Evidence-Based Obesity Prevention in Childhood and Adolescence: Critique of Recent Etiological Studies, Preventive Interventions, and Policies^,,

Abstract

Prevention of obesity in childhood and adolescence remains a worthwhile and realistic goal, but preventive efforts have been beset by a number of problems, which are the subject of this review. The review draws on recent systematic reviews and evidence appraisals and has a United Kingdom (UK) perspective because there is a rich evidence base in the United Kingdom that may be helpful to obesity prevention researchers elsewhere. Recent evidence of a leveling off in child and adolescent obesity prevalence in some Western nations should not encourage the belief that the obesity prevention problem has been solved, although a better understanding of recent secular trends might be helpful for prevention strategy in future. An adequate body of evidence provides behavioral targets of preventive interventions, and there are frameworks for prioritizing these targets logically and models for translating them into generalizable interventions with a wide reach (e.g., school-based prevention interventions such as Planet Health). An improved understanding of the “energy gap” that children and adolescents experience would be helpful to the design of preventive interventions and to their tailoring to particular groups. In the United Kingdom, some recent etiological evidence has been taken as indicative of the need for paradigm shifts in obesity prevention, but this evidence from single studies has not been replicated, and paradigm shifts probably occur only rarely. Ensuring that the evidence base on etiology and prevention influences policy effectively remains one of the greatest challenges for childhood obesity researchers.

Introduction

Obesity prevalence in childhood and adolescence has increased markedly across most of the globe in the past few decades (1), and prevalence is now high. In the United Kingdom (UK), for example, ~10% of children are obese (defined as BMI in the 95th percentile relative to UK reference data from 1990) at the end of primary school (elementary school), and ∼25% of adolescents are obese at the end of high school (2). These are conservative estimates; a recent systematic review found that the BMI provides an adequate means of defining excessive body fatness and the comorbidities of excessive fatness (3), but has a high false-negative rate (3). A large number of excessively fat children and adolescents will not be identified as obese based on their BMI (3). The recent systematic review also found that the widely used Cole-International Obesity Task Force BMI–based definitions of childhood and adolescent obesity are even more conservative than most definitions based on national reference data for BMI (3). Where national reference data for BMI have been obtained after the obesity epidemic, use of the Cole-International Obesity Task Force definitions may be less conservative than use of national reference data (4). A recent concern is the increasing evidence that adverse cardiometabolic consequences are probably manifest to some degree across a wide range of body fatness, below the thresholds of BMI that would usually be defined as overweight or obese (5). Systematic reviews have identified a wide range of comorbidities of childhood obesity, both in the short term [for the obese child (6)] and the long term [for the adult obese as a child (7)].

Mounting a rapid, appropriate, and evidence-based (rather than belief-based), public health response to the epidemic of child and adolescent obesity has been problematic in the United Kingdom and elsewhere. This review aims to provide a critique of some of the major challenges in evidence-based child and adolescent obesity prevention. The critique takes an evidence-based perspective, drawing on systematic reviews where available. The critique also focuses on the UK experience because there is a rich evidence base in the United Kingdom that may provide lessons that are not well known but that may be helpful to obesity prevention in other countries. The critique aims to address the following questions: What is the appropriate response to recent evidence that childhood obesity prevalence may be leveling off? What are the best evidence- based behavioral targets of preventive interventions and how should targets for interventions be chosen? Which intervention models should be prioritized for translation into population-based prevention policy? Should the size of the energy gap, which has to be reduced to prevent obesity, be considered when designing preventive interventions? How should the flood of new evidence on the etiology of obesity be incorporated into obesity prevention interventions in future? How can evidence and policy be matched more successfully in future? The overall aim of addressing these questions is to help provide a framework that might make public health responses to the obesity epidemic more evidence based and less belief based in future.

Is childhood obesity prevalence leveling off or decreasing in the Western world?

