Ashima K Kant, Mark B Andon, Theodore J Angelopoulos, and James M Rippe

Transcription

1 Association of breakfast energy density with diet quality and body mass index in American adults: National Health and Nutrition Examination Surveys, Ashima K Kant, Mark B Andon, Theodore J Angelopoulos, and James M Rippe ABSTRACT Background: Recent reports suggest that dietary energy density (ED) is associated with diet quality, energy intake, and body weight. Breakfast consumption was also associated with diet quality and body weight; however, little is known about the association of breakfast consumption with dietary ED. Objectives: We examined differences in the ED (in energy content/g of food) of diets between breakfast consumers and nonconsumers, and in breakfast reporters we examined the association of ED of breakfast foods with ED of nonbreakfast foods, diet quality, and body mass index (BMI; in kg/m 2 ). Design: We combined dietary data from the 3 continuous National Health and Nutrition Examination Surveys ( ) to determine the ED (in kcal/g) of foods and nutritive beverages and the ED of foods only (n ; 20 y). Linear and logistic regression methods were used to examine the independent associations of breakfast reporting or breakfast ED with 24-h ED, nonbreakfast ED, diet quality, and BMI. Results: The ED of 24-h dietary intake was lower among breakfast reporters than among nonreporters. Women breakfast reporters (but not men) had lower BMI than did nonreporters ( compared with ; P 0.001). With increasing breakfast ED, nonbreakfast ED and fat intake increased, but micronutrient intake and the likelihood of mention of all 5 food groups declined. BMI increased with increasing breakfast ED in men but with increasing nonbreakfast ED in women (P 0.001). Conclusions: Our results support recommendations to encourage breakfast consumption and suggest that the ED of breakfast was associated with diet quality, overall diet ED, and body weight. Am J Clin Nutr 2008;88: INTRODUCTION In observational studies, breakfast consumption was found to be a positive correlate of consumption of at-risk nutrients in children and adults (1 3). The overall energy intake (EI) of breakfast reporters was either higher than that of nonreporters or was not different from that of nonreporters, yet several reports suggest that breakfast consumption may be associated with lower body weight (4 9). Those associations may also differ by the type of breakfast consumed, with cereals reported as having an especially favorable outcome (4, 9). In recent years, dietary energy density (ED; in energy content/g of food) has emerged as a putative regulator of food intake (10 12). Dietary ED was associated with EI in observational and short-term laboratory studies (10 13). Furthermore, observational studies suggest that ED may predict body weight independent of EI (13). In weight-loss intervention trials, decreasing dietary ED promotes greater weight loss (14, 15). In a recent report, the downward secular trend in breakfast reporting from 1971 to 2002 was not predictive of increasing prevalence of obesity in the US population; however, dietary ED tracked increasing prevalence of obesity in the US population during the past 3 decades (16). It was also suggested that the satiety and satiation effects of higher ED meals may be lower than for meals of lower ED (17). Because breakfast is the first meal of the day, ED of this meal may have important implications for subsequent meals consumed. To our knowledge, the association of breakfast consumption with dietary ED or the association of ED of breakfast with subsequent food intake and body weight was not reported previously. Accordingly, in this study, we tested the following hypotheses: 1) there are differences in the ED of diets reported by breakfast consumers and nonconsumers, 2) the ED of breakfast is related to the ED of nonbreakfast eating events in a 24-h recall, and 3) the ED of breakfast is an independent correlate of body weight and quality of dietary intake reported in the 24-h recall. We used nationally representative dietary data to address these questions in adult Americans. SUBJECTS AND METHODS We used data from the continuous National Health and Nutrition Examination Surveys (NHANES) , , and conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention (18 1 From the Department of Family, Nutrition, and Exercise Sciences, Queens College of the City University of New York, Flushing, NY (AKK); Quaker-Tropicana-Gatorade Research and Development Department, Barrington, IL (MBA); the Center for Lifestyle Medicine and Department of Health Professions, University of Central Florida, Orlando, FL (TJA); and the Rippe Lifestyle Institute, Shrewsbury, MA (JMR). 2 Supported by a research grant from the Breakfast Research Institute (AKK). 3 Reprints not available. Address correspondence to AK Kant, Department. of Family, Nutrition, and Exercise Sciences, Remsen Hall, Room 306E, Queens College of the City University of New York, Flushing, NY ashima.kant@qc.cuny.edu. Received March 18, Accepted for publication July 30, doi: /ajcn Am J Clin Nutr 2008;88: Printed in USA American Society for Nutrition

