From the Division of Health Examination Statistics, National
Center for Health Statistics/Centers for Disease Control and
Prevention, Hyattsville, Maryland.
Overweight among preschool children is a concern, because it may
have long-term health consequences.1 Excess body
weight in childhood is associated with overweight in
adulthood,2,3 and excess body fat, or obesity, is
recognized as a health risk for adults.4 Obesity and
overweight in childhood also have been linked to subsequent morbidity
and mortality in adulthood.5,6 Moreover, overweight
preschoolers have been shown to have higher mean levels of cholesterol
than other children.7 The concern about overweight
prevalence in preschoolers is heightened by recent increases in
overweight among school-aged children and adults. Overweight prevalence
increased among boys and girls 6 through 11 years of age from
approximately 5% in 1963 through 1965 to 11% in 1988 through 1991 and
5% to 13% among 12- through 17-year-old-boys.8 Furthermore, at least one third of adults in the United States are
overweight.9
This article presents cross-sectional estimates of overweight
prevalence for preschool children between 2 months and 6 years of age
in the United States using weight adjusted for length or stature.
Current estimates are from the third National Health and Nutrition
Examination Survey (NHANES III) conducted during 1988 through 1994. In
addition, trends in overweight are presented based on estimates from
earlier NHANES.
METHODS
NHANES III is a nationally representative sample of the total
civilian noninstitutionalized population in the 50 states and the
District of Columbia. NHANES III collected information on individuals 2 months of age and older. The design was a stratified, multistage
probability sample based on selection of counties, city or suburban
blocks, households, and persons within households. In total,
approximately 40 000 people were selected, 31 311 of whom received
physical examinations. NHANES III was designed to sample large numbers
of preschool children to facilitate revising the
original10 National Center for Health Statistics
(NCHS) and modified11 NCHS/Centers for Disease Control and
Prevention (CDC) growth charts for children. Mexican-American persons,
black persons, and persons older than 60 years were also oversampled to
have large enough sample sizes for subgroup comparisons. Sample weights
were used to adjust estimates back to the appropriate population
distribution.12,13
The NHANES comprise a series of cross-sectional, nationally
representative examination surveys conducted by the NCHS of the CDC.
NHANES III is the seventh in the series, which began in 1960. NHANES I,
conducted from 1971 through 1974, was the first survey to include
preschool-aged children.14 NHANES II was conducted from
1976 through 1980.15 A special study of Hispanic
populations, the Hispanic Health and Nutrition Examination Survey
(HHANES), in three regions of the United States was conducted from 1982 through 1984. The Mexican-American subpopulation sampled in the HHANES
was located entirely in the southwestern United States.16
The surveys included both a home interview and a standardized physical
examination conducted in a mobile examination center. A physical
examination in the home was available for infants 2 through 11 months
of age. The physical examination included measurement of recumbent
length, stature, and weight. In NHANES III recumbent length was
measured in children younger than 4 years, and stature was measured in
children 2 years and older. Children wore paper examination gowns, and
infants wore only diapers.
The sample of NHANES III included 8765 children 2 months through 5 years of age. Of these 8765 children, 8261 (94%) were interviewed by
proxy after informed consent was obtained from parents or guardians. Of
the 8261 for whom interview data were obtained, 7784 (94%) were
examined. The overall survey response rate was 89% (7784 of 8765). Of
the 7784 children participating in NHANES III, 199 were excluded from
this analysis because their length or stature was above or below the
range of the original NCHS growth charts (67 children) or they were
missing stature, length, or weight data (132 children). Between initial
household screening and the physical examination, 65 children reached 6 years of age, and 4 home-examined children reached 1 year of age before
being examined and were excluded from these analyses. Another 36 children had stature and length data that differed by 5 cm or more and
were also excluded. The analytic data set included 7480 children 2 months through 5 years of age at the time of examination.
The response rate is similar across the NHANES surveys and is
comparable or higher than other national health and nutrition surveys.17 Trend estimates would not be affected by
nonresponse, because the nonresponders would have had to be different
in each survey. Nonresponse bias in NHANES III has been evaluated
elsewhere.18
Current overweight prevalence from NHANES III was estimated by age at
examination, sex, and race/ethnicity. Race/ethnicity were reported by
household members. Children were categorized as non-Hispanic white,
non-Hispanic black, Mexican-American, or other. Numbers for racial and
ethnic groups in the other category were too small for meaningful
analysis when considered separately, but these persons were included in
totals. Trend estimates based on all the surveys were estimated by age,
sex, and race. Trend estimates were not estimated by race/ethnicity,
because Hispanic children cannot be separately identified in NHANES I. However, estimates were made for Mexican-American children in the
HHANES and NHANES III.
