Total energy intake, adolescent discretionary behaviors and the energy gap



To estimate total energy intake and the energy gap—the daily imbalance between energy intake and expenditure—associated with discretionary behaviors of adolescents, namely their leisure active behaviors (playing or participating in sports and heavy chores), leisure sedentary behaviors (television (TV) viewing and playing video and computer games), productive sedentary behaviors (reading or doing homework).


Prospective observational study.


A total of 538 students (mean age at baseline=11.7 years) from public schools in the Boston area studied prospectively from the fall of 1995 to the spring of 1997.


Anthropometric assessment including height and weight, dietary assessment using a youth food frequency questionnaire and measures of TV, video, reading/doing homework and youth physical activity.


We estimate the change in total energy intake for each hour change in discretionary activity using regression methods. A 1-h increase in watching TV is associated with a 106 kcal h−1 increase in total energy intake (95% confidence interval (CI): 61–150 kcal day−1). A similar change of 92 kcal h−1 (95% CI: 37–147 kcal day−1) is seen with playing video and computer games. The change in energy intake associated with an hour change in physical activity is 292 kcal h−1 (95% CI: 262–321 kcal day−1). No significant change is associated with reading/doing homework. Assuming that typical energy expenditures are associated with these behaviors, reading/doing homework appears to be an ‘energy neutral’ activity, whereas watching TV and playing video and computer games is associated with an energy surplus. If we assume that physical activity levels are moderate (3.5 METs), then this is also an energy surplus activity. If physical activity is assumed to be vigorous for the entire time allotted (>6.0 METs), an energy deficit could be achieved. We validated these estimates by calculating regressions predicting change in weight. Results indicate that each hour increase in TV viewing is associated with a weight increase of 0.38 kg (95% CI: 0.17–0.59 kg), with no significant associations for the other behaviors. A model with change in BMI as the dependent variable produced similar results.


Watching TV is an activity associated with a daily energy surplus. Although physical activity is thought of as an energy deficit activity, our estimates do not support this hypothesis. Reading/doing homework is the only discretionary activity examined which appears to be clearly energy neutral. The differential impacts of these discretionary behaviors on energy intake and the energy gap are discussed in relation to food-related advertisements aimed at children and adolescents.


Childhood obesity rates have tripled in the United States in the past four decades.1 Despite wide variations in secular prevalence trends, with levels of overweight and obesity averaging above 20% in the Americas and Europe,2 few countries remain unaffected by these increases.3 On the individual level, obesity is the result of a daily imbalance between energy intake and expenditure, known as the energy gap.4 Simply put, a child who eats more energy than needed for basic metabolic demands and growth and that expended through physical activity will gain excess weight. Among all children aged 2–7 years in the United States, it appears that an energy gap of 110–165 kcal day−1 has driven excess weight gain over the past decade. The estimated gap leading to overweight among adolescents is substantially larger.4

Achieving energy balance is certainly challenging in today's obesogenic environment. Strategies for reducing the energy gap can be divided into two broad categories: those that reduce energy intake and those that increase energy expenditure. Modifying the current food environment is fundamental to reducing the energy gap on a population level. Commonly employed strategies to reduce excess energy intake at the individual level include: (1) providing structured energy-reduced eating plans,5 (2) restricting common food and drink sources of excess energy and limited nutritional value such as sugar sweetened beverages,6 (3) modifying diet quality to maximize satiety and decrease ad libitum intake7 and (4) reduce time spent in activities (such as television (TV) watching) that are associated with both inactivity and excessive consumption of energy.8

Strategies used to increase energy expenditure generally focus on increasing moderate and vigorous physical activity, as this is the only modifiable component of energy expenditure.9, 10 Physical activity is largely responsible for the variation among individuals in energy expenditure, and therefore total energy needs.10 As such, there is an expectation that increases in physical activity can offset the energy gap by creating an energy deficit.

Although promoting physical activity in children is a widely adopted obesity prevention strategy, it is not clear what effect increasing physical activity will have on food intake and thus on energy balance. Because physical activity is a major determinant of energy intake,10 a concomitant increase in energy intake with increases in physical activity because of hunger should be expected. As a result, the impact of changing the frequency, duration or intensity of physical activities on the energy gap and obesity risk warrants exploration. Simply put, when a child increases her physical activity level, is there also a change in her energy intake, and is this less than or equal to the amount of energy expended during the activity?

