Reported Consumption of Low-Nutrient-Density Foods by American Children and Adolescents: Nutritional and Health Correlates, NHANES III, 1988 to 1994

Objective  To examine the contribution of foods of modest nutritional value to the diets of American children and adolescents.

Methods  The data were from the third National Health and Nutrition Examination Survey, 1988 to 1994, and included 4852 children and adolescents, aged 8 to 18 years. Foods reported in the 24-hour dietary recall were grouped into the following low-nutrient-density (LND) food categories: visible fat; table sweeteners, candy, and sweetened beverages; baked and dairy desserts; salty snacks; and miscellaneous. The independent association of the number of LND foods mentioned in the recall with intake of food groups, macronutrients, micronutrients, and body mass index was examined by means of regression procedures to adjust for multiple covariates.

Results  The LND foods contributed more than 30% of daily energy, with sweeteners and desserts jointly accounting for nearly 25%. Intakes of total energy and percentage of energy from carbohydrate and fat related positively, but percentage of energy from protein and dietary fiber (in grams) related inversely to the reported number of LND foods (P<.05). The reported number of LND foods was a negative predictor (P<.001) of the amount of nutrient-dense foods reported. The mean amount of reported intake of several micronutrients—vitamins A, B₆, and folate, and the minerals calcium, magnesium, iron, and zinc—declined (P<.05) with increasing tertiles of reported number of LND foods. The LND food reporting was not a significant predictor of body mass index.

Conclusion  High LND food reporting was related to higher energy intake but lower amounts of the 5 major food groups and most micronutrients.

MANY AMERICAN children and adolescents consume diets that provide marginal amounts of several nutrients, including vitamins A, E, B₆, and folate, and the minerals calcium, magnesium, iron, and zinc.^1,2 These essential nutrients have well-known metabolic functions, and many have been linked to health promotion and disease prevention.^3-6 Recent survey data also suggest an alarming trend in increasing prevalence of adiposity in US children and adolescents.^7,8 Estimates from the Continuing Survey of Food Intakes by Individuals, 1994 to 1996, suggest a dramatic increase in consumption of relatively energy-dense foods of modest nutrient density.⁹ The low-nutrient-density (LND) foods provide highly palatable energy, possibly at the expense of foods that are sources of essential nutrients.^10-12 It is tempting to speculate that dietary patterns characterized by excessive consumption of LND foods may contribute to increased risk of marginal nutrient intakes and positive energy balance.

However, there has been no systematic, comprehensive study of the extent of contribution of energy-dense foods of modest nutritional value to the diets of American children or the nutritional and health implications, if any, of their consumption. Past attempts at examining LND food intake in children have included examination of the contribution of added sugar or carbonated beverages to the diets of children.^11-18 There is some evidence that food selection patterns favored in childhood may track through adult years¹⁹; therefore, a better understanding of patterns of food consumption is clearly warranted. The purpose of this study was to examine nationally representative data from the third National Health and Nutrition Examination Survey (NHANES III, 1988-1994) for consumption of LND foods by 8- to 18-year-old American children and adolescents.

The NHANES III is a multistage, stratified, probability sample of the noninstitutionalized, civilian US population, aged 2 months and older.²⁰ The survey was conducted by the National Center for Health Statistics and included administration of a questionnaire at home and a full medical examination along with a battery of tests in a special mobile examination center.²⁰ Demographic and medical history information was obtained during the household interview. The examination at the mobile examination center included physical and dental examinations, dietary interview, body measurements, and collection of blood and urine samples. Body weight, height, and circumference at various body sites were measured by standardized procedures in the mobile examination center.²⁰

A 24-hour dietary recall was collected by a trained dietary interviewer in a mobile examination center interview with the use of an automated, microcomputer-based interview and coding system.²⁰ The type and amount of foods consumed were recalled with recall aids such as abstract food models, special charts, measuring cups, and rulers to help in quantifying the amounts consumed. Special probes were used to help the recall of commonly forgotten items such as condiments, accompaniments, fast foods, and alcoholic beverages.

All NHANES III respondents aged 8 to 18 years with a 24-hour recall considered complete and reliable by the National Center for Health Statistics were eligible for inclusion in this study (n = 4889). Recalls of pregnant (n = 34) or nursing (n = 3) respondents were excluded, to result in a final analytic sample size of 4852 (2383 boys and 2469 girls) representing a population estimate of 38 738 796 children and adolescents.