2 BREAKFAST ENERGY DENSITY AND DIET QUALITY ). Each survey was a stratified, multistage, national probability sample of the civilian noninstitutionalized population of the United States (18 20). The survey procedures consisted of a household interview and a health examination of the sample person in the mobile examination center (MEC). Weight and height measurements and interview to collect a 24-h dietary recall were obtained in the MEC. Survey response rates for persons examined in the MEC for the NHANES , , and were 76%, 80%, and 76%, respectively (18 20). Analytic sample All respondents aged 20 y were eligible for inclusion in this study (n ). After exclusions for unavailable recalls (n 1812), incomplete and unreliable recalls (n 89), recalls of pregnant and lactating women (n 843), recalls from proxies (n 270), and reports of 0 calories (n 2), the final analytic sample included men and women. The NHANES included a 24-h dietary recall that was administered by an interviewer using computer-assisted methods. In addition, the NHANES included a second recall obtained by telephone. Because we combined data from 3 independent surveys, we did not use the second recall information in our analysis. Breakfast reporting During the 24-h dietary recall, respondents were asked to name each eating episode reported. We considered respondents who mentioned eating occasions named as breakfast, desayuno, or almuerzo as breakfast reporters. Dietary ED Differences are acknowledged in the association of ED with nutrient intake and body weight, depending on the treatment of beverages in computation of this dietary attribute (13, 21). Because there is no consensus on the definition of ED, in this study, we assessed ED with the use of 2 different measures: 1) the ED (in kcal/g) of foods and nutritive beverages included milk and 100% fruit and vegetable juices (energy from foods and nutritive beverages was divided by the weight of foods and nutritive beverages), and 2) the ED of foods only excluded all beverages and anything added to beverages (energy from foods only was divided by the weight of foods only). We considered it important that 1 ED variable should include nutritive beverages because milk and 100% fruit juice were reported as part of breakfast by 40% and 20% of adult Americans, respectively (22) and because breakfast may be the only eating occasion when these foods are reported. We computed each type of ED for the entire 24-h recall (24-h ED), for foods and beverages reported for breakfast (breakfast ED), and foods and beverages reported for nonbreakfast eating episodes (nonbreakfast ED). Outcomes examined We examined the following outcomes in relation to whether breakfast was reported in the recall: 24-h ED, 24-h EI, amount (in g) of all foods and beverages, and body mass index (BMI; kg/m 2 ). Outcomes examined in association with each type of breakfast ED were 24-h ED and EI, nonbreakfast ED and EI, dietary total and saturated fat intake, mention of all 5 food groups (grain, fruit, vegetable, dairy, and meat or alternate), BMI, and a number of recently identified at-risk micronutrients (vitamins A, E, B-6, and C; folate; and the minerals calcium, potassium, magnesium, phosphorus, and zinc) (23). To determine food group consumption, we assigned all foods reported in each recall to the 5 major food groups with the use of methods described previously (24, 25). The fruit group included all fruits and 100% fruit juice; the vegetable group included all fresh, frozen, canned, raw, and cooked vegetables and juices; the meat group included all meat, poultry, fish, eggs, and meat alternates such as beans, nuts, and seeds; the dairy group included milk, cheese, and yogurt (but not ice cream); and the grain group included all cereals, breads, and pasta (excluded cakes, cookies, pies, and pastries). Food mixtures that included foods from multiple groups (eg, vegetable lasagna) were assigned to all the relevant food groups. We examined the types of foods reported for breakfast by creating all possible combinations of whether a food from the major food groups (dairy, fruit, grain, meat or alternate, and vegetable) was mentioned in the recalled breakfast. With 5 food groups and 2 possibilities (yes or no) of consumption for each food group, 32 different combinations were possible. We have previously used these methods to examine patterns of food group consumption in the US population (25). The percentage of breakfast reporters who mentioned type of food group combinations and the ED of each type associated with the reported pattern were also examined. Statistical analyses The differences in the ED of foods reported by breakfast reporters and nonreporters in the 24-h recall were examined with the use of multivariate regression methods. Sociodemographic and lifestyle variables with putative association with breakfast reporting and ED were included as covariates in these models. The covariates were sex, age, race-ethnicity, survey wave, years of education, income as poverty income ratio, BMI, smoking status, and leisure-time physical activity. Among breakfast reporters, the covariate-adjusted, independent association of each type of breakfast ED with 24-h and nonbreakfast ED and 24-h and nonbreakfast EI was examined with the use of multivariate linear or logistic regression methods. We operationalized each type of breakfast ED variable as weighted tertiles and as a continuous independent variable to examine the above outcomes. Output from linear or logistic multiple regression models was used to compute the adjusted mean or proportion (26) of various dependent variables by breakfastreporting status and by tertiles of breakfast ED of each type. The models for 24-h intake of energy, micronutrients, and BMI as dependent variables also included nonbreakfast ED as a covariate. For those outcomes, we present the regression coefficient, the SE, and the P value associated with both breakfast ED and nonbreakfast ED. We also tested for a curvilinear association of each type of breakfast ED with 24-h and nonbreakfast ED, 24-h and nonbreakfast EI, and the likelihood of mention of a food from the major food groups for breakfast by adding a quadratic term to regression models. Breakfast reporters missing information on the breakfast ED variables and any covariate were excluded from all regression models (n 1791 for ED of breakfast foods and nutritive beverages; n 2026 for ED of breakfast foods only). The sex, age, and race-ethnicity distributions of those excluded from regression models were similar to the distributions in all breakfast reporters (n ). Because of reports of high occurrence of low-energy reporting in national surveys, we also computed the ratio of reported EI to