This question leads to even broader questions that relate to how children spend their time. Are there activities that are associated with a daily energy surplus and others that are associated with an energy deficit? What activities should be recommended as replacements for activities that create an energy surplus?

To begin to answer these questions, we consider how children and adolescents spend their discretionary time. Most weight-related research to date has focused on leisure physical activity and leisure sedentary behaviors such as TV viewing. Productive sedentary behaviors, such as reading and doing homework, are distinct from leisure sedentary activities in that books are generally seen as free of the food advertisements that permeate TV. Although early observations indicate that productive sedentary behaviors such as time spent reading or doing homework are not associated with obesity,11 there has been very limited subsequent exploration of this variable.

In this analysis, we utilize longitudinal data on youth to estimate the impact of changes in discretionary behaviors on total energy intake using regression methods that allow control over both measured and unmeasured potentially confounding variables.12 Whether increases in energy intake associated with these behaviors exceed the energy expended is estimated by assuming standard levels of energy expenditure associated with different activities. We estimate which behaviors are ‘energy neutral’ and which are associated with an energy surplus or deficit. Discretionary behaviors that we consider include leisure active behavior (playing or participating in sports and heavy chores), leisure sedentary behavior (TV viewing and playing video and computer games) and productive leisure behavior (reading or doing homework). We further validate these findings with analyses of change in weight and BMI.


Protocol and participants

The present analysis was motivated by the hypothesis that changes in time spent in discretionary activities by adolescents would predict changes in total energy intake. In this prospective observational analysis, the primary hypotheses were that changes in the discretionary behaviors would directly predict changes in total energy intake over the 21 months of this study. Demonstration that change in an independent variable predicts change in a dependent variable may provide stronger evidence for causality than predictions involving the independent variable measured at just one point in time (for example, baseline).13 In addition, as described below, the regression methods provide further control for confounding.

Data for this prospective observational analysis were obtained from the Planet Health study, a randomized controlled field trial conducted in five intervention and five control schools in four communities in the Boston metropolitan area. A detailed description of the study design and participant characteristics has been reported elsewhere.14, 15 Briefly, Planet Health was a group randomized controlled field trial of a school-based interdisciplinary intervention that took place over 2 school years. The study was approved by the Committee on Human Subjects at the Harvard School of Public Health.

The present analysis uses data collected from students attending one of the five control schools and includes baseline and follow-up anthropometric data and measures of diet and behaviors including watching TV, playing video and computer games, participating in physical activity and reading/doing homework. Anthropometric data and student surveys were collected at baseline in the fall of 1995 on 780 participants in grades 6 and 7. Follow-up anthropometric and survey data were collected in the spring of 1997 on 655 participants in grades 7 and 8. The majority of loss to follow-up, which totaled 18.0% for girls and 14.0% for boys, was because of school transfer and school absence. Participants with missing data and implausible energy intakes (less than 500 or more than 7000 kcal day−1) were excluded, leaving a final sample of 538 students. Descriptive sample data are presented in Table 1. Although means were similar at baseline and follow-up, there was a good deal of change over time noted. The amount of time spent in a discretionary activity changed by greater than half an hour from baseline to follow-up in 48% of the sample for the physical activity variable and 69% of the sample for the TV viewing variable.

Table 1 Descriptive data of study sample (n=538)

Anthropometric and behavioral assessment

Baseline BMI was calculated from weights and heights measured in the fall of 1995.15 A detailed description of behavioral assessment is described elsewhere.14, 15 Briefly, students completed a food and activity survey in the fall of 1995 and the spring of 1997. The survey included a youth food frequency questionnaire16 that has been validated in ethnically and socioeconomically diverse populations.16, 17 Time spent playing video and computer games and watching TV was measured using an 11-item TV and video measure. In a validation study among Planet Health participants, a deattenuated18 correlation of r=0.54, with equivalent means, was found between TV viewing measured using the television and video measure and a repeat 24-h recall measure.15 Participants completed a 16-item youth activity questionnaire that estimates time spent in moderate or vigorous activities of 3.5 METs.19 These activities included soccer, gymnastics, running/jogging, basketball, dancing and other sports as well as heavy chores; walking was not included because of the poor validity of measurement.15 In a Planet Health validation sample, the questionnaire had a deattenuated18 correlation of r=0.80, with equivalent means, with repeated 24-h physical activity recalls collected 1 month apart.15

