PAPER Longitudinal study of skipping breakfast and weight ... - Nature

International Journal of Obesity (2003) 27, 1258–1266 & 2003 Nature Publishing Group All rights reserved 0307-0565/03 $25.00 www.nature.com/ijo

PAPER Longitudinal study of skipping breakfast and weight change in adolescents CS Berkey1*, HRH Rockett1, MW Gillman2,3, AE Field1 and GA Colditz1,4 1

Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA; Department of Nutrition, Harvard School of Public Health, Boston, MA, USA; 3Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, Boston, MA, USA; and 4Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA 2

BACKGROUND: Several cross-sectional studies reported that heavier children eat breakfast less often. However, no longitudinal studies have addressed whether skipping breakfast leads to excessive weight gain. OBJECTIVE: To investigate whether skipping breakfast was prospectively associated with changes in body fatness. METHODS: A cohort of 414 000 boys and girls from all over the US, 9- to 14-y-old in 1996, returned annual mailed questionnaires (1996–1999) for the Growing Up Today Study. We analyzed change in body mass index (BMI; kg/m2) over three 1-y periods among children who reported breakfast frequency. RESULTS: Children who reported that they never eat breakfast had lower energy intakes than those who eat breakfast nearly every day. Children who were more physically active reported higher energy intakes, as did those who reported more time watching television/videos and playing videogames. Like previous studies, skipping breakfast was associated with overweight, cross-sectionally. However, overweight children who never ate breakfast lost BMI over the following year compared to overweight children who ate breakfast nearly every day (boys: 0.66 kg/m2 (s.e. ¼ 0.22); girls: 0.50 kg/m2 (s.e. ¼ 0.14)). But normal weight children who never ate breakfast gained weight relative to peers who ate breakfast nearly every day (boys: +0.21 kg/m2 (s.e. ¼ 0.13); girls: +0.08 kg/m2 (s.e. ¼ 0.05)). Breakfast frequency was positively correlated with self-reported quality of schoolwork. CONCLUSIONS: Overweight children who never eat breakfast may lose body fat, but normal weight children do not. Since numerous studies link skipping breakfast to poorer academics, children should be encouraged to eat breakfast. International Journal of Obesity (2003) 27, 1258–1266. doi:10.1038/sj.ijo.0802402 Keywords: breakfast; BMI; weight change; overweight; adolescent; academic; longitudinal

Introduction Numerous studies have documented the increase over recent decades in the prevalence of childhood obesity1–8 and the associated health and social consequences.3,6–25 This rapid increase implicates environmental factors.26–28 During these decades of rising obesity prevalence, physical activity among adolescents has declined, while time spent in sedentary activities such as watching television and playing computer games has increased.5,6 Furthermore, in nationally representative samples of US adolescents, breakfast consumption

*Correspondence: Dr CS Berkey, Channing Laboratory, 181 Longwood Avenue, Boston, MA 02115, USA. E-mail: [email protected] Received 17 March 2003; accepted 10 May 2003

declined from 1965 to 1991,29 a concern since eating breakfast is associated with higher overall diet quality.30–32 Cross-sectional studies have consistently reported positive associations between measures of adiposity in children and skipping breakfast.33–37 A recent analysis of US adults who maintained substantial weight losses for at least 1 y suggested that eating breakfast may be an important factor in weightloss maintenance.38 However, a randomized clinical trial of obese women found that those who stopped eating breakfast lost the most weight, although those who ate breakfast had lower dietary fat and impulsive snack intakes.39 There are no longitudinal studies of adolescents that evaluate whether, as implied by cross-sectional studies, skipping breakfast may result in excessive body weight increases over time. Using data from the Growing Up Today Study (GUTS), we prospectively analyzed the relationship between breakfast

Skipping breakfast and adolescent weight change CS Berkey et al

1259 frequency and subsequent changes in body mass index (BMI).

Methods Subjects Established in the fall of 1996, the ongoing Growing Up Today Study consists of 16 882 children, residing in 50 states, who are offspring of Nurses’ Health Study II participants.40 The Study was designed to assess prospectively the determinants of adolescent weight changes, including dietary factors, physical activity and inactivity. Human Subjects Committees at the Harvard School of Public Health and Brigham and Women’s Hospital approved the Study. In 1996, letters were sent to 34 174 mothers, explaining the goals of the new study and requesting consent for their children to participate; 18 526 mothers returned the consent form, providing the name, age, gender, and mailing address of 26 765 children. Introductory letters and gender-specific questionnaires were then mailed directly to children whose mothers granted consent. These letters assured potential participants that the information they provided would remain confidential. The baseline (1996) sample included 9039 girls (68% response rate) and 7843 boys (58% response rate) who returned completed questionnaires, thereby assenting to participate. Some of these respondents were outside the 9- to 14-y age range at the time of questionnaire completion, leaving us with 8980 girls and 7791 boys at baseline. In 1997, 1998, and 1999, we sent subjects follow-up questionnaires to update all information. Response rates to at least one of these follow-up surveys were 94.1% for girls and 89.5% for boys. The present longitudinal analyses were restricted to 8128 girls and 6458 boys who were of age 9–17 y upon returning two or more consecutive annual surveys, or 1996 and 1999 surveys.

