Ebbeling CB, Sinclair KB, Pereira MA, Garcia-Lago E, Feldman HA, Ludwig DS. Compensation for Energy Intake From Fast Food Among Overweight and Lean Adolescents. JAMA. 2004;291(23):2828-2833. doi:10.1001/jama.291.23.2828
Author Affiliations: Division of Endocrinology, Department of Medicine, Children's Hospital, Boston, Mass (Drs Ebbeling, Feldman, and Ludwig and Mss Sinclair and Garcia-Lago); and Division of Epidemiology, University of Minnesota, Minneapolis (Dr Pereira).
Context Fast food consumption has increased greatly among children in recent
years, in tandem with the obesity epidemic. Fast food tends to promote a positive
energy balance and, for this reason, may result in weight gain. However, if
fast food and obesity are causally related, the question arises of why some
children who frequently eat fast food do not become overweight.
Objective To test the hypothesis that overweight adolescents are more susceptible
to the adverse effects of fast food than lean adolescents.
Design and Setting In study 1, we fed participants an "extra large" fast food meal in a
naturalistic setting (a food court). The participants were instructed to eat
as much or little as desired during this 1-hour meal. In study 2, we assessed
energy intake under free-living conditions for 2 days when fast food was consumed
and 2 days when it was not consumed. Data were collected in Boston, Mass,
between July 2002 and March 2003.
Participants Overweight (n = 26) and lean (n = 28) adolescents aged 13 to 17 years.
Overweight was defined as a body mass index exceeding sex- and age-specific
85th percentiles based on the 2000 Centers for Disease Control and Prevention
Main Outcome Measures Energy intake determined by direct observation in study 1 and by unannounced
24-hour dietary recalls, administered by telephone, in study 2.
Results In study 1, mean (SEM) energy intake from the fast food meal among all
participants was extremely large (1652  kcal), accounting for 61.6% (2.2%)
of estimated daily energy requirements. Overweight participants ate more than
lean participants whether energy was expressed in absolute terms (1860 
vs 1458  kcal, P = .02) or relative to estimated
daily energy requirements (66.5% [3.1%] vs 57.0% [2.9%], P = .03). In study 2, overweight participants consumed significantly
more total energy on fast food days than non–fast food days (2703 
vs 2295  kcal/d; +409  kcal/d; P = .02),
an effect that was not observed among lean participants (2575  vs 2622
 kcal/d; –47  kcal/d; P = .76).
Conclusions In this study, adolescents overconsumed fast food regardless of body
weight, although this phenomenon was especially pronounced in overweight participants.
Moreover, overweight adolescents were less likely to compensate for the energy
in fast food, by adjusting energy intake throughout the day, than their lean
Consumption of fast food has increased rapidly since the 1970s1 among adolescents from all socioeconomic and racial/ethnic
groups across the United States.2,3 Fast
food is ubiquitously available and heavily marketed to adolescents.4 An estimated 75% of adolescents eat fast food 1 or
more times per week.5
The increase in fast food consumption parallels the escalating obesity
epidemic,6 raising the possibility that these
2 trends are causally related. Characteristics of fast food previously linked
to excess energy intake or adiposity include enormous portion size,7 high energy density,8 palatability,9 excessive amounts of refined starch and added sugars,10 high fat content,11 and
low levels of dietary fiber.12 Previous studies,
which used between- and within-subject comparisons, consistently demonstrate
that consumption of fast food is directly related to total energy intake and
inversely related to diet quality.2,5,13- 16 Some
not all,5,15 have found a direct
association between fast food and body weight.
These studies raise a fundamental question: if most children eat fast
food regularly, why do some become overweight, whereas others do not? Perhaps
certain individuals are susceptible and others relatively resistant to the
adverse effects of fast food. Therefore, we hypothesized that adolescents
who eat fast food regularly but are not overweight compensate for the excessive
energy in a fast food meal by commensurately decreasing energy intake throughout
the day; in contrast, overweight adolescents do not have this tendency.
