US adolescent food intake trends from 1965 to 1996

Abstract

Objective To examine adolescent food consumption trends in the United States with important chronic disease implications. Methods Analysis of dietary intake data from 4 nationally representative US Department of Agriculture surveys of persons aged 11 to 18 years (n = 12,498). Results From 1965 to 1996, a considerable shift occurred in the adolescent diet. Total energy intake decreased, as did the proportion of energy from total fat (39%-32%) and saturated fat (15%-12%). Concurrent increases occurred in the consumption of higher-fat potatoes and mixed dishes (pizza and macaroni and cheese). Lower-fat milks replaced higher-fat milks, but total milk consumption decreased by 36%. This decrease was accompanied by an increase in the consumption of soft drinks and noncitrus juices. An increase in high-fat potato consumption led to an increase in vegetable intake, but the number of servings for fruits and vegetables is still lower than the recommended 5 per day. Iron, folic acid, and calcium intakes continue to be below those recommended for girls. Conclusions These trends, far greater than for US adults, may compromise the health of the future US population.

Adolescence is a critical period during which lifetime habits are established. Many teenagers are preoccupied by their physical characteristics and appearance,¹ resulting in nutritional practices such as dieting or consuming fad foods. Dieting, a common practice among adolescent girls, may be the cause of inadequate dietary intake and may have important health effects for selected nutrients such as calcium, iron, and folic acid.² Some teenagers continue to exercise and become more involved in sports, while others stop and eventually develop a sedentary lifestyle.³ Many of the adolescent lifestyle habits may have severe consequences in the long run. For example, being overweight during adolescence is associated with increased risk for being overweight during adulthood^4,5 and for increased incidences of hypertension during adolescence⁶ and of adult-onset diabetes during adolescence.⁷ The National Health and Nutrition Examination Surveys (NHANES) in the United States showed increasing prevalence of overweight in not only adults but also adolescents, in particular in the past 15 years.^8,9 Furthermore, we know from previous studies that blood pressures and cholesterol concentrations track from childhood into adulthood.^10,11 Excessive weight gain during childhood and adolescence is a determinant of adult cardiovascular risk and mortality.^12,13 Similarly, inadequate fruit and vegetable intake may affect the prevalence of selected cancers.¹⁴ This study was initiated to provide a better understanding of adolescent food habits and to follow food trends that may have implications for the later development of adult noncommunicable diseases.

METHODS

Survey design and sample

From the initial sample of about 90,000 persons participating in 4 US Department of Agriculture (USDA) surveys, adolescents aged 11 to 18 years with dietary data were selected for this study. Of 12,498 adolescents, 3,061 came from the 1965 and 6,208 from the 1977-1978 Nationwide Food Consumption Surveys and 1,656 came from the 1989-1991 and 1,573 from the 1994-1996 Continuing Survey of Food Intake by Individuals (CSFII). The USDA surveys from 1965 to 1991 contain stratified area probability samples of noninstitutionalized US households in the 48 coterminous states and in all 50 states in 1994 through 1996. The 1977, 1989-1991, and 1994-1996 surveyors conducted 4 data collections (winter, spring, summer, autumn), each surveying a different sample, whereas the 1965 survey collected individual dietary data in a single spring sample. These surveys were self-weighting, multistage, stratified area samples of the US population. For each survey, response rates differed at the level of the primary sampling unit. Thus, weights based on response rates for each sample unit are used to permit inferences applicable to the total noninstitutionalized US population. Detailed information on the methods pertaining to each survey has been published previously.^15,16,17,18

Dietary data

Each survey included at least 1 in-home, interviewer-administered, 24-hour dietary recall. The 1977 and 1989-1991 surveys also included 2 self-administered 1-day food records. Information was collected on everything teens ate at home or away. For our purpose of studying trends in dietary intake over time, we used only the first day of dietary intake information from each survey—the 1-day (24-hour) dietary recall. This method avoids the biasing of intake results that may occur because of the different dietary data collection methods used and different number of days of reported intake that were collected.