3 1398 KANT ET AL estimated basal energy expenditure as an indicator of low-energy reporting status. Basal energy expenditure was computed with the use of age-, sex-, and weight-specific equations developed by the Committee on Dietary Reference Intakes (27). We considered a ratio of 1.2 to suggest low-energy reporting. All statistical analyses were performed with the use of SAS (version 9.1; SAS Institute, Cary, NC) and software designed for analysis of complex survey data (SAS-callable SUDAAN) (28). This software computes variance estimates that are corrected for multistage, stratified, cluster probability design of complex surveys. Sample weights provided by the National Center for Health Statistics to correct for differential probabilities of selection, noncoverage, and nonresponse were used in all analyses to obtain point estimates. All reported P values were 2 sided. RESULTS Breakfast was mentioned by 80% of adult Americans on the recall day. The percentage reporting breakfast on the recall day differed by sex, race-ethnicity, age, income, education, smoking status, supplement use, and leisure-time physical activity (P ). These results are available as Table S1 under Supplemental Data in the online issue. Are there differences in the ED of 24-h dietary intake recalled by breakfast reporters and nonreporters? The 24-h EI of breakfast reporters was higher than that of nonreporters (2278 and 1998 kcal, respectively; P ; Table 1). The ED of foods and nutritive beverages and the ED of foods only were lower among breakfast reporters (P and P 0.03, respectively). The association of breakfastreporting status with BMI differed by sex (P for interaction ). The mean BMI of women breakfast reporters was lower than that of nonreporters (27.9 compared with 29.4; P 0.001). BMI and breakfast reporting were unrelated in men (P 0.7). The BMI and breakfast-reporting associations were unchanged by the addition of attempted weight loss in the past year or by consideration of self as underweight, overweight, or right weight to regression models. (Results are shown in Table S2 under Supplemental Data in the online issue.) Sociodemographic correlates of ED of breakfast foods and beverages Age, smoking status, and leisure-time physical activity were independent inverse correlates of both types of breakfast ED (Table 2). In addition, years of education and BMI were correlated with the ED of breakfast foods and nutritive beverages, whereas race-ethnicity was associated with the ED of breakfast foods only. [The unadjusted estimates of breakfast ED variables (mean SE) by sociodemographic characteristics of breakfast reporters are available in Table S3 under Supplemental Data in the online issue.] Types of food group combination patterns reported for breakfast The top 13 food group combination patterns reported for breakfast by 90% of adult Americans on the recall day are ranked in Table 3. Nearly 17% of Americans did not report a food that was grouped into the major food groups for breakfast (these respondents consumed pastries, confectionaries, nonnutritive beverages, juice drinks, meal replacement bars and drinks, etc). The 5 most frequently reported food group combinations were dairy and grain; grain only; dairy, fruit, and grain; grain and meat or alternate; and fruit and grain. The highest mean ED of each type was associated with the breakfast that excluded foods from all the major food groups. Is ED of breakfast foods and beverages related to ED of nonbreakfast foods and beverages reported in the 24-h recall? A significant positive linear association of both types of breakfast ED with 24-h and breakfast EI and 24-h ED and nonbreakfast ED of each type was noted (P 0.004; Table 4). Nonbreakfast EI was inversely associated with the ED of breakfast foods and nutritive beverages (P 0.007) but showed no association with the ED of breakfast foods only. TABLE 1 Reported energy intake, energy density (ED), amount of food, and BMI by breakfast-reporting status: NHANES Dependent variables Mentioned breakfast in the recall? Yes (n 10071) No (n 2245) P 2 Energy intake (kcal) (n ) ED of all foods and nutritive beverages in the 24-h recall (kcal/g) (n ) ED of all foods only (no beverages) in the 24-h recall (kcal/g) (n ) Amount of foods only in the 24-h recall (g) (n ) Amount of all beverages in the 24-h dietary recall (g) (n ) Reporting any nutritive beverage in the 24-h dietary recall (%) (n ) BMI (kg/m 2 ) 5,6 Men (n ) Women (n ) Estimates were obtained from linear or logistic regression models with each outcome in the table as a bivariate or continuous dependent variable and sex, race-ethnicity, age, BMI, poverty income ratio, education, leisure-time physical activity, smoking status, and survey wave as independent variables. 2 Associated with mentioned breakfast (yes or no) from regression models mentioned above. 3 Respondents with information on all covariates in the model. 4 Multivariate adjusted x SE (all such values). 5 Regression models included all of the above and energy intake (kcal). 6 Significant sex mentioned breakfast interaction (P ).