Body mass index Children self-reported their height and weight annually on our questionnaire which provided specific measuring instructions but suggested they ask someone to help; since their mothers are nurses, assistance is available. A previous study reported high validity for self-reported heights and weights in adolescents.41 We assessed adiposity by computing BMI ( ¼ weight/height2 (kg/m2)). The International Obesity Task Force supports the use of BMI to assess fatness in children and adolescents.42 Roche et al43 demonstrated that childhood BMI is related to other measures of adiposity that were not feasible to collect on our cohort. A recent study44 supported the validity of BMI computed from selfreported height and weight, with a correlation of 0.92 between BMI computed from youth (grades 7–12) selfreports and measured values. We estimated our primary outcome, 1 y change in adiposity, by BMI1997BMI1996, BMI1998BMI1997, and BMI1999BMI1998, dividing each by the exact time interval

between the pair of measurements. We also estimated 3-y change in adiposity, BMI1999BMI1996, divided by the exact time interval. Before computing BMIs, we excluded any height that was more than 3 standard deviations (s.d.) beyond the gender–age-specific mean height (0.46% of heights excluded), and any 1 y height change that declined by more than 1 in or increased by 43 s.d. above the mean change (1.94% excluded). We further excluded any BMI o12.0 kg/m2 as a biological lower limit (clinical opinion), and any BMI 43 s.d. above or below the gender- and agespecific mean (in log (BMI) scale) (0.89% excluded). As all BMI changes computed from the remaining data represented realistic weight changes, there were no further exclusions: 7545 girls and 5962 boys provided annual BMI-change data. We grouped together children, using their BMI at the earlier year of each 1-y interval, who were above the Center for Disease Prevention and Health Promotion (CDC) genderand age-specific (to the month) 85th percentile for BMI; children above the 85% were at risk of overweight (85–95%) or were overweight (495th percentile).45 Similarly, we grouped together those below the 10th percentile. We refer to children whose BMI exceeded the 85th as ‘overweight’, those below the 10th as ‘lean’, and those between the 10th and 85th as ‘normal weight’. The CDC standards also provided BMI Z-scores.

Breakfast frequency In 1996, 1997, and 1998, we asked ‘How many times each week (including weekdays and weekends) do you eat breakfast?’ Response categories were never (or almost never), 1–2 times per week, 3–4 times per week, and five or more times per week.

Physical activity We developed a physical activity questionnaire for youth that asked the participants to recall the typical number of hours per week, within each season over the past year, during which they participated in 17 activities and team sports, outside of gym class. Response categories ranged from 0 to 10+ h/week. We computed each child’s typical hours per day of physical activity over the entire year. Assessments of an earlier nonseasonal version of this instrument found that estimates of total activity were moderately reproducible and reasonably correlated with cardio-respiratory fitness,46 and another validation study reported a correlation of r ¼ 0.80 between self-reports and 24-h recalls.47 We developed the seasonal version used in this paper to further improve reliability and validity.48 Estimates of total physical activity that exceeded 40 h/ week were deemed implausible and excluded (3.60%) from our longitudinal analyses. International Journal of Obesity


1260 Inactivity Another series of questions were designed to measure weekly hours of recreational inactivity: ‘watching TV’, ‘watching videos or VCR’, and ‘Nintendo/Sega/computer games (not homework)’. For each of these three items, children reported their usual number of h/week, separate for weekdays and for weekends, from options ranging from zero to 31+ h. From this information, we computed each child’s hours of recreational inactivity per day. Moderate validity has been reported for recalled total inactivity from a similar instrument.47 We excluded totals exceeding 80 h/week (0.89%).

Energy intake Members of our group designed a self-administered semiquantitative food frequency questionnaire (FFQ) for older children and adolescents which is inexpensive and easy to administer to large populations.49 This FFQ for youth has been shown to be valid50 and reproducible,49 comparable in performance to a similar adult FFQ. It included questions regarding frequency of intake of 132 specific food items over the past year. Using nutrient composition databases, food portion sizes specific to this age range, and each child’s reported intake of each food and fat level of relevant food items (such as milk), we estimated total energy intake (kcal/ day). We excluded intakes o500 or 45000 kcal/day (0.53% excluded). Some of our models adjusted for total energy intake and cereal intake (hot and cold), reported at the same time as breakfast frequency, since cereal (per calorie and per cost) may provide a healthy breakfast.51

Race At baseline, children reported their race/ethnic group by marking all that apply from six options. We assigned each child to one of five race/ethnic groups following US Census definitions, except that we retained Asians as a separate group rather than pooled with ‘Other’.1

Tanner stage, menarche, age Each year children reported their Tanner maturation stage, a validated self-rating52 of sexual maturity with five stages of pubic hair development, and girls reported whether/when their menstrual periods began. We derived a menstrual history variable having three categories: premenarche before and after the 1-y BMI change, periods began during the interval, and postmenarche both years. We computed each child’s age from birthdate and date of questionnaire return.

Schoolwork On the 1996–1999 questionnaires, we included questions adapted from the Harter Self-Perception Profile for Children.53 We asked the participants how much this statement describes them ‘Some kids feel like they are very good at their International Journal of Obesity

school work’, to which they responded ‘Really true for me’, ‘Sort of true for me’, or ‘Not true for me’. About 55% responded that it was ‘Really true’ for them. As very few (4%) said it was ‘Not true’, we combined the two categories ‘Sort of true’ and ‘Not true’.