To determine the intake of LND foods, it was necessary to identify foods belonging in this category from those reported in the 24-hour dietary recall. As a first step, the 2578 foods reported by the analytic sample were classified as belonging to 1 or more of the 5 major food groups (dairy, fruit, grain, meat, and vegetable) or the LND food group by methods described previously.^10,21,22 Briefly, the assignment of foods into the various groups was dependent on their nutrient content and uses in the diet. The dairy group included milk, yogurt, cheese, and buttermilk, but excluded butter, cream cheese, and dairy desserts. The fruit group included all fresh, frozen, dried, and canned fruits and fruit juices, but excluded fruit drinks. The grain group included all breads, cereals, pastas, and rice, but not pastries. The meat group included meat, poultry, fish, eggs, and meat alternates such as dried beans, nuts, and seeds. The vegetable group included all raw or cooked fresh, frozen, and canned vegetables and juices. Mixed dishes containing foods from several groups were grouped into all the relevant groups. Foods excluded from these major food groups were grouped into the LND foods group. The LND foods were further placed into 5 subgroups as follows: (1) visible fat: butter, oil, dressings, gravies, etc; (2) sweeteners: sugar, syrup, candy, carbonated and noncarbonated sweetened drinks, etc; (3) baked and dairy desserts: cookies, cakes, pies, pastries, ice cream, puddings, cheesecakes, etc; (4) salted snacks: potato, corn, and tortilla chips, etc; and (5) miscellaneous: coffee, tea, condiments, etc.

The LND foods variable was operationalized in 3 different ways: (1) number of mentions of LND foods and beverages in the recall, (2) amount (in grams) of LND foods and beverages, and (3) percentage of total energy from LND foods and beverages.

The NHANES III nutrient database for individual foods, which is derived from the US Department of Agriculture's Survey Nutrient Database, was used for determining energy and nutrient content of all foods.²³ We examined the mean intake of selected micronutrients (vitamins A, E, C, B₆, and folate, and the minerals calcium, magnesium, iron, and zinc) by tertiles of the reported number of LND foods. The reported intake of each nutrient was also examined with reference to the most recent age-sex–specific standard available, the estimated average requirement.^3-6 As an estimate of dietary misreporting, the ratio of reported energy intake to energy expenditure for basal needs was computed. Energy expenditure for basal needs was estimated by prediction equations developed by the Dietary Reference Intakes Committee for normal and overweight or obese 3- to 18-year-olds.²⁴

Data on serum concentrations of vitamin C, folate, and carotenoids were obtained from the National Center for Health Statistics public release compact disks.^25,26 The methods used for measurement of these serum analytes and their associated errors have been described.^25,26 Serum folate, ascorbate, and the carotenoids—α-carotene, β-carotene, β-cryptoxanthin, lutein-zeaxanthin, and lycopene—were chosen because the dietary intake of each nutrient is believed to be a major determinant of serum concentration of the respective nutrient and can thus serve as a biomarker of dietary exposure.^4,5 Serum homocysteine has been reported as an independent predictor of coronary heart disease risk and may have dietary determinants.^27,28

The number, amount, and proportion of daily energy contributed by all LND foods and subgroups were computed, and each LND food variable was examined as weighted tertiles. The mean daily energy, percentage of energy from macronutrients, and mean percentage of the population meeting the dietary reference intake of selected nutrients were obtained by tertiles of each LND foods variable. The estimates of nutrient and food group intake were adjusted for age (continuous), race (non-Hispanic white, non-Hispanic black, Mexican American, other), sex, family income (<$10 000, $10 000-$29 999, $30 000-$49 9999, ≥$50 000, unknown), highest grade of education reported for the family respondent (<12, 12, >12, or unknown grade), and total energy intake (in kilocalories). Adjusted mean concentration of serum analytes by tertiles of the reported number of LND foods was obtained separately for boys and girls. These models were adjusted for age, race, dietary intake of the relevant nutrient, supplement use in the 24 hours before phlebotomy, supplement use in the past month, and hours of fasting before phlebotomy.

