PreprintPDF Available

Limitations of the Food Compass Nutrient Profiling System

  • Oslo New University College
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Nutrient Profiling Systems provide algorithms which are designed to assess the healthfulness of foods based on nutrient composition, and intended as a strategy to improve diets. Many Nutrient Profiling Systems are founded on a reductionist assumption that the healthfulness of foods is determined by the sum of their nutrients, with little consideration for the extent and purpose of processing and its health implications. A novel Nutrient Profiling System called Food Compass attempted to address existing gaps and provide a more holistic assessment of the healthfulness of foods. While a conceptually impressive effort, we propose that the chosen algorithm is not well justified and produces results that fail to discriminate for common shortfall nutrients, exaggerate the risks associated with animal-source foods, and underestimate the risks associated with ultra-processed foods. We caution against the use of Food Compass in its current form to inform consumer choices, policies, programs, industry reformulations, and investment decisions.
Content may be subject to copyright.
Limitations of the Food Compass Nutrient
Profiling System
Content type: Matters Arising
Authors: Flaminia Ortenzi, Marit Kolby, Mark Lawrence, Frédéric Leroy, Stella Nordhagen,
Stuart M. Phillips, Stephan van Vliet, Ty Beal*
*Corresponding author: Ty Beal
Nutrient Profiling Systems provide algorithms which are designed to assess the
healthfulness of foods based on nutrient composition, and intended as a strategy to improve
diets. Many Nutrient Profiling Systems are founded on a reductionist assumption that the
healthfulness of foods is determined by the sum of their nutrients, with little consideration
for the extent and purpose of processing and its health implications. A novel Nutrient
Profiling System called Food Compass attempted to address existing gaps and provide a
more holistic assessment of the healthfulness of foods. While a conceptually impressive
effort, we propose that the chosen algorithm is not well justified and produces results that
fail to discriminate for common shortfall nutrients, exaggerate the risks associated with
animal-source foods, and underestimate the risks associated with ultra-processed foods. We
caution against the use of Food Compass in its current form to inform consumer choices,
policies, programs, industry reformulations, and investment decisions.
Unhealthy diets are a leading cause of death and disease.1 A common strategy to improve diets is
by using Nutrient Profiling Systems (NPS)—which are designed to assess the healthfulness of
foods based on their nutrient composition—to inform front-of-package labeling, industry
reformulations, company ratings, policies, programs, and investments.2 However, the application
of NPS is limited because health is not determined by the consumption of single foods but rather
by overall diet quality interacting with unique individual and contextual factors.3 Furthermore,
NPS largely depend on a reductionist paradigm that the healthfulness of a given food is
determined by the mere sum of its nutrients, with little consideration for the extent and purpose
of processing, which has important implications on food quality and health.4 This is problematic
because food processing is not invariably benign. While acknowledging there is a large variation
in the health effects of different types of ultra-processed foods (UPF), in general the higher the
share of UPF in the diet the higher the risk of non-communicable diseases (NCD).5 For instance,
a recent randomized controlled trial found that diets high in UPF led to substantial overeating
and weight gain compared to unprocessed food diets, even though meals within both diets were
matched for energy, energy density, macronutrients, sugar, sodium, and fiber.6 This evidence
suggests that how foods are processed significantly impacts their healthfulness, independently of
nutrient profiles. The resulting overconsumption, weight gain, and related NCD risk is likely
attributed, at least partly, to the degradation of the food matrix and increased hyperpalatability.
