Background and Objective:
The United States has experienced a large increase in the prevalence of obesity since the 1970s. Our objective was to describe recent trends in obesity and abdominal obesity among adults in the United States.
Trend study of cross-sectional studies.
We used data from up to 22 872 men and non-pregnant women aged 20 years from the National Health and Nutrition Examination Survey (NHANES) 1999–2008.
Main Outcome Measures:
Main outcome measures are mean body mass index and waist circumference, percentages of obesity and abdominal obesity. Obesity was defined as a body mass index 30 kg m−2, and abdominal obesity was defined as a waist circumference 102 cm in men and 88 cm in women.
In men, the age-adjusted mean body mass index, mean waist circumference, and prevalence of obesity and abdominal obesity were 27.8 kg m−2, 99.1 cm, and 26.9 and 37.8%, respectively, during 1999–2000 and 28.5 kg m−2 (P trend=0.001), 100.8 cm (P trend=0.002), and 32.0 (P trend=0.001) and 43.7% (P trend=0.002), respectively, during 2007–2008. In women, the age-adjusted mean body mass index, mean waist circumference, and prevalence of obesity and abdominal obesity were 28.2 kg m−2, 92.2 cm, and 33.2 and 55.8%, respectively, during 1999–2000 and 28.6 kg m−2 (P trend=0.181), 94.9 cm (P trend=0.006), and 35.2 (P trend=0.180) and 61.8% (P trend=0.036), respectively, during 2007–2008. Significant linear trends for increasing prevalence of obesity were noted among men with the least and most education.
Between 1999 and 2008, both obesity and abdominal obesity increased in men, and abdominal obesity increased in women.
The age-adjusted prevalence of obesity among United States adults aged 20–74 years rose from 13.4% during 1960–1962 to 30.9% during 1999–2000.1, 2 Because obesity is a risk factor for many chronic conditions such as diabetes,3 this development has elicited a great deal of worry about the future of the nation's health.4 In cadence with the escalation of the prevalence of obesity, the medical cost attributed to obesity has skyrocketed with the most current estimate placing the annual cost at $147 billion.5
The trajectory of the obesity epidemic in the United States and its potential impact on the nation's health are ongoing concerns. Overweight and obesity constitute one of the nation's ten leading health indicators, and a Healthy People 2010 objective calls for a prevalence of obesity of 15%.6, 7 Our objective was to examine trends in mean body mass index, mean waist circumference, and prevalence of obesity and abdominal obesity among adults in the United States from 1999 through 2008. Furthermore, because previous findings from the United States and other countries suggest that these trends may vary by level of socioeconomic status, we also present trends by level of socioeconomic status, particularly by level of educational status.
Materials and methods
We used data from the National Health and Nutrition Examination Survey (NHANES) 1999–2008.8 Each 2-year cycle included a national sample—recruited using a multistage, stratified sampling design—that was designed to be representative of the civilian, non-institutionalized population in the United States. Participants were interviewed at home and were invited to attend a mobile examination center, where they were asked to complete additional questionnaires, undergo various examinations and provide a blood sample. The study received approval for the use of human subjects, and participants were asked to sign an informed consent form. Details about the survey can be found elsewhere.8
Body measurements were performed using standardized methods and equipment. Body mass index was calculated from measured weight and height. Weight was measured on an electronic weight scale. Participants wore underwear, disposable paper gowns and foam rubber slippers. Height was measured with an electronic stadiometer to the nearest millimeter. The waist circumference was measured with a steel measuring tape to the nearest 0.1 cm at the high point of the iliac crest at minimal respiration. Obesity was defined as a body mass index 30 kg m−2, and abdominal obesity was defined as a waist circumference 102 cm in men and 88 cm in women.9
Two measures of socioeconomic status, educational attainment and poverty income ratio, were included in the analyses. Three levels of educational status were created: below twelfth grade, high school graduate or equivalent, and above twelfth grade. The poverty income index is the ratio of family income to the official income level based on a family's composition and size. Values below 1 indicate that a family's income is below the official poverty level.