4 BREAKFAST ENERGY DENSITY AND DIET QUALITY 1399 TABLE 2 Independent association of energy density (ED) of breakfast foods with respondent characteristics among breakfast reporters: NHANES Independent variable ED of all breakfast foods and nutritive beverages (n ) ED of all breakfast foods only (no beverages) (n ) kcal/g kcal/g All Sex Men Women P Survey P Race-ethnicity Non-Hispanic white Non-Hispanic black Mexican American Others P Age group (y) P Education (y) P Poverty income ratio to P Smoking status Never Former Current P Any leisure physical activity Yes No P 4 P BMI category (kg/m 2 ) P Estimates were derived from regression models with each type of breakfast ED as a continuous dependent. All variables in the table were entered as independent variables. 2 Breakfast reporters with complete covariate information. 3 Adjusted x SE. 4 Significance of the overall F test of association of each variable in the presence of other independent variables in the model. With increasing breakfast ED, the amount of reported foods and nutritive beverages or foods only at breakfast or nonbreakfast events declined (P 0.001). With increasing ED of breakfast foods and nutritive beverages, the percentage of the population reporting any dairy or fruit for breakfast declined, whereas the percentage reporting any pastry increased (P ). These associations were inconsistent or in the opposite direction for the ED of breakfast foods only (Table 4). Evidence suggests a curvilinear association of both types of breakfast ED variables with nearly all examined outcomes presented in Table 4. With a few exceptions (amount of breakfast and nonbreakfast foods, mention of breakfast pastries), the quadratic term for the remaining outcome variables was negative. Both types of breakfast ED were unrelated with the likelihood of mention of any fruit, dairy, grain, vegetable, or meat at nonbreakfast eating events (P 0.05) (data not shown; available from authors). Is ED of breakfast foods and beverages an independent correlate of diet quality and nutrient intake in a 24-h recall? Breakfast ED (both types) was an independent positive correlate of the percentage of energy from dietary total and saturated fat, but it was an inverse correlate of mention of a food from each of the 5 food groups (grain, fruit, vegetable, meat, and dairy) (Table 5). With increasing ED of breakfast foods and nutritive beverages, the 24-h intakes of all examined micronutrients and dietary fiber decreased (P ). Breakfast ED of foods only also related inversely with intakes of fiber and some micronutrients (except folate, vitamin B-6, calcium, and zinc). Is ED of breakfast foods and beverages an independent correlate of BMI? In regression models adjusted for dietary EI and nonbreakfast ED, the ED of breakfast foods and nutritive beverages was an independent positive correlate of BMI (P 0.001) and a lower likelihood of BMI 30 (P 0.02) in men only (Table 6). In women, the nonbreakfast ED (both types) was a positive correlate of BMI (P 0.001). DISCUSSION Overall, our results show that breakfast reporters had lower 24-h ED, and breakfast ED was associated with ED at nonbreakfast eating events, 24-h ED, and diet quality. Moreover, in men, BMI increased with increasing ED of breakfast foods and nutritive beverages, whereas, for women, consuming breakfast per se was associated with lower BMI as was the ED of nonbreakfast food intake. We are not aware of other studies in which breakfast ED was systematically examined in relation to food intake for the rest of the day, diet quality, and BMI. It is possible that studies with a different definition of breakfast and ED may yield different results. Differences in the ED of diets recalled by breakfast reporters and nonreporters Our finding of higher EI among breakfast reporters is consistent with previous reports (4, 9). Notably, although the mean amount of foods was nearly 20% higher in breakfast reporters, the corresponding increase in mean 24-h EI was 13%, which is consistent with our observation of lower 24-h ED for breakfast reporters (Table 1) This finding possibly reflects the nature of

5 1400 KANT ET AL TABLE 3 Food group combinations reported for breakfast in a dietary recall and the associated energy density (ED) of each type among breakfast reporters: NHANES Food group combination 1 Weighted reporting ED of breakfast foods and nutritive beverages 2 (n ) ED of breakfast foods only 2 (n ) % kcal/g kcal/g No food from major food groups Dairy grain Grain (only) Dairy fruit grain Grain meat or alternate Fruit grain Fruit (only) Grain meat or alternate vegetable Dairy grain meat or alternate Fruit grain meat or alternate Dairy (only) Dairy grain meat or alternate vegetable Meat or alternate (only) All other food group combinations Obtained from considering all possible combinations of whether fruit, dairy, grain, meat or alternate, and vegetable were mentioned for breakfast. With 5 food groups at 2 levels each (present or absent), 32 combinations were possible. For example, a reported food group combination of dairy grain implies that a food from the dairy and a food from the grain group were mentioned for breakfast. The table includes the 13 most frequently reported combinations of the major food groups. 