals (HCP), children and young people and parent/carers are taken into account.

In 2018 the first formal PCCS-SG led research prioritisation exercise was conducted with the aim to identify and agree research priorities in PCC in the United Kingdom, both from a HCP and parent/caregiver perspective [9]. The exercise examined and compared HCPs and families’ views around research priorities in PCC and was disseminated to the PCC community through PCCS-SG meetings, presentation at the national PCCS research conference and publication. Following on from this, several studies have been or are being undertaken, have secured funding or are being developed as future trials.

The COVID-19 pandemic had a significant impact on biomedical research around the world with a shift towards COVID and virology related research questions. There was also a reprioritisation of research staff and resources which affected research activity and delivery in other areas, with potentially disproportionately higher impact on pediatric research. Post-pandemic effects related to economic challenges may also have longer lasting impact on research funding and strategies are being developed to focus efforts on research recovery [10,11,12,13].

It is therefore timely to revisit and take stock of our current priorities post-pandemic and five years on. The aim of this study was to establish the research priorities of healthcare professionals working within UK PCCs in 2022 and to describe the extent to which they had changed since 2018.

Methods

A modified three-round e-Delphi study [14, 15] of PCCS members was undertaken, followed by the more rigorous Hanlon method of prioritisation [16], to generate an updated consensus around priority research areas for pediatric critical care in the UK. The three rounds of e-Delphi survey were completed between February and May 2022 and the Hanlon scoring was carried out in October 2022. This study is reported according to the Guidance on Conducting and Reporting Delphi Studies (CREDES) checklist for e-Delphi studies [17].

All PCCS members (n = 1200) were invited to complete the e-Delphi survey and two reminders were sent for each round at weekly intervals. The time between rounds was six weeks to reduce attrition. In round one (R1), respondents were asked to list up to three research topics or questions they deemed as key priorities for the specialty over the next five years using SurveyMonkey (SurveyMonkey, Mateo, CA). Respondents were encouraged to be specific in suggesting a study question and outcomes, considering the relevance to patient care. Participant demographics were also collected. There was an option to provide respondents’ email address to allow for any clarification of their question or topic. Simple content analysis was undertaken to classify responses from R1 into topic domains by three study team members (KM, LT, SR), following the analysis method [18]. This involved collapsing the suggestions into key domains and merging duplicate items; this was undertaken independently, and discrepancies resolved through an online discussion by the reviewers. Special consideration was given to retain original language used to frame questions by the respondents and duplicates were removed.

In round two (R2), topics and domains were sent to participants, who were asked to rate each topic on a scale between one (not an important topic) and six (the most important research topic). Optional free text comments were permissible. To generate topics for R3, we followed a two-step process: (a) selecting questions that scored more than the population mean, and (b) excluded questions with a score above the mean but that had a proportions of scores of 5 (‘very important’) or 6 (‘most important’) lower than any with score below the mean. This therefore accounted for both the average score for each question and the distribution of scores and addressed clustering around the central tendency (supplement 1). Free text comments were also reviewed by the study team to identify any topics that were already being studied and gain further insight into responses.

The R3 survey was sent back to participants with the group mean score and proportion ranked five or six, to re-rank in light of the group score. The same rating scale with one to six, was used. Participants were also asked to prioritise domains identified after the first round in the same way as the individual questions. This was carried out to consider developing research questions in particular areas for future prioritisation exercises.

Following this, the top 20 topics (by mean scores) were then converted into more structured research questions using a PICO format (Population, Intervention, Control, Outcomes) by members of the study team (SR, LT, PR). Members of PCCS-SG were invited to discuss and prioritise these topics during a consensus webinar according to the (i) size of the problem, (ii) seriousness of the issue, and (iii) feasibility of answering the research question. This session was moderated by an independent facilitator (LS) who was not a member of the PCCS and not involved in the design, conduction, or analysis of the e-Delphi study. Live online voting took place using Surveymonkey (SurveyMonkey, Mateo, CA) in this consensus webinar. The Hanlon prioritisation method [16] was then used to calculate Hanlon scores from the responses and the top ten research questions were identified.

