The new NRF models were developed using the open-access USDA National Nutrient Database for Standard Reference, Release 28 (now SR Legacy), that is available on FoodData Central (15). The SR-28 database lists total amounts of nutrients present in the edible portion of the food, including any nutrients added in processing. The nutrient content values are for fruits and vegetables that are fresh, frozen, or canned, sometimes with the addition of other ingredients such as sugar syrup in canned fruit. Herbs and spices can be dried or fresh. NDB (National Data Base) first digit codes were used to identify fruits, vegetables, beans, nuts, grains, and herbs and spices.

Based on the USDA classification used in the What We Eat in America (WWEIA) studies (17), the fruit category was subdivided into citrus fruit, citrus juice, other fruit (raw or cooked), melons, dried fruit, berries, and olives. Fruit jams, jellies and preserves other than cranberry were excluded. Vegetables were subdivided into leafy green vegetables, cruciferous vegetables, red/orange vegetables (and products), and other vegetables. Beans were chickpeas, lima beans, lentils and soy. Additional categories were nuts, cocoa powder, chocolate, coffee, and tea, and selected whole grains.

The herbs and spices category included dried herbs and spices, along with some fresh herbs. Items for which nutrient composition data were available (albeit with some missing values) included allspice, anise, basil, bay leaves, caraway seed, cardamon, celery seed, chervil, chili powder, cinnamon, cloves, coriander, cumin seed, dill seed, fennel seed, fenugreek, ginger, mace, marjoram, mustard, nutmeg, oregano, paprika, parsley, pepper (black, red, or white), rosemary, saffron, sage, tarragon, turmeric, and thyme. The SR-28 spices category also listed vinegars; those were removed.

The SR-28 provides energy and nutrient content for 65 nutrients per 100 g of food for 8,789 foods from all food groups. In addition to protein, fiber and vitamins and minerals, the SR-28 provides data on individual carotenoids (β-carotene, α-carotene, β-cryptoxanthin, lycopene, and lutein+ zeaxanthin). Most analytical systems do not separate lutein and zeaxanthin, so these carotenoids are typically combined. The SR-28 also lists values for vitamin C, vitamin E, selenium, copper, and zinc. As might be expected, there were some missing values.

The USDA database for the flavonoid content of selected foods, release 3.3 (16) contains analytical values for 25 flavonoid compounds for 512 food items. The unit of measure is mg/100 g edible portion on fresh weight basis. Subclasses of flavonoids were flavonols, flavones, flavan-3-ols, flavanones, and anthocyanidins. Detailed documentation is provided online (16). The term “total flavonoids” refers to those flavonoids that were listed in the release 3.3 of the USDA database.

The SR-28 database and the much smaller USDA Expanded Flavonoid database version 3.3. were matched using NDB food identification codes. Some herbs and spices lacked flavonoid data and could not be matched, indicating a potential data need. Analyses that involved flavonoids were restricted to those items where the match was successful. A total of 335 items in multiple categories of interest were available for data analysis.

As new datasets become available, the classic NRF algorithm structure can be maintained but some index nutrients can be easily swapped, added, or removed. Notably, the choice of index nutrients can vary depending what available data can best capture nutritional value of items of interest.

The NRF9.3 algorithm is based on the unweighted sum of percent Daily Values (%DV) for 9 nutrients to encourage (the NR9 sub-score) minus the sum of %DV for 3 nutrients to limit (LIM sub-score) (18). The original Nutrient Rich (NR9) sub-score was based on protein, fiber, vitamins A, C and E, calcium, iron, potassium, and magnesium (2, 3). Later versions replaced vitamin E with vitamin D. Those nutrients are either part of the FDA definition of « healthy » or have been identified by the DGA as shortfall nutrients under-consumed by the US population. Nutrients to limit were saturated fat, added sugars, and sodium.

The NRF algorithm uses the sum rather than the mean of %DVs. NRF algorithms based on subtraction (NR9 – LIM) yielded a better distribution of values than did those based on ratios (NR9/LIM). The final product was given the name NRF9.3. The %DVs were expressed per 100 kcal and were capped at 100%.

The NRF4.3 score was based on the sum of %DVs for 4 nutrients to encourage: protein, fiber, vitamin C and potassium (the NR4 sub score) minus the sum of %DVs for saturated fat, added sugar and sodium (LIM sub score). These 4 nutrients to encourage are normally listed on the Nutrition Facts Panel.

The NRFa11.3 was also based on the sum of %DV nutrients and components to encourage (NR11 sub score) minus the sum of %DV for nutrients to limit. The elements to encourage were vitamins and minerals, carotenoids, and flavonoids. Vitamins and minerals were zinc, copper, selenium, vitamin C, and vitamin E. The 5 carotenoids were alpha carotene, beta carotene, beta cryptoxanthin, lycopene, and lutein/zeaxanthin. Total flavonoids were the sum of flavonoid classes. In a departure from the original NRF9.3, %DVs for both models were calculated per 100 g rather than per 100 kcal.