2 Obtained from regression models with each breakfast ED variable as a continuous outcome. Independent variables included sex, race-ethnicity, age, BMI, poverty income ratio, education, leisure-time physical activity, smoking status, survey wave, and a 14-category breakfast-type variable. 3 Breakfast reporters with complete covariate information. 4 Foods reported in this breakfast type did not meet the study criteria for inclusion in the major food groups (fruit, dairy, grain, meat or alternate, and vegetable). Examples of foods reported for this type of breakfast included pastries, confectionaries, meal replacement bars, and nonnutritive energy-containing beverages, etc. 5 Adjusted x SE (all such values). commonly consumed breakfast foods with relatively lower energy content and the consumption of nutritive beverages at breakfast. The mean amount of foods reported was higher among breakfast reporters, but the amount of beverages reported by breakfast consumers and nonconsumers did not differ. Because nearly 60% of breakfast consumers reported a nutritive beverage (milk or 100% juice) relative to 35% of nonconsumers of breakfast (Table 1), these findings may suggest that breakfast omission is associated with greater consumption of caloric and noncaloric beverages of low-nutrient content (eg, coffee, tea, soft drinks, juice drinks, etc). Is ED of breakfast foods and beverages related to ED of nonbreakfast foods and beverages and 24-h EI? Both types of breakfast ED variables were strong positive correlates of each type of ED of nonbreakfast eating events. The explanation for these findings is unclear, but it may suggest that patterns of food selection (which, in turn, relate to food preferences) tend to be similar across eating events in the day. In other words, those who choose a higher ED breakfast also choose other meals of higher ED. In addition, food selection at breakfast may affect physiologic or psychological factors or both involved with subsequent eating events in the day. Relatively few studies have examined the association of ED of self-selected eating episodes with subsequent intake. de Castro (29) analyzed 7-d food diaries from 867 adults and found the ED of each eating event, including the morning period, in the day to be a positive correlate of total EI during the day. Mazlan et al (30) examined the effect of consuming a high-ed or low-ed breakfast and a mandatory snack on self-selected EI during the rest of the day in 16 lean men. High-ED breakfast treatment was associated with lower lunch energy but not postlunch EI. Total selfselected EI in the high-ed treatment was lower than in the low-ed breakfast treatment; however, with inclusion of the mandatory meal, the EI in the high-ed breakfast treatment was higher. In our study, both types of ED of breakfast events were related with 24-h EI when there was no adjustment for ED of nonbreakfast eating events (Table 4). With adjustment for ED of nonbreakfast eating events, breakfast ED was no longer a significant correlate of 24-h EI (Table 5). These comparisons, however, are of limited value because the dietary assessment methods, definition of eating events, ED, sample size, and statistical methods used in our study were different from those mentioned in the work of de Castro (29) and Mazlan et al (30). Is ED of breakfast foods an independent correlate of breakfast quality and 24-h diet quality? Our results show that for assessment of breakfast quality and its association with quality of the overall dietary intake, the ED of breakfast foods and nutritive beverages was more informative than the ED of breakfast foods only. As shown in Table 4, respondents in the first tertile of ED of breakfast foods and nutritive beverages were more likely to report a nutritive beverage, any dairy, and any fruit but were less likely to mention a pastry or meat or alternate for breakfast. Although respondents in the first

