We describe prevalence and trends in overweight among children and adolescents (6 to 17 years old) in the US population and variation in the prevalence by sex, age, race-ethnicity, income, and educational level. Height and weight were measured in nationally representative surveys conducted between 1963 and 1994: cycles II (1963 to 1965) and III (1966 to 1970) of the National Health Examination Survey (NHES) and the National Health and Nutrition Examination Surveys (NHANES I, 1971 to 1974; NHANES II, 1976 to 1980; and NHANES III, 1988 to 1994). Overweight was defined by the age- and sex-specific 95th percentile of body mass index (BMI) from NHES II and III. BMI values between the 85th and 95th percentiles were considered an area of concern, because at this level there is increased risk for becoming overweight.
Approximately 11% of children and adolescents were overweight in 1988 to 1994, and an additional 14% had a BMI between the 85th and 95th percentiles. The prevalence of overweight did not vary systematically with race-ethnicity, income, or education. Overweight prevalence increased over time, with the largest increase between NHANES II and NHANES III. Examination of the entire BMI distribution showed that the heaviest children were markedly heavier in NHANES III than in NHES, but the rest of the distribution of BMI showed little change.
Data are limited for assessing the causes of the rapid change in the prevalence of overweight. The increased overweight prevalence in US children and adolescents may be one manifestation of a more general set of societal effects. Childhood overweight should be addressed from a public health perspective.
This article provides epidemiologic data on the prevalence of obese or overweight children (6 to 11 years of age) and adolescents (12 to 17 years of age) in the United States. We present descriptive information on the current prevalence of overweight, the variation among groups of youths described by various demographic factors, and trends over time in the United States.
The first question that arises with the health issue of overweight is what definition to use to distinguish obese or overweight individuals from others. Defining obesity or overweight for children and adolescents is difficult, and there is no generally accepted definition of obesity or overweight for youths.1,2 A variety of criteria for overweight and obesity has been used to evaluate prevalence and trends among children and adolescents.3-7
Any criterion used to estimate the number of overweight or obese children and adolescents in the United States must specify the measure(s) to be used and the corresponding threshold value(s) above which overweight or obesity is present. Ideally, the classification would both reflect adiposity and be related to morbidity and mortality outcomes. Criteria based on measures of fatness or adiposity can classify individuals as obese, the condition of excess adipose tissue; criteria that rely on weight-based measures only indirectly measure adiposity. Therefore, these criteria can classify individuals as overweight but not necessarily as obese.
The national surveys that are the source of the data discussed below have included measures of weight and height as well as of skinfold thicknesses measured at several anatomic sites. Because of concerns about the comparability of skinfold thickness measurements across the surveys, weight-based measures that determine overweight are preferred to the skinfolds that determine obesity. A definition of weight status based on height and weight is desirable, because these measures can be obtained with reasonable precision in a variety of settings including field studies, clinical practice, and research. Comparisons among various weight-for-height indices for both adults and children have led to the selection of body mass index (BMI; kg/m2) as most desirable.8 In the results and discussion that follow, BMI is used as the measure of weight status, and individuals are described with respect to overweight, not obesity.
Studies comparing BMI with measures of adiposity have shown high correlations between the measures. However, it must be noted that BMI is not as reliable a measure of fatness for children, especially across different ages and degrees of maturity, as it is for adults, who have attained their peak height.9-12 Studies that show good correlation between BMI and adiposity in youths find that other factors such as gender, race, age, and maturational status are important to consider in predicting adiposity. The estimates of adiposity also are subject to a substantial degree of imprecision.9
BMI changes dramatically with age during childhood and adolescence.13,14 For example, mean BMI for children at 6 and 7 years of age may be ∼16.0, whereas at 16 and 17 years of age, the mean is closer to 22.0.14,15 This age dependence of BMI indicates that the measure is not independent of height. Therefore, a given BMI value has very different implications for body composition depending on age, and overweight criteria based on BMI must be age-specific. The components that contribute to the composite measure of BMI also change throughout the process of development. During growth and maturation, body proportions, bone mass, and the ratio of lean-to-fat tissue change at different times and at different rates. The timing of these body composition changes may be related to development of overweight and obesity.13,14,16 Therefore, caution is necessary when BMI is used as a measure of body composition in children and adolescents.