A recent systematic review (8) summarized a relatively large body of evidence from a number of Western nations showing that the rate of increase in the prevalence of child and adolescent obesity has decreased and in some countries may have leveled off. This is a somewhat surprising observation, given clear trends to rapid increases in obesity prevalence in the past decade (9). It may represent a temporary pause in the epidemic rather than any long-term change (9). There may be a degree of bias in published studies [increasing underrepresentation of the overweight and obese, as in some surveys in the United Kingdom (2)] and/or in the population surveillance data that demonstrate an apparent slowing in the trajectory of the epidemic. It is also likely that simple proxies for excess adiposity, such as the BMI, are too crude to detect subtle population changes in body fatness or changes in body fat distribution. The course of the obesity epidemic in US children in the 1970s and 1980s was detectable from population-based studies that used skinfold thickness measures, when it was not detectable in population-based studies that used BMI (10). In the United Kingdom, secular trends toward increased obesity during the mid-1980s were detectable from trends in waist circumference when these were not detectable from trends in BMI (11).

One further concern is that what appears to be a leveling off in prevalence nationally may mask subtle but important differences between subgroups. In the United Kingdom, for example, despite evidence of the leveling off of prevalence overall, there appears to have been continuing increases in prevalence among children and adolescents of low socioeconomic status (12). However, population surveillance of childhood obesity surveillance in the United Kingdom, based on national surveys of BMI, is not ideal for detecting differences in secular trends between subgroups because our surveys are rather small. The use of routinely collected data for surveillance (e.g., from routine primary care health checks in early childhood and routine school nurse checks later in childhood) provides much larger numbers that are helpful for surveillance (13), but concerns over the recent underrepresentation of the obese remain (2).

If the secular trends toward an increasing child and adolescent obesity prevalence are slowing in some Western nations, it would be helpful to explore the likely reasons, because these could provide a useful guide to future health policy. It has been suggested that greater public awareness of childhood obesity may have contributed to the apparently encouraging recent trends, perhaps via changes in lifestyle. Although substantial media interest in this subject must have increased public awareness at some level, a systematic review found that parental awareness of childhood obesity was consistently limited (14), and more recent UK studies do not provide encouragement that parental perceptions of child weight status are realistic (15). Current UK child, adolescent, and parent perceptions of important health behaviors such as levels of physical activity are usually also very unrealistic (16). Recent trends in health behaviors identified by surveillance in the United Kingdom do not provide entirely convincing evidence that lifestyles of UK children and adolescents have become less “obesogenic” in recent years (17). Recent longitudinal studies of cohorts of children in the United Kingdom tend to show progressive increases with age in the prevalence of excess body fatness, BMI, and waist circumference Z-scores and in the prevalence of overweight and obesity (18), and these do not support the view that the environment is becoming less obesogenic over time in the United Kingdom, although a thorough analysis of birth cohort and period effects would be necessary to understand trends fully (8).

What are the most appropriate behavioral targets for obesity prevention interventions and how should targets be selected?

Selecting targets for preventive interventions should ideally involve an understanding of the complex evidence of the etiology of child and adolescent obesity combined with a logical framework for prioritizing potential targets. It should go without saying, but recent experience in the United Kingdom suggests that it is worth restating, that greatest confidence in etiology should come from systematic review and/or critical appraisal of the entire body of etiological evidence rather than from single etiological studies. A number of systematic reviews recently provided informative and concise summaries of the etiological evidence. Monasta et al. (19), for example, highlighted the strength, quality, and consistency of evidence supporting the view that a number of lifestyle factors contribute to childhood obesity: excess TV viewing, low physical activity, excess consumption of sugar sweetened–drinks, formula feeding in infancy, lack of sleep, maternal smoking during pregnancy, and rapid early growth. These lifestyle factors are at least potentially modifiable, and reviews of the body of evidence should provide increased impetus for interventions that focus on enabling changes in these behaviors. In all cases, plausible mechanisms exist that explain why these factors contribute to obesity risk in early life. Whether rapid early growth reflects a cause of later obesity or is simply an early marker of an energy balance trajectory leading to later obesity is less clear, but regardless of which explanation applies, this evidence is still helpful in highlighting the potential importance of early life for public health efforts in prevention.

Having established important and potentially modifiable behaviors as candidates for preventive intervention, a logical issue would be to prioritize these. Anecdotally, the experience of the author in the United Kingdom has been that some researchers and many policymakers have a tendency to prioritize candidate behaviors based on their beliefs and personal experiences, often with little or no reference to the evidence base, perhaps in part because the apparent simplicity of the etiology of obesity seems not to require a detailed knowledge of the evidence. One helpful framework for prioritizing candidate behaviors with a view to selecting the most appropriate behavioral targets in interventions, which has been largely ignored in the policy response in the United Kingdom, was suggested by Whitaker (20). In the United Kingdom, obesity prevention interventions often focus on trying to change behaviors that are probably not important to the development of obesity and/or not readily modifiable. Whitaker (20) proposed that the behaviors that should be rated as most important in obesity prevention interventions should be those that are likely to be modifiable, be implicated in the etiology of childhood obesity, do no harm, and be helpful to child health and development in other ways.

What are the most appropriate models in childhood obesity prevention?

Having identified and prioritized target behaviors, translation of this information into interventions requires evidence from intervention programs, ideally model interventions from the literature. Interventions that are likely to be generalizable and to have a potentially wide reach and evidence of efficacy and, ideally, of effectiveness and cost-effectiveness, should be prioritized in policy. Such interventions are admittedly scarce in childhood obesity prevention, but systematic reviews and critical appraisals of the literature (21) suggest that the most promising model to date for school-based obesity prevention interventions is probably Planet Health, a potentially generalizable intervention with evidence of efficacy and cost-effectiveness, at least in the United States (22). Other interventions that are likely to be generalizable and that have evidence of efficacy should also be of great interest. In the United Kingdom, for example, a very simple and low-cost primary school–based intervention showed efficacy at 1 y of follow-up (23), although not at longer term follow-up (24).

Translation of the most promising models in prevention from 1 nation to others is a logical way forward in prevention, but requires effort to adapt the original intervention and demonstration of efficacy in the new setting. Recent experience in the United Kingdom suggests that such model interventions are often ignored by researchers and policymakers in favor of interventions that are created de novo. One notable exception is the attempted translation of a successful US program to the United Kingdom (25).

Systematic reviews of preventive interventions (21, 26) provide a convenient means of accessing the body of evidence. These reviews have noted weaknesses and gaps in the evidence, but have been encouraging about the prospects for preventive interventions (21, 26); evidence-based optimism might be the most informed view to take at present. In addition, a number of very recent preventive interventions have evidence of efficacy across a wide age range (27–29); this provides further grounds for optimism. One trend that is discernible in the evidence on intervention studies is the tendency for earlier interventions [in utero, in infancy, in early childhood (30, 31)]. If efficacy is demonstrated, these will provide further evidence of optimism. Attempts to prevent obesity in early life are particularly appealing because some potentially brief interventions (e.g., promotion of breastfeeding) could have long-lasting effects because many such interventions would have benefits for child health and development beyond the issue of obesity and because the establishment of obesogenic behaviors may occur early in life (32). Few such early interventions have reported outcome data at present.

What is the energy gap responsible for childhood obesity?

Obesity is the result of a chronic excessive positive energy balance. Although a small positive energy balance is required for normal growth and development, the magnitude of positive energy balance typical of contemporary populations of children and adolescents can be much greater (33). Evidence of the magnitude of positive energy balance would be helpful in determining how substantive the lifestyle changes required by preventive interventions need to be and also be useful in understanding obesity etiology (33, 34). Because preventive interventions aim to reduce this energy gap, it is surprising that interventions are not tailored to the size of the gap more often and more explicitly. However, empirical studies of the magnitude of positive energy balance are in fact very scarce (33, 35, 36) and have tended to use conservative methods for estimating the energy gap. Even where the energy gap appears to be small, this may not be fixed in time (secular trends in the energy gap have not been studied), and the size of the gap may differ quite markedly among subgroups defined by ethnicity or initial weight status (33). Heterogeneity in the energy gap within populations (34), with a much larger gap in the overweight (33), for example, would provide a basis for more tailored interventions. Designing interventions to fit the energy gap being experienced by each population may therefore be helpful in future, and estimates of the size of the energy gap can be made relatively easily from long-term changes in body composition in children and adolescents participating in cohort studies.

