Longitudinal Patterns of Breakfast Eating in Black and White Adolescent Girls*

Article (PDF Available)inObesity 15(9):2282-92 · October 2007with50 Reads
DOI: 10.1038/oby.2007.271 · Source: PubMed
Abstract
The objective was to describe the pattern of breakfast eating over time ("breakfast history") and examine its associations with BMI and physical activity. This longitudinal investigation of patterns of breakfast eating included 1,210 black and 1,161 white girls who participated in the 10-year, longitudinal National Heart, Lung, and Blood Institute Growth and Health Study (NGHS). Three-day food records were collected during annual visits beginning at ages 9 or 10 up to age 19. Linear regression and path analysis were used to estimate the associations between breakfast history, BMI, and physical activity. Among girls with a high BMI at baseline, those who ate breakfast more often had lower BMI at the end of the study (age 19), compared with those who ate breakfast less often. Path analysis indicated that energy intake and physical activity mediated the association between patterns of breakfast eating over time and BMI in late adolescence. The association between regular breakfast consumption over time and moderation of body weight among girls who began the study with relatively high BMI suggests that programs to address overweight in children and adolescents should emphasize the importance of physical activity and eating breakfast consistently.
Longitudinal Patterns of Breakfast Eating
in Black and White Adolescent Girls
Ann M. Albertson,* Debra L. Franko,† Douglas Thompson,‡ Alison L. Eldridge,* Nort Holschuh,*
Sandra G. Affenito,§ Robert Bauserman,‡ and Ruth H. Striegel-Moore¶
Abstract
ALBERTSON, ANN M., DEBRA L. FRANKO,
DOUGLAS THOMPSON, ALISON L. ELDRIDGE,
NORT HOLSCHUH, SANDRA G. AFFENITO, ROBERT
BAUSERMAN, AND RUTH H. STRIEGEL-MOORE.
Longitudinal patterns of breakfast eating in black and white
adolescent girls. Obesity. 2007;15:2282–2292.
Objective: The objective was to describe the pattern of
breakfast eating over time (“breakfast history”) and exam-
ine its associations with BMI and physical activity.
Research Methods and Procedures: This longitudinal in-
vestigation of patterns of breakfast eating included 1210
black and 1161 white girls who participated in the 10-year,
longitudinal National Heart, Lung, and Blood Institute
Growth and Health Study (NGHS). Three-day food records
were collected during annual visits beginning at ages 9 or 10
up to age 19. Linear regression and path analysis were used
to estimate the associations between breakfast history, BMI,
and physical activity.
Results: Among girls with a high BMI at baseline, those
who ate breakfast more often had lower BMI at the end
of the study (age 19), compared with those who ate
breakfast less often. Path analysis indicated that energy
intake and physical activity mediated the association
between patterns of breakfast eating over time and BMI
in late adolescence.
Discussion: The association between regular breakfast con-
sumption over time and moderation of body weight among
girls who began the study with relatively high BMI suggests
that programs to address overweight in children and ado-
lescents should emphasize the importance of physical ac-
tivity and eating breakfast consistently.
Key words: body weight, BMI, adolescents, youth, phys-
ical activity
Introduction
Breakfast consumption has been linked to better overall
nutrition (1), BMI (1,2), higher levels of physical activity
(3,4), and overall quality of life (5) in children and adoles-
cents. Despite the positive implications of breakfast con-
sumption, the frequency of eating breakfast decreases from
childhood through adolescence. For example, the National
Heart, Lung, and Blood Institute Growth and Health Study
(NGHS)
1
(1) reported that at age 9, approximately 77% of
white girls and 57% of black girls ate breakfast on all 3 days
for which food records were kept, compared with approxi-
mately 32% and 22%, respectively, by age 19. In that study,
breakfast eating was associated with lower BMI, but this
effect did not hold when socioeconomic status, energy in-
take, and physical activity were accounted for statistically.
Several studies have confirmed that breakfast eating de-
clines as children grow older (6 8). Among adolescent
girls, this trend is partially explained by the fact that skip-
ping breakfast has been shown to be a common method of
trying to lose weight (6,8,9). It is possible that regular
breakfast eating may be related to a general healthy lifestyle
(i.e., higher activity and lower caloric intake), which, in
turn, contributes to a lower BMI, but research to date has
not examined the mechanisms that might explain how
breakfast eating is related to weight in children and adoles-
cents.
An additional question is the degree to which the pattern
of breakfast eating over time (“breakfast history”) may
Received for review November 2, 2006.
Accepted in final form February 1, 2007.
The costs of publication of this article were defrayed, in part, by the payment of page
charges. This article must, therefore, be hereby marked “advertisement” in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
*Bell Institute of Health and Nutrition, General Mills, Inc., Minneapolis, Minnesota;
†Northeastern University, Department of Counseling and Applied Educational Psychology,
Boston, Massachusetts; ‡Maryland Medical Research Institute, Baltimore, Maryland; §St.
Joseph College, Department of Nutrition, Hartford, Connecticut; and ¶Wesleyan University,
Department of Psychology, Middletown, Connecticut.
Address correspondence to Debra L. Franko, Northeastern University, Department of
Counseling Psychology, 203 Lake Hall, 360 Huntington Ave., Boston, MA 02115-5000.
E-mail: d.franko@neu.edu
Copyright © 2007 NAASO
1
Nonstandard abbreviations: NGHS, National Heart, Lung, and Blood Institute Growth and
Health Study; MET, metabolic equivalent; EDI, Eating Disorder Inventory.
2282 OBESITY Vol. 15 No. 9 September 2007
impact health indicators. Eating breakfast consistently
(across a number of years) may be important for long-term
management of body weight. With few exceptions, previous
studies of breakfast consumption have been cross-sectional,
thus precluding prospective analyses to test whether the
pattern of breakfast eating over time plays a role in changes
in body weight or activity patterns. Two relevant longitu-
dinal studies were identified. Berkey et al. (10) examined
whether skipping breakfast was associated with BMI
changes over a 3-year period in children who were 9 to 14
years of age at baseline. Specifically, children were asked,
“How many times each week (including weekdays and
weekends) do you eat breakfast?” Overall, results indicated
that children who never ate breakfast had lower total caloric
intake than those who ate breakfast nearly daily. Interest-
ingly, overweight children who never ate breakfast de-
creased in BMI over the following year, relative to over-
weight children who ate breakfast nearly every day. In
contrast, normal-weight children who did not eat breakfast
gained weight when compared with their peers who ate
breakfast routinely. It should be noted, however, that height
and weight were self-reported in this study and overweight
children who never ate breakfast may have underestimated
their weight and under-reported their food intake. Further-
more, because breakfast was not defined and was based on
children’s retrospective recall in response to one global
question, these results need to be interpreted with caution. A
second study (11) followed 7745 Finnish adolescents over a
3-year period (ages 16 to 19 years); they found that regularly
eating breakfast (every morning) was consistently associated
with self-rated good health as well as with “persistent exercise”
(engaging in physical activity 4 or 5 times/wk).