Data analysis

Survey data were imported, validated, and analysed in Microsoft Excel (Microsoft Corporation, Redmond, WA), and STATA version 16 [19] was used for further analysis. During the consensus webinar each topic was scored on a scale from zero through ten on the pre-defined criteria. Priority scores were calculated using the following formula: D = [A + (2 × B)] x C, where: D = Priority Score, A = Size or prevalence of the issues, B = Seriousness of the issue, and C = feasibility of the study [16].

The survey was categorised according to the U.K. Health Research Authority as staff research and did not require formal NHS ethical approval. It was formally approved by PCCS-SG, who sent this out to the members of the society. Consent was implied by the return of the survey.

Results

Eighty-five respondents submitted 247 research topics or statements during R1 (Fig. 1). Respondents were physicians (62%), nurses (27%) followed by other healthcare professionals, including dieticians (5%), advanced clinical practitioners (3%), and physiotherapists (2%). The mean (standard deviation (SD)) years of PCC experience were 14.3 (8.7) and 94% of the respondents were based in the UK. Once duplicates were removed, 135 topics were categorised into 12 domains (Table 1).

Fig. 1
figure 1

Flowchart of Delphi process with number of respondents and topics put forward in each round

Table 1 Categories identified after round 1 and number of topics within each category

One hundred and twelve participants scored the 135 topics in R2. After this 45 of the 135 topics scored high enough priority to be entered into the R3 survey. Sixty-seven participants scored these in R3, and the top 20 topics were used for Hanlon prioritisation, based on the highest mean scores focusing on importance of the topic. Respondents were also asked to prioritise the key domains identified after R1, and five domains received a mean score higher than 4/5 with ‘improving outcomes after PICU’ receiving the highest score (mean = 4.75) followed by ‘pain and sedation’ (mean = 4.25), Monitoring, diagnostics, technology, and artificial intelligence (mean = 4.17), staffing and well-being (mean = 4.11), and ethical issues (4.03).

A list of top ten priority research questions was generated according to the Hanlon calculation (Supplement 2) which was attended by 21 members of PCCS-SG. These participants included established academics and clinicians with active involvement or interest in research in pediatric critical care.

Discussion

In this study, we not only described current research priorities for pediatric critical care according to HCPs in the UK in 2023, but also explored if there has been a shift in key research areas or topics over the last five years. The response rates were similar to the previous exercise conducted in 2018. This might reflect the small number of clinicians being academically active and the need for a more inclusive research culture in paediatrics as has been noted in previous studies [20]. The respondents were however representative of the wider MDT in PCC and included doctors, nurses and allied health professionals. In comparison to 2018, HCPs prioritised topics related to early rationalisation and a more targeted approach for antimicrobial therapy, especially related to respiratory infections (Supplement 3). There was also greater emphasis on involvement of families and patients to identify outcomes that are important to them and incorporating those into further clinical research. This was considered particularly important for longer term outcomes, including for children with traumatic brain injuries. A new priority topic that emerged was related to improving how research evidence is incorporated into daily clinical practice. Implementation science has become an increasingly important field of science as more RCTS are undertaken, yet the results not always implemented into practice [21] contributing to research waste.

Some topics were consistent in remaining a priority since 2018, such as interventions to improve staff morale and retention, optimal sedation strategies and intravenous fluid management. The highest priority for the next five years, consistent with the 2018 exercise, was research to inform the processes related to communication and decision-making regarding withholding or withdrawing life-sustaining treatment, which reflects a growing number of admissions to PCC for children with complex (often life-limiting) medical conditions [22]. One topic that increased in priority was Registered nurse (RN) staffing levels and optimal RN:patient ratios, with the need to study new approaches for delivery of nursing care in PCC, increasing from 10th priority by HCPs in 2018, to fifth in 2022. This is likely reflective of an increasing demand for PCC capacity and staffing levels across the country [23]. Workforce modelling is an area currently being explored within the adult ICU [24] but no studies are currently exploring this within the PCC context.

