Apps for Exercise and Dietary Interventions for Children
Apps for Exercise and Dietary Interventions for Children
Objective Systematically review and meta-analyze the pediatric literature on behavior-change techniques (BCT) as defined by Abraham & Michie (Health Psychology, 27, 379–387, 2008), and describe whether the most effective BCTs are incorporated in physical activity (PA) and dietary mobile apps.
Methods Randomized controlled trials (n = 74) targeting diet or PA were meta-analyzed. Metaregressions were used to determine which BCTs predict aggregate effect size (ES). iTunes™ apps were coded for presence/absence of BCTs that produce larger ES.
Results Modeling was the only predictor of PA ES in children (aged 6–13 years). Consequences for behavior, other's approval, self-monitoring, intention formation, and behavioral contracting significantly predicted PA for adolescents. Modeling and social support predicted dietary ES in adolescents and children, respectively. Practice was also a significant predictor for children. A majority of effective strategies for children were not widely incorporated in apps; however, the picture is more optimistic for adolescents.
Conclusions More collaboration is needed between pediatric psychologists and technologists to incorporate evidence-based BCTs into developmentally appropriate mobile apps.
It is recommended that children aged 6–17 years participate in at least 60 min of moderate to vigorous physical activity and eat a balanced diet of at least five servings of fruits and vegetables daily (Department of Health and Human Services, 2008). Indeed, regular physical activity and adequate fruit and vegetable consumption are associated with numerous desirable health outcomes such as aerobic fitness, healthy blood pressure, decreased overweight/obesity, and better overall psychological health (Janssen & LeBlanc, 2010; Sallis & Patrick, 1994). However, observational studies of children and adolescents provide evidence that the recommended guidelines are not being met. In fact, as few as 3% of children <18 years of age meet the recommendation for moderate to vigorous physical activity, and <30% of adolescents meet the recommendation for intake of fruits and vegetables (Neumark-Sztainer, Story, Hannan, & Croll, 2002; Pate et al., 2002). Children and adolescents who fail to meet these guidelines are at risk for heart disease, diabetes, depression, and some types of cancer (Baranowski et al., 1992; Pate et al., 2002; World Health Organization, 2004). These serious consequences serve to underscore the critical importance of helping to establish healthy behaviors early in life.
The field has made attempts to modify pediatric behavior, and review papers suggest that such interventions are generally effective (Cushing, Brannon, Suorsa, & Wilson, 2014; Stice, Shaw, & Marti, 2006). Such findings highlight the potential impact of aiming intervention at healthy children through universal prevention efforts (Kazak, 2006) to assist children in changing or developing physical activity and healthy eating patterns. Despite the established efficacy of pediatric behavior-change programs, it is currently unclear what specific intervention strategies lead to changes in pediatric health behavior. This is problematic because the field is limited to assertions that "multicomponent" or "behavioral" interventions are effective without highlighting specific efficacious techniques (e.g., modeling, self-monitoring; Cushing & Steele, 2010; Kahana, Drotar, & Frazier, 2008). This creates a substantial problem because it is not possible to streamline a "multicomponent" intervention while it is possible to eliminate an individual technique that is ineffective or to champion one that is found to drive intervention success. Identifying the agents of change in behavioral interventions is a critical step toward refining face-to-face interventions and translating behavioral science to additional mechanisms of intervention.
Moreover, for translation efforts to move forward, the field needs to look outside of intervention development and toward implementation in sectors such as the thriving mHealth marketplace. Among health behaviors, physical activity and dietary patterns may be amenable to change following an mHealth intervention. Indeed, behavioral science has much to add to the technology sector by answering the call of Pagoto and Bennett (2013) to first understand behavioral techniques that drive intervention success, and then determine whether these components can be, or are being, incorporated in existing mHealth technologies. Previous attempts to characterize mHealth technologies have compared apps to the recommendations of expert panels (Schoffman, Turner-McGrievy, Jones, & Wilcox, 2013) as well as evidence-informed practices common to governmental agencies such as the National Institutes of Health and Food and Drug Administration (Breton, Fuemmeler, & Abroms, 2011), with results suggesting minimal congruence between available technologies and evidence-informed recommendations. Additional studies have examined whether mobile health apps align with clinical guidelines, suggesting most apps incorporate few, if any, behavioral strategies for weight loss (Pagoto, Schneider, Jojic, DeBiasse, & Mann, 2013).
To move forward in answering the call of Pagoto and Bennett (2013), clear operational definitions of behavioral strategies must be established and efforts must be made to determine whether these strategies are having an effect in health promotion efforts appearing in the empirical literature. If it can be demonstrated that a specific technique is effective (e.g., modeling) and there exists a clear operational definition that can be communicated to stakeholders, then it will be possible to more rapidly translate effective science into practice and the public sector (i.e., the mHealth marketplace).
One such set of operational definitions for behavior-change techniques (BCTs) used in health behavior interventions was developed by Abraham and Michie (2008), and includes a detailed coding manual. Applying this categorization system to the empirical literature provides a framework from which it may be possible to identify mechanisms of behavior change that can then be used to assess the evidence-based nature of emerging technologies such as commercial eHealth and mHealth programs (not to mention streamline existing interventions and promote cost-effectiveness work).
