Prevalence of Childhood and Adult Obesity in the United States, 2011-2012 FREE

Obesity and childhood obesity, in particular, are the focus of many public health efforts in the United States.^1,2 New regulations have been implemented by the US Department of Agriculture for food packages in the Special Supplemental Nutrition Program for Women, Infants and Children,³ the Centers for Disease Control and Prevention (CDC) has funded state- and community-level interventions,⁴ and there have been numerous reports and recommendations issued by the Institute of Medicine,⁵ the US Surgeon General,⁶ and the White House.¹ Although the prevalence of obesity in the United States is high, with one-third of adults and 17% of children obese,⁷ it appears to have leveled off between 2003-2004 and 2009-2010.^8,9 Given the focus of public health efforts on obesity, surveillance of trends in obesity remains important.

The purpose of this study is to provide the most recent national estimates of childhood obesity and analyze trends in childhood obesity between 2003 and 2012. In addition, as a follow-up to recently published estimates of adult obesity,¹⁰ more detailed trend analyses by age are presented. Estimates of overweight prevalence are also presented but are not analyzed for trends because overweight prevalence showed no trends between 1960 and 2010.¹¹ Analyses are based on measured values of weight and height from the 2011-2012 National Health and Nutrition Examination Survey (NHANES).

A detailed description of NHANES is available elsewhere.¹² It is a cross-sectional, complex, probability sample of the US noninstitutionalized population, with both interview and examination components. Although each year represents a nationally representative sample, data are released every 2 years to ensure adequate sample size for analyses and protect confidentiality. The most recent data presented in this analysis are from 2011-2012, and trends were analyzed between 2003-2004 and 2011-2012 with 5 time periods. In 2011-2012, 70% of individuals selected for the survey participated in its examination component compared with 76% in 2003-2004, 77% in 2005-2006, 75% in 2007-2008, and 77% in 2009-2010.¹³

NHANES includes oversampling of different subpopulations, including specific race/Hispanic origin groups. In 2011-2012, Hispanics, non-Hispanic blacks, and non-Hispanic Asians were oversampled. Oversampling of non-Hispanic Asians is a new addition to the NHANES survey. In the collection of NHANES data, survey participants were asked to report their race and whether they were of Hispanic origin. Individuals who reported they were of Hispanic origin were categorized as Hispanic regardless of their race. If they were not of Hispanic origin, they were categorized into 4 different groups (non-Hispanic white, non-Hispanic black, non-Hispanic Asian, and other). All non-Hispanic individuals who reported more than 1 race group were included in the “other” group, which was included in calculations of the total population but not reported separately.

During the NHANES physical examination, weight and height were measured in a standardized fashion.¹² For children younger than 2 years, recumbent length was measured instead of standing height. In 2011-2012, 1.5% (141) of examined participants were missing weight and height or length measurements.

The National Center for Health Statistics Research Ethics Review Board approved NHANES. Written parental permission was obtained for minors younger than 18 years. Children aged 7 to 17 years were also asked to provide documented assent. Written consent was obtained for all adults aged 18 years and older.

Weight status for individuals aged 2 years and older was defined with body mass index (BMI, measured as weight in kilograms divided by height in meters squared) rounded to 1 decimal place. In children and adolescents aged 2 to 19 years, obesity was defined as a BMI at or above the 95th percentile of the CDC sex-specific BMI-for-age growth charts from 2000.^14,15 Overweight was defined as a BMI between the 85th and 95th percentiles. The estimates are presented as greater than or equal to both the 85th and 95th percentiles. Because there is no recommended definition of obesity in children younger than 2 years, excess weight was defined as a weight for recumbent length at or above the 95th percentile on the CDC sex-specific weight for recumbent length growth charts, similar to what has been presented in previous analyses.⁹ The World Health Organization (WHO) growth standards have been recommended to monitor growth in children younger than 2 years in the United States.¹⁶ Consequently, the percentage of infants and toddlers at or above the 97.7th percentile of WHO weight for recumbent length growth standards¹⁶ also is presented.

In adults aged 20 years and older, obesity was defined as a BMI greater than or equal to 30. Obesity was further divided into grade 1 (BMI 30-34), grade 2 (BMI 35-39), and grade 3 (BMI ≥40). Overweight among adults was defined as a BMI greater than or equal to 25 but less than 30. The estimates are presented as BMI greater than or equal to 25, 30, 35, and 40.

