Modeling Possible Outcomes of Updated Daily Values on Nutrient Intakes of the United States Adult Population

The United States (US) Food and Drug Administration has updated the Daily Values (DVs) for the Nutrition Facts Label on packaged foods. We used the National Health and Nutrition Examination Survey 2009–2012 data with the International Life Sciences Institute, North America Fortification Database, which identifies intrinsic, mandatory enriched, and fortified sources of nutrients in foods and beverages, to model the new DVs’ potential impact on adult (≥19 years of age) intake. We assumed that manufacturers will adjust voluntary fortification to maintain percent DV claims. We assessed the percent of the US population whose usual intake (UI) was < the Estimated Average Requirement (EAR), and ≥ the Upper Limit (UL) based on the current DVs, and modeled estimated UI and %<EAR with the new DVs (Updated DV) for 12 micronutrients. Modeling for vitamins B12, A, B6, riboflavin, niacin, thiamin, and zinc predicts fewer voluntarily fortified foods and reduced adult UI. Assuming manufacturers add more vitamins C and D and calcium to foods, the Updated DV predicts the adult UI will increase for these nutrients. Our modeling predicts a 15% reduction in overall adult vitamin A intake, a recognized “shortfall nutrient” and that even with the increased DV for vitamin D, 70% of US adults are predicted to have an intake <EAR.


Introduction
Nutrients added through enrichment and voluntary fortification of foods make important contributions to dietary intake in the United States (US) [1][2][3][4][5][6]. However, most Americans don't consume recommended amounts of dietary fiber and what have been termed "shortfall nutrients" including vitamins A, C, D, and E, folate, magnesium, potassium, and calcium [7][8][9]. In addition, US men and women in the lowest income groups are more likely to experience inadequate intakes of these nutrients compared with the highest income groups [10].
Use of food labels by US consumers has been positively associated with dietary quality [11]. Specifically, regulated nutrition labels on packaged foods and beverages enable consumers to Although complete 24-h dietary intake data were available for 18,273 participants on Day 1 and Day 2 from the NHANES (2009)(2010)(2011)(2012) combined data, after exclusion of participants whose data were deemed by the interviewer to be incomplete (n = 261), participants who were <4 years of age (n = 1991), and women who were pregnant or lactating (n = 168), the sample included a total of 15,853 participants. The sample was further reduced to only include adults ≥19 years of age to produce an analytic sample of 10,698 participants in this study. In this total sample, 50.0% were male.
We also used the ILSI NA-FD (Database of Fortification, Enrichment, and Intrinsic Nutrient Levels in Foods Reported Consumed in What We Eat in America, NHANES 2009-2010 and 2011-12, version 1.0, completed 16 October 2015) [21], which contains estimates of nutrients available from three sources-as naturally occurring (i.e., intrinsic), enriched, and fortified-for those foods and beverages as reported in the two NHANES cycles (2009-2010 and 2011-2012). The ILSI NA-FD is based on the USDA's FNDDS, and the original database was developed by Nutrition Impact LLC, Battle Creek, MI, USA and has been updated by Exponent, Inc. (Menlo Park, CA, USA). Some values for folate content of foods in the ILSI NA-FD were sourced from the USDA FNDDS (release 28; released September 2015, slightly revised May 2016).

Usual Intake Estimation
Data obtained from the two NHANES 24-h recalls were used to estimate prevalence of intake following the US National Cancer Institute (NCI) usual intake (UI) estimation methodology [25,26]. The NCI's MIXTRAN and DISTRIB computer macros enable UI estimation at the individual level. As part of our analysis, we incorporated/controlled for age, interview day (first compared with second), and weekend day (yes/no) to account for weekend effects in intake.

Creation of the Current DV and Updated DV Datasets
First, we replicated the ILSI NA-FD (Original FD) and created a second database ILSI NA-FD (Modified FD) in which voluntary fortification levels were adjusted to maintain the same %DV in foods reported in the Original FD but applying updated DVs that will be effective in 2020 or 2021 (see Table 1). In creating the Modified FD, three assumptions were made: (1) the naturally occurring intrinsic nutrient content of a food remained constant; (2) the amount of nutrients added to a food through enrichment was held constant; and (3) food manufacturers will maintain the same %DV in food products after 2020 or 2021. Appendix A Table A1 provides details about the calculations used in creating the ILSI NA-FD Modified.