Energy expenditure estimation

Estimated resting energy expenditure was calculated by multiplying the age-adjusted metabolic equivalent (MET) value of 1.34 kcal kg−1 h−1 as reported by Harrell et al.20 and the mean weight at follow-up of 57.5 kg. The age-adjusted MET value of 1.34 kcal kg−1 h−1 can be used for all individuals in the sample because it is the estimate for boys aged 13–15 years and girls aged 12–14 years,20 that span the age of the sample at follow-up. The average estimated resting energy expenditure of the sample, 77 kcal h−1, was used in all subsequent calculations. The energy cost of each discretionary activity was calculated as described by Arvidsson et al.21 The average MET value reported for each activity performed (1.05 METs for watching TV, 1.15 METs for reading/doing homework and 1.25 METs for playing video and computer games) was multiplied by the estimated resting energy expenditure.21 Our physical activity measure include those activities with a MET value of 3.5 METs or greater, but the average continuous intensity of the physical activity performed by the adolescents in the sample is not known. Troiano and colleagues22 analyzed accelerometer data of a representative US sample to describe the physical activity levels of the US population and reported that vigorous activity (6.0 METs) accounts for 8–13% of all physical activity (3.0 METs) in youth aged 12–15 years; similar acelerometer results have been described in Planet Health Validation Sample.23 Based on this finding, we assumed that the average continuous intensity of the physical activity in this sample would not exceed 6.0 METs. We estimated the energy cost of physical activity using three assumptions about the intensity of the activity in the sample: 3.5 METs (the minimum captured by the activity variable), 6.0 METs (based on the assumption that the average intensity would not exceed this value) and the midpoint (4.8 METs) between these assumptions about the minimum and maximum intensity level. The difference between energy cost per hour of activity and resting energy expenditure per hour was calculated to represent the additional energy expended when the activity replaces a resting activity.

Statistical analysis

Adjusted associations of change in discretionary activities (TV viewing, video and computer game playing, physical activity and reading/doing homework) with change in energy intake from baseline to follow-up were estimated in what has been called a ‘fixed effects’ regression model. As described by Allison,12 fixed effects methods, when applied to non-experimental research, eliminate a potentially large source of bias by controlling for all stable characteristics of individuals in a study. Following Allison,12 the basic model used in these regressions is:

yit=estimated total energy intake for child i on occasion t, μt=intercept that is allowed to vary with time, β and γ=row vectors of coefficients, xit=column vector of variables that vary both over individuals and over time (such as hours of TV viewing per day), zi=column vector of variables that describe children but do not vary over time (such as gender), αi=all differences between children that are stable over time and not otherwise accounted for by the γzi, ɛit=random disturbance term.

In the case of two time points, difference scores are created by subtracting the first equation from the second, thereby ‘differencing out’ time invariant components γzi and αi and creating the following equation:

Ordinary least squares regression estimates using these difference scores will produce unbiased coefficient estimates for the time-varying variables in the model; other estimation approaches can also be used.12 We used this approach, also applying SUDAAN software (Center for Information Technology, National Institutes of Health, Bethesda, MD, USA) to take clustering of students within schools into account in regression estimates. Predictor variables in the model included baseline BMI, age, four race/ethnicity indicator variables, four indicator variables for school, and change in TV viewing, physical activity (3.5 METs) and reading/doing homework from baseline to follow-up. Continuous change variables were obtained by subtracting the 1995 value from the 1997 value.

We also estimated fixed effects regressions with weight change as the dependent variable, as validation for our estimates of energy imbalance. The assumption is that positive energy imbalance over a substantial period of time can lead to weight gain in excess of normal growth. We also estimated a model for change in BMI.