A novel NPS called Food Compass attempted to address the concerns of an overly reductionist
approach by comprehensively assessing 54 food attributes across nine health-relevant domains,
including additives, processing, and phytochemicals, among others.2 While a conceptually
impressive effort, we propose that the chosen algorithm and weighting of various components is
not well justified and produces results that do not account for shortfall nutrients, exaggerate the
risks associated with animal-source foods (ASF), and underestimate the risks associated with
A key feature of Food Compass is that it applies the same scoring principles to all foods, aiming
to increase objectivity, avoid arbitrary food categories, allow for direct comparison across items,
and facilitate the assessment of mixed dishes. However, since healthy diets can be achieved in
various ways through combining disparate quantities of different food groups, assessing the
healthfulness of all foods based on a single scale may not be appropriate. Impressively, Food
Compass has undergone validation for content, face, and convergent/discriminant validity, and
the authors plan to test it for construct/predictive validity against health outcomes. This is
welcomed; however, most current assessments of construct/predictive validity consist of
modeling approaches, which require significant assumptions with limited certainty, imply a risk
of circularity, and can be heavily influenced by a limited selection of foods and their
combinations. This raises the question of whether it is possible to robustly validate the health
effects of individual foods, as health outcomes are determined by overall diet quality.3
We propose that a primary limitation of Food Compass is that the weighting of attributes and
domains does not represent a balanced reflection of available evidence and sometimes appears
arbitrary and insufficiently transparent. For instance, phytochemicals are assigned the same
weight as protein and fiber, even though evidence shows that health effects from these domains
are not comparable. Similarly, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)
are weighted the same as alpha‐linolenic acid (ALA), despite the inefficient conversion of ALA
to EPA and DHA, and the more convincing evidence on health benefits of EPA and DHA.7
Other nutrients such as iron and vitamin C, whose abundance in the food supply varies
considerably and deficiencies of which have very different public health significance,8 are also
weighted equally. Additionally, the NOVA classification is only one of 54 attributes,
contributing <5% to the overall score, which fails to sufficiently recognize the crucial role of
food processing in health and disease. The type of processing should arguably be attributed
significantly more weight given the substantial evidence linking UPF with negative health
outcomes,4–6 as well as broader societal impacts associated with aggressive marketing and nutri-
washing of UPF.9 Furthermore, we are concerned that Food Compass did not properly apply
NOVA classifications. For example, many vegetables such as fresh kale and spinach prepared in
oil are classified as UPF, when these should be classified as processed foods. Moreover, several
of these foods incorrectly classified as UPF also received the top score of 100, which suggests
they have not been properly penalized according to the Food Compass algorithm.
Other aspects of Food Compass illustrate its limited consideration of the food matrix. For
instance, while UPF are assigned negative points, unprocessed foods do not receive any positive
points, which does not valorize food matrix quality. Moreover, no distinction is made between
minimally processed and reconstituted whole grains, or between fortified and naturally occurring
nutrients. However, reconstitution and fortification cannot fully replicate the potential benefits of
the food matrix in unprocessed and minimally processed foods, which go beyond their individual
Another limitation is lack of adjustment for differences in bioavailability of iron and zinc, which
varies widely across foods depending on factors such as the proportion of heme to non-heme
iron, phytate, and vitamin C content.11 Also, anti-nutrients (for example, phytate and oxalate) are
not incorporated into the scoring, despite their important role in nutrient absorption. Finally, little
consideration is given to total protein, contributing <4% of the overall score, while protein
quality of foods is omitted.