Overweight Definition
The definition of overweight among preschoolers is based on the
original NCHS weight-for-stature and weight-for-length reference growth
curves.10 In this analysis, weight-for-length curves were used for children 2 years and younger, and weight-for-stature curves were used for children 2 through 5 years. Two-year-old children
were compared with either curve depending on whether they had stature
or length measures. Some 2-year-old children in this sample had both
measures and were included in the prevalence estimates based on both
weight-for-length and weight-for-stature.
The prevalence of overweight is defined as the percentage of children
whose weight-for-stature or weight-for-length was above the 95th
percentile of the respective growth curve. Concern about overweight
prevalence occurs when prevalence exceeds 5%. This definition of
overweight is commonly used and recommended for preschoolers.21,22
The 95th percentile is a conservative definition of overweight. Infants
and preschoolers are in a dynamic state of growth in which body size is
continually in a state of flux. For this reason, it is difficult to
assign a single cutoff value to an age range.23 To
circumvent potential adverse effects of clinically misclassifying
youths as overweight, using the highest cutoff of the 95th percentile
is prudent for children younger than 6 years. Among adults, 85th
percentile cutoffs are often used24 and usually are based
on body mass index (BMI, kilograms per square meter). At this level in
adults, unfavorable health outcomes begin to emerge.25 The
Committee on Clinical Guidelines for Overweight in Adolescent
Preventive Services recommended the use of the 95th percentile of BMI
to classify adolescents as overweight.26
The prevalence of overweight clearly depends on the selection of the
reference population and the overweight criteria. The original NCHS and
modified NCHS/CDC growth charts include several different data sources.
For preschool children 3 to 6 years of age the weight-for-height charts
were based on weight and stature data from the nationally
representative NHANES I. For children from birth to 3 years, the
weight-for-height charts were based on weight and recumbent length data
from the Fels Research Institute (for the years 1929 through 1974). The
Fels data were for primarily white, middle-class children. The original
NCHS and NCHS/CDC charts represent smoothed versions of the data from
the various surveys. Consequently, the prevalence estimates for
children 3 to 6 years of age in NHANES I are not the expected 5%.
Although there is concern27,28 regarding the use of the
original NCHS or NCHS/CDC growth charts, the choice of the reference
will not affect trend estimates of overweight.
A disjunction in the growth curves occurs at 2 years of
age29 because of the two different measures (stature and
length) and because of the different data sources used to make the
charts. It has been suggested that the disjunction between infant and later curves occurs because the Fels sample was "actually taller or
measured taller than the US sample."27 For these and
other reasons, the original NCHS and modified NCHS/CDC growth charts are currently under revision.28
Because of rapid growth in children younger than 2 years, there is no
agreed on definition of overweight in this age group. Consequently, the
term overweight is used for children 2 through 5 years of age when
compared with the weight-for-stature growth charts. When children are
compared with the weight-for-length growth charts, the percentage above
the 95th percentile of the weight-for-length growth curve is presented.
Data Analysis
Data were analyzed using SAS30 and
SUDAAN31 software programs. All analyses included sample
weights that account for the unequal probabilities of selection
attributable to oversampling and nonresponse (based on participation in
the examination component of the survey). Standard errors (SE) were
calculated with SUDAAN to account for the sample weights and complex
sample design. Tests of significant differences in prevalence estimates
within NHANES III by sex, age, and race/ethnicity were done using
multivariate logistic regression analysis. Trends in overweight
prevalence among surveys were modeled and tested using weighted least
squares models for categorical data.32
RESULTS
Current Overweight Prevalence Estimates
Sample sizes are presented in Table 1 by age in
months for each of the surveys. NHANES III oversampled preschool-aged
children and, therefore, has the largest sample sizes.
|
Table 1.