The adjusted changes in total energy intake associated with a 1 h day−1 increase in discretionary behaviors, independent of the other variables in the model, are shown in Table 2. These estimates indicate that each hour increase in TV viewing was associated with an additional energy intake of 106 kcal day−1 (95% confidence interval (CI): 61–150 kcal day−1). Each hour increase in video and computer game playing was associated with an additional energy intake of 92 kcal day−1 (95% CI: 37–147 kcal day−1). Each hour increase in physical activity was associated with an additional energy intake of 292 kcal day−1 (95% CI: 262–321 kcal day−1). No significant change in energy intake was observed for each hour increase in reading/doing homework (95% CI: −31–61 kcal day−1).

Table 2 Adjusteda change in daily energy intake associated with 1 h day−1 increases in discretionary behaviors

Sex-specific estimates in the adjusted changes in daily energy intake associated with a 1 h day−1 increase in discretionary behaviors are shown in Table 2. Because of substantial confidence intervals, no significant sex differences are observed. However, the difference in the coefficients seen for TV viewing and playing video and computer games suggest that sex differences may exist, although the present analysis is likely underpowered to detect them.

Table 3 includes the estimated energy cost of each activity and estimated energy cost above resting energy expenditure for a participant of mean weight. These values are compared with the changes in energy intake as reported in Table 2 to estimate the energy gap associated with each activity. Reading/doing homework appears to be ‘energy neutral’. That is, there is virtually no change (3 kcal) in energy intake associated with each hour of reading/doing homework. Watching TV, playing video and computer games and doing physical activity (if 3.5 METs is assumed) are all associated with an energy surplus, with estimates of 102, 73 and 99 kcals h−1, respectively. Although physical activity is an energy surplus activity under the 3.5 METs assumption, if the activity had been done at an average intensity of 4.8 METs, it would be ‘energy neutral’ and would be associated with an energy deficit only if the average METs performed exceeded this level. If the average level activity in our sample fell at the cutoff for vigorous activity (6.0 METs), it would be associated with a 93 kcal h−1 deficit. A graphical representation of the energy expended and consumed in each activity is shown in Figure 1, with activities listed in order of increasing gap, from net energy deficit to net energy surplus.

Table 3 Estimated energy cost and energy gap of discretionary activities of youth
Figure 1

Estimated energy gap (kcal h−1). Observed change in energy intake per hour of activity—estimated energy expended above rest.

To validate these estimates (we thank an anonymous reviewer for the suggestion), we estimated fixed effects regressions predicting change in weight over this time period, with the same predictor and control variables. These results indicate that, independent of the other variables in the model, each hour increase in TV viewing was associated with a weight increase of 0.38 kg (95% CI: 0.17–0.59 kg; P=0.02) over this 19-month period. No significant changes were seen for the other variables: video and computer game playing (P=0.43), physical activity (P=0.61) or reading/doing homework (P=0.50). The r2 for this model is 0.13. These results validate evidence for a positive energy gap with the TV change variable, and also the lack of association with the change in physical activity, reading/doing homework and video and computer game variables. A similar model with change in BMI as the dependent variable produced similar results: a significant association with change in TV viewing (0.14 kg/m2 (95% CI: 0.07–0.21; P=0.01).

To understand better what students were changing these behaviors, we estimated regressions using baseline demographic variables (age, sex, ethnicity, controlling for school) to predict change in the discretionary behaviors. Results indicated that one variable predicted increases in TV viewing: Black students increased more than non-Hispanic whites (0.69 h day−1; P=0.005). One year older age predicted fewer hours of moderate and vigorous physical activity (−0.19 h day−1; P=0.0006). Both of these results are consistent with earlier literature. None of the sociodemographic variables predicted change in video and computer game playing or reading/doing homework.


In this analysis, we sought to quantify the total energy intake and the energy gap associated with discretionary activities of adolescents. All activities, with the exception of reading/doing homework were associated with an increase in energy intake. When compared with the energy expended during the activity, watching TV and playing video and computer games resulted in an energy surplus. The effect of physical activity on the energy gap is dependent on assumptions made about the intensity of the activity. However, it is unlikely that the adolescents engaged in continuous activity at an intensity that would result in an energy deficit. If the activity is assumed to be of moderate intensity at 3.5 METs, the activity was associated with a positive energy gap. Reading was found to be ‘energy neutral’.

