点击显示 收起
1 From the Division of Epidemiology School of Public Health, University of Minnesota, Minneapolis.
2 Supported by National Institute of Diabetes and Digestive and Kidney Diseases grant IP30-DK540456 (to A Levine, principal investigator). 3 Address reprint requests to LJ Harnack, University of Minnesota, Division of Epidemiology, 1300 South 2nd Street, Suite 300, Minneapolis, MN 55454-1015. E-mail: harnack{at}epivax.epi.umn.edu.
ABSTRACT
Background: The causes of recent increases in the prevalence of overweight in the United States are perplexing because national survey data do not show clear patterns of change in energy intake or expenditure.
Objective: Ecologic data regarding the quantities and types of foods and nutrients available in the United States between 1970 and 1998 were reviewed to provide an alternative perspective on trends in energy intake.
Design: Literature searches in agriculture, business, and medical library databases were conducted to identify data regarding the quantities and types of foods and nutrients available in the United States between 1970 and 1998.
Results: Per capita energy availability estimates from the US Department of Agriculture, US Food Supply Series, indicate that energy availability increased by 15% between 1970 and 1994. Data regarding trends in food purchasing and preparation suggest that Americans are eating more meals outside the home, relying more heavily on convenience foods, and consuming larger food portions. Americans appear, in some cases, to be selecting lower-fat foods over higher-fat alternatives. In addition, availability and sales of reduced-energy and reduced-fat products have increased.
Conclusion: Consistent with trends in overweight, most of the ecologic data reviewed suggest that energy intake has increased over the past several decades and is likely a major contributor to increases in average body weight.
Key Words: Energy intake • obesity • ecologic data • surveys • US Food Supply Series • United States • overweight
INTRODUCTION
Over the past 20 y, the prevalence of overweight in the United States has increased steadily. Data from the second and third US National Health and Nutrition Examination Surveys conducted between 1976 and 1980 (NHANES II) and 1988 and 1994 (NHANES III) indicate that the prevalence of overweight has increased in all age groups (Table 1) and in both sexes (1). In addition, increases in overweight were found for white, black, and Hispanic Americans.
View this table:
TABLE 1.. Prevalence of overweight by age—United States, second (1976–1980) and third (1988–1994) National Health and Nutrition Examination Surveys (NHANES)1
The increasing prevalence of overweight is a major public health concern because obesity is associated with a host of adverse health outcomes (2, 3). Public health interventions that target factors contributing to the rising rates of overweight are needed; however, it is unclear what factors are primarily responsible for the epidemic of overweight in America. Body fat stores are a function of energy intake and energy expenditure. Thus, the increasing prevalence of overweight in the United States could be the result of an upward shift in energy intake, a downward shift in energy expenditure, or both. Although national data on trends in energy expenditure from physical activity are sparse, available data suggest that there has been little change in energy expenditure from recreational physical activity over the past 10 y (4–6). Research on trends in energy intake over similar periods has given mixed results regarding whether the increasing prevalence of overweight in the United States could be attributable to a shift in energy intake. Data from 2 national dietary surveys indicate that energy intake has been declining (7, 8), whereas data from 1 survey suggest it has been increasing (9).
Trends in energy intake found in each of the national dietary surveys could be artifacts of inherent methodologic shortcomings. For each of the surveys, the methods used to collect dietary information and derive nutrient estimates varied between survey years. Consequently, trends in energy intake detected by these surveys may be the result of differences in the methods used rather than a reflection of real changes in food or nutrient intake. The reliance of each of the national dietary surveys on self-reported dietary information presents another opportunity for bias in the results because underreporting of foods consumed is a well-documented problem with self-reported dietary information (10, 11), particularly among overweight individuals (12–15).
Individual-based information, such as that collected through national dietary surveys, is generally preferred over ecologic or population-based data because of the methodologic limitations of ecologic data. However, in consideration of the conflicting findings and methodologic shortcomings of the national dietary data currently available at the individual level, available ecologic data regarding the quantities and types of foods available to Americans over time may serve as a valuable supplement to trend information reported in national dietary surveys. To provide additional information on trends in energy intake in the United States, available ecologic data regarding the quantities and types of foods and nutrients available in the United States between 1970 and 1998 are reviewed in this article.
