Nutrition Facts - Diet And Nutrition - Free Diet Plans

Let the Pyramid Guide Your Food Choices: Capturing the Total Diet Concept Part II

June 28th, 2005

By Lori Beth Dixon, Frances J. Cronin and Susan M. Krebs-Smith

ASSESSING THE TOTAL DIET: MEASURES AND RESULTS

Methods have been developed to measure variety, moderation, and proportionality in the total diet in order to assess the diets of the U.S. population. Like food guidance, the development of such methods is evolutionary, with more recent methods attempting to incorporate the strengths and address the limitations of the previous methods.

Variety measures

Initial attempts at capturing the total diet were related to the original guideline, “Eat a variety of foods,” and involved measuring dietary variety in different ways. Dietary variety, or dietary diversity as it is sometimes called, refers to the inclusion of different items in the diet. Krebs-Smith measured variety among and within food groups from 24-h recall data in the 1977–1978 Nationwide Food Consumption Survey (NFCS). They assessed whether these variety measures were related to variety’s purported benefits of dietary quality, specifically nutrient adequacy and the intakes of energy, fat, sugar, cholesterol and sodium. The authors controlled for the number of foods to measure the effects of variety per se, apart from the additional food that a varied diet often entails. Variety among the major Pyramid food groups (counted as the number of groups present in the diet) explained as much variation in nutrient adequacy as variety within those groups (counted as the number of subgroups or as the number of individual foods within the groups). Neither type of variety was related to intakes of energy, fat, sugar, sodium or cholesterol. This study suggested that variety might best be interpreted as choosing the recommended number of servings from each group. Telling consumers this directly might be an improvement over the ambiguities of “Eat a variety of foods.” The finding that there is little nutritional advantage to increasing variety within the food groups is consistent with the fact that certain foods within each group are more nutrient dense than others. This suggests that choosing foods simply because they are different from one another may not be as important as selectively including more of some foods and less of others.

Other studies have examined dietary variety or diversity among food groups. Kant created two different scores from 24-h recall data in the Second National Health and Nutrition Examination Survey (NHANES II) to measure dietary diversity. A food group score gave a maximum score of 5 to adults who reported at least one food, above a minimum gram weight amount, from each of five food groups (dairy, meat, grain, fruits, vegetables). A serving score gave a maximum score of 20 to adults who reported at least two servings each from dairy, meat, fruit, and vegetable groups and four servings from the grain group. Serving sizes were based on the median gram weights of each food reported. Using the food group score of dietary diversity, only one third of the adult population consumed foods from all five food groups on a given day. Using the serving score of dietary diversity, <3% consumed the designated minimum number of servings from all food groups on a given day.

Researchers have also examined whether variety among food groups is linked to health outcomes. Using the 5-point food group score from 24-h recall data in the NHANES I Epidemiologic Follow-Up Study, Kant reported an association, among adults, between the consumption of two or fewer food groups and increased mortality. Lower scores, from similar scoring systems using food-frequency data, have also been associated with increased risk of mortality or chronic disease.

A recent study measured within food group variation using food frequency data from 71 healthy adult men and women. The authors examined whether this type of dietary variety was related to energy intake and body fatness. In this study, variety was defined as “the percentage of different food types within each of 10 food groups, regardless of the frequency with which they were consumed.” Results showed increased variation within all food groups to be positively associated with energy intake. Increased variation within certain food groups (e.g., sweets, snacks, condiments, entrees and carbohydrates) was positively associated with body fatness, but increased variation within the vegetable group was negatively associated. Variety ratios of the vegetable group to other food groups were negatively associated with body fatness, even after adjustment for dietary fat.
Dietary variety among or within food groups is related to the total number of foods in the diet (i.e., increases in variety will necessitate increases in food intake), and the quantity of foods is related to nutrient adequacy (i.e., more foods will increase nutrient intakes). Accordingly, many studies of variety have controlled for the number of foods and/or energy intake. However, variety scores do not take into account portion size, beyond a minimum amount, or otherwise quantify the intake of each group. Although variety captures the presence or absence of different food groups, it does not capture them as true “dimensions” of the diet because their quantities are not assessed. Furthermore, the variety score tallies only the number of food groups but does not indicate which groups are included. For example, a score of 3 merely indicates the presence of three food groups, not which groups they are or how much of any group is present. Last, these scores are based on the assumption that eating foods from all food groups is preferable and are biased against diets that do not include certain classes of foods such as meat or milk, which could also be nutritionally adequate.