Our analyses of changes in weight and BMI validate these findings for the most part—increases in TV viewing predict increase in weight and BMI, but the other behaviors do not. These results indicate that these other activities could be energy neutral, with the most consistent evidence for this being the case of reading/doing homework.


There are several inherent limitations of the present analysis. Because observational data are used, causality cannot be established. However, the regression method used controls for all the stable characteristics of the study participants, and thus controls for a wide range of potential confounding variables.12 Measurement of time spent in discretionary activities is subject to recall bias and limitations imposed by the survey instrument. Random measurement error in estimates of these variables would tend to attenuate their association with changes in total energy intake and weight. Random measurement error is intrinsic to the ascertainment of energy intake; however, random error in the dependent variable will not bias regression coefficients24 although it will reduce the precision of estimation (the r2 of the equation).

Energy expenditure is crudely estimated using average resting energy expenditure at the mean BMI of our sample and cannot capture the variability in the energy gap produced by each activity. An additional limitation of the point estimates for physical activity used is that they may fail to capture the sustained increase in energy expenditure that may occur after engaging in physical activity.

Homework, television and video and computer games—intuitive effect on energy balance

The present analysis resulted in some expected findings. As hypothesized, time spent reading and doing homework did not predict increased energy intake and seems to be an ‘energy neutral’ activity. This effect may relate to the fact that this behavior is, to date, unlinked to food advertising. Books do not presently contain advertisements and food products explicitly marketed for consumption whereas reading or doing homework, such as ‘reading drinks’ or ‘study bars,’ are not, to our knowledge, on the market. An unintended consequence of reporting this finding may be that it alerts food manufactures to an untapped market.

As previously observed using data from the present sample,14 increases in TV viewing were associated with increased energy intake, an effect that appears to be mediated by the consumption of foods commonly advertised on TV. (Note: the estimate reported here differs from that reported by Weicha et al.14 where all TV viewing >5 h was assigned a value of 5; we allowed up to 10 h of TV per day.)

This finding supports the relationship between TV viewing and energy intake, which has been observed in a number of studies.25, 26 This relationship may reflect a few different mechanisms. Children snack excessively while watching TV and choose meals which are less healthy than those they might otherwise eat when eating in front of the TV.27 In addition to impacting food choices children make while watching television, TV watching also seems to have a broader impact on food intake, at least in part through food advertisements and cross-promotions of food products and media characters.27 Foods commonly advertised to children are of poor nutritional quality28, 29, 30 and adolescents are exposed to targeted advertising for food products that are within their purchasing power such as fast food, sweets and beverages.31

An increase in time spent playing video and computer games is associated with a significant increase in energy intake that appears to exceed the energy expended during the activity. Although we find no significant association with changes in weight or BMI, the potential energy gap associated with playing video and computer games and the suggestion of sex differences warrants further exploration. Sex differences in video and computer game preferences have been observed32 and it is plausible that the type of video and computer games played by girls and boys may impact energy intake differently. For example, perhaps the pace of Educational video and computer games, which are more likely to be preferred by girls, allow for more opportunities for snacking than would games designated as Human Violence or Sports Violence, typically preferred by boys.32 The relationship of video and computer game playing and energy balance is likely to change over time given the advent of new generation active video and computer games and the increase of online gaming which may be laden with targeted advertising. Thus, the associations observed here may not reflect the contemporary relationship between energy intake and playing video and computer games.

Physical activity—counterintuitive effect on energy gap?