Although informative, these studies have several limita-
tions: they span a relatively short time period (3 years), are
based on homogeneous (i.e., mostly white) samples, and
rely on questionnaire data for reports of breakfast consump-
tion. The current study provides 10-year longitudinal data
from a diverse sample whose 3-day food records were verified
by dieticians, thus addressing some of these concerns.
In light of previous literature, we sought to expand what
is known about breakfast eating among children and ado-
lescents with 4 aims: 1) to describe the pattern of breakfast
eating over time (“breakfast history”), 2) to examine the
relationships between breakfast history and BMI and phys-
ical activity, 3) to better understand the specific types of
physical activity that are associated with breakfast eating
over time, and 4) to test the hypothesis that energy intake
and physical activity partly explain the relationship between
breakfast history and BMI over time.
Research Methods and Procedures
Participants and Recruitment
As previously reported (12), the NGHS recruited 2379
black and white girls at three study sites who were 9 or 10
years old at study entry: University of California at Berke-
ley, University of Cincinnati/Cincinnati Children’s Hospital
Medical Center, and Westat, Inc./Group Health Association
in Rockville, MD. Girls were recruited from public and
parochial schools or (in Maryland/Washington, DC only)
from a membership listing of families who were enrolled in
a large health maintenance organization and local Girl Scout
troops. Eligible participants identified themselves (using
census categories for race/ethnicity) as “black” or “white,”
non-Hispanic, with racially concordant parents or guard-
ians. All girls who entered the NGHS assented, and their
parents (or guardian) consented to their participation. In
each race group, wide ranges of income ($10,000 to
$75,000) and educational levels (less than high school
diploma to graduate degree) were represented.
Due to variable annual participation rates, sample sizes
varied from visit to visit. Retention rates (in relation to the
sample size of 2379 at baseline) were very high at visits 2
to 4 (96%, 94%, 91%), declined to a low of 82% at visit 7,
and increased to 89% at visit 10.
Measurements and Procedure
A complete description of NGHS procedures and mea-
sures has been reported elsewhere (12). Briefly, data were
collected annually at participating sites or, if the girl was
unable to travel to the site, at her home. The study protocol
was approved by the Institutional Review Boards of all
participating sites. Only instruments of relevance to the
present report are described below.
Breakfast history was the independent variable of pri-
mary interest. This measure was based on 3-day food
records [previously validated compared with observed in-
takes during school lunch (13)] collected in study years 1
through 5, 7, 8, and 10. Dietitians used age-appropriate
materials to instruct girls to record all food and drink and
time of intake for 3 consecutive days that included 2 week-
days and 1 weekend day. Breakfast was defined as any
eating that occurred between 5:00 and 10:00
AM on week-
days or between 5:00 and 11:00
AM during weekends. Food
records were coded and analyzed for nutrients using Food
Table Version 19 of the Nutrition Coordinating Center
nutrient database (14). Breakfast history was defined as the
total percentage of days that a girl reported eating breakfast
across all food record days that a girl reported throughout
NGHS. The maximum number of food record days was 24
(8 study years 3 food record days each), but due to
missing visits some girls had fewer than 24 total days of
food records. For example, if a girl completed food records
in only years 1 to 5, 7, and 10 (7 years 3 food record
days 21 total food record days) and reported eating
breakfast on 12 of those days, then the girl was coded as
having eaten breakfast on 57% (12 of 21) of all days
reported in the food records.
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
OBESITY Vol. 15 No. 9 September 2007 2283
Demographics. Data regarding race and highest level of
parental education were collected at study entry from girls
and their parents (or guardians). Race (black or white) was
defined by the participant’s self-report at baseline. Parental
education was categorized as 4 years of college vs. 4
years of college. Education was chosen over income as a
proxy of socioeconomic status because NGHS data were
collected in three regions differing in median household
income and also because previous research has shown that
education is a better predictor of health-related outcomes
than income (15). Participants’ age was recorded as age at
last birthday.
BMI. Annually (all visits), centrally trained examiners
obtained height and weight measurements. BMI was calcu-
lated as weight in kilograms divided by height in meters
squared. Instead of raw BMI, BMI-for-age z-scores [using
the Centers for Disease Control and Prevention age- and
gender-specific percentiles as the reference distribution
(16)] were used in this study because they have the same
distribution across different ages. A score of 0 is at the
gender- and age-specific mean of BMI, with positive scores
indicating above-average BMI for that age and negative
scores indicating below-average BMI.
Physical Activity. The analyses included a composite
variable representing total energy expended across all
activities, based on 3-day physical activity diaries col-
lected in tandem with the food records in study years 1 to
5, 7, 8, and 10. The methodology and revisions over time
in the specific activities and categories of activities in-
cluded have been described in detail elsewhere (17).
Briefly, girls indicated the times they woke up and went
to bed each day as well as the approximate amount of
time they spent engaged in various categories of activi-
ties of similar intensity (for example, “running, soccer,
track, or field hockey”). Participants completed the diary
by selecting the duration of time they spent in each
category of activity (1 to 15 minutes, 16 to 30 minutes, or
30 minutes) during specific parts of the day. For all
study years, completed diaries were reviewed by cen-
trally trained staff using a common protocol. Metabolic
equivalent (MET) values were assigned to each activity
grouping, based on a review of existing literature and
modified to reflect the age and gender of NGHS partic-
ipants. Finally, summary MET values were assigned to
each girl for each usable diary. For each visit, each girl’s
physical activity score was the average MET value of
either 2 or 3 usable diary days (if only 1 day of data were
usable at a visit, that visit was excluded).