METHODS
Computer-assisted searches of English-language publications in the MEDLINE (National Library of Medicine, Bethesda, MD), AGRICOLA (National Agricultural Library, Beltsville, MD), and GENERAL BUSINESS FILE (Information Access Co, Wayne, PA) databases were conducted to identify relevant literature. References cited in related publications were also reviewed. The literature search was restricted to research reporting on trends in the quantities and types of foods and nutrients available in the United States between 1970 and 1998. Information on trends in food purchasing, food preparation, and eating habits between these years were also reviewed.
Sources of information identified in the literature search included research conducted by public institutions, such as government agencies, and private organizations, such as food companies. The US Department of Agriculture (USDA) US Food Supply Series was a major source of data identified and used (16). The US Food Supply Series provides measures of per capita availability of several hundred food commodities and per capita per day nutrient availability estimates. The most recent publication providing US Food Supply Series data available since 1970 (16) was reviewed to identify per capita per day energy and macronutrient availability estimates and per capita per day availability estimates for the major food-commodity categories. Estimates of percentage changes were calculated by the investigators on the basis of data provided in the report so that the magnitude of change could be compared across food-commodity categories and nutrients.
To estimate per capita food commodities and nutrient availabilities, the total available supply of a food commodity is first calculated by summing production, imports, and beginning-of-the-year inventories. Nonfood uses (eg, exports, industrial uses, and end-of-the-year inventories) are then subtracted from this total to derive the amount of food available for human consumption. This residual amount is generally referred to as "food disappearance." Annual per capita availability for a commodity is calculated by dividing the total food disappearance estimate by the total US population on July 1. Per capita per day nutrient availability is subsequently calculated from per capita availability data (16). Nutrient availability is calculated by multiplying the amount (in kilograms per capita per year) of each food used (disappearance) by the amount of each of the 24 nutrients or by the amount of food energy contained in the edible portion of the food. The nutrient contents of all foods is then totaled for each nutrient and converted to an amount per capita per day (17).
Because per capita availability data represent the amount of food available for potential consumption rather than the amount of food ingested, estimates of per capita availability are usually overestimates of actual consumption, ie, food that is wasted or that spoils during the marketing process or in the home (eg, food that is thrown away or given to pets) is included in the estimates. Thus, food-disappearance data are not useful for estimating absolute intake. Food-disappearance data may be useful as an indicator of trends in consumption over time, however, assuming that food spoilage and waste remain constant over time. Similar to food availability data, per capita nutrient availability estimates are usually overestimates of actual intake because nutrients that may be lost during processing, marketing, or cooking, or as plate waste, are included in per capita nutrient estimates. Per capita nutrient availability estimates may also be useful for assessing trends in nutrient intake over time, however.
RESULTS
Trends in food and nutrient availability: US Food Supply Series, 1970–1995
Per capita availability of major food commodities in the United States during select periods between 1970 and 1995 are presented in Table 2 (16). The availability of some commodities increased whereas the availability of others declined during these periods.
View this table:
TABLE 2.. Average annual per capita availability of select food commodities in the US food supply and percentage change, 1970–1974 to 19951
Meat and proteins
US Food Supply Series data indicate that the per capita availability of meats that tend to contain lower amounts of fat has increased, whereas the availability of meats that tend to contain higher amounts of fat has declined (Table 2). The annual per capita amount of red meat (beef, pork, veal, and lamb) available for consumption decreased by 11.9% between 1970 and 1995. Similarly, annual per capita availability of eggs decreased by 21.1% between those years. In contrast, the annual per capita availability of poultry and fish and shellfish increased by 84.5% and 23.1%, respectively, in the same period.
