Existing literature

To identify existing evidence in this area, a systematic review of the English and non-English literature was performed using the following search terms: physical activit* OR motor activity (MeSH) OR sedentary behavio* AND child* OR adolescen* OR youth* AND research priorit* OR research agenda* OR research issue*. The databases PsychINFO (1887–), SPORTDiscus (1949–), Cochrane (1992–), CINAHL (1937–), ERIC (1966–) and PubMed (1950–) were searched in May 2012. Additional studies were also identified by contacting experts, Google searching and identifying potential studies in the reference lists of identified studies. Only four previously published papers that arrived at research priorities in child physical activity and/or sedentary behaviour were identified [11, 13–15]. A working paper by Bull et al. [11] identified research priorities in physical activity with a focus on low to middle income countries. Evenson and Mota [13] highlighted research on the determinants and outcomes of physical activity and made recommendations for future study designs. Mountjoy et al. [15] identified existing gaps in physical activity research for children, with a focus on the need for greater collaboration between sport and existing programmes. The final study by Fulton et al. [14] had two aims. Firstly, the study aimed to review the current knowledge of existing methods for assessing physical activity and sedentary behaviour. Secondly, on the basis of this, the study aimed to set research priorities on the use of reliable and valid measurement tools to assess physical activity and sedentary behaviour in children aged 2–5 years.

While these studies were valuable contributions, they also had many limitations, including unsystematic participant selection, unstructured data collection procedures, and limited reporting on the process followed to arrive at the research priorities. Furthermore, the participants involved in the decision-making processes did not always represent the broader community of researchers, either from a geographical or institutional point of view. In addition, the anonymity of participants was not maintained during the consensus process. These limitations warranted a further study with an aim to arrive at a set of research priorities by employing a structured and rigorous methodology and improving reporting quality.

Expert demographics

All 24 experts completed the three Delphi rounds. Data was collected on the 24 experts’ geographical distributions, institutional affiliations and years worked in the study area.

As a group, the 24 experts represented every geographical region and 12 sub-regions. This geographical distribution is illustrated in Figure 2.

In terms of institutional affiliation, twenty-three experts acknowledged they were affiliated with an educational institution, eleven were affiliated with a professional organisation, six with an international organisation, six with a non-government organisation and four with a government organisation. It was noted that due to the nature of their work, experts were often affiliated with more than one institution type.

In regards to years worked in the study area, twelve experts had worked in for greater than 16 years, five had worked for 11 to 15 years, four had worked for 6 to10 years and three had worked for less than five years.

Results from Delphi rounds

In Round 1, each expert put forward five research issues. Collectively this provided a total of 120 issues across all 24 experts, with 60 for each panel. Following qualitative reduction of overlapping issues, 26 issues from Panel A and 34 issues from Panel B, were carried forward to Round 2. On reviewing the amended list, all exerts agreed that the issues they had raised were adequately represented.

From Round 2, the mean Likert-scale ratings were used to determine the top 20 issues for each panel. For Panel A, the mean Likert-scale ratings of the top 20 issues ranged from 3.5 to 5.0, with 18 of 20 issues having a median rating of >4.0 (“important”). For Panel B, the mean Likert-scale ratings of the top 20 issues ranged from 4.0 to 4.8, with all 20 research issues having a median rating of >4.0.

There was only weak intra-panel agreement. The mean inter-individual rho (±95% CI) was 0.20 ±0.05 for Panel A and 0.13 ±0.04 for Panel B. The average standard deviation of the rankings for individual issues was 5.1 (Panel A) and 5.3 (Panel B). When Panel B ranked Panel A’s issues, the correlation was very strong (rho ±95% CI: 0.79 ±0.17), and when Panel A ranked Panel B’s issues, the correlation was strong (rho ±95% CI: 0.52 ±0.31). Figures 3 and 4 clearly illustrate the correlations for each research issue.

Study outcomes

The primary outcome of this study was the development of 29 international research priorities in child and adolescent physical activity and sedentary behaviour. In order for the research priorities to be useful, it is important that they be neither too general nor too specific. The research priorities in this study appear broad enough to enable them to be transferable to researchers’ specific regions and contexts.

