| |
Socioeconomic, Racial/Ethnic, and Geographic Characteristics: A Systematic Review and Meta-Regression Analysis
|
| |
| |
Epidemiologic Reviews Advance Access published online on May 17, 2007
Epidemiologic Reviews, doi:10.1093/epirev/mxm007
Youfa Wang and May A. Beydoun
From the Center for Human Nutrition, Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD
A doubling in American obesity took place in the two decades after the nation's bicentennial. In 1976, about 15.1% of Americans were obese, rising to 30.9% by 2000. If this trend continues, 75% of adults will be overweight or obese by 2015, 41% will be obese, and almost 25% of children and adolescents are expected to be overweight or obese.
ABSTRACT
This review of the obesity epidemic provides a comprehensive description of the current situation, time trends, and disparities across gender, age, socioeconomic status, racial/ethnic groups, and geographic regions in the United States based on national data. The authors searched studies published between 1990 and 2006. Adult overweight and obesity were defined by using body mass index (weight (kg)/height (m)2) cutpoints of 25 and 30, respectively; childhood "at risk for overweight" and overweight were defined as the 85th and 95th percentiles of body mass index. Average annual increase in and future projections for prevalence were estimated by using linear regression models. Among adults, obesity prevalence increased from 13% to 32% between the 1960s and 2004. Currently, 66% of adults are overweight or obese; 16% of children and adolescents are overweight and 34% are at risk of overweight. Minority and low-socioeconomic-status groups are disproportionately affected at all ages. Annual increases in prevalence ranged from 0.3 to 0.9 percentage points across groups. By 2015, 75% of adults will be overweight or obese, and 41% will be obese. In conclusion, obesity has increased at an alarming rate in the United States over the past three decades. The associations of obesity with gender, age, ethnicity, and socioeconomic status are complex and dynamic. Related population-based programs and policies are needed.
US adults
Overall patterns and age and gender disparities. Current situation.
Estimates of the national prevalence of overweight and obesity among adults based on the most recent NHANES data collected in 1999-2004 are shown in table 1. In 2003-2004, among men and women aged 20 years or older, approximately two thirds (66.3 percent) were overweight or obese, 32.42 percent were obese, and 4.8 percent were extremely obese (BMI >40 kg/m2). The combined prevalence increased with age. Of persons in the United States aged 60 years or older, more than 70 percent were overweight or obese, and the trend was similar for men and women. However, there was no such clear trend with respect to obesity. More men than women were overweight or obese (68.8 percent vs. 61.6 percent in 2001-2002) (32).
Trends.
The trends based on NHANES data are shown in table 2 and figure 1. Between 1971 and 2001, mean BMI increased by three units, from 24.4 to 27.6 kg/m2 in men and from 25.3 to 28.2 kg/m2 in women. During 1960-1980, prevalence increased slowly, but, since NHANES II (1976-1980), prevalence increased dramatically at an average annual rate of approximately 1 percentage point. The prevalence of obesity doubled between 1976-1980 and 1999-2000, increasing from 15.1 percent to 30.9 percent. The annual rate of increase was similar for men and women during this period. The rate of increase for the combined prevalence of overweight and obesity during that same period was smaller than that for obesity. Combined prevalence increased by about one third, from 47.4 percent to 64.5 percent. Among men, the rate of increase was similar across age groups. Among women, the group 20-34 years of age had the fastest increase.
Racial/ethnic disparities. Current situation.
Data from NHANES, BRFSS, and the Add Health study show large racial/ethnic differences, especially for women. Non-Hispanic Blacks had the highest prevalence. Minority groups (i.e., non-Hispanic Blacks and Mexican Americans) had a higher combined prevalence than non-Hispanic Whites by almost 10 percentage points. The corresponding prevalences in 2003-2004 were 76.1 percent and 75.8 percent versus 64.2 percent. The racial/ethnic differences among men were much smaller than among women. In 1999-2002, the combined prevalence and the prevalence of obesity among non-Hispanic Black women was 20 percentage points higher than among White women (77.2 percent vs. 57.2 percent, and 49.0 percent vs. 30.7 percent, respectively). Among non-Hispanic Black women aged 40 years or older, more than 80 percent were overweight or obese, and more than 50 percent were obese. In 1999-2002, the prevalence of extreme obesity among African-American women was more than twice that among White and Mexican-American women (13.5 percent vs. 5.5 percent and 5.7 percent).