To identify specific physical activities that were associ-
ated with breakfast consumption, a binary variable was
formed for specific categories of activities, coded as 1 (girl
reported doing the activity on at least one day) or 0 (girl
never reported doing the activity). Some activities were
grouped in the physical activity records (e.g., walking and
bike riding) to reduce the reporting burden on girls. Results
for specific activities are reported at the finest level of detail
permitted by the data.
Eating Disorder Inventory (EDI). Because NGHS lacked
well-validated measures of dieting to lose weight, two sub-
scales of the EDI were used as a proxy for tendencies
toward dieting behavior. The Body Dissatisfaction subscale
consists of 9 items that measure dissatisfaction with specific
body parts or the body in general (18). The Drive for
Thinness subscale consists of 7 items that assess the degree
to which the respondent thinks about and/or wishes to
change her shape and weight. Extensive data regarding
reliability and validity have been reported (19) and results
from a factor analysis with the NGHS cohort have been
published (20). For black girls in NGHS, Cronbach
co-
efficients for the Body Dissatisfaction subscale ranged from
0.85 (age 11, visit 3) to 0.88 (age 15, visit 7); for the Drive
for Thinness subscale, the coefficients were 0.69 (age 11,
visit 3) to 0.71 (age 15, visit 7). For white girls, Body
Dissatisfaction coefficients ranged from 0.92 to 0.93 and for
Drive for Thinness from 0.74 to 0.76, for the same ages.
These subscales were administered in years 1, 3, 5, 7, and
10 of NGHS.
Statistical Analysis
BMI-for-age z-scores in the final year of NGHS (Year
10) were modeled as a function of breakfast history, adjust-
ing for study site and other variables associated with BMI in
past studies (i.e., race, parental education, physical activity,
energy intake, and the EDI subscale scores). In all analyses,
baseline values were used for race and parental education;
physical activity scores and energy intake were averaged
across all study years, analogous to the breakfast history
measure; and EDI subscales were taken from the final study
year. Preliminary analyses indicated quadratic trends for
physical activity and energy intake, so squared terms for
these were also included in the final model. The model was
adjusted for baseline BMI-for-age z-score, to ensure that
associations between breakfast history and Year 10 BMI
were not simply due to frequent breakfast-eaters also having
higher baseline BMI values; it is important to control for
this because BMI is known to track over time (21). Finally,
the model included days eating breakfast (0, 1, 2, or 3)
based on the food records in Year 10; this was done to
ensure that the history of breakfast consumption throughout
adolescence has explanatory value, above and beyond
breakfast consumption measured concurrently with the out-
come (BMI-for-age in Year 10). The model included the
interaction between baseline BMI-for-age z-scores and
breakfast history. Interactions between breakfast history,
race, and physical activity were also considered but not
included in the final model because they were not signifi-
cant in preliminary analyses, nor did their inclusion have a
noticeable effect on the other model estimates. The model-
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
2284 OBESITY Vol. 15 No. 9 September 2007
ing technique was linear regression (PROC REG in SAS
v9.1.3; SAS Institute, Inc., Cary, NC). Multiple imputation
(PROCs MI and MIANALYZE; SAS Institute, Inc.) was
used to ensure unbiased estimation in the presence of miss-
ing data for some of the analytic variables. A complete-case
model (i.e., a model dropping all cases with missing data on
any variable) was also estimated to gauge the extent to
which the model estimates were affected by the method of
handling missing data. The model using multiple imputation
was viewed as the primary analysis and the complete-case
model was viewed as a sensitivity analysis. Mixed models
(PROC MIXED in SAS; SAS Institute, Inc.) were used to
estimate linear age trends in the intake of selected nutrients
at breakfast, taking into account the repeated measurements
of breakfast consumption within girls.
Additional secondary analyses were conducted to gain
insight into processes through which breakfast consumption
might influence BMI. It seemed plausible that differences in
physical activity and energy intake might mediate the asso-
ciation between breakfast history and BMI-for-age z-scores
in Year 10 of NGHS. That is, differences in breakfast habits
might lead to differences in physical activity and energy
intake throughout the day, which, in turn, would affect BMI
in the final study year. Path analysis was used to examine
these possibilities. The model consisted of estimated paths
(interpreted similarly to regression coefficients) from break-
fast history to physical activity and energy intake (defined
as in the linear regression model, including the quadratic
term), as well as paths from the latter two variables to
BMI-for-age z-score in study Year 10. In addition, because
breakfast history might directly influence BMI (above and
beyond its indirect effects through the mediating links of
physical activity and energy intake), a direct path from
breakfast history to BMI-for-age z-score in study Year 10
was estimated. Similar to the linear regression model, the
path model adjusted for study site, race, parental education,
and EDI subscales. Path analysis was done using Mplus v3
(Muthen & Muthen, Los Angeles, CA).
To gain a sense of specific types of physical activity that
were frequent among girls who often ate breakfast, break-
fast history was examined in relation to categories of phys-
ical activity measured in the physical activity records in
Table 1. Descriptive statistics for measures used in the analysis
Measure Sample size
Mean (SD); range; 25th/50th/75th percentile
(continuous measures) or (categorical measures)
Raw BMI, baseline 2352 18.6 (3.8); range: 11.2–35.3; 15.8/17.6/20.5
BMI-for-age z-score, baseline 2352 0.3 (1.1); range: 4.8–2.7; 0.5/0.3/1.1
Raw BMI, study year 10 2066 25.6 (6.8); range: 15.8–55.6; 20.7/23.6/28.7
BMI-for-age z-score, year 10 2066 0.5 (1.1); range: 3.1–2.6; 0.3/0.5/1.4
Breakfast history (% days eating breakfast)
(years 0–10) 2371 70.6 (19.2); range: 0.0–100.0; 58.3/72.7/85.7*
Average energy intake (kcal/1000) (years 0–10) 2371 1.9 (0.4); range: 0.7–4.1; 1.6/1.8/2.1
Average physical activity score (years 0–10) 2368 452.5 (260.9); range: 49.5–2894.1; 280.6/395.1/553.5
EDI drive-for-thinness (year 10) 2069 5.1 (5.7); range: 0.0–21.0; 0.0/3.0/9.0
EDI body dissatisfaction (year 10) 2065 9.0 (7.8); range: 0.0–27.0; 2.0/7.0/14.0
Days eating breakfast (year 10) 1896
None 22.4%
1 day 26.1%
2 days 25.8%
3 days 25.7%
Race: black 2371 51.0%
Study site 2371
California 37.3%
Ohio 36.5%
Washington, DC 26.2%
Parental education: 4 years college 2369 35.1%
* Additional percentiles for breakfast history are 33.3 (5th percentile), 42.9 (10th percentile), 94.4 (90th percentile), and 100.0 (95th
percentile).