Adults have been classified as overweight by a variety of cutoffs based on mortality experience of life insurance policy holders,17-20 statistical criteria from a reference population,21 and, more recently, on data relating morbidity as well as mortality to weight status.22,23Classifying children as overweight or obese based on mortality or morbidity is problematic. Childhood mortality is unlikely to be related to body composition. Outcomes such as childhood morbidity, adult morbidity, adult mortality, and adult obesity have been examined,24-32 but data to support a classification based on these outcomes are sparse and difficult to interpret.33
In the absence of outcome-based criteria for children, a statistical approach offers the most practical choice.1 Overweight or obesity is defined relative to a selected percentile of a reference population based on age, sex, race-ethnicity, or other group characteristics. Choosing a particular percentile entails the assumptions that the proportion of the reference population that exceeds the cutoff is overweight and that the prevalence of overweight is the same for the groups defined by sex, age, or other characteristics incorporated in the definition. The normal processes of growth and development during childhood and adolescence make the second assumption potentially problematic.
A well-known example of statistical criteria is the use of growth charts to characterize children. Growth charts show the full distribution of a measurement (height, weight) or measurement ratio (weight-for-height, BMI) across a range of ages. Instead of defining a condition by a single value to create a dichotomy, growth charts present multiple percentiles. These distributions may be based on a population that can be defined in terms of a particular place and time or they may combine multiple populations over place or time.
Growth charts are used differently in clinical and epidemiologic applications. In clinical practice, growth charts are used to monitor individual children as they develop over time. In this application, emphasis is placed on tracking within a percentile range, and concern occurs when a child crosses percentiles as he or she grows. In epidemiologic applications, growth charts can form the basis for evaluations of cross-sectional data by serving as the reference population from which a particular percentile cutoff is chosen to classify individuals as overweight, underweight, stunted, etc. In addition to providing a variety of percentiles, growth charts have the added feature of usually presenting smoothed percentiles.
Preliminary data from the revision of the National Center for Health Statistics (NCHS) growth charts34 serve as the basis for defining overweight for this article. The revised growth charts incorporate smoothed BMI percentiles and are based on data from cycles 2 and 3 of the NHES (NHES II and III) and the first, second, and third National Health and Nutrition Examination Surveys (NHANES I, II, and III). BMI data for NHANES III participants ≥6 years of age were not included in the revised growth charts because of the known higher overweight prevalence for these ages in NHANES III.7,35
As suggested by Himes and Dietz for clinical application36and Troiano et al for epidemiologic application,7 the 95th percentile of BMI is used to define overweight. This criterion is more likely to be specific for obesity than lower percentile cutoffs. The 85th percentile has been used to classify adults as overweight, and adolescents between the 85th and 95th percentile cutoffs are considered at risk of becoming overweight.36 Therefore, the proportion of the population with BMI between the 85th and 95th percentiles also is described.
Two previous publications7,35 presented prevalence and trends for overweight among children and adolescents and reported data collected in NHANES III. Overweight was defined based on unsmoothed sex- and age-specific 95th percentiles of BMI from NHES II and III. As indicated below, use of the growth chart data as the reference population results in a slightly lower prevalence of overweight. However, the patterns of secular trends and relative relationships among demographic subgroups are not altered.