Quantifying the individual dimensions of the diet

Quantification of the different dimensions of the diet became possible with the development of the Pyramid Servings Database (PSDB) for use with the USDA’s Continuing Survey of Food Intakes by Individuals (CSFII) . This database provides the number of servings of each of the Pyramid’s major food groups and subgroups, and the amounts of discretionary fat and added sugars contained in 100 g of every food mentioned in the survey. It required the development of a recipe file to disaggregate food mixtures into their component ingredients or foods before assigning the components to food groups. Because 75% of the foods reported in the 1989–1991 CSFII were mixtures, this task was especially challenging and required many difficult decisions as to the level of disaggregation and the choice of appropriate food groups. In addition, food intakes were reported in grams and converted to serving sizes, and new variables were developed to quantify discretionary fat and added sugars. In effect, the PSDB’s exacting system of operationalizing the recommendations of the Pyramid brings to light many of the ideas that were emphasized in the original food guidance system and the brochure, but lost in the widespread use of the graphic icon.

Intakes of all major food groups and subgroups, along with the percentage of energy from total fat, discretionary fat and added sugars were assessed by combining dietary data from both the 1994–1996 CSFII and 1989–1991 CSFII with the PSDB. Intakes were compared to individual-specific recommendations determined from the sample patterns in the Pyramid. The dairy recommendation was based on age and physiologic status; recommendations for other food groups were based on energy needs. In both the 1994–1996 and 1989–1991 surveys, mean intakes of males were more in accordance with Pyramid recommendations than mean intakes of females. For example, mean intakes of males >=12 y old met the minimum grain, vegetable and meat recommendations, and mean intakes of males ages 6–19 y met the minimum dairy recommendations. However, with the exception of 2- to 5-y-old males in 1994–1996, mean intakes of all age groups of males failed to meet the minimum fruit recommendation. In both surveys, mean intakes of very few age groups of females met the minimum recommendations for any food group. In 1994–1996, mean intakes of females ages 2–19 y met the minimum grain recommendation, mean intakes of females >=20 y old met the minimum vegetable recommendation, and mean intakes of females ages 2–5 y met the minimum fruit recommendation. Mean intakes of all other age groups of females were below the minimum recommendations for all food groups. In 1989–1991, with the exception of 6–11 y-old females with mean intakes above the minimum dairy recommendation, mean intakes of all ages of females were below the minimum recommendations for all food groups.

For all individuals, whole grains comprised between 15 and 17% of the grain intake, and dark green and deep yellow vegetables comprised between 10 and 12% of the vegetable intake, noticeably less than amounts intended by the food guidance system. Mean intakes of total fat, discretionary fat and added sugars exceeded Pyramid recommendations. Among all age groups, the percentage of energy from total fat decreased from 1989–1991 to 1994–1996. However, the percentage of energy from discretionary fat remained essentially the same, whereas the percentage of energy from added sugars increased.

Differences emerge when food group intakes are compared among groups of individuals defined by socioeconomic status (e.g., years of education of the head of household or income according to the federal poverty threshold adjusted for household size). According to 1994–1996 CSFII data for adults >=20 y old, mean numbers of servings from the grain, fruit, vegetable and milk intakes were highest among the high education (>high school) and high income (>350% of poverty) groups. Mean numbers of servings from meat intake were generally lower in the higher socioeconomic groups. Intakes of discretionary fat did not differ by education or income. Added sugars were highest among the middle income group (131–350% of poverty) and tended to decline with education.