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
OBESITY Vol. 15 No. 9 September 2007 2285
study Year 10. Breakfast history was categorized into 9
groups of increasing percentage of days with breakfast, and
the percentage of girls exhibiting each type of activity on
one or more days was computed for each category. Using
logistic regression (PROC GENMOD in SAS; SAS Insti-
tute, Inc.), a linear trend of association was estimated for
each category of activity; this provides a test of whether
rates of the activity tend to increase (or decrease) as the
percentage of days eating breakfast increases. The criterion
of statistical significance in all analyses was p 0.05.
Results
The analysis included the 2371 NGHS participants
(99.7% of the 2379 total) who completed food records in
one or more study years. Sample size for individual vari-
ables varied by study year and the specific measures used
(Table 1).
As shown in Table 1, on average, girls ate breakfast on
70.6% of days and one fourth of the girls ate breakfast on
85% of days. Even girls at the 5th percentile consumed
breakfast on 33.3% of days, indicating that the large major-
ity of girls consumed breakfast at least occasionally (only 3
girls, or 0.1%, reported consuming breakfast on 0 days).
Over the years, girls who ate breakfast tended to consume
more energy at breakfast time (an average increase of 5.1
kcal per year, p 0.0001). Adjusting for age-related
changes in energy intake at breakfast, over the years, girls
who ate breakfast consumed more sucrose (0.39 g per year,
p 0.0001), caffeine (1.6 mg per year, p 0.0001) and
sodium (4.9 mg per year, p 0.0001), and less calcium
(1.4 mg per year, p 0.005) at breakfast time; intake of
fat, protein, and cholesterol neither increased nor decreased
(p 0.10).
Results of the model of BMI-for-age z-scores in study
Year 10 are shown in Table 2. The two methods of handling
missing data (multiple imputation and complete case anal-
ysis) resulted in similar parameter estimates, including
nearly identical estimates for the effects of primary interest
(breakfast history and breakfast history by baseline BMI-
for-age). Controlling for baseline BMI, energy intake, phys-
ical activity, EDI subscale scores, and study site, race, and
parental education, the main effect of breakfast history was
not significant. However, the significant breakfast history
by baseline BMI interaction indicates that the association of
breakfast history with BMI z-scores in Year 10 depended on
BMI-for-age in Year 0 (baseline), as illustrated in Figure 1.
Among girls with a high BMI-for-age at baseline, those who
ate breakfast more often had lower BMI-for-age at the end
of the study (Year 10), compared with those who ate break-
fast less often. In other words, eating breakfast more often
Table 2. Parameter estimates of the model of BMI-for-age z-score in study year 10
Parameter
Parameter estimate (SE)
Semi-partial
correlation
Multiple imputation
of missing data
Drop cases with
missing values
Intercept 0.1577 (0.0785)‡ 0.1482 (0.0795)† NA
Breakfast history (% days eating breakfast) 0.0013 (0.0011) 0.0015 (0.0011) 0.00038
Breakfast history by baseline BMI-for-age 0.0026 (0.0007) 0.0027 (0.0007)§ 0.00285
Baseline BMI-for-age 0.7919 (0.0490) 0.7984 (0.0537) 0.04607
Days eating breakfast (year 10) 0.0080 (0.0177) 0.0085 (0.0170) 0.00005
Average energy intake (kcal/1000) 0.0942 (0.0461)‡ 0.0788 (0.0467)† 0.00059
Average energy intake (kcal/1000)
2
0.0623 (0.0490) 0.0508 (0.0559) 0.00017
Average physical activity score (log) 0.0781 (0.0352)‡ 0.0727 (0.0337)‡ 0.00097
Average physical activity score (log)
2
0.0480 (0.0384) 0.0834 (0.0417)‡ 0.00083
Race: black 0.2618 (0.0339) 0.2788 (0.0364) 0.01220
Parental education: 4 years college 0.1043 (0.0350)§ 0.1095 (0.0356)§ 0.00197
Study site California* 0.1074 (0.0417)‡ 0.1071 (0.0427)‡ 0.00131
Study site Ohio 0.0287 (0.0449) 0.0272 (0.0421) 0.00009
EDI drive-for-thinness (year 10) 0.0069 (0.0036)† 0.0068 (0.0039)† 0.00063
EDI body dissatisfaction (year 10) 0.0325 (0.0029) 0.0316 (0.0030) 0.02307
* For study site, Washington, DC served as the reference level.
p 0.10; p 0.05; § p 0.01; p 0.0001.
NA, not applicable.
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
2286 OBESITY Vol. 15 No. 9 September 2007
was associated with decreased BMI at the end of the study,
but only among girls who had relatively high BMI at the
beginning of the study. One indicator of effect size is to
estimate the impact of eating breakfast one additional day
per week, which would increase days eating breakfast by
14% (1 of 7 days). Among girls with median baseline
BMI-for-age, eating breakfast one more day per week is
estimated to result in an increase of 0.02 (95% confidence
interval, 0.01, 0.05) in BMI-for-age in Year 10; this
estimate was not significantly different from 0 (p 0.17).
However, among girls with baseline BMI at the 95th per-
centile, eating breakfast one more day per week is expected
to result in a significant decrease of 0.04 (95% confidence
interval, 0.08, 0.01) in BMI-for-age in Year 10 (p
0.04). There was an even greater decrease in Year 10
BMI-for-age among girls with baseline BMI at the 97th
percentile (decrease ⫽⫺0.05; 95% confidence interval,
0.10, 0.01, p 0.01).