Total energy intake was obtained from a single 24-h dietary recall. For the years 1999–2000 and 2001–2002, only a single 24-h dietary recall was collected. For the years 2003–2004 and 2005–2006, two 24-h recalls were collected. To maximize comparability with the first two cycles, we only used the data for the first recall. Data for the 2007–2008 cycle were not yet available.
Two measures of physical activity were examined: the percentage of participants who participated in moderate or vigorous physical activity during the past 30 days, and the percentage of participants who reported walking or bicycling to and from work or school or to do errands during the past 30 days. For the 2007–2008 cycle, the questions about physical activity were changed substantially and were not analyzed. In addition, we estimated the percentages of participants who watched television or videos or used a computer for >2 h per day. For the years 1999–2000 and 2001–2002, a single question was used to ascertain the time spent in these sedentary activities. For the years 2003–2004 and 2005–2006, one question was asked about time spent watching television and videos, and a second question was asked about computer use. Because of concerns about the comparability of the questions over time, trends for these statistics were not tested. For the 2007–2008 cycle, questions about screen time were not asked of adults.
We limited the analyses to men and non-pregnant women aged 20 years. For the years 2007–2008, the previous upper age limit of 59 years for determining pregnancy status was lowered to 44 years. Age adjustment to the projected year 2000 US population using three age groups (20–39 years, 40–59 years, and 60 years) was performed with the direct method. For race or ethnic-specific estimates, we only present results for the three major groups: Whites, African Americans and Mexican Americans. Trends were examined with orthogonal linear contrasts, and with linear and log-binomial regression analysis. Analyses were conducted with SAS (SAS Institute Inc., Cary, NC, USA) and SUDAAN (software for the statistical analysis of correlated data); Research Triangle Institute, Research Triangle Park, NC, USA).
Of 26 246 participants aged 20 years, 24 693 attended the mobile examination center. After excluding 1154 pregnant women, 22 872 adults (11 486 men and 11 386 women) had measurements for weight and height, and 22 252 adults (11 198 men and 11 054 women) had a waist measurement.
The age-adjusted mean body mass index increased from 28.0 kg m−2 during 1999–2000 to 28.5 kg m−2 during 2007–2008 (Ptrend=0.017; Table 1). The regression coefficient adjusted for age, gender and race or ethnicity was 0.08058 kg m−2 per year (P=0.017). A significant linear trend was present for men but not for women. Furthermore, increasing trends for mean body mass index were noted for all demographic subgroups among men.
Significant increases in the age-adjusted mean waist circumference were noted for all adults, men and women (Table 1). The age-adjusted mean waist circumference was 2.1 cm greater among all adults, 1.7 cm greater among men and 2.6 cm greater among women during 2007–2008 than during 1999–2000. Significant linear trends occurred in all three groups. The regression coefficient among all adults adjusted for age, gender and race or ethnicity was 0.29296 cm per year (P=0.001). The largest increases in mean waist circumference occurred among young women and Mexican American women (+3.6 cm in both groups).
During the 8-year study period, the age-adjusted prevalence of obesity trended upwards (Table 2). The trend was significant among men and most subgroups of men, but not women. The prevalence ratio for all adults adjusted for age, gender and race or ethnicity was 1.02 (95% CI: 1, 1.03). During 2007–2008, the prevalence of obesity was higher among women than men (P=0.010) and was especially high among African American women.
Significant increasing trends in abdominal obesity were present for all adults (adjusted prevalence ratio=1.02; 95% CI: 1.01, 1.03), men (adjusted prevalence ratio=1.02; 95% CI: 1.01, 1.04) and women (adjusted prevalence ratio=1.01; 95% CI: 1.00, 1.02). During 2007–2008, the age-adjusted prevalence of abdominal obesity was considerably higher among women than men (P<0.001), and was particularly high among women aged 60 years and among African American and Mexican American women.