Number of Survey Participants in Sex and Age Groups
by Survey
[View Table]
|
The percentage of children younger than 1 and 1 to 2 years who exceeded
the 95th percentile on the weight-for-length growth curves is presented
in Table 2 overall and by age, sex, and racial/ethnic group. More than 5% of children younger than 1 and 1 to 2 years by sex
and race/ethnicity were above the 95th percentile of weight-for-length. Among 1- to 2-year-old children, the percentage above the 95th percentile on the weight-for-length growth curves was 9.4%. It is
unclear whether this relatively high percentage is a reflection of a
problem in the population or an artifact of the growth charts for this
age group.29
|
Table 2.
Percentage of Children Younger Than 3 Years Above
the 95th Percentile of the Weight-for-Length Growth Reference, Third
National Health and Nutrition Examination Survey (1988 Through 1994)
[View Table]
|
There was no difference in the percentage above the 95th percentile of
the weight-for-length growth curve between the younger than 1-year age
group and the 1- to 2-year age group. However, within age groups there
were significant differences by gender and race/ethnicity. The
percentage above the 95th percentile of weight-for-length was
significantly higher among girls than boys (P < .01) after accounting for differences by age group and race/ethnicity in a multivariate logistic regression model. The percentage above the
95th percentile was highest among Mexican-American children, intermediate among non-Hispanic black children, and lowest in non-Hispanic white children. The differences in the percentage above
the 95th percentile between non-Hispanic white and non-Hispanic black
children (P < .05) and between non-Hispanic
white and Mexican-American children were statistically significant
(P < .01).
The prevalence of overweight among children 2 to 3 and 4 to 5 years of
age based on weight-for-stature is presented in Table 3.
In general, the prevalence of overweight was higher among children 4 to
5 years of age than among children 2 to 3 years of age. In both age
groups, the prevalence of overweight was lowest among non-Hispanic
white boys. The prevalence of overweight among girls was higher than
among boys for all racial/ethnic and age groups. More than 10% of all
girls 4 to 5 years of age were overweight. Among the three different
racial/ethnic groups, Mexican-American children had the highest
prevalence of overweight (12.0% for boys and 13.2% for girls);
non-Hispanic black children had intermediate prevalence, and
non-Hispanic white children had the lowest prevalence of overweight.
Differences by age group, sex, and race/ethnicity were statistically
significant (P < .05) in multivariate logistic regression models. For children 2 to 3 years of age, prevalence estimates for most subgroups were near or less than 5%. There was one
notable exception in which 10.5% of Mexican-American girls were
overweight.
|
Table 3.
Prevalence of Overweight Based on Percentage of 2- to 5-Year-Old Children Above the 95th Percentile of the
Weight-for-Stature Growth Reference, Third National Health and
Nutrition Examination Survey (1988 Through 1994)
[View Table]
|
The effect of the disjunction29 in the growth charts is
evident in the NHANES III results. The percentage of 1- to 2-year-old children above the 95th percentile on the weight-for-length growth curve (Table 2) was 9.4%, whereas only 3.4% of the 2- to 3-year-old age group was overweight based on weight-for-stature (Table 3). Although data are not shown, the prevalence of overweight among 2-year-old children based on weight-for-stature was substantially lower
than the prevalence among the same children based on weight-for-length.
Trends Across National Surveys
Table 4 contains estimates of the percentage of
children overall and separately by race above the 95th percentile of
weight-for-length for NHANES I through III and for Mexican-American
children in the HHANES and NHANES III. Among children younger
than 24 months, estimates based on weightfor-length showed an
increase over time in the percentage of children above the 95th
percentile. This change was greater for girls than boys. In the
youngest age group (younger than 1 year) the percentage above the 95th
percentile rose from 6.2% in NHANES II to 10.8% in NHANES III. Among
girls 12 to 23 months of age, the percentage above the 95th percentile increased from 6.1% to 9.5% between NHANES I and NHANES III. Among black girls 12 to 23 months of age, the percentage above the 95th percentile of weight-for-length rose from 8.9% to 15.2%. There was no
change in the percentage above the 95th percentile among boys 12 to 23 months of age.
|
Table 4.
Percentage of Children 6 to 23 Months of Age Above
the 95th Percentile of the Weight-for-Length Growth Reference, 1971 Through 1974 to 1988 Through 1994*
[View Table]
|
Trends in national estimates of overweight prevalence based on
weight-for-stature are presented in Table 5. Among 2- to
3-year-old girls and boys there was no significant change in the
prevalence of overweight between 1971 through 1974 and 1988 through
1994. However, among 4- to 5-year-old children there was an increase, especially in girls overall and by race. From NHANES I to NHANES III
the prevalence among 4- to 5-year-old girls rose from 5.8% to 10.8%.