Data From National Surveys
Cross-sectional Prevalence Estimates
NHANES III, conducted from 1988 to 1994 by the NCHS of the Centers for Disease Control and Prevention (CDC), was a cross-sectional multipurpose survey designed to provide nationally representative reference data and prevalence estimates for a variety of nutrition and health status measures and health conditions. The sampling plan followed a complex, stratified, multistage, probability cluster design from which a sample representative of the civilian noninstitutionalized population in the United States was selected. NHANES III oversampled Mexican-Americans and African-Americans as well as young children (<6 years of age) and older persons (≥60 years of age) to produce more reliable estimates for these groups. Detailed descriptions of the survey design and operation are available.37,38
The survey protocol included a household interview followed by a standardized physical examination of selected participants in a mobile examination center. Body weight and height were measured in the mobile examination center with standardized equipment and procedures.39
The sampling plan for NHANES III was based on age at interview. Examinations generally took place within 3 weeks after the interview, and some participants had birthdays in the intervening period. Because children and adolescents can experience growth in height and weight during a brief period, age at the time of examination was used in all analyses.
Race-ethnicity was based on proxy or self-report and was categorized as non-Hispanic white, non-Hispanic black, Mexican-American, or other. The “other” racial-ethnic group sample was too small for separate reporting, but it was included in estimates for combined racial-ethnic groups. Data from 31 young women who reported being pregnant were excluded from all analyses, as were data from one overweight adolescent male whose exceptionally high sample weight caused undue influence on prevalence estimates.
Previous National Samples for Trends
Four national surveys conducted before NHANES III used a similar design and comparable standardized methods to collect data on heights and weights of nationally representative samples of children and adolescents: the NHES cycle 2 (NHES II, 1963 to 1965, age 6 to 11 years) and cycle 3 (NHES III, 1966 to 1970, age 12 to 17 years), and the first and second NHANES (NHANES I, 1971 to 1974, age 1 to 74 years; NHANES II, 1976 to 1980, age 6 months to 74 years).40-43 These surveys allow analysis of trends in the prevalence of overweight in the United States. To correct for changes in the age distribution over time, prevalence estimates from each survey were age-adjusted by the direct method to the 1980 census population figures based on single years of age.44
Data collection on ethnicity was limited in NHES and NHANES I; thus, Hispanics could not be identified separately. In these earlier surveys, race was categorized as white, black, or other. Therefore, trend analyses were performed by race (white and black) rather than by race-ethnicity. Sample sizes for the “other” race category were insufficient for separate reporting, but were included in estimates for all races combined. Data from young women in NHANES I and NHANES II who reported being pregnant were excluded from the analyses. Information on pregnancy was not collected in NHES II or III.
Statistical analyses were performed with SAS45and SUDAAN.46 For each survey, sample weights were calculated that took into account the unequal probabilities of selection resulting from the cluster design, planned oversampling of selected subgroups, and unit nonresponse. All analyses incorporated the sample weights. Standard error (SE) measures were calculated with SUDAAN, a program that takes into account the sample weights and complex sample design for calculating variance estimates.
Prevalences were calculated with sex- and age-specific criteria using 6-month intervals. To increase sample sizes and stabilize prevalence estimates, prevalence calculated by 6-month age intervals was averaged across 3-year age categories (6 to 8, 9 to 11, 12 to 14, and 15 to 17 years) for cross-sectional estimates for all races combined. Cross-sectional prevalence by race-ethnicity and trends in prevalence were averaged across 6-year age categories (6 to 11 and 12 to 17 years), corresponding to the age groups of NHES II and NHES III.
Current Data on Overweight Prevalence
Data collected from 1988 to 1994 for NHANES III indicate that the prevalence of overweight among children and adolescents was substantially higher than in the reference population across virtually all racial-ethnic, age, and sex groups (Table1). Prevalence for non-Hispanic white girls was higher than the 5% prevalence established by the overweight definition, but the differences were smaller than for other groups and not likely to be statistically significant.
Beyond the ∼11% of children and adolescents classified as overweight, an additional 14% had a BMI between the 85th and 95th percentiles of the reference population, which puts them in an area of concern because they may be at risk for becoming overweight. The proportion of youths in this category ranged from 10% for non-Hispanic black males 6 to 11 years of age to ∼17% for Mexican-American females 6 to 11 years and non-Hispanic black and Mexican-American females 12 to 17 years of age.