An enriched food is a product to which nutrients have been added that were typically present in the food in its original form but were lost during processing. Food enrichment in the US is expected to follow the FDA's food restoration principle, cf. [27]. In the US, food fortification is governed by the FDA and refers to the practice of increasing the content of micronutrients in a food to improve its nutritional quality [27]. In this study, we assumed that the amount of nutrients added through enrichment would be held constant because the US regulations specifying the amounts of vitamins A, thiamin, riboflavin, niacin, D, folic acid, and iron to be added when enriching foods were not altered when the DVs were updated [12]. In addition, the amount of nutrients added during enrichment are often linked to standards of identity in the US [17]. A standard of identify establishes the name of the food, e.g., yogurt, and defines the ingredients or components, their function and levels permitted in food (21 Code of Federal Regulations 130.3) [28] or foods that substitute (21 CFR 130.10) [29].  2 1 µg RAE = 1 µg retinol = 12 µg supplemental β-carotene = 24 µg α-carotene = 24 µg β-cryptoxanthin. 3 1 mg NE = 1 mg niacin = 60 mg tryptophan. 4 1 mg a-tocopherol (label claim) = 1 mg a-tocopherol = 1 mg RRR -a-tocopherol=2 mg all-rac-a-tocopherol. 5 1 µg DFE = 1 µg naturally occurring folate = 0.6 µg folic acid. 6 1 µg vitamin D = 40 IU.
When the intrinsic and enrichment levels of a nutrient in a food would yield the same (or greater) %DV in the Modified FD as in the Original FD, we also assumed food manufacturers would discontinue voluntary fortification. Therefore, in our model, we set the fortification contribution to zero and categorized the food as non-fortified in the Modified FD. This condition was the only one where we reassigned a food/beverage from fortified in the Original FD to non-fortified in the Modified FD. Since the Original FD reports DFE for folate/folic acid, and does not report whether a vitamin E fortified food contains µg of synthetic or natural vitamin E, nutrient levels were left unchanged for these two nutrients when creating the Modified FD.
Second, the NHANES individual food intake files were merged with the Original FD, which contained the total amount of each nutrient in the specified food/beverage and proportions according to the three sources (naturally occurring, enriched, and fortified). The total amount of each nutrient in food and source type (naturally occurring, enriched, fortified) for the analysis was calculated by multiplying the number of grams per nutrient from the NHANES food file by the nutrient proportion in the ILSI NA-FD, then dividing by 100. This yielded a total nutrient intake from all foods being consumed and the proportion that was intrinsic, enriched or fortified for each individual. The same modeling approach was used with the Modified FD.
Third, nutrient by nutrient and using the DV relevant to each model, individuals were categorized according to the proportion of their nutrient intake derived from fortified foods as follows: (a) 0% of DV consumed in the form of fortified food (0% DV), (b) >0-50% of the DV in the form of fortified food (>0-50% DV), or (c) >50% of DV in the form of fortified food (>50% DV).
Fourth, using the two 24-h dietary intake recalls per person, Current DV and Updated DV models were run independently using the Original and Modified FDs, respectively, to obtain usual nutrient intake for each individual and nutrient distribution. In addition, mean UI, mean percentage below the Estimated Average Requirement (% <EAR), and mean percentage greater than or equal to the Tolerable Upper Limit (% ≥UL) were determined for the population in both models.

Statistical Analysis
We accounted for the NHANES clustered sampling design and oversampling in all analyses and adjusted for differential noncoverage and nonresponse across the two continuous NHANES cycles [30][31][32]. Frequencies were reported for sample size, number of foods that were fortified and enriched, and number of foods with 'Good' or 'Excellent' source nutrient content claims. Means and standard errors of the mean (SEM) were calculated for average UI, percentage below the EAR (% <EAR), and percentage greater than or equal to the UL (% ≥UL). SEs were estimated using Balanced Repeated Replication and NHANES weights were applied. The percent of nutrient intake derived in the form of fortified food was calculated by dividing the total fortified nutrient intake consumed by the DV times 100. We analyzed the data for all adults ≥19 years of age for both sexes combined. All analyses were conducted using SAS version 9.4 and its complex survey-specific procedures (SAS Institute, Cary, NC USA). The NHANES survey protocol has been approved by the NCHS Research Ethics Review Board. Since this study was a secondary data analysis of publicly available federal data, Human Subject Institutional Review Board approval was not required by the Medical University of South Carolina.