The procedure used for obtaining covariate-adjusted estimates and SEs from survey data is based on Taylor linearization methods according to Graubard and Korn.²⁹ All statistical analyses were performed with SAS³⁰ and software designed for analysis of survey data (SUDAAN).³¹ This software generates variance estimates that are corrected for multistage stratified 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.²⁰

The independent association of each LND foods variable with food and nutrient intake was examined by regression procedures to adjust for covariates mentioned above. Linear regression procedures were used when the outcome variables were continuous (eg, dietary nutrient intake or serum analyte concentration). For categorical outcomes, such as whether the amount meeting the standard of a nutrient intake was reported, logistic regression procedures were used. Most of the results shown in the article are for the LND foods variables expressed as number of mentions.

More than 30% of daily energy in the diets of 8- to 18-year-old US children and adolescents was contributed by LND foods, with sweeteners and desserts jointly accounting for nearly 25% (Table 1). The third tertile of the number of LND foods mentioned had a higher proportion of non-Hispanic whites, respondents with higher family income, and supplement users (Table 2). The 2-day average number of hours of television watched, the weekly hours reportedly spent on physical activity, body mass index, self-perceived body weight status, or history of attempted weight loss in the past year were unrelated to the reported number of LND foods (Table 2).

Table 3 presents the adjusted mean ± SE of energy and macronutrient intake by tertiles of the reported number of LND foods. Intakes of total energy and percentage of energy from carbohydrate and fat were positively related, but percentages of energy from protein and dietary fiber were related inversely to the number of LND foods mentioned. Percentage of energy from saturated fat was not related to the reported number of LND foods. The total amount of all reported foods and beverages, amount of LND foods, and percentage of energy from LND foods related positively, but the amounts of foods and beverages from the 5 major food groups (grain, fruits, vegetables, meat and alternates, dairy) were related inversely to the reported number of LND foods (P<.001).

Table 4 shows the mean amount of selected food groups and nutrients, and percentage of US children and adolescents meeting the dietary reference intake for selected nutrients on recall day, by tertiles of the reported number of LND foods. The mean amounts of foods from the fruit and dairy groups declined with increasing LND food tertiles in both boys and girls. The odds of mentioning any food from the fruit or the dairy groups on recall day were also lower in girls in the third tertile of number of LND foods mentioned. The odds of mentioning a food from the grain, vegetable, and meat or alternate groups did not differ among tertiles of number of LND foods (data are shown for dairy and fruit groups only).

In girls, the reported mean amounts of all micronutrients examined (except vitamin E) declined with tertiles of reported number of LND foods (P<.05). In boys, the reported mean amounts of folate, vitamin B₆, calcium, and magnesium declined with tertiles of reported number of LND foods. The percentage of children and adolescents meeting the respective standard of intake of folate, vitamins B₆ and A, and the minerals magnesium, iron, and zinc on recall day decreased with increasing tertiles of number of LND foods. Children and adolescents in the highest third of LND food reporting had at least 50% lower odds of meeting the estimated average requirement for folate, vitamin B₆, magnesium, iron, and zinc on recall day.

The independent association of the reported number of LND foods with the serum concentrations of selected vitamins and carotenoids was also examined. In multiple regression models, the number of LND foods was not an independent predictor of serum concentration of most of the analytes examined. Serum homocysteine concentration was available for 1209 children and adolescents in the analytic sample. Multicovariate-adjusted serum homocysteine concentration for the first, second, and third tertiles of number of LND foods mentioned was 0.90 ± 0.027, 0.97 ± 0.081, and 1.07 ± 0.041 mg/L (6.7 ± 0.2, 7.2 ± 0.6, and 7.9 ± 0.3 µmol/L), respectively (P for trend = .009). It is noteworthy that dietary vitamin C and folate were significant positive predictors of serum concentration of vitamin C and folate, respectively, in both boys and girls. Dietary carotenoid intake related positively to serum concentration of α-carotene and β-carotene, but was unrelated to serum β-cryptoxanthin and lycopene. (Data on serum analyte concentration are not shown and are available from the author.)

The results presented above were largely the same when the LND food variable was expressed as amount reported (in grams) or as percentage of total energy intake.