Statistical analysis To assess the potential for bias, we compared the baseline (1996) values of age, BMI, breakfast frequency, and energy intake of those children who returned surveys in consecutive follow-up years with those who did not. The differences were small (see Results). The cross-sectional association between baseline breakfast frequency and BMI was summarized by tabulating the percent of children in each breakfast frequency group who were lean, normal, or overweight. We fitted a regression model of baseline energy intake on breakfast frequency and body weight, with age, Tanner stage, race, menarche, physical activity and inactivity included. To assess prospectively the effect of breakfast frequency at time t on change in BMI from time t to t+1 years, we used mixed linear regression models,54 estimated by SAS proc mixed,55 with annual change in BMI as the continuous outcome variable. We used mixed models because each child had one, two, or three BMI changes; the assumption of independent observations required by ordinary linear regression models was thus not satisfied. To account for changes in BMI that typically occur during growth and maturation, all models included height growth from year t to t+1, menstrual history for years t and t+1, Tanner stage, age, and BMI Z-score at year t.26,56–59 We used continuous exact age with linear and quadratic terms. All models included race, as well as activity and inactivity during the year of BMI change.60 Some models further included energy intake and (hot and cold) cereal intake. All models obtained separate estimates of breakfast effects for lean, normal, and overweight children. The effects of change in breakfast frequency were estimated from a model with an interaction between breakfast frequency at time t and time t+1. A separate analysis related 3-y (1996–1999) change in BMI to mean (3-y) breakfast frequency, adjusting for 3-y height growth, means and changes in activity, in inactivity, and in energy intake. Logistic models, estimated using generalized estimating equations61 and SAS proc genmod55 to account for up to three observations per child, related academic performance (very good at schoolwork, or not) at year t+1 to breakfast frequency at years t and t+1, adjusting for race, age, Tanner stage, and academic performance at year t.

Results The participants were mostly white children (94.7% white, not Hispanic), with only 0.9% black children (not Hispanic), 1.5% Hispanic, 1.5% Asian, and 1.4% other (including


1261 Native American); this distribution reflects the small minority representation of their mothers who are participants in the Nurses’ Health Study II.40 Children not included in our longitudinal analyses were slightly older (girls by 0.3 y; boys by 0.46 y), and at baseline ate breakfast slightly less often (girls by 0.39 days/week; boys by 0.18 days/week) (age-adjusted Po0.05), but they were similar in age-adjusted BMI and total energy intake. At baseline, 23.2% of the boys and 17.4% of the girls were overweight (485th percentile BMI), while 7.2% of the boys and 8.6% of the girls were very lean (o10th percentile). Children who never ate breakfast were heavier (26.4% of boy never-eaters were overweight and 25.3% of girl never-eaters were overweight) than those who ate breakfast nearly every day (21.2% of boys and 15.8% of girls were overweight) (Table 1). As suggested by the mean energy intakes in Table 1, normal weight and overweight children who never ate breakfast reported lower (by 1.3 MJ) energy intakes than those (of similar weight) who ate breakfast daily (Po0.0001, adjusted for age, Tanner stage, race, menarche, physical activity and inactivity). More physically active children also reported higher energy intakes: 0.49 (s.e. ¼ 0.03) MJ/day for boys, 0.32 (s.e. ¼ 0.02) MJ/day for girls per hour daily activity. Contrary to our expectations, more inactivity (TV/videos/ videogames) time was also associated with higher energy intakes: 0.08 (0.02) MJ/day for boys and 0.12 (0.01) MJ/day for girls, per hour inactivity. Our prospective model, which related BMI changes over 1 y to the breakfast frequency at the start of each year, provided contrasting findings for children who were overweight vs those who were normal weight (Table 2). For

Table 1

overweight adolescents, those who never ate breakfast had smaller BMI increases over the subsequent year relative to those who ate breakfast daily (0.66 kg/m2 for boys and 0.50 kg/m2 for girls). The linear trend for breakfast frequency was significant in overweight girls (P ¼ 0.049). For normal weight boys and girls, there was a suggestion, although not statistically significant, that those who never ate breakfast gained weight relative to daily eaters. When we further adjusted for daily cereal (hot and cold) intake and total energy intake, the estimated effects for breakfast were hardly changed (Table 2). However, we do not know when the cereal was consumed, and some children who never ate breakfast ate as many as four bowls daily. Our analysis relating change in breakfast frequency to BMI change did not include BMI changes from 1998 to 1999, because the breakfast question was dropped from our survey in 1999. Normal weight boys who never ate breakfast, either before or after a change in BMI, had significantly larger (by +0.62 kg/m2) BMI gains than those who ate breakfast daily both years, but there were no consistent patterns for normal weight children (Table 3). Overweight boys and girls who never ate breakfast (before or after) had significantly smaller BMI increases (boys 1.12 kg/m2; girls 0.67 kg/m2) relative to overweight boys and girls who ate breakfast daily both years. Our analysis of 3-y BMI change (from 1996 to 1999) suggested that normal weight girls (1996) who ate breakfast 1–2 days/week (on average during follow-up) gained more weight (+0.072 (0.037) kg/m2) than peers who ate daily. However, overweight boys and girls who skipped breakfast gained less weight than daily eaters (boys never-eaters:

Cross-sectional associations between baseline breakfast frequency, body weight status (for age and gender),a and energy intakeb in children 9–14 y of age Boys

Girls

Lean

Normal

Overweight

Lean

Normal

Overweight

Percent

7.2

69.6

23.2

8.6

74.0

17.4

Breakfast frequency Never (193 boys, 388 girls) Percent Energy intake

4.7 8.88

68.9 9.05

26.4 7.40

5.9 7.45

68.8 7.37

25.3 7.27

1–2 days/week (510 boys, 774 girls) Percent Energy intake

5.7 10.06

61.8 9.00

32.6 8.75

7.6 8.32

68.2 8.35

24.2 7.82


4.7 10.22

64.1 9.75

31.3 9.18

6.8 8.88

73.9 8.51

19.3 7.72


7.8 9.66

71.1 9.79

21.2 9.32

9.2 8.93

75.0 8.81

15.8 8.44

a

Lean are o10th percentile BMI,45 normal are between 10th and 85th percentiles, and overweight are >85th percentile. Mean daily energy intakes (MJ/day).

