Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

A three-component Breakfast Quality Score (BQS) to evaluate the nutrient density of breakfast meals

A three-component Breakfast Quality Score (BQS) to evaluate the nutrient density of breakfast meals TYPE Conceptual Analysis PUBLISHED 28 September 2023 DOI 10.3389/fnut.2023.1213065 A three-component Breakfast OPEN ACCESS Quality Score (BQS) to evaluate EDITED BY George Pounis, the nutrient density of breakfast Harokopio University, Greece REVIEWED BY meals Malcolm Riley, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia 1 1 2 Owen Kelly, * Romane Poinsot , Matthieu Maillot , Gabriel Masset and Sam Houston State University, United States Adam Drewnowski *CORRESPONDENCE 1 2 3 Matthieu Maillot MS-Nutrition, Marseille, France, Cereal Partners Worldwide, Lausanne, Switzerland, Center for Public [email protected] Health Nutrition, University of Washington, Seattle, WA, United States RECEIVED 27 April 2023 ACCEPTED 01 September 2023 PUBLISHED 28 September 2023 Background: Nutrient profiling methods can be applied to individual foods or to CITATION composite meals. This article introduces a new method to assess the nutrient Poinsot R, Maillot M, Masset G and density of breakfast meals. Drewnowski A (2023) A three-component Breakfast Quality Score (BQS) to evaluate the Objective: This study aimed to develop a new breakfast quality score (BQS), nutrient density of breakfast meals. based on the nutrient standards previously published by the International Breakfast Front. Nutr. 10:1213065. Research Initiative (IBRI) consortium. doi: 10.3389/fnut.2023.1213065 Methods: BQS was composed of three sub-scores derived from the weighted COPYRIGHT © 2023 Poinsot, Maillot, Masset and arithmetic mean of corresponding nutrient adequacy: an eLIMf sub-score (energy, Drewnowski. This is an open-access article saturated fat, free sugars, and sodium), a PF (protein and fiber) sub-score, distributed under the terms of the Creative and a VMn micronutrient sub-score, where n varied from 0 to 14. The Commons Attribution License (CC BY). The use, 1−14 distribution or reproduction in other forums is eects of assigning dierent weights to the eLIMf, PF, and VMn were explored permitted, provided the original author(s) and in four alternative models. The micronutrients were calcium, iron, potassium, the copyright owner(s) are credited and that magnesium, zinc, vitamin A, thiamin, riboflavin, niacin, vitamin B5, vitamin B6, the original publication in this journal is cited, in accordance with accepted academic practice. vitamin B12, vitamin C, and vitamin D. Micronutrient permutations were used to No use, distribution or reproduction is develop alternate VMn sub-scores. The breakfast database used in this study 1−14 permitted which does not comply with these came from all breakfasts declared as consumed by adults (>18 years old) in the terms. French dietary survey INCA3. All models were tested with respect to the Nutrient Rich Food Index (NRF9.3). BQS sensitivity was tested using three prototype French breakfasts, for which improvements were made. Results: The correlations of the models with NRF9.3 improved when the VMn >3 sub-score (n > 3) was included alongside the PF and eLIMf sub-scores. The model with (PF+VMn) and eLIMf each accounting for 50% of the total score showed the highest correlations with NRF9.3 and was the preferred final score (i.e., BQS). BQS was sensitive to the changing quality of three prototype breakfasts defined as tartine, sandwich, and cereal. Conclusion: The proposed BQS was shown to valuably rank the nutritional density of breakfast meals against a set of nutrient recommendations. It includes nutrients to limit along with protein, fiber, and a variable number of micronutrients to encourage. The flexible VMn sub-score allows for the evaluation of breakfast quality even when nutrient composition data are limited. KEYWORDS breakfast, nutrients, score, nutritional recommendations, meal, nutrient density, nutrient profiling Frontiers in Nutrition 01 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Using the French National Dietary Survey INCA3, alternative 1. Introduction scores were tested with respect to the NRF9.3 nutrient density score. The nutrient density of the breakfast meal was then The nutritional value of the breakfast meal can be assessed in compared across the tertiles of BQS. BQS sensitivity to changes a number of ways (1, 2). In their simplest form, scores of breakfast in breakfast composition was tested using three French breakfasts, quality award points to food groups or nutrients to encourage and identified as tartine, sandwich, and cereal. subtract points for food groups and nutrients to limit. A 10-point breakfast quality index (BQI) published in 2012 (2) awarded points for breakfast energy in the range of 20–25% of total daily intake and for the presence of cereals, fruits/vegetables, dairy products, 2. Methods monounsaturated fats, and calcium. An additional point was given for having cereals, fruit/vegetables, and dairy in the same meal (2). 2.1. INCA 3 and CIQUAL databases The same BQI subtracted points for the breakfast content of sugars and saturated and trans fats (2). Dietary data on breakfast consumption in France came from Nutrient profiling (NP) methods generally use both positive the INCA3 study on dietary intakes of the French population (24). and negative components to assess the nutrient density of meals (3), INCA3 is based on three non-consecutive 24 h dietary recalls for dishes (4), or individual foods (5). The Nutri-Score (6), the Health 1,907 men and 2,207 women (2 weekdays and 1 weekend day). Only Star Rating (7), and the Nutrient Rich Food Index (NRF9.3) (7) data for adults aged ≥18 years (n = 2,121) were analyzed. As part each have positive and negative sub-scores. Negative sub-scores are of the dietary intake assessment, participants needed to name the typically driven by saturated and trans fats, total or added sugars, eating occasion. There were 10 possibilities (including “breakfast”). and sodium, but can also include energy (6). While the choice Only foods consumed during the declared breakfast meal were of nutrients to encourage can vary, the Nutri-Score, Health Star analyzed. Breakfasts consisting of coffee or tea only (sweetened or Rating, and NRF9.3 each include both protein and fiber (6–8). The not) were excluded. A total of 4,478 breakfasts were analyzed. Nutri-Score does not include any vitamins or minerals, preferring The associated ANSES CIQUAL 2016 nutrient composition to award points based on the foods’ content of pre-selected food database (25) provides energy and nutrient values per 100 g edible groups (fruit, vegetables, and nuts) (6). By contrast, the NRF9.3 portion for all the foods consumed by INCA3 participants. Free score complements protein and fiber with calcium, iron, potassium, sugars were estimated from added sugars (26). Portion sizes were magnesium, vitamin A, vitamin C, and vitamin E (now replaced based on a previous study in France (27). by vitamin D) (7). Depending on the NP model, the number n of vitamins and minerals has varied from 3 to as many as 23 (9, 10). The International Breakfast Research Initiative (IBRI) has proposed a set of nutrient standards to assess the nutritional value 2.2. Development of the breakfast quality of breakfast meals (11, 12). In a series of IBRI studies, the breakfast score quality of representative populations in the United States (13), Canada (14), France (15), Spain (16), the United Kingdom (17), and 2.2.1. Establishing nutrient standards for BQS Denmark (18) was measured using the Nutrient Rich Food Index Table 1 summarizes the nutrient standards developed by IBRI (7). Other studies have since explored the nutritional contributions (11, 12). Also shown are the formulas for calculating nutrient of the breakfast meal in Latin America (19, 20) and the Philippines adequacy. Upper and lower bounds for energy were set for (21). The IBRI targets for vitamins and minerals at breakfast were the breakfast meal following IBRI recommendations. Maximum based on actual consumption levels during breakfast and on overall recommended values (MaxRV) were set for saturated fats, free nutrient adequacy (22). sugars, and sodium. Minimum recommended values (MinRV) were The present goal was to simplify the large set of nutrient-by- set for protein, fiber, and 14 micronutrients. nutrient IBRI recommendations into an overall breakfast quality IBRI-derived MinRV and MaxRV values for adults were based score (BQS). Alternative scores were all comprised of the same on 10 or 20% of daily recommendations from WHO/Codex (28) three components. The negative sub-score was eLIMf (energy, and from the WHO guidelines for nutrients of public health saturated fat, free sugars, and sodium). The two positive sub-scores concern (free sugars, saturated fats, and sodium) (29). For breakfast were PF (protein and fiber) and VMn, the latter composed of a energy, a maximum adequacy score of 100% was achieved when variable number of vitamins and minerals. As in other NP models, breakfast energy intakes fell within the 300–500 kcal range. The the final quality score was based on the difference between the lower bound (0%) was set at 0 kcal. The upper bound (also 0%) negative and positive sub-scores (6–8). The present innovation was was set at 800 kcal, i.e., 500 kcal + 300 kcal. When breakfast energy to vary the number of vitamins and minerals in VMn from 0 to 14 was lower than 300 kcal or higher than 500, the adequacy scores and to explore the impact of micronutrient permutations (9, 10). diminished in proportion to intakes relative to the recommended We also explored the effects of assigning differential weights to the values. When the breakfast energy was higher than the upper bound eLIMf, PF, and VMn sub-scores on the total score. This was done (800 kcal), the adequacy score became negative. because the weighting of the sub-scores affects the final evaluation For saturated fat, sodium, and free sugars, a maximum (23). Although some NP models are driven by negative elements, adequacy score (100%) was achieved when breakfast intakes were energy, sugar, and fat, others are weighted to favor protein, fiber, <MaxRV. The upper bound (0%) was set at 2 × MaxRV. Adequacy vitamins, and minerals. diminished in proportion from MaxRV to the upper bound (2 × Frontiers in Nutrition 02 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 1 Calculation of nutrient adequacy for breakfast quality score. Sub- Nutrient Unit Reference values Boundaries Percentage of score adequacy ∗ ∗ MinRV MaxRV Lower Upper bound bound obsi eLIMf Energy kcal 300 500 0 500 + 300 = 800 Adeq = min( × i min reco upper −obs i i 100, max ×100, 100) upper −reco SFAs %EBI 10 2 × 10 = 20 Free sugars %EBI 10 2 × 10 = 20 Sodium mg 400 2 × 400 = 800 PF Proteins g 10 0 Adeq = upper −obs i i min( × max upper −reco 100, 100) Fibers g 5 0 VMn Calcium mg 250 0 Adeq = obs min( × 100, 100) min reco Iron mg 2.8 0 Magnesium mg 62 0 Potassium mg 700 0 Zinc mg 2.2 0 Vitamin A mg 80 0 Vitamin B1 mg 0.24 0 Vitamin B2 mg 0.36 0 Vitamin B3 mg 3.75 0 Vitamin B6 mg 0.26 0 Folate μg 80 0 Vitamin B12 μg 0.48 0 Vitamin C mg 20 0 Vitamin D μg 1 0 max min obs , observed breakfast intake of nutrient i; reco , minimum recommended value for nutrient i; upper , upper bound for nutrient i; reco , maximum recommended value for nutrient i i i i i; EBI, energy breakfast intake; n, number of micronutrients. Free sugars refer to all monosaccharides and disaccharides added to foods by the manufacturer, cook, or consumer, and sugars naturally present in honey, syrups, and fruit juice. The MinRV and MaxRV values came from IBRI recommendations for adults and were based on 10 or 20% of daily recommendations from WHO/Codex. MaxRV). When the observed values at breakfast were higher than lower the content of nutrients to be limited, the higher the the upper bounds, the adequacy scores became negative. eLIMf sub-score. For protein, fiber, vitamins, and minerals, a maximum 1 i=4 adequacy score (100%) was achieved when breakfast nutrient eLIMf = × Adeq , (1) 4 i=1 intakes were >MinRV. The lower bound (0%) was set at zero i = energy, saturated fats, free sugars, sodium , consumption. Nutrient adequacy between these two points was eLIMf ∈ [−∞; 100] calculated by dividing the breakfast value by the recommended value and multiplying by 100. The second sub-score was named PF (Equation 2) since it contained protein and fiber adequacy. 