Prevalence estimates are presented with 95% CIs, which were constructed with the logit transformation.¹⁷ Differences in prevalence between male and female participants overall in 2011-2012 were tested separately in infants, children, and adults at the α = .05 level, using 2-sided t tests. To test for race/Hispanic origin and age differences in 2011-2012, the null hypothesis of no race/ethnic or age difference was first tested with an analysis of variance. If this hypothesis was rejected, tests for differences between any 2 subgroups were conducted with t tests. Tests for differences by race/Hispanic origin were evaluated by comparing the 4 race/Hispanic origin groups described above. Tests for differences by age in children were evaluated with the following comparisons: aged 2 to 5 vs 6 to 11 years, 2 to 5 vs 12 to 19 years, and 6 to 11 vs 12 to 19 years. Similarly, in adults comparisons were made between aged 20 to 39 and 40 to 59 years, 20 to 39 and 60 years or older, and 40 to 59 and 60 years or older. P values for test results are shown in the text but not the tables. Adjustments were not made for multiple comparisons.

Trends in the unadjusted prevalence of high weight for recumbent length or obesity from 2003-2004 through 2011-2012 (using 5 time periods: 2003-2004, 2005-2006, 2007-2008, 2009-2010, and 2011-2012) were tested with t statistics and orthogonal contrast matrices. Trends were analyzed separately for infants, children, and adults because of different definitions. P values for trends, along with the absolute change in obesity prevalence between 2003-2004 and 2011-2012, are reported.

Trends in high weight for recumbent length or obesity were also tested in logistic regression models adjusted for age and race/Hispanic origin with the Satterwaite adjusted F statistic.¹⁸ We found a significant interaction between survey period and age among youth and adult women, so we conducted sex- and age-specific logistic regression models of obesity adjusted for race/Hispanic origin (results shown in the Supplement). Survey period was treated as a continuous variable.

Analyses were conducted with SAS version 9.3 and SUDAAN version 11.0. All analyses used NHANES examination sample weights that adjust for nonresponse, noncoverage, and unequal probabilities of selection. Standard errors were estimated with Taylor series linearization to take into account the complex sample design. Pregnant females were excluded from all analyses. Obesity estimates for total adults aged 20 years and older were age standardized to the projected estimates of the 2000 US Census by the direct method, using the age groups 20 to 39 years, 40 to 59 years, and 60 years and older. Crude estimates of obesity among all adults are also presented.

There were 9120 persons with measured weights and heights (or recumbent length) in NHANES 2011-2012. More than half of these (5181) were adults aged 20 years and older and 584 were infants and toddlers. Of the 9120 respondents, 1179 were non-Hispanic Asian. Detailed sample sizes by sex, age, and race/Hispanic origin are shown in Table 1.

The prevalence of high weight for recumbent length among infants and toddlers from birth to aged 2 years was 8.1% (95% CI, 5.8%-11.1%) (Table 2). There was a significant difference between boys and girls; 5% of boys (95% CI, 3.5%-7.0%) and 11.4% of girls (95% CI, 7.3%-17.4%) had high weight for recumbent length (P = .03). There were no significant differences between the race/Hispanic origin groups (P = .32). When WHO growth charts were used to define excess weight for recumbent length, 7.1% (95% CI, 4.9%-10.3%) of infants and toddlers had high weight for recumbent length (Table 2).

