Nutrition Labeling: Issues and Directions for the 1990s (1990)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7 Presentation of Nutrition Infor~nation on Food Labels CRITERIA FOR PRESENTING INFORMATION To assure that the nutrition information provided on a food label is conveyed in a manner Hat will allow the majority of consumers to use it successfully, a number of criteria need to be considered, including literacy of users, com- putational abilities, knowledge of English, and knowledge of the specialized vocabulary of nutrition labeling. The actual label presentation scheme needs to make it possible for consumers to understand the nutrition contents of individual food products, compare nutrition contents across product categories, and choose among relevant food alternatives. REFERENCE UNITS FOR DECLARING NUTRIENT CONTENT (SERVING SIZE) In assessing the adequacy of current food labels, the element of serving size affects the usability of all other label components. Over a decade ago, the Food and Drug Administration (FDA) indicated that, "serving size has been one of the issues that has most concerned consumers and manufacturers alike" (DHEW/USDAIFIC, 1979, p. 77~. Concept of Serving Originally, the concept of serving was geared to the actual amount likely to be consumed at a single sitting. However, the concept of serving size is currently used to provide a reference point for information about the nutritional 203

204 NU1771T ON LABELING and other qualities of the food product. Once seeing size is regarded as a standard unit rather than as an estimate of likely consumption, it is possible to visualize varying amounts for similar products, such as a 2-oz serving size for canned tuna and a 3.5-oz serving size for salmon. For maximum usefulness and understanding, however, labeled seeing size should not depart widely from the amount normally consumed at one time. Nutrient information on food labels under FDA and U.S. Depar~anent of Agriculture (USDA) jurisdiction is declared in relation to the average or usual serving, or, when the food is customarily used as an ingredient, in relation to the average or usual portion. The FSIS Standards and Labeling Policy Book and policy memoranda simply stipulate that "when a label contains a statement or claims identification of the number of servings, it must be qualified to identify the size of the servings, e.g. 3, 2 oz settings—or, 1-6 oz seeing—or, 3 portions, 2 oz each" (USDA, 1989b, p. 139~. In the dietary assessment literature, seeing size is typically regarded as a term for a standardized or commonly ingested portion of food. In contrast, portion size refers to that amount of food reported to be ingested at an eating occasion. Definitions There is considerable confusion among three terms: serving, portion, and helping. The term serving was defined by FDA as a reasonable quantity of food suited for or practicable of consumption as a part of a meal by an adult male engaged in light physical activity, or by an infant or child under age 4 when the article purports or is represented to be for consumption by an infant or child under age 4 (21 CEK §101.9(b)~1~. In contrast, FDA defined the term portion as the amount of food customarily used only as an ingredient in the preparation of a meal component, e.g., Y2 tablespoon of cooking oil or \/4 Cup of tomato paste. FDA has further specified that servings and portions must be expressed in terms of common household measuring units or other easily identifiable units such as cups, tablespoons, ounces, or slices. Disparities in Serving Sizes Sewing size is provided as a tool for consumers and users of dietary guid- ance information, nutrient composition data bases, food consumption research, and on food Labels. There are great variations and wide disparities in the in- forrnation presented to consumers in each of these domains for the same foods or for items in the same product categones. A comparison of the seeing sizes specified by various dietary guidance plans, used in food composition data bases, amounts actually consumed, and the range currently shown on food labels is presented in Table 7-1.

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABELS Serving Size Information as Portrayed in Dietary Guidance Materials 205 Dietary guidance systems provide information to consumers about the individual foods they are advised to consume. The common approach has been to recommend a number of daily servings of each of several food groups. In some, but not all, cases the sizes of settings are specified. In 1958 the Basic Four food guide, officially known as Food for Fitness: A Daily Food G - e, was developed by USDA nutritionists (USDA, 1958~. That guide recommended "some mild for everyone," with servings from the mild group of 2 or more cups a day for adults to 4 or more cups a day for teenagers. However, no sizes were specified for the number of servings of food from the meat, vegetable and fruit, or bread and cereal groups. In 1980 and 1985, dietary guidance information was presented to consumers in Nutrition and Your Health: Dietary Guidelines for Americans, but these advisory statements made no specific quantitative recommendations. Instead, they offered seven qualitative, directional statements, such as "eat a variety of foods," "eat foods with adequate starch and fiber," and "avoid too much . . ." (USDAIDHHS, 1980, 1985~. The third edition is expected to provide essentially the same type of directives (USDA, 1990~. In the late 1980s, USDA developed and published a food guidance system, using both a menu planning strategy and commonly used food guides that did specify serving sizes (Cronin et al., 1987~. The pattern for daily food choices recommended consumption of 6 to 11 servings of grains, breads, and cereals; 2 servings of milk, cheese, and yoglrt; 2 to 3 servings of meat, poultry, fish, and eggs; 3 to 5 servings of vegetables; 2 to 4 servings of fruits; and moderate amounts of fats, sweets, and alcohol. In general, amounts of food that counted as a serving were based on typical serving sizes reported by individuals in the 1977-1979 Nationwide Food Consumption Survey conducted by USDA. A typical serving of food was defined as "the median amount of food consumed at a single eating occasion." The amounts of typical servings were specified for food groups and are shown in Table 7-1. The National Research Council (NRC) report, Diet and Health: Implica- tions for Reducing Chronic Disease Risk MARC, 1989a), advised that consumers should "every day eat five or more servings of a combination of vegetables and fruits" and should "increase intake of starches and other complex carbohy- drates by eating six or more daily servings of a combination of breads, cereals, and legumes." The average serving of these foods was defined as "equal to a half cup for most fresh or cooked vegetables, fruits, dry or cooked cereals and legumes, one medium piece of fresh fruit, one slice of bread, or one roll or muffin" (NRC, 1989a, p. 15~. For the most part, the various sources of authoritative dietary advice offer fairly consistent messages about recommended serving sizes for products in the same categories (Table 7-1~. However, it remains to be determined whether

206 NllTRITION LABELING TABLE 7-! Serving Sizes as Depicted by Several Means for Selected Food Dietary Guidance Recommendations Diabetic USDA Food Diet and Exchange Basic Four Guidance Food Item Healtha Lists Food Groups Systems Grains Bread Muffin, bagel 1 Cereals, pasta, rice ll2 c Legumes, beans ll2 c Dairy 1 slice 1 slice _ e 1/2 c, ckd 3/4 C, my 1/3 C 1 slice 1 small Y2 c, ckd Mink, fluid 1 c 1 c 1 c Cheese 1 oz. low-fat 1~/2 oz natural, 2 oz processed Yogurt 8-oz carton — 8 fl oz Fruit Fresh ll2 c 1 piece Average piece Juice ll2 c 6 oz Vegetable Cooked Raw Meat Ground beef Poultry Tuna (canned) Peanut butter Condiments Butter, margarine Catsup 2 c ll2 c 1 c 1 sm. hamburger /2 chicken breast ll2 c 1 T 1 t ll2 c 1 c raw 1 c leafy raw Total of 5 to70z Lean daily a NRC (National Researc h Council). 1989. Diet and Health: Recommmendations for Reducing National Research Council. National Academy Press, Washington, D.C. 749 pp. b American Diabetes Association and The American Dietetic Association. 1989. Exchange Alexandria, Va. 32 pp. c Cronin, F. et aL 1987. Developing a food guidance system to implement the Dietary d Krebs-Smith, S.M., and H. Smiciklas-Wright. 1985. Typical serving sizes: Implications for c Not specified. f Sandwich steaks.

PRESENTATION OF NlTI771TION INFORMATION ON FOOD LABEL; Products 207 Food Composition Data Bases Food Labels Giant Food, Inc., (ranges for USDA USDA Home Food Guide selected As Consumed Handbook 8 & Gdn. Bltn. 72 (average) products) 2 slices 100 g 1 slice 2 slices 1-2 slices ~ 1 1 1 1 c, ckd rice " 1 c 1 c (approx) 1 oz (dry wt) t' 1 c 1/2 c 1/2 c lo " lc 8floz lo 1 ~/2 oz (males) n 1 OZ 1 OZ 1 OZ 1 oz (females) " " ~ oz 6 oz 6 oz container 1 med piece " 1 unit 1 piece 374 C n 1 C 6 fl oz 6 fl oz Y2 c " 1 c I/2 c 1/2 c 2 leaves (males) " 1 c 73 c lettuce Y2 c (females) " 3 oz ckd 2 ozf 6 oz daily (males), " 3 oz 1 oz 4 oz daily (females) " 3 oz 3 oz 2 oz " 1 T 2T 2T " IT IT IT " IT IT — Chronic Disease Risk. Report of the Committee on Diet and Health, Food and Nutrition Board, Lists for Menu Planning. American Diabetes Association and Me American Dietetic Association, Guidelines. J. Nutr. Ed. 19(6):281-302. food guidance. J. Am. Diet. Assoc. 85:1139- 1141.

208 NlmUTION LABELING consumers visualize or consume portions in the same sizes and dimensions as recommended. Serving Size in Nutrient Composition Data Bases The USDA Agriculture Handbook No. 8 series, Composition of Foods, comprise a group of technical publications that provide nutrient composition information for an extensive list of raw, processed, and prepared foods. It is now frequently updated and will consist of 22 sections when fully completed. This series of publications provides nutrient composition information based on 100-g portions of foods (USDA, 1976~. In an effort to compile a document that would be more useful to professional and technical personnel as well as researchers, USDA issued Nutritive Value of American Foods in Common Units (USDA, 1975~. In that publication, nutrient composition information was presented for approximately 1,500 foods in frequently used household measures and market units of food The measurements for specific quantities listed are the customary units now in use for the edible portion of the food item. For example, information for breads is presented both by the loaf and by the slice; for juices, cereals, and fluid milk, by 1 cup; for vegetables and fruits, by the piece or 1-cup portions; and for meats, by 1-pound or 1-cup portions or by the piece. A more concise document for consumer use is Nutritive Value of Foods, Home and Garden Bulletin No. 72, first published in 1960 and last revised in 1981 MISDO, 1981~. It provided a table of nutritive values for household measures of 908 commonly used foods grouped under 15 different main headings. Most foods were listed in ready-to-eat form, but some were basic products widely used in food preparation, such as flour, fat, and cornmeal. The Bulletin was careful to point out that The approximate measure shown for each food is in cups, ounces, pounds, some other well-known unit' or a piece of a certain size. The measures shown do not necessarily represent a serving, but the unit given may be used to calculate a variety of serving sizes. For example, values are given for 1 cup of applesauce. If a serving is Y2 cup, divide the values by 2 or multiply by 0.5; for a 43 Cup serving multiply values by 0.67 (USDA, 1981, p. 4~. Serving Size Reported by Consumers Data bases containing food consumption data taken from national surveys are another source of typical serving sizes. Pao et al. (1982) used 3-day reports from a weighted sample of about 38,000 individuals to determine the weights (in grams) of various foods eaten per meal or snack Krebs-Smith and Smic~las-Wright (1985) converted the amounts reported

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEl~; 209 by Pao and fellow researchers to common household measures for several age and sex groups and used these amounts to determine the most common serving size for different foods. They showed that, for many food items, typically reported servings deviate considerably from the amount accepted as standard servings. Unlit juices, breads, and cereals were frequently consumed in larger amounts than expected, whereas quantities of raw vegetables, meat, fish, and poultry varied widely. Typical serving sizes of breads and cereals were usually twice the size found in earlier recommendations (2 slices versus 1 slice). The study by Hunter et al. (1988) of 194 women to determine the seeing sizes of 68 foods found that for most foods there was no usual seeing size even for foods that are in well-defined units, such as crackers. Guthrie (1984) reported that food portions self-sened by young adults deviated by more than 25 percent from the generally accepted serving size for 28 to 80 percent of the serving selections. Recently, FDA conducted a survey of the amount of food consumed per eating occasion from USDA's 1977-1978 Nationwide Food Consumption Survey and foods in the marketplace. The data were used to determine the standard seeing sizes for 159 food product categories and to define single- sening containers for its proposed rule on seeing size (55 Fed. Reg. 29,476- 29,533, July 19, 1990~. Research on Portion Size Estimation Dietary assessment research has examined the issue of whether consumers are able to provide realistic, valid estimates of the amount of food they consume. A number of the studies that have addressed this topic have documented that a large proportion of respondents cannot accurately judge the amounts of foods and beverages they consume. Both Madden et al. (1976) and Gersovitz et al. (1978) observed overreporting of low intakes and underreporting of high intakes in the subjects they studied—a phenomenon commonly referred to as "regression to the mean." The conclusions from various studies on portion size estimation indicate that people do not give accurate estimates of the amounts of foods they consume. In general, there is a greater tendency to overestimate than to underestimate portion sizes, with the magnitude of the error varying with the specific food item (Goalie, 1984; Minsky and Brownell, 1982; Lewis et al. 1988; Webb and Yuhas, 1988~. Some studies have suggested that food preparation experience seems to help subjects better estimate the amounts of foods they consume, which may explain why women are better than men at estimating quantities of foods. Mining and the use of measuring utensils or food models have been shown to improve some estimates of sizes, but the results are not consistently relivable across all types of foods or memory aides used (Yuhas et al., 19891.

210 Nl~RlTION LABELING Serving Sizes on Food Labels There is tremendous variability in Me seeing sizes currently declared as the reference standard on food labels in different product categories, among foods, and between foods in the same product category as shown in Table 7-2. Ready- to-eat breakfast cereals, perhaps, show the greatest variability, with serving sizes usually designated as 1 oz. regardless of the volume of the product; 1-oz volumes range from 21/2 tablespoons to 1 cup. Not only are the acts quantities not Me same for serving sizes for foods in the same product categories but the units of measurement also vary. For example, serving sizes may be expressed in ounces, units (such as pieces or sticks), cups, tablespoons, or teaspoons. When nutrition labeling was first introduced in 1973, it was left to indus- try to adopt reasonable serving sizes (21 CF1< § lOl.9(b3~1~. After the nutrition TABS 7-2 Serving Sizes Currently Used on Food Labels (Selected sample of commonly used products Food Item ServingslPackage Serving Size or Container Breads Bread, white 2 slices (2 oz) 12 Bread, whole wheat 2 slices (1.7 oz) 9.5 Bread, cinnamon (Pepperidge Farm) 1 slice 16 Bagel (Lenders) 1 4 English muffin (Thomas') 1 6 Cereals looks Bran (Nabisco) 1 oz (Y2 c) 17 AllBran (Kellogg) 1 oz (y3 C) 13 Cream of Wheat (Nabisco) 1 oz (2~/2 T dry) _b Crispix (Kellogg) 1 oz (1 c) 12 Frosted Mini Wheats (Kellogg) 1 oz (Y2 c) 17 Fruit and Fibre (Post) 1.25 oz (y3 C) 12 Fiber One (General Mills) 1 oz (Y2 c) 13.5 Golden Grahams (General Mills) 1 oz (3~4 C) 18 Grape Nuts (Post) 1 oz (y4 C) 16 Instant Oatmeal (General Mills) 1 pkt (1.6 oz) 8 Instant Oatmeal (Quaker) 1 pkt (1 y4 OZ) TO Quaker Oat Squares (Quaker) 1 oz (Y2 c) Meat, poultry, fish Frankfurters Luncheon meat Sausage Fish sticks, frozen Salmon, canned Tuna, canned 10 1 slice 8 1 cooked path (33 g) 10 4 sticks 8 ll2 c 2 oz 4 3.3

PRES~NT~ION OF NUTRITION INFORMATION ON FOOD LABELS TABLE 7-2~0ntinued 211 Food Item Servings/Package Serving Size or Container Beverages Diet soda 6 fl oz Orange juice, prepared from frozen cone. 6 fl oz Hawaiian Punch (drink box) 6 fl oz 1 pkt (6 fl oz) 10 6 fl oz 1.4 Hot chocolate mix Pink grapefruit juice cocktail Dairy Cream cheese Ice cream Mills Sour cream 2 8 1.4 1 oz ll2 c 1 c (S fl oz) Varies 2 T 8 Spreads Butter 1 T Margarine 1 T Mayonnaise 1 T 64 Fruits and vegetables Bluebernes, Dozen 4 oz 4 Fruit cocktail, canned i/2 C 7 Peas, canned \/2 C 4 Peas, frozen i/2 C 6 Tomatoes, canned i/2 C 4 Over Macaroni and cheese dinner, box 3/4 C (as prepared) 4 Noodles Alfredo, box l/2 C (as prepared) 4 Peanut butter, jar 2 T (32 g) 10 Pork and beans, canned i/2 C (130 g) 3.5 Potato ships, bag IS chips (1 oz) 7 Soup, canned, condensed 8 oz (as prepared) 2.5 Soup, canned, single serving 103/4 OZ 1 Spaghetti sauce, jar 4 oz 12 a Connation compiled from visits to local Washington, D.C., area supermarkets by L.S. Sims, 1990. Not specified. labeling regulations were implemented, FDA conducted an informal survey of product labels that provided nutrition information. The results of that survey indicated ~at, in many cases, serving sizes were not reasonable or uniform within a product class (39 Fed. Reg. 20,878-20,887, June 14, 1974~. In 1974, I;DA published proposed serving sizes for several foods, including fluid milk beverages, noncarbonated breakfast beverage products, hot and ready- t~eat cereals, and formulated meal replacements (39 ~d. Reg. 20,895-20,900, June 14, 1974~. Since then, a serving size has been proposed for soft drinks (40

212 NUT~ION LABELING Fed. Reg. 4315~316, Jan. 29, 1975), and petitions have been received seeking to establish a portion size for flour and serving sizes for bread and peanut butter. However, the agency has taken no further action on either the proposed serving sizes or amending the portion size definition. As a result, FDA (and, consequently, USDA as well) does not specify serving sizes for any of the thousands of food products on the market. Rather, the manufacturer decides the serving size to be designated, subject only to the loose requirements that the serving size must be (1) a "reasonable quantity of food suited for or practicable of consumption as part of a meal by an adult male engaged in light physical activity" (or by a child, for foods intended for children) (21 CFIt §101.9(b)~1~), and (2) it must be expressed "in terms of a convenient unit of such food or a convenient unit of measure that can be easily identified as an average or usual serving and can be readily understood by purchasers of such food" (Heimbach et al., 1990~. Over time, a hybrid system for seeing size has evolved. For most foods, the declared serving size is neither an arbitrarily fixed amount that is uniform for similar products nor an average or usual serving, but rather is something in between a standard unit which is set by food manufacturers rather than by the federal government and is unique to each product type. FDA believed that declaration of the nutrition content in terms of a usual seeing of each different food product would be more flexible and more meaningful to consumers so that they could relate the label information to individual intakes. This rationale may still be valid, but it has allowed manufacturers to vary serving size declarations over time and manipulate label claims on the basis of per seeing nutrient contents. The significance of the chosen seeing size is magnified by the fact that all nutrient values declared on the label are dependent upon this determination. For example, jelly and jam produced by the same manufacturer listing 1 or 2 teaspoons, respectively, as serving sizes illustrates the problem. Both labels offer the same information, and a careful comparison would reveal that the nutrient contents of the two products are virtually identical. Clearly, neither label is inaccurate. Yet, consumers who choose products mainly by examining the calorie content may think that one product has twice the calories of the other, when the calorie content of equal portions of the products is essentially the same. The leeway given to manufacturers to set serving sizes offers them the opportunity to portray each food item in the most favorable light. In 1979, FDA expressed concern that"some manufacturers were using overly large serving sizes to inflate the nutritional value of their product in order to enhance its attractiveness (DHEW/USDA/E1C, 1979, p. 77~." The commercial rationale for doing so was the belief Hat consumers were seeking foods on the basis of increased nutritive value and, in particular, higher protein, vitamin, and mineral contents. FDA noted that the serving sizes recommended for many canned