1 i=2 PF = × Adeq , (2) 2.2.2. Definitions of BQS sub-scores 2 i=1 Nutrient adequacies were grouped into one of three sub-scores: i = proteins, fibers , PF ∈ [0; 100], eLIMf, PF, or VMn (Table 1). The sub-scores were equal to the arithmetic mean of corresponding nutrient adequacies (Equations In previous studies, NP profiles have used a wide range of 1–3). The eLIMf sub-score (Equation 1) included energy, saturated micronutrients (9, 10). The third sub-score VMn (Equation 3) was fat, sodium, and free sugars (rather than total or added) adequacies. based on a variable number n of vitamins and minerals (n ranges The eLIMf sub-score could be either negative or positive. The from 0 to 14) adequacies. The 14 micronutrients for which IBRI Frontiers in Nutrition 03 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 recommendations were available were calcium, iron, potassium, 2.3. Testing alternative BQS models magnesium, zinc, vitamin A, thiamin, riboflavin, niacin, vitamin B5, vitamin B6, vitamin B12, vitamin C, and vitamin D. The VMn Scores obtained using alternative BQS models on the INCA3 sub-score could only be positive. breakfast meals were compared to scores for the same breakfast meals generated by the NRF9.3 nutrient profiling model (7) and P to scores obtained with the same BQS model where n = 14 (called 1 i=n VMn = Adeq , (3) n i=1 “complete” BQS). The NRF9.3 is the sum of percent daily values i ∈ {0, micronutrient , . . . micronutrient } , 1 n (DV) for nine nutrients to encourage (proteins, fibers, calcium, iron, magnesium, potassium, vitamin C, vitamin A, and vitamin n ≤ 14, VMn ∈ [0; 100] D) minus the sum of percent DV for three nutrients to limit (saturated fats, added sugars, and sodium). All values are calculated All possible combinations of the 14 micronutrients were per 100 kcal and capped at 100% (7). Correlations were estimated systematically tested. The number of possible combinations of using the Spearman rank-order correlation coefficient. For partial micronutrients depended on their number. For VM = 0 and for VMn scores where n < 14, correlations were obtained for each VM = 14, there was only one possible option. For VM = 1 and for combination of micronutrients, and their distribution was analyzed VM = 13, there were 14 possible choices. For VM = 2 and for VM using boxplots. = 12, there were 91 possible combinations of micronutrients. For The model that correlated best with NRF9.3 was selected and VM = 3 and VM = 11, there were 364 and so on. For VM = 7, there called “BQS” in the rest of the article. were 3,432 possible combinations. BQS values for 4,478 breakfasts in the INCA3 database were split into tertiles. The general linear model then compared values of energy and nutrients (for which IBRI recommendations were available), sub-scores eLIMf, PF, and VM, and grams of eight 2.2.3. Alternative BQS models with dierent dietary components across the BQS tertiles. The eight dietary sub-score weights components of interest were fruits and vegetables, whole-grain, Alternative BQS models were calculated as a weighted refined-grain, milk and dairy, plant fats, animal fats, sugary foods, mean of the three sub-scores following Equation 4. In and sweet-tasting beverages (e.g., soda and fruit juices). A post-hoc every case, the minimum score was 0, so there were no comparison (Tukey’s HSD test) was performed when the difference negative scores. Scores ranged from 0 to 100, with the was significant. highest scores given to those breakfasts that met all of the Statistical analyses used the R software version 4.1. The level of IBRI recommendations. significance was set to 5% for all tests. Breakfast quality score = α × eLIMf + β × PF + γ × VMn 3. Results (4) 3.1. Alternative BQS models with complete VMn sub-scores Where α, β, and γ are weights ranging from zero to 1, and their sum is equal to 1. Breakfast meals in the INCA 3 database were evaluated using Four alternative BQS models were tested, each with a different the four alternative BQS models. All four BQS models were highly sub-score weighting scheme (Table 2). correlated with each other (range r = 0.76 to r = 0.99; results not In the balanced model, the eLIMf sub-score accounts for 50% shown). With all the alternative BQS models, none of the 4,478 of the BQS, and the sum of PF and VMn sub-scores (PF+VMn), breakfasts got a 100% adequacy score. Mean BQS values ranged where n ≤ 14, also accounts for 50% of the BQS. The weights from 51.6% (balanced model) to 58.5% (unweighted model) and α, β, and γ differ between models and are shown in Table 2. In the were comparable for the four models. unweighted model, all 6+n elements are equivalent. Each element The balanced model showed the highest correlations with accounts for [1/(6+n)] × 100%] of the total score as VMn rises NRF9.3 nutrient density scores (r = 0.55) and the lowest from 0 to 14. correlations with energy density (r = −0.15) of breakfasts in the In the micronutrient model, the VMn sub-score (n ≤ 14) INCA database (Figure 2). The balanced model had a moderate now accounts for 50% of total BQS. The eLIMf and PF sub- correlation with LIM. scores together account for 50% of the BQS total. In the three-way model, the eLIMf sub-score, the PF sub-score, and the VMn sub-score (n ≤ 14) each account for 1:3 of the 3.2. Alternative BQS models with partial BQS total. Pie charts are an alternative way of visualizing BQS weights VMn sub-scores when n = 14, and they are presented in the last column of Table 2. Figure 1 shows the shift in weights for the four Figure 2 shows that correlations with NRF9.3 improved as alternative weighting schemes for BQS sub-scores as the number the VMn sub-score incorporated more vitamins and minerals. of vitamins and minerals in the VMn sub-score rises from 0 For all four alternative BQS models, correlations with NRF9.3 to 14. were weakest when no micronutrients were included (VM ). n =0 Frontiers in Nutrition 04 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 2 Weighting scheme of component sub-scores for the four alternative BQS models (with n, the number of micronutrients). Alternative models Weight of sub-score Weight of each sub-score, where n = 14 eLIMf (α) PM (β) VMn (γ ) ( ) 1 ( ) 2+n 2+n Balanced model 4 2 n Unweighted model 6+n 6+n 6+n 1 4 1 1 2 1 1 Micronutrient model × = × = 2 6 3 2 6 3 2 1 1 1 Three-way model 3 3 3 Correlations improved when the VMn component included at 3.3. Testing the performance of BQS (i.e., least three vitamins or minerals. For balanced, micronutrient, the balanced model) and three-way models, correlations with NRF9.3 increased when the VMn number of micronutrients increased. For the 3.3.1. Distribution of BQS in INCA3 adult unweighted model, the correlation with NRF9.3 increased up to 5– breakfasts 6 micronutrients and then decreased. The average correlations of The distribution of the BQS values in INCA3 adult breakfasts is NRF9.3 with the balanced model were very close to those with the shown in Figure 4. Only 3% of the breakfasts were cut at 0%. The micronutrient model; however, according to the permutation, the score of 0% means that the breakfast provided more negative points correlations with the micronutrient model were less homogeneous (from the eLIMf sub-score) than positive points from PF and VM. than those with the balanced model (see the size of boxes The BQS distribution exhibits a normal shape. The average (51.6%) in Figure 2). and median (53.1%) balanced BQS were close to 50%. All average correlations were highest for the balanced model. For the balanced model, when VM , correlation values ranged n =3 from 0.43 to almost 0.60. This variability was related not only to 3.3.2. Nutritional and dietary components by the number but also to specific combinations of micronutrients. tertiles of BQS According to Figure 3, the correlation between complete and Percent nutrient adequacy in the first tertile of BQS ranged partially balanced BQS stayed high (minimum 0.93 without from 0 to 43.7% and in the third tertile from 62.1 to 98.4%. Table 3 micronutrients) even when the number of micronutrients included confirmed that sub-score eLIMf increased as well as sub-scores in the score decreased. However, it was not clear which vitamins PF and VM when BQS values increased. Indeed, the amounts of and minerals were the most important. Based on these results, nutrients to limit decreased between low and medium and between the balanced model was selected as BQS and was subject to medium and high tertiles of BQS, and, except for vitamin A, the further testing. amounts of nutrients to encourage increased between low and high Frontiers in Nutrition 05 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 1 Weight (in %) of each sub-score in the four alternative BQS models, where the number of micronutrients ranged from 0 to 14. tertiles of BQS. For proteins, fiber, iron, magnesium, potassium, by our group and available at https://ms-nutrition.com/web-app/ and vitamins B1, B3, B9, and C, the difference between the low breakfast-calculator/. The version used for the article contained the and medium tertiles was not significant, but the difference between nutritional composition of foods that corresponded to the dietary the medium and high tertiles was significant. Likewise, amounts intake data (INCA3) and has since been updated to reflect the most of healthier food groups such as fruits and vegetables, whole- up-to-date food composition database (i.e., CIQUAL 2020). None grain foods, milk and dairy products, and plant-based fats were of the eLIMf sub-scores were negative, so BQS corresponded to the higher. Conversely, amounts of less healthy food groups such as stacking of the sub-scores in Figure 5. refined-grained foods, animal fats, sugary foods, and sweet-tasting The balanced BQS values improved with increasing versions beverages were lower in breakfast in low tertiles than in breakfast except between v2 and v3 in “sandwich” breakfast, where they in medium or high tertiles. Breakfasts in the highest tertile of slightly decreased. For the “tartine” and “sandwich” breakfasts, BQS were thus of higher nutritional quality than breakfasts in the BQS increased from 46.2 and 55.3% (v0) to 57.6 and 60.8% medium or low tertile of BQS. (v1), respectively, because it enabled the fiber recommendation to be met (Supplementary Figures 1, 2). For the “tartine” breakfast, changes between v1 and v2 (64.7%) and v2 and 3.3.3. Sensitivity through three examples of v3 (73.8%) aimed at increasing the eLIMf sub-score, and changes between v3 and v4 (84%) and v4 and v5 (88.6%) breakfasts aimed at increasing the VM sub-score. Adding an orange and The sensitivity of the BQS was tested using three alternative plain yogurt in v4 and v5 increased calcium and vitamins breakfasts. Breakfast 1 (“tartine”) was a baguette with jam; C, B1, B2, and B12 while staying within the recommended Breakfast 2 (“cereal”) was ready-to-eat (RTE) cereal and milk; energy range. and Breakfast 3 (“sandwich”) was a savory sandwich. Four For the “sandwich” breakfast, the macronutrient profile was to five versions of each breakfast (v0, v1, v2, v3, v4, and more favorable thanks to the replacement of butter with low- v5) were constructed (Table 4). Breakfast nutrient content was fat butter from v1 (60.8%) to v2 (70.6%), even though the mean calculated using the “Breakfast Calculator” online tool, developed Frontiers in Nutrition 06 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 2 Spearman correlations between each alternative BQS model and NRF9.3 for all combinations of VMn ranging from 0 to 14 micronutrients. The horizontal reference line is set to 0.55. Frontiers in Nutrition 07 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 3 Distribution of Spearman correlation between the complete balanced model (based on 14 micronutrients) and the balanced model when VM ranged from 0 to 14 micronutrients. FIGURE 4 Distribution of BQS in INCA3 adults breakfast. BQS, BQS balanced model. micronutrient adequacy decreased. The addition of milk to the cereals with low-sugar cereals, which reduced free sugars coffee (i.e., white coffee instead of black coffee) in v3 decreased the to below the threshold level. Adding fruits, almonds, and total score because the increases in mineral and vitamin contents sunflower seeds in v2, v3, and v4 improved fiber, potassium, brought by the milk, such as calcium and vitamin B2, were not and calcium adequacy (Supplementary Figure 3). The cereal sufficient to compensate for the increase in energy and SFA. On breakfast v4 was the only one to restrict eLIMf (nutrients the other hand, adding oranges increases the score even if it adds to limit) and therefore scored the highest overall (93.2%) energy to the breakfast. among the 16 breakfasts (Figure 5). The last versions of For the “cereal” breakfast, the balanced BQS increased “sandwich” and “tartine” breakfasts were still too high in sodium from 69.4 (v0) to 80.3% (v1) because v1 replaced chocolate (Supplementary Figures 1, 2). Frontiers in Nutrition 08 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 3 Average BQS , sub-scores, energy, and nutrients included in IBRI recommendations and dietary components by tertiles of BQS. Tertile of BQS Low [0; 43.7] Medium [43.7; 62.1] High [62.1; 98.4] p-value $ ∗∗∗ BQS 27.6 52.9 74.3 ∗∗∗ Sub-score eLIMf −0.18 23.1 35.8 a b b ∗∗∗ Energy (kcal) 482 410 411 ∗∗∗ SFA (% EBI) 14.6 13.6 11.5 ∗∗∗ Free sugars (% EBI) 28.4 16.6 10.6 ∗∗∗ Sodium 562 485 438 ∗∗∗ Sub-score PF 3.61 4.11 4.98 a a b ∗∗∗ Proteins (g) 9.90 10.0 14.3 a a b ∗∗∗ Fibers (g) 3.47 3.35 4.19 ∗∗∗ Sub-score VM 23.0 25.7 33.5 ∗∗∗ Calcium (mg) 130 163 324 a a b ∗∗∗ Iron (mg) 1.60 1.65 2.15 a a b ∗∗∗ Magnesium (mg) 70.2 73.6 92.7 a a b ∗∗∗ Potassium (mg) 597 630 829 ∗∗∗ Zinc (mg) 0.95 1.04 1.65 Vitamin A (μg RAE) 133 92.7 98.2 0.174 a a b ∗∗∗ Vitamin B1 (mg) 0.25 0.26 0.38 ∗∗∗ Vitamin B2 (mg) 0.34 0.41 0.69 a a b ∗∗∗ Vitamin B3 (mg) 2.88 3.11 3.68 ∗∗∗ Vitamin B6 (mg) 0.21 0.23 0.36 a a b ∗∗∗ Vitamin B9 (mg) 56.7 58.8 77.6 ∗∗∗ Vitamin B12 (μg) 0.41 0.53 1.09 a a b ∗∗∗ Vitamin C (mg) 18.8 18.2 26.4 ∗∗∗ Vitamin D (μg) 0.41 0.55 0.97 Dietary components ∗∗∗ Fruits and vegetables (g) 8.38 14.2 34.7 ∗∗∗ Whole-grain foods (g) 6.70 9.45 14.3 ∗∗∗ Refined-grain food (g) 50.2 38.1 25.2 ∗∗∗ Milk and dairy products (g) 47.0 71.9 196 a b b ∗∗∗ Plant fats (g) 1.15 2.33 2.80 ∗∗∗ Animal fats (g) 7.25 4.92 2.27 a b b ∗∗∗ Sugary foods (g) 57.3 41.8 40.0 ∗∗∗ Sweet-tasting beverages (g) 55.3 46.2 37.7 $ ∗∗∗ BQS, BQS balanced model. p < 0.001. Same index letters (e.g., a and a) indicate that there is no significant difference between the two tertiles, and different index letters (e.g., a and b) indicate that the difference is statistically significant between the two tertiles. No indexes indicate that the difference is significant between the three tertiles (Tukey’s range test). 4. Discussion BQS that was selected from among four alternatives showed the highest correlations with the NRF9.3 index and low correlations Nutrient profiling methods, initially developed for individual with energy density. The present BQS score is composed of three foods, can also be used to assess the nutritional value of meals. This distinct sub-scores. Each sub-score had been used in previous article introduces a new way to assess breakfast quality that was nutrient profiling models but in different ways. The PF (protein specifically designed to follow a set of published recommendations and fiber) component has been used by nutri-score, HSR, and the for the breakfast meal from the IBRI consortium (11, 12). The NRF9.3 nutrient density index (6–8). The eLIMf sub-score was Frontiers in Nutrition 09 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Frontiers in Nutrition 10 frontiersin.org TABLE 4 Breakfast examples. Breakfast type Food 1 Food 2 Food 3 Food 4 Food 5 Food 6 Food 7 Food 8 Tartine (v0) French baguette (80 g) Butter, unsalted (16 g) Strawberry jam (30 g) Black coffee (200 ml) Tartine (v1) Whole grain baguette Butter, unsalted (16 g) Strawberry jam (30 g) Black coffee (200 ml) (80 g) Tartine (v2) Whole grain baguette Dairy spread, 39–41% Strawberry jam (30 g) Black coffee (200 ml) (80 g) fat (16 g) Tartine (v3) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) (80 g) fat (16 g) Tartine (v4) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) Average fruit (100 g) (80 g) fat (16 g) Tartine (v5) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) Average fruit (100 g) Plain yogurt (125 g) (80 g) fat (16 g) Sandwich (v0) French baguette Roasted chicken (60 g) Butter, unsalted (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v1) Whole grain baguette Roasted chicken (60 g) Butter, unsalted (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v2) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v3) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) White coffee (200 ml) (100 g) Sandwich (v4) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) White coffee (200 ml) Orange (80 g) (100 g) Cereal (v0) Cereal, chocolate, Semi-skimmed milk Tea, no sugar (200 ml) enriched (45 g) (150 ml) Cereal (v1) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) low-sugars, fortified (150 ml) (45 g) Cereal (v2) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) low-sugars, fortified (150 ml) (45 g) Cereal (v3) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) Sunflower seed (7 g) low-sugars, fortified (150 ml) (45 g) Cereal (v4) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) Sunflower seed (7 g) Almond (10 g) low-sugars, enriched (150 ml) (45 g) Bold text indicates that the food or the amount of food is new compared to the previous version of the BF. Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 5 BQS and its sub-scores PF, eLIMf, and VMn for three types of breakfast: “tartine” (A), “sandwich” (B), and “cereal” (C) and their improved versions. BQS, BQS balanced model. close to the negative LIM sub-score (saturated fat, added sugar, and The number of vitamins and minerals was based on the IBRI sodium) used in NRF9.3, but with the addition of energy, a feature recommendations. The performance of VMn scores was tested shared with nutri-score. The nutri-score version of eLIMf includes when the number of vitamins and minerals was allowed to vary energy and total sugars. from 0 to 14. This was done to determine the minimum number of The novelty here was to create a variable VMn sub-score where vitamins and minerals that are necessary to assess breakfast quality. n varied from 0 to 14 and different permutations of nutrients It was found that less than three vitamins and minerals should be were deployed. The present approach differs from that of nutri- critical to derive a robust BQS. The best choice is to conserve the 14 score, which does not include vitamins or minerals but awards micronutrients. However, not all food databases, especially those points for the content of fruits, vegetables, legumes, nuts, and seeds. currently available for low- and middle-income countries, have a Frontiers in Nutrition 11 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 full set of nutrients, so our results demonstrated that using <14 explored in contexts where limited nutrient content information micronutrients could be a viable alternative. The choice of these is available. Finally, the balanced BQS was sensitive to small nutrients should be guided by relevant public health considerations improvements in breakfast quality, suggesting that it may serve as and data availability. Not all nutrients are necessarily consumed in a basis to educate the population on breakfast quality. adequate amounts in the course of a single meal. When it comes There are some limitations. First, validity testing is difficult to dietary inadequacies, the Dietary Guidelines for Americans for meal-specific indicators. The correlations were performed identified calcium, potassium, and vitamin D as shortfall nutrients with another model of nutrient density, the NRF9.3 score. With for the United States (30). Priority nutrients for low- and middle- single meals (as opposed to the total diet), there are no potential income countries were calcium, iron, zinc, folate, iodine, vitamin health outcomes. Finally, the BQS model was applied to an A, and vitamin B12 (31). Of course, the construction of NP models adult population in a single country, France. However, the BQS depends on the availability of nutrient composition data. For that is easily applicable to other breakfast meals based on national reason, it is advantageous to have flexibility in the number of score dietary surveys. elements (32). In the BQS construction, the calculation of the eLIMf sub- 5. Conclusion score followed a different concept compared to other nutrient profiles, which took into account unfavorable nutrients such as The present study introduces a new breakfast quality score saturated fats, sodium, and free sugars. In BQS, the eLIMf sub- (BQS), designed to assess the nutrient adequacy of a single meal– score values ranged from zero whether nutritional content is twice breakfast. Similar in structure to other compensatory nutrient the maximal recommended amount to 100 whether the nutrient profiling models, the BQS introduces a novel flexible VMn sub- contents are below the limits. To penalize breakfast with a high score based on a variable number of vitamins and minerals. The amount of saturated fats, free sugars, or sodium, the sub-score flexibility of the BQS makes it an attractive tool for evaluations becomes linearly negative when nutritional content exceeds twice of breakfast quality in settings where comprehensive nutrient the recommendation. Thanks to this approach, the BQS is able to composition data are not available. discriminate between two different breakfasts with a high amount of unfavorable nutrients. One challenge of nutrient profiling is to adequately weight Ethics statement the respective contributions of positive and negative components. Some existing systems appear to be mainly driven by energy density The studies involving humans were approved by Comité and nutrients (33, 34). In this study, we tested four alternative consultatif sur le traitement de l’information en matière de weighting models for the BQS. The selected algorithm, which gave recherche dans le domaine de la santé. The studies were equal weight to the positive (protein, fibers, and micronutrients) conducted in accordance with the local legislation and institutional and negative (eLIMf) components of the BQS, showed a low requirements. Written informed consent for participation was correlation of the BQS with LIM, or energy density, meaning