Several elements in the scoring algorithm are unfavorable towards ASF, regardless of their
processing level. First, no distinction is made between red and processed meat, although
available evidence shows that the consumption-associated risk differs substantially between the
two.12 Second, red meat is always attributed negative scores, even when minimally processed,
despite being a top source of high-quality protein and bioavailable micronutrients commonly
lacking in diets globally, including iron, zinc, and vitamin B12.13 Third, dietary cholesterol is
always scored negatively despite its limited influence on plasma cholesterol,14 inevitably
penalizing ASF. Finally, seafood and yogurt are the only two ASF in the list of healthful food-
based ingredients, while other nutrient-dense ASF that make a significant contribution to diet
quality (for example, eggs and milk) are not included. The excessive penalization of ASF could
have implications on nutrient adequacy in all contexts, but especially in low- and middle-income
countries, where ASF intake is generally low, and the burden of nutrient inadequacies is high,
particularly for iron, zinc, folate, calcium, vitamins A, B12, and D, and essential amino acids and
fatty acids.13,15
While there are many justified scores, the above-mentioned limitations of Food Compass result
in numerous unjustified scores. For example, kale receives the same score as watermelon, despite
the large difference in nutrient density and fiber between the two foods (Figure). Surprisingly,
Frosted Mini Wheats, Honey Nut Cheerios, nonfat frozen yogurt, calcium-fortified orange juice,
and chocolate-covered almonds all receive top scores (≥70, “to be encouraged”). In contrast,
foods such as millet, whole wheat bread, skinless chicken breast, boiled eggs, and whole milk,
are assigned lower scores (31–69, “to be moderated”), which are comparable to those of sweet
potato fries, Lucky Charms, canned pineapple in sugar syrup, almond M&M’s, and ice cream
cones with nuts. Moreover, all of the aforementioned items score higher than ground beef,
cheddar cheese, and whole eggs fried in butter (≤30, “to be minimized”).
Figure. Illustrative examples of limitations of Food Compass scores (non-exhaustive list).
Although a variety of NPS have been developed, currently there is no universally accepted gold
standard, as all existing NPS face technical and conceptual limitations. Food Compass represents
an impressive effort to develop a more comprehensive and objective NPS; nonetheless, we
propose that its overall algorithm and structure needs to be redesigned. If used as intended to
inform consumer choice, policies, programs, reformulations, and investment decisions, Food
Compass scores could result in certain dietary improvements such as increased consumption of
minimally processed plant-source foods, but may have negative implications on nutrient
adequacy in essential commonly lacking nutrients (for example, iron) and, to some extent,
reinforce the consumption of UPF.
Competing Interests statement
TB, SN, MK, and FO have nothing to disclose. SMP reports grants from US National Dairy
Council, Roquette Freres, and Dairy Farmers of Canada during the conduct of the study; non-
financial support from Enhanced Recovery, outside the submitted work. In addition, SMP has a
patent Canadian 3052324 issued to Exerkine, and a patent US 20200230197 pending to Exerkine
but reports no financial gains. SvV reports grant support from USDA (2020-38640-31521; 2021-
67034-35118), the North Dakota Beef Commission, the Turner Institute of Ecoagriculture, and
the Greenacres Foundation during the conduct of the study. SvV reports financial remuneration
for academic talks, but does not accept honoraria, consulting fees, or other personal income from
food industry groups/companies. ML is a Board member of Food Standards Australia New
Zealand (FSANZ) and the views he expresses in this commentary are his alone and not
necessarily those of FSANZ. FL reports financial support of the Research Council of the Vrije
Universiteit Brussel, including the SRP7, IOF3017 and IRP11 projects. FL is a board member of
various academic non-profit organizations including the Belgian Association for Meat Science
and Technology (president), the Belgian Society for Food Microbiology (secretary), and the
Belgian Nutrition Society. On a non-remunerated basis, he also has a seat in the scientific
committee of the Institute Danone Belgium and the Scientific Board of the World Farmers’
Organization. The views he expresses in this commentary are his alone and not necessarily those
of the aforementioned organizations. All authors consume omnivorous diets.
Author contributions
FO and TB prepared the initial manuscript. All authors critically reviewed and approved the final
1. Afshin, A. et al. Health effects of dietary risks in 195 countries, 1990–2017: a systematic
analysis for the Global Burden of Disease Study 2017. The Lancet 393, 1958–1972 (2019).
2. Mozaffarian, D. et al. Food Compass is a nutrient profiling system using expanded
characteristics for assessing healthfulness of foods. Nat Food 2, 809–818 (2021).
3. Beal, T., Neufeld, L. M. & Morris, S. S. Uncertainties in the GBD 2017 estimates on diet and
health. The Lancet 394, 1801–1802 (2019).