PRESENTATION OF NlJTRITION INFORMATION ON FOOD l,ABELS 213 fruits and vegetables, as well as for some varieties of canned tuna, flit juices, and frozen vegetables, had approximately doubled (in amount) since nutrition labeling was initiated Ten years later, the concern is that manufacturers have begun shrinking serving sizes in response to consumers' tendency to use label information for avoiding particular food constituents such as fat, sodium, and cholesterol (Heimbach, 1985, 1986, 1987~. This tendency may be encouraged by current definitions of descriptors such as low sodium, which is based on a sodium content of less than 140 mg per serving. For example, an 8-oz food package that contains 360 mg of sodium does not qualify for a low-sodium descriptor if it is labeled as providing two 4-oz servings (with 180 mg of sodium per serving), but it may be labeled as low sodium if the package is declared to contain four 2-oz servings. Recent data from PDA's Food Label and Package Survey have shown that 19 of 44 product classes and both bread categories (white and nonwhite) moved toward smaller declared serving sizes in the period from 1977 to 1986 (Heimbach et al., 19903. The information contained in Table 7-2 dramatizes the considerable variability in serving size information on food labels, even for foods within the same product categories. In some cases, food manufacturers keep the serving size unit constant and vary the volume of a serving or a container. For example, the serving size of ready-to-eat breakfast cereal is routinely listed as 1 oz. but the volume varies tremendously (from 2~/2 tablespoons to 1 cup of cede due to the wide variation in the densities of different products. Another example is juice, which is routinely listed as a 6-oz serving, but the number of servings varies from 1.4 to 8 per container. In other cases, the recommended serving size is varied in order to keep the number of servings per container constant, as is the case with single-serving canned soups, packets of instant oatmeal, or individual beverage boxes. Standardized serving sizes are needed so that information remains comparable within product categories, such as breakfast cereals, and across product categories. Alternatives to Serving Size Specifications If serving size is less than ideal as a reference standard, the alternatives need to be considered One alternative would be to declare nutrient content per pacl~ge or container and let consumers judge the proportion of that package that is being consumed at any eating occasion. Such a concept is useful for single-serving containers which are currently popular, but in the Committees view could make it very difficult for the consumer to relate the nutrition data to the amount actually consumed, especially for packages containing large numbers ~ . or servings. Another alternative would be to declare nutrient content per 100 g or other standard unit. This approach is used internationally where the metric system

214 NUTRITION LABEllNG is in place, but it has not been widely used in the United States. Current law in Australia requires labeling per 100 g of product. However, it has recently been suggested that labeling on a per seeing basis may be more appropriate for some foods (Farmakalidis, 1989~. Canada required labeling per 100 g of product until recently, when research evaluating consumer understanding and usage of nutrition labels on ready-to-eat breakfast cereals revealed that consumers rely on and refer to information provided on a per serving basis rather than on a per 100-g basis. For example, 100 g is only about half the weight of a glass of mild but about seven times the weight of a pat of butter. The newest Canadian regulations, which were promulgated in 1988, express seeing size "in the same units as the net quantity declaration, whether it be grams or milliliters, and as well in an equivalent household measure or common unit" (Gunner, 1989~. Fannakalidis (1989) provides further support for using the per serving rather than per 100 g designation: Consider, for example, very lightweight products such as desiccated coconut or a chocolate drink powder. Nutrition labelling on a per serving (in this case approximately 10 g) and a per 100 g serving (or ten ordinary servings) can only serve to seriously mislead consumers into thinking that the product, when consumed in an ordinary manner (i.e. a serving) will provide nutrients equivalent to that in a 100 g serving. Yet another example is puffed wheat cereal one cup of which normally weighs only 15 grams. A 100 g serving would be equivalent to 6.6 cups, an amount which is highly unlikely even for the most dedicated puffed wheat eater. The extra nutrition information provided by the 100 g labelling merely clutters the label and confuses the consumer (p. 980~. Data from an FDA survey revealed that the per serving standard was preferred by majorities of consumers, food and nutrition professionals, and food industry representatives (Heimbach and Stokes, 1981~. This preference was based on the belief that it would be less useful to consumers to have the nutrient content of all products calculated against a standard weight, which would often be larger or smaller than a usual serving. Committee Recommendations The main purpose of stating serving size on the food label is to provide a reference unit for the presentation of the nutrient composition about a product. If sizes are uniform and realistic, this manner of presentation provides consumers with comparable information about the nutritive value of products in the same category (e.g., breakfast cereals). Such label information should also enable consumers to compare the amounts of nutrients they consume with the amount of nutrients currently recommended. Without product-based information of this ~e, it may be more difficult for consumers to compare between alternative meals or meal components (e.g., between breakfast cereals and a bacon and eggs breakfast) or to plan daily diets. Despite the potential for misuses of the

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEL; . 215 concept of seeing, the Committee believes that it remains the best reference unit for expressing the nutrient composition of foods. The Committee recommends that: · Given the alternatives available (serving size, nutrient values per package or container, 100 g portions), serving should continue to be the reference unit for presenting nutrition information on foods. Serving sizes should be expressed in common household measures, fol- lowed by the weight in grams (in parentheses) to facilitate comparisons across product categories. Serving sizes should be standardized across food categories on the basis of volume or weight measures. For example, mink beverages should be listed in 1-cup servings, whereas breakfast cereals may be standardized by weight (e.g., 1 oz), as long as the corresponding volume measure is specified (e.g., 1 cup or 2 tablespoons, dry weight). All seeing sizes should be rounded down to the nearest whole numbers; fractions or decimals should be avoided. The number of settings per package or container should be specified. For a single-seeing container, 50 to 150 percent of the commonly consumed unit would be acceptable. The number of sowings per container should be expressed rounded down to the nearest whole number. Consistent with the recommendation that serving sizes should be standard- ized, quantities specified by dietary guidance recommendations should sense as the main criteria for selecting the amount of food to be described as a seeing. This preference for recommended, rather than as consumed, has the advantage that it can be more readily applied in educational programs and will ensure consistency between seeing sizes as presented in dietary guidance materials and on the food label. · FDA and USDA should jointly establish seeing sizes for a limited number of different food categories (i.e., fruit juices, breads, cereals, fruits, vegeta- bles, spreads, and salad dressings) since serving size-based information will be more valuable for consumers if it applies to broad categories of food. The Committee favors fewer, rather than more, categories so that nutrition information can readily be used by consumers for product comparisons and reference purposes. If a food manufacturer desires a serving size different from that set by the agencies, it should be permitted to petition the responsible agency to allow a deviation or to create a new subclass of foods with its own serving size. · FDA and USDA should establish uniformity in serving size specifications within product categories and between agencies. As noted in the NRC report, Designing Foods: Animal Product Options in the Marketplace, "consistency [among serving sizes] would facilitate comparisons among products, labels, point-of-purchase information, and federal and private data bases" (NRC, 1988, p. 5~. However, it will still be important for nutrition

216 NUTRITION LABELING education programs to provide the type of guidance and information to help consumers make appropriate types of food product comparisons. · Research should be conducted to determine how consumers comprehend food label information and how they interpret seeing sizes declared on the food package. U.S. RECOMMENDED DIETARY ALLOWANCES The first explicit, comprehensive effort to establish national dietary recom- mendations occurred in 1941. The poor nutritional status of military recruits and the prospect of some limitations on food availability were the immediate impe- tus, adding to existing concerns about health problems due to nutrient deficiency diseases. In response, the Food and Nutrition Board (FOB) issued guidelines in the form of Recommended Dietary Allowances (RDA). These recommendations included protein, six vitamins (A, C, D, and thiamin, riboflavin, and niacin), and two minerals (iron and calcium), expressed as the amounts needed (NRC, 1941~. Revisions of this document issued in 1943, 1948, 1953, 1958, 1963, 1968, 1974, 1980, and 1989 showed an evolution in the science and the approaches used to arrive at the recommendations. The range of substances covered has increased, with the most recent 10th edition (NRC, 1989b) covering 11 vitamins and 7 minerals, and providing estimated safe and adequate daily dietary intakes ~SADDIs) for 2 vitamins and 5 minerals. Accumulating scientific data have also made the recommendations more precise. RDAs are defined as the levels of intake of essential nutrients that, on the basis of scientific knowledge, are judged by FNB to be adequate to meet the known nutrient needs of practically all healthy people. Also in 1941, FDA established regulations for Minimum Daily Require- ments (MDR) for use in the labeling of foods (41 Fed. Reg. 5921-5926, Nov. 22, 1941~. The concept was designed to address the problem of labeling foods for special dietary uses and foods to which nutrients had been added. The nu- trients of particular interest were vitamins A, C, and D, and thiamin, riboflavin, and niacin. Experience revealed, however, that many people misinterpreted the name MDR. Some believed that nutrients had to be consumed in at least the stated amounts daily, even though the goal had been to describe a desirable av- erage to be consumed over time. People also misunderstood the term minimum to mean that there was a real health risk if at least those amounts were not consumed daily. Current Provision of Desired Information When FDA initiated nutrition labeling regulations in 1973, it was necessary to establish an appropriate way to convey the content of a certain nutrient

PREVENTION OF NUTRITION INFO~AHON ON FOOD HEM 217 contained in a food. The obvious choice might have been listing the amount of the vitamins and minerals in a food by weight. However, the agency chose to create a system whereby the nutrients for which there was an RDA (protein, vitamins, and minerals) would be listed as a percentage of a standard, so that consumers could determine the contribution a food would make to the daily intake of a given nutrient. Because the RDAs are set for numerous age and sex groups, FDA devised a scheme that generally took the highest RDA value for any of the nutrients involved and established it as the standard or U.S. RDA for that nutrient. Current U.S. RDA values are derived from the 1968 RDAs set for adults (NRC, 1968~. U.S. RDAs were proposed for infants, children, and pregnant and lactating women, but were never finalized; instead, FDA guidelines exist for these three groups. Except for infant, baby, and junior-~e foods for which special regulations apply, adult U.S. RDA values are used as the basis for the percentage reporting of protein, vitamins, and minerals listed on food labels. USDA's policy follows FDA's regulations for use of the U.S. RDA (USDA, 1989b). Although the U.S. RDAs have not been revised as subsequent editions of the RDAs have been published, the 1968 figures are considered generous by most standards. The values are based on the RDAs for adult males or females, whichever value is higher, except for thiamin, niacin, iodine, and magnesium, which are based on the adolescent male RDAs; these values exceed those for adults. The U.S. RDAs for calcium and phosphorus are also higher than the adult RDAs, but lower than those for teenagers, and pregnant and lactating women. The 1989 RDA provided values for vitamins A, C, D, and E, thiamin, riboflavin, niacin, Be, B12, folate, calcium, phosphorus, magnesium, iron, zinc, and iodine (NRC, 1989b). The 1968 RDA did not include recommendations for zinc, copper, biotin, or pantothenic acid (NRC, 19683. Estimates were made to establish a value so that the percentage of the U.S. RDA for labels could include these essential nutrients if the manufacturer wished to provide them on the label. Table 7-3 provides a comparison of the current U.S. RDAs with the 1968 and 1989 RDAs. The regulations state that the U.S. RDA may be amended "from time to time as more information on human nutrition becomes available" (21 CFR § 101.9(c)~7~(b3(ii)~. On the food label, the percentage of U.S. RDA is "a statement of the amount per serving (portion) of the protein, vitamins, and minerals." According to FDA regulations, U.S. RDA percentages are expressed in 2 percent increments up to the 10 percent level, 5 percent increments from 11 to 50 percent levels, and 10 percent increments above the 50 percent level (21 CF~ §101.9(c)~7~(i)~. Nutrients present in amounts less than 2 percent are indicated by a zero or an asterisk referring to another asterisk at the bottom of the table and followed by the statement "contains less than 2 percent of the U.S. RDA of this (these) nutrient (nutrients)." When a food product contains less than 2 percent of the U.S. RDA for each of five or more of the eight required nutrients (protein, vitamins

218 Nl~RlTION LABELING TABLE 7-3 Comparison of U.S. RDAsa with 1968 and 1989 RDAsb,C Nutrient 1968 RDA Adult Adult Male 1989 RDA Adult Adult Female U.S. RDA Male Female Required nutrients on current nutrition information panel Protein (g) Vitamin A (RE) Vitamin C (mg) Thiamin (mg) Riboflavin (mg) Niacin (mg) Calcium (mg) Iron (mg) Optional nutrients on current nutrition information panel Vitamin D (]U) Vitamin E (IU) Vitamin B6 (mg) Folic acid (mg) Vitamin BY ~g) Phosphorus (g) Iodine (fig) Magnesium (mg) Zinc (mg) Copper (mg) Biotin (mg) Pantothenic acid (mg) 65 55 1~000 800 60 55 1.4 1.0 1.5 13 800 800 10 18 18 65 5,000 IUd 60 1.5 1.7 20 1,000 400 30 2 0.4 6 1.0 150 400 15 2 0.3 10 30 25 2 2 0.4 0.4 6 6 0.8 0.8 120 100 350 300 63 1~000 60 1.5 1.7 19 800 10 50 800 60 1.1 1.3 15 800 15 5 Age 10 a-TEf 2 0.2 2 2 0.8 150 150 350 280 15 12 1.5 -3.08 0.03-0.18 4 -78 5 Age 8 a-TEf 1.6 0.18 0.8 a 21 CFR §101.9 (7). b NRC (National Research Council). 1968. Recommended Dietary Allowances, Ah ed. Food and Nutrition Board, Assembly on Life Sciences. National Academy of Sciences, Washington, D.C. 101 pp. c NRC (National Research Council). 1989. Recommended Dietary Allowances, 10th ad. Report of the Subc~xnmittee on the Tenth Edition of the Recommended Dietary Allowances, Food and Nutrition Board, Commission al Life Sciences. National Academy Press, Washington, D.C. 284 W. d 1,000 RE. c 5 lag as cholecalciferol = 200 IU of vitamin D. f a-Tocopherol equivalents (1 a-TE = 1 mg of d-a-tocopherol). 8 Estimated safe and adequate daily dietary intakes (ESADDI) for both males and females.

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABELS 219 A and C, thiamin, riboflavin, niacin, calcium, and iron), the manufacturer may choose to declare no more than three of those eight required nutrients and none of the optional nutrients. In that case, it must be accompanied by the statement "contains less than 2 percent of the U.S. RDAof...," listing whichever of the eight required nutrients are not declared, directly following the declared nutrient in the same type size. When vitamins and minerals are added to a food, or a claim is made about them, the percentage of the U.S. RDA contributed by the food must be declared. U.S. RDA percentages of other optional micronutrients may also be listed. The percentage of the U.S. RDA for protein is currently required to be declared on the label (21 CFR § lOl.9(c)~7~(ii)(a), ably. There are two U.S. RDA adult values for protein because of the difference in protein quality. The U.S.RDA of the protein efficiency ratio (PER) in a food is 45 g if the product is equal to or greater than that of casein, and 65 g if the PER of the total protein in the product is less than that of casein. Total protein with a PER less than 20 percent of the PER of casein is not allowed on the label in terms of percentage of U.S. RDA, and the statement of protein in grams per serving is to be modified by a statement that the food is not a significant source of protein content regardless of the actual amount of protein present. There are special labeling requirements for infant, baby, and junior-type food (21 CFR § lOl.9(h)~1~. The U.S. RDA levels for infants from birth to 12 months of age and children under 4 years of age may be declared for foods intended for use by these two age groups. If such dual declaration is used on any label, it is to be included in all labeling, and with equal prominence given to bow values in all promotional material. The U.S. RDA PER values for infants are 18 g and 25 g, and for children under age 4, 20 g and 28 g depending on protein quality. In cases in which the food does not contain 40 percent of the PER of casein, the U.S. RDA may not be declared on the label of infant foods, and the label must carry the statement, adjacent to the protein content statement, that the food is not a significant source of protein for infants, regardless of the actual amount of protein. If the adult U.S. RDAs continue to be used as the basis for labeling vitamins and minerals, Hey need revision since several adult nutrient levels and/or their units of measurement are out of date. Nutrient levels most notably in need of revision are those for biotin, vitamins BE and D, folate, pantothenic acid, iron, niacin, and protein (Table 7-3~. The current recommendation for biotin is one-ten of the current U.S. RDA; for vitamin Big, one-third; and for vitamin D, folate, and pantothenic acid, one-half. The current U.S. RDA for protein is slightly above the 1989 RDA, with the 1989 reference protein allowance for adults now set at 0.75 g/kg/day. A moderate protein intake (no more than twice the RDA for all age groups) is now recommended (NRC, 1989a). The U.S. RDA does not use the official measurement unit, retinal equivalents (RE), for vitamin A activity, nor does it use me most appropriate measurement units for

220 NIJTR1TION LABELING vitamins D or E, or niacin. Weights and heights for 1989 RDA reference adults are actual U.S. population mediums, but these figures do not imply that the height-to-weight ratios are ideal. The technical nature of the U.S. RDAis not entirely overcome by using percentage comparisons on labels. Less emphasis on micronutrients on labels decreases the usefulness of comparisons with the U.S. RDA. Current regula- tions for the U.S. RDAs for protein, vitamins, and minerals include numerous exemptions and special requirements. The number of regulatory options further decreases consumer understanding, and hence, the value of the U.S. RDA in food labeling. U.S. RDAs are easily misconstrued as minimum requirements, especially because consumers often assume that more is better in their interpre- tation of the nutrient content of foods. The 1968-based U.S. RDA percentages on labels underestimate the relative amounts of several nutrients in foods. The Committee believes that resolution of these problems and the emphasis on total diet, rather than daily consumption, should be the focus of reform. Ample assurance that all healthy individuals are covered should be the emphasis of any standard values used as the basis of nutrition labeling. The current terms RDA and U.S. RDA have also been the source of con- fusion. FDA recently proposed new terms to replace the U.S. RDA: Recom- mended Dietary Intakes (RDIs) and Dietary Reference Values (DRVs) (56 Fed. Reg. 29,47~29,533, July 19, 19903. These two terms seem to compound the problem by continuing to use the letters R and D in some combination in the term. The Committee suggests that alternative terms such as Dietary Values, Reference Values, Reference Intake, or Standard Values be considered for the name of a dietary standard to describe nutrient content. Whether the RDAs or other dietary recommendations serve as the basis for reference values, the use of only one term both on the label and in the regulations will help to reduce the confusion, while still allowing the general concept to represent population- based, standard reference values. "Source of" Listings of Micronutrients The Committee was persuaded that the current manner of providing infor- mation about the levels of vitamins and minerals present in foods was dys- functional. EDA's current regulations require the listing of protein and seven vitamins and minerals as a percentage of the U.S. RDA, which means that the user must understand both the concepts of the U.S. RDA and percentages. There is a good deal of evidence that many consumers do not (DHEW/USDA,/E1C, 1979; Elizabeth Yetley, Center for Food Safety and Applied Nutrition, FDA, personal communication, July 1990~. Furthermore, the use of percentages cre- ates undesirable incentives for manufacturers to overfortify foods in order to achieve "100 percent of your [or the government's] requirements."