that not required from the participants or the participants’ legal the selected BQS would be sensitive to changes in both positive and guardians/next of kin in accordance with the national legislation negative components. and institutional requirements. The sensitivity of the BQS score to small changes in the mean composition of breakfasts was illustrated with reference to three Author contributions types of breakfasts. Based on BQS score distributions in French adults, 40% (close to the first tertile, which was 43.7%) appeared GM, RP, and MM conceptualized the study. RP and to be an appropriate cut-off point to identify breakfast that could MM conducted statistical analyses and wrote the first draft be considered nutritionally adequate. Given that the present results manuscript. AD wrote the final manuscript. GM revised the were based on INCA3 data in adults, further work is needed to manuscript. All authors contributed to the article and approved the assess breakfast quality among children and teenagers. submitted version. This study had both strengths and limitations. First, the selected BQS was based on nutrient recommendations that were breakfast- specific as opposed to daily. Second, the balanced BQS was Funding robust, showing good performance even with a limited number of micronutrients. That will be of importance in places where Cereal Partners Worldwide (CPW) funded the study. AD comprehensive nutrient composition data may not be available. In received funding from CPW to conduct the study and write some countries, nutrient composition data are partial, and some the manuscript. MS-Nutrition was financially supported to nutrients are missing altogether. In those cases, it is useful to have conceptualize and conduct the analysis. a flexible and pretested BQS that can be based on the nutrients that are available. This would allow for consistent and harmonized Acknowledgments testing of breakfast quality across multiple locations, including low- and middle-income countries. However, we did not analyze We thank Sinead Hopkins and Diane Zimmermann for their the performance of BQS, considering particular combinations of valuable comments, which helped improve the manuscript. nutrients. The performance of the proposed score needs to be Frontiers in Nutrition 12 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Conflict of interest evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the GM was employed by CPW at the time of the study. RP and publisher. MM are employees of MS-Nutrition. The remaining author declares that the research was Supplementary material conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict The Supplementary Material for this article can be found of interest. online at: https://www.frontiersin.org/articles/10.3389/fnut.2023. 1213065/full#supplementary-material Publisher’s note SUPPLEMENTARY FIGURE 1 Nutrient adequacies of “Tartine” breakfasts: version 0, 1, 2, 3, 4, and 5. All claims expressed in this article are solely those SUPPLEMENTARY FIGURE 2 Nutrient adequacies of “Sandwich” breakfasts: version 0, 1, 2, 3, and 4. of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, SUPPLEMENTARY FIGURE 3 Nutrient adequacies of “Cereal” breakfasts: version 0, 1, 2, 3, and 4. the editors and the reviewers. Any product that may be References 1. Matthys C, De Henauw S, Bellemans M, De Maeyer M, De Backer G. Breakfast breakfast research initiative. Nutrients. (2018) 10:1056. doi: 10.3390/nu100 habits affect overall nutrient profiles in adolescents. Public Health Nutr. (2007) 10:413– 81056 21. doi: 10.1017/S1368980007248049 16. Ruiz E, Ávila JM, Valero T, Rodriguez P, Varela-Moreiras G. Breakfast 2. Monteagudo C, Palacin-Arce A, del Mar Bibiloni M, Pons A, Tur JA, Olea-Serrano consumption in Spain: patterns, nutrient intake and quality. Findings from the ANIBES F, et al. Proposal for a Breakfast Quality Index (BQI) for children and adolescents. study, a study from the international breakfast research initiative. Nutrients. (2018) Public Health Nutr. (2013) 16:639–44. doi: 10.1017/S1368980012003175 10:1324. doi: 10.3390/nu10091324 3. Fern EB, Watzke H, Barclay DV, Roulin A, Drewnowski A. The nutrient balance 17. Gaal S, Kerr MA, Ward M, McNulty H, Livingstone MBE. Breakfast concept: a new quality metric for composite meals and diets. PLoS ONE. (2015) Consumption in the UK: Patterns, nutrient intake and diet quality. a study 10:e0130491. doi: 10.1371/journal.pone.0130491 from the international breakfast research initiative group. Nutrients. (2018) 10:999. doi: 10.3390/nu10080999 4. Furuta C, Jinzu H, Cao L, Drewnowski A, Okabe Y. Nutrient profiling of japanese dishes: the development of a novel ajinomoto group nutrient profiling system. Front 18. Fagt S, Matthiessen J, Thyregod C, Kørup K, Biltoft-Jensen A. Breakfast Nutr. (2022) 9:912148. doi: 10.3389/fnut.2022.912148 in Denmark: Prevalence of consumption, intake of foods, nutrients and dietary quality. A study from the international breakfast research initiative. Nutrients. (2018) 5. Drewnowski A, Fulgoni V. Nutrient profiling of foods: creating a nutrient-rich 10:1085. doi: 10.3390/nu10081085 food index. Nutr Rev. (2008) 66:23–39. doi: 10.1111/j.1753-4887.2007.00003.x 19. Fisberg M, Kovalskys I, Previdelli AN, Pereira JL, Zimberg IZ, Fisberg R, et al. 6. Hercberg S, Touvier M, Salas-Salvado J. Group of European scientists supporting Breakfast Consumption Habit and Its Nutritional Contribution in Latin America: the implementation of Nutri-Score in Europe: The Nutri-Score nutrition label. Int J Results from the ELANS Study. Nutrients. (2020) 12:2397. doi: 10.3390/nu120 Vitam Nutr Res. (2022) 92:147–57. doi: 10.1024/0300-9831/a000722 7. Fulgoni III VL, Keast DR, Drewnowski A. Development and validation of the 20. Kovalskys I, Fisberg M, Previdelli AN, Pereira JL, Zimberg IZ, Guajardo V, nutrient-rich foods index: a tool to measure nutritional quality of foods. J Nutr. (2009) et al. Breakfast in Latin America: evaluation of nutrient and food group intake 139:1549–54. doi: 10.3945/jn.108.101360 toward a nutrient-based recommendation. J Acad Nutr Diet. (2022) 122:1099– 8. Health Star Rating System. Available online at: http://www.healthstarrating.gov. 1113.e3. doi: 10.1016/j.jand.2021.11.012 au/internet/healthstarrating/publishing.nsf/content/home (accessed September 19, 21. Angeles-Agdeppa I, Custodio MRS, Toledo MB. Breakfast in the 2023). Philippines: food and diet quality as analyzed from the 2018 expanded 9. Drewnowski A, Maillot M, Darmon N. Testing nutrient profile models in national nutrition survey. Nutr J. (2022) 21:52. doi: 10.1186/s12937-022- relation to energy density and energy cost. Eur J Clin Nutr. (2009) 63:674– 00804-x 83. doi: 10.1038/ejcn.2008.16 22. Vieux F, Maillot M, D Rehm C, Drewnowski A. Designing optimal 10. Drewnowski A, Maillot M, Darmon N. Should nutrient profiles breakfast for the United States using linear programming and the be based on 100 g, 100 kcal or serving size? Eur J Clin Nutr. (2009) NHANES 2011-2014 database: a study from the international breakfast 63:898–904. doi: 10.1038/ejcn.2008.53 research initiative (IBRI). Nutrients. (2019) 11:1374. doi: 10.3390/nu110 11. Gibney MJ, Barr SI, Bellisle F, Drewnowski A, Fagt S, Hopkins S, et al. Towards an evidence-based recommendation for a balanced breakfast— 23. Drewnowski A, Gonzalez TD, Rehm CD. Balanced hybrid nutrient density a proposal from the international breakfast research initiative. Nutrients. (2018) score compared to nutri-score and health star rating using receiver operating 10:1540. doi: 10.3390/nu10101540 characteristic curve analyses. Front Nutr. (2022) 9:867096. doi: 10.3389/fnut.2022. 12. Gibney MJ, Barr SI, Bellisle F, Drewnowski A, Fagt S, Livingstone B, et al. Breakfast in Human Nutrition: The International Breakfast Research Initiative. 24. French Agency for Food Environmental and Occupational Health & Safety Nutrients. (2018) 10:559. doi: 10.3390/nu10050559 (ANSES). Étude Individuelle Nationale des Consommations Alimentaires 3 (INCA 3); French Agency for Food Environmental and Occupational Health & Safety: 13. Drewnowski A, Rehm CD, Vieux F. Breakfast in the United States: food and Maisons-Alfort, France. (2017). Available online at: https://www.anses.fr/fr/system/ nutrient intakes in relation to diet quality in national health and examination survey files/NUT2014SA0234Ra.pdf (accessed September 19, 2023). 2011-2014. A study from the international breakfast research initiative. Nutrients. (2018) 10:1200. doi: 10.3390/nu10091200 25. French Agency for Food, Environmental and Occupational Health & Safety. ANSES-CIQUAL French Food Composition Table. (2013). Available online at: https:// 14. Barr SI, Vatanparast H, Smith J. Breakfast in Canada: prevalence of consumption, ciqual.anses.fr/ (accessed September 19, 2023). contribution to nutrient and food group intakes, and variability across tertiles of daily diet quality. A study from the international breakfast research initiative. Nutrients. 26. Louie JCY, Moshtaghian H, Boylan S, Flood VM, Rangan AM, Barclay AW, et al. (2018) 10:985. doi: 10.3390/nu10080985 A systematic methodology to estimate added sugar content of foods. Eur J Clin Nutr. (2015) 69:154–61. doi: 10.1038/ejcn.2014.256 15. Bellisle F, Hébel P, Salmon-Legagneur A, Vieux F. Breakfast consumption in french children, adolescents, and adults: a nationally representative 27. Maillot M, Sondey J, Braesco V, Darmon N. The simplified nutrient cross-sectional survey examined in the context of the international profiling system (SENS) adequately ranks foods in relation to the overall Frontiers in Nutrition 13 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 nutritional quality of diets: a validation study. Eur J Clin Nutr. (2018) 72:593– 31. Beal T, Ortenzi F. Priority micronutrient density in foods. Front Nutr. (2022) 602. doi: 10.1038/s41430-018-0104-3 9:806566. doi: 10.3389/fnut.2022.806566 28. Food and Agriculture Organization of the United Nations. World Health 32. Drewnowski A, Amanquah D, Gavin-Smith B. Perspective: how to develop Organization Guidelines on Nutrition Labelling CAC/GL 2-1985. (2017). Available nutrient profiling models intended for global use: a manual. Adv Nutr. (2021) 12:609– online at: https://www.fao.org/fao-who-codexalimentarius/sh-proxy/es/?lnk=1& 20. doi: 10.1093/advances/nmab018 url=%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards 33. Poinsot R, Vieux F, Dubois C, Perignon M, Méjean C, Darmon N. Nutritional %252FCXG%2B2-1985%252FCXG_002e.pdf (accessed September 19, 2023). quality of vegetarian and non-vegetarian dishes at school: are nutrient profiling 29. World Health Organization. Healthy Diet—Fact Sheet n. 394. (2020). Available systems sufficiently informative? Nutrients. (2020) 12:2256. doi: 10.3390/nu120 online at: http://www.who.int/mediacentre/factsheets/fs394/en/ (accessed September 82256 19, 2023). 34. Tharrey M, Houeto A, Dubois C, Vieux F, Maillot M, Darmon N. Comparison 30. Dietary Guidelines for Americans 2020-2025. Available online at: https://www. of recipes classification between 3 French nutrient profiling systems: SAIN,LIM, dietaryguidelines.gov/sites/default/files/2020-12/Dietary_Guidelines_for_Americans_ SENS and 5-C. Cahiers Nutr Dietetique. (2017) 52:269–77. doi: 10.1016/j.cnd.2017. 2020-2025.pdf (accessed September 19, 2023). 04.005 Frontiers in Nutrition 14 frontiersin.org http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Frontiers in Nutrition Unpaywall

A three-component Breakfast Quality Score (BQS) to evaluate the nutrient density of breakfast meals

Poinsot, Romane; Maillot, Matthieu; Masset, Gabriel; Drewnowski, Adam
Frontiers in Nutrition – Sep 28, 2023
Download PDF
14 pages

Loading next page...