In 2011-2012, 31.8% (95% CI, 29.1%-34.7%) of youth were either overweight or obese, and 16.9% (95% CI, 14.9%-19.2%) of youth were obese (Table 3). In 2011-2012, there was no difference in obesity prevalence between boys and girls (P = .77), but there were race/Hispanic origin (P = .001, analysis of variance) and age (P < .001, analysis of variance) differences. The prevalence of obesity was lower in non-Hispanic Asian youth (8.6%; 95% CI, 5.7%-12.7%) than in youth who were non-Hispanic white (14.1%; 95% CI, 10.8%-18.2%; P = .04), non-Hispanic black (20.2%; 95% CI, 16.7%-24.2%; P < .001), or Hispanic (22.4%; 95% CI, 20.3%-24.6%; P < .001). The prevalence of obesity was also lower among non-Hispanic white youth compared with non-Hispanic black youth (P = .048) and Hispanic youth (P < .001). There was no difference in prevalence between non-Hispanic black youth and Hispanic youth (P = .31). More than 8% (8.4%; 95% CI, 5.9%-11.6%) of 2- to 5-year-olds were obese compared with 17.7% (95% CI, 14.5%-21.4%) of 6- to 11-year-olds (P < .001) and 20.5% (95% CI, 17.1%-24.4%) of 12- to 19-year-olds (P < .001). There was no difference in obesity prevalence between 6- to 11- and 12- to 19-year-olds (P = .22). Additional information on the unweighted number of participants with high weight for recumbent length or who were obese is detailed in eTable 1 in the Supplement. The percentage of adolescents aged 12 to 19 years with BMI greater than or equal to 30, meeting the adult definition of obesity, was 13.9% (95% CI, 10.9%-17.7%) in 2011-2012.

Age-adjusted and crude prevalence estimates of overweight and obesity among adults by sex, age, and race/Hispanic origin are shown in Tables 4 and 5. The age-adjusted estimates indicate that more than two-thirds (68.5%; 95% CI, 65.2%-71.6%) of adults were either overweight or obese, 34.9% (95% CI, 32.0%-37.9%) were obese, and 6.4% (95% CI, 5.2%-7.7%) were extremely obese (grade 3 obesity) in 2011-2012. There were significant differences by sex, age, and race/Hispanic origin. For example, the prevalence of grade 3 obesity differed by sex (P = .004), with the prevalence higher in women (8.3%; 95% CI, 6.9%-9.8%) than men (4.4%; 95% CI, 2.8%-6.8%), and by age (P = .03), with the prevalence highest among the middle-age group (7.7%; 95% CI, 6.2%-9.4%) compared with 20- to 39-year-olds (5.6%; 95% CI, 4.4%-7.1%) and adults aged 60 years or older (5.6%; 95% CI, 3.8%-8.0%). The prevalence of grade 3 obesity also varied by race/Hispanic origin (P < .001), with the highest prevalence among non-Hispanic black adults (12.1%; 95% CI, 10.3%-14.0%) compared with 5.6% (95% CI, 4.1%-7.5%) in non-Hispanic white adults, 0.9% (95% CI, 0.3%-2.3%) in non-Hispanic Asian adults, and 5.8% (95% CI, 4.7%-7.1%) in Hispanic adults.

Results of linear unadjusted trend tests between 2003-2004 and 2011-2012 are shown in Table 6. Among infants and toddlers from birth to aged 2 years, there was no significant change in high weight for length prevalence (−1.4 percentage points; P = .72). Among children and adolescents aged 2 to 19 years, there was no significant change overall (−0.2 percentage points; P = .78), but there was a significant decrease in obesity prevalence among 2- to 5-year-old children (−5.5 percentage points; P = .03). Among adults, there was no significant change in obesity prevalence in the total population (+2.8 percentage points; P = .09), but there was an increase in prevalence among adults aged 60 years and older (+4.4 percentage points; P = .004).

Sex- and age-specific unadjusted trend results are shown in eTable 2 in the Supplement. In these analyses, the only significant trends were found in women aged 60 years and older (6.6 percentage point increase: 31.5% to 38.1%; P = .006); there was no significant trend among men aged 60 years and older (P = .25). Among girls aged 2 to 5 years, there was a 5.5 percentage point decrease in obesity prevalence, although it was not significant (P = .07). In sex and race/Hispanic origin adjusted analyses of trends, results were similar to those in the unadjusted analyses (eTable 3 in the Supplement). Tests for an interaction between survey period and age found an interaction in children (P = .03) and women (P = .02).

In 2011-2012, the prevalence of obesity in the United States was 16.9% in youth and 34.9% in adults. The overall prevalence of obesity among youth remained unchanged compared with that in 2009-2010 (16.9%),⁹ and there was no significant change since 2003-2004. Similarly, there was no significant change in obesity prevalence among adults between 2003-2004 and 2011-2012. In subgroup analyses, the prevalence of obesity among children aged 2 to 5 years decreased from 14% in 2003-2004 to just over 8% in 2011-2012, and the prevalence increased in women aged 60 years and older, from 31.5% to more than 38%. Because these age subgroup analyses and tests for significance did not adjust for multiple comparisons, these results should be interpreted with caution.