When the data were stratified by educational status, a significant linear trend for increasing body mass index was observed only among the most educated men (Table 3). Significant linear trends for waist circumference were observed among the most educated men as well as all three groups of women. When comparing the means for 1999–2000 with 2007–2008, the largest increases in body mass index occurred among the most educated men and the least educated women, and the largest increase in waist circumference (+4.2 cm) occurred among the least educated women. When the data were stratified by levels of poverty income ratio, significant increasing linear trends for body mass index and waist circumference were observed among men, with poverty index ratios of 2 to <3 and 4 to <5 (Table 4). No significant linear trends for body mass index were present among any group of women. However, the age-adjusted waist circumference increased significantly among women with a poverty income ratio of 3 to <4 and increased almost significantly in several additional groups of women.
Significantly increasing trends in the age-adjusted prevalence of obesity were noted among the least and most educated men. No significant trends were observed among any groups of women (Table 5). Although the educational gradient with respect to the prevalence of obesity among men showed no particular pattern over time, the gradient among women tended to widen during the first three consecutive cycles and was largest during 2007–2008. Significantly increasing trends in abdominal obesity were observed among the two most educated groups of men, and near significant increases were observed among the two most educated groups of women.
From 1999 to 2006, mean energy intake increased among men but not women (Table 6). The percentage of adults who reported walking or bicycling to and from work or school or to do errands during the past 30 days showed no significant trends. The percentage of men and women who participated in moderate or vigorous physical activity during the past 30 days increased from 1999 to 2006.
The prevalence of obesity and abdominal obesity remains disturbingly high among adults in the United States, and our trend analysis shows that both may still be increasing among men. The percentage of women who were obese was highest during 2007–2008, but the trend analysis did not indicate that obesity increased significantly during the 8-year period. However, the trend analysis suggested that the prevalence of abdominal obesity among women increased significantly during the 8-year period despite relatively stable prevalence estimates during the last three study cycles.
Although sample sizes for each 2-year cycle were substantial, sampling variability remains a concern for any consecutive 2-year cycle. For subgroups with much smaller sample sizes, potential sampling variability is a greater concern.
Trends in the prevalence of obesity by educational attainment have varied among countries. In Denmark, the prevalence of obesity increased significantly among the least and most educated men and among the least educated women from 1982 to 1992.10 Whereas the educational gradient with respect to the prevalence of obesity widened among women, it did not do so among men. In Spain, the absolute increase in obesity among participants aged 45–64 years from 1987 to 1995–1997 was greatest among men and women with an elementary education.11 Among participants aged 25–44 years, the greatest increases occurred among men with a secondary education and women with an elementary education. In all, but one age and gender group, an inverse gradient between the prevalence of obesity and educational status was present. In Malmö, Sweden, the prevalence of obesity varied inversely with educational status only among women in 1986.12 The least educated men and women experienced the largest absolute increases in obesity during the next 8 years. Furthermore, the educational gradient with respect to obesity widened over time, especially among men. In Belgium, a clear gradient in the prevalence of obesity based on self-reported weight and height by level of education was present among women, but less so among men, during 3 calendar years.13 Absolute increases in the prevalence of obesity were more pronounced among men and women with a primary or lower secondary education. The educational gradient widened progressively over time among men but not women. In Switzerland, the prevalence of obesity increased in all educational groups from 1992 to 2007, and, furthermore, the educational gradient widened among men and women.14 In each year, the prevalence of obesity decreased as educational attainment increased. In Finland, the prevalence of obesity increased in each educational group and the absolute educational gradients changed little from 1978–1980 to 2000–2001.15 During each study period, the prevalence of obesity and educational status were inversely related in men and women.