The prevalence of overweight among 4- to 5-year-old girls in NHANES III
was statistically different from the prevalence of overweight in the
other two surveys (P < .01). Among black girls
the prevalence more than doubled from 5.0% to 12.6% between 1971 through 1974 and 1988 through 1994. Among black boys the prevalence of
overweight increased from 3.0% to 8.7% between 1976 through 1980 and
1988 through 1994. Statistical tests were not performed on changes in
overweight prevalence for individual racial/ethnic groups.
|
Table 5.
Prevalence of Overweight Based on Percentage of 2 to
5-Year-old Children Above the 95th Percentile of the Weight-for-stature Growth Reference, 1971 Through 1974 to 1988 Through 1994
[View Table]
|
The prevalence of overweight based on weight-for-stature also has
increased among most groups of Mexican-American preschool children.
Among 2-to 3-year-old Mexican-American girls the prevalence increased
from 4.2% to 10.5%. Among Mexican-American 4- to 5-year-old boys the
prevalence increased from 4.9% in the HHANES to 12.0% in NHANES III.
DISCUSSION
These results are the most recent nationally representative
prevalence estimates of overweight among preschool children in the
United States. We found that during the past two decades in the United
States, there was no change in the prevalence of overweight among 2- to
3-year-old children, but there was an increase among 4- to 5-year-old
children, especially in girls. Currently 10.8% of 4- to 5-year-old
girls are overweight. In addition, these data show that the prevalence
of overweight among preschool children is higher among girls than boys
and that overweight prevalence is highest among Mexican-American
children. All of these results parallel what has been seen among older
children and adults in the same surveys.8,9 These
findings are based on highly reliable, standardized measures of weight,
stature, and length across a series of national
surveys.33,34
The higher prevalence of overweight among girls than boys may be
attributable to gender differences in behavior and is unlikely to be an
artifact of the growth charts. In at least one study preschool boys
were found to be more physically active than preschool girls,35 and similar gender differences in overweight
prevalence have been observed in older children and
adults.8,9
The use of the NCHS growth charts in some Hispanic36 and
African-American37 populations should be done with caution,
particularly among children younger than 2 years. The charts for this
age group are based exclusively on white, middle-class children, and
there is some evidence that growth differences do exist between these ethnic groups. However, the use of the NCHS growth charts does provide
a uniform basis for estimating prevalence in different populations, and
measurements taken from privileged children in developing countries in
Africa, Latin America, and the Caribbean do not differ significantly
from the NCHS reference population.38
The high prevalence of overweight found among Mexican-American
preschool children may have an environmental basis. Trend estimates within racial/ ethnic groups indicate an increase in overweight between
the HHANES and NHANES III for Mexican-American children, strongly
suggesting that environmental influences are affecting the observed
increase in overweight. Furthermore, findings from the CDC Pediatric
Nutrition Surveillance System indicate that low-income Hispanic
children 2 to 4 years of age in the United States showed a relative
increase in overweight prevalence of nearly 20% between 1980 and
1991.21
The difference in overweight prevalence between the white and
non-Hispanic white children in NHANES III may reflect demographic changes in the population of the United States. The prevalence of
overweight among preschool-aged children is higher in the
Mexican-American population, and this population makes up a greater
percentage of the population than it did in the early 1970s (4% in
1972 and 8% of children younger than 5 years in
1991).43,44 The total population estimates may reflect this
change.