6 BREAKFAST ENERGY DENSITY AND DIET QUALITY 1401 TABLE 4 Energy intake and energy density (ED) of nonbreakfast foods and beverages reported in the 24-h recall, by tertile (T) of ED of breakfast foods and beverages, among breakfast reporters: NHANES ED of all breakfast foods and nutritive beverages (n ) ED of all breakfast foods only (no beverages) (n ) Dependent variable T1 ( 1.05 kcal/g) T2 ( kcal/g) T3 ( 2.12 kcal/g) P for trend 3 T1 ( 2.03 kcal/g) T2 ( kcal/g) T3 ( 3.23 kcal/g) P for trend 3 Linear Quadratic Linear Quadratic 24-h Energy intake (kcal) Breakfast energy intake (kcal) Nonbreakfast energy intake (kcal) h ED Nonbreakfast ED Amount of breakfast foods (g) Amount of nonbreakfast foods (g) Reported any nutritive beverage in the recall (%) Reported any dairy food for breakfast (%) Reported any fruit for breakfast (%) Reported any grain for breakfast (%) Reported any vegetable for breakfast (%) Reported any meat or alternate food for breakfast (%) Reported any pastry, etc, for breakfast (%) EI:BEE Estimates were derived from linear or logistic regression models with each dependent variable as a continuous or a binary outcome. Independent variables included sex, race-ethnicity, age, BMI, poverty income ratio, education, leisure-time physical activity, smoking status, survey wave, and the breakfast ED variable as tertiles. EI:BEE, ratio of 24-h energy intake to basal energy expenditure. 2 Breakfast reporters with complete covariate information. 3 Associated with each breakfast-ed variable (continuous) and each breakfast-ed variable as a quadratic term. 4 Adjusted x SE (all such values). 5 The regression coefficient for the breakfast ED term was positive. 6 The regression coefficient for the breakfast ED term was negative. 7 Amount of breakfast grams and nonbreakfast grams includes foods and nutritive beverages or foods only, depending on the type of variable.

7 1402 KANT ET AL TABLE 5 Independent association of energy density (ED) (kcal/g) of breakfast and nonbreakfast foods and beverages with diet quality variables in a 24-h dietary recall, among breakfast reporters: NHANES ED of foods and nutritive beverages (n ) ED of foods only (n ) Reported for breakfast Reported for nonbreakfast Reported for breakfast Reported for nonbreakfast SE P SE P SE P SE P 24-h Energy intake (kcal) Mentioned all 5 food groups Number of unique foods Fat energy (%) Saturated fat energy (%) Fiber (g) Vitamin C (mg) Folate ( g) Vitamin A ( g RAE) Vitamin E (mg AT) Vitamin B-6 (mg) Calcium (mg) Potassium (mg) Magnesium (mg) Phosphorus (mg) Zinc (mg) Estimates were derived from regression models with each dependent variable as a continuous outcome. Independent variables included sex, raceethnicity, age, BMI, poverty income ratio, education, leisure-time physical activity, smoking status, survey wave, breakfast energy density (continuous), and nonbreakfast energy density (continuous). Separate regressions were run for each type of breakfast ED variable as a predictor. RAE, retinol activity equivalent; AT, -tocopherol. 2 Breakfast reporters with complete covariate information. tertile of ED of breakfast foods also were more likely to report a fruit, associations with reporting of other food groups were inconsistent. Moreover, only the ED of breakfast foods and nutritive beverages was associated with all micronutrient outcomes examined in our study. Because the ED of postbreakfast foods represents 80% of the daily EI, we expected strong associations of this variable with daily nutrient intake. Surprisingly, the ED of nonbreakfast foods and nutritive beverages was not associated with 24-h intakes of vitamin E and zinc, whereas the ED of breakfast foods and nutritive beverages did. (These associations may reflect the contribution of commonly consumed breakfast foods that include fortified breakfast cereals.) Moreover, because the ED of breakfast foods and nutritive beverages was not related to 24-h EI independent of the ED of nonbreakfast foods and beverages, it is clear that the strong association of this ED variable with nutrients and diet quality is not through higher EI but because lower ED breakfast foods make a disproportionately high contribution to intake of these high-risk nutrients relative to their energy content. Nearly 17% of adult Americans reported a breakfast type that included pastries, confectionaries, meal replacement drinks or bars, and nonnutritive beverages but no foods that met the study TABLE 6 Independent association of BMI with energy density (ED) (kcal/g) of breakfast and nonbreakfast foods, by sex, among breakfast reporters: NHANES ED of foods and nutritive beverages (n 4117 men and 4148 women 2 ) Reported for breakfast Reported for nonbreakfast events ED of foods only (n 4010 men and 4013 women 2 ) Reported for breakfast Reported for nonbreakfast events SE P SE P SE P SE P BMI (kg/m 2 ) as a continuous dependent Men Women BMI 30 or 30 (kg/m 2 ) as a binary dependent Men Women Estimates were derived from regression models with BMI as a continuous (linear regression) or binary (logistic regression) dependent. Independent variables included sex, race-ethnicity, age, poverty income ratio, education, leisure-time physical activity, smoking status, survey wave, breakfast energy density (continuous), and nonbreakfast ED (continuous), and energy intake (in kcal). Separate regressions were run for each type of breakfast ED variable as a predictor. 2 Breakfast reporters with complete covariate information.