 
/lp/unpaywall/a-three-component-breakfast-quality-score-bqs-to-evaluate-the-nutrient-YFHl00D1GR

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
2296-861X
DOI
10.3389/fnut.2023.1213065
Publisher site
See Article on Publisher Site

Abstract

TYPE Conceptual Analysis PUBLISHED 28 September 2023 DOI 10.3389/fnut.2023.1213065 A three-component Breakfast OPEN ACCESS Quality Score (BQS) to evaluate EDITED BY George Pounis, the nutrient density of breakfast Harokopio University, Greece REVIEWED BY meals Malcolm Riley, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia 1 1 2 Owen Kelly, * Romane Poinsot , Matthieu Maillot , Gabriel Masset and Sam Houston State University, United States Adam Drewnowski *CORRESPONDENCE 1 2 3 Matthieu Maillot MS-Nutrition, Marseille, France, Cereal Partners Worldwide, Lausanne, Switzerland, Center for Public [email protected] Health Nutrition, University of Washington, Seattle, WA, United States RECEIVED 27 April 2023 ACCEPTED 01 September 2023 PUBLISHED 28 September 2023 Background: Nutrient profiling methods can be applied to individual foods or to CITATION composite meals. This article introduces a new method to assess the nutrient Poinsot R, Maillot M, Masset G and density of breakfast meals. Drewnowski A (2023) A three-component Breakfast Quality Score (BQS) to evaluate the Objective: This study aimed to develop a new breakfast quality score (BQS), nutrient density of breakfast meals. based on the nutrient standards previously published by the International Breakfast Front. Nutr. 10:1213065. Research Initiative (IBRI) consortium. doi: 10.3389/fnut.2023.1213065 Methods: BQS was composed of three sub-scores derived from the weighted COPYRIGHT © 2023 Poinsot, Maillot, Masset and arithmetic mean of corresponding nutrient adequacy: an eLIMf sub-score (energy, Drewnowski. This is an open-access article saturated fat, free sugars, and sodium), a PF (protein and fiber) sub-score, distributed under the terms of the Creative and a VMn micronutrient sub-score, where n varied from 0 to 14. The Commons Attribution License (CC BY). The use, 1−14 distribution or reproduction in other forums is eects of assigning dierent weights to the eLIMf, PF, and VMn were explored permitted, provided the original author(s) and in four alternative models. The micronutrients were calcium, iron, potassium, the copyright owner(s) are credited and that magnesium, zinc, vitamin A, thiamin, riboflavin, niacin, vitamin B5, vitamin B6, the original publication in this journal is cited, in accordance with accepted academic practice. vitamin B12, vitamin C, and vitamin D. Micronutrient permutations were used to No use, distribution or reproduction is develop alternate VMn sub-scores. The breakfast database used in this study 1−14 permitted which does not comply with these came from all breakfasts declared as consumed by adults (>18 years old) in the terms. French dietary survey INCA3. All models were tested with respect to the Nutrient Rich Food Index (NRF9.3). BQS sensitivity was tested using three prototype French breakfasts, for which improvements were made. Results: The correlations of the models with NRF9.3 improved when the VMn >3 sub-score (n > 3) was included alongside the PF and eLIMf sub-scores. The model with (PF+VMn) and eLIMf each accounting for 50% of the total score showed the highest correlations with NRF9.3 and was the preferred final score (i.e., BQS). BQS was sensitive to the changing quality of three prototype breakfasts defined as tartine, sandwich, and cereal. Conclusion: The proposed BQS was shown to valuably rank the nutritional density of breakfast meals against a set of nutrient recommendations. It includes nutrients to limit along with protein, fiber, and a variable number of micronutrients to encourage. The flexible VMn sub-score allows for the evaluation of breakfast quality even when nutrient composition data are limited. KEYWORDS breakfast, nutrients, score, nutritional recommendations, meal, nutrient density, nutrient profiling Frontiers in Nutrition 01 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Using the French National Dietary Survey INCA3, alternative 1. Introduction scores were tested with respect to the NRF9.3 nutrient density score. The nutrient density of the breakfast meal was then The nutritional value of the breakfast meal can be assessed in compared across the tertiles of BQS. BQS sensitivity to changes a number of ways (1, 2). In their simplest form, scores of breakfast in breakfast composition was tested using three French breakfasts, quality award points to food groups or nutrients to encourage and identified as tartine, sandwich, and cereal. subtract points for food groups and nutrients to limit. A 10-point breakfast quality index (BQI) published in 2012 (2) awarded points for breakfast energy in the range of 20–25% of total daily intake and for the presence of cereals, fruits/vegetables, dairy products, 2. Methods monounsaturated fats, and calcium. An additional point was given for having cereals, fruit/vegetables, and dairy in the same meal (2). 2.1. INCA 3 and CIQUAL databases The same BQI subtracted points for the breakfast content of sugars and saturated and trans fats (2). Dietary data on breakfast consumption in France came from Nutrient profiling (NP) methods generally use both positive the INCA3 study on dietary intakes of the French population (24). and negative components to assess the nutrient density of meals (3), INCA3 is based on three non-consecutive 24 h dietary recalls for dishes (4), or individual foods (5). The Nutri-Score (6), the Health 1,907 men and 2,207 women (2 weekdays and 1 weekend day). Only Star Rating (7), and the Nutrient Rich Food Index (NRF9.3) (7) data for adults aged ≥18 years (n = 2,121) were analyzed. As part each have positive and negative sub-scores. Negative sub-scores are of the dietary intake assessment, participants needed to name the typically driven by saturated and trans fats, total or added sugars, eating occasion. There were 10 possibilities (including “breakfast”). and sodium, but can also include energy (6). While the choice Only foods consumed during the declared breakfast meal were of nutrients to encourage can vary, the Nutri-Score, Health Star analyzed. Breakfasts consisting of coffee or tea only (sweetened or Rating, and NRF9.3 each include both protein and fiber (6–8). The not) were excluded. A total of 4,478 breakfasts were analyzed. Nutri-Score does not include any vitamins or minerals, preferring The associated ANSES CIQUAL 2016 nutrient composition to award points based on the foods’ content of pre-selected food database (25) provides energy and nutrient values per 100 g edible groups (fruit, vegetables, and nuts) (6). By contrast, the NRF9.3 portion for all the foods consumed by INCA3 participants. Free score complements protein and fiber with calcium, iron, potassium, sugars were estimated from added sugars (26). Portion sizes were magnesium, vitamin A, vitamin C, and vitamin E (now replaced based on a previous study in France (27). by vitamin D) (7). Depending on the NP model, the number n of vitamins and minerals has varied from 3 to as many as 23 (9, 10). The International Breakfast Research Initiative (IBRI) has proposed a set of nutrient standards to assess the nutritional value 2.2. Development of the breakfast quality of breakfast meals (11, 12). In a series of IBRI studies, the breakfast score quality of representative populations in the United States (13), Canada (14), France (15), Spain (16), the United Kingdom (17), and 2.2.1. Establishing nutrient standards for BQS Denmark (18) was measured using the Nutrient Rich Food Index Table 1 summarizes the nutrient standards developed by IBRI (7). Other studies have since explored the nutritional contributions (11, 12). Also shown are the formulas for calculating nutrient of the breakfast meal in Latin America (19, 20) and the Philippines adequacy. Upper and lower bounds for energy were set for (21). The IBRI targets for vitamins and minerals at breakfast were the breakfast meal following IBRI recommendations. Maximum based on actual consumption levels during breakfast and on overall recommended values (MaxRV) were set for saturated fats, free nutrient adequacy (22). sugars, and sodium. Minimum recommended values (MinRV) were The present goal was to simplify the large set of nutrient-by- set for protein, fiber, and 14 micronutrients. nutrient IBRI recommendations into an overall breakfast quality IBRI-derived MinRV and MaxRV values for adults were based score (BQS). Alternative scores were all comprised of the same on 10 or 20% of daily recommendations from WHO/Codex (28) three components. The negative sub-score was eLIMf (energy, and from the WHO guidelines for nutrients of public health saturated fat, free sugars, and sodium). The two positive sub-scores concern (free sugars, saturated fats, and sodium) (29). For breakfast were PF (protein and fiber) and VMn, the latter composed of a energy, a maximum adequacy score of 100% was achieved when variable number of vitamins and minerals. As in other NP models, breakfast energy intakes fell within the 300–500 kcal range. The the final quality score was based on the difference between the lower bound (0%) was set at 0 kcal. The upper bound (also 0%) negative and positive sub-scores (6–8). The present innovation was was set at 800 kcal, i.e., 500 kcal + 300 kcal. When breakfast energy to vary the number of vitamins and minerals in VMn from 0 to 14 was lower than 300 kcal or higher than 500, the adequacy scores and to explore the impact of micronutrient permutations (9, 10). diminished in proportion to intakes relative to the recommended We also explored the effects of assigning differential weights to the values. When the breakfast energy was higher than the upper bound eLIMf, PF, and VMn sub-scores on the total score. This was done (800 kcal), the adequacy score became negative. because the weighting of the sub-scores affects the final evaluation For saturated fat, sodium, and free sugars, a maximum (23). Although some NP models are driven by negative elements, adequacy score (100%) was achieved when breakfast intakes were energy, sugar, and fat, others are weighted to favor protein, fiber, <MaxRV. The upper bound (0%) was set at 2 × MaxRV. Adequacy vitamins, and minerals. diminished in proportion from MaxRV to the upper bound (2 × Frontiers in Nutrition 02 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 1 Calculation of nutrient adequacy for breakfast quality score. Sub- Nutrient Unit Reference values Boundaries Percentage of score adequacy ∗ ∗ MinRV MaxRV Lower Upper bound bound obsi eLIMf Energy kcal 300 500 0 500 + 300 = 800 Adeq = min( × i min reco upper −obs i i 100, max ×100, 100) upper −reco SFAs %EBI 10 2 × 10 = 20 Free sugars %EBI 10 2 × 10 = 20 Sodium mg 400 2 × 400 = 800 PF Proteins g 10 0 Adeq = upper −obs i i min( × max upper −reco 100, 100) Fibers g 5 0 VMn Calcium mg 250 0 Adeq = obs min( × 100, 100) min reco Iron mg 2.8 0 Magnesium mg 62 0 Potassium mg 700 0 Zinc mg 2.2 0 Vitamin A mg 80 0 Vitamin B1 mg 0.24 0 Vitamin B2 mg 0.36 0 Vitamin B3 mg 3.75 0 Vitamin B6 mg 0.26 0 Folate μg 80 0 Vitamin B12 μg 0.48 0 Vitamin C mg 20 0 Vitamin D μg 1 0 max min obs , observed breakfast intake of nutrient i; reco , minimum recommended value for nutrient i; upper , upper bound for nutrient i; reco , maximum recommended value for nutrient i i i i i; EBI, energy breakfast intake; n, number of micronutrients. Free sugars refer to all monosaccharides and disaccharides added to foods by the manufacturer, cook, or consumer, and sugars naturally present in honey, syrups, and fruit juice. The MinRV and MaxRV values came from IBRI recommendations for adults and were based on 10 or 20% of daily recommendations from WHO/Codex. MaxRV). When the observed values at breakfast were higher than lower the content of nutrients to be limited, the higher the the upper bounds, the adequacy scores became negative. eLIMf sub-score. For protein, fiber, vitamins, and minerals, a maximum 1 i=4 adequacy score (100%) was achieved when breakfast nutrient eLIMf = × Adeq , (1) 4 i=1 intakes were >MinRV. The lower bound (0%) was set at zero i = energy, saturated fats, free sugars, sodium , consumption. Nutrient adequacy between these two points was eLIMf ∈ [−∞; 100] calculated by dividing the breakfast value by the recommended value and multiplying by 100. The second sub-score was named PF (Equation 2) since it contained protein and fiber adequacy. 