b

International Journal of Obesity


1262 Table 2 Prospective effects of breakfast frequency on 1-y change in BMI (kg/m2) relative to those of similar body weight who ate breakfast 5+ days/week (data were from 5411 boys and 7112 girls, age 9–17 y) Boys

Girls

Normal weight b7s.e. (P-value)

Overweight b7s.e. (P-value)

Normal weight b7s.e. (P-value)

Overweight b7s.e. (P-value)

Breakfast frequency Never Adjusteda Fully adjustedb

+0.2170.13 (0.11) +0.2270.14 (0.11)

0.6670.22 (0.01) 0.7070.23 (0.01)

+0.0870.05 (0.15) +0.1070.06 (0.09)

0.5070.14 (0.01) 0.4770.15 (0.01)

1–2 days/week Adjusteda Fully adjustedb

+0.0270.06 (0.74) +0.0370.06 (0.62)

+0.0170.14 (0.93) 0.0370.15 (0.83)

+0.0170.04 (0.76) +0.0270.04 (0.62)

0.3370.14 (0.02) 0.3170.14 (0.03)

3–4 days/week Adjusteda Fully adjustedb

+0.0470.05 (0.37) +0.0470.05 (0.42)

0.0870.11 (0.46) 0.1070.11 (0.40)

+0.0270.03 (0.64) +0.0270.03 (0.47)

0.1670.11 (0.14) 0.1670.11 (0.16)

0

0

0

0

5–7 days/week (referent) a

Adjusted for age, race, menarche (girls), Tanner stage, prior BMI Z-score, physical activity, inactivity, and height growth during the year. Further adjusted for cereal (hot and cold) and energy intake.

b

Table 3

Annual BMI changes (b7s.e.; kg/m2) relative to peers who ate breakfast daily both before and after the BMI changea Normal weight boys After

Breakfast frequency Before Never 1–2 3–4 5+/week

Never

1 or 2

3 or 4

5+/week

Never

1 or 2

3 or 4

5+/week

+0.6270.25 0.227.19 +0.1270.21 +0.1770.24

+0.7870.46 +0.0770.12 +0.1970.14 +0.1670.11

0.6270.25 +0.1770.14 +0.0970.09 +0.0670.08

0.1970.23 +0.0570.12 0.0670.09 0.0 (ref)

0.1670.12 0.0970.11 +0.0670.13 +0.0470.13

+0.1070.14 +0.0970.08 0.0470.09 0.0470.08

+0.3470.14 0.1170.10 +0.0570.07 0.0070.05

0.0570.15 0.0370.11 +0.0170.07 0.0 (ref)

Overweight boys

Overweight girls

After

After

Breakfast frequency Before Never 1–2 3–4 5+/week

Normal weight girls After

Never

1 or 2

3 or 4

5+/week

Never

1 or 2

3 or 4

5+/week

1.1270.38 0.6270.34 1.3070.56 0.0070.57

0.9970.36 +0.0270.30 0.0570.36 0.3570.29

0.5170.70 +0.5870.31 0.1670.19 0.0970.20

+0.3670.29 +0.2670.34 0.0370.19 0.0 (ref)

0.6770.26 1.2970.33 0.7070.42 0.5870.28

0.2770.26 +0.5470.26 0.1570.25 0.4270.26

0.2770.37 0.7170.38 0.3170.20 0.0770.18

0.6070.37 0.1470.32 0.2170.23 0.0 (ref)

Bold b7s.e. indicates Po0.05. As we do not have 1999 breakfast data, BMI changes from 1998 to 1999 were not included; adjusted for age, Tanner stage, race, girls’ menarche status, prior BMI Z-score, height growth, activity, and inactivity.

a

0.425 (0.234), 3–4 days/week: 0.139 (0.071); girls who ate 1–2 days/week: 0.114 (0.067) and 3–4 days/week: 0.177 (0.056)). For these analyses, we defined overweight as any BMI above the 85th percentile,45 the widely accepted cutpoint. We were concerned that for children below the 85th percentile, the effect of skipping breakfast on weight change might be more similar to overweight than to normal weight International Journal of Obesity

children, for whom we observed opposite effects. So we estimated the effects of skipping breakfast separately for smaller groups of children (75–80%/80–85%/85–90%, etc), which confirmed that the 85th percentile was an appropriate cutpoint for this work. The prospective effect of skipping breakfast on selfreported academic performance 1 y later is shown in Figure 1. Boys who never ate breakfast were less likely


1263 1.2 Boys Girls

1

RR

0.8

0.6

0.4

0.2

0

Never

1 or 2

3 or 4

5 to 7

Breakfast: Days per Week

Figure 1 Relative risk for doing very well at schoolwork at year t+1, for children who skipped breakfast at year t compared to those who ate breakfast daily at year t. The gender-specific logistic regression models adjusted for schoolwork at t, age, Tanner stage, race, and menarche status (girls). Data were from 5830 boys and 7501 girls.

(RR ¼ 0.68; 95% CI: 0.53–0.89) to report a year later that they did very well at their schoolwork, and similarly for girls (RR ¼ 0.73; 95% CI: 0.62–0.86). Since considerable research demonstrates that the beneficial effects of eating breakfast are quite immediate,62–65 we then further included in the model breakfast reported the same time as schoolwork. The previous-year breakfast was no longer important and, instead, current-year breakfast was significantly beneficial (RR ¼ 0.66 (CI: 0.48–0.91), trend P ¼ 0.0004 for boys; RR ¼ 0.55 (CI: 0.44–0.67), trend P ¼ 0.0001 for girls).