PREVENTION OF NUTRITION INFORMATION ON FOOD HEN 221 Accordingly, me Committee favors a system in which vitamins and minerals, when they are required or allowed to be listed, are described only qualitatively, using simple terms that convey usable information about a food as a source of micronutrients. The Committee recommends that the descriptors very good source of, good source of, or contains be used to characterize foods containing the required or optional micronutrients. These descriptors would be defined in terms of percent ranges of the U.S. RDA or any other system of recommended nutrient values that FDA and USDA might adopt. Implementation of this scheme would require two types of decisions, and to an important degree, the choices the Committee made (or that FDA and USDA might maker are matters of judgment rather than of evidence. The first decision concerns the terminology to use. Current FDA regulations state that no claim may be made that a food is a significant source of a nutrient unless that nutrient is present in the food at a level equal to or in excess of 10 percent of the U.S. RDA in a serving (21 Cow lOl.9(c)~7~(v)~. FDA policies for shelf labeling allow a food to be characterized as an excellent source of a vitamin or mineral if a serving contains 40 percent or more of the U.S. RDA; the term good source of is assigned to foods containing 25 percent or more in a serving, and source of can be used by foods containing 10 percent or more of the U.S. RDA. The Committee recommends changing both the terms for and the content criteria to define these tiered source descriptors. Two considerations led to this recommendation. First, the term excellent source appeared to provide its own incentive for unnecessary vitamin and mineral fortification. Second, on close review of the vitamin and mineral content of a variety of foods, including many fruits and vegetables, grains, meat, and poultry, it appends clear that very few would be eligible for use of this accolade, even though many are recognized as important sources of nutrients. Furthermore, many vitamins and minerals do not occur naturally in high levels in any one food; a diet supplying nutritionally adequate levels of these nutrients must be assembled from a variety of different foods. This is the Me of dietary pattern that labeling should encourage, not penalize. Applying FDA's own criteria, very few standard servings of unfortified foods could be characterized as an excellent source of any nutrient. Thus, the Committee opted for the more modest set of descriptive terms—very good source of, good source of, and contains. The next choice the Committee faced was defining these terms. To aid in this exercise, the Committee examined micronu~ient levels (per serving) in a selected sample of foods. The foods chosen for this exercise are listed in Table 74. The right-hand column indicates the percentage of the current RDA for the highest level of vitamin or mineral present, then the next highest, and so on. For the exercise, vitamins A, B6, and C and folate, calcium, iron, sodium, and potassium are described. The purpose was to develop a preliminary understanding of how common foods which potentially comprise parts of a healthy diet might ram in the type of system the Committee was developing. It made little sense to define

222 NUTRITION LABELING TABLE 74 Nutrients Contained in a Standard Sewing of Selected Foodsa Ranked by Decreasing Percentage of the 1989 RDAb Food Nutrient Percent RDA Very good source of (greater than 20 percent of standard) Beef liver, pan-fried, medium (3 oz) Carrots, raw (100 g) Orange juice, canned (6 oz) Beef liver Beef liver Milk, skim (1 c) Beef liver Beef liver Tomatoes, raw (100 g) Chicken, white, no skin, fried (3 oz) Bread, whole wheat (1 slice) Kidney beans, canned (100 g) Cheddar cheese (1 oz) Prunes, whole, deed, uncooked (100 g) Potatoes, baked, flesh and skin (100 g) Bread, white (1 slice) Good source of (from 11 to 20 percent of standard) Potatoes Bread, whole wheat Cottage cheese, 4% fat, creamed (~/2 C) Prunes Bread, white Chicken, dark, no skin, fried (3 oz) Carrots Mild, skim Prunes Prunes Kidney beans Bread, whole wheat Tomatoes Hamburger, 19% fat, pan-fried, medium (3 oz) Tuna, canned, water-packed, light meat (~/4 C) Tuna Hamburger Flounder, fresh or frozen, raw (3 oz) Contains (between 2 and 10 percent of standard) Bread, white Cheddar cheese Apples, raw, with skin (100 g) Potatoes Kidney Beans Chicken, dark Cottage cheese Vitamin A Vitamin A Vitamin C Folate Vitamin B Calcium Con Vitamin C Vitamin C Vitamin B6 Sodium Folate Calcium Vitamin A Vitamin C Sodium Vitamin B6 Con Sodium lion lion Vitamin B6 Vitamin C Vitamin A Potassium Vitamin B6 Sodium Calcium Vitamin A Vitamin B6 Ifon Vitamin B6 Con Vitamin B6 Calcium Vitamin A Vitamin C Don Iron Iron Calcium 1,000.0 310.0 100.0 95.0 67.0 35.0 35.0 32.0 30.0 29.0 27.0 26.0 25.0 22.0 22.0 22.0 19.0 19.0 18.0 17.0 17.0 17.0 16.0 16.0 15.0 14.0 14.0 14.0 13.0 13.0 12.0 12.0 12.0 10.0 9.6 9.6 9.5 9.3 8.7 8.7 8.5

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEl~ TABLE 7~Continued 223 Food Nutrient Percent RDA Potatoes Potassium 8.4 Orange juice Vitamin B6 8.3 Carrots Vitamin B6 8.3 Tuna Sodium 8.0 Milk, skim Potassium 7.6 Carrots Folate 7.4 Cottage cheese Folate 7.4 Cheddar cheese Sodium 7.2 Bread, whole wheat Potassium 6.6 Prunes Calcium 6.4 a Based on data from Agriculture Handbook No. 8 (series), U.S. Deparunent of Agriculture. b Based on average values for adult men and women. NRC (National Research Council). 1989. Recommended Dietary Allowances, 10th ed. Report of the Subcommittee on the Tenth Edition of the Recommended Dietary Allowances, Food and Nutrition Board, Commission on Life Sciences. National Academy Press, Washington, D.C. 284 pp. the third-tier descriptor, very good source of, in such a way that very few foods would be eligible to employ it. Using the Committee's system, very few unfortified foods provide 100 percent of the RDA of any nutrient in a serving. Indeed, only five foods in the sample provide more than 40 percent. Only 16 foods in the sample provided more than 20 percent. With this array in mind, the Committee debated the appropriate range of a dietary standard (in this case, the 1989 RDA) to assign to the three tiers it had age on. It recommends that very good source of be used to describe any food that provides, in a serving, more than 20 percent of the dietary standard for a given vitamin or mineral. It recommends that good source of be used for any food that provides, in a serving, 11 to 20 percent of the dietary standard for a given nutrient. Any food that provides between 2 and 10 percent of the dietary standard for any nutrient would be permitted to say it contains that nutrient. A food that contained less than 2 percent of the dietary standard for any nutrient would not be required, or allowed, to list any vitamins or minerals on the nutrition information panel that were not present at a level above 2 percent of the dietary standard. These would appear on the nu~idon information panel following the macronutrient listing as follows ~ O A very good source (over 20% [standard]) of: .... A good source (11-20~o [standard]) of: .... Contains (2-lO`Yo [standard]) of: . . This manner of listing would allow manufacturers to draw attention to the vitamin or mineral content of foods that are significant sources. It would assist

224 NlJTRITION LABELING consumers with recognizing, and choosing among, foods that are important sources of micronutrients. And it would discourage overfortification of processed foods by limiting those that contain substantially more than 20 percent of the dietary standard for any micronutrient to the term very good source of. It should be repeated that the development of such a system of verbal rather than numeric description of vitamin and mineral content embodies decisions that are, if not arbitrary, ultimately matters of judgment. The Committee would not think it unwise for FDA and USDA to utilize a different set of terms to characterize macronutrient content, though it would consider any system unwise that retained terms such as excellent, which encourage needless fortification. Nor is the Committee absolutely convinced that the percent ranges of a dietary standard it has arrived at are the ideal ranges. Rather the recommendation put forward here is intended to reflect what is believed to be a sounder way of conveying label information about the vitamin and mineral content of foods. Committee Recommendations The Committee recommends that: The U.S. RDAs (or different reference term) should be updated, even if they are to play a smaller role in nutrition labeling in the future. FDA and USDA should require the use of the descriptors very good source of, good source of, or contains to characterize the content of required or optional micronutrients in foods. · Use of the descriptive terms on the nutrition information panel would require that micronitrients meet the following or similar criteria: use of very good source of must provide, in a seeing, more than 20 percent of the dietary standard for a given vitamin or mineral; use of good source of must provide, in a seeing, 11 to 20 percent of the dietary standard for a given nutrient; use of contains must provide, in a seeing, between 2 and 10 percent of the dietary standard for any nutrient; and a manufacturer would not be required or allowed to declare any nutrient present at less than 2 percent of the dietary standard. INGREDIENT LABELING One important source of information for consumers about the composition of packaged foods is the statement of ingredients that the Federal Food, Drug, and Cosmetic Act (FD&C Act), the Federal Meat Inspection Act (FMI Act), and the Poultry Products Inspection Act (PPI Act) require on most food labels. FDA and USDA require that ingredients be listed in order of predominance by their common, specific names. A complete, simple listing of ingredients can provide consumers with important information about the nutrient content of

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEL; 225 many foods (e.g., when an obvious source of protein or fat appears near the top of the list) or about its role in health-restricted diets (e.g., when eggs or salt appear as an ingredient). On foods that do not now have nutrition labeling, He ingredient statement may provide the only factual information about nutritional value. Furthermore, ingredient labeling can be a useful supplement to more elaborate nutrition labeling when the latter is provided. FDA and USDA have similar requirements for ingredient labeling; however, they are subject to important exceptions. Some of these exceptions have nutri- tional implications and should, in the Committee's judgment, be eliminated by administrative action or, if necessary, by legislation. The Committee acknowl- edges that other criticisms have been lodged against the current rules governing ingredient labeling, such as the authority found in all three statutes, to declare spices, flavorings, and colors by function rather than by name. In declining to recommend modification of these provisions and other exceptions that have no direct relationship to the nutritional value of foods, the Committee does not mean either to dispute or to accept the criticisms; it is merely adhering to the boundaries of its charge. The most important, and dubious, exception from the general requirement that the ingredients of a food be declared is the provision in the FD&C Act that allows, foods covered by a standard of identity, the omission of mandatory ingredients and the inclusion of only those optional ingredients that the standard requires to be declared 0;D&C Act § 403(g)~. This provision not only implies that mandatory ingredients need not generally be declared, but that FDA may not require their listing in specific instances. This is the agency's interpretation and it has been upheld by the courts (L`A;BEL, Inc. v. Edwards, 1969-1974 F.D.L.I. Jud. Rev. 733 (D.C. Cir. 1973~. FDA responded to this ruling by announcing that it would require labeling of all optional ingredients in standardized foods and, furthermore, that it would take steps to amend existing standards of identity to recharacterize most heretofore mandatory ingredients as optional. By 1989, the agency had thus modified four-fifths of the existing food standards. USDA generally requires all ingredients to be listed on a food label, regardless of whether it is subject to a standard of identity. In the Committee's view, FDA is to be commended for seeking to enlarge the share of foods that bear full ingredient labeling, but the statutory procedure that has made this slow and painstaking effort necessary is an anachronism that should be changed. The majority of packaged foods are not covered by standards of identity; accordingly, all of their ingredients must be declared on the food label. There is no longer any plausible basis, if there ever was one, for exempting standardized foods from this ingredient labeling requirement. There is also a troubling irony in FDA's announcement that an ingredient which for years has characterized a standardized food is now declared "optional" so that it can be required to appear on the label. As this report notes, many important foods remain subject to EDA or

226 NUTRITION LABELING USDA standards of identity. Many of these foods, especially dairy products, contain nutrients in amounts that consumers should be attentive to and seek to modify in their diets. The Committee recommends that Congress amend the FD&C, FMI, and PPI Acts to make clear that the general requirement of full ingredient labeling applies to standardized as well as nonstandardized foods. In the meantime, it encourages FDA to continue the process of amending its standards to require ingredient labeling. In addition, FDA and USDA should amend their regulations to require that the ingredients of standardized foods that are incorporated into other packaged foods are declared by name (and in order of predominance) on the label of the final product. The reasoning that supports full ingredient labeling of all packaged foods, including those covered by standards of identity, applies equally in this case. FDA and USDA both require that ingredients be listed in order of predom- inance (by weight). Thus, the chief, often characterizing, ingredient is usually listed first, followed by ingredients used in smaller quantities. Each ingredient (other than spices, artificial flavorings, and colorings) must be named specifi- cally. Thus, for example, corn syrup is identified by name and listed separately. This manner of listing, however, can obscure the total amount of sugars in the food. When different sugars are listed individually—e.g., dextrose, honey, and corn syrup—consumers may not always realize that all of these ingredients are sugars, and the quantities used may result in their dispersal throughout the listing of ingredients. In such cases, listing of ingredients both by category and by name may be necessary. A revised regulation could require that all sugars be aggregated for purposes of the sequence of listing under the generic heading sugars and also be described by specific name. The label of a fruit pie might then bear the following listing: Cherries, Water, Flour, Sugars (Corn Syrup, Dextrose, etc.), Corn Starch, Salt, Artificial Flavoring. It would require only modest changes in FDA and USDA regulations to accomplish this change, and no change in existing legislation would be necessary. Although it could be argued that other ingredients, such as complex carbohydrates (starches), should also be grouped, current recommendations advise that it is the amount of simple carbohydrate consumed that should be reduced. Neither FDA nor USDA generally requires that food labels disclose the proportions or relative quantities of individual ingredients, beyond the informa- tion Hat the consumer can infer from the sequence in which ingredients are listed. In some cases, FDA has adopted regulations requiring Cat the amount of characterizing ingredient in a food be declared as part of the product name (e.g., Peanut Spread Containing 45 Percent Peanuts), but in each instance, the FDA Commissioner has made a finding that the amount of this ingredient (but not others) has a material bearing on price or consumer acceptance (Huts and Mer- rill, 1990~. It is doubtful that FDA or USDA could now mandate a percentage

PRESE~ION OF NUTRITION INFORMATION ON FOOD LABEl~; 227 declaration of ingredients for all foods. Some critics of current labels, however, argue that legislation should require, or at least allow the agencies to require, such labeling. The Committee takes no position on these proposals, which may or may not be justified in terms of protecting consumer economic interests. The Committee has not been persuaded that, if the ingredients of all processed foods were listed and full nutrition information were required, the costs of percentage ingredient labeling would be worth its possible contribution to consumer assessment of the nutrient content of foods. However, this does not mean that the case for percentage labeling, at least of major ingredients, cannot be made on other grounds. Finally, the current FDA and USDA format for declaring ingredients has been criticized. Although this format is not dictated by agency regulations, the typical mode of disclosure has a common and, in the view of critics, unfriendly appearance. Following the heading "ingredients," ingredients appear in a margin- to-margin, flattened column printed entirely in capital letters, separated only by commas, without other breaks or classifications, and occasionally parenthetical phrases describing ingredient functions. Moreover, only a minority of consumers appear to understand the rationale of the prescribed order of listing (Heimbach, 1982~. The Committee has been impressed by proposals for making this portion of the food label more user friendly by, such changes as, using capital and lowercase letters, separating major from minor ingredients, and employing contrasting colors. The Committee recommends that when FDA and USDA test different basic formats for nutrition labeling, they should also seek information about consumer reactions to and use of different formats for depicting the ingredients in foods. If the ingredient panel is to contribute to consumer understanding of the nutritional characteristics of different foods, serious efforts should be made to improve its readability. Committee Recommendations The Committee recommends that: Congress should amend the FD&C, FMI, and PPI Acts to make clear that the general requirement of full ingredient labeling applies to standardized as well as nonstandardized foods. FDA and USDA should take steps to amend their requirements for ingredient labeling to require that the ingredients of standardized foods that are incorporated into other processed foods are declared by name on the label of the final product. · When FDA and USDA test different formats for nutrition labeling, they should also seek information about consumer reactions to and use of different formats for depicting the ingredients in foods.

228 NIJTI~TION BEING STANDARDS OF IDENTITY This report focuses on the desired content and appropriate format for nutri- tion information on food labels, but the Committee was also urged to examine other regulatory practices that impinged on efforts to improve Americans' diets. One of the most controversial issues is the system of food standards of identity created and enforced by FDA. In broad terms, a standard of identity represents the official recipe for a food; it defines the composition of products entitled to use the name of a food. For example, the FDA standard of identity for ice cream specifies the ingredients that any product labeled ice cream must include, lists other optional ingredients that it may include, and prescribes the amount of mink fat 10 percent~hat the final product must contain. In specifying the amount of this presumably valuable constituent, the ice cream standard is like numerous others that FDA has adopted under the authority of the FD&C Act. Many such standards cover daily products, including cheeses, yogurt, and milk. FDA's authority to establish definitions and standards of identity for foods is not unique (FD&C Act §401~. USDA exercises similar powers under both the FMI and the PPI Acts and has established standards for many meat and poultry products. Claims that food standards can impede consumer efforts to choose more nutritious foods have focused primarily on many of those issued by EDA several decades ago, when concerns about nutrition and the goals of regulation were quite different. USDA standards may not raise the same issues since they do not require a minimum fat content as some FDA standards do; most USDA standards establish maximum fat limits and minimum meat/poultry requirements. With the time constraints it faced, the Committee was not able to study these claims thoroughly. It was persuaded, however, that certain aspects of FDA's current food standards system require reform. The criticisms of many FDA standards of identity fall into three categories: (1) Standards for such foods as ice cream and other daily products require the presence of high levels of undesired constituents, chiefly fat; (2) FDA's enforcement of these food standards discourages the marketing of substitutes for the standardized products containing reduced levels of fat or other less desirable constituents; (3) current labeling requirements for standardized foods under the FD&C Act fail to require a full listing of ingredients, a criticism dealt with in the previous section. A fours criticism of the current FDA regimen concerns the procedure by which standards of identity must be adopted and, correspondingly, amended. This procedure, dictated by section 701(e) of the FD&C Act, is known as formal rulemaking. It entails, in addition to the conventional publication of a proposal followed by the submission of written comments and promulgation of a final regulation, the opportunity for a formal evidentiary hearing, He administrative equivalent of a judicial Dial. This last requirement means that a proceeding to establish, amend, or repeal a standard of identity may take 2 or 3 years tl)

PRESENTATION OF NUTRITION INFORMATION ON FOOD Labels 229 complete. One notorious proceeding, to set a standard for peanut butter, took over a decade. This procedural feature of the food standards provision of the FD&C Act has implications for the substantive criticisms of the FDA system. Two concerns motivated Congress in 1938 when it conferred this power on FDA. One was a concern for consumers, whom Congress (and FDA) believed often could not distinguish a debased product by sight or taste alone—for example, fruit preserves in which less expensive sugar was substituted for more costly fruit. The second concern was for producers, who claimed that manufacturers of cheapened products were competing unfairly. These concerns combined to yield the statutory authority under which FDA has defined and standardized nearly 300 foods, products that at one time accounted for almost half of consumers' food expenditures (Merrill and Hutt, 1980~. In 1990, less skepticism exists about consumers' abilities, aided by informative labeling, to protect themselves against debased or diluted products. As The Surgeon General's Report on Nutrition and Health (DHHS, 1988) and the NRC Diet and Health report (NRC, 1989a) confirm, attention is now focused on the consumption of too much fat rather than the possibility that some products will be made using less of an ingredient that was historically considered a valuable constituent. Accordingly, it seems clear to the Committee that any system that significantly impedes the marketing of reduced-, low-, and non- or no-fat substitutes should be reexamined and, presumably, changed. In theory, FDA food standards need not impede the marketing of such products, but there is reason to believe that they sometimes do (see Lorman, Appendix D). It is clear that the current standard of identity for ice cream prevents the marketing of a product called ice cream that contains less than 10 percent mills fat which would violate the food standards provision of section 403(g) of the FD&C Act. A possible response would be to amend the standard of identity and reduce the required level of fat, or eliminate any minimum fat requirement. However, the procedural impediments to such an action are formidable. Current manufacturers of ice Cretan would likely oppose any change in the standard that would make it easier for new and presumably more healthful products to compete with their formulations. Such opposition, even from only a few manufacturers, would ensure that FDA would have to follow the Act's full panoply of procedural requirements. Amending the ice cream standard would not be the only way to facilitate the marketing of lower-fat alternatives. In theory, a frozen dessert product containing only 5 percent milk fat, or even less, should be marketable, and in fact it is. The problem comes in deciding the name for it. The law allows three possibilities but, as administered by FDA, forecloses the one possibility that manufacturers would most prefer (see Lorman, Appendix D). The lower-fat alternative could be standardized under an entirely new name, but this possibility confronts He same procedural hurdles as an effort to amend the ice cream standard does. It is unattractive for another reason because FDA is likely to be reluctant to allow