1 i=2 PF = × Adeq , (2) 2.2.2. Definitions of BQS sub-scores 2 i=1 Nutrient adequacies were grouped into one of three sub-scores: i = proteins, fibers , PF ∈ [0; 100], eLIMf, PF, or VMn (Table 1). The sub-scores were equal to the arithmetic mean of corresponding nutrient adequacies (Equations In previous studies, NP profiles have used a wide range of 1–3). The eLIMf sub-score (Equation 1) included energy, saturated micronutrients (9, 10). The third sub-score VMn (Equation 3) was fat, sodium, and free sugars (rather than total or added) adequacies. based on a variable number n of vitamins and minerals (n ranges The eLIMf sub-score could be either negative or positive. The from 0 to 14) adequacies. The 14 micronutrients for which IBRI Frontiers in Nutrition 03 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 recommendations were available were calcium, iron, potassium, 2.3. Testing alternative BQS models magnesium, zinc, vitamin A, thiamin, riboflavin, niacin, vitamin B5, vitamin B6, vitamin B12, vitamin C, and vitamin D. The VMn Scores obtained using alternative BQS models on the INCA3 sub-score could only be positive. breakfast meals were compared to scores for the same breakfast meals generated by the NRF9.3 nutrient profiling model (7) and P to scores obtained with the same BQS model where n = 14 (called 1 i=n VMn = Adeq , (3) n i=1 “complete” BQS). The NRF9.3 is the sum of percent daily values i ∈ {0, micronutrient , . . . micronutrient } , 1 n (DV) for nine nutrients to encourage (proteins, fibers, calcium, iron, magnesium, potassium, vitamin C, vitamin A, and vitamin n ≤ 14, VMn ∈ [0; 100] D) minus the sum of percent DV for three nutrients to limit (saturated fats, added sugars, and sodium). All values are calculated All possible combinations of the 14 micronutrients were per 100 kcal and capped at 100% (7). Correlations were estimated systematically tested. The number of possible combinations of using the Spearman rank-order correlation coefficient. For partial micronutrients depended on their number. For VM = 0 and for VMn scores where n < 14, correlations were obtained for each VM = 14, there was only one possible option. For VM = 1 and for combination of micronutrients, and their distribution was analyzed VM = 13, there were 14 possible choices. For VM = 2 and for VM using boxplots. = 12, there were 91 possible combinations of micronutrients. For The model that correlated best with NRF9.3 was selected and VM = 3 and VM = 11, there were 364 and so on. For VM = 7, there called “BQS” in the rest of the article. were 3,432 possible combinations. BQS values for 4,478 breakfasts in the INCA3 database were split into tertiles. The general linear model then compared values of energy and nutrients (for which IBRI recommendations were available), sub-scores eLIMf, PF, and VM, and grams of eight 2.2.3. Alternative BQS models with dierent dietary components across the BQS tertiles. The eight dietary sub-score weights components of interest were fruits and vegetables, whole-grain, Alternative BQS models were calculated as a weighted refined-grain, milk and dairy, plant fats, animal fats, sugary foods, mean of the three sub-scores following Equation 4. In and sweet-tasting beverages (e.g., soda and fruit juices). A post-hoc every case, the minimum score was 0, so there were no comparison (Tukey’s HSD test) was performed when the difference negative scores. Scores ranged from 0 to 100, with the was significant. highest scores given to those breakfasts that met all of the Statistical analyses used the R software version 4.1. The level of IBRI recommendations. significance was set to 5% for all tests. Breakfast quality score = α × eLIMf + β × PF + γ × VMn 3. Results (4) 3.1. Alternative BQS models with complete VMn sub-scores Where α, β, and γ are weights ranging from zero to 1, and their sum is equal to 1. Breakfast meals in the INCA 3 database were evaluated using Four alternative BQS models were tested, each with a different the four alternative BQS models. All four BQS models were highly sub-score weighting scheme (Table 2). correlated with each other (range r = 0.76 to r = 0.99; results not In the balanced model, the eLIMf sub-score accounts for 50% shown). With all the alternative BQS models, none of the 4,478 of the BQS, and the sum of PF and VMn sub-scores (PF+VMn), breakfasts got a 100% adequacy score. Mean BQS values ranged where n ≤ 14, also accounts for 50% of the BQS. The weights from 51.6% (balanced model) to 58.5% (unweighted model) and α, β, and γ differ between models and are shown in Table 2. In the were comparable for the four models. unweighted model, all 6+n elements are equivalent. Each element The balanced model showed the highest correlations with accounts for [1/(6+n)] × 100%] of the total score as VMn rises NRF9.3 nutrient density scores (r = 0.55) and the lowest from 0 to 14. correlations with energy density (r = −0.15) of breakfasts in the In the micronutrient model, the VMn sub-score (n ≤ 14) INCA database (Figure 2). The balanced model had a moderate now accounts for 50% of total BQS. The eLIMf and PF sub- correlation with LIM. scores together account for 50% of the BQS total. In the three-way model, the eLIMf sub-score, the PF sub-score, and the VMn sub-score (n ≤ 14) each account for 1:3 of the 3.2. Alternative BQS models with partial BQS total. Pie charts are an alternative way of visualizing BQS weights VMn sub-scores when n = 14, and they are presented in the last column of Table 2. Figure 1 shows the shift in weights for the four Figure 2 shows that correlations with NRF9.3 improved as alternative weighting schemes for BQS sub-scores as the number the VMn sub-score incorporated more vitamins and minerals. of vitamins and minerals in the VMn sub-score rises from 0 For all four alternative BQS models, correlations with NRF9.3 to 14. were weakest when no micronutrients were included (VM ). n =0 Frontiers in Nutrition 04 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 2 Weighting scheme of component sub-scores for the four alternative BQS models (with n, the number of micronutrients). Alternative models Weight of sub-score Weight of each sub-score, where n = 14 eLIMf (α) PM (β) VMn (γ ) ( ) 1 ( ) 2+n 2+n Balanced model 4 2 n Unweighted model 6+n 6+n 6+n 1 4 1 1 2 1 1 Micronutrient model × = × = 2 6 3 2 6 3 2 1 1 1 Three-way model 3 3 3 Correlations improved when the VMn component included at 3.3. Testing the performance of BQS (i.e., least three vitamins or minerals. For balanced, micronutrient, the balanced model) and three-way models, correlations with NRF9.3 increased when the VMn number of micronutrients increased. For the 3.3.1. Distribution of BQS in INCA3 adult unweighted model, the correlation with NRF9.3 increased up to 5– breakfasts 6 micronutrients and then decreased. The average correlations of The distribution of the BQS values in INCA3 adult breakfasts is NRF9.3 with the balanced model were very close to those with the shown in Figure 4. Only 3% of the breakfasts were cut at 0%. The micronutrient model; however, according to the permutation, the score of 0% means that the breakfast provided more negative points correlations with the micronutrient model were less homogeneous (from the eLIMf sub-score) than positive points from PF and VM. than those with the balanced model (see the size of boxes The BQS distribution exhibits a normal shape. The average (51.6%) in Figure 2). and median (53.1%) balanced BQS were close to 50%. All average correlations were highest for the balanced model. For the balanced model, when VM , correlation values ranged n =3 from 0.43 to almost 0.60. This variability was related not only to 3.3.2. Nutritional and dietary components by the number but also to specific combinations of micronutrients. tertiles of BQS According to Figure 3, the correlation between complete and Percent nutrient adequacy in the first tertile of BQS ranged partially balanced BQS stayed high (minimum 0.93 without from 0 to 43.7% and in the third tertile from 62.1 to 98.4%. Table 3 micronutrients) even when the number of micronutrients included confirmed that sub-score eLIMf increased as well as sub-scores in the score decreased. However, it was not clear which vitamins PF and VM when BQS values increased. Indeed, the amounts of and minerals were the most important. Based on these results, nutrients to limit decreased between low and medium and between the balanced model was selected as BQS and was subject to medium and high tertiles of BQS, and, except for vitamin A, the further testing. amounts of nutrients to encourage increased between low and high Frontiers in Nutrition 05 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 1 Weight (in %) of each sub-score in the four alternative BQS models, where the number of micronutrients ranged from 0 to 14. tertiles of BQS. For proteins, fiber, iron, magnesium, potassium, by our group and available at https://ms-nutrition.com/web-app/ and vitamins B1, B3, B9, and C, the difference between the low breakfast-calculator/. The version used for the article contained the and medium tertiles was not significant, but the difference between nutritional composition of foods that corresponded to the dietary the medium and high tertiles was significant. Likewise, amounts intake data (INCA3) and has since been updated to reflect the most of healthier food groups such as fruits and vegetables, whole- up-to-date food composition database (i.e., CIQUAL 2020). None grain foods, milk and dairy products, and plant-based fats were of the eLIMf sub-scores were negative, so BQS corresponded to the higher. Conversely, amounts of less healthy food groups such as stacking of the sub-scores in Figure 5. refined-grained foods, animal fats, sugary foods, and sweet-tasting The balanced BQS values improved with increasing versions beverages were lower in breakfast in low tertiles than in breakfast except between v2 and v3 in “sandwich” breakfast, where they in medium or high tertiles. Breakfasts in the highest tertile of slightly decreased. For the “tartine” and “sandwich” breakfasts, BQS were thus of higher nutritional quality than breakfasts in the BQS increased from 46.2 and 55.3% (v0) to 57.6 and 60.8% medium or low tertile of BQS. (v1), respectively, because it enabled the fiber recommendation to be met (Supplementary Figures 1, 2). For the “tartine” breakfast, changes between v1 and v2 (64.7%) and v2 and 3.3.3. Sensitivity through three examples of v3 (73.8%) aimed at increasing the eLIMf sub-score, and changes between v3 and v4 (84%) and v4 and v5 (88.6%) breakfasts aimed at increasing the VM sub-score. Adding an orange and The sensitivity of the BQS was tested using three alternative plain yogurt in v4 and v5 increased calcium and vitamins breakfasts. Breakfast 1 (“tartine”) was a baguette with jam; C, B1, B2, and B12 while staying within the recommended Breakfast 2 (“cereal”) was ready-to-eat (RTE) cereal and milk; energy range. and Breakfast 3 (“sandwich”) was a savory sandwich. Four For the “sandwich” breakfast, the macronutrient profile was to five versions of each breakfast (v0, v1, v2, v3, v4, and more favorable thanks to the replacement of butter with low- v5) were constructed (Table 4). Breakfast nutrient content was fat butter from v1 (60.8%) to v2 (70.6%), even though the mean calculated using the “Breakfast Calculator” online tool, developed Frontiers in Nutrition 06 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 2 Spearman correlations between each alternative BQS model and NRF9.3 for all combinations of VMn ranging from 0 to 14 micronutrients. The horizontal reference line is set to 0.55. Frontiers in Nutrition 07 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 3 Distribution of Spearman correlation between the complete balanced model (based on 14 micronutrients) and the balanced model when VM ranged from 0 to 14 micronutrients. FIGURE 4 Distribution of BQS in INCA3 adults breakfast. BQS, BQS balanced model. micronutrient adequacy decreased. The addition of milk to the cereals with low-sugar cereals, which reduced free sugars coffee (i.e., white coffee instead of black coffee) in v3 decreased the to below the threshold level. Adding fruits, almonds, and total score because the increases in mineral and vitamin contents sunflower seeds in v2, v3, and v4 improved fiber, potassium, brought by the milk, such as calcium and vitamin B2, were not and calcium adequacy (Supplementary Figure 3). The cereal sufficient to compensate for the increase in energy and SFA. On breakfast v4 was the only one to restrict eLIMf (nutrients the other hand, adding oranges increases the score even if it adds to limit) and therefore scored the highest overall (93.2%) energy to the breakfast. among the 16 breakfasts (Figure 5). The last versions of For the “cereal” breakfast, the balanced BQS increased “sandwich” and “tartine” breakfasts were still too high in sodium from 69.4 (v0) to 80.3% (v1) because v1 replaced chocolate (Supplementary Figures 1, 2). Frontiers in Nutrition 08 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 TABLE 3 Average BQS , sub-scores, energy, and nutrients included in IBRI recommendations and dietary components by tertiles of BQS. Tertile of BQS Low [0; 43.7] Medium [43.7; 62.1] High [62.1; 98.4] p-value $ ∗∗∗ BQS 27.6 52.9 74.3 ∗∗∗ Sub-score eLIMf −0.18 23.1 35.8 a b b ∗∗∗ Energy (kcal) 482 410 411 ∗∗∗ SFA (% EBI) 14.6 13.6 11.5 ∗∗∗ Free sugars (% EBI) 28.4 16.6 10.6 ∗∗∗ Sodium 562 485 438 ∗∗∗ Sub-score PF 3.61 4.11 4.98 a a b ∗∗∗ Proteins (g) 9.90 10.0 14.3 a a b ∗∗∗ Fibers (g) 3.47 3.35 4.19 ∗∗∗ Sub-score VM 23.0 25.7 33.5 ∗∗∗ Calcium (mg) 130 163 324 a a b ∗∗∗ Iron (mg) 1.60 1.65 2.15 a a b ∗∗∗ Magnesium (mg) 70.2 73.6 92.7 a a b ∗∗∗ Potassium (mg) 597 630 829 ∗∗∗ Zinc (mg) 0.95 1.04 1.65 Vitamin A (μg RAE) 133 92.7 98.2 0.174 a a b ∗∗∗ Vitamin B1 (mg) 0.25 0.26 0.38 ∗∗∗ Vitamin B2 (mg) 0.34 0.41 0.69 a a b ∗∗∗ Vitamin B3 (mg) 2.88 3.11 3.68 ∗∗∗ Vitamin B6 (mg) 0.21 0.23 0.36 a a b ∗∗∗ Vitamin B9 (mg) 56.7 58.8 77.6 ∗∗∗ Vitamin B12 (μg) 0.41 0.53 1.09 a a b ∗∗∗ Vitamin C (mg) 18.8 18.2 26.4 ∗∗∗ Vitamin D (μg) 0.41 0.55 0.97 Dietary components ∗∗∗ Fruits and vegetables (g) 8.38 14.2 34.7 ∗∗∗ Whole-grain foods (g) 6.70 9.45 14.3 ∗∗∗ Refined-grain food (g) 50.2 38.1 25.2 ∗∗∗ Milk and dairy products (g) 47.0 71.9 196 a b b ∗∗∗ Plant fats (g) 1.15 2.33 2.80 ∗∗∗ Animal fats (g) 7.25 4.92 2.27 a b b ∗∗∗ Sugary foods (g) 57.3 41.8 40.0 ∗∗∗ Sweet-tasting beverages (g) 55.3 46.2 37.7 $ ∗∗∗ BQS, BQS balanced model. p < 0.001. Same index letters (e.g., a and a) indicate that there is no significant difference between the two tertiles, and different index letters (e.g., a and b) indicate that the difference is statistically significant between the two tertiles. No indexes indicate that the difference is significant between the three tertiles (Tukey’s range test). 4. Discussion BQS that was selected from among four alternatives showed the highest correlations with the NRF9.3 index and low correlations Nutrient profiling methods, initially developed for individual with energy density. The present BQS score is composed of three foods, can also be used to assess the nutritional value of meals. This distinct sub-scores. Each sub-score had been used in previous article introduces a new way to assess breakfast quality that was nutrient profiling models but in different ways. The PF (protein specifically designed to follow a set of published recommendations and fiber) component has been used by nutri-score, HSR, and the for the breakfast meal from the IBRI consortium (11, 12). The NRF9.3 nutrient density index (6–8). The eLIMf sub-score was Frontiers in Nutrition 09 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Frontiers in Nutrition 10 frontiersin.org TABLE 4 Breakfast examples. Breakfast type Food 1 Food 2 Food 3 Food 4 Food 5 Food 6 Food 7 Food 8 Tartine (v0) French baguette (80 g) Butter, unsalted (16 g) Strawberry jam (30 g) Black coffee (200 ml) Tartine (v1) Whole grain baguette Butter, unsalted (16 g) Strawberry jam (30 g) Black coffee (200 ml) (80 g) Tartine (v2) Whole grain baguette Dairy spread, 39–41% Strawberry jam (30 g) Black coffee (200 ml) (80 g) fat (16 g) Tartine (v3) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) (80 g) fat (16 g) Tartine (v4) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) Average fruit (100 g) (80 g) fat (16 g) Tartine (v5) Whole grain baguette Dairy spread, 39–41% Strawberry jam (15 g) Black coffee (200 ml) Average fruit (100 g) Plain yogurt (125 g) (80 g) fat (16 g) Sandwich (v0) French baguette Roasted chicken (60 g) Butter, unsalted (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v1) Whole grain baguette Roasted chicken (60 g) Butter, unsalted (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v2) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) Black coffee (200 ml) (100 g) Sandwich (v3) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) White coffee (200 ml) (100 g) Sandwich (v4) Whole grain baguette Roasted chicken (60 g) Low-fat butter (16 g) Mayonnaise (20 g) Tomato (20 g) Lettuce (5 g) White coffee (200 ml) Orange (80 g) (100 g) Cereal (v0) Cereal, chocolate, Semi-skimmed milk Tea, no sugar (200 ml) enriched (45 g) (150 ml) Cereal (v1) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) low-sugars, fortified (150 ml) (45 g) Cereal (v2) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) low-sugars, fortified (150 ml) (45 g) Cereal (v3) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) Sunflower seed (7 g) low-sugars, fortified (150 ml) (45 g) Cereal (v4) Cereal, whole-wheat, Semi-skimmed milk Tea, no sugar (200 ml) Strawberry (50 g) Sunflower seed (7 g) Almond (10 g) low-sugars, enriched (150 ml) (45 g) Bold text indicates that the food or the amount of food is new compared to the previous version of the BF. Poinsot et al. 10.3389/fnut.2023.1213065 FIGURE 5 BQS and its sub-scores PF, eLIMf, and VMn for three types of breakfast: “tartine” (A), “sandwich” (B), and “cereal” (C) and their improved versions. BQS, BQS balanced model. close to the negative LIM sub-score (saturated fat, added sugar, and The number of vitamins and minerals was based on the IBRI sodium) used in NRF9.3, but with the addition of energy, a feature recommendations. The performance of VMn scores was tested shared with nutri-score. The nutri-score version of eLIMf includes when the number of vitamins and minerals was allowed to vary energy and total sugars. from 0 to 14. This was done to determine the minimum number of The novelty here was to create a variable VMn sub-score where vitamins and minerals that are necessary to assess breakfast quality. n varied from 0 to 14 and different permutations of nutrients It was found that less than three vitamins and minerals should be were deployed. The present approach differs from that of nutri- critical to derive a robust BQS. The best choice is to conserve the 14 score, which does not include vitamins or minerals but awards micronutrients. However, not all food databases, especially those points for the content of fruits, vegetables, legumes, nuts, and seeds. currently available for low- and middle-income countries, have a Frontiers in Nutrition 11 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 full set of nutrients, so our results demonstrated that using <14 explored in contexts where limited nutrient content information micronutrients could be a viable alternative. The choice of these is available. Finally, the balanced BQS was sensitive to small nutrients should be guided by relevant public health considerations improvements in breakfast quality, suggesting that it may serve as and data availability. Not all nutrients are necessarily consumed in a basis to educate the population on breakfast quality. adequate amounts in the course of a single meal. When it comes There are some limitations. First, validity testing is difficult to dietary inadequacies, the Dietary Guidelines for Americans for meal-specific indicators. The correlations were performed identified calcium, potassium, and vitamin D as shortfall nutrients with another model of nutrient density, the NRF9.3 score. With for the United States (30). Priority nutrients for low- and middle- single meals (as opposed to the total diet), there are no potential income countries were calcium, iron, zinc, folate, iodine, vitamin health outcomes. Finally, the BQS model was applied to an A, and vitamin B12 (31). Of course, the construction of NP models adult population in a single country, France. However, the BQS depends on the availability of nutrient composition data. For that is easily applicable to other breakfast meals based on national reason, it is advantageous to have flexibility in the number of score dietary surveys. elements (32). In the BQS construction, the calculation of the eLIMf sub- 5. Conclusion score followed a different concept compared to other nutrient profiles, which took into account unfavorable nutrients such as The present study introduces a new breakfast quality score saturated fats, sodium, and free sugars. In BQS, the eLIMf sub- (BQS), designed to assess the nutrient adequacy of a single meal– score values ranged from zero whether nutritional content is twice breakfast. Similar in structure to other compensatory nutrient the maximal recommended amount to 100 whether the nutrient profiling models, the BQS introduces a novel flexible VMn sub- contents are below the limits. To penalize breakfast with a high score based on a variable number of vitamins and minerals. The amount of saturated fats, free sugars, or sodium, the sub-score flexibility of the BQS makes it an attractive tool for evaluations becomes linearly negative when nutritional content exceeds twice of breakfast quality in settings where comprehensive nutrient the recommendation. Thanks to this approach, the BQS is able to composition data are not available. discriminate between two different breakfasts with a high amount of unfavorable nutrients. One challenge of nutrient profiling is to adequately weight Ethics statement the respective contributions of positive and negative components. Some existing systems appear to be mainly driven by energy density The studies involving humans were approved by Comité and nutrients (33, 34). In this study, we tested four alternative consultatif sur le traitement de l’information en matière de weighting models for the BQS. The selected algorithm, which gave recherche dans le domaine de la santé. The studies were equal weight to the positive (protein, fibers, and micronutrients) conducted in accordance with the