Other researchers have also found a higher than expected prevalence of
overweight among preschool children. A cross-sectional study of urban
kindergartners found excess overweight in 4- to 5-year-old
children.37 In this study, the prevalence of overweight was
highest among Hispanic children, but the overall growth of children was
on par with the NCHS/CDC reference curve. In addition, among low-income
preschool children, overweight has been found to be highest in Hispanic
children compared with white, black, and Asian
children.45,46 The prevalence of overweight among American
Indian preschool children has also been shown to be in excess of 5%
and increasing over time. Using data from the CDC Pediatric Nutrition
Surveillance System, the prevalence of overweight among American Indian
preschoolers 2 to 4 years of age was estimated at 9.0% in
198847 and 11.9% in 1991.21
The reasons for the increase in the prevalence of overweight among 4- to 5-year-old children are complex. Excess weight gain is ultimately a
function of energy intake exceeding energy expenditure, with
sociocultural factors influencing lifestyle and diet choices. However,
the influence of bottle feeding, early introduction of solid foods,
balanced food selection, and other dietary practices on overweight in
preschool-aged children is less clear. Dietary intake data from the
NHANESs suggest that mean energy and fat intakes among preschoolers
have not increased in the last 20 years.17 A review of the
literature suggests that overweight among preschool children, as well
as older children, may be associated less with increased energy intake
and more with low physical activity.48 In the Framingham
Children's Study, inactive preschool children gained substantially
more subcutaneous fat by the time they reached first grade than did
more active children.35 Some researchers attribute the
increasing prevalence of overweight among school-aged children to
decreased physical activity as a result of television watching.49 Others, however, insist that the link between
television and fatness is weak.50
Because of an increase in the percentage of high birth weight infants,
the relationship between birth weight and overweight during childhood
has become a concern. From 1971 through 1985 there was an increase in
the proportion of high birth weight infants (
4000 g) for both white
and black infants.51 A small positive relationship has been
found between birth weight and BMI. However, there was no indication
that an increase in the distribution of birth weights would lead to
increased prevalence of overweight during childhood.52
Cross-sectional surveys such as those used in this analysis
cannot investigate growth patterns in individuals. Moreover, the statistical definition of overweight is not based on risks of morbidity
and mortality, and cross-sectional population-based estimates of
overweight do not account for variation associated with genetic
predisposition of the individual or familial patterns of overweight. In
pediatric clinics, these are important considerations and should be
considered along with an assessment of the individual based on serial
measures of stature or length and weight over time.
The increasing prevalence of overweight among 4- to 5-year-old children
indicates that prevention activities need to begin in the preschool
years. As recommended in the Dietary Guidelines for
Americans,25 these efforts include encouraging physical activity to maintain a healthy weight, eating at least five servings of
fruits and vegetables per day, and after the age of 2 years gradually
decreasing dietary fat to a level of no more than 30% of energy. The
Committee on Nutrition of the American Academy of Pediatrics also
recommends energy balance over specific caloric intake restrictions for
obese children.53 The Special Supplemental Nutrition
Program for Women, Infants, and Children and Head Start are potential
avenues for educating low-income parents and children about healthy
behavior choices such as lower-fat foods and physical activity. In
addition, preschool and day care programs might include more education
related to diet and physical activity.
It is important to encourage healthy choices regarding diet and
physical activity during the preschool years, because behavior patterns
may begin in these years.54,55 In addition, the normal growth of a child is such that a child's adiposity during the first
year of life increases rapidly then decreases until approximately 6 years of age, when adiposity again develops. Young children who go
through this "adiposity rebound" early (4 and 5 years of age) have
higher adiposity in adulthood than children who go through the
adiposity rebound later.56 The increase in the prevalence of overweight reported among adults, school-aged children, and preschoolers makes it even more critical to influence behavior among
preschoolers. It has been suggested that the most successful treatment
of preadolescent obesity is early and regular treatment and that this
preferred approach be adapted for prevention activities during the
preschool years.57 An observational study of white middle-class 3- to 5-year-old children found that modifiable risk factors such as dietary intake and physical activity accounted for more
of the variance in BMI in a 3-year period than did nonmodifiable risk
factors such as obese parents. The authors concluded that encouragement
of healthy diets and physical activity can decrease weight gain in
preschool children.58
Pediatric practitioners should emphasize the importance of diet and
physical activity as two important components of a healthy lifestyle,
with considerably more emphasis on adequate amounts of physical
activity. Continued clinical, community, and national monitoring and
surveillance of weight among preschool children is essential. To
investigate factors associated with overweight among preschool children
further, assessments of dietary habits and activity levels are needed
for different sociocultural groups.
Received for publication Jun 27, 1996; accepted Oct 8, 1996.
Reprint requests to (C.L.O.) Division of Health Examination
Statistics, National Center for Health Statistics/Centers for Disease
Control and Prevention, 6525 Belcrest Rd, Room 900, Hyattsville, MD
20782.
This article is based on data from the National Nutrition
Monitoring and Related Research Program.
NHANES, National Health and Nutrition Examination
Survey.
NCHS, National Center for Health Statistics.
CDC, Centers for
Disease Control and Prevention.
HHANES, Hispanic Health and Nutrition
Examination Survey.
BMI, body mass index.