Patterning techniques

Measures of patterns of Pyramid food group intakes have built upon earlier measures of variety. For example, Kant evaluated 24-h recall data in the NHANES II for the presence or absence, according to a minimum gram weight amount, of five food groups (dairy, meat, grain, fruit, vegetable). Each adult’s pattern was then determined on the basis of whether the individual did/did not report consuming at least a minimum amount of food from each of these groups. Two categories for each of five food groups generated 32 possible patterns. The most prevalent patterns among the population were as follows: all food groups present (34%), no fruit (24%), no dairy or fruit (9%), no dairy (8%), and no fruit or vegetable (6%). The pattern with all food groups was associated with mean nutrient intakes above the RDA. This pattern also had the lowest proportion of the population with mean nutrient intakes below the respective RDA. Patterns with fruit and vegetable intakes were also associated with higher concentrations of serum vitamin C.

Krebs-Smith characterized patterns of adults in a similar way, but was able to quantify the number of servings of Pyramid food groups by combining dietary data from the 1989–1991 CSFII with the PSDB. Each person’s servings were compared with the recommended number of servings of each food group that corresponded to their reported energy intakes. Each person’s pattern was then determined on the basis of whether they met/did not meet the recommendation for each of the five major food groups. Of the 32 possible patterns, six represented 44% of the population. The most prevalent pattern (11%) was meeting the recommendations only for vegetables and meat. Only 1% of the adults met the recommendations for all five food groups. Such patterning techniques involve simultaneously assessing multiple food groups and allowing relationships among the various dietary dimensions to emerge. For example, the pattern of meeting all five food groups was associated with intakes of dietary fat, added sugars and micronutrient intakes all in accordance with recommendations. All other patterns were associated with micronutrient and/or fiber intakes below the recommended amounts, intakes of fat and/or sugars above the recommended amounts or some combination. All five patterns that averaged >=20 g of fiber met the grain recommendation and one or both of the fruit and vegetable recommendations. Similarly, only five patterns had < =30% energy from fat, and all of them met the fruit recommendation.

Similar pattern analyses were conducted on the diets of children and adolescents, ages 2 to 19 y, who participated in the 1989–1991 CSFII. The most prevalent food pattern was meeting dairy only (12%). The next most prevalent food pattern was not meeting any of the food group recommendations (11%). Almost 40% of children and adolescents had patterns of meeting none or only one of the recommendations. Only 2% met all recommendations; only 10% met four or five of the recommendations. Those who met all food group recommendations had micronutrient intakes above the recommended amounts but also had intakes of fat and added sugars above the recommended amounts. Thus, in contrast to findings among adults, meeting the major food groups recommendations was not necessarily sufficient to ensure a healthful total diet.

Scoring methods

Scoring methods that incorporate aspects of the Pyramid have been developed to assess the total diet. One scoring method is the Healthy Eating Index (HEI). The HEI is a 10-component index, i.e., five components measure how diets conform to the Pyramid food group servings of grains, vegetables, fruits, milk and meat, and the other five components include intakes of total fat, saturated fat, cholesterol, sodium and a measure of dietary variety. The scoring system totals 100 points with each of the components having a score that ranges from 0 to 10. Scores of 0 are assigned if no foods are consumed or if nutrient intakes are above the recommended amounts; scores of 10 are assigned if the recommended amounts of foods or nutrients were consumed; intermediate scores are assigned proportionately to the amounts of food or nutrient intakes.
Another scoring method is the Diet Quality Index (DQI). The original DQI was based on eight Diet and Health recommendations, and included intakes of six nutrients (total fat, saturated fat, cholesterol, protein, sodium, calcium), and servings from two food groups (fruits and vegetables, grains). The DQI was modified recently to separate servings of fruits and servings of vegetables according to the Pyramid recommendations, include iron intake and scores of dietary diversity and dietary moderation, and no longer include protein intake. This revised index (DQI-R) has 10 components, each with a score of 0–10 points. Like the HEI, higher scores reflect diets more in accordance with the current dietary guidance.