The NHANES data do not include an adequate number of people with minority backgrounds other than non-Hispanic Black and Mexican American. The 2001 BRFSS data show that the prevalence of obesity was much lower among Asian Americans than other racial/ethnic groups (5 percent vs. approximately 30 percent) (33). More Native Americans (34.3 percent) were obese than Whites (21.8 percent), but the prevalence was similar to that for Blacks (34.8 percent) and Pacific Islanders (33.0 percent) (33). The Add Health study shows that young Asian adults had the lowest prevalence of obesity, particularly among young women (9 percent) (table 3).
Other studies also show that the prevalence of overweight and obesity among Asian Americans was much lower than the national average, but differences between different Asian groups were considerable. A recent review found that prevalence was highest among Native Hawaiians and Samoans (34). In a study based on self-reported height and weight data collected from 254,153 participants aged 18-59 years included in the 1992-1995 National Health Interview Survey of the six largest Asian-American racial/ethnic groups, researchers found that the combined prevalence in Asian Americans was less than half that of the national average, and the prevalence of obesity among Asian-American men was only one fifth that for all US men; among Asian-American women, prevalence ranged from only 1 percent to 4 percent, while the national average was 15 percent (table 4). Asians born in the United States were four times more likely to be obese than their foreign-born counterparts. Among the foreign born, more years in the United States were associated with higher risk (35). Similar findings among young people have been reported by the Add Health study (36, 37).
Trends.
In general, and based on NHANES data, the increase in prevalence of overweight and obesity was similar across racial/ethnic groups in both men and women over the past three decades, although they had different prevalences. Among women, between 1988-1994 and 1999-2002, the increase was slower for Mexican Americans than for non-Hispanic Whites and Blacks. The same rising trend occurred among Blacks although at a much higher baseline prevalence of obesity compared with that for Whites (figure 2).
FIGURE 2. Trends in the prevalence of obesity (body mass index 30 kg/m2) in US adults, by gender and ethnicity, National Health and Nutrition Examination Survey, 1971-2004. (Sources: 2, 54).
Central obesity. Current situation.
The NHANES 1999-2000 data show that the overall prevalence of central obesity was 38.3 percent among men versus 59.9 percent among women. As shown in figure 5, there are large gender and ethnicity differences. Prevalence was the highest among Black women (70.4 percent) and lowest among Mexican-American men (35.5 percent).
Trends.
Between 1960 and 2000, the prevalence of central obesity and mean waist circumference increased steadily (figure 5). Mean waist circumference increased by 10 cm in men and 17 cm in women. Between 1988-1994 and 1999-2000, waist circumference increased in most age and racial/ethnic groups, except for men aged 30-59 years, women aged 40-59 years and 70 years or older, and women who were Mexican American or of "other" ethnicity. The prevalence of central obesity increased by approximately 10 percentage points, with the highest increase among women aged 20-29 years (51).
Ethnicity-specific trends showed no disparities between non-Hispanic Whites and Blacks in the early 1960s, while later surveys showed wide racial/ethnic differences; Whites, both men and women, had a lower mean waist circumference. However, in all surveys, waist circumference linearly increased with age, except in the age group 70-79 years, in which waist circumference decreased slightly. The prevalence of central obesity in all age groups showed a greater racial/ethnic disparity among women than among men (52). In general, Mexican-American women ranked first in prevalence in 1988-1994 but in 1999-2000 were surpassed by non-Hispanic Black women (52, 53).