Two of the challenges in trends analysis are accounting for differences in how foods are defined over time and characterizing the nutrient contribution of each food. Food codes changed considerably (often for the same food), and technology to measure nutrients has also changed. We have developed several ways to link foods coded and collected in the past decade with foods all the way back to 1965. Values from the 1994 nutrient database were then applied to the 3 data sets before 1994 to provide consistently high-quality estimates of nutrient values over time. In the most recent survey, the nutrient database specific to each survey year was used to calculate nutrient values to reflect changes in foods on the market.

This approach allows the use of newer measures of nutrients and for considering shifts in food descriptors. Changes in the actual nutrient content of foods will not be captured by the use of a single nutrient composition table, and trends may be either overstated or understated. In weighing the advantages and disadvantages of using different nutrient composition measures, we concluded that using the 1994 nutrient data provides a conservative measure of some of the nutrient trends in foods (for example, fat) and allows for the examination of trends in several nutrients (for example, fiber) that would not have been possible otherwise.

The Food Guide Pyramid database released by the USDA with the 1994-1996 survey was used to calculate the servings of fruit, vegetables, milk, and grains and the ounces of lean meat equivalents for all 4 surveys.¹⁹ The same linking program used to apply the 1994 nutrient values back in time was used for these calculations.

Food grouping scheme

To examine changes in the quantity of foods consumed over time, we developed a University of North Carolina (UNC) food grouping system that aggregates the 4,000 or so foods consumed. This system separates virtually all foods in the USDA's Individual Food Consumption Surveys into 74 useful descriptive and nutrient-based groups. Initially major food groups were based on food groupings used by the USDA. Then fat and dietary fiber compositions were used to develop more refined food groupings, and nutrient thresholds were used to separate major food groups into more distinct, nutrient-based food groups (available from authors on request).

Statistical analysis

Descriptive statistics were generated for each survey year using a commercial software package (SUDAAN version 7.51; Research Triangle Institute, Research Triangle Park, NC) to appropriately correct the standard errors for sample design effects and for applying survey weights. Differences between years in the means of nutrients and servings of the Food Guide Pyramid were then tested for significance in a statistical software program (SAS version 6.12; SAS Institute, Cary, NC) using a Student t test.

Statistical difference was denoted as P < 0.001. Nutrient values were compared with those of the most recent recommendations made by the Food and Nutrition Board (recommended dietary allowances [RDA] or recommended dietary intake [RDI]) and expressed as a percentage of the recommendations specific for each age and gender grouping.²⁰

RESULTS

Table 1 shows sociodemographic information for the adolescents in the study by survey year. Important trends of interest are the increase in mean household income, the subsequently more even distribution of percentage of households considered below the poverty line starting in 1977, and the increasing number of Hispanics represented in the survey, as well as the increase in body mass index for both younger and older adolescents.

Table 1.

Sociodemographic characteristics for adolescents (aged 11-18) with dietary data from the 4 USDA surveys

Characteristics	1965	1977	1989-1991	1994-1996
No. of individuals	3,061	6,208	1,656	1,573
% of males	51.1	49.4	50.9	50.3
Race (%)*
White	82.0	76.2	74.9	66.0
Black	15.8	15.8	16.6	16.3
Hispanic	NA	7.1	8.5	12.9
Household size, mean (SEM) no.	5.58 (0.04)	5.22 (0.02)	4.43 (0.03)	4.45 (1.04)
Distribution by region (%)
Northeast	22.4	23.0	20.3	18.1
North Central	28.7	28.1	24.3	24.8
South	34.4	31.5	36.1	35.2
West	14.5	17.4	19.3	21.9
Distribution by urbanization (%)
Central city	NA†	27.7	26.8	28.4
Suburban	NA	38.5	47.2	48.8
Rural	35.9	33.8	26.0	22.8
Household income, mean (SEM) $	7,116 (85)	17,517 (159)	40,429 (531)	44,635 (712)
Distribution by percentage of poverty (%)
<185	56.3	37.0	33.0	37.1
185-350	32.7	38.3	33.6	29.8
>350	11.0	24.8	33.4	33.0
Body mass index (SEM)
11-14 year olds	19.49 (0.08)	19.89 (0.07)	20.24 (0.13)	20.74 (0.15)
15-18 year olds	21.35 (0.08)	21.40 (0.06)	22.26 (0.15)	22.76 (0.17)
The 4 USDA surveys are the Nationwide Food Consumption Survey, 1965 and 1977, and the Continuing Survey of Food Intake by Individuals, 1989-1991 and 1994-1996. NA = not available.