Results of the model of BMI-for-age in study Year 10
indicate that both physical activity and energy intake were
associated with BMI. Possibly, breakfast history contributes
to BMI through the intermediating influence of these vari-
ables; specifically, it may be the case that girls who eat
breakfast more often have higher overall energy intake as
well as greater physical activity, and the latter variables, in
turn, are associated with BMI-for-age in study Year 10. A
path model was used to examine whether the data in NGHS
were consistent with this hypothesis. The results are shown
in Figure 2. As hypothesized, eating breakfast more often
predicted significantly greater physical activity (standard-
ized coefficient 0.063, p 0.004); and, in turn, very high
physical activity predicted lower BMI-for-age. Interest-
ingly, the association of physical activity and BMI-for-age,
adjusting for the other variables in the model, was U-shaped
(Figure 3), based on the significant estimates for physical
activity (standardized coefficient 0.038, p 0.02) and
physical activity squared (standardized coefficient
0.029, p 0.05). Further, eating breakfast more often
predicted greater overall energy intake (standardized coef-
ficient 0.096, p 0.0001), which, in turn, was associated
with BMI-for-age in study Year 10. A linear representation
seems appropriate for the association between energy intake
and Year 10 BMI-for-age, given the significant linear com-
ponent (standardized coefficient 0.037, p 0.02) and
non-significant quadratic term (standardized coefficient
0.021, p 0.15). Finally, girls who were more physically
active consumed significantly more calories, possibly to
support their increased energy needs. After these associa-
tions were taken into account, breakfast did not exert any
additional, direct influence on Year 10 BMI-for-age (p
0.31). The model is consistent with the possibility that the
association between breakfast history and Year 10 BMI-for
age is explained by the mediating effects of energy intake
and physical activity.
Because girls who ate breakfast more often were sig-
nificantly more active, it is of interest to know whether
certain kinds of physical activity were associated with
breakfast history. Table 3 shows the association of spe-
Figure 1: Modeled association of BMI in study Year 10 as a function of breakfast history, by baseline BMI.
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
OBESITY Vol. 15 No. 9 September 2007 2287
cific types of physical activity with breakfast history.
Significance tests examined the association between
breakfast history and physical activity after adjusting for
site, race, and parental education. In the category of
sports, girls who ate breakfast more often were more
likely to participate in basketball or in weight training/
golf/badminton. In the category of individual physical
activities, breakfast history was associated with greater
rates of walking for exercise or bike riding; running/
soccer/track/field hockey; jogging/aerobics; and walking
to go someplace/mall walking. Finally, no categories of
housework/chores (such as cooking/dishwashing or mop-
Figure 3: Modeled association of BMI in study Year 10 as a function of physical activity (estimated in path model).
Figure 2: Parameter estimates in the path model of BMI in study Year 10. * p 0.05; ** p 0.01; *** p 0.0001.
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
2288 OBESITY Vol. 15 No. 9 September 2007
Table 3. Percent of girls participating in physical activities by breakfast history and type of physical activity
% days
eating
breakfast
Sports Individual physical activities
Softball,
baseball,
volleyball,
frisbee Basketball
Swimming
or tennis
Bowling
or
archery
Weight
training,
golf, or
badminton
Walking to
go to some
place or
walking at
a mall
Walking
for exercise
or bike
riding
Running,
soccer,
track, or
field
hockey
Jogging,
aerobics,
skating, or
rollerblading
Dancing,
cheerleading,
jumping
rope
Exercises,
gymnastics,
or baton
twirling
0–30% 0.0 0.0 1.9 0.0 0.0 75.9 7.4 1.9 1.9 24.1 9.3
31–40% 0.0 5.0 4.0 2.0 5.0 86.1 19.8 1.0 10.9 23.8 13.9
41–50% 2.8 2.8 8.4 2.2 4.5 77.7 19.0 6.1 6.7 22.9 10.6
51–60% 4.3 2.1 6.4 2.1 6.4 82.4 22.5 5.9 11.2 21.9 15.5
61–70% 1.3 3.5 5.1 2.2 5.7 84.8 21.5 7.3 9.8 22.5 14.2
71–80% 5.9 4.6 7.7 2.6 8.5 81.8 23.8 7.2 11.3 21.8 13.8
81–90% 2.0 4.9 10.0 2.0 7.7 86.5 25.2 10.0 16.0 17.5 13.8
91–99% 4.6 4.6 8.8 4.1 12.0 84.8 30.0 13.4 11.1 16.1 16.6
100% 1.6 1.6 14.8 3.3 13.1 91.8 31.1 9.8 14.8 16.4 18.0
p 0.5365 0.0353 0.9582 0.0777 0.0168 0.0318 0.0019 0.0257 0.0494 0.1948 0.2698
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
OBESITY Vol. 15 No. 9 September 2007 2289
ping/vacuuming) were significantly related to breakfast
consumption (results not shown).
Discussion
Breakfast eating is known to be related to positive health
outcomes in children (1,5). In the current study, we have
shown that the pattern of breakfast eating over time (“break-
fast history”) has an important association with weight and
physical activity. The association between breakfast con-
sumption and BMI-for-age among girls at the 10-year fol-
low up was dependent on baseline BMI. Specifically,
among girls who were relatively heavy at the beginning of
the study (95th or 97th percentile of BMI-for-age), frequent
breakfast consumption was significantly associated with
decreased BMI after 10 years, even after adjusting for
potential confounding variables (e.g., physical activity).
There was no association among girls who began the study
with weight in the normal range (50th percentile). Further-
more, the association between increased breakfast consump-
tion and Year 10 BMI may be explained by the mediating
effects of energy intake and physical activity.
Our findings suggest a pattern whereby breakfast eating,
overall energy intake, and activity levels are associated with
each other as well as with BMI. Girls who ate breakfast
more often consumed more energy, but they were also more
likely to participate in a variety of physical activities. That
is, breakfast consumption was associated with increased
caloric intake (predicting increased BMI-for-age), but this
appeared to be partially offset by the association between
breakfast consumption and physical activity (high physical
activity predicted decreased BMI-for-age). Although it is
not altogether clear which comes first, our data are consis-
tent with the interpretation that overweight girls who engage
in a regular pattern of eating breakfast may be more likely
to attain a healthier weight, perhaps in part by balancing
energy in with energy out. Possibly, girls who are heavy in
early adolescence are encouraged by their physicians and
parents to eat breakfast regularly and to be physically ac-
tive— behaviors that, in combination, lead to decreases in
BMI. Parents who encourage regular breakfast eating, par-
ticularly through the adolescent years when breakfast con-
sumption often decreases dramatically (1), may be more
likely to highlight the importance of overall healthy eating
and regular physical activity (2). Future research might
examine the role that families play in patterns of breakfast
eating over the course of childhood.