230 NU7~TloN LABELING the use of the name ice cream in any new standard. That name, in the agency's view, belongs to the original 10 percent milk fat formulation. Another possibility would be to market the lower-fat alternative, without a standard of identity, under a name that distinguished it from standardized ice cream. Under FDA's view of the law, this descriptive name may not include the words ice cream. No-fat ice cream would not be allowed because the characterizing ingredient, cream, would not be part of the product, although a reduced-fat ice cream would be allowed. According to FDA, such a product would pope to be ice cream and, because it would not meet the standard, would be adulterated. (ADA would allow reduced-fat ice cream because it would contain at least some cream.) A name such as frozen Danbury dessert would probably suffice, but many food marketers believe that this or any similar alternative name would make it more difficult for the product to attract the consumer's attention as a lower-fat substitute for the real thing. In short, they assert that fewer consumers would make the nutritionally sound choice to buy the substitute. Based on case law, there is another alternative to which FDA would not object and which would permit the use of the name ice cream. The new product could be marketed as mutation ice cream Merrill and Hutt, 1980~. Many food marketers believe, however, that such a name would clearly mark the product as inferior to traditional ice cream and make it even harder to sell. It is difficult to assess the merits of the arguments that FDA standards of identity, and the agency's refusal to permit the use of standardized names in the names of reformulated substitutes, are in fact discouraging the marketing of more healthful alternatives to traditional foods. Supermarket shelves are full of new products bearing novel names that have gained consumer acceptance. A firm prepared to spend large sums of money to promote a product as new, different, and more healthful can often compete effectively with a traditional formulation without using the standardized name. This strategy, however, selectively favors large producers. The criticisms of current FDA food standards policy seem plausible enough to warrant a recommendation that both the agency's current enforcement policy and the law itself should be reexamined. Whether or not substantial reform is merited, the Committee believes that two changes need to be made in the FD&C Act: The current procedural requirements for adopting, amending, and repealing standards of identity cannot be justified, either in terms of efficient administration or of ensuring fair opportunities in PDA n~lemaking. The majority of FDA regulations, which resolve issues of great moment and that govern commercial practices of profound significance to consumers and producers, are promulgated through informal rulemaking. Section 701(e) is an anachronism and should be repealed. At the very least, Congress should amend the FD&C Act to exempt standards of identity from its requirements. This change in the law would accomplish two Dings: First, it would make

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABELS 231 it easier for FDA to amend old food standards that embody restrictions that no longer make sense, nutritionally or economically. Correspondingly, it would facilitate the adoption of standards for new substitutes for traditional foods when that seems appropriate. Second, FDA's Midyear effort to recharacterize mandatory ingredients of standardized foods as optional and require their complete listing on the label would be expedited if it were easier to amend food standards. However, the Committee believes this exercise will no longer be necessary if full ingredient labeling is required for foods with standards of identity. The latter advantage would be significant only if Congress failed to make the second change that the Committee recommends. As noted elsewhere, the listing of ingredients on a food's label conveys information about the nutrition contents that can help attentive consumers make sounder nutrition choices. The ED&C Act effectively exempts standardized foods from this otherwise universal requirement. This exemption, in the Committee's judgment, should be repealed promptly. Committee Recommendations The Committee recommends that: · FDA's food standards should be carefully examined for their effects on the marketing of low-, lower-, and no-fat substitutes for high-fat foods. · Congress should amend the FD&C Act to eliminate the requirement that standards be adopted and amended through formal rulemaking. · Congress should eliminate the exemption from full ingredient labeling for standardized foods. PRINCIPAL DISPLAY PANEL DESCRIPI ORS Food labels have probably always been used to promote as well as to describe foods. In the current marketing environment, merchandisers of food think carefully about every facet of label design, from color to typography to size and location. Many choices are influenced by government requirements, but this does not mean that the entire contents are dictated by governmental directives. Manufacturers of foods, even those within USDA's jurisdiction, have a good deal of freedom to decide the information to include on a label and how to display it. Label and package size, of course, operate as important constraints on choice, but it is the rare food label that does not reflect the thoughtful efforts of the manufacturer to make the food appear attractive, tasty, or nutritious, and often all three. In recent years, the growing interest in nutrition in general, and such components as fiber, fat, and calcium in particular, has led manufacturers to try to characterize their products as being nutritionally beneficial. Examples

232 NUTRITION MUNG are almost numberless, but the most common are familiar: low calorie, fat free, no cholesterol, fiber rich, and light (or lite). Current sales data reveal that about 32 percent of the packaged foods on the market bear some type of descriptor (FLAPS, 1988~. The verbal formulas used to describe nutrient contents take many forms, but they have been given the name descriptors or, sometimes, adjectival descriptors. The proliferation of terms and the growth in their usage have drawn a good deal of attention from regulatory bodies and health professionals, as well as from competitors. The reactions have been ambivalent, if not schizophrenic. On the one hand, the popularity of these terms probably signals a rising interest among consumers in the links between diet and long-term health. On the other hand, the potential for confusion, exaggeration, and outright deception has prompted some to argue that nutrient descriptors should be forbidden. The problem stems in part from failures in the system for regulating food labels and has several facets. For many descriptors in common commercial use, neither FDA nor USDA has any official definition against which to measure individual product labels. This is not as serious a problem for USDA, which, by virtue of its power to approve labels, is able to resolve the issue through policy development. For FDA, however, the lack of standard definitions for many common descriptors is a serious problem, because the agency has no effective way to prevent manufacturers from using terms that are meaningless or potentially deceptive. A distinction should be drawn between the lack of formal definitions embodied in regulations and the failure to have an age upon definition. FDA has adopted regulations defining certain descriptors, such as terms describing sodium content, but in most cases it has relied on informal advice. For example, agency officials appear to have come to agreement on the meaning of high fiber, but the agency has not adopted any regulation to this effect, and thus, compliance is dependent both on knowing the informal position and being willing to adhere to it. With little risk of enforcement, a manufacturer who holds a different view has little reason to abide by the FDA rules. The problem is amplified where USDA has formally defined a descriptive term but FDA has not. Examples include the terms natwal and lite, and quantitative descriptors of fat content. Compliance with the USDA criteria is likely to be near universal, whereas manufacturers of FDA-regulated foods are as likely to be as concerned about their competitors' actions as about the suggestions or threats of agency officials. Finally, there are descriptors, such as reduced-calorie and reduced-fat, on whose meaning the two agencies differ. In such cases the same term may mean different things on two different foods displayed in different aisles of the supermarket. Without fonnal definitions for common descriptors, some food manufactur- ers have been able to exploit consumer interest in foods that appear to be more healthful. For example, companies have been using the term lite in two very

PRESENT~irION OF NUTRITION INFORMATION ON FOOD LABEL; 233 different ways. On some products it implies reduced calories, fat, or sodium; on others, He term is intended to convey lighter texture, flavor, or even color. For example, Hito Lights, San Giorgio Light 'n Fluffy Egg Noodles, Keebler Crispy Light Crackers, and Wesson Light and Natural Vegetable Oil do not pro- vide fewer calories or less fat than their original counters do; only lighter taste. Yet, an EDA survey of 1,000 adults revealed that 70 percent who had seen foods labeled lite assumed that it meant lower in calories. The rest thought vari- ously that it referred to sugar, salt, fat, cholesterol, or weight (Heimbach, 1982~. The multiple-message problem is compounded in this case, because USDA has adopted a regulation that permits the term life to be used to refer to breading and other components, as well as to the more familiar caloric, fat, or sodium content. Some uses of nutrient descriptors border on the deceptive. It is a common practice to highlight a single desirable component in which the labeled food contains a significant amount of an undesirable component that the food lacks or contains in reduced amounts. Usually, there is no effort to provide a balanced statement on the nutritional characteristics of He food. This practice can be viewed merely as an example of aggressive merchandising, but some examples are arguably misbranding. Many observers, including members of the Committee, are Doubled, for example, by foods with labels that state no cholesterol but that contain substantial amounts of saturated fat. Equally disconcerting, if not so serious, are such claims on products such as bananas and peanut butter, in plaice nature never put cholesterol. The following pages describe the regulatory policies of FDA and USDA, and Table 7-5 details the agencies' key definitions for descriptors. Current Regulation of Descriptors Caloric Content and Bocly Weigh' In 1978, FDA issued a final rule that regulates label statements relating to usefulness in reducing or maintaining caloric intake or body weight The general requirement ensures that nutrition labeling must appear on any food that carries a caloric-related claim. If the product achieves its special dietary character because a nonnutritive ingredient is present in the food, then the ingredient and its percentage by weight must be specified on the label. The regulation also establishes definitions for low-calorie and reduced- calorie foods and other comparative claims. A low-calorie food is defined by an absolute standard: the food must provide less than or equal to 40 calories per serving and less than or equal to 0.4 calorie per gram. The term low calorie or similar phrase must be displayed on the food's principal display panel. A reduced-calorie food must have at least one-third fewer calones, but otherwise be nutritionally equivalent to the food it replaces. Such claims must also be

234 Hi ·e a If ·O .= ~ v v o ·O ·t ~ ~ ~ ~ c a ~ ~ 2 ~ ~ ~ ~ ~ S.;~; 82 i~ 2=C5 2 2 5 ,Y~ ~ ~ ~ _ T.E2 a88 e S 5 5 05 ,- E E S ~ ~ ~ ~ ~ ~ ~ ~ O 5 2 5 S b 2 .' is ~ O O O S a 2.' S ~ 2C,,~ ~ ~ 8 ° 8 ° ~ ~ , ~ ~ ~ 3 .~ ~= EZRA ~ ~ - 31 ° O.n {s o 5 2 ° -A O

235 o C _ or Ct Cal ~ C ~ o ~ o ~ ~ o .= o ~ o 3 .¢ .5 ~ ~ TIC ~ o a ~ ~ o o o ~ 8 00 ~ Rae 00 ~ ~ e-e z z z ~ z S S o ~ ~ ~ o C ~ ·, c ~ ~ ~ y o ° ~ ~ 5 ~ ~ ~ 5 §' ~ ~ ~ i '' ~ ~ ~ 5 ~ ~ == ~ c 00 ~ ~ ~ ~ ·5 ~§ ~ o ,~, S ,~ E! p. 3 ~ 5 ~ = fi 0 = E E ° ° E E E E = E E ~ ° O _ ° E E E i~ E [l ~ ~ 8 E ~ ~ ~i ~ E ~ O E 8 } ~: E 0 c, ~ 0 ~ o ~ 0 ~ O _ ~ _ ~ 0 ~ ~ t3 ~ ~ ~ .5 " ,= V Z V ~ ~i V

236 ·= v a, a, ·t ·~ U ·~ ~ ~ es ~ ·~ ~ Z Cal 4 - u a c a, 00 - ·5 g ·—E Al D~, ~ s 32 Al ·- L) ; ;> am O ~ ~ Z 8 P4 "C = E E ~ o . ' ~ ~ ' ·$ 5 5, 5 5 ~ 3 5 . 5 5 . o 5 ~ 2 -I ~ ~ of o is ~ o g ~ a, REV ~ ~ ~ ~ a_ to ~ ~ _ Us ~ ~ Go ~ ~ ~ cot ~ i-3 ° S ~ ~ he _ ~ O ~ ~

237 ~ o _ d." -A ~ · o ~~ ~ ~— 86 ~ 3 ~ - ~o at o so P. Do A o I, 3 04 .~ a ·3 Z Z Z -e 3 jig, 5~333 , 35 1 3 3 S z /33~13 lo. !. i i t5 3 3 50 3 §,~§ 1 l'4=''l 3 ~ .~ ~ 3 ~ _% it_ CAD o _ Cal _ ~ ~ ~ ~ of

238 ·e ·e U is ._ U to .O ~ Z _ - ' - ' ~ ~1= c ~ s~ 9 - ~= 2 0 ~ ~ ~ o ° ~ ~ ~ ^ 8 E e A 8 5 E ~ 8 8 E ~ E 5 Z 3 3 - ~ g ·E 8 a E e e ^ · ~ ~ ~ .~ .~ ,~ I ~ `1 ~ 5'= · ~ ~ To E To _ .g E ~ ~ 3 8

239 ~ 1 ~ zap = s ~ .,{ . s~ ~ ~ ~ ~ ha- ~ ~2 ~ ~ ~ ~ a ~ ~ ~- o~ i. ~ ~ ~ , , , ~ n ~ ~ ~ ~ e y ~ ~ ~ ~ ~ ~ I s ~ ha .5 :s ~ S To p! So en ts s ~ ~ ~ i | ~ ~ ~ ~ .

240 NUTRITION LABELING accompanied by a statement that clearly describes the comparison upon which the claim is based (21 CFR § 105.66(a), (c), Idly. Terms such as sugar free, sugarless, and no sugar are also defined. FDA suggests that consumers would assume that products with these descriptors would be significantly reduced in calories. Therefore, the food must be labeled low calorie or reduced calorie or carry the comparative claim or the term, or every time the term is used it must be followed by a phrase indicating the product is not a low-calorie food (21 CF1( §105.66(f)~. USDA has guidelines for label claims related to the caloric content of a product in its usefulness for the reduction or maintenance of body weight. Such claims trigger a requirement for disclosing limited nutrition information includ- ing calones, and protein, carbohydrate, and fat content. Additional statements may be required if necessary for consumer understanding. USDA's definition of low calorie is identical to the FDA definition: less than or equal to 40 calories per serving and less than or equal to 0.4 calorie per gram. Reduced calories or other comparative claims using the term lite must achieve at least a 25 percent reduction along with an explanatory statement and quantitative information (USDA, 1982a). Negative claims, such as no sugar, are permitted (1JSDA, 1986~. Cholesterol In July 1990, EDA issued a tentative final rule defining cholesterol describe tars. The definition for cholesterolfree would be allowed on foods that contain less than 2 mg of cholesterol per serving; 5 g or less of total fat and 20 percent or less of total fat on a dry-weight basis; and 2 g or less of saturated fatty acids and 6 percent or less of saturated fatty acids on a dry-weight basis. Low choles- terol would be allowed for foods containing 20 mg or less of cholesterol per serving and 0.2 mg or less of cholesterol per gram of food; 5 grams or less of total fat per serving and 20 percent or less of total fat on a dry-weight basis; and 2 g or less of saturated fatty acids per serving and 6 percent of saturated fatty acids on a dry-weight basis. Reduced cholesterol would be allowed on foods that contained at least 75 percent less cholesterol compared with the level in the food it replaces accompanied by an explanatory statement of comparison (e.g., cholesterol content has been reduced from 100 mg to 25 mg per serving). The notice also provided guidelines for use of the terms cholesterol free food and low cholesterol food, which would not permit the use of defined terms for foods that normally contain low cholesterol, such as applesauce, unless such labels refer to all foods of that type (e.g., applesauce, a cholesterol-free food) (55 ~d. Reg. 29,455-29,473, July 19, 1990). The proposal also would delete the requirement that any reference to the cholesterol content of a food be accompanied by a statement that the information

PRESE~ION OF N~ION INFO~AHON ON FOOD HEM 241 is provided for individuals who are modifying their diets on the advice of a physician, although this statement is little used in recent practice. Also to be deleted is the present requirement that the percentage of calories from fat must accompany fatty acid labeling. A declaration of either fatty acid composition or cholesterol content would require that quantitative information about both be provided, except for low-fat foods. Additionally, if a food with a cholesterol descriptor is represented as a substitute for a traditional food, it must be its nutritional equivalent. USDA has no formal policy on cholesterol descriptors. The agency is waiting for FDA to issue its final rule on cholesterol claims before proposing its own. At present, USDA is using FDA's 1986 proposed definition of low cholesterol, which is less than or equal to 20 mg of cholesterol per serving. No cholesterol claims are not expected to be used for meat products. USDA permits comparative cholesterol claims if the cholesterol content per serving has been reduced by at least 25 percent from a similar product, as described in Agriculture Handbook No. 8 or similar source. USDA rejected FDA's proposed definition of a low cholesterol meal because it allowed for too high a cholesterol level. USDA's working policy limits a low cholesterol meal to no more than 20 mg of cholesterol per 100 g of product (Sally Jones, Standards and Labeling Division, FSIS, USDA, personal communication, 1990~. Fat FDA has no regulations on fat descriptors which is currently under review at FDA. The agency expects to propose a rule on fat descriptors (including fat free, low fat, reduced fat and low in saturated fat) by October 31, 1990 (R.E. Newberry, Division of Regulatory Guidelines, FDA, personal communication, 1990~. However, FDA has announced policy guidelines for low fat and reduced fat claims. The present working definition of low fat is a food that provides less than or equal to 2 g of fat per serving and less than or equal to 10 percent fat on a dry-weight basis. The policy guideline for reduced fat suggests that it may be used for foods that have achieved at least a 50 percent reduction in fat from the regular product as well as display a statement of comparison (Pennington et al., 1990; R.E. Newberry, Division of Regulatory Guidelines, FDA, personal communication, 1990~. FDA has no regulation for lean claims. Rules defining saturated and polyunsaturated fatty acids have existed for some time; the July 1990 tentative final rule defined monounsaturated fatty acids. USDA has issued guidelines for labeling claims concerning the fat and lean contents of meat and poultry products. Emphatic expressions, such as lean, as well as comparative claims, such as leaner, may be used. Low fat products may contain no more than 10 percent fat. Lean products may contain no more than 10 percent fat (except for ground beef and hamburger). Extra lean products are

242 NUTRITION LABF~NG limited to those with no more than 5 percent fat (except for ground beef and hamburger). In all cases, total fat must be disclosed on the label, for example, contains 4 percent fat (USDA, 1987b). If ground beef or hamburger is labeled lean or extra lean, the fat content must be reduced by 25 percent from the regulatory standard of 30 percent fat. Thus, they may contain no more than 22.5 percent fat. The actual fat and lean percentages must be included prominently on the label. USDA illustrates this as follows: ground beef with 20 percent fat could be labeled as '`Lean Ground Beef, contains 80 percent lean and 20 percent fat." Fat percentages, with no fat descriptors, may be provided on labels (USDA, 1982b, 1987b). Comparative claims regarding lean or fat content must be based on at least a 25 percent reduction or difference In fat or lean. If a market-basket survey shows comparable products have a different amount of fat than that in the standard, then the survey must form the basis for comparison. The amount of fat in a similar product as described in applicable references such as Agriculture Handbook No. 8 may be used. An explanatory statement must accompany a comparative claim. For example, leaner Italian sausage could be labeled as, "This product contains 24 percent fat, which is 30 percent less fat than allowed by the USDA standard for Italuan sausage" (USDA, 1987b, p.2~. USDA permits a loophole in these guidelines for fanciful names, brand names, and trademarks that include the term lean. Products, such as lean entrees or dinners, need only provide USDA's abbreviated labeling requirements. USDA assumes that for such foods, the term is used to suggest usefulness in weight reduction or maintenance (USDA, 1987b). To protect consumers from literally watered down a products, lean claims are supposed to be limited to products composed solely of fat and lean material with no added substances such as water or extenders. Comparisons with leading brands, or the company's regular product, are no longer permitted. USDA viewed these comparisons as limited in value, because sometimes the comparison product was unavailable at the same supermarket or was not typical of marketplace products. Lite and Light FDA has no regulations covering the use of lite or similar terms. FDA's working policy considers lite to mean reduced calorie under 21 CFR §105.66, which requires one-~ird fewer calories, unless the term is defined elsewhere (such as a standardized food, or light cream) or is obvious (light synlp as opposed to heavy syrup) (letter to J. Edward Thompson, Kraft, ~c., by Sanford A. Miller, July 7, 1987; R.E. Newberry, Division of Regulatory Guidance, FDA, personal communication, 1990~. The USDA policy memorandum on lite allows food manufacturers leeway