8 BREAKFAST ENERGY DENSITY AND DIET QUALITY 1403 criteria for inclusion in the major food groups on the recall day (Table 3). Not surprisingly, this breakfast pattern was associated with the highest mean breakfast ED of each type. To our knowledge, previous attempts to examine breakfast types have not assessed reported patterns of food group combinations (4, 9); therefore, they are not comparable to our results. Differences in the BMI of breakfast reporters and nonreporters and by ED of reported breakfast Sex differences are apparent in body weights of persons reporting breakfast as well as the ED of the breakfast recalled. Women (but not men) with higher BMI were more likely to not report breakfast. For those reporting breakfast, men with a higher BMI reported breakfasts of higher ED; conversely, the ED of nonbreakfast foods did not predict BMI. However, women with a higher BMI reported higher ED of all nonbreakfast eating episodes. These findings need confirmation from other data sets, but they suggest that women be encouraged to consume breakfast and moderate the ED of the day s intake, whereas for men reducing the ED of breakfast may be an appropriate target. Cho et al (9) have reported lower BMI in adults consuming ready-to-eat or cooked cereals compared with breakfast skipping or a breakfast based on meat and eggs in the NHANES III. Although the ED was not reported, those results are generally consistent with our findings because the ED of typical cold cereals with milk or a cooked cereal such as oatmeal is 1 kcal/g compared with a breakfast sandwich of ham, eggs, and cheese with an ED of 2.4 kcal/g. Recent reports suggest a high likelihood of low-energy reporting in national surveys, with especially higher prevalence in association with higher body weight (31 33). Low-energy reporting may reflect reporting bias or weight control attempts by overweight respondents, but it can affect possible conclusions about the association of body weight with dietary attributes. In an attempt to understand whether the observed associations of breakfast consumption and ED of breakfast foods with BMI reflect such a bias, we adjusted for history of weight loss and self-image in our analyses. The results were essentially unchanged. If higher BMI respondents inaccurately report lower ED foods at breakfast or nonbreakfast events, we may expect an inverse association between BMI and breakfast or nonbreakfast ED. However, as mentioned earlier, we found a positive association of BMI with the ED of breakfast foods in men and with the ED of nonbreakfast foods in women. In conclusion, our results suggest that breakfast consumers had lower ED of the day s dietary intake. Furthermore, it is important that interventions that promote breakfast include information on food choices that result in lower ED of breakfast. We thank Lisa Licitra Kahle for expert SAS and SUDAAN programming support. The authors responsibilities were as follows AKK: was responsible for all aspects of this study, including conceptualization of the study question, analytic strategy, data analysis, interpretation of study results, and preparation of the manuscript; MBA, TJA, and JMR: were involved in conceptualization of the study question, interpretation of study results, and preparation of the manuscript. AKK was partially supported by a research grant from the Breakfast Research Institute. The Breakfast Research Institute is sponsored by Quaker and Tropicana, manufacturers of commonly consumed breakfast foods and beverages. MBA is employed by Quaker-Tropicana-Gatorade, a division of PepsiCo, that manufactures a number of foods and beverages typically consumed at breakfast. TJA had no personal or financial conflict of interest. JMR has received consulting fees and research grants from Quaker and Tropicana, sponsors of the Breakfast Research Institute. Dr. Rippe s research organization, Rippe Lifestyle Institute, manages the Breakfast Research Institute on behalf of Quaker and Tropicana. REFERENCES 1. Kerver JM, Yang EJ, Obayashi S, Bianchi L, Song WO. Meal and snack patterns are associated with intake of energy and nutrients in US adults. J Am Diet Assoc 2006;106: Affenito SG, Thompson DR, Barton BA, et al. Breakfast consumption by African-American and white adolescent girls correlates positively with calcium and fiber intake and negatively with body mass index. J Am Diet Assoc 2005;105: Ruxton CHS, Kirk TR. Breakfast: a review of associations with measures of dietary intake, physiology, and biochemistry. Br J Nutr 1997; 78: Song WO, Chun OK, Obayashi S, Cho S, Chung CE. Is consumption of breakfast associated with body mass index in US adults? J Am Diet Assoc 2005;105: 