Background

Injuries in children and youth range from minor inconveniences to major trauma, can limit the normal activities of daily living and have been recognized as an important health problem [1]. Common causes of injury to children and youth include falls, transport incidents, self-harm, struck by an object during sports, unintentional poisoning and violence [2]. As injury is the leading cause of death and hospitalization among young people in Canada, foundational epidemiological information is of great value to inform the development and targeted implementation of injury prevention and health promotion initiatives [1, 3].

One of the only cross-national health surveys that includes information on injury occurrence and prevention within adolescent populations is the Health Behaviour in School-aged Children Survey (HBSC) [4]. HBSC is an international survey conducted every four years in 43 countries that informs understanding of the behaviours and attitudes of youth ages 11–15 years, and the factors that impact their health [5]. Canada conducted its sixth survey cycle in 2010, with 26,078 students participating from 436 schools located in eight provinces and three territories. HBSC provides a rare and comprehensive glimpse into health problems, including injury, in the early adolescent years in Canada.

In the fall of 2011, members of the Canadian Institutes of Health Research (CIHR) Team in Child and Youth Injury Prevention, academic researchers associated with HBSC Canada, and the Public Health Agency of Canada met to collaborate on the development of a detailed plan for a Canadian youth injury report using HBSC data. Editorial teams were established, analyses were conducted and the report, Injury among Young Canadians: A national study of contextual determinants, was drafted [6]. The goal of the report was to use Canadian specific data from the 2009/2010 cycle of the HBSC to report nationally representative findings about injury occurrence, and the determinants and consequences of injury, particularly as these relate to the contexts where young people live, learn and play [6]. This was completed in order to support critical evidence-based initiatives to prevent injuries in this population. The development of specific recommendations was identified as a means to ascertain critical evidence-based actions based on the key insights illuminated by the report (Table 1). We approached this in a systematic and scientifically sound manner, using a modified-Delphi process. Delphi is a research tool for obtaining the judgment of a panel of independent experts on a specific topic [7], which facilitates a group communication process dealing with a complex problem [8]. The purpose of this component of the project was therefore, to develop specific policy oriented recommendations arising out of the report using a modified-Delphi process [7] administered to a panel of expert stakeholders.

Table 1 Key Findings from Injury Among Young Canadians: A national study of contextual determinants
Full size table

Methods

The national report had seven core chapters examining risk factors for injury, particularly within home, school, neighbourhood and peer-group contexts. Prior to the first round of the modified-Delphi process, the 22 different chapter authors of the report were asked to suggest 3–7 draft recommendations for each of their chapters. The draft recommendations intentionally reflected both positive and negative findings, the lesson(s) that could be learned and suggested actions to be taken. Upon receipt of 37 draft recommendations from chapter authors, the research team removed duplicates and consolidated similar recommendations, reducing the number to 27. The 5 editors of the report are the main authors of this manuscript and constituted the research team. The draft recommendations were subsequently edited to ensure consistency of language and style so that they were written as policy-oriented action statements directed at the appropriate group to take the recommended action(s). This set of draft recommendations formed the basis for expert review within the modified-Delphi process. The University of British Columbia/Children’s & Women’s Hospital Research Ethics Board discussed this project and conveyed that ethics approval was not required as the project goal was simply to document the process of developing recommendations with those who were considered colleagues of the research team.

Modified-Delphi method

The Delphi process is based on the assumption that group judgments are more valid than individual judgments. Delphi is an appropriate technique when: 1) the problem being addressed does not lend itself to precise analytical techniques and would benefit from collective expert opinion, 2) the required experts may not have a history of communication and/or collaboration, 3) logistics do not support frequent or face-to-face meetings, 4) disagreements among experts may require mediation and anonymity and 5) variety of expert opinion has to be preserved and provided as feedback in the iterative process, avoiding domination by any one opinion [8].