Both scoring methods have been used to assess the diets of Americans and have produced similar results. In 1994–1996, the mean HEI score was 63.6. In 1994, the mean DQI-R score was 63.4 . In both analyses, 18% of the population had scores <51, indicative of a poor diet, and 12% had scores >80, indicative of a healthy diet. For both indices, less than half of the population met the recommendations for any of the food groups, or for total fat, saturated fat and sodium. More than half of the population met the recommendations for dietary cholesterol. Both scoring methods correlated positively with nutrient intakes, with higher HEI and DQI-R scores associated with higher nutrient intakes in relation to the RDA. Higher HEI and DQI-R scores were also associated with being female, and having higher education or income levels.

Both scoring methods have also been used to evaluate dietary change in the population. In 1994–1996, higher percentages of individuals met several of the recommendations related to components of the HEI compared with 1989–1991 data (in particular, total and saturated fat). Similar findings were observed for components of the DQI. However, change in mean scores across the surveys was very small, suggesting that although Americans improved some aspects of their diets, other aspects of their diets remained unchanged or worsened. For example, the percentage of individuals with adequate intakes of calcium or recommended servings of milk declined. When examined over a longer time period (1965–1991), change in mean DQI scores among both Caucasians and African-Americans, and among low, middle and high income Americans indicated a small improvement in dietary quality. Among the individual components of the DQI, average intakes of dietary fats and sodium remained above the recommended amounts, whereas average intakes of calcium and servings of fruits, vegetables and grains remained below the recommended amounts in all subgroups of the population.
Notable differences emerge between the two scoring methods on the measure of dietary variety. More than half of the population met the variety measure in the HEI in 1994–1996, but <15% met the variety measure in the DQI-R in 1994. This discrepancy is due to differences in how variety is measured. The HEI counts the number of different foods and assigns a maximum score (on a 10-point scale) to individuals who report consuming at least half of a serving of 8 or more different foods over 1 d, or 16 or more over 3 d. The DQI counts the number of different food groups and assigns a maximum score (on a 10-point scale) to individuals who report at least half of a serving of foods in those food groups, weighted according to their representation in four primary food subgroups (grains, vegetables, meat/dairy, fruits), over a 2-d period. In addition, the methodology used to place foods in Pyramid food groups differs. The DQI-R uses the USDA PSDB to place foods in food groups. In the DQI-R, legumes are counted as vegetables with the exception of soy, which is considered a meat substitute. The HEI originally used a commodity equivalent methodology, unique to the developers of this index, to classify foods but has since switched to using the PSDB. In the HEI, legumes are first assigned to the meat group. When meat serving recommendations are met, legumes are then counted as vegetables.

Measures of moderation and proportionality

Aspects of the total diet, other than variety, include moderation and proportionality. Moderation refers to the avoidance of excesses, especially of those food components (e.g., fats, added sugars, alcohol, sodium) believed to be related to suboptimal health outcomes. A unique component of the DQI-R is the assignment of a maximum dietary moderation score to individuals with intakes of discretionary fat, added sugars, alcohol intakes and sodium intakes below cut-off points determined from Pyramid recommendations. The dietary moderation score contributed significantly to the variation in the overall DQI-R score, providing evidence that dietary moderation is positively associated with diet quality.

Proportionality refers to the amounts of foods consumed in relation to each other. The shape of the Pyramid suggests that more servings of foods should be consumed from the grain group and fewer servings of foods should be consumed from the dairy and meat groups. Although not obvious from the Pyramid graphic, proportionality also refers to recommendations for subgroups within major food groups (e.g., dark green vegetables should be consumed several times per week). Proportionality implies that, to maintain energy intake, increased intake from one food group or subgroup necessarily results in decreased intake from another. The food groups interact, as demonstrated by results of Krebs-Smith who showed that food group patterns meeting the Pyramid recommendation for total fat also met the Pyramid recommendation for fruit.
Issues associated with patterning and scoring methods.