4. Fardet, A. & Rock, E. Exclusive reductionism, chronic diseases and nutritional confusion: the
degree of processing as a lever for improving public health. Critical Reviews in Food Science
and Nutrition 0, 1–16 (2020).
5. Lane, M. M. et al. Ultraprocessed food and chronic noncommunicable diseases: A systematic
review and meta-analysis of 43 observational studies. Obesity Reviews 22, e13146 (2021).
6. Hall, K. D. et al. Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An
Inpatient Randomized Controlled Trial of Ad Libitum Food Intake. Cell Metabolism 30, 67-
77.e3 (2019).
7. Zhou, Q. et al. EPA+DHA, but not ALA, Improved Lipids and Inflammation Status in
Hypercholesterolemic Adults: A Randomized, Double-Blind, Placebo-Controlled Trial.
Molecular Nutrition & Food Research 63, 1801157 (2019).
8. Allen, L., De Benoist, B., Dary, O. & Hurrell, R. Guidelines on food fortification with
micronutrients. (WHO/FAO, 2006).
9. Monteiro, C. A. et al. The UN Decade of Nutrition, the NOVA food classification and the
trouble with ultra-processing. Public Health Nutrition 21, 5–17 (2018).
10. Aguilera, J. M. The food matrix: implications in processing, nutrition and health. Critical
Reviews in Food Science and Nutrition 59, 3612–3629 (2019).
11. World Health Organization & Food and Agriculture Organization of the United Nations.
Vitamin and mineral requirements in human nutrition. (WHO/FAO, 2005).
12. Bouvard, V. et al. Carcinogenicity of consumption of red and processed meat. The Lancet
Oncology 16, 1599–1600 (2015).
13. Beal, T. Achieving dietary micronutrient adequacy in a finite world. One Earth 4, 1197–
1200 (2021).
14. Kim, J. E. & Campbell, W. W. Dietary Cholesterol Contained in Whole Eggs Is Not Well
Absorbed and Does Not Acutely Affect Plasma Total Cholesterol Concentration in Men and
Women: Results from 2 Randomized Controlled Crossover Studies. Nutrients 10, 1272
15. Parikh, P. et al. Animal source foods, rich in essential amino acids, are important for
linear growth and development of young children in low- and middle-income countries.
Maternal & Child Nutrition 18, e13264 (2022).
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Nutrient profiling systems (NPS) aim to discriminate the healthfulness of foods for front-of-package labelling, warning labels, taxation, company ratings and more. Existing NPS often assess relatively few nutrients and ingredients, use inconsistent criteria across food categories and have not incorporated the newest science. Here, we developed and validated an NPS, the Food Compass, to incorporate a broader range of food characteristics, attributes and uniform scoring principles. We scored 54 attributes across 9 health-relevant domains: nutrient ratios, vitamins, minerals, food ingredients, additives, processing, specific lipids, fibre and protein, and phytochemicals. The domain scores were summed into a final Food Compass Score (FCS) ranging from 1 (least healthy) to 100 (most healthy) for all foods and beverages. Content validity was confirmed by assessing nutrients, food ingredients and other characteristics of public health concern; face validity was confirmed by assessing the FCS for 8,032 foods and beverages reported in NHANES/FNDDS 2015–16; and convergent and discriminant validity was confirmed from comparisons with the NOVA food processing classification, the Health Star Rating and the Nutri-Score. The FCS differentiated food categories and food items well, with mean ± s.d. ranging from 16.4 ± 17.7 for savoury snacks and sweet desserts to 78.6 ± 17.4 for legumes, nuts and seeds. In many food categories, the FCS provided important discrimination of specific foods and beverages as compared with NOVA, the Health Star Rating or the Nutri-Score. On the basis of demonstrated content, convergent and discriminant validity, the Food Compass provides an NPS scoring a broader range of attributes and domains than previous systems with uniform and transparent principles. This publicly available tool will help guide consumer choice, research, food policy, industry reformulations and mission-focused investment decisions.