The pattern of results suggests that both physical activity
and breakfast history were associated with BMI-for-age
z-scores in the final year. Results of the path analysis model
are consistent with the possibility that, holding breakfast
history constant, increased physical activity over the 10
years is associated with decreased BMI z-scores in the final
year, at least at very high levels of physical activity as
suggested by the U-shaped association (Figure 3). Future
work should examine in more detail the combined and
independent influences of breakfast consumption and phys-
ical activity on adiposity.
Our findings and those of others indicate that simple
linear association poorly represents the relation between
breakfast and BMI. Generalizations such as “eating break-
fast more often is associated with decreased BMI” may not
be accurate. Instead, the relation is complex and is likely to
differ across groups (e.g., individuals differing on baseline
weight and physical activity). For example, we found that
regular breakfast consumption may be especially important
among relatively higher weight children, in that breakfast
history was related to a decrease in later BMI in this group.
Another complexity is that diverse patterns are likely to
underlie the breakfast history measure and different patterns
may be associated with different outcomes. For example,
consider two girls who ate breakfast on 67% of the days;
one girl might eat breakfast on 2 of 3 days during every
study year, while the second girl eats breakfast on 3 of 3
days during study Years 1 to 5, 1 of 3 days in study Year 7,
then she skips breakfast altogether thereafter. Future work
should examine such patterns and the associated outcomes.
Strategies for addressing childhood overweight include a
focus on caloric consumption, increase in physical activity,
and perhaps most importantly, the involvement of family in
making changes in eating behaviors and activity levels (22).
Closer inspection of obesity treatment programs (23,24)
indicates that nutrition education is an important compo-
nent, with an emphasis on regular meals and overall healthy
eating habits, among other topics. Highlighting the impor-
tance of regular breakfast consumption may be an important
addition to obesity treatment programs for children and
adolescents. Although we found that breakfast history was
associated with decreases in BMI among heavier girls, it is
not clear that breakfast consumption contributes to weight
maintenance among normal-weight girls; among girls who
began the study with median baseline BMI-for-age, break-
fast history was not associated with BMI-for-age in the final
study year.
The association between breakfast history and BMI in
Year 10 held even after controlling for breakfast consump-
tion in Year 10; this suggests that the longer-term history of
breakfast consumption is important, not just breakfast con-
sumption measured concurrently with BMI. Although past
research has examined the cross-sectional associations be-
tween breakfast eating in childhood and a variety of health
outcomes, additional studies are needed to determine
whether the same associations might be true longitudinally
between adolescence and adulthood.
Girls who ate breakfast more often were more likely to
engage in various physical activities, particularly individual
activities such as running, walking, and jogging, as well as
some team sports. It is not altogether clear why some forms
of physical activity were associated with patterns of break-
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
2290 OBESITY Vol. 15 No. 9 September 2007
fast eating whereas others were not, although it is possible
that season of the year may account for the differences
(certain sports are only played at certain times, which may
not have coincided with when the NGHS assessment took
place). The differences may also be accounted for by the
type of physical activity. Activities that have more of an
emphasis on appearance such as dancing, cheerleading, and
gymnastics, were not associated with more frequent break-
fast consumption in our study. These results are in contrast
to a recent study by Croll et al. (25), in which breakfast
consumption was significantly more frequent among girls
involved in weight-related sports (dancing, cheerleading,
gymnastics, yoga, ice skating, and wrestling) when com-
pared with girls involved in team sports (volleyball, basket-
ball, baseball, softball, hockey, football, soccer, and field
hockey) or girls not involved in any sports. However, the
analyses conducted by Croll et al. (25) adjusted for ethnicity
and socioeconomic status only, while the analyses in the
present study adjusted for Drive for Thinness and Body
Dissatisfaction. Our findings suggest that, when you take
these factors into account, the relationship between break-
fast eating and weight-related physical activities may not
hold in adolescent girls.
The frequency of breakfast eating is known to decrease as
children move into and through adolescence. Yet, our data
show the importance of breakfast consumption in regard to
weight and physical activity among overweight girls. In
earlier work (26), we showed that eating cereal at breakfast
was predictive of lower BMI, suggesting that the consump-
tion of specific foods might be more beneficial than con-
suming breakfast per se. Campaigns to address some ado-
lescent nutrition behaviors (e.g., increasing fruit and
vegetable intake) have been successful (27,28). One impli-
cation of our findings is that strategies designed to increase
breakfast consumption in adolescents are needed to reverse
the trend of skipping this important meal.
The quality of foods consumed at breakfast tended to
decrease over the years; as they grew older, girls consumed
more caffeine, sucrose, and sodium and less calcium at
breakfast. These changes may, in part, be due to the age-
related increases in soda and coffee consumption and de-
creases in milk consumption that have been documented in
the NGHS cohort (29). It is possible that the observed
associations between breakfast history and BMI might be
magnified in girls who tend to consume higher-quality
breakfasts; this deserves attention in future work.
A limitation is that our data focused only on girls of two
races; further studies are needed that include boys and other
ethnic groups. Another limitation is that this study was a
secondary analysis of survey data; some questions could not
be answered using this dataset. Although our prospective
data indicate longitudinal associations, causality and causal
direction cannot be determined on the basis of these data.
For example, the study design did not enable us to deter-
mine whether increased breakfast consumption causes in-
creased physical activity or vice versa. The paths estimated
in path analysis were based on the theory that increased
breakfast consumption leads to increased physical activity,
but the fact that the data were consistent with this theory
does not imply that the data would not also be consistent
with competing theories, e.g., a theory in which increased
physical activity leads to increased breakfast consumption.
An experimental design would be required to unambigu-
ously answer such questions about causal direction. These
limitations are offset by the lengthy follow-up period, low
attrition rate, large sample size, racial diversity, and the
objective assessment of weight and verification of nutri-
tional data.
Overall, our findings indicate an important relationship
between breakfast consumption and moderation of body
weight among girls of relatively higher BMI. Furthering our
understanding of the links between nutrition intake, meal
consumption, and physical activity is an important goal
toward addressing childhood and adolescent obesity.