Since its initial development, the behavior-change taxonomy has been used to examine adult interventions targeting physical activity and diet (Michie, Abraham, Whitington, & McAteer, 2009). Michie and colleagues' (2009) review revealed that interventions including self-monitoring and at least one of four other techniques derived from control theory (goal setting, goal review, and feedback) were more effective than interventions not including these techniques. Similarly, applying the coding scheme to studies of adults with obesity and at risk for Type 2 diabetes revealed that instruction, self-monitoring, and practice were effective techniques for producing behavior change (Dombrowski, Sniehotta, Avenell, MacLennon, & Arau 'jo-Soares, 2012).
To date, no reviews exist to address what BCTs are most important for modifying pediatric health behavior. A more granular view of BCTs as moderators of aggregate effect sizes is needed to assist in determining what specific BCTs carry the most weight in the context of pediatric intervention programs targeting physical activity and diet. Given the developmental differences across the life span, a nuanced examination may identify strategies that are salient for adolescents and children that differ from adults. Finally, no systematic reviews of available mHealth apps have used the existing empirical literature as the foundation for their review.
The current study examined the current state of translational science in pediatric behavior change by first empirically assessing what BCTs have the largest impact on health behavior change, specifically physical activity and diet (fitness, step counts, fruit and vegetable consumption, etc.) using systematic review and meta-analysis. The study is limited to randomized controlled trials (RCTs), as RCTs present the strongest test of the efficacy of BCTs. Apps were then assessed for the presence or absence of the BCTs found to have the largest impact on study effect size. This approach sought to address the call from Pagoto and Bennett (2013) to (1) identify the behavioral components that lead to intervention effectiveness and (2) assess whether effective behavioral components are, or can be, incorporated into the mHealth sphere. We expected that strategies such as self-monitoring, goal setting, and feedback on goal attainment would be significant in driving intervention effectiveness as has been concluded in the adult literature (Michie et al., 2009) and is consistent with the Cybernetic Control Theory (Carver & Scheier, 1982). This hypothesis is further supported by evidence in the pediatric literature that the BCT strategies listed above appear to drive intervention effectiveness across a number of pediatric health behaviors (Cushing & Steele, 2010; Kahana et al., 2008). An additional exploratory aim is to determine differences in the effectiveness across developmental levels (children and adolescents) to examine whether other strategies consistent with social learning models may carry more weight for developmentally younger children. It was also expected that there would be some evidence that app developers incorporate these same strategies into mobile health apps, but that there would be considerable room for improvement.
Abstract and Introduction
Abstract
Objective Systematically review and meta-analyze the pediatric literature on behavior-change techniques (BCT) as defined by Abraham & Michie (Health Psychology, 27, 379–387, 2008), and describe whether the most effective BCTs are incorporated in physical activity (PA) and dietary mobile apps.
Methods Randomized controlled trials (n = 74) targeting diet or PA were meta-analyzed. Metaregressions were used to determine which BCTs predict aggregate effect size (ES). iTunes™ apps were coded for presence/absence of BCTs that produce larger ES.
Results Modeling was the only predictor of PA ES in children (aged 6–13 years). Consequences for behavior, other's approval, self-monitoring, intention formation, and behavioral contracting significantly predicted PA for adolescents. Modeling and social support predicted dietary ES in adolescents and children, respectively. Practice was also a significant predictor for children. A majority of effective strategies for children were not widely incorporated in apps; however, the picture is more optimistic for adolescents.
Conclusions More collaboration is needed between pediatric psychologists and technologists to incorporate evidence-based BCTs into developmentally appropriate mobile apps.
Introduction
It is recommended that children aged 6–17 years participate in at least 60 min of moderate to vigorous physical activity and eat a balanced diet of at least five servings of fruits and vegetables daily (Department of Health and Human Services, 2008). Indeed, regular physical activity and adequate fruit and vegetable consumption are associated with numerous desirable health outcomes such as aerobic fitness, healthy blood pressure, decreased overweight/obesity, and better overall psychological health (Janssen & LeBlanc, 2010; Sallis & Patrick, 1994). However, observational studies of children and adolescents provide evidence that the recommended guidelines are not being met. In fact, as few as 3% of children <18 years of age meet the recommendation for moderate to vigorous physical activity, and <30% of adolescents meet the recommendation for intake of fruits and vegetables (Neumark-Sztainer, Story, Hannan, & Croll, 2002; Pate et al., 2002). Children and adolescents who fail to meet these guidelines are at risk for heart disease, diabetes, depression, and some types of cancer (Baranowski et al., 1992; Pate et al., 2002; World Health Organization, 2004). These serious consequences serve to underscore the critical importance of helping to establish healthy behaviors early in life.