Discussion Earlier cross-sectional studies reported that heavier children eat breakfast less frequently than do leaner children, and when they do eat breakfast, heavier children, compared to leaner children, have lower energy intakes and fewer nutrients supplied by it.33–37 Children who skipped breakfast had higher daily percentage of energy from fat and lower intakes of energy, protein, vitamins, and minerals,32 consumed snacks that were higher in fat, and had higher plasma total cholesterol levels.31 Also, adolescents who ate less cereal and milk had higher percentage of body fat.66 Thus, it has been widely assumed that skipping breakfast may lead to excessive weight gains. A cross-sectional analysis of our data similarly found that children who skipped breakfast were heavier. However, our longitudinal analyses of overweight adolescents found that skipping breakfast was instead associated with relative BMI declines over the following year. In contrast, among normal weight adolescents there was a tendency, although not statistically significant, to gain weight after skipping breakfast. Including total energy and cereal (hot and cold) in our models produced little change in estimated effects of breakfast on weight change. This may

suggest some alternate pathway between breakfast and weight change, rather than through total energy or healthy breakfast foods, or may simply reflect the difficulty of reporting dietary intakes compared to reporting breakfast frequency. It is not apparent why normal weight and overweight children might respond differently to skipping breakfast. Differences in resting energy expenditure (REE) in overweight and normal weight children may be involved, and some adult studies suggest that REE declines in individuals as they lose weight. We are not aware of other longitudinal studies of breakfast and weight change in children. A randomized clinical trial of 52 obese women reported that women who initially ate breakfast lost more weight by subsequently skipping breakfast rather than continuing to eat breakfast, but the opposite was found among those who initially skipped breakfast.39 Our children who reported never eating breakfast also reported lower daily energy intakes. Higher energy intakes were related to more time in physical activity, and also to more inactivity time such as TV viewing. Thus, the role of inactivity in promoting obesity appears to involve increased energy intake as well as reduced energy expenditure.67,68 A recent study found that computer use may be displacing leisure-time physical activity in students.69 Our children who skipped breakfast were also less likely to report that they were doing very well in school, either at the same time or a year later. This is consistent with a large educational literature (including randomized controlled trials and assessments of the School Breakfast Program) on the benefits of breakfast, which include increases in standardized test scores and reductions in psychosocial problems, absenteeism, and tardiness.62–65,70–73 Among US adults, nearly as many reported eating high-fat desserts (15.1%) and eggs (15.3%) for breakfast as hot or cold cereals (21.8%).74 We did attempt to perform an in-depth analysis of specific breakfast foods, but because some children who never ate breakfast reported eating over four servings per day of traditional breakfast foods (such as cereal, bagels, English muffins), and some who ate breakfast daily reported none, it was not feasible to address the quality of the foods consumed at breakfast. A major strength of our analysis was the longitudinal design, which allowed us to study changes over time in breakfast frequency and in BMI, while accounting for growth and maturation. BMI typically goes up from year to year among children in this age range, and we took these changes into account. Although our observational study cannot prove causation, evidence that links breakfast frequency to change over time in BMI is stronger than evidence from cross-sectional studies. However, residual and unmeasured confounding are still possible in spite of our extensive control for many important factors. Another limitation of our study was the necessity to collect data on youth by selfreport on mailed questionnaires, but with our large geographically dispersed cohort, alternatives were not feasible. The impact of the resulting measurement error, if random, International Journal of Obesity


1264 should be to bias estimates of effects toward the null, explaining why many of our estimates, although statistically significant, were quite small. Even if heavier children tended to under-report their weights or their energy intakes, because our analyses utilize 1-y BMI changes and compares overweight children to other overweight children (of different breakfast frequencies), the impact of any such biases should be diminished. Although our assessment of adiposity, using only height and weight, was crude, more reliable body composition techniques75 are not feasible for large epidemiologic studies. Although self-reported BMI has been validated among adolescents,44 we are not aware of any validation studies of change over time in youth self-reported BMI. Furthermore, our assessment of physical activity, inactivity, and energy intake may be sufficiently imprecise that these factors, even though our models controlled for them, could still be responsible for the apparent effects of skipping breakfast. We have not attempted to estimate duration or intensity of each physical activity session, which may be related to weight change, and the wide range of food portion sizes hinders accurate estimation of total energy intake. Although we cannot claim that our cohort of nurses’ offspring is representative of US children, the associations among factors within our cohort should still be internally valid. However, the frequency of eating breakfast in our cohort was reasonably consistent with national data. In the Continuing Survey of Food Intakes by Individuals,76 1989– 1991, 92.8% of 6- to 11-y-old males and 95.6% of females reported eating breakfast the previous day, and 79.4% of 12to 19-y old males and 73.8% of females did so, mirroring the sex-and-age trends in our study. By eighth grade, 85% of a cohort in Minnesota ate breakfast, down from 99% in third grade.77 But 16 y olds in a New York City high school with a large minority population skipped breakfast (at least 3 days in past week) more often (58%) than our 16 y olds.37 Among New Orleans ninth graders providing a single 24-h recall, 23% of females and 14% of males skipped breakfast.32 In a study of 13 y olds in Australia, the most common reasons provided (telephone surveys) for skipping breakfast were lack of time or not being hungry in the morning.78 The Council on Scientific Affairs from the American Medical Association supports prevention of obesity as critical for adults, and for weight loss they support a nutritionally balanced, low-energy diet while increasing energy expenditure through regular physical exercise.79 Similar recommendations have been made for children.80 A review of studies of weight loss and prevention of abnormal weight gain in youth did note some strategies that provided long-term successes.7 Our data suggested that children who skipped breakfast had lower daily energy intakes. However, normal weight children who never ate breakfast tended to gain weight. Overweight children who skip breakfast might lose weight, but alternative methods of reducing energy intake are preferred given the well-documented adverse effects of skipping breakfast on academic performance. International Journal of Obesity