Acknowledgments
This research was supported by General Mills, Inc., and
by a grant from the National Heart, Lung, and Blood Insti-
tute (HL/DK71122). Also supported by contracts
HC55023–26 and Cooperative Agreements U01-HL-
48,941– 44. Participating NGHS Centers included Chil-
dren’s Medical Center, Cincinnati, Ohio (Stephen R.
Daniels, Principal Investigator, John A. Morrison, Co-In-
vestigator); Westat, Inc., Rockville, Maryland (George B.
Schreiber, Principal Investigator, Ruth Striegel-Moore, Co-
Investigator), and University of California, Berkeley, CA
(Zak I. Sabry, Principal Investigator, Patricia B. Crawford,
Co-Investigator); and Maryland Medical Research Institute,
Baltimore, Maryland (Bruce A. Barton, Principal Investiga-
tor) served as the data coordinating center. Program Office:
National Heart, Lung, and Blood Institute (Eva Obarzanek,
Project Officer 1992 to present, Gerald H. Payne, Project
Officer 1985 to 1991).
References
1. Affenito SG, Thompson DG, Barton BA, et al. Breakfast
consumption by African-American and White adolescent girls
correlates positively with calcium and fiber intake and nega-
tively with Body Mass Index. J Am Diet Assoc. 2005;105:
938 45.
2. Vanelli M, Iovane B, Bernardini A, et al. Breakfast habits of
1,202 northern Italian children admitted to a summer sport
school: breakfast skipping is associated with overweight and
obesity. Acta Biomed Ateneo Parmense. 2005;76:79 85.
3. Cohen B, Evers S, Manske S, Bercovitz K, Edward HG.
Smoking, physical activity and breakfast consumption among
secondary school students in a southwestern Ontario commu-
nity. Can J Public Health. 2003;94:41– 4.
4. Godin G, Anderson D, Lambert LD, Desharnais R. Iden-
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
OBESITY Vol. 15 No. 9 September 2007 2291
tifying factors associated with regular physical activity in
leisure time among Canadian adolescents. Am J Health Pro-
mot. 2005;20:20 –7.
5. Chen X, Sekine M, Hamanishi S, et al. Lifestyles and
health-related quality of life in Japanese school children: a
cross-sectional study. Prev Med. 2005;40:668 –78.
6. Lytle LA, Seifert S, Greenstein J, McGovern P. How do
children’s eating patterns and food choices change over time?
Results from a cohort study. Am J Health Promot. 2000;14:
222– 8.
7. Nicklas TA, Morales M, Linares A, et al. Children’s meal
patterns have changed over a 21-year period: the Bogalusa
Heart Study. J Am Diet Assoc. 2004;104:753– 61.
8. Siega-Riz AM, Popkin BM, Carson T. Trends in breakfast
consumption for children in the United States from 1965–
1991. Am J Clin Nutr. 1998;67(suppl):748 –56.
9. Dwyer JT, Evans M, Stone EJ, et al. Child and Adolescent
Trial for Cardiovascular Health (CATCH) Cooperative Re-
search Group: adolescents’ eating patterns influence their nu-
trient intakes. J Am Diet Assoc. 2001;101:798 802.
10. Berkey CS, Rockett HRH, Gillman MW, Field AE, Colditz
GA. Longitudinal study of skipping breakfast and weight
change in adolescents. Int J Obes Relat Metab Disord. 2003;
27:1258 66.
11. Aarnio M, Winter T, Kujala U, Kaprio J. Associations of
health related behavior, social relationships, and health status
with persistent physical activity and inactivity: a study of
Finnish adolescent twins. Br J Sports Med. 2002;36:360 4.
12. NHLBI Study Group. Obesity and cardiovascular disease
risk factors in black and white girls: the NHLBI Growth and
Health Study. Am J Public Health. 1992;82:1613–20.
13. Crawford PB, Obarzanek E, Morrison J, Sabry ZI. Com-
parative advantage of 3-day food records to 24-hour recall or
5-day food frequency validated by observation of 9- and
10-year-old girls. J Am Diet Assoc. 1994;94:626 –30.
14. Schakel GH, Sievert YA, Buzzard IM. Sources of data for
developing and maintaining a nutrient database. J Am Diet
Assoc. 1988;88:1268 –71.
15. Ross CE, Mirowsky J. Refining the association between
education and health: the effects of quantity, credential, and
selectivity. Demography. 1999;36:445– 60.
16. Centers for Disease Control and Prevention. CDC Growth
Charts, 2000. http://www.cdc.gov/nchs/about/major/nhanes/
growthcharts/clinical charts.htm (Accessed October 15, 2006).
17. Kimm SY, Glynn NW, Kriska AM, Fitzgerald SL, Aaron
DJ, Similo SL. Longitudinal changes in physical activity in a
biracial cohort during adolescence. Med Sci Sports Exerc.
2000;32:1445–54.
18. Garner DM, Olmsted MP. Eating Disorders Inventory
Manual. Odessa, FL: Psychological Assessment Resources;
1984.
19. Eberenz KP, Gleaves DH. An examination of the internal
consistency and factor structure of the eating disorder
inventory-2 in a clinical sample. Int J Eat Disord. 1994;16:
371–9.
20. Franko DL, Striegel-Moore RH, Barton BA, et al. Measur-
ing eating concerns in Black and White adolescent girls. Int J
Eat Disord. 2004;35:179 89.
21. Freedman DS, Khan LK, Serdula MK, Dietz WH, Srini-
vasan SR, Berenson GS. Inter-relationships among child-
hood BMI, childhood height, and adult obesity: the Bogalusa
Heart Study. Int J Obes Relat Metab Disord. 2004;28:10 6.
22. American Dietetic Association (ADA). Position of the
American Dietetic Association: individual-, family-, school-,
and community-based interventions for pediatric overweight.
J Am Diet Assoc 2006;106:925– 45.
23. Epstein LH, Myers MD, Raynor HA, Saelens BE. Treat-
ment of pediatric obesity. Pediatrics. 1998;101:554 –70.
24. Nemet D, Barkan S, Epstein Y, Freidland O, Kowen G,
Eliakim A. Short- and long-term beneficial effects of a
combined dietary-behavioral-physical activity intervention
for the treatment of childhood obesity. Pediatrics. 2005;
115:443–9.