The prevalence of obesity among US adults increased among participants at each of three levels of educational status from 1971–1974 to 1999–2000.16 Among women, those with the highest education experienced the largest absolute increase in the prevalence of obesity, whereas among men, the absolute increases were generally of a similar magnitude. Educational status and the prevalence of obesity were inversely related among women during all four periods, but among men only during one study period. The educational gradient regarding the prevalence of obesity among women narrowed after the survey in 1976–1980, but no clear pattern emerged among men. Our analysis of more recent data from NHANES showed an inverse association between educational status and the prevalence of obesity in most study cycles and a widening of the educational gradient with respect to obesity over time among women but not men. Among women, those with the least education experienced the largest increases in mean body mass index, waist circumference and prevalence of obesity. Among men, however, the largest increases in anthropometric measures occurred among the most educated participants.
Temporal changes in energy intake and expenditure in the population largely determine the trajectory of the trends in obesity. Among all adults and women, the mean energy intake did not change significantly during the first 8 years of the 10-year study period. Among men, however, the mean energy intake increased significantly and was 100 kcal higher during 2005–2006 than during 1999–2000. Participation in moderate or vigorous leisure-time physical activity increased significantly during the first 8 years, particularly among women. Thus, increased energy intake coupled with a small increase in participation in physical activity is consistent with an increase in obesity among men, and fairly stable energy intake coupled with increased participation in physical activity is consistent with a leveling of the prevalence of obesity among women.
The obesity epidemic has stimulated a great deal of thought about potential solutions for this seemingly intractable challenge. Finding the correct strategies to redress the imbalance between energy intake and expenditure for a significant slice of the population has proved challenging because of the numerous personal, familial, social, societal, cultural, governmental and environmental contributing factors. Increases in energy intake can explain the increases in the prevalence of obesity in the United States.17 Researchers and policy makers are focusing on dietary influences that are considered to be leading contributors to excess energy intake including the ubiquitous availability of cheap energy dense foods and beverages, growing portion sizes and increased eating of meals away from home. Facilitating access to nutritious foods by using methods such as changing the offerings in vending machines in schools and work sites,18 improving access to supermarkets and markets that offer increased selection of fruits and vegetables in underserved areas,19 promoting the consumption of family meals20 and reducing the amount of money spent on meals outside home21 are ideas that could contribute to reducing energy intake. Policy approaches include proposals to tax certain foods and beverages and forcing disclosure of nutritional information including caloric content of meals.22
Unfortunately, an information void exists about trends in physical activity in the United States during much of the last three decades. Nevertheless, it is commonly assumed that total energy expenditure has declined in the United States in part because of growth in occupations with low energy expenditure, expansion in mechanization, increases in sedentary activities and heavy use of motorized transportation. The role of the physical environment on patterns and levels of physical activity is drawing extensive scrutiny. Despite a large percentage of US adults not meeting recommendations for physical activity, a decreasing percentage of participants of the Behavioral Risk Factor Surveillance System from 1989 to 2002 reported not doing any leisure-time physical activity and a mounting percentage of participants reported engaging in regular physical activity in more recent years.23
The recent upward trend in obesity has all but ruled out the possibility that the Healthy People 2010 objective for the prevalence of obesity of 15% will be achieved. New goals for 2020 are in the offing. A goal of 15% could be achieved if the average energy intake could be lowered by ∼500 kcal per day or if levels of physical activity were to increase by an amount equivalent to walking about 110 min per day.17 The path to achieving such changes in energy intake and physical activity is unlikely to be an easy one. Our findings suggest that enhanced efforts towards identifying and implementing effective and multifaceted prevention strategies are warranted.
About this article
The findings and conclusions in this article are those of the authors and do not represent the official position of the Centers for Disease Control and Prevention.
Sexual Dimorphism in Cardiovascular Disease Risk and Risk Factors Among Vegetarians: an Exploration of the Potential Mechanisms
Current Atherosclerosis Reports (2019)