The purpose of this investigation, which was composed of 2 studies,
was to evaluate the effects of fast food on energy intake in overweight vs
lean adolescents. In study 1, we assessed energy intake during a fast food
meal consumed in a naturalistic setting. In study 2, we compared energy intake
under free-living conditions on days when fast food was consumed and days
when it was not consumed.
We enrolled 54 adolescents (26 overweight, 28 lean) aged 13 to 17 years
who reported eating fast food at least 1 time per week. Fifty-one (24 overweight,
27 lean) of the 54 participants enrolled in study 1 also completed study 2.
Newspaper advertisements and fliers, stating that the purpose of the project
was to collect information on why and how teenagers eat fast food, were used
to recruit participants.
Weight and height were measured using an electronic scale (model 6702,
Scale-Tronix, White Plains, NY) and a wall-mounted stadiometer (Holtain Limited,
Crymych, Wales), respectively. Body mass index (BMI) was calculated as weight
in kilograms divided by the square of height in meters. The Centers for Disease
Control and Prevention defines childhood overweight as
a BMI exceeding sex- and age-specific 95th percentiles and at risk of overweight as a BMI between the 85th and 95th percentiles,
using the 2000 growth charts.18 In this investigation,
we grouped adolescents who were overweight and at risk of overweight and herein refer to them as overweight.18 Adolescents
with a BMI not exceeding the 85th percentiles were considered lean. We did not enroll adolescents with a BMI below the 50th percentile
or above the 98th percentile and also excluded those diagnosed as having any
major medical illness or eating disorder. None of the participants was taking
prescription medications or dieting for the purpose of weight loss. As incentive,
we offered each participant two $30 gift certificates, one following completion
of each study.
The protocol was approved by the institutional review board at Children's
Hospital, Boston, Mass. Written informed consent and assent were obtained
from parents and participants, respectively. Data were collected between July
2002 and March 2003.
Participation involved 1 study visit. We instructed the participants
to eat a standard breakfast of cold cereal and milk at 8:30 AM on
the day of the visit and then to refrain from eating and drinking (except
water) until after the visit. At 1 PM, we fed the participants
a fast food meal from a national chain at a food court. All feedings were
conducted in groups of 4 participants, on average, to foster socializing that
is often part of the fast food experience among adolescents. Participants
were grouped by sex and weight status to avoid any self-consciousness about
eating that may be associated with these variables (eg, girls eating less
in the presence of boys, overweight adolescents eating less in the presence
of their lean peers).
The same meal, modeled after prevailing "extra large" fast food fare
(Table 1), was served to each
participant. The following standard instructions were read to the participants
before the meal: "In a few minutes, we will bring each of you a meal. Eat
as much or as little as you like, until you have had enough. There is more
food available, and you may eat as much as you want. Please do not share your
food with others in the group. If you need more of anything, just ask." The
length of the meal was 1 hour. During this time, a research assistant discreetly
monitored food intake to ensure that ample food was always available.
Whenever approximately three fourths of the meal portion of chicken
nuggets, fries, or cookies was consumed, a refill portion of the item was
added to the tray (Table 1). Empty
cola containers were immediately replaced with full containers. Participants
could obtain ketchup and sweet and sour sauce from the middle of the table
at any time during the meal. This standardized protocol allowed us to provide
more of the items that each individual enjoyed the most and, thus, would be
likely to order in large portions if given the option. Following the meal,
each participant estimated the relative size of the meal consumed during the
study compared with the size of fast food meals that he or she typically consumed,
using a verbally anchored, 10-cm visual analog scale, ranging from "much smaller
than usual" to "much larger than usual."
The difference in weight between the amount of each menu item provided
and that remaining on the tray after the meal was used to calculate energy
intake. In preparation for data collection, 20 reference units of each menu
item were purchased and weighed to evaluate variability in portion sizes.
Coefficients of variation, ranging from 0.8% for a packet of sweet and sour
sauce to 9.2% for an order of "extra large" french fries, confirmed that portions
are highly standardized. Thus, amounts of food provided during the feeding
study were estimated based on mean weights of the reference units. Using this
method, we were able to serve food immediately after purchasing it, thereby
maintaining the temperature, palatability, and visual appearance that are
expected by consumers. Leftovers were weighed on an electronic scale (item
E1D120, Ohaus Corporation, Florham Park, NJ). The Nutrition Data System for
Research Software (NDS-R; versions 4.04 and 4.05, Nutrition Coordinating Center,
University of Minnesota, Minneapolis) was used to convert the gram weight
consumed to energy intake (in kilocalories). We relied on the NDS-R, rather
than nutrition information available from the restaurant, to allow direct
comparison with 24-hour dietary recall data collected for study 2.
Four dietary and physical activity recall interviews, 2 for fast food
days and 2 for non–fast food days, were administered by telephone to
assess energy intake under free-living conditions. We used the NDS-R multiple-pass,
24-hour dietary recall method, which prompted the participant to list in sequence
what foods and beverages were consumed during the preceding day, identify
gaps in the initial list, and then provide details concerning each reported
item. At the end of each recall, participants were asked to confirm the information
provided and categorize the amount of food intake for the day as "usual,"
"more than usual," or "less than usual." Physical activity was quantified
using a 24-hour recall protocol modeled after the method of Pate et al.19 In brief, participants were asked to recall the activity
performed most during respective 15-minute time blocks throughout the day
and then to rate the relative intensity of each activity as light, moderate,
hard, or very hard. A metabolic equivalent (MET level) was assigned to each
reported activity to calculate a physical activity factor. As points of reference,
resting has a MET level of 1.0, and brisk walking has a level of 5.0.20 Total energy expenditure (in kilocalories per day)
was estimated by multiplying basal metabolic rate, calculated from validated
Food and Agriculture Organization, World Health Organization, United Nations
University equations that include weight and height as independent variables,21 by the physical activity factor derived from the
four 24-hour recalls.
Two criteria were used to define a fast food day. Criterion 1 specified
that the participant eat at 1 of the 5 leading fast food establishments: McDonald's,
Burger King, KFC, Wendy's, or Taco Bell.22 Criterion
2 specified that the participant consume at least 1 menu item containing meat
(beef, pork), chicken, fish, beans, or egg plus 1 additional item (eg, fries,
beverage, dessert). A non–fast food day was one that did not meet criterion
1. Days when participants ate at other restaurants, including pizza and sandwich
shops, were classified as non–fast food days. Because we were evaluating
the effects of fast food meals, as opposed to single menu items, intake was
not assessed on days when criterion 1 but not criterion 2 was satisfied. Recalls
were unannounced, to avoid reactivity, and conducted on nonconsecutive days.
On average, we contacted each participant a mean (SEM) of 6.9 (0.3) times
to obtain data for 4 days, including 2 fast food days that satisfied both
Underreporting of dietary intake is a well-recognized phenomenon, particularly
among overweight adolescents, but little is known regarding differential underreporting
among foods.23- 25 This
phenomenon could bias data in study 2 in either direction: against our primary
hypothesis if energy intake from fast food were selectively underreported,
or in favor of the hypothesis if energy intake from fast food were reported
more completely than energy intake from other foods. To evaluate the potential
for bias, we examined underreporting of total energy intake and energy intake
from fast food in overweight and lean participants, using data from both studies.
Recalled total daily energy intake (study 2), averaged across 2 fast food
days and 2 non–fast food days, was expressed as a percentage of estimated
total energy expenditure to assess the accuracy of self-report of total energy
intake. Recalled energy intake from fast food (study 2), averaged across the
2 fast food days, was expressed as a percentage of observed energy intake
during the fast food feeding (study 1) to assess the accuracy of self-report
of fast food energy intake.
Statistical analyses were conducted using SAS statistical software (release
8.2, SAS Institute Inc, Cary, NC). For study 1, 2-sample t tests were used to compare energy intake during the meal between
overweight and lean adolescents. For study 2, analysis of variance was performed
using the mixed linear model procedure to evaluate whether the interaction
between weight status (overweight vs lean) and type of day (fast food days
vs non–fast food days) influenced total daily energy intake. In an additional
model, we adjusted for self-reported relative amount of food intake. Preplanned
contrasts were estimated from the fitted models for overweight and lean adolescents
to determine the effects of fast food on total daily energy intake within
groups. For the interstudy comparison, a mixed linear model was used to compare
the accuracy of self-report between overweight and lean participants. The
model for evaluating self-report of fast food intake was adjusted for the
relative meal size rating in study 1. Using a 5% type I error rate, we estimated
that a sample of 50 participants (25 overweight, 25 lean) would provide 80%
power to detect a between-group difference in energy intake of approximately
150 kcal in study 1 and a difference in effect of approximately 260 kcal between
overweight and lean participants in study 2. All results are presented as
mean (SEM). Statistical significance was defined as P<.05.
Participant characteristics are presented in Table 2. There were no significant differences in demographic variables
(sex, race, age) or height between the overweight and lean participants. The
overweight adolescents tended to be less physically active than their lean
counterparts (P = .06) and tended to have a higher
total energy expenditure (P = .07).
When instructed to eat as much or little fast food as desired, the participants
consumed 1652 (87) kcal, amounting to 61.6% (2.2%) of the estimated total
energy expenditure. Overweight participants ate more than lean participants,
whether energy intake was expressed in absolute terms or relative to estimated
needs (Table 3). Relative meal
size ratings did not differ between the overweight and lean adolescents (8.5
[0.4] vs 7.8 [0.5], P = .22).
There was a significant interaction between type of day (fast food vs
non–fast food day) and weight status (overweight vs lean) for total
daily energy intake (P = .05 unadjusted, P = .04 after adjustment for self-reported relative amount of food
intake). Overweight participants consumed 409 (142) kcal/d more on fast food
than non–fast food days. In contrast, energy intake was not significantly
different on fast food and non–fast food days for the lean participants
(Table 4). With regard to physical
activity, there was no interaction between type of day and weight status (P = .46).
We sought evidence for incomplete reporting of food intake by examining
observed dietary intake data in study 1 and recalled intake data in study
2 (Table 5). Recalled total daily
energy intake, expressed as a percentage of estimated total energy expenditure,
tended to be lower for the overweight compared with lean participants (−15.3%
[8.9%], P = .09). Recalled energy intake from fast
food in study 2, compared with observed intake in study 1, was also lower
for the overweight compared with lean participants (−17.3% [8.7%], P = .05). Adjustment for relative meal size rating in study
1 did not materially affect this difference (−15.0% [8.6%], P = .09). Thus, as expected, overweight participants tended to underreport
total energy intake compared with lean participants; however, the group difference
in reporting accuracy was similar for total and fast food energy intake (−15.3%
vs −17.3%, P = .84), providing evidence against
the possibility of a false-positive result.
In 1989, a published statement warned that a lifetime of fast food consumption
may place children at increased risk for obesity.26 However,
until recently, the potentially adverse effects of fast food in youth have
received limited attention in the medical literature.2,5 With
increasing recognition that excess adiposity confers serious health risks
and that environmental factors may be driving the obesity epidemic,27 the role of fast food in promoting obesity has emerged
as a topic of great interest and debate. Some nutrition professionals argue
that fast food is contributing to the obesity epidemic,27,28 whereas
others support industry claims that fast food can be part of a healthful diet.29
Herein, we present the first investigation, to our knowledge, designed
to evaluate the effects of fast food on energy intake in overweight vs lean
adolescents. Assuming a dietary pattern of 3 meals and 1 or 2 snacks per day,
average meal size to maintain energy balance should not exceed approximately
30% of daily energy requirements or approximately 790 kcal in our study sample.
Compared with this figure, the participants in study 1 massively overate (1652
kcal or 61.6% of estimated total energy expenditure) in the naturalistic setting
of a food court. Overeating, observed in both groups of participants, was
especially pronounced among the overweight. Moreover, the overweight participants
consumed more total energy on days with than without fast food, in contrast
to the lean participants, who consumed virtually the same amount on both days.
This observation suggests that overweight individuals do not compensate completely
for the massive portion sizes characteristic of fast food today.
There are several ways that an individual could maintain energy balance
throughout a day that included large portions of fast food: by decreasing
food intake subsequent to a fast food meal, by decreasing food intake in anticipation
of a fast food meal, or by adjusting the size of a fast food meal based on
how much of other foods have been or will be consumed. Our study does not
allow us to determine in which of these ways the lean and overweight participants
differed. We also cannot determine whether the failure to compensate fully
for energy from large fast food meals is an inherent trait, causing obesity
in susceptible individuals, or a secondary event that occurs after development
of obesity. Nevertheless, these findings suggest that, at least, fast food
consumption serves to maintain or exacerbate obesity in susceptible individuals.
Although excess energy intake in response to large portions is not unique
to fast food,7,30- 32 we
focused on this dietary pattern because of its dominant position in adolescents'
diets and the possibility of a causal link to the obesity epidemic. Indeed,
fast food is designed to promote consumption of a maximum of energy in a minimum
of time, a precept of not only the business model but also the very name.
Other dietary scenarios (eg, a buffet) might also provoke overeating and incomplete
energy compensation if they resembled fast food in critical respects, including
high energy density, low fiber content, extensive food processing (facilitating
rapid swallowing with minimal chewing), and low satiating value. In those
scenarios, however, the distinction with fast food may be more one of terminology
or marketing than physiology. By contrast, overeating to the magnitude observed
in study 1 would be virtually impossible with satiating, low-energy-density,
high-fiber foods that require much chewing before swallowing (eg, fruits,
vegetables, legumes, whole grain products).
Several issues that pertain to study design should be noted. Strengths
include evaluation of energy intake in a naturalistic setting in study 1 and
within-subject comparisons in study 2, reducing the possibility of confounding
by demographic and behavioral factors. Limitations include a relatively small
sample size, restricted generalizability, and reliance on self-report for
assessment of energy intake in study 2 (a methodologic issue common to all
studies that aim to assess diet under free-living conditions).
Consistent with previous studies23,24 that
show that overweight participants have a particularly strong tendency to underreport
what they eat, self-reported energy intake on non–fast food days in
study 2 was lower for the overweight compared with the lean adolescents. However,
owing to the within-subject design, underreporting would lead to a false-positive
result only if energy intake from fast food were reported more completely
(ie, less underreporting) than total energy intake by the overweight vs lean
adolescents. The interstudy comparison suggests that this was not the case.
Total daily energy intake in study 2, expressed as a percentage of total energy
expenditure, was lower for the overweight than the lean adolescents. Recalled
energy intake from the fast food meals in study 2, expressed as a percentage
of intake observed in study 1, was also lower for the overweight adolescents.
However, the magnitude of underreporting of energy intake from fast food compared
with total daily energy intake by the overweight vs lean participants was
similar, even after adjustment for meal size rating, suggesting that fast
food was not reported more completely than other foods. Moreover, prior studies33- 35 suggest that the
opposite is likely to occur: overweight individuals may report high-calorie
foods perceived as "fattening" (eg, fast food) less, rather than more, completely
than other foods. This effect, if present, would bias the study toward the
In conclusion, our investigation suggests that overweight adolescents
are less likely to compensate for the energy in large portions of fast food
than their lean counterparts. These findings do not imply that fast food is
without detrimental effect in lean adolescents. Previous research has shown
that fast food consumption among children in a nationally representative sample
affects diet quality in ways that would plausibly increase risk for obesity,
regardless of baseline body weight.2 Although
the causes of obesity are multifaceted (as emphasized by the fast food industry22), public health measures to limit fast food consumption
in children may be warranted. Such measures could include nutrition education
campaigns, legislation to regulate marketing of fast food to children, and
elimination of fast food from schools.