For the purpose of developing specific recommendations in this study, a modified-Delphi method was chosen over traditional survey methods. This implied that: 1) experts were to be selected based on their unique ability to provide informed responses focused on the development of recommendations and 2) agreement was to be arrived at through the use of controlled and anonymous feedback provided by the facilitator during three rounds of review and feedback [7]. The Delphi process was modified here to be restricted to three iterative rounds of expert input, seeking ratings and comments using FluidSurveys™ [9] online survey software during the first round and subsequent input via email responses during the second and third rounds. The research team served in the role of facilitator, undertaking the synthesis between rounds. Mirroring the process used by Green et al. [10], the process of synthesis included discussion among the facilitator members, exploring all expert opinions, disagreements and suggestions for change, before synthesized recommendations were drafted for each subsequent round. The modified-Delphi process in this study was an iterative process aimed at agreement on a suite of evidence-based recommendations to support a preferred future of injury prevention for children and youth. The process was completed during the period May to September, 2013 (Fig. 1).

Fig. 1
figure 1

3 stage modified-Delphi process to develop report recommendations

Full size image

Establishing the panel of expert stakeholders

An expert is defined as one who is considered to be knowledgeable about the subject under consideration and capable of representing the views of his or her peers [11]. As the report was national in scope, it was important to assemble a panel of injury prevention expert stakeholders who could represent all Canadian regions [12], though input from recognized international colleagues was also desired. In addition to geographic location, criteria for consideration included professional position and experience with youth injury prevention. The research team brainstormed a list of potential experts with positions in government, injury prevention organizations, public health, academic research, educational systems and youth-serving organizations as the youth injury report indicated required action and support by these groups. The injury prevention community in Canada is relatively small and stable, with many professionals having more than 15 years of experience in the field, therefore it was not difficult to brainstorm a list. Several Delphi studies recommend using 10–18 carefully selected expert respondents, enough to provide a range of opinions but also few enough for the research team to be able to summarize and integrate those opinions [13–16]. Thirteen experts were identified by the research team based on the stated criteria, including knowledge user stakeholders and researchers from the Alberta Centre for Child, Family and Community Research, Vancouver Island Health Authority, Saskatchewan Prevention Institute, IWK Health Centre (Nova Scotia), Simon Fraser University (British Columbia), Parachute Canada (National), Alberta Centre for Injury Control and Research, United States Centre for Disease Control, and Johns Hopkins Center for Injury Research and Policy (Maryland, USA). Experts each had between 15–36 years of experience in the injury prevention field and 10–25 years of experience in senior leadership positions that included a focus on knowledge translation. Of the 5 researchers identified, a combined list of their publications related to injury prevention program implementation, evaluation and policy analysis totaled 657. It was recognized, however, in providing their input that each injury expert was to act as an individual, with views that strictly speaking were unattached to any organizational position or policy. These individuals were not any of the 22 report chapter authors.

The research team recognized the need to ensure that youth contributed to the development of the recommendations. Therefore, the number of expert panelists was expanded in the third and final round to include 3 members of Parachute Canada’s Youth Advisory Team [17]. The Youth Advisory Team is a group of young Canadians aged 15–25, who are passionate about the issue of injury in Canada and who contribute to special projects and consultations.

Conducting the modified-Delphi process

The nominated Principal Investigator of the CIHR Team in Child and Youth Injury Prevention extended personal invitations to the 13 potential expert panelists by email to participate in the development of the action-oriented specific injury prevention recommendations, and to inform them of the process and timeline [18, 19]. Nine individuals from injury prevention organizations, medical health centres, schools of public health and academic institutions agreed to participate.

Round 1

The goal of the first round of the modified-Delphi was to decrease the number of recommendations to a core group considered to be most important, and to improve the wording so that each recommendation would resonate with the group(s) responsible to lead or take the action. Experts were sent a survey using FluidSurveys™ [9] that provided the list of recommendations and a link to the report Injury among Young Canadians: A national study of contextual determinants [6]. Panelists were asked to respond to one open-ended [20] and two closed-ended questions for each of the recommendations: 1) Will this recommendation resonate with the group(s) responsible to lead or take the action (yes/no), 2) Do you see ways to improve the strength of the language, and if so, please re-write or add comments, and 3) Using a 3-point scale (extremely important, very important, or moderately important), how important is this recommendation.