The development of the PSDB has made a major contribution to analyses of the U.S. diet, considering the disaggregation of food mixtures, the assignment of foods to their respective groups (including fats and sweets) and the quantification of all food groups (including the tip of the Pyramid) in terms that are consistent with current guidance (e.g., servings). Each of the various dimensions of the diet can be quantified and examined, singly or simultaneously, providing a foundation for assessments of total diet and the direct comparison of diets to recommendations.

However, quantifying the multiple dimensions of diet simultaneously has proven problematic. The seemingly infinite variations in intake along these multiple dimensions necessitate some sort of summary function with which to simplify the interpretation of results. The patterning approach retains the multiple dimensions on which it is based but, on each dimension, dichotomizes the continuous data into categories of “met”/”not met.” Thus, for individuals whose energy-based recommendation for fruit is 2 servings, those who report as few as 2 servings or as many as 10 servings of fruit would be in the “met” group, whereas those who report no servings or as many as 1.9 servings of fruits would be in the “not met” group. Krebs-Smith stated that this approach “obscures the large variability in the amounts of each food group consumed” in an effort to “separate people according to the most rudimentary categories of interest.” In addition, patterns are defined on the basis of the five major food groups and do not include recommendations related to fats and sugars. Had fats and sugars been evaluated as separate food groups, and a category of “exceeding the energy-based recommendations” for each of the food groups been created in addition to “met”/”not met,” the number of patterns would have increased exponentially. Such a large number of patterns would certainly reduce the precision of estimates associated with those patterns and limit their interpretation. Consequently, fats and sugars were not included in defining the pattern, but only as an outcome.

The scoring methods retain the quantification of each dimension but sum these together, in effect reducing the number of dimensions to one. Although this approach is intuitively appealing and offers a simple way to evaluate diets, it implies that certain components of the diet are independent, equally important and additively related to health. This is probably not the case. For example, two components of these indices, total and saturated fat, are highly correlated; consequently, the score is more heavily weighted toward dietary fat than any other dietary component. The assignment of a range of points to each component does take variability in food and nutrient intakes into account, but like the variety scores and food patterning techniques, intakes of food groups above energy-based recommendations are not distinguished further. Total scores at the extremes of the distribution are straightforward in their interpretation, but those in the middle of the distribution (where the majority of individuals fall) are difficult to interpret because distinctly different dietary patterns could result in the same score. For example, one person could score a 70 by completely meeting 7 of 10 recommendations but completely failing to meet 3 recommendations. Another person could score a 70 by partially meeting each of the 10 recommendations. To clarify total scores, individual components of the index are often evaluated. This raises questions concerning whether total scores add information beyond traditional analyses of individual food and nutrient intakes.

Summary of how U.S. diets fare

Evidence from studies of dietary variety shows that one third of the population eats at least some food from all food groups. Evidence from studies of food group patterns shows that many fewer (1–3%) eat the recommended number of servings from all food groups on a given day. Fruits are the most commonly omitted food group, whereas vegetables and meat are the most commonly met by adults and dairy the most commonly met by youth. Among adults, mean number of servings of fruits and dairy fall noticeably below that recommended, whereas among youth, mean number of servings from all food groups but dairy fall below the recommendations. Specific choices within vegetables (i.e., dark green, deep yellow) and within grains (i.e., whole grains) are noticeably below amounts recommended by current food guidance. Intakes of total fat, discretionary fat and added sugars continue to exceed current recommendations. Analyses of diets in relation to socioeconomic status have consistently shown that diets improve with increases in education and income. These results demonstrate shortcomings in variety, proportionality and moderation in the diets of most Americans.

Entry Filed under: General Nutrition

Let the Pyramid Guide Your Food Choices: Capturing the Total Diet Concept Part I Let the Pyramid Guide Your Food Choices: Capturing the Total Diet Concept Part III