Understanding the mechanisms through which interventions achieve success in changing the physical activity behaviours of children is imperative. The aim of this paper was to review evidence of the efficacy (and size of effect) of physical activity interventions targeting 5-12 year old children on potential physical activity mediators and to examine whether success in promoting physical activity varied in terms of potential mediator outcomes. Thirty-one intervention studies published between 1985 and April 2012 satisfied the criteria for inclusion in this review with 18 mediators identified and 77 outcomes on potential mediators reported (nb: these are not mutually exclusive as some studies targeted multiple mediators and reported results separately by sex). There were 33 positive intervention outcomes on the targeted potential mediators and 73% of the time a positive effect on physical activity was also reported. In contrast, where a null effect on a potential mediator was reported (44 results reviewed), a positive effect on children’s physical activity was identified on just 54% of occasions. Although none of these studies performed a mediating analysis, the results suggest that where a positive intervention effect on the mediator was found, there was more likely to be a positive effect on physical activity.
This review confirms that physical activity is a complex entity and that the potential mechanisms of change are multifactorial. Consistent with previous reviews [27–29], self-efficacy, knowledge, intentions, enjoyment, and social support were the most commonly targeted. The results of this review are presented according to an ecological framework and clearly show that much of the focus of previous children’s physical activity interventions has been on cognitive/psychological factors with very few studies targeting the broad range of social, physical, cultural or policy environmental influences, particularly not concurrently. The current review included a broader range of potential mediators for consideration than previous reviews that have only included studies that conducted statistical tests of mediation in adolescent and child interventions [27, 72]. A more inclusive review of potential mediators was deemed important for informing the development of more effective strategies that could be incorporated into future interventions, and also for identifying the gaps in the types of potential mediators that should or could be targeted.
Given the target age of children in these interventions, it is somewhat surprising that the majority of intervention research has focused on cognitive/psychological aspects of children’s physical activity at an age where children’s autonomy is just emerging and the opportunity to be physically active is likely to be highly dependent on adult carers (i.e., parents, grandparents, teachers). Fewer than 40% of studies in the current review reported a positive impact on cognitive/psychological potential mediators. The most effective changes reported were in children’s knowledge of physical activity, which may not translate into change of behaviour [73, 74]. While enjoyment has been found to be a significant correlate of children’s physical activity, other potential cognitive mediators targeted in the studies reviewed, such as self-efficacy, knowledge, intentions and attitudes have not been strongly supported as correlates [27, 75–77]. Self-efficacy has mediated changes in physical activity in several adolescent studies [72, 78, 79]; however, no studies targeting children have undertaken mediating analysis to confirm mediation pathways.
Of the studies that targeted social environmental potential mediators of children’s physical activity, 37% reported some intervention success with the majority achieving trivial to moderate effect sizes with these variables. Social support, the most commonly targeted potential social environmental mediator in this review, has been identified previously as a consistent correlate of physical activity in children [76, 77]. Less than half of the studies reviewed in the current paper showed a positive effect on this potential mediator or a positive effect on the physical activity outcome for studies that targeted this potential mediator. This result may reflect the variation and/or quality of the social support measures used in these studies. A previous review on the validity and reliability of instruments used to assess potential mediators of children’s physical activity reported a lack of appropriate, valid and reliable instruments for measuring constructs such as social support , indicating the need to consider the ways in which potential mediators are measured prior to drawing conclusions regarding their use as potential mediators.
The present review also identified a number of studies where the intervention had no significant effect on the potential mediator; however, a significant effect on physical activity was reported. For example, the only study to target change in children’s attraction to physical activity was not successful in effecting such change, but there were significant physical activity outcomes in that study . The mechanism/s through which this change occurred is unclear. The intervention may have achieved its effect on physical activity through another potential mediator or, as discussed above, the measure used to assess the potential mediator may have lacked adequate validity and reliability and was therefore unable to show an effect. Possible reasons for the lack of demonstrated effect on the potential mediators may be due to the wrong mediator being targeted, lack of statistical mediating analysis, lack of power in the sample to detect change, inadequate intervention dose and/or lack of validity and reliability of mediator measures.
In addition to issues regarding instrument reliability and validity, measurement specificity should also be considered. Stathi et al highlight the importance of measuring and reporting the type, intensity and context of physical activity, ensuring the differentiation of the variable constituents of children’s physical activity as activity undertaken for different purposes and intensities is predicted by different correlates and mediated by different variables. It has been suggested that not considering these dimensions of physical activity may result in inaccurate and even misleading estimates of intervention effects .
It is important to consider that self-report measures are able to provide estimates of the type, intensity and context of physical activity however the use of these measures is limited due to issues with correlated measurement error when assessing associations and thus biased conclusions . Objective measures also provide limitations as current technology is not able to assess the type and context, particularly at a large scale. However, future research should identify the optimal method of combining self-report and device-based data which may help overcome these issues. Recent methodological advancements in objective physical activity assessment where the use of computer based learning algorithms (For example, artificial neural networks) are being used to estimate activity type may help overcome some of the limitations of objective measures [83, 84].
The findings from this review make it difficult to recommend any particular potential mediator as a target for children’s physical activity. This is not to imply that any of the potential mediators reviewed are unimportant, there is simply insufficient evidence that these factors lie on the mediating pathway of children’s physical activity behaviour change. Only one-third of studies reported small to modest changes in the targeted potential mediators, with approximately 75% of these studies reporting positive effects on physical activity outcomes. It is intriguing that a greater number of studies that reported success in changing a potential mediating variable also reported change in children’s physical activity; however, this could also be a reporting bias in the studies. The most frequently targeted potential mediators were cognitive/psychological factors, with only physical activity knowledge having mainly positive outcomes.
Limitations and strengths
There are limitations to the present review, some of which are due to gaps in the literature itself. Only papers published in the English language were included in this review, and the majority of studies were conducted in the US or Europe. Studies were diverse in character (e.g. mediators targeted and strategies used) and so it was not possible to make recommendations regarding which mediator/s or strategies should be targeted to effect change in physical activity. Several studies may not have been powered to detect significant associations between the intervention and potential mediators; however, effect size calculations were performed for this review to try and aid interpretation of whether results were meaningful.
A quality metric was not applied to this review for several reasons. Inclusion criteria of published studies were deliberately broad so that a more informative representation of the breadth and consistency of potential mediators that children’s physical activity intervention studies have reported could be portrayed. With the scarcity of studies that have systematically reported the targeting of specific mediators, designing strategies that address these mediators or performing statistical mediating analyses, we believe that application of a quality metric to this review would have been pointless given the mediator literature is still so under-developed. The present review was also unable to determine whether studies that targeted specific mediators of change in children’s physical activity applied appropriate strategies to effect these changes. Further, there was such variation between studies in the intervention strategies used it was difficult to draw conclusions about what specifically worked in effective interventions compared to ineffective studies or to link such findings to a match or mis-match between targeted mediators and strategies adopted.
Strengths of the review included the systematic approach adopted and the more inclusive criteria for study inclusion, and the synthesis of evidence of intervention effectiveness on the mediator according to physical activity outcomes.
Future interventions promoting children’s physical activity should clearly identify and provide a rationale for the theoretical framework used and the hypothesised mediators of change, as well as clearly linking the targeted mediator with the approach used. Potential mediators that target the full ecological framework, in particular the physical, cultural and policy environments, should be tested. Studies outside the US and Europe should be encouraged, and the use of appropriate statistical mediation techniques and valid and reliable measures that are sensitive to change is recommended to test the pathways of behavioural change, thereby informing future intervention development.