Public Health Procedures,Alone,Will Not Prevent Child Obesity

Ignored Biological Etiologies

Beets et al. ¹ accepted our encouragement of new biobehavioral research on biological etiologies of child obesity, but took the opportunity to criticize behavioral approaches to child obesity prevention. They took an essentially public health perspective pointing out limitations in more behavioral approaches to child obesity prevention.

Their arguments for a more public health approach to child obesity prevention, however, appeared to be predicated on a simple energy balance model of biological etiology. The simple energy balance model, although perhaps relevant in some cases of child obesity or in combination with other etiological factors, has been shown to be inadequate to account for the many complexities in obesity.^6–8 Beets et al. previously stated that their proposed structured day hypothesis produced the same predictions and same methods for change ⁹ as our proposed circadian/circannual rhythm hypothesis, ⁴ and thereby the new biological explanation was irrelevant. We would argue that understanding how or whether the school year prevents excess weight gain is critical, leading to more targeted intervention. Understanding that just increasing time spent in school or structure may not be helpful, if it does not support circadian or circannual alignment. The authors note that often public health behavioral approaches have failed to address the root causes of obesity. ¹ We could not agree more. Understanding the possible biological etiologies of obesity will likely lead to more effective obesity interventions.

It is unfortunate that those authors did not address our proposed possible other biological etiologies for child obesity. We believe behavioral or public health approaches, independent of the biological etiology, could lead to inappropriate or limited interventions. As biological research advances, new behaviors may be identified for targeting, which may not be incorporated into the authors' existing behavioral/public health intervention framework. For example, the extent to which the microbiome is identified as a child obesity etiology, specific dietary interventions (e.g., intake of probiotics) seem to offer the most potential for prevention. Similarly, prevention of a viral etiology would require infection protective interventions. The extent to which there are multiple etiologies would preclude the generalizability of their methods and interventions to the likely diverse etiologies of child obesity. We proposed that child obesity prevention researchers (whether behavioral or public health) need to collaborate with investigators of these new biological etiologies to explore how prevention could be advanced.

Responses to Public Health Research Issues

Although Beets et al. ¹ raised many interesting and important public health issues, they deliberately minimized the contributions of behavioral science and appeared to ignore strategically important published literature. We address what we see as the limitations in their arguments.

Beets et al. ¹ questioned the commonly employed 12 weeks of intervention implementation to test innovative programs, suggesting much longer intervention durations (but of unspecified length) are needed to test interventions. Alternatively, we know of at least two large (thousands of participants), well-funded (millions of dollars), long duration (years), and well-conceptualized, but based on the simple energy balance model, which had no effect on BMI as the primary outcome.^10,11 Although national or international control of child obesity will require pervasive and continued intervention, 12 weeks seems long enough to impact body adiposity and, thereby, appropriate for testing new approaches in an efficacy evaluation trial. Available research funds are limited and very long interventions pose issues of sample attrition with subsequent maintenance bias. ¹² It is up to the investigators who propose a new intervention to propose and justify the duration of intervention after which it is reasonable to expect an impact on BMI. Rather than targeting all children in a setting, perhaps interventions need to target groups at high risk to test the effects of a new intervention within a 12-week or other prespecified period? This could be done with children in schools, calculating power only for the high-risk subsample, and conducting primary analyses of effect on that subgroup. Delivering the intervention to everybody avoids the problem of identifying and possibly stigmatizing the high-risk children.

Beets et al. ¹ suggested that obesity prevention interventions not be offered in locations that already encourage healthier behavior, for example, schools and summer camp. They prioritize interventions targeting summer-based behavior, with which we agree, but schools offer an important venue to deliver such programs to large numbers of children. In addition to summer, however, the interventions need to target subgroups of high-risk children. Our research indicates that only about 9% of students begin a trajectory toward obesity in the summer after kindergarten, and another 9% begin that trajectory in the summer after second grade. ¹³ Validating these trajectories in other samples, and identifying their characteristics and causes of excess weight gain may be used to target interventions and provide a means for more appropriately targeting summer interventions.