Fats and oils
Annual per capita availability of fats and oils (all types) increased by 21.8% from an average of 23.9 kg between 1970 and 1974 to 29.1 kg in 1995. This increase was not uniform across all types of fats and oils (Table 2). More specifically, the per capita availability of butter and margarine declined, whereas the availability of shortening and salad and cooking oils increased significantly. Note that the apparent increase in per capita consumption of fats and oils may not accurately reflect trends in actual use. A major reason for the increasing trend in use of fats and oils was the increase in the number of fast-food restaurants and other away-from-home eating establishments, where much of the food is prepared by deep-fat frying, over the 2 decades. A 1993 study estimated that 50% of deep-frying fat used in food-service operations is discarded after use and is not available for consumption (18). A 1997 study indicated that used frying fat disposed of by restaurants amounted to 2.7 kg per capita, or 9% of the 1995 total per capita disappearance of fats and oils (16). In a study conducted to compare per capita food consumption with federal dietary recommendations depicted in the food guide pyramid, a technique was developed to adjust food-disappearance data from the US Food Supply Series for food losses throughout the marketing system and the home, including disposal of frying oils by restaurants (19). In this study, per capita availability of added fats and oils increased from 49 to 60 g between 1970–1975 and 1996, after adjustment for food losses, suggesting that the increase in per capita fat and oil availability may not entirely be an artifact of increased use of oil for frying.
Dairy products
Trends in the availability of specific types of dairy products varied greatly between 1970 and 1995 (Table 2). Most notably, the per capita availability of whole milk dropped and that of 2%-fat, 1%-fat, and nonfat milk increased precipitously. Interestingly, the availability of cheese, a generally higher-fat dairy product, increased by 111.6% between 1970–74 and 1995.
Fruit and vegetables
Per capita availability of most types of fruit and all categories of vegetables increased between 1970 and 1995 (Table 2). Availability of fresh, frozen, and processed vegetables; fresh and frozen fruit; and fruit juices increased, although availability of canned fruit declined
Flour and cereal products
Annual per capita availability of all types of flour and cereal products increased significantly between 1970 and 1995 (Table 2). The greatest increases were seen for rice, corn, and oat products (data not shown).
Sweeteners
Per capita availability of both high-energy-containing and low-energy-containing sweeteners has increased according to US Food Supply Series data (Table 2). The amount of refined sugar available in high-energy-containing sweeteners per capita decreased moderately, whereas the availability of corn sweeteners increased dramatically. The amount of low-energy-containing sweeteners available per capita doubled between 1970–1974 and 1980–1984, perhaps reflecting the increased number of low-energy-containing sweeteners available in the marketplace and an acceptance of these products by consumers. The simultaneous increase in high-energy-containing and low-energy-containing sweeteners suggests that low-energy-containing sweeteners are perhaps being consumed in addition to rather than in place of high-energy-containing sweeteners.
Per capita energy availability
US Food Supply Series data indicate that the amount of energy available per capita increased steadily between 1970 and 1995 (Table 3). Per capita per day availability of food energy was 13.8 J in 1970 and 15.9 J in 1994, a 15.2% increase. The increasing overestimation of the availability of fats and oils because of increases in the use of frying oil may, in part, explain the increase in the amount of energy available per capita. The increase is unlikely to be entirely attributable to this overestimation, however, given that increases were seen in per capita availability of all macronutrients. The greatest increases in per capita availability of macronutrients were seen for carbohydrate and protein, which resulted in a decrease in the percentage of energy from fat from 42% in 1970 to 38% in 1994.
View this table:
TABLE 3.. Daily per capita availability of food energy and macronutrients in the US food supply and percentage change, 1970–1994
Trends in food purchasing and preparation
Over the past several decades, dramatic changes in food purchasing and food preparation trends have occurred in the United States, including an increase in the number of meals eaten away from home and a decrease in time spent in meal preparation at home. Overall, households spent 38% of their food dollar on food eaten away from home in 1992. By comparison, in the 1970s 20% of the food dollar was spent on food eaten away from home (20). This trend in meals eaten away from home shows no sign of weakening. In 1993 Americans spent $197.8 billion on meals eaten away from home, an increase of 9% above that in 1992 (20). Currently, the restaurant industry as a whole is growing at a modest rate of 3% annually, whereas fast-food restaurants are growing at a rate of 7% annually (21). The proliferation of fast-food restaurants is evident in most parts of the country, there being an average of 1.02 fast-food hamburger establishments/10000 residents in the United States (22). The implications of more meals being eaten away from home on the nutritional quality of the diet are likely unfavorable. Although healthful food choices can be found at fast-food restaurants, most foods available tend to contain undesirable amounts of energy and fat (23, 24). A survey conducted by Restaurants & Institutions in 1994 found that consumers expressed increased interested in more healthful menu options, such as salads, grilled chicken, and fish; salads were found to be the second-most ordered food in restaurants (25). Of interest, hamburgers were found to be the most frequently ordered item and French-fries were the third-most ordered item.