local legislation and institutional and negative (eLIMf) components of the BQS, showed a low requirements. Written informed consent for participation was correlation of the BQS with LIM, or energy density, meaning that not required from the participants or the participants’ legal the selected BQS would be sensitive to changes in both positive and guardians/next of kin in accordance with the national legislation negative components. and institutional requirements. The sensitivity of the BQS score to small changes in the mean composition of breakfasts was illustrated with reference to three Author contributions types of breakfasts. Based on BQS score distributions in French adults, 40% (close to the first tertile, which was 43.7%) appeared GM, RP, and MM conceptualized the study. RP and to be an appropriate cut-off point to identify breakfast that could MM conducted statistical analyses and wrote the first draft be considered nutritionally adequate. Given that the present results manuscript. AD wrote the final manuscript. GM revised the were based on INCA3 data in adults, further work is needed to manuscript. All authors contributed to the article and approved the assess breakfast quality among children and teenagers. submitted version. This study had both strengths and limitations. First, the selected BQS was based on nutrient recommendations that were breakfast- specific as opposed to daily. Second, the balanced BQS was Funding robust, showing good performance even with a limited number of micronutrients. That will be of importance in places where Cereal Partners Worldwide (CPW) funded the study. AD comprehensive nutrient composition data may not be available. In received funding from CPW to conduct the study and write some countries, nutrient composition data are partial, and some the manuscript. MS-Nutrition was financially supported to nutrients are missing altogether. In those cases, it is useful to have conceptualize and conduct the analysis. a flexible and pretested BQS that can be based on the nutrients that are available. This would allow for consistent and harmonized Acknowledgments testing of breakfast quality across multiple locations, including low- and middle-income countries. However, we did not analyze We thank Sinead Hopkins and Diane Zimmermann for their the performance of BQS, considering particular combinations of valuable comments, which helped improve the manuscript. nutrients. The performance of the proposed score needs to be Frontiers in Nutrition 12 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 Conflict of interest evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the GM was employed by CPW at the time of the study. RP and publisher. MM are employees of MS-Nutrition. The remaining author declares that the research was Supplementary material conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict The Supplementary Material for this article can be found of interest. online at: https://www.frontiersin.org/articles/10.3389/fnut.2023. 1213065/full#supplementary-material Publisher’s note SUPPLEMENTARY FIGURE 1 Nutrient adequacies of “Tartine” breakfasts: version 0, 1, 2, 3, 4, and 5. All claims expressed in this article are solely those SUPPLEMENTARY FIGURE 2 Nutrient adequacies of “Sandwich” breakfasts: version 0, 1, 2, 3, and 4. of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, SUPPLEMENTARY FIGURE 3 Nutrient adequacies of “Cereal” breakfasts: version 0, 1, 2, 3, and 4. the editors and the reviewers. Any product that may be References 1. Matthys C, De Henauw S, Bellemans M, De Maeyer M, De Backer G. Breakfast breakfast research initiative. Nutrients. (2018) 10:1056. doi: 10.3390/nu100 habits affect overall nutrient profiles in adolescents. Public Health Nutr. (2007) 10:413– 81056 21. doi: 10.1017/S1368980007248049 16. Ruiz E, Ávila JM, Valero T, Rodriguez P, Varela-Moreiras G. Breakfast 2. Monteagudo C, Palacin-Arce A, del Mar Bibiloni M, Pons A, Tur JA, Olea-Serrano consumption in Spain: patterns, nutrient intake and quality. Findings from the ANIBES F, et al. Proposal for a Breakfast Quality Index (BQI) for children and adolescents. study, a study from the international breakfast research initiative. Nutrients. (2018) Public Health Nutr. (2013) 16:639–44. doi: 10.1017/S1368980012003175 10:1324. doi: 10.3390/nu10091324 3. Fern EB, Watzke H, Barclay DV, Roulin A, Drewnowski A. The nutrient balance 17. Gaal S, Kerr MA, Ward M, McNulty H, Livingstone MBE. Breakfast concept: a new quality metric for composite meals and diets. PLoS ONE. (2015) Consumption in the UK: Patterns, nutrient intake and diet quality. a study 10:e0130491. doi: 10.1371/journal.pone.0130491 from the international breakfast research initiative group. Nutrients. (2018) 10:999. doi: 10.3390/nu10080999 4. Furuta C, Jinzu H, Cao L, Drewnowski A, Okabe Y. Nutrient profiling of japanese dishes: the development of a novel ajinomoto group nutrient profiling system. Front 18. Fagt S, Matthiessen J, Thyregod C, Kørup K, Biltoft-Jensen A. Breakfast Nutr. (2022) 9:912148. doi: 10.3389/fnut.2022.912148 in Denmark: Prevalence of consumption, intake of foods, nutrients and dietary quality. A study from the international breakfast research initiative. Nutrients. (2018) 5. Drewnowski A, Fulgoni V. Nutrient profiling of foods: creating a nutrient-rich 10:1085. doi: 10.3390/nu10081085 food index. Nutr Rev. (2008) 66:23–39. doi: 10.1111/j.1753-4887.2007.00003.x 19. Fisberg M, Kovalskys I, Previdelli AN, Pereira JL, Zimberg IZ, Fisberg R, et al. 6. Hercberg S, Touvier M, Salas-Salvado J. Group of European scientists supporting Breakfast Consumption Habit and Its Nutritional Contribution in Latin America: the implementation of Nutri-Score in Europe: The Nutri-Score nutrition label. Int J Results from the ELANS Study. Nutrients. (2020) 12:2397. doi: 10.3390/nu120 Vitam Nutr Res. (2022) 92:147–57. doi: 10.1024/0300-9831/a000722 7. Fulgoni III VL, Keast DR, Drewnowski A. Development and validation of the 20. Kovalskys I, Fisberg M, Previdelli AN, Pereira JL, Zimberg IZ, Guajardo V, nutrient-rich foods index: a tool to measure nutritional quality of foods. J Nutr. (2009) et al. Breakfast in Latin America: evaluation of nutrient and food group intake 139:1549–54. doi: 10.3945/jn.108.101360 toward a nutrient-based recommendation. J Acad Nutr Diet. (2022) 122:1099– 8. Health Star Rating System. Available online at: http://www.healthstarrating.gov. 1113.e3. doi: 10.1016/j.jand.2021.11.012 au/internet/healthstarrating/publishing.nsf/content/home (accessed September 19, 21. Angeles-Agdeppa I, Custodio MRS, Toledo MB. Breakfast in the 2023). Philippines: food and diet quality as analyzed from the 2018 expanded 9. Drewnowski A, Maillot M, Darmon N. Testing nutrient profile models in national nutrition survey. Nutr J. (2022) 21:52. doi: 10.1186/s12937-022- relation to energy density and energy cost. Eur J Clin Nutr. (2009) 63:674– 00804-x 83. doi: 10.1038/ejcn.2008.16 22. Vieux F, Maillot M, D Rehm C, Drewnowski A. Designing optimal 10. Drewnowski A, Maillot M, Darmon N. Should nutrient profiles breakfast for the United States using linear programming and the be based on 100 g, 100 kcal or serving size? Eur J Clin Nutr. (2009) NHANES 2011-2014 database: a study from the international breakfast 63:898–904. doi: 10.1038/ejcn.2008.53 research initiative (IBRI). Nutrients. (2019) 11:1374. doi: 10.3390/nu110 11. Gibney MJ, Barr SI, Bellisle F, Drewnowski A, Fagt S, Hopkins S, et al. Towards an evidence-based recommendation for a balanced breakfast— 23. Drewnowski A, Gonzalez TD, Rehm CD. Balanced hybrid nutrient density a proposal from the international breakfast research initiative. Nutrients. (2018) score compared to nutri-score and health star rating using receiver operating 10:1540. doi: 10.3390/nu10101540 characteristic curve analyses. Front Nutr. (2022) 9:867096. doi: 10.3389/fnut.2022. 12. Gibney MJ, Barr SI, Bellisle F, Drewnowski A, Fagt S, Livingstone B, et al. Breakfast in Human Nutrition: The International Breakfast Research Initiative. 24. French Agency for Food Environmental and Occupational Health & Safety Nutrients. (2018) 10:559. doi: 10.3390/nu10050559 (ANSES). Étude Individuelle Nationale des Consommations Alimentaires 3 (INCA 3); French Agency for Food Environmental and Occupational Health & Safety: 13. Drewnowski A, Rehm CD, Vieux F. Breakfast in the United States: food and Maisons-Alfort, France. (2017). Available online at: https://www.anses.fr/fr/system/ nutrient intakes in relation to diet quality in national health and examination survey files/NUT2014SA0234Ra.pdf (accessed September 19, 2023). 2011-2014. A study from the international breakfast research initiative. Nutrients. (2018) 10:1200. doi: 10.3390/nu10091200 25. French Agency for Food, Environmental and Occupational Health & Safety. ANSES-CIQUAL French Food Composition Table. (2013). Available online at: https:// 14. Barr SI, Vatanparast H, Smith J. Breakfast in Canada: prevalence of consumption, ciqual.anses.fr/ (accessed September 19, 2023). contribution to nutrient and food group intakes, and variability across tertiles of daily diet quality. A study from the international breakfast research initiative. Nutrients. 26. Louie JCY, Moshtaghian H, Boylan S, Flood VM, Rangan AM, Barclay AW, et al. (2018) 10:985. doi: 10.3390/nu10080985 A systematic methodology to estimate added sugar content of foods. Eur J Clin Nutr. (2015) 69:154–61. doi: 10.1038/ejcn.2014.256 15. Bellisle F, Hébel P, Salmon-Legagneur A, Vieux F. Breakfast consumption in french children, adolescents, and adults: a nationally representative 27. Maillot M, Sondey J, Braesco V, Darmon N. The simplified nutrient cross-sectional survey examined in the context of the international profiling system (SENS) adequately ranks foods in relation to the overall Frontiers in Nutrition 13 frontiersin.org Poinsot et al. 10.3389/fnut.2023.1213065 nutritional quality of diets: a validation study. Eur J Clin Nutr. (2018) 72:593– 31. Beal T, Ortenzi F. Priority micronutrient density in foods. Front Nutr. (2022) 602. doi: 10.1038/s41430-018-0104-3 9:806566. doi: 10.3389/fnut.2022.806566 28. Food and Agriculture Organization of the United Nations. World Health 32. Drewnowski A, Amanquah D, Gavin-Smith B. Perspective: how to develop Organization Guidelines on Nutrition Labelling CAC/GL 2-1985. (2017). Available nutrient profiling models intended for global use: a manual. Adv Nutr. (2021) 12:609– online at: https://www.fao.org/fao-who-codexalimentarius/sh-proxy/es/?lnk=1& 20. doi: 10.1093/advances/nmab018 url=%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards 33. Poinsot R, Vieux F, Dubois C, Perignon M, Méjean C, Darmon N. Nutritional %252FCXG%2B2-1985%252FCXG_002e.pdf (accessed September 19, 2023). quality of vegetarian and non-vegetarian dishes at school: are nutrient profiling 29. World Health Organization. Healthy Diet—Fact Sheet n. 394. (2020). Available systems sufficiently informative? Nutrients. (2020) 12:2256. doi: 10.3390/nu120 online at: http://www.who.int/mediacentre/factsheets/fs394/en/ (accessed September 82256 19, 2023). 34. Tharrey M, Houeto A, Dubois C, Vieux F, Maillot M, Darmon N. Comparison 30. Dietary Guidelines for Americans 2020-2025. Available online at: https://www. of recipes classification between 3 French nutrient profiling systems: SAIN,LIM, dietaryguidelines.gov/sites/default/files/2020-12/Dietary_Guidelines_for_Americans_ SENS and 5-C. Cahiers Nutr Dietetique. (2017) 52:269–77. doi: 10.1016/j.cnd.2017. 2020-2025.pdf (accessed September 19, 2023). 04.005 Frontiers in Nutrition 14 frontiersin.org

Journal

Frontiers in Nutrition – Unpaywall

Published: Sep 28, 2023

There are no references for this article.