^*USDA began collecting data on Hispanic origin in 1977; the remaining percentage represents other minorities.

^†In 1965, USDA coded this variable as urban, rural non-farm, and rural farm. The only comparable variable we could create was rural by adding the 2 rural categories. The other 64% is thus a combination of central city and suburban.

Energy intake decreased over the years, with the greatest drop occurring from 1965 to 1977 (table 2). The data for all 4 years showed typical differences in pattern of energy intake, with boys consuming higher amounts than girls (data not shown). Table 2 also shows the proportion of energy from total fat, saturated fat, carbohydrate, and protein. A steady decline in fats and protein as a proportion of total energy intake occurred in the diet over time, which was compensated by an increase in carbohydrate intake. Gender did not affect the distribution of macro-nutrients (data not shown). Between 1989-1991 and 1994-1996, fat intake increased from 80 to 83 g per day, saturated fat intake stayed at 30 g per day, carbohydrate intake increased from 270 to 311 g per day, and protein intake remained relatively unchanged (increased from 81-82 g per day).

Table 2.

Selected nutrient intakes of 11-18 year olds with 1 day of dietary data from the 4 USDA surveys

Nutrient	1965	1977	1989-1991	1994-1996
Energy (megajoules)	9.92 (0.16)*†	8.78 (0.09)¶	8.77 (0.18)#	9.58 (0.18)
Total fat, % of energy	38.7 (0.2)*†‡	37 (0.2)§¶	34.3 (0.4)#	32.7 (0.2)
Saturated fat, % of energy	15 (0.1)*†‡	14.1 (0.1)§¶	12.9 (0.2)#	11.6 (0.1)
Carbohydrate, % of energy	46.3 (0.3)†‡	47.1 (0.2)§¶	51.4 (0.5)#	54.2 (0.3)
Protein, % of energy	16.1 (0.1)*	16.7 (0.1)§	15.4 (0.2)#	14.2 (0.1)
Vitamin A (retinol equiv)	1115 (43.1)†‡	1026 (26.8)	902 (33.0)	932 (32.3)
Vitamin C (mg)	89 (2.2)	90 (2.0)	103 (4.1)	105 (4.4)
Iron (mg)	14 (0.2)†	14 (0.2)¶	15 (0.4)#	17 (0.4)
Folate (μg)	244 (4.5)*	265 (4)	272 (7.4)	274 (8.3)
Calcium (mg)	1100 (22.1)*†‡	1009 (15.6)	974 (23.6)	960 (18.6)
Fiber (g)	13.5 (0.3)‡	13 (0.2)¶	14 (0.3)	15 (0.4)
Results expressed as mean (SEM). The 4 USDA surveys are the Nationwide Food Consumption Survey, 1965 and 1977, and the Continuing Survey of Food Intake by Individuals, 1989-1991 and 1994-1996.

^*Significant differences between 1965 and 1977, P < 0.001.

^†Significant differences between 1965 and 1989, P < 0.001.

^‡Significant differences between 1965 and 1996, P < 0.001.

^§Significant differences between 1977 and 1989, P < 0.001.

^¶Significant differences between 1977 and 1996, P < 0.001.

^#Significant differences between 1989 and 1996, P < 0.001.

Other nutrients of interest that we examined included vitamins A and C, iron, folate, calcium, and fiber. On average, only vitamin A and calcium concentrations decreased, so that in 1996 the mean percentage of the recommended intakes being met for each nutrient was 103% and 74%, respectively. The other 4 nutrient concentrations increased, and concentrations for all but folate and fiber met the most recent dietary recommendations. Some slight differences by gender existed. For example, in 1996, 40% of adolescent girls met the recommended dietary intakes for iron, 20% for calcium, and 17% for folate. For adolescent boys, 75% met the recommendations for iron, and about one third met the recommendations for calcium and folate.