American children and adolescents reported nearly a third of their daily energy intake from relatively energy-dense, low-nutrient foods, with sweeteners, candy, carbonated or noncarbonated beverages, and desserts providing approximately a quarter of the total daily energy intake. A third of the 8- to 18-year-old population examined reported 6 or more LND foods on the recall day, providing an average of 41% of the total daily energy intake. To my knowledge, few other directly comparable estimates have been published. Differences in methods used to group foods make comparisons difficult to interpret. In the National Adolescent Health Survey, nearly 60% of those surveyed reported having consumed "junk food snacks" on the previous day.³² The proportion reporting LND food intake was similarly high in the Youth Risk Behavior Surveillance System in 1993.³³ Adolescents studied by Bandini and colleagues³⁴ reported approximately 20% to 27% of their total daily energy intake from chips, candy, soda, baked goods, and ice cream. From the list of food sources of energy in 24-hour recalls of 2- to 18-year-olds reported by Subar et al,³⁵ it can be estimated that approximately 30% of the energy came from foods comparable to those considered LND in the present study. Other available estimates of consumption of LND foods are limited to carbonated beverages or added sugars.^11-18

Although the percentage of energy from carbohydrate increased with the reported number of LND foods, grams of fiber decreased, confirming higher consumption of refined carbohydrates and lower consumption of fruits, vegetables, and grains that provide dietary fiber and complex carbohydrates. Furthermore, increasing carbohydrate intake in association with the reported number of LND foods was accompanied by a lower percentage of energy from protein but not fat. Instead, the percentage of energy from fat increased slightly with in creasing number of LND foods. It is likely that more of the fat in the diets of those reporting a high intake of LND foods came from prepared snack foods. Such foods may contain more of hydrogenated fats that include trans–fatty acids.³⁶Trans–fatty acids are also believed to be atherogenic and have been implicated in increased risk of coronary heart disease.³⁷

It is evident from the data presented inTables 3 and 4 that there were both qualitative and quantitative differences in dietary patterns of children who reported high LND food intake. Foods from the dairy and fruit groups were less likely to be mentioned and the amount of all nutrient-dense foods reported decreased with increasing number of LND foods. These data suggest that, on the average, reporting of LND foods was associated negatively with reporting of nutrient-dense foods. However, these are cross-sectional data and do not allow definitive conclusions about displacement of nutrient-dense foods by LND foods.

Given the food intake data, it is not surprising that the proportion of children and adolescents, especially girls, meeting the dietary reference intakes of most examined micronutrients (except vitamins C and E) on recall day also declined with increasing tertiles of number of LND foods reported. The positive association of dietary ascorbate, folate, and carotenoids with serum ascorbate, folate, and selected carotenoids, respectively, validates the reported dietary intakes of these micronutrients. The published literature on nutritional implications of LND food intake is limited to examining the relationship of sugar intake (or carbonated beverages) with nutritional quality of diets, with equivocal results.^11-18 Nutritional quality of diets was found to be negatively impacted by carbonated beverage consumption by some,^11,12 while others have concluded that high sugar consumers had higher energy intake without adverse nutritional consequences.^13,15,16

The reported hours spent on exercise or watching television varied little by the number of LND foods reported. These data do not support the widely held belief that a high LND food intake is associated with more television time or limited physical activity. However, the number of LND foods mentioned was an independent positive predictor of energy intake; increased energy intake in association with LND food intake without a concurrent increase in energy expenditure would play a role in promoting a positive energy balance.³⁸

In this study, the reported number of LND foods did not relate to sex-age–adjusted body mass index. Bandini et al³⁴ also found that obese adolescents did not report more energy from LND foods relative to nonobese adolescents. Others have similarly found no association between reported body mass index and added sugar intake.^13,17 Conversely, Ludwig et al³⁹ found consumption of sweetened drinks to be related to obesity in children. The observed lack of an association between body mass index and LND food intake may partly be attributable to dieting by overweight respondents as well as measurement error (discussed in the next paragraph).

All methods of dietary assessment including the 24-hour dietary recall used in NHANES III are subject to measurement errors.⁴⁰ Data from a 24-hour recall are believed to be suitable for estimating usual dietary intakes of groups, but not individuals.²⁴ Prevalence of micronutrient inadequacy in particular is likely to be overestimated in 24-hour recall data because of intraindividual variation in food intake.⁴¹ Therefore, this study makes no claims to estimating prevalence of marginal nutrient intakes in association with LND reporting. It is preferable to interpret the results presented in Table 4 as indicating that 8- to 18-year-old Americans grouped in the highest third of reported number of LND foods consumed reported lower amounts of nutrient-dense foods and were more likely to report less than the estimated average requirement of several nutrients on the recall day. Dietary measurement error often contributes to the inability to find significant associations between dietary exposures such as intake of LND foods and body mass index or serum analyte concentrations.⁴² As mentioned already, a positive association of dietary vitamin intake (ascorbate, folate, and carotenoids) with serum concentrations of the respective analytes in the present study increases confidence in the dietary data.