When experts advised that a recommendation be deleted, merged with another or placed as content elsewhere in the report, the changes were made by the project facilitator. In a publication on the Delphi method, Okoli and Pawlowski, 2004, suggest retaining items selected by 50 % of the expert panelists [16]. This study was able to achieve a greater percentage of agreement and retained recommendations if 66 % of experts agreed that they would resonate with the group(s) responsible to lead or take the action. When comments conflicted, the research team acted as facilitator and used a process of discussion to reach a mutually satisfactory decision. Descriptive text that provided a rationale for each recommendation based on report findings was added in preparation for Round 2 of the modified-Delphi focused on the remaining 19 recommendations.

Round 2

The goal of the second round was to generate further expert opinion and feedback on the list of 19 recommendations resulting from Round 1. The same panel of experts was asked to use their expertise and background in child and youth injury prevention to consider the recommendations in light of the youth injury report. Using the ‘review’ and ‘track changes’ tool in the Microsoft Word software package, experts provided edits and comments to the recommendations, and submitted their edited version to the facilitator. All experts’ comments and edits were incorporated wherever possible. Where there were conflicting comments, they were reviewed by the facilitator and decisions were made on how to ensure that all comments were synthesized as part of the recommendations for inclusion in Round 3. All decisions and edits were documented and communicated back to the expert panel as part of the Round 3 recommendations that members could make final comment on.

Round 3

The goal of the third and final round was to finalize the wording of the recommendations. In addition to the same 9 expert panelists, members of the Parachute Youth Advisory Team were asked to submit their edits and comments, if any, using the ’review’ and ‘track changes’ tool within Microsoft Word, and to submit their edited version to the facilitator. The Youth Advisory Team were not included in all modified-Delphi rounds as the 3 round process was assessed by the facilitator as too time consuming for youth volunteers; however, 3 Youth Advisory Team members contributed to the final round. Their comments included suggested wording additions and excellent examples of injury incidents they were aware of that illustrated the importance of the recommendations. There was strong agreement regarding wording of the recommendations from expert panelists at the end of Round 3.

Results

Responses to the 27 recommendations were received from 8 experts in the first round. Two recommendations were rated as resonating with the intended audience, and as “extremely important” or “very important” by 100 % of the expert panel: 1) Federal and provincial Health Ministries as well as regional and community-level health organizations are urged to make investments to establish and continue surveillance efforts to identify new and emerging patterns of injury, and 2) School districts, schools and Parent Advisory Councils are encouraged to implement evidence-based peer-mentorship programs that address the social context of the school environment and improve feelings of belonging and safety. Two of the recommendations received less than 66 % of the expert panel rating them as resonating with the intended audience, and less than 66 % of experts rating them as “extremely” or “very” important, and were subsequently deleted. The facilitator incorporated feedback on 2 other recommendations indicating that the content was important but should be placed as text elsewhere in the report, as it was not judged to be a recommendation.

Experts suggested that several recommendations could be merged due to their similar content, resulting in 4 being merged with existing recommendations. The remaining 19 recommendations were edited for consistency of language and style, based upon a total 135 open-ended question responses from the experts.

Two recommendations related to food security each received less than 66 % of expert ratings of resonance (57 % and 50 %, respectively) and importance (57 % and 63 %, respectively); however, a decision to retain the 2 recommendations was made based upon the expert panel opinions and the evidence revealed in the report regarding the relationship between injury and going to school or bed hungry [6]. While experts “applauded” the inclusion of these 2 recommendations, they were guarded in their assessments of the importance and how they would resonate with those who might lead and take action. Comments included: “To qualify - I hope that it will resonate”, “This is logical, so we don't need research- we need pilot interventions or communication so we address the hunger”, and “I really applaud you for including it”.