The field has made attempts to modify pediatric behavior, and review papers suggest that such interventions are generally effective (Cushing, Brannon, Suorsa, & Wilson, 2014; Stice, Shaw, & Marti, 2006). Such findings highlight the potential impact of aiming intervention at healthy children through universal prevention efforts (Kazak, 2006) to assist children in changing or developing physical activity and healthy eating patterns. Despite the established efficacy of pediatric behavior-change programs, it is currently unclear what specific intervention strategies lead to changes in pediatric health behavior. This is problematic because the field is limited to assertions that "multicomponent" or "behavioral" interventions are effective without highlighting specific efficacious techniques (e.g., modeling, self-monitoring; Cushing & Steele, 2010; Kahana, Drotar, & Frazier, 2008). This creates a substantial problem because it is not possible to streamline a "multicomponent" intervention while it is possible to eliminate an individual technique that is ineffective or to champion one that is found to drive intervention success. Identifying the agents of change in behavioral interventions is a critical step toward refining face-to-face interventions and translating behavioral science to additional mechanisms of intervention.
Moreover, for translation efforts to move forward, the field needs to look outside of intervention development and toward implementation in sectors such as the thriving mHealth marketplace. Among health behaviors, physical activity and dietary patterns may be amenable to change following an mHealth intervention. Indeed, behavioral science has much to add to the technology sector by answering the call of Pagoto and Bennett (2013) to first understand behavioral techniques that drive intervention success, and then determine whether these components can be, or are being, incorporated in existing mHealth technologies. Previous attempts to characterize mHealth technologies have compared apps to the recommendations of expert panels (Schoffman, Turner-McGrievy, Jones, & Wilcox, 2013) as well as evidence-informed practices common to governmental agencies such as the National Institutes of Health and Food and Drug Administration (Breton, Fuemmeler, & Abroms, 2011), with results suggesting minimal congruence between available technologies and evidence-informed recommendations. Additional studies have examined whether mobile health apps align with clinical guidelines, suggesting most apps incorporate few, if any, behavioral strategies for weight loss (Pagoto, Schneider, Jojic, DeBiasse, & Mann, 2013).
To move forward in answering the call of Pagoto and Bennett (2013), clear operational definitions of behavioral strategies must be established and efforts must be made to determine whether these strategies are having an effect in health promotion efforts appearing in the empirical literature. If it can be demonstrated that a specific technique is effective (e.g., modeling) and there exists a clear operational definition that can be communicated to stakeholders, then it will be possible to more rapidly translate effective science into practice and the public sector (i.e., the mHealth marketplace).
One such set of operational definitions for behavior-change techniques (BCTs) used in health behavior interventions was developed by Abraham and Michie (2008), and includes a detailed coding manual. Applying this categorization system to the empirical literature provides a framework from which it may be possible to identify mechanisms of behavior change that can then be used to assess the evidence-based nature of emerging technologies such as commercial eHealth and mHealth programs (not to mention streamline existing interventions and promote cost-effectiveness work).
Since its initial development, the behavior-change taxonomy has been used to examine adult interventions targeting physical activity and diet (Michie, Abraham, Whitington, & McAteer, 2009). Michie and colleagues' (2009) review revealed that interventions including self-monitoring and at least one of four other techniques derived from control theory (goal setting, goal review, and feedback) were more effective than interventions not including these techniques. Similarly, applying the coding scheme to studies of adults with obesity and at risk for Type 2 diabetes revealed that instruction, self-monitoring, and practice were effective techniques for producing behavior change (Dombrowski, Sniehotta, Avenell, MacLennon, & Arau 'jo-Soares, 2012).
To date, no reviews exist to address what BCTs are most important for modifying pediatric health behavior. A more granular view of BCTs as moderators of aggregate effect sizes is needed to assist in determining what specific BCTs carry the most weight in the context of pediatric intervention programs targeting physical activity and diet. Given the developmental differences across the life span, a nuanced examination may identify strategies that are salient for adolescents and children that differ from adults. Finally, no systematic reviews of available mHealth apps have used the existing empirical literature as the foundation for their review.
The current study examined the current state of translational science in pediatric behavior change by first empirically assessing what BCTs have the largest impact on health behavior change, specifically physical activity and diet (fitness, step counts, fruit and vegetable consumption, etc.) using systematic review and meta-analysis. The study is limited to randomized controlled trials (RCTs), as RCTs present the strongest test of the efficacy of BCTs. Apps were then assessed for the presence or absence of the BCTs found to have the largest impact on study effect size. This approach sought to address the call from Pagoto and Bennett (2013) to (1) identify the behavioral components that lead to intervention effectiveness and (2) assess whether effective behavioral components are, or can be, incorporated into the mHealth sphere. We expected that strategies such as self-monitoring, goal setting, and feedback on goal attainment would be significant in driving intervention effectiveness as has been concluded in the adult literature (Michie et al., 2009) and is consistent with the Cybernetic Control Theory (Carver & Scheier, 1982). This hypothesis is further supported by evidence in the pediatric literature that the BCT strategies listed above appear to drive intervention effectiveness across a number of pediatric health behaviors (Cushing & Steele, 2010; Kahana et al., 2008). An additional exploratory aim is to determine differences in the effectiveness across developmental levels (children and adolescents) to examine whether other strategies consistent with social learning models may carry more weight for developmentally younger children. It was also expected that there would be some evidence that app developers incorporate these same strategies into mobile health apps, but that there would be considerable room for improvement.