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: Ma Y, Bertone ER, Stanek EJ III, et al. Association between eating patterns and obesity in a free-living US adult population. Am J Epidemiol 2003;158: Berteus Forslund H, Lindroos AK, Sjostrom L, Lissner L. Meal patterns and obesity in Swedish women a simple instrument describing usual meal types, frequency and temporal distribution. Eur J Clin Nutr 2002; 56: Ortega RM, Redondo MR, Lopez-Sobaler AM, et al. Associations between obesity, breakfast-time food habits and intake of energy and nutrients in a group of elderly Madrid residents. J Am Coll Nutr 1996;15: Cho S, Dietrich M, Brown CJ, Clark CA, Block G. The effect of breakfast type on total daily energy intake and body mass index: results from the Third National Health and Nutrition Examination Survey (NHANES III). J Am Coll Nutr 2003;22: Poppitt SD, Prentice AM. Energy density and its role in the control of food intake: evidence from metabolic and community studies. Appetite 1996;26: Stubbs J, Ferres S, Horgan G. Energy density of foods: effects on energy intake. Crit Rev Food Sci Nutr 2000;40: Bell EA, Castellanos VH, Pelkman CL, Thorwart ML, Rolls BJ. Energy density of foods affects energy intake in normal-weight women. Am J Clin Nutr 1998;67: Kant AK, Graubard BI. Energy density of diets reported by American adults: association with food group intake, nutrient intake, and body weight. Int J Obes (Lond) 2005;29: Rolls BJ, Bell EA. Dietary approaches to treatment of obesity. Med Clin North Am 2000;84: Ledikwe JH, Rolls BJ, Smiciklas-Wright H, et al. Reductions in dietary energy density are associated with weight loss in overweight and obese participants in the PREMIER trial. Am J Clin Nutr 2007;85: Kant AK, Graubard BI. Secular trends in patterns of self-reported food consumption of adult Americans: NHANES to NHANES Am J Clin Nutr 2006;84: Rolls BJ. The role of energy density in the overconsumption of fat. J Nutr 2000;13(suppl):268S 71S. 18. Centers for Disease Control and Prevention (CDC). National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey (NHANES). Hyattsville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention. NHANES Internet: nhanes99_00.htm (accessed March 21, 2007). 19. Centers for Disease Control and Prevention (CDC). National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey (NHANES). Hyattsville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention. NHANES Internet: nhanes01-02.htm (accessed March 21, 2007).

9 1404 KANT ET AL 20. Centers for Disease Control and Prevention (CDC). National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey (NHANES). Hyattsville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention. NHANES Internet: nhanes /nhanes03_04.htm. (accessed March 21, 2007). 21. Cox DN, Mela DJ. Determination of energy density of freely selected diets: methodological issues and implications. Int J Obes Relat Metab Disord 2000;24: US Department of Agriculture. Beltsville Human Nutrition Research Center, Food Surveys Research Group. Breakfast in America, Internet: pdf/breakfast_2001_2002.pdf (accessed May 6, 2007). 23. Moshfegh A, Goldman J, Cleveland L. What We Eat in America, NHANES : Usual Nutrient Intakes from Food Compared to Dietary Reference Intakes. Beltsville, MD: US Department of Agriculture, Agricultural Research Service, 2005;Internet: usda.gov/ba/bhnrc/fsrg (accessed May 2007). 24. Kant AK, Block G, Schatzkin A, Ziegler RG, Nestle M. Dietary diversity in the US population, NHANES II, J Am Diet Assoc 1991; 91: Kant AK, Schatzkin A, Block G, Ziegler RG, Nestle M. Food group intake patterns and associated nutrient profiles of the US population. J Am Diet Assoc 1991;91: Graubard BI, Korn E. Predictive margins with survey data. Biometrics 1999;55: Institute of Medicine. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein, and amino acids. Washington, DC: National Academy Press, Research Triangle Institute. SUDAAN. Release 9.0. Research Triangle Park, NC, de Castro JM. The time of day of food intake influences overall intake in humans. J Nutr 2004;134: Mazlan N, Horgan G, Stubbs RJ. Energy density and weight of food effect short-term caloric compensation in men. Physiol Behav 2006;87: Briefel RR, McDowell MA, Alaimo K, et al. Total energy intake of the US population: third National Health and Nutrition Examination Survey, Am J Clin Nutr 1995;62(suppl):1072S 80S. 32. Kant AK. The nature of dietary reporting by adults in the Third National Health and Nutrition Examination Survey, J Am Coll Nutr 2002;21: Pryer JA, Vrijheid M, Nichols R, Kiggins M, Elliott P. Who are the Low energy reporters in the dietary and nutritional survey of British adults? Int J Epidemiol 1997;29: