Prevalence of obesity and severe obesity in US children, 1999-2014

Introduction

Obesity prevalence, which has increased in all age groups over the past 30 years [1], may be leveling off. There have been multiple reports of decreases in obesity in specific populations and leveling off in others [1-5]. However, these studies have focused largely on low-income young children or children from defined geographic areas. Reports using nationally representative data have shown no changes in the prevalence of overweight or obesity, but most have not examined severe obesity [1, 6]. Increasingly, severe obesity is being recognized as a distinct classification of excess weight. Originally defined as the 99th percentile BMI for age and gender [7], percentile classifications have been noted to perform poorly in statistical tests, and their use has not been recommended [8]. Flegal et al. suggested use of percentages above the 95th percentile as a better means of quantifying children at higher levels of obesity, and this was recommended for use by the American Heart Association [8, 9]. Using nationally representative data, this designation of severe obesity finds associated cardiometabolic risk factors at higher prevalence [10]. Other studies have also noted significant cardiometabolic risk, though definitions of severe obesity vary [11-13].

Data from 2011 to 2012 demonstrated an increase in severe obesity among children and adolescents in the previous 14 years [14]. In particular, severe obesity (class II and III obesity) [8] has significantly increased in adolescents and non-Hispanic black children. Ongoing surveillance and analysis of obesity prevalence are paramount in order to best gauge progress of individual and public health efforts. Given the discrepancy between changes in overall rates of obesity in children and rates of severe obesity, our objective was to provide the most up-to-date data on the current national prevalence of obesity and severe obesity among children and adolescents in the United States. In addition, we aimed to examine the changes in prevalence trends for the time period 1999 to 2014.

Methods

We are using methods and analysis similar to previous studies [10, 14], detailed in brief as follows.

Results

Table 1 describes differences in obesity prevalence by age, sex, and race for the 2013–2014 sample: 33.4% of children met criteria for overweight, 17.4% for class I obesity, 6.3% for class II obesity, and 2.4% for class III obesity. These categories are not mutually exclusive (e.g., the 17% with class I obesity are included in the 33% with overweight). Older children had greater prevalence of all classes of obesity compared with younger children. Point estimates for all classes of obesity were larger in 2013–2014 compared with 2011–2012, but these were not significant (Table 2). A clear, statistically significant increase in all classes of obesity continued from 1999 through 2014 (Figure 1). Detailed time trends by sex and age are available in Table 3 and by sex and race in Table 4. Most notably, all classes of obesity have increased over time for adolescents of both sexes. Changes over time are similar by race, though prevalence is consistently higher among black and Hispanic children.

Discussion

There is no evidence of a change in obesity prevalence in any age group, despite substantial clinical and policy efforts targeting the issue. Prevalence estimates are not significantly changed in 2013–2014 than 2011–2012 for any obesity class. Although we do not have adequate power to detect small differences in prevalence, none of the estimates decreased from 2012 to 2014. However, the upward trend in prevalence beginning in 1999 continues through 2014.

Although the prevalence of obesity in children is regularly reported, as new data become available, there are several critical reasons for ongoing examination of obesity prevalence and trends in prevalence. First, additional years of data allow for a more nuanced examination of trends. Reporting trends using all available data improves the ability to assess whether changes in prevalence are anomalous or genuine trajectory shifts. For example, the prevalence of obesity in children was significantly higher in 2003–2004 than in the previous 2 years. Only with the inclusion of several previous years and following years is it apparent that the greater prevalence did not represent a long-term trend. The significant decline previously reported in prevalence for 2- to 5-year olds for 2003–2012 is not evident in our results, for girls or boys, when using all data from 1999 to 2014. There is no specific time frame that can be considered ideal for tracking changes; only through consistent monitoring over time can transient dips or peaks in prevalence be sorted out.

Understanding the ongoing trends in obesity is also very important for public health and policy makers. While our study is unable to specifically link any policies or interventions to changes in obesity prevalence, our results fail to show an indication that current policies have had a sweeping effect on obesity, though it is always possible that the policies were effective and other worsening simultaneous trends were at play. Public health tracking of obesity is helpful in informing policies, titrating limited resources, determining subgroups at risk, and as context for determining other potentially related trends such as social inequality, environmental exposure, and nutrition transition related to urbanization [16]. Without tracking, we cannot assess efficacy of public policies or the aggregation of individual efforts.

Trends in the prevalence of smoking provide an important comparison to demonstrate the use of long-term trends. Smoking rates declined significantly over the past 50 years. However, the interval years did not always show a decline, including many years with no apparent decline. By looking at the complete picture, we are able to see that the leveling off of smoking rates did not persist, and the decline has continued [17, 18]. Only with ongoing data will we be able to recognize whether the apparent leveling off of obesity represent real change or whether prevalence will continue to rise.

By including severe obesity, our results also highlight particular areas of concern. In 2013–2014, nearly 10% of adolescents met criteria for class II obesity, and nearly 5% also met criteria for class III obesity. As recently shown, these higher obesity levels are associated with increased cardiometabolic risk [10], and there is little evidence that severe obesity can be addressed through population-level policies alone. While there may be hope a child's obesity resolves over time, the bulk of scientific literature demonstrates childhood overweight and obesity adversely affect adult morbidity and mortality [19]. One exception is bariatric surgery. While this option is both costly in the short term, not available in many centers, and certainly a high-intensity treatment program, the prevalence of severe obesity and the success of the procedures [20, 21] suggest that programmatic and policy efforts should address the availability of bariatric surgery at the same time as other options are under research. Policies to improve outcomes of children with severe obesity will likely focus on ensuring access to appropriate clinical interventions, in addition to the population-level interventions useful for less severe forms of excess weight.

There are important limitations to our findings, primarily related to the risk of inferential error. We are unable to state affirmatively that there is no change in the prevalence of overweight and obesity, given the available data. However, the differences reported are unlikely to represent clinically significant changes in obesity prevalence, either increasing or decreasing. For certain subgroups, particularly young children, the available sample size further limits our ability to draw inferences from our findings.

Despite any limitations to inferential findings, our reports of the prevalence of obesity in 2013–2014 are critical to ongoing research in childhood obesity, particularly severe obesity. Researchers need current estimates of prevalence and variation in order to conceive and design studies. These data provide a reference point for comparison with other populations that are not nationally representative.

The 4.5 million children and adolescents with severe obesity will require novel and intensive efforts for long-term obesity improvement [22, 23]. With scarce resources, and increasing costs of comorbid conditions [24], there is an urgent need for targeted interventions to stem the rise in severe obesity among children, in addition to policies and clinical efforts designed to prevent obesity. Policy efforts are yet to yield substantive changes in obesity prevalence [25, 26], but few have specifically targeted severe obesity. Unfortunately, many clinical interventions are limited in their success, and limited availability and costs prevent widespread use of those that are effective [23]. Coordinated efforts, through integration of clinical and community systems, and supported by policies addressing population-level and individual needs, are necessary to address a complex, recalcitrant problem [27].