Striking differences in overweight prevalence by race and race-ethnicity have been found among adult women.39Similarly, in the National Heart, Lung, and Blood Institute's Growth and Health Study data for children 9 and 10 years of age,47mean BMI for black girls was significantly greater than that for white girls. Overweight prevalence among children and adolescents in the NHANES III data was higher for non-Hispanic blacks and Mexican-Americans than for non-Hispanic whites; however, the 95% confidence intervals for the prevalence estimates overlapped.
Overweight Prevalence and Socioeconomic Status
An inverse relationship between socioeconomic status and overweight or obesity frequently is found among adult women and sometimes among adult men.48 Studies of children find a weaker and less consistent relationship between socioeconomic status and overweight in girls than in women, and often find the same relationship in boys as in girls.48 In the NHANES III data, overweight prevalence among Mexican-American and non-Hispanic black children and adolescents is not related to family income. However, there appears to be an inverse relationship between overweight prevalence and family income for non-Hispanic white adolescents (Table2). These data must be interpreted cautiously, because many of the estimates have large standard error (SE) measures. Values that have a coefficient of variation (estimate/SE) >30% are flagged in Table 2.
There also is little evidence for a pattern in the relationship between overweight prevalence and education of the family reference person (Table 3). With one exception among the 12 age, sex, and racial-ethnic categories, the ≥13 years of education category has the lowest prevalence of overweight. However, in many categories, the 0 to 11 years of education stratum has lower overweight prevalence than the high school graduate stratum (12 years). A pattern of decreasing overweight prevalence with increasing education of the family reference person is found only among non-Hispanic white male children and adolescents.
Trends in Overweight Prevalence
NHANES III is the most recent in a series of nationally representative surveys that have included measurements of height and weight taken with comparable, standardized methods. Overweight prevalence has increased primarily since NHANES II in the second half of the 1970s (Fig 1, Table4). Furthermore, comparison of the data from the entire 6 years of NHANES III35 with estimates published previously from the first 3 years7 suggests an apparent trend of increasing overweight prevalence even within the 6 years of the survey. Characteristics of the sample design of NHANES III complicate comparison of values from the two halves of the survey, but an intrasurvey increase of ∼2 to 6 percentage points was found for most of the sex, age, and racial-ethnic groups.
Trends in overweight prevalence among preschool children also have been examined with data from NHANES III and earlier national surveys.49 Preschool children were classified as overweight if their weight for stature exceeded the 95th percentile of the NCHS growth chart.34,49 The pattern of a notable increase between NHANES II and NHANES III also was found for preschool girls, particularly 4- and 5-year-olds. There was no significant trend of increasing prevalence of overweight for children 2 and 3 years of age.
Trends in Overweight Prevalence From Other Data Sources
Although other data to track trends in overweight prevalence among children and adolescents are limited, longitudinal data from the Bogalusa Heart Study50 and baseline data from the National Heart, Lung, and Blood Institute's Growth and Health Study47 show increases in overweight prevalence over a period corresponding approximately to that between NHANES II and NHANES III.
Change in the BMI Distribution
Increases in overweight prevalence and in the mean BMI (Table5) do not fully describe changes in the distribution of BMI within the population. An increase in mean BMI may be attributable to either a shift in the entire distribution of BMI or an increase in just the upper portion of the distribution. It is not clear whether the entire population is becoming heavier or whether only the heavier individuals are even heavier than before, and lighter individuals show little change.
Mean–difference (m–d) plots51 can be used to address this issue by examining graphically changes in the distribution of BMI between NHES II or III and NHANES III. These plots enable us to see how the distribution of BMI has changed between the two surveys. For each year of age within the same sex, we compared the BMI at 49 distinct percentile points from NHES with the corresponding percentiles from the same sex-age group in NHANES III. For each percentile level, we then calculated the mean of the percentile values from the two distributions and the difference between the percentile values from the two distributions. To create an m–d plot, we plot the differences against the means. Each point on the plot represents the mean at a single percentile level and the difference in BMI between the two distributions for the same percentile level. If the distribution of BMI were shifted to the right by a constant amount between the two surveys, then the m–d plot would show a horizontal line >0 by the amount of the constant difference. However, if the distribution were shifted by a greater amount at the upper end (representing the heaviest children becoming heavier), then the m–d plot would show a pattern of larger differences at higher percentiles.