Results
With the new FDA labeling rules, the DV was reduced for seven nutrients (vitamins A, thiamin, riboflavin, niacin, B 6 , B 12 , and zinc) and increased for vitamins C and D, and calcium (Table 1). In general, foods in the US are most often fortified with vitamins A and D and enriched with riboflavin, niacin, thiamin, folate, and iron ( Table 2). In our modeling, voluntary fortification was adjusted in the Modified FD to achieve the same numerical %DV for each nutrient and food as found in the Original FD. For the seven nutrients where the DV was reduced, the number of foods predicted to be voluntarily fortified decreased for all except vitamin B 6 ( Table 2). These voluntarily fortified foods (Modified FD) would contain less of these seven nutrients per serving or RACC compared to the Original FD. When a nutrient DV is not changed, the number of fortified foods does not change, as was the case for vitamin E, folate DFE, and iron. When a nutrient Updated DV is higher, such as for vitamins C and D, and calcium, the amount of nutrient added to a voluntarily fortified food was increased in the Modified FD, whereas the amount found intrinsically in foods or added via enrichment was unchanged. The decisions by food manufacturers to voluntarily add more nutrients per serving to maintain the same %DV in a fortified food or to fortify new foods (vs. Original) will depend upon palatability and may require regulatory changes to 'standards of identity' to permit the addition of more vitamins or minerals (21 Code of Federal Regulations 130.3) [28]. In our modeling, we discounted consideration of palatability or standards of identity limitations, to estimate the number of foods that would be predicted to carry 'Good' and 'Excellent' source claims. As shown in Table 3, we found predicted increases for the B vitamins, vitamin A, and zinc for 'Good' and 'Excellent' claims, even though nutrient density is unchanged (intrinsic and/or enrichment foods) or diminished through voluntary fortification. Presuming manufacturers add calcium and vitamins C and D to fortified foods to maintain the same %DV, the number of 'Good' and 'Excellent' source foods is reduced. This reduction occurs because the intrinsic content of a food or the enrichment amount allowed by 'standards of identity' regulations is insufficient to maintain the higher nutrient content claim. For example, US regulations (Code of Federal Regulations 131.110) [33] permit the addition of 400 International Units (IU) of vitamin D per quart (400 IU or 10 µg per 32 fluid ounces) of cow's milk [17]. With the Updated DV (20 µg) being twice the Current DV (400 IU), an eight-fluid ounce glass of milk containing 100 IU (2.4 µg) vitamin D will be downgraded from an 'Excellent' source (25% Current DV) to a 'Good' source (12.5% Updated DV) of vitamin D.  Vitamin B 12  1025  2101  579  1189  1150  2251  633  1326  Vitamin A  267  318  350  460  280  365  342  474  Zinc  670  880  482  721  682  911  466  691  Riboflavin  1284  1451  494  849  1318  1452  526  906  Niacin  974  960  915  1202  989  957  969  1266  Thiamin  756  872  765  938  829  958  762  940  Vitamin B 6  844  1025  480  591  910  1093  547  673  Vitamin E  420  420  223  223  484  484  245  245  Folate 2  839  839  845  845  865  865  843  843  Iron  887  887  434  434  869  869  437  437  Calcium  752  516  334  229  794  504  332  236  Vitamin C  481  389  717  478  546  422  764  513  Vitamin D  357  268  239  200  404  296  291  243  Table 4 shows that the Updated DV model predicts a small change in the number of adults aged ≥19 years consuming foods that are not voluntarily fortified with the exception of vitamin A and to a lesser extent, zinc. However, the Updated DV model predicts that more individuals should be obtaining >50% of the DV for vitamin D from fortified foods, i.e., more than 10 µg per day. The Updated DV model predicts an appreciable drop in the number of individuals obtaining vitamin A from fortified foods, i.e., >50% and >0-50% DV categories for vitamin A. This downward shift in consumption of fortified foods is exacerbated given the lower DV target in the Updated DV than the Current DV model. The Updated DV model finds a small increase in the number of people not consuming foods voluntarily fortified with B vitamins, except for niacin, vitamin B 6 , and folic acid, and a large increase (40%) in the number of people not consuming foods fortified with vitamin A-a "shortfall nutrient" of national concern [8]. The number of foods voluntarily fortified with calcium, vitamin C, and vitamin D was held constant in the model; predictably, the number of people consuming these fortified foods will not change.  