US children and adolescents
Overall patterns and age and gender disparities. Current situation.
Estimates of the national prevalence of at risk for overweight (BMI 85th percentile) and overweight (BMI >95th percentile) based on the most recent NHANES data collected in 1999-2004 are shown in table 5. In 2003-2004, more than one third (about 35 percent) of older US children and adolescents aged 6-19 years were at risk for overweight or overweight, and almost 17 percent were overweight; the figures were lower for young children aged 2-5 years (26.2 percent and 13.9 percent, respectively). Prevalence was similar among older children and adolescents (32). Similar patterns across age and gender were observed for 2003-2004 compared with 1999-2002 analyses (2). The overall national average prevalence is similar among boys and girls; however, large gender differences exist in some racial/ethnic groups. In 1999-2000 and 2003-2004, the prevalence of both outcomes showed a larger gender gap among non-Hispanic Blacks and Mexican-American children and adolescents compared with non-Hispanic Whites.
Trends.
Trends data are shown in figures 6 and 7. In all age groups, the prevalence of overweight had increased since the 1960s. The increasing trend accelerated since NHANES II (1976-1980). Between 1976-1980 and 2003-2004, the average annual rate of increase was approximately 0.5 percentage points for children and adolescents aged 2 years or older. This rate was slower for young children. During this period, the prevalence of overweight among children aged 2-5 years increased from 7.2 percent to 10.3 percent; among children aged 6-11 years, it almost tripled, increasing from 6.5 percent to 15.8 percent. Among adolescents aged 12-19 years, it more than tripled, increasing from 5.0 percent to 16.1 percent. The NHANES 1999-2004 data showed that even during this short period of time, combined prevalence rose from 31.0 percent to 33.6 percent and the prevalence of overweight from 16 percent to 17.1 percent (2, 54).
Between 1971-1974 and 1999-2002, on average, US children's and adolescents' BMI increased by 1.4 points and by 2 points among adolescent boys and girls, respectively (figure 7) (3). The pace of increases in mean BMI was slower than that of the prevalence, suggesting that more of the increase is attributable to the upper tail of the distribution.
Racial/ethnic disparities. Current situation.
Similar to what was observed for adults, the NHANES data show considerable racial/ethnic disparities in obesity among US young people. Non-Hispanic White children and adolescents had the lowest prevalence compared with their non-Hispanic Black and Mexican-American counterparts. For example, combined prevalence was 28.2 percent, 35.4 percent, and 39.9 percent among those aged 6-19 years in the three racial/ethnic groups, respectively. Among boys, Mexican Americans aged 6-11 years had the highest combined prevalence and prevalence of overweight (43.9 percent and 26.5 percent, respectively). Among girls, non-Hispanic Black adolescents (aged 12-19 years) had the highest prevalence and non-Hispanic Whites had the lowest prevalence (41.9 percent vs. 23.6 percent) (32).
The Add Health study 1995-1996 data show that Asian adolescents had an obesity prevalence lower than the national average and all other main racial/ethnic groups (figure 8). The prevalence of at risk for overweight was 22.8 percent and 10.4 percent among Asian adolescent boys and girls compared with 26.5 percent and 22.2 percent among Caucasian adolescent boys and girls, respectively (36).
Large racial/ethnic disparities emerge at very young ages and exist even in homogeneous SES groups. For example, a recent, large study of 21,911 preschool children aged 12-59 months who participated in the Hawaiian Women, Infants, and Children program in 1997-1998 found large racial/ethnic differences in the prevalence of overweight (55). Of the eight racial/ethnic groups (White, Black, Asian, Filipino, Hawaiian, Hispanic, Samoan, and other), Samoan children had the highest prevalence (17.5 percent of those aged 1 year and 27.0 percent of those aged 2-4 years), while Asian children 1 year of age (2.3 percent) and Black children 2-4 years of age (7.3 percent) had the lowest rates. The overall prevalence was 5.9 percent among children 1 year of age and 11.4 percent in those aged 2-4 years.