Newly available estimates of obesity among non-Hispanic Asians show that almost 9% of non-Hispanic Asian youth and 10.9% of non-Hispanic Asian adults were obese according to BMI cut points. BMI, however, is not a perfect measure of body fat.¹⁹ It is highly correlated with body fat but does not account for differences in distribution of body fat or differences between race/Hispanic origin groups, sex, and age. Different sex, age, and race/ethnicity groups may have different body fat at the same BMI.^20,21 For example, some research suggests that Asians may have more body fat than whites, especially at lower BMIs.²² Risk of morbidity and mortality may not be completely captured by BMI.²³ Given concerns that health risks begin at a lower BMI among Asians compared with others, some Asian countries have adopted lower cut points of BMI to define overweight or obesity,²⁴ and although WHO has recommended continuing to use the standard cutoffs for international comparisons, a WHO expert committee has recommended lower cutoffs for Asians as points for “public health action.”²⁵

The prevalence of high weight for recumbent length with the CDC growth charts is slightly higher than the prevalence with WHO growth standards. Overall, among infants and toddlers from birth to aged 2 years, 8.1% were at or above the 95th percentile of sex-specific CDC weight for recumbent length growth charts, whereas 7.1% were at or above the 97.7th percentile on the corresponding WHO growth charts. For all groups (sex and race/Hispanic origin), the estimates were slightly higher with the CDC growth charts than WHO growth standards. The 2 sets of growth charts differ in that the CDC charts represent a growth reference based on the general US population in the 1970s, 1980s, and early 1990s.¹⁵ WHO growth standards, on the other hand, represent growth of children in select settings around the world with optimal feeding practices, among other factors.²⁶

Recent decreases in the prevalence of obesity have been reported in some populations of youth in the United States. Between 2008 and 2011, significant decreases were reported among low-income preschool-aged children participating in federal nutrition programs in 18 states and the US Virgin Islands. The absolute decreases ranged from 0.3 to 2.6 percentage points.²⁷ Similarly, between 2003 and 2008 a decrease in obesity prevalence among children younger than 6 years was reported in a multisite pediatric practice in eastern Massachusetts.²⁸ The decrease we observed among preschool-aged children is consistent with the decreases observed in low-income children in the United States overall and in some states individually.

Analyses of trends in obesity prevalence among middle and high school students have shown mixed results. Among public middle school students in New York City, a recent analysis found a decrease in obesity prevalence between 2006-2007 and 2010-2011.²⁹ Other researchers using a school-based survey found an increase in obesity prevalence among US adolescents between 2001-2002 and 2005-2006 but no change between 2005-2006 and 2009-2010.³⁰ In addition, data from the Youth Risk Behavior Surveillance System showed an increase in the prevalence of obesity between 1999 and 2011 but no change between 2009 and 2011.³¹ NHANES results among middle school– and high school–aged children show no significant change in prevalence between 2003-2004 and 2011-2012.

Data from many countries have shown a decline or stabilization of obesity levels, especially in children.³² For example, in Germany a study of children aged 4 to 16 years found a significant decline in overweight or obesity between 2004 and 2008 among the youngest children (aged 4-7 years) and a stabilization in other ages.³³ In England, the rate of increase in adult BMI has decreased.³⁴

Analysis of time trends depends on what is chosen as the initial point of examination. In this analysis, we selected 2003-2004 as the starting point because previous analyses had shown no change in sex- or race/Hispanic origin–specific trends in obesity prevalence between 2003-2004 and 2009-2010.^8,9 The selection of the initial point can have an effect on findings. For example, analyses of childhood obesity trends between 1976-1980 and 2011-2012 show an increase in childhood obesity, whereas trends between 2003-2004 and 2011-2012 do not.

In the current analysis, trend tests were conducted on different age groups. When multiple statistical tests are undertaken, by chance some tests will be statistically significant (eg, 5% of the time using α of .05). In some cases, adjustments are made to account for these multiple comparisons, and a P value lower than .05 is used to determine statistical significance. In the current analysis, adjustments were not made for multiple comparisons, but the P value is presented.

Overall, there have been no significant changes in obesity prevalence in youth or adults between 2003-2004 and 2011-2012. Obesity prevalence remains high and thus it is important to continue surveillance.