Full-text available
Growth faltering under 5 years of age is unacceptably high worldwide, and even more children, while not stunted, fail to reach their growth potential. The time between conception and 2 years of age is critical for development. The period from 6 to 23 months, when complementary foods are introduced, coincides with a time when growth faltering and delayed neurocognitive developments are most common. Fortunately, this is also the period when diet exercises its greatest influence. Growing up in an adverse environment, with a deficient diet, as typically seen in low- and middle-income countries (LMICs), hampers growth and development of children and prevents them from realising their full developmental and economic future potential. Sufficient nutrient availability and utilisation are paramount to a child's growth and development trajectory, especially in the period after breastfeeding. This review highlights the importance of essential amino acids (EAAs) in early life for linear growth and, likely, neurocognitive development. The paper further discusses signalling through mammalian target of rapamycin complex 1 (mTORC1) as one of the main amino acid (AA)-sensing hubs and the master regulator of both growth and neurocognitive development. Children in LMICs, despite consuming sufficient total protein, do not meet their EAA requirements due to poor diet diversity and low-quality dietary protein. AA deficiencies in early life can cause reductions in linear growth and cognition. Ensuring AA adequacy in diets, particularly through inclusion of nutrient-dense animal source foods from 6 to 23 months, is strongly encouraged in LMICs in order to compensate for less than optimal growth during complementary feeding.
Full-text available
This systematic review and meta‐analysis investigated the association between consumption of ultraprocessed food and noncommunicable disease risk, morbidity and mortality. Forty‐three observational studies were included (N = 891,723): 21 cross‐sectional, 19 prospective, two case‐control and one conducted both a prospective and cross‐sectional analysis. Meta‐analysis demonstrated consumption of ultraprocessed food was associated with increased risk of overweight (odds ratio: 1.36; 95% confidence interval [CI], 1.23‐1.51; P < 0.001), obesity (odds ratio: 1.51; 95% CI, 1.34‐1.70; P < 0.001), abdominal obesity (odds ratio: 1.49; 95% CI, 1.34‐1.66; P < 0.0001), all‐cause mortality (hazard ratio: 1.28; 95% CI, 1.11‐1.48; P = 0.001), metabolic syndrome (odds ratio: 1.81; 95% CI, 1.12‐2.93; P = 0.015) and depression in adults (hazard ratio: 1.22; 95% CI, 1.16‐1.28, P < 0.001) as well as wheezing (odds ratio: 1.40; 95% CI, 1.27‐1.55; P < 0.001) but not asthma in adolescents (odds ratio: 1.20; 95% CI, 0.99‐1.46; P = 0.065). In addition, consumption of ultraprocessed food was associated with cardiometabolic diseases, frailty, irritable bowel syndrome, functional dyspepsia and cancer (breast and overall) in adults while also being associated with metabolic syndrome in adolescents and dyslipidaemia in children. Although links between ultraprocessed food consumption and some intermediate risk factors in adults were also highlighted, further studies are required to more clearly define associations in children and adolescents. Study registration Prospero ID: CRD42020176752.
Full-text available
Whole egg is a food source of dietary cholesterol and inconsistent research findings exist about the effect of dietary cholesterol from whole egg on blood cholesterol concentration. We assessed the effect of co-consuming cooked whole egg (CWE) on dietary cholesterol absorption from two randomized-crossover studies. For study 1, 16 men consumed raw vegetables with no egg, 75 g CWE, or 150 g CWE. For study 2, 17 women consumed cooked vegetables with no egg or 100 g CWE. Triacylglycerol-rich lipoprotein fractions (TRL) were isolated from collected blood. In study 1, total-cholesterol areas under the curve (AUC)0–10h in TRL were not different but triacylglycerol AUC0–10h in TRL was greater for 150 g CWE vs. 75 g CWE and no egg. Similarly, in study 2, total-cholesterol AUC0–10h in TRL was not different but triacylglycerol AUC0–10h in TRL was greater for 100 g CWE vs. no egg. In both studies, whole egg consumption did not affect plasma total-cholesterol AUC0–10h, while triacylglycerol AUC0–10h was increased. These results suggest that the dietary cholesterol in whole egg was not well absorbed, which may provide mechanistic insight for why it does not acutely influence plasma total-cholesterol concentration and is not associated with longer-term plasma cholesterol control.