PRESE==ION OF NUTRITION INFORMATION ON FOOD LABEl~; 243 to creatively use such terms as lite, light, and lightly (USDA, 1986). The memo notes that such terms generally imply that a product has significantly fewer calories than expected in the product it replaces. However, the agency condones the use of lite to refer to fat, salt, sodium, breading, andJor other components. The reference component must be reduced by at least 25 percent. USDA requires the term lite to be explained either adjacent to its use or by an asterisk with an accompanying explanation on the principal display panel or information panel. Quantitative information must be provided about the component along with a qualitative comparison to (1) the amount permitted by an applicable standard if it is representative of the majority of products in the marketplace, (2) the amount found in similar products in a market-basket survey, or (3) the amount in similar products as described by an applicable reference source such as Agriculture Handbook No. 8. Products that meet an absolute low standard as defined by USDA may pro vice only disclosure of the actual amount of the substance without a comparative statement. For this purpose, calories can be no more than 40 per serving and no more than 0.4 calorie per gram of product. For fat and breading, the limit for low is no more than 10 percent of the product. For salt and sodium, the product can contain no more than 35 mg of sodium per 100 g of the product to qualify for a low descriptor. Products that use the term lite in the brand name need only meet USDA's abbreviated nutrition labeling requirements. The agency suggests that when lite is used in this manner it is assumed to represent usefulness in weight reduction or maintenance (USDA, 1982a). The Bureau of Alcohol, Tobacco, and Firearms (B. AM:) has jurisdiction over alcohol labeling (45 Fed. Reg. 83,53~83-545, Dec. 19, 1980~. In 1988, BATE proposed a rule to regulate the use of the words lite and light in the labeling (and advertising) of wine, distilled spirits and malt beverages. BATI; proposed two alternative ways a beverage could qualify for a lite claim: (1) if it contains 20 percent fewer calories than the producer's regular product or, if the producer does not make a regular product, 20 percent fewer calories than a competitor's same or similar regular product; or (2) if the product is labeled with both the number of calories in the producer's lite and regular products or, if the producer makes no regular product, the number of calories in a competitor's specifically named regular product (53 ~d. Reg. 22,678, June 17, 1988). This proposal has not been finalized. Organic, Natural, and Fresh FDA has taken no formal position on the use of the descriptors organic and natural. According to a 1978 article, the agency "has not tried to arrive at a legal definition of these terms because enforcement would be difficult or impossible,

244 NUTRITION LABEl]NG and costly. Organically grown foods, once they are removed from the field, cannot be told from commercially fertilized plants" (Stephenson, 1978). In a December 16, 1988, letter to the National Association of State Departments of Agriculture on organic food standards, FDA stated, "We believe it would serve no useful purpose to create standards for foods which are virtually the same, regardless of methods of production" (Food Chemical News, 1988, pp. 1~17~. The agency also has not defined the term natural. Informal agency policy considers natwal to mean that there is nothing artificial or synthetic in the product, and that includes any color (e.g., beet juice in lemonade to make it pink would not be natural) (R.E. Newberry, Division of Regulatory Compliance, FDA, personal communication, 19903. FDA has a policy guideline that defines the term fresh, which states that the term should not be applied to foods that have been subjected to any fan' of heat or chemical processing" (Compliance Policy Guide 7120.06~. Additionally, to avoid misrepresentation and provide information needed to ensure proper storage, food labels are supposed to include in the name or statement of identity appropriate descriptive terms such as pasteurized or frozen. Recently, certain food labels using the descriptorfresh for foods from cooked tomatoes have been challenged by consumer organizations and competing manufacturers; however, the issue has not been resolved Mod Chemical News, 1990~. USDA has no formal policy on organic claims, and it has never approved the use of the term organic on any label (Sally Jones, Standards and Labeling Division, FSIS, USDA, personal communication, 1990~. Though FDA suggests that natural is too difficult to define, USDA has had a policy memorandum defining the term natural since 1982 (USDA, 1982b). Under its policy, natural claims may be made if the product meets two criteria: (1) no artificial flavor, coloring ingredient, or chemical preservative or any other synthetic ingredient is contained in the product; and (2) the product and its ingredients are not more than rrun~mally processed. USDA considers minimal processing to mean any traditional process used to make food edible, such as freezing or drying. Additionally, the agency requires an explanatory statement on the packaging, such as "the product is natural because it contains no artificial ingredients and is only minimally processed." USDA's policy memorandum on use of the term fresh permits its use on any product that is not cured, canned, hermetically sealed, dried, or chemically preserved, that is, not shelf- stable (USDA, 1989a). For poultry, fresh cannot be used on poultry frozen at or below 0°F. The agency makes exceptions if fresh is used as part of a brand name. (Note: There are about 20 states that encourage producers and marketers of organic food through labeling laws and/or programs that certify grow- ers'/producers' claims for food buyers [TDA, 19891. In addition, there was legislation pending in the 101st Congress which contained provisions defining the term organic.)

PRESEw~ION OF Nl~RITION INFORMATION ON FOOD Labels 245 Sodium In June 1982, FDA issued a proposed rule to amend the food labeling reg- ulations concerning sodium labeling. That proposed rule established definitions for four claims: sodium free, low sodium, moderately low sodium, and reduced sodden. It provided for the proper use of these terms in food labeling. The appropriate uses of the terms without added salt, unsalted, and no salt added were also included. Additionally, FDA specified that sodium content of foods be included in nutrition labeling information whenever it was used on food labels. FDA explained that its goals for this program were to increase the availability of, and make more effective, sodium content labeling, as well as to reduce the amount of sodium added to processed foods, when it was safe and technically possible (47 Pled. Reg. 26,580-26,595, June 18, 1982~. Two years later, ADA issued a final rule on sodium content labeling, defining the number and terms as well as the numeric basis for the above descriptors (49 Bed. Reg. 15,510-15,535, Apr. 18, 1984~. Four sodium-related descriptors were established by the final rule. Sodium free may be used to describe foods that contain less than 5 me of sodium per serving (21 CF~ § 101.13(a)~1~. Very low sodium may be used for foods that contain 35 mg or less of sodium per serving (21 CPR § 101.13(a)~2~. Low sodium may be used for foods that contain 140 mg or less of sodium per serving (21 CFR § 101.13(a)~3~. Reduced sodium may be used in labeling if the product contains 75 percent less sodium than the regular product and is represented as a direct substitute for that food. Additionally, if reduced sodium is claimed on a food, its label must also provide a sodium comparison per serving with the original product (21 CFR § 101.13(a)~4~. If any of these four terms is used in labeling, or any other truthful statement about sodium content is used in labeling, the product label must provide quantitative sodium information. This information must be provided as part of full nutrition labeling (21 CPR § 101.9). If the food is represented for special dietary use for low salt or low sodium intake, then either full nutrition labeling or only the number of milligrams of sodium per portion needs to be provided (21 CI;R §§ 105.69, 101.13(b)~5)~. The final rule also permits references on labeling to salt content, such as unsalted, no salt awed, or without added sail. Such terms, or their equivalents, are permitted only if (1) no salt is added during processing, (2) the food it resembles normally is processed with salt, and (3) either nutrition labeling or sodium content labeling is provided (21 CF~ § 101.13(b)~. In its anal rule on sodium labeling, FDA discusses several important aspects of descriptors. The agency noted that the 1982 survey documents that consumers object to the use of too many descriptor terms as too confusing (49 Fed. Reg. 15,51~15,535, April 18, 1984. Yet, truthful and nonmisleading descriptors were described as "useful and desirable for highlighting products, particularly when quantitative information does not appear on the display panel.

246 NUTRITION LABELING A total of 93 percent of respondents to the 1982 FDA consumer survey who reported concern about sodium intake wanted descriptor labeling in addition to quantitative content. FDA concluded that its sodium definitions were "simple terms that are easily understood and that will not mislead the consumer." In situations in which manufacturers use descriptors that are not defined in its final rule, the agency has, basically, punted. For example, one comment on the proposed rule asked FDA to prohibit the use of undefined claims such as "naturally low in sodium." It strongly urged manufacturers to use only those descriptors defined in section 101.13(b) to minimize consumer confusion. It did not limit manufacturers in any other way, however. FDA mentioned its case-by- case review for false and misleading claims as a means to protect consumers from labeling pandemonium. USDA policy on sodium labeling differs from the FDA approach. USDA requires quantitative information on the sodium content per serving only if a claim is made about sodium and/or salt content of a meat or poultry product (USDA, 1984~. Otherwise, sodium content information is provided on a volun- tary basis. Like FDA, USDA permits manufacturers to provide sodium content information without other nutrition information on foods for special dieted use. The USDA definitions of very low sodium, low sodium, sodium free, and unsalted, no salt added, or without added salt, or equivalent terms, are identical with the FDA definitions. USDA also allows reduced-sodium claims to be used if the product has achieved a 75 percent reduction in sodium content and the label provides quantitative information comparing it to the original product. Like FDA's sodium labeling regulations, USDA's policy memorandum on sodium specifically describes how a comparative sodium claim may be used, such as: "This bologna has 25 percent less sodium per serving than our regular bologna" (USDA, 1984~. Comparative claims may not be made unless (1) a product's sodium content is at least 25 percent less than the product to which it is compared and (2) the comparative claim is accompanied by (in immediate conjunction with the claim or referenced by an asterisk) an identification of the productts) with which the comparison is being made and a quantitative statement of the difference in sodium content per seeing (using equivalent serving sizes) of the products being compared. USDA encourages comparues to reduce the sodium contents of their products to even lesser amounts, though such foods would not qualify for sodium comparative claim labeling. USDA's earlier sodium labeling policy contained provisions for the qual- it~ive sodium claims before FDA had such regulations. Some of them were adopted by FDA, such as unsalted, no salt added, and without added salt (Wolf et al., 1983~.

PRESENTATION OF NUTRITION INFORMATION ON FOOD LAB Fly 2A7 No or Negative FDA has no written policy on negative descriptors and no plans to develop such a policy (R.E. Newberry, Division of Regulatory Guidance, FDA, personal communication, 1990~. In 1987, USDA modified its policy on negative ingredient labeling, because the agency believed that negative descriptors can be useful and meaningful to consumers as an aid in understanding product contents. Also, such claims offer a simple and direct means of alerting consumers to the absence of ingredients they might not want to consume for health, ethnic, or personal reasons (USDA, 1987a). The phrase no preservatives exemplifies negative descriptors. USDA guidelines permit negative labeling if it is not clear from the product name that the ingredient is not contained in the product. The agency uses as an example the term no beef on the label of turkey pastrami as an example. Negative labeling is also allowed if the food processor can show that the statement would be beneficial for health, religious preference, or similar reasons. Other statements are also allowed to describe product packaging, such as no refrigeration needed. Negative labeling may also be used to highlight the absence of ingredients that are prohibited by regulation, as long as the label prominently indicates this absence. USDA gives the following example for ground beef: "USDA federal regulations prohibit the use of preservatives in this product" VISTA, 1987a). FDA Descriptors for Shelf Labeling in Grocery Stores Some retail grocery stores have wanted to provide consumers with point- of-purchase shelf labeling nutrition information. At the request of Giant Food, Inc. (Washington, D.C.), and other stores, FDA has, on a case-by-case basis, developed policies that exempted stores from certain nutrition labeling require- ments on a temporary basis to permit in-store food labeling experiments. The terms used by Giant Food, Inc., are listed in Table 7~. International Use of Descriptors Descriptors are used in many countries through the world. Canada and the United States have defined a larger number of descriptors than other countries or international entities Trough efforts to achieve consistency and address significant health matters. The Council of the European Community and the Codex Alimentarius Commission have proposed or defined terms for reduction or absence of nutrients, calories, sodium, natural, organic, Desh, wholesome, healthful, and sound. An overview of the terms Canada has defined appear in Table 7-7. Further discussion of the international use of nutrition labeling is provided in Appendix C.

248 NUTRITION LABELING TAs~ F 7-6 Descriptive Terms Used in a Supermarket Shelf Labeling Programa Term Cntena Per Serving Micronutrients Fair source 10 percent or mote of U.S. RDA Good source 25 percent or more of U.S. RDA Excellent source 40 percent or more of U.S. RDA Fiber Source 2 g Good source 5 g Excellent source 8 g Calcium Source 100 mg Good source 250 mg Excellent source 400 mg Low calorie 40 calories or less and 0.4 calories/g or less Low fat 2 g or less of fat and less than 10 percent fat on a dry weight basis a Adapted from letter to 0. Mathews, Giant Food, Inc., Washington, D.C., by S.A. Miller, Center for Food Safety and Applied Nutrition, FDA, DHHS, July 16, 1986, and letter to J.S. Kahan and B.L Rubin, Hogan and Hartson, Washington, D.C., by RJ. Ronk, Center for Food Safety and Applied Nutrition, FDA, DHHS, August 26, 1988. Summary of the Current Use of Descriptors It is not easy to summarize the problems associated with the current widespread use of nutrient descriptors on food labels. The variety of terms used makes generalization difficult, and efforts to group the terms into smaller, similar categories is challenging. No doubt this explains, in part, the two agencies' failure to establish official definitions for more of them, much less to agree upon a framework for establishing definitions for terms that may be used in the future. The central concern, in the Committee's view, is that the unregulated use of a growing variety of nutrient descriptors will nullify the efforts of consumers to make intelligent use of the factual information required on the nutrition label. There is a second concern as well, however. The absence of authoritative definitions for many descriptors works to the disadvantage of manufacturers that are reluctant to use terms that distort or exaggerate nutritionally unimportant differences. If calling a product high fiber produces a 0.5 percent increase in market share, the temptation to exploit this consumer interest will be difficult to resist. Accordingly, the Committee believes that FDA and USDA could contribute to consumer understanding of nutrition labeling by adopting and enforcing

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEl~; TABLE 7-7 Criteria for Descriptive Terms Used in Canadaa 2~49 Term Criteria Low calorie Low calorie food Calorie reduced Low fat Fat free Low saturates Low cholesterol Cholesterol free Sodium free Low sodium food Fiber Source or moderate source High source Very high source Descriptive terms Contains or source High or good source Very high or excellent source Less than 1 calorie/100 g Greater than 50 percent reduction, less than 15 calories per serving, and less than 30 calories per reasonable daily intake Greater than 50 percent reduction No more than 3 g of fat per serving and not more than 15 percent fat in the dry matter (about 30 percent calories from fat) Less than 3 g per serving No more than 2 g of saturated fatty acids per serving and not more than 15 percent calories from saturated fatty acids No more than 20 mg of cholesterol per serving and per 100 g and low in saturated fatty acids No more than 3 mg of cholesterol per 100 g and low in saturated fatty acids Less than 5 mg/100 g Greater than 50 percent reduction and less than 40 mg/100 g with some exceptions At least 2 g/serving At least 4 glserving At least 6 g/serving At least 5 percent of RDIb At least 15 percent of RDI (30 percent for vitamin C) At least 25 percent of RDI (50 percent for vitamin C) a Steele, PJ., and M.C. Cheney. 1989. Canada's system of nutrition labeling. Rapport 4(2):1-2. b Recommended Daily Intake.

250 NlTI771TlON LABORING official and uniform definitions for a much larger number of nutrient descriptors. For many descriptors it will not be possible to say with confidence which of several possible definitions is the best. It seems improbable that consumers possess clear notions about the meaning of different descriptors. The choice of definitions, in some cases, will be arbitrary; the key is to be sure a definition is chosen. The Committee does not believe that the exercise must always be arbitrary. There is a category of descriptors, comprising many of those now in wide use, for which it would appear to be possible to establish generic definitions. These are terms used to characterize the relative amount of a component in a serving of food, such as low-fat and high-fiber. As these examples suggest, there appear to be two subcategories for which different criteria probably apply. One category includes quantitative descriptors of nutrients whose consumption should be encouraged, such as vitamins, minerals, and carbohydrates, and the second would include descriptors of nutrients whose consumption should be controlled or curtailed, such as fat, cholesterol, and sodium. This section incorporates this distinction in suggesting a uniform set of criteria for quantitative descriptors of such components as fat, cholesterol, fiber, and complex carbohydrates. However, in those instances in which FDA or USDA has already adopted official definitions (and there is no conflict between the agencies), the Committee recommends that no change be made. Stability in meaning is more important than theoretical consistency across nutritional components. It is the Committee's hope that such criteria might provide a general framework for establishing official definitions for quantitative descriptors (sometimes termed nutrient content claims) in the future. Suggested Framework for Defining Descriptors If present practices offer any lessons, innumerable descriptors and deriva- tives thereof may be proposed in the future. To control future proliferation and possible misunderstanding of these descriptors, the Committee suggests that all such descriptors could be assigned to one of three classes: (1) those that al- lege or imply a health benefit, (2) those that describe various other features and that require specific definitions and criteria, and (3) those that describe relative amounts of nutrients and other constituents. The first group of descriptors~hose that imply a health claim (e.g., heart healthy or safe for didbetics~resumably would be regulated under the current proposed FDA regulations for public health messages (55 Fed. Reg. 517~5912, Feb. 13, 1990~. The second group of descriptors represents a more diverse and creative list of terms that characterize other features (e.g., fresh, natural, or organic). This group ought to be defined on the basis of reasonable and reliable scientific evidence. Indeed, there is legislation currently pending before Congress to establish a legal definition for the term organic.