Perhaps the findings related to reported changes in physical activity are most surprising. Because of the energy expended during physical activity, there is a widespread assumption that increasing activity will result in a net reduction in any energy gap among youth in general, and will decrease the energy gap in overweight children and adolescents. However, if this activity results in an increase in energy intake that is equal to or in excess of the energy expended during the activity, then adding physical activity could fail as a strategy to reduce excess weight. A comparison of estimated energy expenditure with energy intake findings from the present analysis indicates that each hour of resting activity that is replaced with physical activity (3.5 METs assumed) yields a positive energy gap of approximately 100 kcals h−1. Although substantial inter-individual variability would be expected because of weight, height, sex, body composition and type of physical activity done, these rough estimates suggest overcompensation for physical activity among the study population as a whole if the activity is assumed to be done at 3.5 METs. Simply put, it appears that when youth increased their activity, they also increased their food intake by an amount that exceeded that expended during the activity. If the activity was done at a level of 4.8 METs, the activity would be considered ‘energy neutral’. Physical activity would only be associated with a net energy deficit if the average activity intensity had exceeded this level.

Our analysis of weight change and BMI however found no association with the change in these variables—indicating that the physical activity variable may be energy neutral. The finding that increasing physical activity could be energy neutral, although counterintuitive, does support the relationship observed between weight status and physical activity that has been observed to date. Although weak inverse associations have been observed between physical activity and BMI among children and adolescents,33, 34, 35 the majority of interventions aimed at increasing physical activity to prevent obesity in childhood have not shown favorable results.36 A recent systematic review of efficacy of exercise for the treatment of overweight in children and adolescents suggests that a moderate to high intensity physical activity prescription is effective in reducing body fat, but not BMI.37

It remains to be established why a behavior that is a key determinant of energy needs is not more closely linked with reduced excess weight in children. The lack of empirical evidence for such a benefit of physical activity may be the result of body composition changes,37 poor compliance,38 inadequate exercise prescription37, 38 or dietary changes.37 The issue of concomitant changes in energy intake with increases in physical activity poses a unique concern in that these null findings may imply complete compensation by energy intake in response to the added activity. If this mechanism explains the lack of a relationship observed to date, increasing exercise prescriptions or ensuring improved compliance would continue to be an ineffective excess weight gain prevention strategy if dietary intake increases to offset the energy imbalance.


Certainly, ongoing exploration of these relationships is warranted. Lessons learned from the relationship between TV viewing and energy intake may be used to begin generating hypotheses about the association of physical activity and dietary intake. As TV exposure is known to be associated with overall food eaten as well as consumption while watching TV, perhaps some of the effect of physical activity operates by altering the food a child eats in preparation for, during or in response to physical activity. A recently published content analysis of food-related advertisements aimed at children by Connor39 examined the appeals most commonly used in these advertisements. Connor39 reported that action appeal was used in 57% of the ads, the second most common appeal after the fun appeal. She describes the action appeal as one that associates the product with excitement or active lifestyle.39 One can easily evoke images of the action appeal: a child drinking a sports drink on a soccer field, a group of young athlete's sharing a pizza after a game, or a child playing in a ‘fun house’ at a fast food restaurant before being served a hamburger, French fries and soda for lunch.

Advertising messages that create associations between particular events or activities and a food product may pose a previously unexplored hurdle to achieving energy balance and may render other approaches for reducing the obesity epidemic ineffective. As advertising to children becomes increasingly pervasive in video and computer games and computer gaming and through product placement and product cross-promotion, so could its negative effects. Wide variation in number of food advertisements shown on children's television exists across countries, for example from 7 and 77 ads per 20 h in Sweden and the Netherlands to 215 and 231 ads per 20 h in the United States and Australia.40 As such, the consequences of this mechanism may disproportionately affect countries that do not restrict advertising to children.

If food intake becomes so closely entwined with physical activity that every Little League baseball game is followed by a requisite fast food stop, every bike ride requires a sports drink and opportunities for ingesting calories are incorporated into practices and games, then increasing physical activity could be an unsuccessful approach for reducing positive energy balance. If advertising messages have the power to influence energy intake in response to physical activity, then the effects of physical activity may represent a much more complex relationship between changes in energy expenditure and changes in energy intake.


  1. 1

    Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM . Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006; 295: 1549–1555.

  2. 2

    Kosti RI, Panagiotakos DB . The epidemic of obesity in children and adolescents in the world. Cent Eur J Public Health 2006; 14: 151–159.

  3. 3

    Wang Y, Lobstein T . Worldwide trends in childhood overweight and obesity. Int J Pediatr Obes 2006; 1: 11–25.