The prevalence among American native Indian children was higher than the national average. Data collected in the PATHWAY study from 1,704 schoolchildren (in grades 2 and 3) in 41 schools from seven American Indian communities show that half of them were at risk of overweight or overweight (51.5 percent of girls vs. 46.5 percent of boys) and that 30.5 percent of girls and 26.8 percent of boys were overweight (56). Although there was a wide range in BMI across study sites, prevalence was consistently higher than the national averages in all seven communities and among both girls and boys (56).
Trends.
NHANES data were available to enable us to compare the trends among White, Black, and Mexican-American children and adolescents since 1971 (figure 9). In all racial/ethnic groups, overweight increased. The racial/ethnic differences in the increase were small except for adolescent boys. Between 1988-1994 and 1999-2002, White adolescent boys had a slower increase than the other two racial/ethnic groups (2, 54). The prevalence of at risk for overweight among children and adolescents increased from 15.5 percent in 1971-1975 to 33.6 percent in 2003-2004. The Youth Risk Behavior Surveillance System data enabled us to examine the trends based on self-reported weight and height since the early 1990s. In general, they suggest similar patterns (28).
Socioeconomic disparities. Current situation.
In general, the patterns for US children and adolescents and for adults share some similarities, but some features are unique. SES was inversely related to prevalence of obesity among Whites but not among African Americans or Hispanics. The patterns of the SES disparities in obesity are presented in table 6. Our recent study based on NHANES 1999-2002 data shows that high-SES young boys had the lowest prevalence compared with their counterparts, whereas the SES difference in prevalence was small among young girls (3). Among adolescents, no consistent association was found between SES and overweight for boys, but low-SES adolescent girls had a much higher prevalence than their medium- and high-SES counterparts (20.0 percent vs. 14.2 percent and 12.9 percent, respectively). This difference is mainly due to the strong inverse association between SES and overweight among White adolescent girls (3). High-SES Black adolescent girls were at increased risk compared with their lower-SES counterparts (38.0 percent vs. 18.7 percent and 24.5 percent, respectively). Main findings from other selected studies are highlighted in appendix 2 (36, 57, 58).
Trends.
Our analysis of the NHANES data collected between 1971 and 2002 show that overweight increased in all SES groups cross-classified by sex-age-race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American) except for low-SES Mexican-American girls aged 2-9 years (40). We found no consistent patterns for an SES difference in the increasing trend. Between 1976 and 1980 and between 1999 and 2002, the low- to high-SES ratio for prevalence increased from 0.8 to 1.8 among boys aged 2-9 years but then decreased from 1.9 to 1.0, respectively; among adolescent boys and girls, the ratio increased between 1976 and 1994 but then decreased between 1994 and 2002. Furthermore, the patterns varied across age-gender-racial/ethnic groups. Tertiles of family per capita income (assessed by using poverty income ratio) were used to define low-, medium-, and high-SES groups.
Geographic and urban-rural differences: current situation and trends.. To our knowledge, limited studies have examined the regional differences in overweight among US children and adolescents. Our previous research based on the NHANES III data shows that the rural-urban differences are small, and they vary across age groups (16, 59). In children aged 6-9 years, the combined prevalence was higher in urban than in rural areas (26.1 percent vs. 22.8 percent), but the prevalence of overweight was almost the same (11.9 percent vs. 12.1 percent). Among adolescents aged 10-18 years, whereas the combined prevalence was slightly higher in rural than in urban areas (27.2 percent vs. 24.4 percent), the prevalence of overweight was comparable between areas (11.2 percent vs. 10.2 percent).
A recent study based on the Add Health study 1994-1995 baseline data examined the differences in US adolescents' risk of obesity and in their physical activity patterns according to neighborhood characteristics (60). Study participants were grouped into six categories: 1) rural working class; 2) exurban; 3) newer suburban; 4) upper-middle class, older suburban; 5) mixed-ethnicity urban; and 6) low-SES, inner-city areas. Compared with US adolescents living in newer suburbs, those living in rural working-class, exurban, and mixed-ethnicity urban areas were approximately 30 percent more likely to be overweight, independent of individual SES, age, and ethnicity. These findings illustrate important effects of the neighborhood on health and the inherent complexity of assessing residential landscapes across the United States. Simple classic urban-suburban-rural measures may mask the important complexities.
Central obesity.. Unlike for adults, and because national guidelines for classifying central obesity are lacking, little is known about the status of central obesity in US children and adolescents. Nevertheless, a recent study based on NHANES III shows considerable racial/ethnic differences in waist circumference. The distribution of age- and gender-specific percentiles of waist circumference among children and adolescents aged 2-18 years differs across racial/ethnic groups, especially for the uppermost percentiles (75th and 90th) and for adolescents close to age 18 years (figure 10). In fact, compared with their counterparts, Mexican-Americans boys and non-Hispanic Black girls aged 18 years had the highest waist circumference values in the 90th percentile (61).
Tracking of BMI and obesity from childhood to adulthood
A large number of studies have shown the tracking of BMI and obesity status from childhood to adulthood (62-67), providing additional support for early prevention. Overall, it is estimated that about one third of obese preschool children and about one half of obese school-age children become obese adults, although findings from different studies varied considerably. Recent studies also suggest some racial/ethnic differences in the tracking patterns. For example, when longitudinal data collected from 2,392 children (initially aged 5-14 years) over 17 years from childhood to adulthood were used (62), the tracking of childhood BMI was stronger in Blacks than in Whites. Among overweight children, 65 percent of White girls versus 84 percent of Black girls became obese adults; among boys, the corresponding figures were 71 percent versus 82 percent.
Average annual increase in prevalence and future projections
Using linear regression models, we estimated the average annual increase in the prevalence of overweight and obesity (BMI >30 kg/m2) among adults and children between 1980 and 2004 (table 7). Among adults, the prevalence of obesity increased at a faster pace for women than men (β = 0.911 vs. β = 0.653) and particularly among non-Hispanic White women (β = 0.654), who had the fastest increase. Non-Hispanic Black women had the fastest increase in the prevalence of obesity (β = 0.878). Projections based on these models indicate that by 2015, the prevalence of obesity among adults will reach 40.8 percent among all and as high as 62.5 percent among non-Hispanic Black women. The projection is even more alarming for the prevalence of overweight. Overall, the prevalence will be 74.7 percent by 2015: 86.5 percent among non-Hispanic Black women and 82.2 percent among Mexican-American men.
Regarding children aged 6-11 years, non-Hispanic Black girls and Mexican-American boys have by far the fastest annual increase (β = 0.564 vs. 0.548) in the prevalence of overweight, which is expected to reach 31.1 percent and 32.9 percent, respectively, by 2015 (compared with an overall 22.7 percent). A similar pattern is observed for adolescents aged 12-19 years.
DISCUSSION
Currently, more than two thirds of US adults and approximately one third of US children and adolescents are overweight or obese, and some minority and low-SES groups are disproportionally affected. The prevalence of obesity and overweight among US children and adults has more than doubled since the 1970s, and the rate continues to rise. Numerous studies have shown that obesity increases morbidity and mortality (17). Obesity has become the second leading preventable cause of disease and death in the United States, second only to tobacco use (1). Obesity is likely to continue to increase and soon become the leading cause if no effective approaches to controlling it can be implemented.
Consistent with previous studies, our systematic analysis shows large racial/ethnic disparities in obesity among women, children, and adolescents in the United States. Some minority and low-SES groups such as non-Hispanic Black women and children, Mexican-American women and children, low-SES Black men and White women and children, Native Americans, and Pacific Islanders are disproportionally affected. On the other hand, some minority groups such as Asian Americans have a lower prevalence of obesity. Of great concern, our analysis shows that the prevalence of obesity and overweight has increased at an average annual rate of approximately 0.3-0.8 percentage point across different sociodemographic groups over the past three decades. If a similar increase in trend is assumed, by 2015, the majority of US adults (75 percent: BMI >25 kg/m2) and nearly a quarter (24 percent) of US children and adolescents are expected to be overweight or obese (95th percentile of BMI). Some population groups will be more seriously affected. For example, by 2015, 86.5 percent of non-Hispanic Black women will be overweight or obese, and 62.5 percent will be obese. However, current available data are limited and do not enable us to examine the trends in other minority groups or to understand the factors that have led to the current obesity epidemic.
A good understanding of underlying causes that triggered the increase in obesity prevalence in the United States over the past three decades and the factors that have contributed to the disparities across groups is critical in fighting this growing public health crisis and achieving an important national priority to eliminate health disparities. Although obesity is caused by many factors, in most persons, weight gain results from a combination of excess calorie consumption and inadequate physical activity. To maintain a healthy weight, there must be a balance between energy consumption through dietary intake and energy expenditure through metabolic and physical activity (17). A number of individual-, population-, and international-level factors and environmental determinants might have played a role in the obesity trends, such as changes in people's eating behaviors, physical activity and inactivity patterns, occupation, development of technology, culture exchange, and global trade (16, 17, 37). The NHANES data show a dramatic increase in the prevalence of overweight and obesity across all population groups and a declining disparity of obesity across SES groups over the past two decades. This finding indicates that individual characteristics are not the dominant factor to which the rising obesity epidemic is ascribed. Social environmental factors might have a more profound effect in influencing individuals' body weight status than do individuals' characteristics such as SES. A growing consensus is that environmental factors have played a pivotal role in influencing people's lifestyles and fueling the obesity epidemic in the United States and worldwide (17, 68, 69). The environment in the United States has been characterized as "obesogenic" because of its promotion of high energy intake and low energy expenditure (69). The current society provides Americans with abundant food at a relatively low cost and numerous opportunities to reduce energy expenditure at work and at home, which facilitates sedentary behaviors.
Nationally representative survey data examining trends in people's eating patterns between 1970 and the 1990s have indicated several patterns likely to put people in the United States at increased risk of obesity, such as increased consumption of total energy, soft drink, and snack foods; more frequent eating at fast-food and other restaurants; and inadequate consumption of vegetables and fruits compared with dietary recommendations (70-77). The increase in portion size in the United States over the past three decades probably is an important contributor to overconsumption of food and has fueled the growing obesity epidemic. Examination of the current portions of food products against previous portions and dietary intake data collected from individuals consistently show that portion sizes have risen sharply in the United States (73, 75, 78) (appendix 3).
Although our current understanding of the underlying complex causes of the disparities in obesity between population groups in the United States (e.g., gender, age, ethnicity, and SES groups) is still very limited, recent research has shed some important light on related factors at the individual, community, regional, and national levels. At the community level, disadvantage may constrain people's ability to acquire and maintain healthy diet and exercise behaviors. Differential rates of available local area physical fitness facilities, restaurants, and types of food stores by neighborhood characteristics may help explain why obesity does not affect all population groups equally (79, 80). A recent study shows significant disparities in the availability of food stores. African-American and Hispanic neighborhoods had fewer chain supermarkets compared with White and non-Hispanic neighborhoods, by about 50 percent and 70 percent, respectively (81). The availability of supermarkets has been associated with more healthful diets, higher vegetable and fruit consumption, and lower rates of obesity (82, 83). Shopping at supermarkets versus independent groceries has been associated with more frequent vegetable and fruit consumption (84). The Add Health study shows that lower-SES and minority population groups had less access to physical activity facilities, which in turn was associated with decreased physical activity and increased overweight (85).