Modern food systems have contributed to extensive environmental degradation, resulting in calls for a plan- etary health diet that dramatically reduces the consumption of animal-source foods. However, animal-source foods provide key micronutrients vital to healthy diets. Planetary boundaries and local contexts must be considered to facilitate regenerative and sustainable livestock production.
Exclusive reductionism in nutritional science consists of viewing foods as only the sum of nutrients. This position paper argues that the extreme application of this paradigm since 1950 has greatly contributed to confusion about a healthy diet among consumers and to the development of chronic diseases worldwide. First, history of nutritional sciences in Western countries shows that by approximately 1850, laboratory research had mainly been conducted by reducing foods to nutrients that were interchangeable from one food to another. Second, descriptive and experimental studies show that the increased prevalence of chronic diseases mainly derive from ultra-processed foods. With such foods being representative of a final output in the degree of food processing, the relevance of reformulating food versus developing less unstructured processed foods is discussed. Third, the reductionist validation of food additives, randomized controlled trials, and food scoring is also questioned. Additionally, epidemiological studies that associate dietary patterns with the risk of chronic diseases and that aggregate approaches in nutrition, technology, food science and food scoring appear to be more adapted for nutritional recommendations in society. It is concluded that a complementary holistic perspective is needed to communicate to society about diet/food health potential and to efficiently prevent populations from chronic diseases.
The Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2017 Diet Collaborators use evidence from primarily observational data and short-term trials of intermediate outcomes to draw conclusions about the causal relationships between individual dietary components and death and disease. We applaud this analysis for raising the crucial importance of diet for overall health, but have some concerns about the potential policy and programmatic implications if results are interpreted literally. Individual dietary components are not eaten separately, but rather are bundled together as diets. However, when individual components are statistically separated, strange conclusions emerge. For example, we question the biological plausibility that low consumption of whole grains specifically is a leading global dietary risk factor for death and non-communicable diseases (NCDs). Whole grains are not themselves a dietary requirement. Our ancestors lived virtually free from NCDs and consumed essentially no grains for millions of years, yet had healthy diets high in micronutrients and fibre. Even cultures that have maintained traditional diets and lifestyles post-industrial revolution have low NCD prevalence. Although increased consumption of whole grains and fruit would probably benefit some populations, a healthy diet can be achieved in many other ways, including by consuming a decreased quantity of refined grains and an increased amount of vegetables. We are surprised that dietary factors associated with the obesity and NCD epidemics (ie, refined grain, sugar, and oil) were not taken into consideration—these might confound the reported associations. We have no doubt that the global food system needs to be changed urgently in favour of health and the environment, but we call for caution with how observational data is used to guide these changes.
The concept of food matrix has received much attention lately in reference to its effects on food processing, nutrition and health. However, the term matrix is used vaguely by food and nutrition scientists, often as synonymous of the food itself or its microstructure. This review analyses the concept of food matrix and proposes a classification for the major types of matrices found in foods. The food matrix may be viewed as a physical domain that contains and/or interacts with specific constituents of a food (e.g., a nutrient) providing functionalities and behaviors which are different from those exhibited by the components in isolation or a free state. The effect of the food matrix (FM-effect) is discussed in reference to food processing, oral processing and flavor perception, satiation and satiety, and digestion in the gastrointestinal tract. The FM-effect has also implications in nutrition, food allergies and food intolerances, and in the quality and relevance of results of analytical techniques. The role of the food matrix in the design of healthy foods is also discussed.