The names for research domains identified in this study were derived from the language used by respondents to phrase research topics. However, priority domains identified in 2022 based were similar to those in 2018, with the introduction of the use of artificial intelligence as a new research area in PCC (Table 2).

Table 2 Comparison of topic domains identified after round 1 in 2018 vs 2022

In comparison to other PICU research prioritisation exercises undertaken internationally, UK research priorities are similar to those identified by the Australian and New Zealand Intensive Care Society Paediatric Study Group (ANZICS PSG), with an emphasis on long term outcomes, fluid therapy, sedation strategies, antimicrobial decision making, staff retention and psychological support for families [25]. We also noted some similarities in research domains identified from a survey of clinician-researchers from low- and middle-income countries, including ventilation, nutrition and ethics in paediatric critical care as well as a need for building capacity [26] (Table 3). This highlights the potential for multi-national collaborative studies and a need for developing stronger collaborations between paediatric critical care societies to address these research topics.

Table 3 Research priorities for pediatric critical care in the last 5 years – an international comparison

Research in pediatric critical care has unique challenges such as a complex and relatively small patient population [1] and challenges related to parental involvement [27], on top of cultural and systematic barriers related to pediatric research in the UK [28]. This has resulted in historically limited numbers of RCTs in PICU [29]. Encouragingly, a number of research topics that were prioritised in 2018 have, or are being, addressed in current studies and trials. These include the OxyPICU trial [30] that compares liberal vs conservative peripheral oxygen saturation targets for children in PICU, the Enhance study, that aims to identify and investigate different models of providing end of life care for pediatric patients [31], and the SWell study that aims to investigate wellbeing interventions for PICU [32, 33]. Large multi-centre trials such as First ABC and SANDWICH have also studied ventilation and sedation strategies in PICU [34, 35]. Despite this, many of the key questions still need to be addressed, preferably in multicentre trials, which will be resource intensive.

In addition to existing resource limitations, research delivery in the UK was heavily affected by the COVID-19 pandemic [10,11,12], with research recovery and reset still requiring attention three years on [13]. It is crucial that new and innovative approaches are considered, including platform and adaptive trial designs to address these key areas and challenges related to co-enrolment within finite resources and existing infrastructure [5].

This study has some limitations that warrant mentioning; although parents and families of children of PCC patients were involved in the research prioritisation in 2018, we were unable to involve parents during this updated prioritisation study. This was not possible due to resource limitations and challenges with access to families during COVID-19 pandemic [36, 37]. The study was also limited by a low survey response rate (especially in R1). This was however, higher than the previous Delphi study in 2018. Although we collected data regarding length of experience in PCC, we did not collect demographic data for survey respondents such as sex, age and ethnicity. The percentage of nurses and allied health professionals who responded remained low compared to the ANZICS PSG study [19] (27% vs 59%) which may have impacted on the results. There are recognised challenges to research engagement of nursing and allied health professionals (AHPs) which are the focus of national strategies (CNO 2021, NHSE 2022). Within the PICU specialty we hope to specifically target this with targeted support, education and training for nurses and AHPs from the recently funded NIHR Paediatric Critical Care Incubator [38,39,40]. Finally, the Hanlon scoring only involved 21 HCP and this again may have impacted on the findings. Despite these, one strength of this study was the addition of this extra Hanlon scoring system with input from active academics and clinicians with interest in research in the field of pediatric critical care in the UK.

Conclusions

In the UK, research priorities for pediatric critical care have shifted over the last five years with more emphasis on complex decision-making in end-of-life care, rapid diagnostics and antimicrobial therapy, innovative strategies towards RN staffing ratios and studying the emerging potential for and impact of artificial intelligence in this field. Other research questions related to fluid therapy and sedative agents remain a priority and need to be addressed. The similarities between research areas of interest identified from prioritisation exercises conducted in different countries highlights the importance of multi-national collaboration for further research in pediatric critical care.