Acknowledgements We are grateful to Karen Corsano, Gary Chase, Gideon Aweh, and our other colleagues in the Growing Up Today Study Research Group. We are especially grateful to the children (and their mothers) for careful completion of the questionnaires.Funded by Grant DK46834 from NIH, an SIPII grant from CDC, a grant from the Boston Obesity Nutrition Research Center (P30 DK46200), research grant, 43-3AEM0-80074 from the Economic Research Service, US Department of Agriculture, and in part by Kelloggs.

References 1 Troiano RP, Flegal KM. Overweight children and adolescents: description, epidemiology, and demographics. Pediatrics 1998; 101 (Suppl): 497–504. 2 Shear CL, Freedman DS, Burke GL, Harsha DW, Webber LS, Berenson GS. Secular trends of obesity in early life: the Bogalusa Heart Study. Am J Public Health 1988; 78: 75–77. 3 Gortmaker SL, Dietz Jr WH, Sobol AM, Wehler CA. Increasing pediatric obesity in the US. Am J Dis Child 1987; 141: 535–540. 4 Pate RR, Ross JG, Dotson CO, Gilbert GG. The National Children and Youth Fitness study: the New norms: a comparison with the 1980 AAHPERD norms. J Phys Educ, Recreat Dance 1985; 56: 28–30. 5 Samuelson G. Dietary habits and nutritional status in adolescents over Europe. An overview of current studies in the Nordic countries. Eur J Clin Nutr 2000; 54 (Suppl): s21–s28. 6 Murata M. Secular trends in growth and changes in eating patterns of Japanese children. Am J Clin Nutr 2000; 72 (Suppl): 1379S–1383S. 7 Fulton JE, McGuire MT, Caspersen CJ, Dietz WH. Interventions for weight loss and weight gain prevention among young: current issues. Sports Med 2001; 31: 153–165. 8 Crawford PB, Story M, Wang MC, Ritchie LD, Sabry ZI. Ethnic issues in the epidemiology of childhood obesity. Pediatr Clin N Am 2001; 48: 855–878. 9 Gunnell DJ, Frankel SJ, Nanchahal K, Peters TJ, Davey Smith G. Childhood obesity and adult cardiovascular mortality. Am J Clin Nutr 1998; 67: 1111–1118. 10 Fagot-Campagna A, Pettit DJ, Engelgau MM, Burrows NR, Geiss LS, Valdez R, Beckles GL, Saaddine J, Gregg EW, Williamson DF, Narayan KM. Type 2 diabetes among North American children and adolescents: an epidemiologic review and a public health perspective. J Pediatr 2000; 136: 664–672. 11 Redline S, Tishler PV, Schluchter M, Aylor J, Clark K, Graham G. Risk factors for sleep-disordered breathing in children. Associations with obesity, race, and respiratory problems. Am J Respir Crit Care Med 1999; 159: 1527–1532. 12 Gold DR, Damokosh AI, Dockery DW, Berkey CS. Body mass index as a predictor of incident asthma in children. Pediatric Pulmonology (in press). 13 Dietz WH. Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics 1998; 101: 518–525. 14 Freedman DS, Dietz WH, Srinivasan SR, Berenson GS. The relation of overweight to cardiovascular risk factors among children and adolescents: The Bogalusa Heart Study. Pediatrics 1999; 103: 1175–1182. 15 Sangi H, Mueller WH. Which measures of body fat distribution is best for epidemiologic research among adolescents? Am J Epi 1991; 133: 870–883. 16 Dwyer T, Blizzard CL. Defining obesity in children by biological endpoint rather than population distribution. Int J Obes Relat Metab Disord 1996; 20: 472–480. 17 Berkey C, Gardner J, Colditz G. Blood pressure in adolescence and early adulthood related to obesity and birth size. Obes Res 1998; 6: 187–195.