Underreporting of food intake is one of the sources of measurement error in dietary assessment.⁴⁰ However, the nature, extent, or determinants of underreporting in children and adolescents are poorly understood. In the present study, a ratio of reported energy intake to estimated basal energy expenditure was calculated as a measure of reporting status and found to be positively related to LND food intake (P for trend, <.001). Finally, there may be differential misreporting of LND foods because of the social desirability bias. However, it is noteworthy that in adults in the NHANES III, ratio of reported energy intake to energy expenditure for basal needs related positively to reporting of both nutrient-dense and LND foods.⁴³ Nevertheless, there is a possibility that estimates of reported LND food intake from the present study may be lower than actual intakes.

The results indicate that LND foods are widely consumed by American children and adolescents, and suggest opportunities for improving nutrient intake profiles by interventions to moderate LND food intake. In my opinion, the needed interventions to change this behavior must be from several directions. Consistent with the philosophy of not labeling foods "good" or "bad," the food guide pyramid groups many LND foods (eg, desserts and salty snacks) as part of the dairy or grain groups.⁴⁴ The extent to which most consumers comprehend and implement the recommended fat and sugar substitutions necessary to include foods such as baked or dairy desserts in the diet is not known. Nevertheless, simplification of dietary guidance messages regarding inclusion of LND food in the diet may be useful. Second, interventions need to target families and children. A recent study⁴⁵ suggested that family-based interventions that focused on increasing fruit and vegetable intake were also effective in reducing the intake of LND foods for parents and children. Practitioners may find the strategy of emphasizing nutrient-dense foods useful for moderating the intake of LND foods, with a resultant decrease in energy intakes and improvement of intake of critical micronutrients. However, interventions targeting children and adolescents to lower LND food intake may need a more creative approach; there is preliminary evidence to suggest that children perceive foods labeled "healthy" as not being "tasty."⁴⁶ Because taste is a major determinant of food consumption,⁴⁷ nutrition messages for children and adolescents have to promote the nutrient-dense foods with a different marketing angle. Third, as has been suggested elsewhere, marketing of soft drinks, candies, and prepared baked or fried snacks in school cafeterias and vending machines should be discouraged and alternate foods including milk and fruit should be promoted.^7,48 Finally, the food industry should be a partner in these interventions by avoiding aggressive marketing of food products without nutritionally redeeming qualities.

In conclusion, the results suggest that LND food mentions by 8- to 18-year-old Americans were accompanied by adverse food and nutrient profiles and higher energy intake.

Article

Corresponding author and reprints: Ashima K. Kant, PhD, Department of Family, Nutrition, and Exercise Sciences, Remsen Hall, Room 306E, Queens College of the City University of New York, Flushing, NY 11367 (e-mail: ashima_kant@qc.edu).

Accepted for publication November 7, 2002.

This study was supported in part by award NYR-00-35200-9067 from the US Department of Agriculture–Cooperative State, Research, Education, and Extension Service–National Research Initiative Competitive Grants Program, Washington, DC.

This study was presented in part at the Experimental Biology Meetings, New Orleans, La, April 21, 2002.

I thank Lisa L. Kahle for expert programming support; and Barry I. Graubard, PhD, National Cancer Institute, National Institutes of Health, Bethesda, Md, and Daniel F. Heitjan, PhD, Division of Biostatistics, University of Pennsylvania, Philadelphia, for statistical consultation.

There has been no systematic, comprehensive study of the extent of contribution of LND foods to the diets of American children and adolescents or the nutritional and health implications, if any, of consumption of such foods. Nearly a third of the total daily energy in the diets of 8- to 18-year-old American children and adolescents was contributed by LND foods (sweeteners, beverages, desserts, discretionary fats, and salty snacks). Increasing mentions of LND foods on the recall day were associated with higher energy intake but lower amounts of nutrient-dense foods and high-risk micronutrients.