With this approach, we found a general pattern of little or no difference between the surveys at the lower percentiles of the BMI distribution, but increasing differences at higher percentiles. The patterns generally were similar by sex. With increasing age, there was more evidence of an upward shift of the entire distribution, combined with the disproportionate increase in the higher percentiles. Representative plots from the NHES and NHANES III data are shown in Figs 2 and3. Figure 2 is an m–d plot for 6-year-old males, and it shows a shift only in the upper end of the BMI distribution. Figure 3 is a plot for 14-year-old girls, and it shows a slight shift in the entire distribution (upward displacement of the plot) combined with a greater shift at the upper end.
These plots show that the BMI distribution for children and adolescents in the NHANES III is more right-skewed than the distribution in the NHES. The heaviest children are markedly heavier in NHANES III (1988 to 1994) than in NHES II (1963 to 1965) and III (1966 to 1970), but the rest of the distribution of BMI is shifted little or not at all among the surveys. Examination of m–d plots for non-Hispanic white youths showed the same pattern that was observed for the entire population. This shows that the change in the BMI distribution was not attributable to changes in the racial-ethnic composition of the population.
Trends in Other Body Measures
Echoing the changes in the BMI distribution, mean and median triceps and subscapular skinfold measures increased between NHANES II and III (data not shown). Examination of the 90th percentile from each survey suggests that, as with BMI, the distribution of skinfolds has become more skewed. The increases observed between NHANES II and III are greater at the 90th percentile than at the mean or median.
Mean stature did not change between NHES II and III and NHANES II. When examined by race, sex, and 3-year age groups, mean stature in NHANES III was greater than in NHANES II. Although some within-group differences were ≥2 cm, not all differences were statistically significant. Differences between the surveys were larger for 9 to 11 and 12 to 14 years than for other age groups, and they were larger for the black race than for the white race. Among black youths, mean height in NHANES III was 2.5 to 3.0 cm greater than in NHANES II. Any differences in mean stature appear to be attributable to higher values in the right tail of the distribution instead of to an overall shift in the distribution. The differences in stature are unlikely to account for the secular trend in BMI.
Overweight prevalence has increased dramatically over a relatively brief time period. In some population subgroups, ≥30% are either overweight or at risk of overweight. Evaluation of the entire range of BMI indicates that most of the change has occurred among the heaviest portion of the population. The probability of overweight continuing into adulthood is greater with more severe overweight.31,32 Therefore, high prevalence of overweight is likely to continue, if not increase, under current conditions. Furthermore, NHANES III data indicate that the prevalence of overweight has increased among adults over a similar period. Behaviors that contributed to the increase in overweight prevalence for adults may be transmitted within the family setting and affect the weight status of children.
Paradoxically, adverse conditions associated with overweight have declined in the population despite the increase in overweight prevalence. Among adults and youths, total serum cholesterol values are lower in NHANES III than in previous surveys. Mean blood pressure also has declined (NCHS, unpublished observations). These observations and the dangers of interference with growth and eating disorders that may result from excessive attention to body weight and food restriction among youths remind us that dealing with overweight among children and adolescents is a sensitive area.
Why Is This Happening?
The increase in overweight prevalence reflects a population shift toward positive energy balance. Dietary intake and physical activity represent the behavioral and, therefore, modifiable aspects of this balance equation. It is apparent that calorically dense foods are abundant and readily available in the United States. However, neither the NHANES nor other national data indicate an increase in caloric intake among children and young adolescents.52,53Estimated caloric intake has increased for older adolescents (16 to 19 years of age) and adults.52 Percentage of calories from fat has continued a downward trend that has been observed since the mid-1960s.52 These observations do not preclude the possibility that excessive caloric intake contributed to the weight status of individual children in NHANES III.
Concurrent with availability of calorically dense foods, the United States has moved toward a sedentary lifestyle.54 Although data to evaluate secular trends in activity for young children do not exist, a study of high school students found that participation in physical education declined from 1984 to 1990.55 Changes in safety, parental work habits, television viewing, availability of video games, and other cultural aspects of the environment may have further decreased opportunities for exercise.56 Data from the 1990 Youth Risk Behavior Survey revealed that >70% of students in grades 9 through 12 reported at least 1 hour of television-watching, and >35% reported watching television ≥3 hours each school day.55 The relationship between time spent watching television and overweight is controversial, with some researchers finding a strong relationship,57,58 and others questioning the presence of a direct relationship.59
The 1990 Youth Risk Behavior Survey found that only approximately half of all students in grades 9 through 12 reported being enrolled in physical education classes.55 The decline in physical education participation is particularly troubling because school-based, health-oriented physical education may provide both immediate effects of the activity and sustained effects through encouragement of lifelong activity patterns. For physical education programs to contribute to the public health goal of lifelong activity, they should include activities of moderate intensity and should not focus exclusively on team-oriented sports activities.60 Several studies have shown that fewer than half of school children received daily physical education, and games and competitive sports were the mainstays of existing programs.5
The recent increase in the prevalence of overweight is not limited to just one age group or one country. Data from NHANES also showed an increase in prevalence of overweight among adults over the same period.39 The prevalence of overweight has been reported to be increasing in varying degrees, not only in the United States but in Britain and elsewhere in Europe.61-64
These increases in overweight may be attributable to more than changes in the behavior of individuals. From a population perspective, the trends observed for all age groups in the United States and in many other societies worldwide suggest social and environmental factors that are affecting many individuals similarly. The increased prevalence of overweight in children in the United States should be viewed in the context of similar increases occurring in other age groups in the United States and in many other societies around the world. The increase in the prevalence of overweight in children in the United States may be one manifestation of a more general set of effects operating broadly in many societies. With this understanding, childhood overweight should be addressed from a public health perspective.
- Received October 24, 1997.
- Accepted November 6, 1997.
Dr Troiano is currently at the National Cancer Institute, National Institutes of Health, Bethesda, Maryland.
- BMI =
- body mass index •
- NCHS =
- National Center for Health Statistics •
- NHES =
- National Health Examination Survey •
- NHANES =
- National Health and Nutrition Examination Surveys •
- CDC =
- Centers for Disease Control and Prevention •
- SE =
- standard error •
- m–d =
- mean–difference •
- SEM =
- standard error of the mean
- ↵Public Health Service. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Publication PHS 90–50212. Washington, DC: US Department of Health and Human Services; 1990
- Must A,
- Dallal GE,
- Dietz WH
- ↵Cole TJ. Weight-stature indices to measure underweight, overweight, and obesity. In: Himes JH, ed. Anthropometric Assessment of Nutritional Status. New York, NY: Wiley-Liss; 1991:83–111
- Daniels SR,
- Khoury PR,
- Morrison JA
- ↵Najjar MF, Rowland M. Anthropometric Reference Data and Prevalence of Overweight, United States, 1976–80. DHHS Publication No (PHS) 87-1668 (Vital and Health Statistics; Series 11, No 238). Washington, DC: Public Health Service, Government Printing Office; 1987
- Johnston RE
- ↵Anonymous. New weight standards for men and women. Stat Bull Metrop Insur Co. 1959;40(Nov–Dec):1–4
- ↵Anonymous. 1983 Metropolitan height and weight tables. Stat Bull Metrop Insur Co. 1984;64(Jan–Jun):2–9
- National Institutes of Health Consensus Development Panel on the Health Implications of Obesity
- ↵US Department of Agriculture ARS Dietary Guidelines Advisory Committee. Report of the dietary guidelines advisory committee on the dietary guidelines for Americans, 1995, to the Secretary of Health and Human Services and the Secretary of Agriculture. Springfield, VA: National Technology Information Service; 1995
- ↵US Department of Agriculture, US Department of Health and Human Services. Dietary Guidelines for Americans. 4th ed. Home and Garden Bulletin No 232. Washington, DC: Government Printing Office; 1995
- Abraham S,
- Collins G,
- Nordsiek M
- Nieto FJ,
- Szklo M,
- Comstock GW
- Sonne-Holm S,
- Sorensen TI,
- Christensen U
- Guo SS,
- Roche AF,
- Chumlea WC,
- Gardner JC,
- Siervogel RM
- Hamill PVV,
- Drizd TA,
- Johnson CL,
- Reed RB,
- Roche AF
- Centers for Disease Control and Prevention
- Himes JH,
- Dietz WH
- ↵Mohadjer L, Montaquila JM, Waksberg J, et al. National Health and Nutrition Examination Survey. III. Weighting and estimation methodology. Prepared by Westat Inc for National Center for Health Statistics, Hyattsville, MD. NHANES III Reference Manuals and Reports (CD-ROM). Hyattsville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics; 1996
- ↵National Center for Health Statistics. Plan and Operation of the Third National Health and Nutrition Examination Survey, 1988–1994. Department of Health and Human Services Publication No (PHS) 94-1308 (Vital and Health Statistics; Series 1, No 32). Hyattsville, MD: National Center for Health Statistics; 1994
- ↵National Center for Health Statistics. Plan, Operation, and Response Results of a Program of Children's Examinations. Vital and Health Statistics; Series 1, No 5. Washington, DC: Public Health Service; 1967
- ↵National Center for Health Statistics. Plan and Operation of a Health Examination Survey of US Youths 12–17 Years of Age. Vital and Health Statistics; Series 1, No 8. Washington, DC: Public Health Service; 1969
- ↵National Center for Health Statistics. Plan and Operation of the Health and Nutrition Examination Survey, United States, 1971–1973. Vital Health Statistics; Series 1, No 10. Washington, DC: US Government Printing Office; 1973
- ↵McDowell A, Engel A, Massey J, Maurer K. Plan and Operation of the Second National Health and Nutrition Examination Survey, United States, 1976–1980. Vital Health Statistics; Series 1, No 15. Washington, DC: US Government Printing Office; 1981
- ↵US Bureau of the Census. 1980 Census Population, Vol 1. Characteristics of the Population. Chapter B—General Population Characteristics. Part 1. United States Summary. Washington, DC: US Government Printing Office; 1981
- ↵SAS Institute Inc. SAS/STAT User's Guide, Version 6. 3rd ed. Cary, NC: SAS Institute; 1990
- ↵Shah BV, Barnwell BG, Hunt PN, Nileen P, LaVange LM. SUDAAN User's Manual. Release 5.50. Research Triangle Park, NC: Research Triangle Institute; 1991
- ↵Ogden CL, Troiano RP, Briefel RR, Kuczmarski RJ, Flegal KM, Johnson CL. Prevalence of overweight among preschool children in the United States, 1971 through 1994. Pediatrics. 1997;99:e1. URLhttp://www.pediatrics.org/cgi/content/full/99/4/e1
- Freedman DS,
- Srinivasan SR,
- Valdez RA,
- Williamson DF,
- Berenson GS
- ↵Cleveland WS. Visualizing Data. Summit, NJ: Hobart Press; 1993
- ↵Federation of American Societies for Experimental Biology, Life Sciences Research Office. Prepared for Interagency Board for Nutrition Monitoring and Related Research. Third Report on Nutrition Monitoring in the United States. Vol 2. Washington, DC: US Government Printing Office; 1995
- McGinnis JM
- Dietz WH,
- Gortmaker SL
- Malina RM
- Sallis JF,
- McKenzie TL
- Prentice AM,
- Jebb SA
- Kuskowska-Wolk A,
- Bergstrom R
- Kuskowska-Wolk A,
- Bergstrom R
- Copyright © 1998 American Academy of Pediatrics