The final set of research priorities address a broad range of areas from epidemiology, determinants and correlates, through to intervention effectiveness and translational research. Of the 29 identified research priorities, ten related directly to translational research centred on intervention design and effectiveness. These focussed on specific behaviours (active transport, screen time, sport, physical education), settings (schools, communities, whole of population), or vehicles (mass advertising, policy). Translational research, centred on intervention design and effectiveness, can potentially guide governments and stakeholders to fund interventions that are the most effective, sustainable and transferable for changing behaviours [7]. This is important because to date, the research community has not been very successful at developing interventions for children and adolescents that bring about long-term and sustained change in health behaviours [10]. In addition, little attention has been given to the importance of the intervention setting and establishing what works in what situation and with whom [22].

Nine of the research priorities had a focus on capturing and quantifying the health benefits of engaging in physical activity and limiting sedentary behaviour, These research priorities were concerned with the impact of physical activity and sedentary behaviour on obesity, cognition, and general health and well being, and on describing behavioural patterns (across the day or the life-course or in specific populations such as pre-school children). Epidemiological research was considered important to address the cause, distribution and patterns of childhood physical activity and sedentary behaviour on current and future health [2, 6, 9, 23].

Six research issues related to determinants and correlates research such as psychosocial and cultural/parental factors, the impact of technology, and the importance of enjoyment and lifestyle in general. Research that focuses on the determinants and correlates of behaviours is important. This is because while many correlates appear to be intuitively obvious, at present they have mixed support from high quality research [3].

Four issues did not fit into the aforementioned categories. They were related to the theory of behaviour change, injury prevention, measurement of behaviours and the physical education in culture of movement. Objective measurement of behaviours was ranked highly and is thought to be a “necessary first step for conducting meaningful epidemiological surveillance, public health research and intervention research” [14] p.124.

Strengths and limitations

Unlike previously identified priority reports [11, 13–15] this study employed a Delphi method to arrive at a more valid set of research priorities. Strengths related to the Delphi method include participant blinding, iterative data collection and controlled feedback between rounds. For example, the identities and responses of the experts were anonymised so that the identified research priorities could not be dominated by certain individuals [24]. Furthermore, the provision of controlled feedback allowed experts to individually consider their views in light of their panel’s collective opinion.

Other strengths related to the methodology were the use of criterion and purposive sampling methods. This procedure meant that all participants held a senior position in their respective organisations and had published in the study area. In addition, experts collectively represented every major world region and a wide range of discipline areas, affiliations and interests. This approach meant that the identified research issues were more likely to reflect the most important physical activity and sedentary behaviour issues facing the children and adolescents worldwide.

A novel component of this study was split-panel approach, which allowed comparisons to be made between the rankings given by the two expert panels. The experts from each panel were taken from the same population, given the same study information, answered identical online questionnaires and participated simultaneously and independently. One can therefore be confident that comparing the Round 3 rankings of Panel A and Panel B experts would provide valid measures of inter-panel agreement.

The weak intra-panel agreement was weak, which is likely a reflection of the natural variation of individual’s opinions and areas of interest within the broad study area. This weak agreement could also highlight the advantages of the methodology which retained anonymity and used an online mode of data collection. There were fewer pressures to conform to others opinions due to decreased likelihood of peer dominance and status. Evidence to reinforce confidence in the results is the strong to very strong (rho = 0.52–0.79) inter-panel agreement. While experts were invited from every United Nations sub-region (United Nations 2011), no experts from the following sub-regions took part: Southern Africa, Middle Africa, Caribbean, Eastern Europe, Australia, Central Asia and Western Asia. This was significant because many of these sub-regions are heavily involved in physical activity and sedentary behaviour research. Consequently, caution should be applied when recommending that the identified research priorities truly provide a global perspective. Nonetheless, these research priorities provide an international context from which priorities at the regional, national and local levels can be developed.

In addition the priorities were set for the broad area of child and adolescent physical activity and sedentary behaviour. Due to the generality of this topic, it may be that the research priorities are not relevant when conducting research into minority populations. For example, children and adolescents with disabilities may warrant different research issues not identified in this study.

Implications for research

We hope that the identification of a set of ranked research priorities may contribute to more co-ordinated international research. For example, research priorities can help inform post-graduate students regarding where the current evidence gaps exist. This may be especially helpful for researchers who reside in less developed or marginalised research regions. In addition, encouraging more guided research can help to conceptualise how findings can be used as a basis for policy decisions. Lastly, research priorities can help to direct valuable funding into priority areas and away from studies on over-researched or lower priority topics.