  4. 4

    Wang YC, Gortmaker SL, Sobol AM, Kuntz KM . Estimating the energy gap among US children: a counterfactual approach. Pediatrics 2006; 118: e1721–e1733.

  5. 5

    Epstein LH, Wing RR, Koeske R, Valoski A . Effects of diet plus exercise on weight change in parents and children. J Consul Clin Psychol 1984; 52: 429–437.

  6. 6

    Ebbeling CB, Feldman HA, Osganian SK, Chomitz VR, Ellenbogen SJ, Ludwig DS . Effects of decreasing sugar-sweetened beverage consumption on body weight in adolescents: A randomized, controlled pilot study. Pediatrics 2006; 117: 673–680.

  7. 7

    Ebbeling CB, Leidig MM, Sinclair KB, Hangen JP, Ludwig DS . A reduced-glycemic load diet in the treatment of adolescent obesity. Arch Pediatr Adolesc Med 2003; 157: 773–779.

  8. 8

    Robinson TN . Reducing children's television viewing to prevent obesity: A Randomized Controlled Trial. JAMA 1999; 282: 1561–1567.

  9. 9

    Prentice A, Jebb S . Energy intake/physical activity interactions in the homeostasis of body weight regulation. Nutr Rev 2004; 62 (7 part 2): S98.

  10. 10

    Willet W . Nutritional Epidemiology, 2nd edn. Oxford University Press: New York, 1998.

  11. 11

    Dietz Jr WH, Gortmaker SL . Do we fatten our children at the television set? Obesity and television viewing in children and adolescents. Pediatrics 1985; 75: 807.

  12. 12

    Allison P . Fixed Effects Regression Methods for Longitudinal Data Using SAS. SAS Institute Inc.: Cary, NC, 2005.

  13. 13

    Shadish W, Cook T, Campbell D . Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Houghton Mifflin Company; Boston, MA, 2002.

  14. 14

    Wiecha JL, Peterson KE, Ludwig DS, Kim J, Sobol A, Gortmaker SL . When children eat what they watch: impact of television viewing on dietary intake in youth. Arch Pediatr Adolesc Med 2006; 160: 436–442.

  15. 15

    Gortmaker SL, Peterson K, Wiecha J, Sobol AM, Dixit S, Fox MK et al. Reducing obesity via a school-based interdisciplinary intervention among youth: planet health. Arch Pediatr Adolesc Med 1999; 153: 409–418.

  16. 16

    Rockett HRH, Wolf AM, Colditz GA . Development and reproducibility of a food frequency questionnaire to assess diets of older children and adolescents. J Am Diet Assoc 1995; 95: 336–340.

  17. 17

    Field A, Peterson K, Gortmaker S, Cheung L, Rockett H, Fox M et al. Reproducibility and validity of a food frequency questionnaire among fourth to seventh grade inner-city school children: implications of age and day-to-day variation in dietary intake. Public Health Nutr 1999; 2: 293–300.

  18. 18

    Rosner B, Willett WC . Interval estimates for correlation coefficients corrected for within-person variation: implications for study design and hypothesis testing. Am J Epidemiol 1988; 127: 377–386.

  19. 19

    Ainsworth BE, Haskell WL, Leon AS, Jacobs DR, Montoye HJ, Sallis JF et al. Compendium of physical activities—classification of energy costs of human physical activities. Med Sci Sports Exerc 1993; 25: 71–80.

  20. 20

    Harrell J, McMurray R, Baggett C, Pennell M, Pearce P, Bangdiwala S . Energy costs of physical activities in children and adolescents. Med Sci Sports Exerc 2005; 37: 329–336.

  21. 21

    Arvidsson D, Slinde F, Larsson S, Hulthen L . Energy cost of physical activities in children: validation of sensewear armband. Med Sci Sports Exerc 2007; 39: 2076–2084.

  22. 22

    Troiano R, Berrigan D, Dodd K, Mâsse L, Tilert T, McDowell M . Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008; 40: 181–188.