1265 18 Dwyer T, Gibbons LE. The Australian Schools Health and Fitness Survey: Physical fitness related to blood pressure but no lipoproteins. Circulation 1994; 89: 1559–1544. 19 Dwyer JT, Stone EJ, Yang M, Feldman H, Webber LS, Must A, Perry CL, Nader PR, Parcel GS. Predictors of overweight and overfatness in a multiethnic pediatric population. Child and Adolescent Trial for Cardiovascular Health Collaborative Research Group. Am J Clin Nutr 1998; 67: 602–610. 20 Gortmaker SL, Must A, Perrin JM, Sobol AM, Dietz WH. Social and economic consequences of overweight in adolescence and young adulthood. N Engl J Med 1993; 329: 1008–1012. 21 Wolf AM, Colditz GA. The cost of obesity: the US perspective. PharmacoEconomics 1994; 5: 34–37. 22 Canning H, Mayer J. ObesityFits possible effect on college acceptance. N Engl J Med 1966; 275: 1172–1174. 23 Roe D, Eickwort K. Relationships between obesity and associated health factors with unemployment among low income women. J Am Med Women’s Assoc 1976; 31: 193–204. 24 Nieto FJ, Szklo M, Comstock GW. Childhood weight and growth rate as predictors of adult mortality. Am J Epidemiol 1992; 136: 201–213. 25 Must A, Jacques PF, Dallal GE, Bajema CJ, Dietz WH. Long-term morbidity and mortality of overweight adolescents: a follow-up of the Harvard Growth Study of 1922 to 1935. N Engl J Med 1992; 327: 1350–1355. 26 Rosenbaum M, Leibel RL. The physiology of body weight regulation: relevance to the etiology of obesity in children. Pediatrics 1998; 101 (Suppl): 525–539. 27 Jacobson KC, Rowe DC. Genetic and shared environmental influences on adolescent BMI: interactions with race and sex. Behav Genet 1998; 28: 265–278. 28 Maffeis C. Aetiology of overweight and obesity in children and adolescents. Eur J Pediatr 2000; 159 (Suppl): S35–S44. 29 Siega-Riz AM, Popkin BM, Carson T. Trends in breakfast consumption for children in the US from 1965–1991. Am J Clin Nutr 1998; 67: 748S–756S. 30 Ruxton CHS, Kirk TR. Breakfast: a review of associations with measures of dietary intake, physiology and biochemistry. Br J Nutr 1997; 78: 199–213. 31 Resnicow K. The relationship between breakfast habits and plasma cholesterol levels in schoolchildren. J School Health 1991; 61: 81–85. 32 Nicklas TA, Reger C, Myers L, O’Neil C. Breakfast consumption with and without vitamin–mineral supplement use favorably impacts daily nutrient intake of ninth-grade students. J Adolesc Health 2000; 27: 314–321. 33 Ortega RM, Requejo AM, Lopez-Sobaler AM, Quintas ME, Andres P, Redondo MR, Navia B, Lopez-Bonilla MD, Rivas T. Difference in the breakfast habits of overweight/obese and normal weight schoolchildren. Int J Vitam Nutr Res 1998; 68: 125–132. 34 Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C. Relationship between feeding pattern and BMI in 220 free-living people in four age groups. Eur J Clin Nutr 1996; 50: 513–519. 35 Wolfe WS, Campbell CC, Frongillo Jr EA, Haas JD, Melnik TA. Overweight schoolchildren in New York State: prevalence and characteristics. Am J Public Health 1994; 84: 807–813. 36 Gibson SA, O’Sullivan KR. Breakfast cereal consumption patterns and nutrient intakes of British schoolchildren. J R Soc Health 1995; 115: 366–370. 37 Pastore DR, Fisher M, Friedman SB. Abnormalities in weight status, eating attitudes, and eating behaviors among urban high school students: correlations with self-esteem and anxiety. J Adolesc Health 1996; 18: 312–319. 38 Wyatt HR, Grunwald GK, Mosca CL, Klem ML, Wing RR, Hill JO. Long-term weight loss and breakfast in subjects in the national weight control registry. Obes Res 2002; 10: 78–82. 39 Schlundt DG, Hill JO, Sbrocco T, Pope-Cordle J, Sharp T. The role of breakfast in the treatment of obesity: a randomized clinical trial. Am J Clin Nutr 1992; 55: 645–651.

40 Rich-Edwards J, Goldman MB, Willett WC, Hunter DJ, Stampfer MJ, Colditz GA, Manson JE. Adolescent body mass index and ovulatory infertility. Am J Obstet Gynecol 1994; 171: 171–177. 41 Strauss RS. Comparison of measured and self-reported weight and height in a cross-sectional sample of young adolescents. Int J Obes Relat Metab Disord 1999; 23: 904–908. 42 Dietz WH, Bellizzi MC. The use of body mass index to assess obesity in children. Am J Clin Nutr 1999; 70: 123S–125S. 43 Roche AF, Siervogel RM, Chumlea WC, Webb P. Grading body fatness from limited anthropometric data. Am J Clin Nutr 1981; 34: 2831–2838. 44 Goodman E, Hinden BR, Khandelwal S. Accuracy of teen and parental reports of obesity and body mass index. Pediatrics 2000; 106: 52–58. 45 Kuczmarksi RJ, Ogden CL, Grummer-Strawn LM, Flegal KM, Guo SS, Wei R, Mei Z, Curtin LR, Roche AF, Johnson CL. CDC growth charts: US advance data from vital and health statistics, No. 314. National Center for Health Statistics, 2000. CDC website: http: //www.cdc.gov/growthcharts. 46 Peterson KE, Field AE, Fox MK, Black B, Simon DS, Yeamans L, Bosch RJ, Smith-Fawzi MK, Gortmaker S, Colditz GA. Validation of the Youth Risk Behavioral Surveillance System (YRBSS) questions on dietary behaviors and physical activity among adolescents in grades 9 through 12. Report to Division of School and Adolescent Health at the Centers for Disease Control and Prevention, 1996. 47 Gortmaker SL, Peterson K, Wiecha J, Sobol AM, Dixit S, Fox MK, Laird N. Reducing obesity via a school-based interdisciplinary intervention among youth: Planet Health. Arch Pediatr Adolesc Med 1999; 153: 409–418. 48 Rifas-Shiman SL, Gillman MW, Field AE, Frazier AL, Berkey CS, Tomeo CA, Colditz GA. Comparing physical activity questionnaires for youth: seasonal vs annual format. Am J Prev Med 2001; 20: 282–285. 49 Rockett HRH, Wolf AM, Colditz GA. Development and reproducibilty of a food frequency questionnaire to assess diets of older children and adolescents. J Am Diet Assoc 1995; 95: 336–340. 50 Rockett HRH, Breitenbach M, Frazier AL, Witschi J, Wolf AM, Field AE, Colditz GA. Validation of a Youth/Adolescent Food Frequency Questionnaire. Prev Med 1997; 26: 808–816. 51 Nicklas TA, McQuarrie A, Fastnaught C, O’Neil CE. Efficiency of breakfast consumption patterns of ninth graders: nutrient-to-cost comparisons. J Am Diet Assoc 2002; 102: 226–233. 52 Morris NM, Udry JR. Validation of a self-administered instrument to assess stage of adolescent development. J Youth Adolesc 1980; 9: 271–280. 53 Harter S. Manual for the self-perception profile for Children. University of Denver: Denver, CO; 1985. 54 Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics 1982; 38: 963–974. 55 SAS Institute Inc. SAS/STAT software: changes and enhancements through release 6.12; Proc Genmod and Proc Mixed. SAS Institute Inc.: Cary, NC; 1997. 1167pp. 56 Siervogel RM, Roche AF, Guo S, Mukherjee D, Chumlea WC. Patterns of change in weight/stature2 from 2 to 18 years: findings from long-term serial data for children in the Fels Longitudinal Growth Study. Int J Obes Relat Metab Disord 1991; 15: 479–485. 57 Buckler JMH. Weight/height relationships through adolescence; a longitudinal study. In: Tanner JM (ed). Auxology 88: perspectives in the science of growth and development. Smith-Gordon: London; 1989. p 373. 58 Casey VA, Dwyer JT, Coleman KA, Valadian I. Body mass index from childhod to middle age: a 50 year followup. Am J Clin Nutr 1992; 56: 14–18. 59 Cronk CE, Roche AF, Kent R, Berkey CS, Reed RB, Valadian I, Eichorn D, McCammon R. Longitudinal trends and continuity in weight/stature2 from 3 months to 18 years. Hum Biol 1982; 54: 729–749. 60 Berkey CS, Rockett HRH, Field AE, Gillman MW, Frazier AL, Camargo CA, Colditz GA. Activity, dietary intake, and weight