Beets et al. ¹ encouraged moving away from individual behavior-based interventions and move toward more public health friendly policy interventions, that is, policies, environmental change, and systems. A problem with some policy interventions (e.g., taxes on food) is that they penalize many citizens not in need of an obesity prevention intervention. A review of 53 reviews of the impact of policy effects indicated inconsistent effects on behavior (especially not calories consumed) and adiposity. ¹⁴ Ultimately, if a policy has an impact on obesity-relevant child behaviors, there must be some behavioral mechanism that accounts for that impact. Perhaps policy interventions need to be better informed about, and test, behavioral mechanisms of effect. The measurement of environmental influences on behavior is problematic and has indicated no relationship. ¹⁵

Systems-based interventions may offer some future solutions. ¹⁶ However, there is an extensive theoretical framework on how systems function. ¹⁷ Research is needed on system-related mechanisms to show whether, and the extent to which, they account for obesity-related behavior and/or obesity. Although systems-type models have been demonstrated to be useful in the biological sciences, ⁷ to our knowledge no corresponding study has demonstrated systems-related effects across community, family, peer, and other likely influences.

Beets et al. ¹ proposed that scalability of an intervention needs to be a primary consideration in intervention design. Scalability of an intervention is important when effective intervention procedures have been identified. Scalability should not be a priority concern when attempting to specify mechanisms and/or likely effective behavior change techniques. Once effective procedures have been identified, research can address innovative ways to deliver such interventions to large groups of people.

Beets et al. ¹ decried the lack of use of innovative technological measures of behaviors. We agree that advances in objective behavior measurement are important. They limit their presentation to include measures of physical activity. With regard to food, our study with taking all day images for measuring dietary intake indicates it takes 9 hours to process the images for one child for 1 day ¹⁸ and still requires an interview with the child to clarify the content of some of the images. ¹⁹ This suggests that these methods are not currently ready for use in behavioral research. Advances are being made in the automation of these methods: identification of food preparation practices ²⁰ ; whether an image contains a food ²¹ or eating behaviors ²² ; identification of the food ²³ ; and portion size assessment,^19,24 all necessary to make image processing practical. Other technologies, for example, measures of chewing and swallowing,^25,26 may offer complementary or separate methods. Although promising, this research is still several years from providing valid off-the-shelf methods for use in dietary change research.

Beets et al. ¹ criticized the usual practice of measuring behavior for “seven or fewer” days. Although we agree that behavior is dynamic, interventions need to be assessed with methods that reflect pre- and postintervention periods. The number of days of assessment needs to be kept to a minimum, which collects reliable data, minimizes burden on the participant (with ensuing concerns for lack of compliance), and minimizes cost of implementing the evaluation. Methods have been specified and employed to identify the numbers of days of observation needed to reliably assess a behavior.^27,28 These methods should be applied to the newly available measures.

Beets et al. ¹ touted the “robust evidence linking obesogenic behaviors to overweight or obesity.” Alternatively, a recent literature review reported extensive inconsistencies in the extent to which behaviors were related to adiposity or obesity. ²⁹ By definition, obesogenic behaviors are related to obesity. However, it is not clear which behaviors may be obesogenic, perhaps indicating a need for behavior-related research on new biological etiologies. ³

Finally, Beets et al. ¹ argued that from a public health perspective, it makes more sense to target multiple behaviors since the health problems are influenced by multiple behaviors. Alternatively, if we have not definitively demonstrated we can effectively and consistently change one behavior,^30,31 there seems to be hubris in attempting to change more than one behavior. Alternatively, a recent behavioral theory has proposed cross-behavior compensation as a way to assess and impact multiple behaviors.^32,33 This behavioral theory, although offering possible new insights, requires additional research to test its propositions and assess the extent to which it can sufficiently account for the behaviors influencing obesity.

Conclusion

We believe that although public health approaches have contributed important advances to the prevention of child obesity, they need to be informed by the latest biological and behavioral research. We hope that we have contributed to an examination of possible biological etiologies for child obesity; to concerns raised about current behavioral issues in child obesity prevention; and thereby to further advance these important areas. Time and further behavioral and public health research will tell.

Authors interested in further pursuing these issues are invited to contribute relevant literature reviews or original empirical contributions, and hope they will submit them to Childhood Obesity.