25. Croll JK, Neumark-Sztainer D, Story M, Wall M, Perry C,
Harnack L. Adolescents involved in weight-related and
power team sports have better eating patterns and nutrient
intakes than non-sport-involved adolescents. J Am Diet Assoc.
2006;106:709 –17.
26. Barton BA, Eldridge AL, Thompson D, et al. The relation
of breakfast and cereal consumption to nutrient intake and
body mass index: the National Heart, Lung and Blood
Institute Growth and Health Study. J Am Diet Assoc. 2005;
105:1383–9.
27. Nicklas TA, O’Neil CE. Process of conductinga5ADay
intervention with high school students: Gimme 5 (Louisiana).
Health Educ Behav. 2000;27:201–12.
28. Nicklas TA, Johnson CC, Myers L, Farris RP, Cunning-
ham A. Outcomes of a high school program to increase fruit
and vegetable consumption: Gimme 5—a fresh nutrition con-
cept for students. J Sch Health. 1998;68:248 –53.
29. Striegel-Moore RH, Thompson D, Affenito SG, et al. Cor-
relates of beverage intake in adolescent girls: the National
Heart, Lung, and Blood Institute Growth and Health Study.
J Pediatr. 2006;148:183–7.
Breakfast Eating Patterns in Adolescent Girls, Albertson et al.
2292 OBESITY Vol. 15 No. 9 September 2007
    • "Research on lunch skipping and dinner skipping in young children is scarce; however, studies among older children (7-to 13-year-olds) show a higher prevalence of lunch skipping and dinner skipping353637. Since meal skipping is known to increase with age throughout childhood and adolescence [33, 38], the age difference between studies may explain why we found a lower prevalence of lunch and dinner skipping. Few studies have investigated the co-occurrence of meal skipping behaviors among young children and therefore direct comparison of the correlation coefficients found in the current study is precluded. "
    [Show abstract] [Hide abstract] ABSTRACT: Background: Regular meal consumption is considered an important aspect of a healthy diet. While ample evidence shows social inequalities in breakfast skipping among adolescents, little is known about social inequalities in breakfast skipping and skipping of other meals among young school-aged children. Such information is crucial in targeting interventions aimed to promote a healthy diet in children. Methods: We examined data from 4704 ethnically diverse children participating in the Generation R Study, a population-based prospective cohort study in Rotterdam, the Netherlands. Information on family socioeconomic position (SEP), ethnic background, and meal skipping behaviors was assessed by parent-reported questionnaire when the child was 6 years old. Multiple logistic regression analyses were performed to assess the associations of family SEP (educational level, household income, employment status, family composition) and ethnic background with meal skipping behaviors, using high SEP children and native Dutch children as reference groups. Results: Meal skipping prevalence ranged from 3% (dinner) to 11% (lunch). The prevalence of meal skipping was higher among low SEP children and ethnic minority children. Maternal educational level was independently associated with breakfast skipping ([low maternal educational level] OR: 2.21; 95% CI: 1.24,3.94). Paternal educational level was independently associated with lunch skipping ([low paternal educational level] OR: 1.53; 95% CI: 1.06,2.20) and dinner skipping ([mid-high paternal educational level] OR: 0.39; 95% CI: 0.20,0.76). Household income was independently associated with breakfast skipping ([low income] OR: 2.43, 95% CI: 1.40,4.22) and dinner skipping ([low income] OR: 2.44; 95% CI: 1.22,4.91). In general, ethnic minority children were more likely to skip breakfast, lunch, and dinner compared with native Dutch children. Adjustment for family SEP attenuated the associations of ethnic minority background with meal skipping behaviors considerably. Conclusion: Low SEP children and ethnic minority children are at an increased risk of breakfast, lunch, and dinner skipping compared with high SEP children and native Dutch children, respectively. Given these inequalities, interventions aimed to promote regular meal consumption, breakfast consumption in particular, should target children from low socioeconomic groups and ethnic minority children. More qualitative research to investigate the pathways underlying social inequalities in children's meal skipping behaviors is warranted.
    Full-text · Article · Aug 2015
    • "A Finnish study reported regular breakfast eating was consistently associated with good health and the individuals were more inclined to engage in physical activity [26]. Albertson et al. reported that physical activity is closely related to significantly decreased BMI which closely relate to our findings [27]. Our study also reports that females tend to sleep less than males, probably because sleep disorders are prevalent among females [28]. "
    Article · Nov 2014 · Journal of Adolescent Health
    • "Developing and maintaining regular physical activity and healthful dietary intake habits during adolescence contributes to several positive health outcomes, including healthy weight maintenance, into the adult years123. Conversely, adolescents' use of dieting and unhealthy weight control behaviors including skipping meals and using diet pills and laxatives may be ineffective in achieving weight loss and may even contribute to excessive weight gain and other negative health outcomes [4, 5]. "
    [Show abstract] [Hide abstract] ABSTRACT: Parental encouragement for healthy eating and physical activity has been found to be associated with the long-term healthy habits of adolescents, whereas parental encouragement to diet has been associated with disordered eating behaviors among adolescents. However, little is known about how parental encouragement changes as adolescents grow older (longitudinal trends), or how parental encouragement has changed over time (secular trends). This study examined 5-year longitudinal and secular trends in adolescents' report of their parents' encouragement to eat healthily, be physically active, and diet. Project Eating Among Teens surveyed a cohort of Minnesota adolescents (n = 2,516) in the years 1999 and 2004. Mixed-model regressions were used to assess changes in adolescents' reports of parental encouragement from early to middle adolescence (middle school to high school) and from middle to late adolescence (high school to post-high school), and secular changes in parental encouragement among middle adolescents between the years 1999 and 2004. Longitudinally, there were significant decreases in parental encouragement to eat healthy food, be active, and diet between early and middle adolescence. Between middle and late adolescence, among males parental encouragement for all behaviors decreased, whereas among females parental encouragement to diet increased. Few secular changes in parental encouragement were observed between 1999 and 2004. Given the importance of parental support for healthy eating and physical activity, efforts should be made to help parents maintain a high level of encouragement for their children's healthy behavior throughout adolescence. Parents of late adolescent females should aim to decrease the pressure on their daughters to diet during these critical developmental years.
    Full-text · Article · Sep 2011
Show more