PRESE~r~lON OF NUTRITION INFORMATION ON FOOD LABELS 251 The third group consists of descriptors that relate to the quantity of nutrients and other constituents. The following proposal is intended to cover all food constituents—nutrients and nonnutrients for which there may be future interest in this quantitative category. Therefore, the term nutrient wit1 be used generically to reflect all possible food components, regardless of whether there are current published RDAs or other dietary standards for their consumption. The message conveyed by quantitative descriptors should be consistent, clear, and reliable. Consistency can be obtained by including under the umbrella of the following criteria all nutrients contained in all food products. This is a reasonable assumption because low sodium, for example, should have the same meaning, whether it is applied to soup, frozen peas, or meat. Clarity of message may be achieved by using simple rules. Thus, the Committee suggests that descriptors be limited to two categories of nutrient contents that deviate from the norm, that is, low and very low, or high and very high. The first category, demarcated by words such as low or high or their equivalents, should represent a relative nutrient content that can be expected to have significantly different biological effects, whereas the second category, very low or very high, is demarcated by a level one-half (or twice3 this amount. A third category is obvious but pejorative and irrelevant in the marketplace. Although it might be useful to consumers if they had some way of knowing that a food contained a low level of a desirable nutrient or a high level of an undesirable nutrient, manufacturers would have no interest in describing a food as containing an average or low level of a desirable nutrient or an overage or high level of an undesirable nutrient. Descriptors that depict nutrient amounts should be indexed against author- itative dietary recommendations for upper and lower benchmarks. This notion suggests that nutrient intake recommendations should be expressed as ranges, not as single numbers, because they reflect the fact that every nutrient or other dietary constituent is toxic when consumed at high enough levels. Ranges are es- sential if regulation of these quantitative descriptors is to be consistent rather than ad hoc. Beneficially low levels of nutrients should be indexed against recom- mended upper limits, whereas beneficially high levels should be indexed against recommended lower limits. Accordingly, a consistent and reasoned scheme ap- plicable to all nutrients requires the development of consistent and rationally developed ranges of optimal intakes for all nutrients. Unfortunately, except for two vitamins and five minerals with ESADDIs (NRC, 1989b), dietary intake recommendations are not given as optimal intake ranges for most nutrients. For most nutrients (protein, 11 vitamins, and 7 minerals), only single numbers (i.e., RDAs) are published, which are designed for the maintenance of good nutrition of practically all healthy people in the United States. These single numbers are not minimum requirements for individuals. In He absence of alternative reference limits, however, for purposes of illustration, these RDAs may be considered as comparable lower benchmarks

252 NlTI771TION LABEIING for the various nutrients to which they apply. For a few nutrients (total fat, cholesterol, sodium, and dietary fiber), recommended upper benchmarks have been published, although traditional RDAs do not exist. This suggested scheme, as it applies to a few nutrients, is illustrated in Table 7-8. It is important to emphasize the tentative nature of the suggested descriptor reference points; they have been selected only from varied and some- what f~nentary information to illustrate the scheme and to give approximate estimates of descriptor benchmarks. If the logic of this scheme is appealing, it would be necessary to produce appropriate benchmarks for all nutrients. The scheme illustrates the need for a broader definition of nutrients. If minimal in- takes of constituents such as dietary fiber, carotenoids, total fat, and any other naturally occurring food constituents are essential for optimal health, it would be reasonable to establish RDAs or some equivalent standard for them, espe- cially when current dietary recommendations provide clear indications on such values as the percentage of calories from fat and desirable fiber intake. Also, it will be necessary to develop rational benchmarks for each of the four categories of descriptors (high and very high, low and very low3. The Committee suggests that these benchmark limits be kept conceptually consistent for all nutrients in order to simplify the message. This scheme could also lend itself to establishing the terms of reference and criteria for nutrition information panel descriptors. For macronutrients, the benchmark criteria could be based on the existing FDA criteria of excellentlvery good source (greater than 40 percent), good/high source (greater than 20 percent), low source (less than 2 percent), and very low source (fess than 1 percent) of the standard for a nutrient. Thus, macronutrient content claims on the principal display panel would be supported by the quantitative values listed on the nutrition information panel. Similarly, the criteria proposed by the Committee for the listing of micronutrients on the nutrition information panel very good source of (greater than 20 percent), good source of (11 to 20 percent), and contains (2 to 10 percent~ould serve as the same criteria and descriptors for the micronutrients on the principal display panel. The micronutrients would not be supported by quantitative values on the nutrition panel, but the reference range (e.g., 11 to 20 percent of standard) would be provided based on the rationale provided in the section on U.S. RDAs in this chapter. Use of the terms no or free could be based on the criteria that macronutrients so classified on the label are present at a level that is less than 1 percent of the maximum benchmark. Comparative Descriptors Descriptors also have been used to compare similar products, although the product being compared is not always identified. The Committee recommends that use of comparative descriptors be strictly regulated to ensure that He products being compared are clearly identified and that the extent of nutrient

253 oo m _ i§A ~ ~ ·s .a ~ E S ~3 C) ~ ~ a; _ - 3 V a_ ,~ ° ~ V, ~ 3 ~ m _` A z ~ fig S 3 o £ ._ Cot o ._ o flood on 1 U! C, ~ ~ ~ ~ ~ oo ~ 8888 °8 ~ 1 1 1 1 .= 8 `: _ ~ ~ ~ ~ ~ ~ I I l l E or! ~ ._ at: I I ° ~ y i~ He~- ° 8 s ~ e 2 2 ~i ~ ~: ~ ~ 53 o x ~ ~ 3 ~ C '' ~ ~ 3 ~o o ~ ~c ~ ~ x ° E F E 3 E 3 ~ E ~ ~ 3] 3 E E ~ ~ S ~ ~ ~ ~ ~ ~, ~ ~ ~ ~ ~ 13 ~ li3 t3 -, 0 0 0 ~ —0 i3 ~ .o {., ~ ~, ~ 00 ~ ._ ._ ~ ~ O.l _ Z 8 1 1 1 za ~ ° 1 ~ I1 ~ ~ ~ ~ o oo 1 1 ~ c~ ~ C`1 1 1 ' ' 1 1 ~ o ~ o

254 NUTRITION LABELING modification is specified. The Committee suggests that for the use of comparative descriptors (e.g., reduced and lower than), modifications of at least 20 percent be required. Descriptors should not be allowed for nutrients unless they are normally present in physiologically significant amounts. A suggested benchmark for a physiologically significant amount might be the same 1 percent of the maximum allowance used to define the limits for very low in Table 7-8. On the basis of this reasoning, a no cholesterol descriptor would not be allowed on foods that do not normally contain cholesterol, because they normally would contain less than 1 percent of the maximum daily allowance in a seeing. Such a scheme would provide a logical and consistent system for regulation of quantitative descriptors of food constituents which may be proposed or discovered in the future. Without some such a scheme, regulation will, of necessity, remain ad hoc and invite continued confusion. LABEL FORMAT OPIIONS This section examines the history of the current label format, sets forth the Committee's criteria for label redesign, and proposes specific examples of improved labels. Recommendations are made for a mechanism for testing label revisions prior to implementation in the next section. Selection of Current Label Format In 1972, when the nutrition labeling program was in preparation, FDA investigated the various options that might be appropriate for the nutrition panel of the food label. A study conducted for FDA sought to dete~n~ine the best way to convey the nutritional value of a food to consumers (CRI, 1972~. In that study, three methods were tested: (1) a numeric system, (2) a verbal system, and (3) a pictorial system. In the numeric system, the amount of the RDA provided by the product for each of seven key nutrients was presented as a numeric percentage. In the verbal system, adjectives were used to rate the product as a source of a nutrient. In the pictorial system, symbols such as stars or smiling faces were used to indicate the quantity of nutrient present in the product. The study measured consumer reaction to all three systems on (1) how well the information could be understood and used in making food purchase decisions, (2) how each method affected the actual purchasing behaviors of consumers, and (3) how well each of the three alternative formats was liked. A majority of those surveyed, including consumers representing underprivi- leged minority groups, seemed to understand the nutrition information regardless of the method used. Consumer responses to all three systems were reported as encouragingly high, but the verbal system was consistently less well received by consumers than the other two format systems were. There were no major differences among the formats in terms of influencing product choice, although

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABELS 255 consumers indicated that, under certain conditions, they would switch to prod- ucts that were nutritionally superior. Consumers reported that they preferred nutrient information presented in terms of numeric percentages rather than in terms of words or pictures, which were considered too vague to communicate precise nutrition information. Consumers stated that they expected the nutrition information to be precise (even though it cannot be; see Chapters 4 and 5) and felt that words were too vague for this purpose. In addition, because they con- sidered nutrition to be a serious subject, consumers felt that use of symbols such as faces was childish and condescending. On the basis, at least in park of these results, FDA decided to use numbers and percentages, which were determined to be the best of the three alternatives under consideration. The usefulness of listing nutrient information for all key nutrients, regardless of their presence in the product, was tested. Consumers seemed to find it easier to determine the best nutritional value when the label listed only nutrients that were actually present rather than all key nutrients, even when some were not present. Consumers provided with information on the percent composition of fat, carbohydrate, and protein in a product showed a tendency to switch to products with lower fat contents compared with consumers who did not receive such information. The results of this study influenced FDA's decision to select the current label format using the numeric system that presently appears on the current nutrition panel. As part of the interagency review of all aspects of food labeling in the late 1970s, FDA commissioned a study to explore nutrition label format alternatives. That study developed five alternative labels for consumer testing which included the present label and four others described as simplified numeric/numeric, simplified numericherbal, simplified numeric/graphic, and simplified graphic/graphic (or unitary nutrient density) (USDAIDHHS, 1982~. Unfortunately, further consumer testing of these label alternatives was never funded. Experience with Other Label Formats Some research has assessed how well the current label format and vari- ous alternatives convey nutrition information to consumers. Studies conducted before nutrition labeling was instituted were primarily focused on ascertaining consumers' reactions to the novel concept. The early research did not evaluate certain alternative formats, such as various graphic presentations. However, later studies did test such formats, including nutrient density, but none of the studies on nutrition labels tested various elements of a nutrition label in graphic format against themselves or against other elements. A review of these format studies is included in the section "Consumer Understanding of Nutrition and Use of Food Labels" in Chapter 4 (see Asam and Bucklin, 1973; Babcock and Murphy,

256 NUTRITION MING 1972; Betteman, 1979; CNCPL, 1990; Geiger et al., 1990; Hammonds, 1978; Lenahan et al., 1972; McCullough and Best, 1980; Mohr et al., 1980; Muller, 1985; ORC, 1990; Rudd, 1986, 1989; Vankatesan, 1977, 1986; Yankelovich, Inc., 1971) Revised Nutrition Label Information The studies cited above have established that consumers want more relevant nutrition information on the products they purchase, even though they display less than a complete technical understanding of current label information. Further improvements in consumer ability to make dietary choices will be seriously hampered if deficiencies in the current labeling requirements are not corrected. Daly (1976) concluded that unless product information is easily accessible at the point of purchase, can be easily understood by the consumer, and is presented in a format allowing direct comparisons of alternatives, it is unlikely to be used in making food choices. In evaluating various label format alternatives, the Committee used the following criteria: health relevance of content, clarity to consumers, consistency, space requirements, and compatibility with existing labeling practices. Relevant Content Info,7nation Declaration of nutrient content information on the label should reflect the goals of the current dietary recommendations as summarized in the Surgeon General's report (DHHS, 1988) and the NRC Diet and Health report (NRC, 1989a). Although knowledge of the relationship between nutrition and long- term health will continue to evolve, these two reports set forth a reasonable consensus for action. The objective of label revisions suggested in this report is to provide consumers with the food label information necessary to apply these dietary recommendations to their food purchase and consumption decisions. The Committee's recommendations for information on nutrient content are based on the discussion in Chapter 6. Included are recommendations on the disclosure and presentation of toad calories, fats, cholesterol, carbohydrates, protein, fiber, sodium, potassium, calcium, iron, and other micronutrients. Contained earlier in this chapter are discussions of and recommendations for serving size and qualitative disclosure of micronutrients. Clarity of Information In addition to standardizing and limiting the number of categories of serving size and providing for the organized grouping of fats and carbohydrates, several other issues are of concern that have an impact on label formats.

PRESE==ION OF NUTRITION INFORMAT70N ON FOOD FEW 257 Units of Measurement The appropriate measure for most macronutri- ents is grams. However, for macronutrients such as cholesterol, sodium, and potassium, the measure should be milligrams, to avoid the need to use decimal declarations. The Committee recommends that components listed in milligrams be grouped together following the macronutrients declared in grams. Nutrient Groupings For clarity of information and to facilitate the education process, consumers should expect that nutrients will be presented in logical groupings. Consumers should also expect that the quantities declared for nutrient subgroupings added together will equal the amount appearing on the line for the group as a whole (e.g., required fat components [except cholesterol] and, if provided, carbohydrate components [except fibers. Presentation Issues Consumer understanding of label information is undoubtedly influenced by the manner of presentation. Many of the formats that have been suggested are modifications of the current format, which either expand on the existing components (breakdown of fats) or add new items not previously required (fiber, cholesterol) to the list. Several visual representations (graphics) of nutritional value have been suggested, but their usefulness to consumers has yet to be successfully demonstrated. Graphics include such options as bar graphs, pie charts, and symbols. Use of graphics for a combination of several nutrients makes it difficult to convey the information that consumption of some nutrients, such as complex carbohydrates, is to be encouraged whereas others, such as saturated fatty acids, are to be discouraged. Graphics based on a calorie reference require selection of a single calorie consumption standard, even though individuals vary widely in their requirements. A decision must also be made as to whether graphic information is to be expressed with or without water. For example, on a weight basis, milk, including the water, makes the product appear to be very low in protein, whereas excluding the water makes it appear to be very high in fat and carbohydrates. Space is another serious constraint for any graphic format. A graphic representation would result in a significant expansion of the size of the nutrition information panel, since graphic presentations must be supported by numeric information. As a result, the Me size for the numeric section would be reduced on many products, and the number of products of a size too small to support the revised label would increase the number of products that would be exempt from mandatory labeling. Another problem with graphics is that manipulations of the scale used, i.e., ratio of height to width, can produce serious visual distractions and consequent potential for abuse. For these reasons, the Committee recommends that the use of any graphic representations be optimal. An additional presentation issue concerns the provision of dietary recom- mendation standards against which consumers can compare the nutritional value

258 NUTRITION LABELING of a food to their dietary pattems. The current consensus on dietary recom- mendations provides a set of standards for the provision of such information to make such a comparison. However, the listing of such information will take up precious space on the label, requiring another column of comparative values. Consistency Consistency has Free dimensions: internal consistency, consistency of format across products, and consistency over time. Internal consistency means that the same measurement should be used for all similar groups of nutrients. All fats, for example, should be expressed both in grams and as calories from fat. Consistency of format across products means that all nutrition labels should list the same nutrients in the same order. A zero declaration, in other words, is preferable to omission. As a result, consumers would be presented with a familiar format each time, thus simplifying education programs. Consistency over time means that the format selected should not include information about constituents that do not yet have a well-established diet-health relationship and, therefore, require frequent revision. Consistency of format becomes more critical with the age of the user. Younger individuals tend to be better able to separate relevant information from a cluttered presentation. Older individuals are more field dependent; that is, they rely more on consistency of placement and presentation to aid their information gathering. Cross-sectional studies indicate that after people reach their late 30s, the rate of change toward greater field dependence accelerates (Cole and Gaeth, 1990; Eisner, 1972~. Space Requirements Although space limitations should not be decisive when there is a com- pelling health reason for including information, space is an important consid- eration in designing any label. Currently, over half of the food packages that bear nutrition labeling confine this information to an area no larger than 2 square inches, which is the average space allotted to the nutrition panel on food packages (FLAPS, 1988~. A significant expansion of space requirements would necessitate major redesign of many labels, delaying compliance and increasing food costs. Space requirements are a major problem with graphic displays. Although these should be allowed on a voluntary basis, as mentioned above, the Com- mittee aloes not recommend that they be required. There may be a perception that manufacturers could enlarge the size of the package to accommodate any new labeling requirements, but this perception is distorted Government limits on

PRESENTATION OF NI~RlTION INFORMATION ON FOOD LABEL 259 slack fill of packages place an upper limit on the size of the container. Even when this is not a factor, consumer perception of deception due to slack fill limits the freedom of the manufacturer to expand the size of the package at will. Compatibility with Existing Labeling Practices To the extent that recommendations for label reform can be made compatible with existing practices, the costs of compliance will be minimized, the speed of compliance will be enhanced, existing pavate-sector educational materials will remain useful, and consumers will not need to relearn a different protocol. It is important to recognize that consumers have a 17-year investment in the current label format. Consumers have become accustomed to the current format over a long period of time, and a wide variety of information and educational programs have been developed to support the understanding and use of current label information. In addition, manufacturers also have experience in using the current label format. Although familiarity should not prevent beneficial change, neither should this 17-year investment be abandoned lightly. Committee Recommendations A number of recommendations were made in the previous sections and chapters concerning the content and manner by which to better convey nutrition information to consumers on food labels. To summarize those issues that apply to format changes, the Committee recommends that: · Sewing size should be prominently displayed on the nutrition information panel and should appear in household units. · The amount of the serving should appear in grams or milliliters in paren- theses following household units. · Nutrient information should appear for the food as it is packaged, with the option of providing information relevant to the manner in which the food is prepared. · Macronutrients should be listed in grams or milligrams. · Macronutrients should be listed first, and then other food components, electrolytes, and micronutrients, and similar food components should be grouped together, except dietary fiber and cholesterol should not appear in groupings. · Various issues related to placement and prominence of food components on the nutrition label (e.g., increased prominence of fat components, decreased prominence of protein, and ordering of macro- and micronutrients) should be subjected to consumer testing. · Comparison with dietary recommendations should be optional.

260 NUTRITION LABELING Sample Formats The sample formats in Figure 7-1 illustrate He Committee's recommenda- tions for nutrient content disclosure on and format of the nutrition information panel. The Committee is not endorsing any one of these samples; they serve only to illustrate ways in which the various recommended components could appear on a nutrition panel. Given the Committee's recommendation for the need to test label formats, graphic options were not included. TESTING OF LABEL FORMATS Although there is considerable information about the beneficial changes that might be made in the content and formats of nutrition labels, consumer understanding and acceptance of any new design will determine the success or failure of this effort. This information cannot be determined without direct tests of the proposed revisions. The Committee believes that alternative label formats should be subjected to both qualitative and quantitative consumer testing prior to issuance of any final nutrition labeling requirements. However, this testing must be carefully structured to produce measurable results and, given the level of expectation for this process, must be carried out within a reasonable period of time. It is also assumed that before any testing procedures begin, the agencies will have determined through He comment and rulemaking process the nutrient content information to be conveyed on the label, as this will affect the required nu~idon information to be presented in any format tested. As to the experimental design, it would be a mistake to structure format comparisons amounting to little more than popularity poll choices among alternatives. First, the revisions must be based on sound science. That is, the objective must be to provide He kind of label information to the public that will allow consumers to apply current dietary recommendations to their everyday food purchase and consumption decisions. Therefore, all formats to be tested should be consistent with this objective. For this reason, it is unnecessary and unwise simply to test all label revisions that have been proposed. A screening process must first be carried out to eliminate formats that do not pass the tests of sound science and at least an initial screen of reasonableness. Second, the level of detail provided should be consistent with the type of information needed to make informed dietary choices and a reasonable size requirement for labeling. Consumers have a strong tendency, in purely attitudinal surveys, to favor greater levels of detail, even when that detail is of questionable relevance (ORC, 1990~. That is, consumers tend to say Hey want more information, regardless of whether Hey are likely to use it. However, unnecessary detail would expand the size of the nutrition information panel and

PRESENTATION OF NUTRITION INFORMATION ON FOOD LABEl~ 261 would therefore exempt a higher percentage of foods in packages too small to accommodate the new labeling requirements. Third, the popularity of a format alone is not enough to recommend its use. Consumers must be able to use the new format to make improved dietary choices. After an initial screening of label formats for preference, it is therefore essential that actual formats be tested in an environment closely approximating marketplace conditions. For this to take place, some minimal level of education will need to be given to the test subjects. Since new label formats will be used, test subjects will need some instruction as to their use. If this is not done, clearly inferior formats might surface as the labels of choice simply because they are similar to a preexisting frame of reference. The Committee recommends that: A brief test panel education program reviewing the current dietary rec- ommendations and explaining the basics of the new label formats should precede label format testing. · A format testing procedure should be initiated that ensures adequate con- sumer input and evaluation of whichever label format is chosen as the stan- dard and should include the following elements: 1. An advisory panel, consisting of individuals familiar with dietary guidance and an understanding of how consumers use label infor- mation to make decisions, should be assembled to work with the relevant agencies. This group should be charged with assisting the agencies to select the label formats to be tested and with overseeing the nature of the testing process. 2. The first stage in any testing process would be the mock-up of several label formats in order to submit them to comprehensive qualitative review by consumers. Such a review would probably be performed by using a number of focus groups of consumers with varying interest and ability levels to determine their preference for the amount of information presented and the label format of choice. 3. In-depth consumer evaluation of the most preferred choices would be necessary to assess consumers' ability to apply the food label in- formation in making food choices. Individual protocols to determine how they process the label information can be used. Such protocols to determine cognitive responses to nutrition information have pre- viously been developed and tested (Sims and Shepherd, 1987~. By such techniques, it is possible to determine before-and-after label use. 4. On the basis of results from the cognitive response protocol test- ing, revisions to the label format should be made. Large-scale sur- veys can then ascertain overall consumer acceptance and compre-

262 NUTRITION LABELING B 2% LOWFAT MILK Nutrition Information Per Serving SERVING SIZE ONE CUP SERVINGS PER CONTAIN ER 8 CALORIES PROTEIN CARBOHYDRATE 120 8 GRAMS 1 1 GRAMS 5 GRAMS SODIUM 130 mg Percentage of U.S. Recommended Daily Allowances (U.S. RDA) PROTEIN 20 RIBOFLAVIN 25 VITAMIN A 10 NIACIN * VITAMIN C 4 CALCIUM 30 THIAMINE 6 IRON * 'CONTAINS LESS THAN 2% OF TH E U.S. RDA FOR THESE NUTRIENTS 2% LOWFAT MILK Serving size 1 cup (8 fl oz) Servings per container Nutrition Information Per Serving Calories 120 Total Fat 5 9 (45 kcal) Saturated Fat 3 9 (27 kcal) Unsaturated Fat 2 9 (18 kcal) Carbohydrate 11 Protein Total Dietary Fiber Cholesterol Sodium 9 9 0 9 20 mg 120 mg A very good source (over 20% [standard]) of: Calcium. FIGURE 7-1 Current nutrition label and samples of revised nutrition labels based on the Com- mittee's recommendations. Ike U.S. RDA was used as the standard for source of listings. (A) Sample nutrition information panel for 2% lowfat mink (~/2 gallon) under current FDA regulations (minimum requirements). (I3) Sample nutrition information panel for 2% lowfat mink (Y2 gallon) incorporating the Committee's mandatory content recommendations. Total dietary fiber is included on the nutrition information panel, but could be exempted for milk products (see Chapter 6). Iron is not declared in the source of listings because its value is less than 2 percent. (C) Sample nutrition infonnatian panel for 2% lowfat milk (~/2 gallon) incorporating the Committee's mandatory and voluntary recommendations. Note that carbohydrate changes to total carbohydrate when its com- ponents are listed. Listing of complex carbohydrate and sugar content are optional. Declaration of calories from total carbohydrate, complex carbohydrate, sugars, and protein are optional. Aside from calcium and iron (not listed; less than 2% standard), all source of listings of micronutrient content are optional. (D) Sample nutrition information panel for macaroni and cheese dinner incorporating the Committee's mandatory and voluntary content recommendations contained in Sample C, and the optional as prepared format.

PRESEaNT~ION OF NUTRITION INFORMATION ON FOOD LABEl~; D 2% LOWFAT MILK Serving size 1 cup (8 fl oz) Servings per container -- Nutrition Information Per Serving Calories 120 Total Fat ~ 9 (45 scat) Saturated Fat 3 g (27 kcal) Unsaturated Fat 2 9 (18 kcal) Total Carbohydrate 11 g (44 kcal) Complex Carbohydrate O g (Okcal) Sugars 11 9 (44 kcal) Protein 9 9 (36 kcal) Total Dietary Fiber O 9 Cholesterol 20 mg Sodium 120 mg Potassium 430 mg A very good source (over 20% [standard]) of: Vitamin D, Calcium, Riboflavin, Phosphorus. A good source (11-20% [standard]) of: Vitamin A, Vitamin B12. Contains (2-10% [standard]): Vitamin Be, Vitamin C, Magnesium, Pantothenic Acid, Thiamin, Zinc. MACARONI & CHEESE DINNER Serving size (as prepared) 3/4 cup (50 9) Servings per container 4 Nutrition Information Per Serving As Packaged As Prepared Calories 190 290 TotalFat 2 9 (18kcal) 13 9 (117kcal) Saturated Fat 1 9 ( 9 kcai) 9 9 (81 kcai) Unsaturated Fat 1 9 ( 9 kcai) 4 9 (36 kcal) Total Carbohydrate 36 9 (144 kcal) 34 9 (136 kcal) Complex Carbohydrate 30 9 (120kcal) 28 9 (112kcal) Sugars 6 9 (24 kcal) 6 9 (24 kcal) Protein 9 9 (36 kcal) 9 9 (36 kcai) Total Dietary Fiber 1 9 1 9 Cholesterol 5 mg 5 mg Sodium 425 mg 525 mg Potassium 850 mg 900 mg As Packaged A very good source (over 20% [standard]) of: Niacin, Riboflavin, Thiamin. Contains (2-10% [standard]): Calcium, Iron. As Prepared A good source (11-20% [standard]) of: Riboflavin, Thiamin. Contains (2-10% [standard]): Vitamin A, Calcium, Iron, Niacin. 2~3

264 NUTHTION LABELING hension of the label information. In addition, testing of the label win consumers under conditions approximating marketing environ- ments should be carried out. This latter testing would involve deter- mination of consumer ability to describe the nutritional contents of individual food products, compare nutritional contents across prod- uct categories, and choose among relevant food alternatives. Given the normal time requirements for research of this ~e, and given the extensive material already available in this report and elsewhere on suggested label revisions, it is reasonable to set a minimum timetable of 1 year for this testing to be completed. Following that period, the agencies will then be ready to propose the new format. EDUCATING CONSU1VIERS TO USE NUTRITION INFORMATION ON FOOD LABELS Given the current wave of authoritative reports linking diet and chronic dis- ease, coupled with an era of public responsiveness to dietary recommendations, a unique opportunity exists to positively influence the future health of the U.S. population. Rising consumer awareness of the relationship between nutrition and long-term health has stimulated the food industry to develop a variety of new, more healthful products, has encouraged increased use of experimentation with health claims on food labels, and in some cases, has promoted the development of innovative consumer information programs. Whether raising public aware- ness of He dietary risk factors for disease can lead to the desired fundamental changes in dietary behavior in the United States depends on many factors, such as the availability of foods with high nutritional quality in the food supply and consumer ability to make more healthful food choices. The proposed Year 2000 Objectives for the Nation provide a number of nutrition objectives that relate to consumer knowledge of foods, nutrition labeling, and availability of improved food products. The nutrition objectives are designed to enable consumers to adopt sound dietary practices (reduce dietary fats and sodium; increase calcium and dietary fiber), and identity the dietary factors associated with chronic disease. Objectives specific to labeling are to: increase to at least 80 percent the proportion of people age 21 and older who use food labels to make food selections; increase nutrition labeling that provides information to facilitate choos- ing foods consistent with the Dietary Guidelines for Americans to at least 80 percent of processed foods and 40 percent of fresh meat, poul- try, fruits, vegetables, baked goods, and ready-to-eat foods; increase to at least 5,000 brand items the availability of processed food products that are reduced in fat, saturated fat, and cholesterol; and

PRESE;NT~ION OF Nl~RlTION INFORMATION ON FOOD HEM 265 increase to at least 5,000 brand items the availability of processed foods with lowered sodium (DHHS, 1989~. Strategies for Promoting Dietary Changes Two general approaches have been suggested to promote dietary changes: (1) environmental or structural intervention, and (2) personal or direct-influence strategies (Glanz and Mullis, 1988; Sims and Smiciklas-Wright, 1978~. Envi- ronmental or structural interventions are strategies that encourage positive be- haviors by creating opportunities for action and removing barriers to follow health-promoting practices. Such strategies focus on modifying the environment first, without requiring the individual to make conscious (often unwanted or unpopular) choices or to participate voluntarily in educational activities. In terms of promoting dietary change, such strategies take the form of modifying some aspect of the food supply or improving consumer access to food. Examples of such strategies would be to directly provide or distribute food, improve consumer's ability to purchase food, directly alter the nutritional quality of food products (e.g., by enrichment or fortification), or limit the food selections available to population groups in certain locations, such as institutions, work site cafeterias, or vending machines. In economic terms, environmental strategies can be said to affect the supply side, whereas direct or personal strategies are used in an attempt to alter the demand side of the supply-and- demand equation. Personal or direct-influence strategies are based on providing information or applying educational, persuasion, and behavior modification techniques directly to individuals or small groups. Over the past two decades, educational techniques have evolved from simple information transmission (based on the premise, "If they know the facts, they will change their behavior',) to a variety of direct behavior modification techniques that are designed to lead directly to the development of health-promoting skills and practices. Although such strategies have grown increasingly sophisticated and behaviorally oriented, they appear to be inefficient and ineffective means of reaching large population groups (Glanz and Mullis, 1988~. The provision of nutrition information on food labels is an interesting amalgam of the environmental and personal strategies. On the one hand, it is unquestioningly a personal informational strategy. One stated goal of nutrition labeling is Hat the nutrient composition information provided on the food label should enable the public to make informed food choices. To the extent that nutrition labeling leads to improvements in the nutritional qualities and varieties of foods that are consumed, labeling can be expected to have positive health benefits.

266 NUTRITION LABF~NG Yet, government has been encouraged to adopt this information provision strategy in order to facilitate and enhance the application of nutritional consid- erations to food consumption behavior (Quelch, 1977~. Rather than prohibit all information on food labels that would promote the importance of the relation- ship between diet and long-term health, FDA adopted the current policy strategy in the early 1970s that the manufacturer could present factual information on the nutrient composition of foods and then proceeded to promulgate regulations governing the content, format, and placement of nutrition labeling (Huts, 1986~. With such action, a personal strategy was transformed into an environmental strategy, those described as being more efficient and practical than one-to-one programs (Syme, 1986) and having the potential for reaching wider audiences and yielding a greater health-promoting impact (Glanz and Mullis, 1988~. Promoting Health Through Informational Campaigns Various public information campaigns aimed at promoting healthy behavior (including those for coronary heart disease, high blood pressure, and cancer) have focused on enhancing knowledge, changing attitudes, and improving skills. Two beliefs characterize many of the health promotion campaigns conducted in the United States over the past two decades. The first belief is that if people are just given the facts, they will proceed to change their behavior in accordance with this guidance. The second belief is that if people can be induced to hold favorable or unfavorable attitudes about a particular practice, they will change their behaviors to correspond with the appropriate attitude. Unfortunately, the research evidence does not fully support the efficacy of presenting just information or improving attitudes to cause individuals to follow health-promoting practices (Bettinghaus, 1986~. Consumers must have an information base in order to make long-lasting dietary changes. Yet, information alone cannot possibly be expected to produce behavioral changes unless adequate quantities of appropriate foods are available from which consumers can make choices. Educational approaches make sense only as far as there are environmental resources available to enable the consumer to implement the advice. The challenge is to combine effectively both types of strategies in nutrition labeling programs in order to capitalize on the relative strengths of each approach, with the ultimate goal being to achieve long-lasting positive behavioral changes among consumers. Communication diffusion research (Rogers, 1983), in particular, has demon- strated that mass media—whether print or audiovisual as in radio or television are best used to create awareness or expand knowledge about a particular new concept or program. In order to effect longer-lasting attitudinal and behavioral changes, however, information is best delivered by interpersonal communication channels, such as instructional sessions or friendly advice. Mass media can be effectively used to introduce a new concept or piece

PRES~NT~lON OF NUTRITION INFORMATION ON FOOD LABEI~i 267 of information to target populations and thus create a better knowledge base about the meaning of that new concept or idea. In order to effect behavior change, however, it is essential to employ interpersonal communication channels, such as small group discussions, individual counseling sessions, or personal demonstrations (Sims, 1979~. Obstacles To Effecting Dietary Changes There are a myriad of factors that affect dietary habits, from environmental and cultural to personal and idiosyncratic. It is little wonder that dietary change, no matter how sought after or desired, is difficult to achieve. Many foods, and often those most laden with saturated fatty acids, sodium, or cholesterol—offer psychological comfort. Yet, others are imbued with symbolism of a cultural heritage, such as foods eaten on special occasions. Most diet-related health problems develop gradually, and often they do not present immediate or dramatic symptoms. In turn, risk-factor reduction and disease prevention through dietary means require individuals to make long- term and often arduous changes in their habitual food intakes. Furthermore, some dietary changes (e.g., weight loss) provide obvious physical feedback, but others (e.g., increased fiber intake and cholesterol reduction) do not. Dietary recommendations, such as those contained in the Surgeon General's (DHHS, 1988) or the NRC (NRC, 1989a) reports, advise the public to forego immediate satisfaction in order to experience health benefits in the distant future. Most Americans consider themselves reasonably healthy and question whether such major alterations in their life-styles will really provide long-term benefits. Some believe (Glanz and Mullis, 1988) that in order to achieve health improvement by promoting healthy diets, nutrition interventions must reach large segments of the population and effectively influence the diverse factors that determine eating patterns. Most one-on-one nutrition education efforts are directed toward individuals identified as being at risk for disease or as having conditions requiring therapeutic dietary intervention. Nutrition information on food labels has the unique function of being able to offer something for all: consumers trying to avoid or reduce the percentage of certain elements in their diet (e.g., fat, cholesterol or sodium) or to maximize other elements (e.g. fiber, vitamins, or minerals); patients with congestive heart failure assiduously monitoring their sodium and/or potassium intake; consumers who want to see if the no cholesterol claim on the front of the label is, in fact, correct in terms of the nutrient composition of the product; or consumers who wish to compare two comparable products (e.g., breakfast cereals) for their nutrient composition. It was the responsibility of this Committee to make recommendations to ensure that the information-seeking consumer is able to make informed dietary decisions with the facts presented on the food label in the most understandable and usable format.

268 NUTRITION LABELING Using Food Label Information To Make Food Choices Consumers define the quality of a diet in terms of types of foods, not nutrients (Liefeld, 1983~. In order to facilitate dietary changes, consumers must understand the contributions that specific foods and food types make to the overall diet. From this perspective, science-derived diet and disease messages must be supplemented by information about the nutritional characteristics of specific foods and food types, and how to buy and prepare appropriate foods and meet appropriate quantitative goals. Without such understanding, attempts to modify dietary intake may not meet with success. From the perspective of helping consumers make intelligent food selections, one difficulty with diet and disease messages is that these messages usually are based on food components whose scientific role is not well understood by consumers. For most consumers, cholesterol, fatty acids, fiber, and sodium are relatively new terms when they are applied to making food choices. As a con- sequence, consumers may need to acquire knowledge about specific nutrients, food components, or food and nutrition processes in order to implement the generic dietary advice implied by such messages. To apply the general recom- mendations to reduce the intake of saturated fatty acids or sodium or to consume more dietary fiber, for example, an individual must understand the major food sources of these components, the contribution of different foods to the total diet, and how one's present diet may be excessive or deficient in these food comply nents. Appropriate dietary behavior depends on consumer's ability to recognize foods with desirable properties, to understand relevant terminology and apply it to food choices, to critically evaluate claims, and to assess the relative benefits of possible dietary changes in their own diets. Committee Recommendations The very concept of a comprehensive national nutrition policy suggests that not only should adequate supplies of safe, nutritious foods be available but that consumers should be given the educational means for making informed food choices (Helling, 1989; Quelch, 1977~. The Committee understands that a nutrition labeling program is only one component of a comprehensive education program, but believes that a well-designed nutrition label can help consumers to make informed food choices. However, nutrition information on food labels is just that, an information provision strategy, not an education program. The provision of information is only the first stage in the behavioral change process. Nutrition labeling can provide information about food and nutrition to the public, but it cannot be a substitute for comprehensive nutrition education programs. It is imperative that nutrition education programs be designed to complement nutrition labeling in order to give consumers the information and skills to make healthful food choices.

PRESENTATION OF N~ION INFO~AHON ON FOOD ~~ 269 A comprehensive, coordinated program of nutrition education will enable consumers to make individual food choices within the context of their own comprehensive program for health maintenance and disease prevention. The Committee urges the establishment of a public- and private-sector initiative to better help consumers understand and apply the information on the revised nutrition label. However, the Committee refrains from providing a discussion of or recommendations about Be specific aspects of nutrition education programs in deference to the pending Institute of Medicine/~;NB report by the Committee on Dietary Guidelines Implementation which will address this subject in depth (IOM, in press). The Committee recommends that comprehensive nutrition education pro- grams be developed in order to assist consumers to understand the information on food labels to plan diets and make appropriate food choices. It is the re- sponsibility of those designing such public information programs to ensure that consumers can process the information easily and accurately. This means that more attention must be given to thorough message testing research to determine the most effective format for delivering nutrition information on food labels. The Committee recommends that: Public- and private-sector initiatives should be established to help consumers understand and apply the information on the nutrition label. · Comprehensive nutrition education programs should be developed in order to help consumers to understand the information on food labels to enable them to plan diets and make appropriate food choices. COSTS OF NUTRITION LABELING REFORM Any reform of food labeling to provide more complete nutrition information and any expansion of the coverage of current nutrition labeling requirements will impose costs on producers, manufacturers, retailers, and, ultimately, consumers. It is not only the Committee's recommendations that would result in such costs; FDA's recent nutrition labeling proposal and the nutrition labeling legislation currently before Congress would impose similar costs. The Committee was not charged with assessing the costs of its or any other set of proposals for reform. But it would be irresponsible not to acknowledge that expanded and improved nutrition labeling will have costs and that the magnitude of these costs ought to be taken into account by FDA and USDA in their formulation of He details of and, even more important, the timing of a revised nutrition labeling system. In 1990, FDA commissioned a study on the costs of implementing the type of changes in nutrition labeling that it was planning to propose (55 ~d. Reg. 29,47~29,533; July 19, 1990~. The agency's notice of proposed rulemaking contains a preliminary analysis of the private-sector and consumer costs in

270 NATION LABELING the first four areas of implementing revised nutrition labeling regulations: (1) extending mandatory nutrition labeling to all packaged foods, (2) standardizing serving sizes, (3) revising the U.S. RDAs, and (4) listing all optional ingredients in standardized foods. FDA estimated that 21,000 arms would be affected and that the cost per U.S. household will be $3.15 in the first year of implementation of its proposal, and $0.60 per U.S. household each year afterward. This estimate, if accurate, may provide some guide as to the costs of the Committee's more ambitious set of recommendations. The added costs of expanded nutrition labeling must be compared to the savings through improved health that consumers are expected to realize by having and using improved, more comprehensive nutrition information. The Committee believes that potential long-term savings in health care costs and gains in longevity would outweigh the cost of its recommendations. Overview of Costs To Manufacturers and Retailers The costs incurred by food producers will include those for administrative activities, nutrient analysis of foods, changes in label design, printing new labels, and in some instances, reduced ingredient flexibility. Within the first category are the costs of discovering and interpreting new requirements, assessing their impact on products, and developing a product compliance system. In addition, retail food stores, which under the Committee's proposal would be required to post nutrition information about produce, meat, poultry, and seafood, would incur costs in preparing and maintaining this information. Some costs, however, notably the costs of assembling the information about nutrient content, will be borne by the suppliers of fresh foods. Ultimately, most additional costs will be passed on to consumers. FDA noted in its proposal that a firm's costs for nutrient analysis would depend to a great extent on which and how many of its products currently carry nutrition labeling or have nutrient analysis data available (55 Fed. Reg. 29,47~29,533; July 19, 1990~. For foods that have not previously provided nutrition labeling, the start-up costs of obtaining the required information may be substantial. Costs for relabeling would include label design, printing, and inventory costs. The latter could be minimized by scheduling the effective date of new regulations to conform with the already-scheduled "uniform compliance dates" for incorporating other mandated label changes and by allowing existing label stocks to be exhausted. For some foods there could conceivably be costs associated with reduced flexibility in the choice of ingredients. The Committee recommends that man- ufacturers be allowed to continue to use "and/or" labeling for fats and oils, on the condition that the food label state the highest level of saturated fatty acids achievable by any combination of listed fats and oils. This approach should not constrict choice of fats and oils unless manufacturers, worried about disclosing

PRESEl1T~ION OF NUTRITION INFORMATION ON FOOD LANDIS 271 high levels of saturated fatty acids, decide to curtail their use of some highly saturated fats or oils. But the associated product costs of such changes would be offset by direct nutritional benefits for consumers. FDA's proposal is designed to require more detailed listing of fatty acid content, which could curtail flexibility in formulation. By limiting the breakdown of fat components to saturate and unsaturated fatty acids, however, the agency expects that most manufacturers will be able to maintain sufficient flexibility in their selection of ingredients to minimize any increase in the cost of the final product. Costs for Different Food Categories Four major sectors of foods will be affected by the adoption of new requirements for nutrition labeling: (1) foods that currently carry nutrition labeling; (2) food that do not currently carry nutrition labeling but have been analyzed; (3) produce and fresh seafood under FDA jurisdiction, and fresh meat and poultry under USDA jurisdiction, and (4) restaurant foods. Under the Committee's recommendation (as well as FDA's proposal and proposed legislation), virtually all packaged foods would be required to bear nutrition labeling. In addition, foods now bearing nutrition labeling would be required to provide different information. Cost for Foods That Currently Carry Nutrition Labeling It is the Committee's judgment that its recommendations for the content of nutrition labels would require very little information that producers do not already possess. Possibly, the requirements that sodium and dietary fiber be listed will require reanalysis of some products. However, the requirements for listing the content of fat, fatty acids, protein, carbohydrates, vitamins, and minerals do not appear likely to demand new testing. For foods in this sector (now 60 percent of all packaged foods), the Committee's recommendations would mainly require changes in the presentation of information that manufacturers already have. The timing of the imposition of such requirements could affect the cost of compliance. If new labels had to be prepared and applied on a schedule that took no account of the normal, commercially driven evolution of food labels, or other government-mandated label changes, the extra cost could be substantial. But labels undergo relatively frequent changes, and FDA customarily establishes a"uniform compliance date" for all required label changes far in advance of the effective date. The Committee believes that recommended changes in nutrition labeling should be implemented on the same schedule already fixed for other mandated changes, which should allow ublizabon of old label stocks. USDA should also follow this same approach for implementing changes in nutrition labeling for meat and poultry products.