The most interesting trends of food group consumption are shown in figures 1, 2, 3, 4, and 5 by gender and in table 3 for the entire age group (data for other foods are available from the authors on request). The figures reflect the per capita grams of food consumed per day using the UNC food groups. The table presents actual servings of food groups or, as in the case of meat, ounces of lean meat equivalents using the USDA's Food Guide Pyramid database. The results show a decline in milk consumption that was not compensated for by an increase in consuming other dairy products. Low-fat milk was substituted for whole milk (medium-fat source). Grain intake increased but primarily from high-fat mixed dishes such as pizza, macaroni-and-cheese, and certain ethnic foods. Raw fruit consumption declined steadily, but juice consumption increased; this accounts for the relatively stable number of servings in table 3. Figure 4 shows a decrease in the grams of dark green and orange vegetables consumed from 1965 to 1989 and an increase from 1989 to 1996. Since 1965, the grams of high-fat potatoes consumed have increased. By 1996, white potatoes accounted for half of the vegetable consumption, whereas in 1965 it accounted for only a third (table 3). Thus, the overall vegetable consumption has increased since 1965 for this population. The consumption of lean meat and eggs decreased. Finally, soft drink consumption rose sharply.

Figure 1.

Trends in milk consumption, per capita grams, among US adolescents in the US Department of Agriculture surveys, 1965-1996.

Figure 2.

Trends in consumption of grain-based mixed (BM) dishes, by per capita grams, among US adolescents in the US Department of Agriculture surveys, 1965-1996.

Figure 3.

Trends in consumption of raw fruit and juices, by per capita grams, among US adolescents in the US Department of Agriculture surveys, 1965-1996.

Figure 4.

Trends in vegetable consumption, by per capita grams, among US adolescents in the US Department of Agriculture surveys, 1965-1996.

Figure 5.

Trends in beverage consumption, by per capita grams, among US adolescents in the US Department of Agriculture surveys, 1965-1996.

Table 3.

Mean Food Guide Pyramid servings of 11-18 years olds with 1 day of dietary data from the 4 USDA surveys

Food group	1965	1977	1989-1991	1994-1996
Dairy	2.5 (0.07)‡	2.3 (0.05)¶	2.2 (0.07)	2.0 (0.05)
Grain	6.6 (0.12)*‡	6.2 (0.07)§¶	7.1 (0.20)	7.7 (0.11)
Fruit (incl. juices)	1.4 (0.06)	1.1 (0.04)	1.2 (0.07)	1.4 (0.04)
Vegetables (all)	2.7 (0.06)*‡	3.0 (0.06)	3.0 (0.10)#	3.3 (0.07)
White potato	0.8 (0.02)*†‡	1.1 (0.03)¶	1.3 (0.08)	1.5 (0.06)
Meat (ounces)
Meat, poultry, and fish**	5.6 (0.12)*†‡	5.0 (0.07)§¶	4.0 (0.10)	4.0 (0.09)
Eggs††	0.7 (0.02)*†‡	0.5 (0.01)§¶	0.3 (0.03)	0.3 (0.02)
Nuts and seeds‡‡	0.3 (0.03)*†‡	0.1 (0.01)	0.1 (0.02)	0.2 (0.01)
Results expressed as mean (SEM).
The four USDA surveys are the Nationwide Food Consumption Survey 1965 and 1977, and the Continuing Survey of Food Intake by Individuals 1989-1991 and 1994-1996.

^*Significant differences between 1965 and 1977, P < 0.001.

^†Significant differences between 1965 and 1989, P < 0.001.

^‡Significant differences between 1965 and 1996, P < 0.001.

^§Significant differences between 1977 and 1989, P < 0.001.

^¶Significant differences between 1977 and 1996, P < 0.001.

^#Significant differences between 1989 and 1996, P < 0.001.

^**Ounces of cooked lean meat equivalents from beef, pork, veal, lamb, game; frankfurters, sausages, and luncheon meats; poultry; and fish and shellfish.

^††One egg is equivalent to 1 oz of cooked lean meat. Includes eggs and egg substitutes.

^‡‡One-third cup of nuts and one-fourth cup of seeds is equivalent to 1 oz of cooked lean meat.