  23. 23

    Cradock A, Wiecha J, Peterson KE, Sobol AM, Colditz GA, Gortmaker SL . Do missing data bias activity estimates? Youth recall and TriTrac accelerometer estimates of physical activity levels. Med Sci Sports Exerc 2004; 36: 525–532.

  24. 24

    Fox J . Applied Regression Analysis, Linear Models and Related Methods. CA Sage Publications: Thousand Oaks, 1997.

  25. 25

    Epstein LH, Paluch RA, Consalvi A, Riordan K, Scholl T . Effects of manipulating sedentary behavior on physical activity and food intake. J Pediatr 2002; 140: 334–339.

  26. 26

    Matheson DM, Killen JD, Wang Y, Varady A, Robinson TN . Children's food consumption during television viewing. Am J Clin Nutr 2004; 79: 1088–1094.

  27. 27

    The Role of Media in Childhood Obesity. The Henry J. Kaiser Family Foundation: Menlo Park, 2004;

  28. 28

    Harrison K, Marske AL . Nutritional content of foods advertised during the television programs children watch most. Am J Public Health 2005; 95: 1568–1574.

  29. 29

    Powell LM, Szczypka G, Chaloupka FJ, Braunschweig CL . Nutritional content of television food advertisements seen by children and adolescents in the United States. Pediatrics 2007; 120: 576–583.

  30. 30

    Powell LM, Szczypka G, Chaloupka FJ . Exposure to food advertising on television among US children. Arch Pediatr Adolesc Med 2007; 161: 553–560.

  31. 31

    Powell LM, Szczypka G, Chaloupka FJ . Adolescent exposure to food advertising on television. Am J Prev Med 2007; 33 (4, Suppl 1): S251.

  32. 32

    Buchman DD, Funk JB . Video and computer games in the 90's: Children's time commitment & game preference. Child Today 1996; 24: 12.

  33. 33

    Berkey CS, Rockett HRH, Field AE, Gillman MW, Frazier AL, Camargo CA et al. Activity, dietary intake, and weight changes in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics 2000; 105: e56.

  34. 34

    Gordon-Larsen P, Adair LS, Popkin BM . Ethnic differences in physical activity and inactivity patterns and overweight status. Obes Res 2002; 10: 141–149.

  35. 35

    Kimm SYS, Glynn NW, Obarzanek E, Kriska AM, Daniels SR, Barton BA et al. Relation between the changes in physical activity and body-mass index during adolescence: a multicentre longitudinal study. Lancet 2005; 366: 301–307.

  36. 36

    Campbell K, Waters E, O'Meara S, Kelly S, Summerbell C . Interventions for preventing obesity in children. Cochrane Database Syst Rev 2002;(2): CD001871. Review. Update in Cochrane Database Syst Rev2005;(3): CD001871. PMID: 12076426.

  37. 37

    Atlantis E, Barnes EH, Singh MAF . Efficacy of exercise for treating overweight in children and adolescents: a systematic review. Int J Obes 2006; 30: 1027.

  38. 38

    Steinbeck KS . The importance of physical activity in the prevention of overweight and obesity in childhood: a review and an opinion. Obes Rev 2001; 2: 117.

  39. 39

    Connor SM . Food-related advertising on preschool television: building brand recognition in young viewers. Pediatrics 2006; 118: 1478–1485.

  40. 40

    Lobstein T, Dibb S . Evidence of a possible link between obesogenic food advertising and child overweight. Obes Rev 2005; 6: 203–208.

Download references


This study was supported by grant 57891 from the Robert Wood Johnson Foundation, HD-30780 from the National Institute of Child Health and Human Development, Rockville, MD, and by Cooperative Agreement U48/DP00064-00S1 from the Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention nor other funders.

Author information

Correspondence to S L Gortmaker.

Additional information

Conflict of interest

Kendrin Sonneville has received a training grant from NIDDK in Academic Nutrition. Steven Gortmaker has declared no financial interests.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Sonneville, K., Gortmaker, S. Total energy intake, adolescent discretionary behaviors and the energy gap. Int J Obes 32, S19–S27 (2008).

Download citation


  • TV
  • physical activity
  • total energy intake
  • reading
  • energy gap

Further reading