Population-based policies and programs that emphasize environmental changes are most likely to be successful. Strategies to tackle obesity need to be incorporated into other existing health promotion programs, particularly those preventing chronic diseases by promoting healthful eating and physical activity. Childhood and adolescence are key times for persons to form lifelong eating and physical activity habits. Overweight children are likely to remain obese as adults. Thus, obesity prevention in schoolchildren is a public health priority. In addition, because the majority of children spend many of their waking hours in schools, schools should be key partners in the prevention of childhood obesity. The large racial/ethnic differences in the prevalence of overweight and obesity suggest that culturally sensitive and appropriate approaches are needed in promoting healthful eating in fighting the obesity epidemic. It is crucial to tailor treatment and prevention efforts to each particular ethnicity group's specific situation and needs. Policy makers and public health workers need to be aware of racial/ethnic differences that may affect one's health behaviors and body weight status, such as the differences in their local communities, perceptions of body weight, food preparation, eating practices, physical activity and inactivity patterns, and child-feeding practices. Without developing effective strategies to modify the current "obesogenic" environment in the United States, it is likely that the obesity epidemic will continue. Government agencies, industry, public health professionals, and individual persons all need to play an active role in the growing national efforts to combat the obesity epidemic.
METHODS
Literature search strategy and study inclusion criteria
Using PubMed (National Library of Medicine, Bethesda, Maryland), we searched the English-language literature for the period 1990-2006 that quantitatively assessed obesity and overweight in the United States. The MeSH headings "United States," "obesity" or "overweight" or "body mass index (BMI)" and "race" or "ethnicity" were used as an initial search step. Although more than 900 titles were examined, only those studies with estimates of obesity and/or overweight prevalence in the United States were included upon initial screening of abstracts and full text if needed. These studies were supplemented with others, which were brought to our attention by colleagues and experts consulted. Consequently, more than 80 journal papers, reports, and online data sheets were included in the present study for the literature review, while only about 20 were used for quantitative meta-analysis. Using a standardized data extraction form, we extracted and tabulated related data including measure of obesity or overweight, sociodemographic variables, and quantitative findings.
Definitions of overweight and obesity
According to the World Health Organization, obesity is a disease and is defined as the condition of excess body fat to the extent that health is impaired (17). For practical purposes and among both children and adults, BMI (weight (kg)/height(m)2) is now widely used to assess obesity (11, 17, 18). BMI is closely correlated with body fat and obesity-related health consequences (13, 17).
Adults.. The World Health Organization currently defines overweight and obesity by using BMI cutpoints of 25 and 30 kg/m2, respectively (14, 17). These cutpoints were recommended by a National Heart, Lung, and Blood Institute's and North American Association for the Study of Obesity expert committee (19). Prior to this, various measures and BMI cutpoints have been used in the United States. In the present study, all overweight and obesity prevalence rates reported for adults were based on these BMI cutpoints; in most cases, we present the combined prevalence when using the term "overweight."
The National Heart, Lung, and Blood Institute/North American Association for the Study of Obesity committee also recommends using waist circumference cutpoints of 40 inches (102 cm) for men and 35 inches (88 cm) for women to define "central obesity." Increasingly, research shows that waist circumference or central obesity is a better predictor of obesity-related diseases than overall obesity assessed by using BMI (20, 21). Waist circumference may be equally or more useful than BMI because of its higher predictive value for future health risks, ease of measurement, and understanding by the general public (19-21).
Children and adolescents.. Two sets of BMI percentile references have been used in the United States, which are slightly different. The "old" BMI reference is age- and gender-specific percentiles for ages 6-19 years based on NHANES I data collected in 1971-1974 (22-24). "Overweight" is defined as a BMI greater than or equal to the sex-age-specific 95th BMI percentile, and "at risk for overweight" is defined as a BMI greater than or equal to the 85th percentile but less than the 95th percentile. The "new" BMI reference is provided in the 2000 Centers for Disease Control and Prevention Growth Charts (18), which were developed for all US children aged 2-19 years by using data collected from five national data sets between the 1960s and 1994. These "new" BMI 85th and 95th percentiles are recommended by the Centers for Disease Control and Prevention for screening overweight persons. For children younger than age 2 years, there is no BMI-for-age reference to define overweight, and the weight-for-length 95th percentile has been used (25). The majority of our reported figures for children and adolescents are based on the 2000 Centers for Disease Control and Prevention Growth Charts. Currently, no national recommendation exists regarding classification of central obesity in children and adolescents.