1266

61 62 63

64

65 66

67

68 69

70

changes in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics 2000; 105. URL: http://www.pediatrics.org/cgi/content/full/105/4/e56. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73: 13–22. Pollitt E. Does breakfast make a difference in school? J Am Diet Assoc 1995; 95: 1134–1139. Wyon DP, Abrahamsson L, Jartelius M, Fletcher RJ. An experimental study of the effects of energy intake at breakfast on the test performance of 10-year-old children in school. Int J Food Sci Nutr 1997; 48: 5–12. Fischer K, Colombani PC, Langhans W, Wenk C. Cognitive performance and its relationship with postprandial metabolic changes after injestion of different macronutrients in the morning. Br J Nutr 2001; 85: 393–405. Dye L, Lluch A, Blundell JE. Macronutrients and mental performance. Nutrition 2000; 16: 1021–1034. Baric IC, Cvjetic S, Satalic Z. Dietary intakes among Croatian schoolchildren and adolescents. Nutr Health 2001; 15: 127–138. Robinson TN, Hammer LD, Killen JD, Kraemer HC, Wilson DM, Hayward C, Taylor CB. Does television viewing increase obesity and reduce physical activity? Cross-sectional and longitudinal analyses among adolescent girls. Pediatrics 1993; 91: 273–280. Birch LL, Fisher JO. Development of eating behaviors among children and adolescents. Pediatrics 1998; 101: 539–549. Fotheringham MJ, Wonnacott RL, Owen N. Computer use and physical inactivity in young adults: public health perils and potentials of new information technologies. Ann Behav Med 2000; 22: 269–274. Pollitt E, Mathews R. Breakfast and cognition: an integrative summary. Am J Clin Nutr 1998; 67: 804S–813S.


71 Kennedy E, Davis C. US Department of Agriculture School Breakfast Program. Am J Clin Nutr 1998; 67: 798S–803S. 72 Meyers AF, Sampson AE, Weitzman M, Rogers BL, Kayne H. School Breakfast Program and school performance. Am J Dis Child 1989; 143: 1234–1239. 73 Murphy JM, Pagano ME, Nachmani J, Sperling P, Kane S, Kleinman RE. The relationship of school breakfast to psychosocial and academic functioning: cross-sectional and longitudinal observtions in an inner-city school sample. Arch Pediatr Adolesc Med 1998; 152: 899–907. 74 Siega-Riz AM, Popkin BM, Carson T. Differences in food patterns at breakfast by sociodemographic characteristics among a nationally representative sample of adults in the US. Prev Med 2000; 30: 415–424. 75 Goran MI. Measurement issues related to studies of childhood obesity: assessment of body composition, body fat distribution, physical activity, and food intake. Pediatrics (Suppl) 1998; 101: 505–518. 76 Tippett KS, Mickle SJ, Goldman JD, Sykes KE, Cook DA, Sebastian RS. Food and nutrient intake by individuals in the US, 1 day, 1989–91. Continuing survey of food intakes by individuals, 1989–91. Nationwide Food Surveys Report No. 91-2, 1995. 263pp (p 207). 77 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: 222–228. 78 Shaw ME. Adolescent breakfast skipping: an Australian study. Adolescence 1998; 33: 851–861. 79 Council on Scientific Affairs. Treatment of obesity in adults. JAMA 1988; 260: 2547–2551. 80 Barlow SE, Dietz WH. Obesity evaluation and treatment: Expert Committee Recommendations. The Maternal and Child Health Bureau, Health Resources and Services Administration, and Dept Health and Human Services. Pediatrics 1998; 102: E29.