Aside from eating more meals away from home, Americans appear to be spending less time preparing meals that are consumed at home (21, 26). In 1987, 43% of all meals included at least one item made from scratch; in 1997, this value dropped to 38% (21). To meet consumer demands for foods that require minimal preparation time, supermarkets are increasing their selection of prepared foods. Along with offering convenience foods such as frozen dinners, supermarkets are offering precooked meals, such as rotisserie chicken, to better compete with restaurants (21).
Availability and sales of reduced-energy and reduced-fat products
In response to consumer interest in reducing dietary fat and energy intakes, a plethora of reduced-energy and reduced-fat food products has been introduced to the marketplace (27). Consumer acceptance of these products appears to be high. An estimated three-fourths of adults regularly consume reduced-fat foods (28). A study conducted to evaluate the size and growth of the market for nutritionally improved products between 1989 and 1993 used supermarket scanner data to track the growth in sales of these products relative to that of their traditional counterparts. Sales of nutritionally improved products, including reduced-energy and reduced-fat products, grew faster than did sales of traditional counterparts of these foods in the supermarkets included in this study (29).
Health concerns and a desire to reduce fat and energy intakes are the primary reasons consumers give for choosing reduced-energy and reduced-fat products (28). However, whether intakes of these foods actually lead to reduced fat and energy intakes has been questioned (30, 31). Anecdotal observations suggest that consumers may not use reduced-energy and reduced-fat products as one-to-one substitutes for their traditional counterparts. It is possible that reduced-energy and reduced-fat products are being consumed in addition to rather than in place of other foods, thereby resulting in an increase rather than a decrease in energy and fat intakes. In addition, larger portions of reduced-energy and reduced-fat products are perhaps being consumed, thereby nullifying any beneficial effect of these products. Results of ongoing population-based studies evaluating the effects of reduced-energy and reduced-fat products on body weight and energy and fat intakes are not yet available. Thus, it is difficult to infer what the influence of increased availability of these products on trends in energy and fat intakes might be.
Changes in food-portion sizes
Anecdotal evidence suggests that food-portion sizes have been on the rise in recent years and may be increasing energy intakes (32–36). Scant data are available, however, regarding actual trends in portion sizes and potential implications of larger portions on food and nutrient intakes. The National Restaurant Association conducted a study in which menus collected in 1988 from 66 restaurants were compared with menus collected in 1993 from the same restaurants. The number of menus offering entrées with more than one portion size, such as "queen-size" and "king-size" steak, increased by 12% during this period (37). Further research is warranted to determine whether portion sizes are changing and to examine the effect of such changes on energy intakes.
DISCUSSION
Findings from the USDA 1977–1978 Nationwide Food Consumption Survey and the USDA 1994 Continuing Survey of Food Intakes of Individuals indicate that, on average, energy intakes per person were 7% lower in 1994 than in 1977–1978 (7). Dietary data from the 1987 and 1992 Cancer Epidemiology Supplements to the National Health Interview Survey suggest that energy intakes in US adults declined between 1987 and 1992. On average, 423 kJ less energy per person was consumed in 1992 than in 1987 (8). In contrast with findings from the USDA and National Health Interview Survey dietary surveys, NHANES II and III data indicate an increase in mean energy intakes in adolescents and adults between 1976–1980 and 1988–1991. Mean energy intakes were 423–1255 kJ higher in adolescents and adults according to NHANES III compared with NHANES II data (38).
Trends in energy and fat intakes found in each of the national dietary surveys could be artifacts of methodologic shortcomings inherent to each. For each of the surveys, the method used to collect dietary information and derive nutrient estimates varied between survey years. Consequently, trends in energy intakes indicated by each of these surveys may be the result of differences in methods rather than of real changes in food and nutrient intakes. In particular, procedural changes between the time that the NHANES II and NHANES III data were collected threaten the comparability of data from these surveys. An automated data-collection system was introduced by the time NHANES III was conducted. In addition, different nutrient databases were used to conduct NHANES II and NHANES III. The effects of these differences between NHANES II and III have not yet been examined (39).