272 NUTRITION LABELING Cost for Foods That Do Not Currently Carry Nutrition Labeling For foods that do not now bear nutrition labeling of any sort, additional costs will be incurred. It is likely that manufacturers do not currently have complete information about the nutrient content of some products. The cost of analysis of these foods may not be trivial, though adequate methods and laboratory resources are available to analyze, at a reasonable cost, virtually all packaged foods for all of the nutrition components that the Committee recommends. It is reported that many packaged foods that do not currently bear nutrition labeling have nonetheless been analyzed by their manufacturers (Daniel Padberg, Texas A&M University, personal communication, 19903. The growth in voluntary nutrition labeling over the past decade and the willingness of most sectors of the packaged-food industry to accept, and in many instances support, mandatory nutrition labeling suggests that these costs will not be significant. The costs of relabeling can be controlled in the same fashion as for foods that already bear nutrition labeling, by allowing reasonable lead time and schedules, coupled with the mandated label changes. Costs for Labeling Fresh Foods Produce and Seafood The Committee is recommending nutrition la- beling, broadly defined, for selected produce and fresh seafood. Retailers will incur modest costs in displaying and maintaining the required information, but the major costs are likely to occur in assembling the data base to support the required nutrient declarations. The Committee acknowledges that FDA may have to adjust the timing of its requirement for such information in light of the costs, and time required, to comply. The Committee recognizes, as does FDA, that providing point-of-purchase nutrition information for produce and fresh seafood will impose significant new costs on retailers and on consumers. Some 235,000 food stores were estimated to be in operation in 1987. FDA's proposal does not provide any estimate of total compliance costs. It does estimate that 50 percent of stores could supply the required information at an annual cost of $200 per store, but it acknowledges that this figure does not include the cost of generating the data bases to support such displays, and a portion of this cost is likely to be borne by food stores and, ultimately, consumers. Meat and Poudtry Unlike either the FDA proposal or pending legislation, which do not mention meat and poultry products, the Committee is recommend- ing nutrition labeling of all packaged meat and poultry products and point-of- purchase information for fresh meat and poultry products. A share of USDA- regulated packaged foods now bear nutrition information, and the abbreviated

PRESE=~ION OF NUTRITION INFORMATION ON FOOD LABEl~; 273 USDA format demands less information than FDA, only requiring information about calories, protein, carbohydrate, and fat. Fresh meat and poultry do not provide significant amounts of dietary fiber. Analysis for sodium, calcium, and iron would be an expense only if they were not included in the original analysis. Costs for labeling fresh meat and poultry products would be from printing and maintaining point-of-purchase information in retail food stores, since the data would be taken from existing nutrient data bases. Broadly speaking, however, it would appear that the cost of implementing the Committee's recommenda- tions for meat and poultry should not differ substantially, on a per product basis, from those incurred by FDA-regulated packaged foods and for produce and fresh seafood. A caveat should be added here, however. It is possible that USDA's rigorous criteria for approving nutrient statements on the labels of meat and poultry products would impose higher per product costs than those of FDA. The deparUnent's prior approval system, like other premarket approval systems, appears to invite the Me of skepticism that drives up the cost of product testing. The Committee believes that it is important for FDA and USDA to agree on uniform criteria for assessing the accuracy of label statements of nutrition content as well as on consistent standards for approving data bases as the source of nutrient composition data for fresh foods (see Chapter 5~. Costsfor Obeying Restaurant Foods In one respect, the costs of the Committee's recommendations will differ sharply from those of FDA's proposal or the pending legislation. The Committee is recommending that limited-menu restaurants be required to display point- of-purchase nutrient content (and ingredient) information on Heir foods and that all other restaurants be required to have such information available to consumers on request. The first half of this recommendation is not likely to entail substantial additional costs, either for the limited-menu restaurant or for consumers. The products sold by these operations are standardized, carefully controlled for content and quality, and generally uniform throughout the country, and the world. Major limited-menu restaurant franchise chains have previously reported that they have already analyzed their products for nutrient content. The only significant new cost involved is likely to be that of preparing and maintaining the posted nutrient information. The recent willingness of several major chains to display information of the Me recommended by the Committee suggests that this expense is, on a per restaurant basis, modest. The costs of the Committee's recommendation to require all other restau- rants to have their menus evaluated and to offer nutrient information on request to consumers are considerably less ceItain and more speculative. The Committee believes that computer programs for evaluation of the nutrition profile of menus

274 NUTRITION lW3ELING is widely available and inexpensive, but even modest expense may prove high for small operators (see Chapter 5). Printing the statement "nu~idon inforrnabon is available on request" on menus would be essentially costless. If requests were frequent, however, the cost of preparing, duplicating, and maintaining menu in- fonnation could prove more than trivial, particularly for operators Eat changed menus frequently. This is an area in which close study of potential costs is warranted. REFERENCES Asam, E.H., and L.P. Bucklin. 1973. Nutrition labeling for canned goods: A study of consumer response. J. Market. 37:32-37. Babcock M.J., and M.M. Murphy. 1973. Two nutrition labeling systems. J. Am. Diet Assoc. 62:15~161. Betteman, J.R. 1979. An Information Processing Theory of Consumer Choice. Addison- Wesley, Reading, Mass. 402 pp. Bettinghaus, E.P. 1986. Health promotion and the knowledge-attitude-behavior contin- uum. Preventive Med. 15:475-491. CNCFL (Committee on the Nutrition Components of Food Libeling). 1990. Workshop on Label Formats, CNCFL, Food and Nutrition Board, Institute of Medicine, April 25, 1990. Unpublished. Cole, C A., and GJ. Gaeth. 1990. Cognitive and age-related differences in the ability to use nutritional information in a complex environment. J. Market. Res. May:175- 184. CRI (Consumer Research Institute). 1972. Interim Report of the First Two Phases of the CRI/FDA Nutrition Labeling Research Program. CRI, Washington, D.C. 223 pp. Cronin, F.J., A.M. Shaw, S.M. Krebs-Smith, P.M. Marsland, and L. Light. 1987. Developing a food guidance system to implement the Dietary Guidelines. J. Nutr. Ed. 19(6):281-302. Daly, P.A. 1976. The response of consumers to nutrition labeling. J. Consumer Affairs 10(Winter): 170-178. DHEW/USDA/FI`C (U.S. Department of Health, Education, and Welfare, U.S. Depart- ment of Agriculture, and Federal Trade Commission). 1979. Food Labeling Back- ground Papers. Govemment Printing Office, Washington, D.C. 124 pp. DHHS (U.S. Department of Health and Human Services). 1988. The Surgeon General's Report on Nutrition and Health. Government Printing Office, Washington, D.C. 727 PP. DHHS (U.S. Department of Health and Human Services). 1989. Promoting Health/Pre- venting Disease: Year 2000 Objectives for the Nation. Draft for public review and commend Public Health Service, Washington, D.C. Eisner, D. 1972. Developmental relationships between held independence and Sxity- mobility. Perceptual Motor Skills 34(June):767-770. Farmakalidis, E. 1989. Nutrition labelling—an industry perspective. Food Aust. 41~10~: 979-981. FLAPS Mod Libeling and Package Survey). 1988. Status of Nutrition and Sodium

PRESE;NT=ION OF NUTRITION INFORMATION ON FOOD LABEl~; 275 Labeling of Processed Foods. Division of Consumer Surveys, Center for Food Safety and Applied Nutrition, FDA, Washington, D.C. 19 pp. Food Chemical News. 1988. FDA declines to establish "organic" food labeling standards. 26(December): 16-17. Food Chemical News. 1990. FDA action against Ragu "Fresh Italian" sauce asked by Public Voice 12(March):20. Geiger, C.J., B.W. Wyse, C.R.M. Parent, and R.G. Hansen. 1990. The use of adaptive conjoint analysis (ACA) to determine the most useful nutrition label for purchase decisions. Abstract No. 4587. Presented at Federation of American Societies for Experimental Biology, Washington, D.C. Gersovitz, M., J.P. Madden, and H. Smiciklas-Wright. 1978. Validity of the 24-hour dietary recall and seven-day records for group comparisons. J. Am. Diet. Assoc. 73:48-55. Glanz, K., and R. Mullis. 1988. Environmental interventions to promote healthy eating: A review of models, programs, and evidence. Health Ed. Q. 15~4~:395~15. Gunner, S.W. 1989. Nutrition labelling~he Canadian experience. Food Aust. 41~10~: 984-987. Guthrie, H.A.1984. Selection and quantification of typical food portions by young adults. J. Am. Diet. Assoc. 84:1440-1444. Hammonds, T.M. 1978. Testimony before the Subcommittee on Nutrition, Committee on Agriculture, Nutrition, and Forestry, U.S. Senate. August 9-10, 1978. Heimbach, J.T. 1982. Public Understanding of Food Label Information. Food and Drug Administration, Washington, D.C. 24 pp. Heimbach, J.T. 1985. Cardiovascular disease and diet The public view. Public Health Rep. 100:5. Heimbach, J.T. 1986. The growing impact of sodium labeling of foods. Food Technol. 40~12~: 102. Heimbach, J.T. 1987. Risk avoidance in consumer approaches to diet and health. Clin. Nutr. 6:159. Heimbach, J.T., and R.C. Stokes. 1981. Nutrition Labeling for Today's Needs: Opinions of Nutritionists, the Food Industry and Consumers. Food and Drug Administration, Washington, D.C. 52 pp. Heimbach, J.T., A.S. Levy, and R.E. Schucker. 1990. Declared serving sizes of packaged foods, 1977~6. Food Technol. 44:82-90. Helsing, E. 1989. Nutrition policies in Europe—the state of the art. Eur. J. Clin. Nutr. 43(Suppl.~:57-66. Hunter, D.M., L. Sampson, M.H. Stampfer, G A. Colditz, B. Rosner, and W.C. Willett. 1988. Variability in portion sizes of commonly consumed foods among a population of women in the United States. Am. J. Epidemiol. 127:1240. Hutt' P~. 1986. National nutrition policy and the role of the Food and Drug Adminis- ~ation. Currents 2~2~:2-11. Hutt, P~., and R.A. Merrill. 1990. Food and Drug Law: Cases and Materials. Foundation Press, Inc., Mineola, N.Y. In press. IOM (Institute of Medicine). In press. Improving America's Diet and Health: From Recommendaions to Action. Committee on Dieta~y Guidelines Implementation, Food and Nutrition Board. National Academy Press, Washington, D.C.

276 NUWTION LABELING Krebs-Smith, S.M., and H. Smiciklas-Wright. 1985. Typical serving sizes: Implications for food guidance. J. Am. Diet. Assoc. 85:1139-1141. Lansky, D., and K.D. Brownell. 1982. Estimates of food quantity and calories: Errors in self-report among obese patients. Am. J. Clin. Nutr. 35:727-732. Lenahan, RJ., J.A. Thomas, D A. Taylor, D.L. Call, and PI. Padberg. 1972. Consumer reaction to nutrition information on food product labels. Search Agric. 2~15~:1-26. Lewis, CJ., A.M. Beloian, and EA. Yetley. 1988. Serving size issues in estimating dietary exposure to food substances. J. Am. Diet. Assoc. 88:1545-1552. Liefield, J.P. 1983. Nutrition Labeling and Consumer Behavior: A Review of the Evidence. Prepared for the Bureau of Nutritional Sciences, Health ~ Welfare Canada, Ottawa, Ontario, Canada. Madden, J.P., SJ. Goodman, and H A. Guthrie. 1976. Validity of the 24-hour recall. J. Am. Diet. Assoc. 68:143-147. McCullough, J., and R. Best. 1980. Consumer preference for food label information: A basis for segmentation. J. Consumer Affairs 14~1~:180-192. Merrill, R. A., and PB. Hutt. 1980. Food and Drug Law: Cases and Materials. Foundation Press, Inc., Mineola, N.Y. 959 pp. Mohr, K.G., B.W. Wyse, and R.G. Hansen. 1980. Aiding consumer nutrition decisions: Comparison of a graphical nutrient density labeling format with the current food label system. Home Econ. Res. J. 8~3~:162-172. Muller, T.E. 1985. Structural information factors which stimulate the use of nutrition infotnlation: A field experiment. J. Market. Res. May:143-157. NRC (National Research Council). 1941. Recommended Dietary Allowances. Committee on Food and Nutrition. National Research Council, Washington, D.C. 13 pp. NRC (National Research Council). 1968. Recommended Dietary Allowances, 7th ed. Report of the Committee on the Seventh Edition of the Recommended Dietary Allowances, Food and Nutrition Board. National Academy of Sciences, Washington, D.C. 101 pp. NRC (National Research Council). 1988. Designing F~ods: Animal Product Options in the Marketplace. Committee on Technological Options to Improve the Nutritional At~ibutes of Animal Products, Board on Agriculture. National Academy Press, Washington, D.C. 367 pp. NRC (National Research Council). 1989a. Diet and Health: Implications for Reducing Chronic Disease Risk. Report of the Committee on Diet and Health, Food and Nu- trition Board, Commission on Life Sciences. National Academy Press, Washington, D.C. 749 pp. NRC (National Research Council). 1989b. Recommended Dietary Allowances, 10~ ed. Report of the Subcommittee on the Tenth Edition of the Recommended Dietary Allowances, Food and Nutrition Board, Commission on Life Sciences. National Acadlemy Press, Washington, D.C. 285 pp. ORC (Opinion Research Corporation). 1990. Food Labeling and Nutrition: What Ameri- cans Want. Survey conducted for the National Food Processors Association, Wash- ington, D.C. 178 pp. Pao, E.M., K.H. Fleming, P.H. Guenther, and SJ. Mickle. 1982. F~ods Commonly Eaten by Individuals: Amount Per Day and Per Eating Occasion. Home Economics

PRESENT~ION OF NUTRITION INFOR~lATION ON FOOD LABELS 277 Research Report No. 44. Human Nutrition Information Service, U.S. Department of Agriculture, Washington, D.C. 431 pp. Pennington, J. A.T., V.L. WiLkening, and J.E. Vanderveen. 1990. Descriptive terms for food labeling. J. Nutr. Ed. 22~1~:51-54. Quelch, J. A. 1977. The role of nutrition information in national nutrition policy. Nutr. Rev. 35~11~:289-293. Rogers, E.M. 1983. Diffusion of Innovations, 3rd ed. The Free Press, New York 367 pp. Rudd, J. 1986. Aiding consumer nutrition decisions with the simple graphic label format. Home Econ. Res. J. 14~3~:342-346. Rudd, J. 1989. Consumer response to calorie base variations on the graphical nutrient density food label. J. Nutr. Ed. 21:259-264. Sims, L.S. 1979. The community nutritionist as change agent. Fam. Community Health 1(4):83-92. Sims, L.S., and S.K. Shepherd. 1987. An Information Processing Approach to the Evaluation of Nutrition Education Materials. Final Report submitted to the Human Nutrition Information Service. USDA, Washington, D.C. Sims, L.S., and H. Smiciklas-Wright. 1978. An ecological systems perspective: Its application to nutrition policy, program design and evaluation. Ecol. Food Nutr. 7(3):17~180. Stephenson, M. 1978. The confusing world of health foods. PDA Consumer 12~6~:19-20. Syme, S.L. 1986. Strategies for health promotion. Preventive Med. 15:492-507. TDA (Texas Department of Agriculture). 1989. Statement of the Texas Department of Agriculture before the U.S. Food and Drug Administration hearing on food labeling. San Antonio, Tex., Nov. 1, 1989. USDA (U.S. Department of Agriculture). 1958. Food for Fitness: A Daily Food Guide. Leaflet No. 424. Institute of Home Economics, Agricultural Research Service. Government Printing Office, Washington, D.C. USDA (U.S. Department of Agriculture). 1975. Nutritive Value of American Foods in Common Units. Agriculture Handbook No. 456. Government Printing Office, Washington, D.C. 291 pp. USDA (U.S. Department of Agriculture). 1976. Composition of Foods. Agriculture Handbook No. 8 series. Govemment Printing Office, Washington, D.C. Revised continuously. USDA (U.S. Department of Agriculture). 1981. Nutritive Value of Foods. Home and Garden Bulletin No. 72, revised. Govemment Printing Office, Washington, D.C. 72 PP USDA (U.S. Department of Agriculture). 1982a. FSIS Policy Memorandum 039. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1982b. FSIS Policy Memorandum 055. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1984. FSIS Policy Memorandum 049C. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1986. FSIS Policy Memorandum 071A. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1987a FSIS Policy Memorandum 019A. Food Safety and Inspection Service, Washington, D.C.

278 NUTRITION LABELING USDA (U.S. Department of Agriculture). 1987b. FSIS Policy Memorandum 070B. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1989a FSIS Policy Memorandum 022C. Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1989b. FSIS Standards and Labeling Policy Book. Standards and Labeling Division, Food Safety and Inspection Service, Washington, D.C. USDA (U.S. Department of Agriculture). 1990. Report of the Dietary Guidelines Ad- visory Committee on the Dietary Guidelines for Americans, 1990. Report to the Secretary of Agriculture and the Secretary of Health and Human Services. Govem- ment Printing Office, Washington, D.C. 48 pp. USDAtDHHS (U.S. Department of Agriculture and U.S. Department of Health and Hu- man Services). 1980. Nutrition and Your Health: Dietary Guidelines for Americans. Government Printing Office, Washington, D.C. 20 pp. USDA/DHHS (U.S. Department of Agriculture and U.S. Department of Health and Hu- man Services). 1982. Design and Evaluation of Nutrition Label Formats: Informa- tion Kit. Government Printing Office, Washington, D.C. 20 pp. USDA/DHHS (U.S. Department of Agriculture and U.S. Department of Health and Hu- man Services). 1985. Nutrition and Your Health: Dietary Guidelines for Americans, 2nd ed. Government Printing Office, Washington, D.C. 24 pp. Vankatesan, M. 1977. Providing Nutritional Information to Consumers. Paper presented at a Special NSF/MIT Conference on Consumer Research for Consumer Policy, Cambridge, Mass., July. Vankatesan, M., W. Lancaster, and K.W. Kendall. 1986. An empirical study of alterna- tive formats for nutritional information disclosure in advertising. J. Public Policy Marketing 5:29-43. Webb, C.A., and J. A. Yuhas. 1988. Ability of WIC clientele to estimate food quantities. J. Am. Diet. Assoc. 88:601~02. Wolf, I.D., N.R. Raper, and J.C. Rosenthal. 1983. USDA activities in relation to the sodium issue: 1981-1983. Food Technol. 37:59. Yankelovich, Inc. 1971. Nutrition labeling: A consumer experiment to determine the effects of nutrition labeling on food purchases. Chain Store Age January:55-77. Yuhas, J. A., J.E. Bolland, and T.W. Bolland. 1989. The impact of training, food type, gender, and container size on the estimation of food portion sizes. J. Am. Diet. Assoc. 89:1473-1477.