Main data sources
Most of the findings presented in our study were based on nationally representative data. Data from other studies were included to describe the situation in some low-SES and minority racial/ethnic groups that were inadequately covered in these national surveys (more details are provided in appendix 1).
NHANES.. NHANES, a series of cross-sectional, nationally representative examination surveys conducted by the National Center for Health Statistics since the 1970s, includes NHANES 1 (1971-1974), II (1976-1980), and III (1988-1994). Beginning in 1999, NHANES became a continuous survey, and data were collected on people older than age 74 years. Data on weight and height were collected through direct physical examination in a mobile examination center (26).
BRFSS.. BRFSS is an ongoing, nationally representative telephone health survey system, tracking health conditions and risk behaviors among US adults yearly since 1984. This survey is conducted by the 50 state health departments with support from the Centers for Disease Control and Prevention using standard procedures. BRFSS provides state-specific information; that is, it enables geographic differences to be examined (27).
Youth Risk Behavior Surveillance System.. The ongoing Youth Risk Behavior Surveillance System was initiated in 1991 to monitor priority health risk behaviors that contribute markedly to the leading causes of death, disability, and social problems among youth and adults in the United States. This system collected information on risk behaviors and self-reported weight and height (28).
National Longitudinal Survey of Adolescent Health.. The National Longitudinal Survey of Adolescent Health (Add Health study) is a nationally representative, school-based study of youths (grades 7-12; approximately aged 12-17 years at baseline, wave I, in 1994-1995) followed up into young adulthood (approximately aged 18-26 years, wave III, in 2001-2002). In waves I and II (1996), information on self-reported weight and height and, in wave III, directly measured weight and height was collected. Compared with NHANES, the Add Health study oversampled more minority groups such as Asians and Native Americans (29, 30).
Statistical analysis
Analyses were conducted to examine the characteristics and trends of obesity and overweight prevalence. To estimate the average annual increase in prevalence and to predict the future situation regarding obesity and overweight among US adults and children, based on NHANES data, we fit linear regression models with prevalence as the dependent variable and survey years as the independent variable for different sociodemographic groups. In this projection analysis, we assumed that trends would be similar to those observed over the past two to three decades. Time periods for each survey were represented by their median. Beta coefficients were used to indicate the average annual increases in prevalence. The models fit the data well in each sociodemographic stratum, and they explained 48-100 percent (i.e., R2) of the variance in prevalence. In the majority (90 percent) of the models, R2 > 0.90. We had fit the models based on both all of the available data (since NHANES I, 1971-1974) and those available since NHANES II, 1976-1980. Because prevalence increased little between NHANES I and NHANES II, the results based on only the second set of models were reported. Next, on the basis of our second set of linear regression models, we projected the situation for 2010 and 2015 by assuming that prevalence would continue to increase at a similar rate. One should be cautious when interpreting such projection results, particularly for issues related to predicting future events using linear models. Note that we considered alternative statistical models such as linear models with the quadratic term of year/time, but the results differed little, as indicated by the fact that the majority of the coefficients of determination (R2) in our estimated models were between 96 percent and 99 percent. Thus, for simplicity, we chose to use the simple linear models. To assess SES disparities, we calculated prevalence ratios comparing low- with high-SES groups for each age, gender, and ethnicity group and across time. Our regression analysis was conducted by using STATA release 9.0 software (31).
|
|
| |
| |
|
|
|