The second round again elicited responses from 8 expert stakeholders. Comments included requests for clarification related to confusing points, the addition of a rationale for 2 recommendations, and ensuring the recommendations were clearly based on evidence from the original youth injury report. For example, a reference to 4-sided pool fencing as a policy solution was deleted from the recommendations because swimming pools were not discussed in the report. Five expert panelists responded to the third and final round, three of whom were members of the Parachute Youth Advisory Team [17]. One of the expert panelists responded to indicate that she had no further comments. It is noted that panelists were asked to “provide edits and comments, if any”. Therefore, it is assumed that the remaining experts were content with the recommendations as written (Table 2).

Table 2 Final list of recommendations
Full size table

An additional table illustrates the progress and modification of each recommendation through the 3-Round modified-Delphi process [see Additional file 1].

Discussion

The Delphi process (and its modifications) has demonstrated utility to reach consensus in previous research, including: program planning, needs assessment, policy determination, resource utilization, curriculum development and the development of clinical guidelines in clinical education [11, 12, 20–23]. This research project demonstrated that the Delphi process used here, modified to receive expert opinion through online survey and email [10], was a successful method for developing critical evidence-based action-oriented recommendations in the context of a national, youth injury report.

The 2 recommendations related to the relationship identified in the report between injury and food security, elicited unexpected responses from the expert panel that required further consideration by the facilitator. Expert comments indicated the importance of the research findings, the novel nature of the findings, and the relevance to policy and action, yet only half of experts agreed that these recommendations would resonate with the groups responsible to take action. Posing a different question to experts may be required to understand how these recommendations could best be utilized.

The strengths of this study include the involvement of youth as expert advisors and the strong engagement of experts as demonstrated by responses from 8 out of 9 experts to Delphi rounds 1 and 2. The decreased response rate in the final round is a known limitation for Delphi studies. However, the research team clearly invited panelists to respond to the third round only if they had further comments. It is assumed that the non-response was in fact an indication that they had no further comments or suggestions for improvement, and were satisfied with the recommendations resulting from the first 2 rounds.

Experts provided input based upon their individual expertise and understanding of the evidence. The extent to which each expert may or may not have assessed and considered prevailing political, economic, socio-cultural, environmental and other external influences is unknown, and was not formally included within this modified Delphi process.

The Delphi method has been criticized for the fact that the opinions of a small number of experts may not be representative [24]. In this study, efforts were made to select expert panelists who represented different regions, disciplines and constituent stakeholders relevant to youth injury. It is also acknowledged that the injury prevention community in Canada is relatively small and well-connected improving the chances of identifying who the generally recognized child and youth injury prevention experts are, and ensuring representativeness of the expert panel members across regions and disciplines. Experts were supported by the previous work, and preliminary list of recommendation ideas, that emanated from the 22 chapter authors of the national injury report and this expanded the diversity and number of overall contributors to the outcomes.

Poor summary and presentation of expert input by the facilitator can result in flawed synthesis between rounds [24]. In this study, the facilitator was represented by the 5 members of the research team, all of whom are expert injury prevention researchers, and familiar with the injury prevention field in Canada. This approach, that 5 minds are better than 1, is likely to have precluded flawed synthesis.

These recommendations have the potential to move the policy agenda of local governments forward with respect to child and youth injury prevention, in particular if done in partnership with the organizations and professionals listed in each recommendation. Future research to study the implementation of these recommendations and the impact on the burden of injury to children and youth in Canada would be a valuable next step.

Conclusions

This project was successful in utilizing a modified-Delphi process to achieve agreement on 19 specific recommendations to complement the report Injury among Young Canadians: A national study of contextual determinants. These actions or recommendations will require multidisciplinary collaborations in order to support the proposed policy development, additional research, programming and clear decision-making for youth injury prevention.

An ideal environment is one where governments, business leaders and academics work together to ensure healthy public policy, enhance community capacity, support individual skills, and take all appropriate action to reduce the likelihood of injury and death; where society protects and nurtures high-risk members of the community and those who lack resources to fully act on their own behalf; where inequities are seen as challenges that threaten the health and safety of all, and which must be solved.