Nutrient standards are typically based on local reference dietary amounts. The U.S. Reference Daily Values (19, 20) that are published by the FDA and used on nutrition labels are summarized in Table 1. In a departure from the original model, the present approach was to convert nutrient amounts per 100 g to percent daily values (% DV) also per 100 g. As in the past, percent DVs were capped at 100% so that foods containing very large amounts of a single nutrient would not have a disproportionately high score.

Reference daily values for carotenoids were based on USDA calculations of the relations between provitamin A carotenoids and retinol equivalents (RAE). No daily values are available for flavonoids at this time. The present preliminary standards were based on estimated median intake of total flavonoids at 250 mg/d that was based on prior estimates of population intakes of flavonoids in the Us based on NHANES and other data (21, 22); observed associations between flavonoid intakes and diet quality and (23) and flavonoid intakes of tea consumers versus non-consumers (24).

The three nutrients to limit in the LIM sub-score of the NRF9.3 model were saturated fat, added sugar and sodium (2, 3). Maximum recommended values (MRVs) for nutrients to limit were 20 g for saturated fat, 90 g for total sugar, 50 g for added sugar, and 2,300 mg for sodium, all based on a 2000 kcal/d diet (18). Most of the food items selected contained no (or minimal amounts) of saturated fat, added sugars, or sodium.

Differences in NR sub scores and NRF scores by food category were tested using one way ANOVA with post hoc Dunnett’s test, using herbs and spices as the control group that was compared to all other groups. Statistical significance determined at p value <0.05. Analyses used SPSS (Statistical Package for the Social Sciences v 16.0).

Figure 1 shows mean values for the standard NRF4.3 score (based on protein, fiber, potassium, vitamin C) by food category. Data are plotted against energy density (kcal/100 g) with the size of the bubble corresponding to the number of items in each category. On per 100 g basis, the highest NRF4.3 scores were obtained for herbs and spices, cocoa powder, beans, cruciferous vegetables, and berries. Nuts and dried fruit scored lower because of the fat and added sugar content. Olives and chocolate obtained negative scores (because of salt and sugar). As shown in Figure 1, the NRF4.3 score did not discriminate among fruits and vegetables particularly well. Berries scored higher than most other fruits largely because of high vitamin C content. Supplementary Table S1 shows means and standard errors (SEM) for NR4 sub score components protein, fiber, vitamin C, and potassium by food category. Highest in protein and fiber were herbs and spices, nuts, cocoa/chocolate, and beans. Herbs and spices and cocoa/chocolate were also highest in potassium. Berries were highest in vitamin C (acai berry was the highest), followed by cruciferous and red/orange vegetables, herbs and spices, leafy green vegetables, and citrus fruit.

Supplementary Table S2 shows energy density (in kcal/100 g), LIM and NRF4.3 values by food category. As expected, nuts had the highest energy density followed by chocolate, whole grains, dried fruit and herbs and spices (mostly dry). Highest LIM values were for chocolate (sugar), olives (salt), and nuts.

Figure 2A shows a scatterplot of %DV for carotenoids plotted against energy density. The size of the bubble denotes the number of items in each category. All calculations are per 100 g. Highest in carotenoid content were green leafy vegetables, red orange vegetables, herbs and spices, melons, and cruciferous vegetables.

Figure 2B shows a scatterplot of %DV for flavonoids plotted against energy density. Highest in flavonoids were cocoa powder and chocolate, herbs and spices, and berries. Flavonoid content in the USDA database was not available for all herbs and spices.

Figure 3 shows a scatterplot of %DV for antioxidant vitamins and minerals (AVM) plotted against energy density. The AVM were vitamin C, vitamin E, selenium, copper, and zinc. Highest in AVM (per 100 g) were nuts, herbs and spices, cocoa powder and chocolate, whole grains, and beans.

Data for sum %DV for total carotenoids, total flavonoids and AVM by food category are provided in Supplementary Table S3. Herbs and spices were consistently high in most carotenoids other than lycopene and were high in antioxidant vitamins and minerals.

The final NRFa11.3 score was calculated as the sum of %DV for nutrients and other compounds to encourage (AVM, carotenoids and flavonoids) minus the sum of % DV for the nutrients to limit (saturated fat, added sugar, sodium). All individual %DV were capped at 100%.

Figure 4 shows composite NRFa11.3 scores plotted against energy density by food category. Highest values were assigned to herbs and spices, green leafy vegetables, cocoa powders, nuts, and cruciferous vegetables. Berries were among the highest-scoring fruit. Data (means and SEM) and statistics are shown in Supplementary Table S4.

Top scoring items are shown in Table 2. Spices and herbs, fresh and dried, had some of the highest NRFa11.3 scores. Pepper, paprika and chili powder along with dried and some fresh herbs had the highest estimated content of nutrients and functional ingredients per 100 g. Leafy green vegetables also scored high, benefiting from high carotenoid content and low energy density. High scores were also obtained by red orange vegetables (pumpkin, squash, carrots), pulses, nuts, and fruit, especially melons and berries. Melons are high in carotenoids whereas berries score high because of high vitamin C and flavonoid content.