Table 5 includes the UIs and the percent of individuals in the US with intakes below the EAR and above or equal to the UL (as available) for the ten modeled nutrients. Since the DVs were unchanged for vitamin E, folate and iron, UIs did not differ between the two models and these nutrients were excluded from these analyses. Comparing the two models with the Updated DV, the overall mean UI of vitamin B 12 and vitamin A is predicted to decrease, whereas the mean UI for vitamin C and vitamin D (almost double) and calcium is predicted to increase. The mean UI for thiamin, riboflavin, niacin, vitamin B 6 , and zinc is predicted to remain relatively unchanged. The model predicts a reduction in %<EAR for vitamins C and D and calcium along with an increase in the percentage of the population exceeding the UL for calcium. Despite a doubling of DV for vitamin D and an assumption that food manufacturers reformulate fortified food products to maintain the same %DV, approximately 70% of adults still are predicted to have a vitamin D intake <EAR. For all seven nutrients with a decreased DV, there is an increase in the percentage of the population below the EAR (Table 5). With the change in DVs, the only predicted decrease in the percentage of the population consuming more than or equal to the UL is for niacin.
Appendix B Tables A2-A5 provide comparison modeling data on adult age group subsets: 19-30 years; 31-50 years; 51-70 years; >70 years. Within these age groups, the same trends were observed as with the overall adult intakes with the Current DV and Updated DV models. However, comparisons of adults aged 19-30 years (vs. all ages reported in Table 5) predict an increased proportion of this age group would have nutrient intakes below the EAR, especially for vitamin A, with the exception of vitamin B 6 and calcium. Based on the %<EAR, adults aged 31-50 years are predicted to be generally better nourished than the overall population, whereas proportionally more adults aged 51-70 years were estimated to have intakes <EAR for vitamin B 6 and ≥UL for calcium. Compared with the overall adult population shown in Table 5, a lower percentage of adults aged >70 years would be expected to have intakes <EAR for vitamins A, C and D and more individuals of this age group would be expected to ingest <EAR for the B vitamins.

Discussion
Food availability and personal food choices affect dietary nutrient density and nutrient intake. When nutrient labeling requirements, such as the DV and targets for 'Good' and 'Excellent' nutrient claims are changed, the nutrient density of many food choices also may be affected [17]. The goal of this study was to determine the potential impact of regulatory application of newer FNB nutrient recommendations and codification as Updated DVs used in nutrition labeling on voluntary food fortification and the UI of select essential nutrients among American adults. The DV was decreased for 14 nutrients and increased for eight [17]. Using the ILSI NA-FD and NHANES datasets, we modeled 13 nutrients. Our Updated DV model predicts a reduction in the number of voluntarily fortified foods for six of the seven nutrients with a decreased DV, the exception being vitamin B 6 .
As part of this modeling approach, we did not transform a non-fortified food to a fortified food. Thus, the number of voluntarily fortified foods in the Modified FD was not altered for nutrients where the DV was increased (vitamins C and D and calcium), the DV was held constant (folate, iron), or the chemical form of the vitamin being added could not be determined (vitamin E and folate). However, for nutrients where the Updated DV is lower than the Current DV, food manufacturers have a choice to reduce voluntary fortification and maintain the same %DV claim on a food product. The Updated DV model predicts that a lowering of the DV may lead to a~40% reduction in the number of foods voluntarily fortified with vitamin A and an increased number of foods meeting 'Good' and 'Excellent' nutrient content claims with a concomitant lower nutrient density.
The consumption of enriched and fortified foods is associated with increased nutrient intake and only a small percentage of US adults having total UIs below the EAR for vitamins B 6 and B 12 , thiamin, riboflavin, niacin, folate, iron, copper, and selenium [2,3,6]. Based on this analysis, mandatory food enrichment, not voluntary food fortification, is primarily responsible for UI above the EAR for thiamin, riboflavin, niacin, folate, and iron. To benefit the US population, both enriched and fortified foods need to be available in the marketplace and then selected by the consumer for consumption. For nutrients where the DV increased, e.g., vitamins C and D and calcium, and assuming food manufacturers increase voluntary fortification to maintain the same %DV, these fortified foods should become more nutrient dense. However, it is worth noting that the intrinsic and enriched contribution of these nutrients will not be sufficient for foods which currently carry 'Good' or 'Excellent' claims to continue to do so under the Updated DV regulations. There appears to be an opportunity to modify US regulations pertaining to standards of identity and enrichment of foods to allow the addition of more vitamins and minerals and align with updated DVs.
Limitations to our analysis include that dietary intakes based on 24-h dietary recalls are subject to misreporting; however, the newer multiple-pass method of dietary interviewing used to collect the data for the NHANES cycles included in this modeling has been shown to be much improved and significantly reduces bias [37][38][39]. Another limitation is that the vitamin E and folate/folic acid content of foods in the database could not be verified because it is not possible to ascertain the chemical structure added to foods or the conversion factor used for the label. This is unfortunate given the low intake of vitamin E among the US population [2,3,6], and the importance of folic acid in fetal development [40,41]. In addition, the food database used does not capture every food in the marketplace and food manufacturers may make different voluntary fortification adjustments. The analysis also does not evaluate the impact on consumer food choices when comparing food labels with new DVs and related nutrient content, e.g., 'good' source, claims. Hopefully, results from this analysis may help inform food manufacturers in product development and help identify additional opportunities, e.g., regulatory changes to standards of identity and levels of enrichment, to improve the nutrient intake of US adults.
There are several strengths to this analysis. The estimates are nationally representative of the US population and based on the ILSI NA-FD that reports the intrinsic, enriched (mandatory) and voluntary fortification of each food serving (Original FD). The nutrient database was replicated, new DVs were applied, and the need for voluntary fortification was determined for each nutrient in every food (Modified FD). This created two databases that were used to estimate two sets of nutrient intakes (Current DV and Updated DV) using the same dietary recalls and statistical methods. This means that the UIs, EARs, and ULs are identical in 24-h dietary recall data and foods within the database.

Conclusions
This analysis highlights the fact that changing the DV may affect the nutrient density of foods and nutrient content claims. When the DV increases, it may lead to a downgraded nutrient content claim, i.e., from an 'excellent' to a 'good' source or from a 'good' source to no claim, unless voluntary fortification is increased. When the DV is decreased, the number of foods eligible to carry 'Good' and 'Excellent' nutrient content claims increase without any change in nutrient density. There may also be an~40% reduction in the number of foods voluntarily fortified with vitamin A. Such changes in nutrient density of foods affect UI and % below the EAR. The Updated DV model predicts that more people will not consume foods voluntarily fortified with vitamins A, thiamin, riboflavin niacin, B 6 , B 12 , and zinc. Our modeling also predicts that there will be a decrease in the proportion of the population obtaining up to 50% or >50% of their nutrient DV by means of voluntarily fortification. This translates to lower UI in the Updated DV Model for vitamins A, niacin B 12 , and zinc, but not thiamin, riboflavin, and vitamin B 6 . The greatest reduction in UI is predicted for vitamin A. This analysis found that 75% of adults currently consume foods fortified with vitamin A, and the Updated DV model predicts this will decrease to 65%, which is troubling because vitamin A has been recognized as a nutrient already deemed of national public health concern by the 2015 Dietary Guidelines Advisory Committee [8]. Separate adult subgroup age analysis predicted that the highest proportion of individuals with intakes < EAR for vitamin A would be individuals age [19][20][21][22][23][24][25][26][27][28][29][30] year. Even with a significant increase in the DV for vitamin D, our modeling still predicts that approximately 70% of the US adult population will have a UI < the EAR for this nutrient. A number of studies have demonstrated the value of fortification and enrichment in the US food supply [1,3]. As the US continues to address the related issues of food deserts, obesity, and shortfall nutrients [8,42,43], it is disconcerting to find that approximately 70% of adults also report not consuming any fortified foods. Funding: This work was supported by the International Life Sciences Institute (ILSI) North America Fortification Committee. ILSI North America is a public, nonprofit science foundation that provides a forum to advance understanding of scientific issues related to the nutritional quality and safety of the food supply. ILSI North America receives support primarily from its industry membership. ILSI North America had no role in the design, analysis, interpretation, or presentation of the data and results. Church & Dwight provided an unencumbered gift to help pay publication costs without playing a role in the design, analysis, interpretation, or presentation of the data and results.