DISCUSSION

This is the first study to examine dietary trends among US adolescents that uses actual dietary intake data that are comparable from a nationally representative sample over 3 decades. The small sample size for non-Hispanic African American and Hispanic subgroups forced us to focus on the total adolescent population. Our results show a decrease in total energy intake and total fat over the period studied, the latter of which is positive, but also a decreased consumption of raw fruits, nonpotato sources of vegetables, and calcium-rich dairy sources. Soft drink consumption increased greatly. Many national surveys that focus on adolescents, such as the Youth Risk Behavior Surveillance System (funded by the Centers for Disease Control and Prevention) and the National Longitudinal Survey of Adolescent Health (funded by the National Institutes of Health) lack the appropriate dietary data to quantify these trends.

The total energy intake observed between 1965 and 1996 was below recommendations and decreased with time in each age and gender group. Overall, energy intake declined 17% during this 30-year period. In the last period between 1989-1991 and 1994-1996, mean energy intake increased by 9%. Future surveys need to show if this is a real trend or related to the major methodologic changes between these 2 surveys. Both the NHANES and CSFII reported this increased intake, but both changed their approaches to provide more in-depth probing for dietary intake to avoid some of the undercounting found in the past. The decreased energy intake seems counter-intuitive when other studies have shown a rising prevalence of overweight and obesity among adolescents.^8,9,21

The most likely explanation, apart from the 9% increase in total energy intake from 1989 to 1996, is a concomitant decrease in energy expenditure caused by a decrease in physical activity.³ Most children as they age decrease their physical activity.^3,22,23 Another explanation is differential underreporting by year of the survey, with more underreporting occurring in 1996 than in 1965. However, with each survey, the method for collecting the dietary information improved; in the latest survey, more probing and a multipass approach were used to decrease the amount of underreporting.²⁴ Therefore, more underreporting may have occurred in 1965 than in 1996, which would have led to underestimating the difference in energy intake.

As with older Americans, adolescents still had a higher percentage of energy intake from fat than the current dietary guidelines recommend.^25,26 The proportion of energy coming from carbohydrate and protein was within recommendations. In terms of the absolute intake of food, however, we found that between 1989 and 1996, total fat intake increased by 4%, saturated fat intake remained stable, carbohydrate intake increased by 15%, and protein intake increased by 1%. Intakes of fiber (the recommendation being more than 10 g per 4.18 megajoules),²⁷ folic acid, and calcium for the entire age group and iron for girls were lower than optimum for proper growth and development during adolescence.

Several of these trends combined create important nutritional problems. Inadequate fiber consumption and decreasing consumption of raw fruit and nonpotato vegetables are major concerns. Others in the field have substantiated these findings for this age group and for younger ages.^28,29 Adolescents do not consume other sources of fiber, such as whole grains, pasta, rice, bread, and high-fiber cereals, in sufficient quantities to make a difference in the total fiber intake. Thus, they are at risk of chronic diseases because of low intake of not only fiber but also of antioxidant and nutrient sources of plant foods that may serve as protective factors for certain cancers.¹⁴.

Another example of a potential nutritional problem is the decreased consumption of milk, which has not been replaced by other rich sources of calcium but rather by soft drinks and noncitrus juices and drinks. Both adolescent boys and girls have decreased their milk consumption, but this is of greater concern for girls, who may be at a higher risk for developing osteoporosis later in life. Adolescence is the optimal period to increase bone density.². This trend, coupled with the low consumption of dark green and orange vegetables among adolescents, may be more problematic than once thought. Preliminary findings suggest that this dietary behavior leads to increased phosphorus excretion, which subsequently may compromise bone building and maintenance.³⁰ All age groups are susceptible to dental caries and hyperlipidemia, 2 nutritional consequences of increased simple carbohydrate intake generated by increased soft drink consumption.³¹

Another area of concern is the low intake of dietary folate and iron for adolescent girls. If adolescents with low intakes become pregnant, their risks of iron deficiency anemia and of having an infant with neural tube defects are increased. More nutrition intervention strategies are needed to improve the intakes of these nutrients. Recommending increased supplement use and fortification are key options.

CONCLUSION

Food consumption trends are compromising the nutrition and health of US adolescents and may contribute to important increases in nutrition-related chronic diseases. Given our increasing understanding of the role that diet and body composition play in preventing disease and promoting a higher quality of life, these results provide a clearer basis for intervention during adolescence to instill proper dietary habits.

Figure 1.

Staples of the diet of American adolescents do not meet their nutritional needs

^{© PRNewsFoto/Burger King}