Integrating new evidence of etiology and prevention into preventive interventions and policy

Increasing interest in the topic of child and adolescent obesity has led to a rapid increase in the evidence base. One important challenge to the development of evidence-based preventive interventions is the issue of how to integrate new evidence into the existing body of evidence. As a general rule, research-based interventions and policies should be guided by the overall body of evidence, based on the amount, quality, and consistency of the evidence as summarized in systematic reviews and critical appraisals (21, 26). Researchers and policymakers in the United Kingdom have, in contrast, been tempted to see single studies as a reasonable basis for a paradigm shift in prevention; this is particularly problematic when single studies are inconsistent with the existing body of evidence or when they are not replicated in subsequent studies.

Two recent examples from etiological studies in the United Kingdom are worthy of note. First, the question of the extent to which parent-child obesity associations are sex specific. One recent study of a small cohort in the southwest of England suggested that such associations were highly sex specific (37), and an accompanying editorial hailed this study as a paradigm shift with major practical consequences for future obesity prevention strategy (38). However, this small single study was inconsistent with the larger body of evidence, including a much larger cohort from the southwest of England (39) born at around the same time, and so the extent to which it should induce a paradigm shift is questionable.

The second example concerns the question of when to carry out obesity prevention interventions, i.e., at what stage of the life course. This is a complex issue that depends on a number of factors, and preventive interventions should ideally be at least considered across the life course. Recent UK evidence focused on identifying periods of the life cycle when excessive weight gain is most marked to inform the issue of when to set preventive interventions (40, 41). A recent review found that researchers were rarely explicit about the rationale for setting preventive interventions at one life stage rather than another (42). One study of a small cohort from southwest England suggested that most excess weight gain had occurred by the age of school entry, and the authors suggested that preventive efforts should therefore be focused on the preschool period (40). However, this observation was not replicated in a study of a much larger cohort from southwest England studied at around the same time (with a difference in birth periods of 3 y between cohorts), which found that mid to late childhood was characterized by greatest excess weight and BMI gain (41). In addition, annual incidence of new cases of obesity in the United Kingdom, at least for children born in the early 1990s, appears to be much higher in mid to late childhood (43) compared with early childhood and adolescence. This observation creates further doubts about the suggestion that the period of early childhood should be prioritized for preventive interventions in the United Kingdom. Moreover, greater awareness of age-related and/or period-related differences or changes in the incidence and persistence of obesity would be helpful in deciding when in the life course to set interventions (43). Avoiding difficulties in the prioritizing of intervention efforts and making prioritization more evidence based in future will depend on a more cautious approach that recognizes that paradigm shifts are rare and that integrates new evidence into the existing evidence base more explicitly.

Public health policy in childhood obesity prevention should be developed cautiously and logically, informed by the body of evidence and useful logic frameworks that are specific to childhood obesity (20) and others not specific to obesity but likely to be helpful. Policy should not be based on single studies, and paradigm shifts are likely to be rare. This is not an argument for inaction: sufficient evidence has existed for some time to give cause for concern over the epidemic of child and adolescent obesity; evidence is adequate at present to inform public health interventions (44).

Informing policy by evidence

The UK experience has been that matching the research evidence on child and adolescent obesity to national and local policy responses has been difficult. A number of problems have become apparent for obesity researchers, notably the lack of awareness of reviews on etiology, weighting single studies excessively relative to the rest of the body of evidence, identifying target behaviors and model intervention programs that should be prioritized, disseminating research syntheses to policymakers, adapting intervention models from other settings such as the United States, failure to consider explicitly the stage of the life course most appropriate for interventions, and influencing policymakers in the face of what appears to many policymakers to be a simple energy balance problem that must have simple solutions. More effective matching of evidence to policy would be desirable if policy responses are to become more logical and more effective in future. The field of policy development for obesity prevention is growing rapidly and has local, national, and global policy dimensions, but is beyond the scope of the current review (45, 46).

Conclusions

This review highlights a number of challenges for the development of more effective, evidence-informed strategies aimed at prevention of child and adolescent obesity. The possible leveling off of the childhood obesity epidemic in some nations should not create complacency that the problem has been solved. Preventive interventions and prevention policy should be informed more explicitly by systematically reviewed and critically appraised summaries of the evidence and by the translation of the most important evidence of etiology into interventions and with greater use of the most important prevention models.