The reliance of each of the national dietary surveys on self-reported dietary information presents another opportunity for bias in the results. Underreporting of foods consumed is a well-documented bias of self-reported dietary information (10–15, 40). This phenomenon is likely attributable, in large part, to social desirability bias in food reporting. Worsley et al (11) found that reported intakes of certain foods (fresh fruit, vegetables, and sweet foods) are either over- or underreported depending on the perceived "healthfulness" of the food item. Prentice et al (15) found that obese subjects were more apt than were nonobese subjects to underreport food intake, presumably to provide a more socially desirable response. It is likely that an increasing proportion of Americans are apt to underreport food intake because interest in nutrition has increased over the past several decades with the recognition of diet as a major contributor to morbidity and mortality in the United States (41). A disproportionate underreporting of food intake between the periods of different dietary surveys could compromise the validity of trends in energy intake observed. More specifically, the decrease in energy intakes noted in 2 of the national dietary surveys (7, 8) could be attributable to a greater proportion of Americans underreporting dietary intake in later survey years rather than to an actual change in energy intakes in the population.
Another potential methodologic limitation of national dietary survey data relates to the questionable validity of the visual aids used to estimate food-portion sizes. Anecdotal evidence suggests that food-portion sizes have been on the rise (32–36). Because individuals appear to have greater difficulty estimating food-portion sizes as the size of the portion increases (32, 42–44), it is possible that portion size estimation and, in turn, food and nutrient intake estimates are becoming less valid over time as customary portions of food consumed increase.
Unlike individual-based dietary information, ecologic data are not prone to social desirability bias. In addition, some available ecologic data on food supplies were collected annually and thus provide more data points for evaluating trends in food and nutrient intakes. As with all ecologic data, however, it is possible that bias may occur because an association observed between variables on a group level does not necessarily represent an association that exists at an individual level.
Most of the ecologic data reviewed suggest that energy intakes have been increasing over the past several decades. Estimates of per capita energy availability from the US Food Supply Series indicate that energy intakes increased by 15% between 1970 and 1994. Increases in per capita availability were seen for most food commodities. Data regarding trends in food purchasing and food preparation suggest that Americans are eating out more often and are relying more heavily on convenience foods. Portion sizes may be increasing, which could reflect increases in energy intake.
The upward shift in energy intake suggested by the ecologic data reviewed is consistent with trends in overweight, which suggest that ecologic data may more closely reflect actual changes in energy intake than do national dietary survey data collected at the individual level. It must be recognized, however, that trends in overweight may be attributable to decreases in energy expenditure rather than to changes in energy intake (45).
The ecologic data reviewed present a complicated picture with regard to trends in dietary fat intake. Some of the data suggest that total fat intake has been declining since 1970. Americans appear to be selecting lower-fat foods over higher-fat alternatives, and the availability and sales of reduced-fat products have increased over the past several decades. To the contrary, however, some of the data suggest that total fat intake has increased. The availability of dietary fat and fats and oils per capita appears to have increased. In addition, Americans are eating out more often and the greatest growth in meals eaten away from home appears to be in fast-food restaurants, where menu items are often high in fat content.
National dietary survey (7–9) and ecologic (16) data are consistent in indicating that the percentage of total energy from fat has declined over the past several decades. Furthermore, shifts in serum lipid concentrations observed in the US population (46) are consistent with a decrease in the percentage of energy from fat. Note, however, that these shifts in serum lipids may also be attributable to changes in the types of fats consumed over the past several decades. The ecologic data suggest that intakes of saturated fat have declined while intakes of monounsaturated and polyunsaturated fats have increased. National dietary survey data suggest similar shifts in the types of fats consumed (8, 38).
Despite the limitations of ecologic studies, the data reviewed in this article serve as a valuable supplement to national dietary survey data available at the individual level. Consistent with trends in overweight, most of these data suggest that energy intakes have been increasing over the past several decades.
REFERENCES