Are the 2019 Canada’s Food Guide Recommendations on Healthy Food Choices Consistent with the EAT-Lancet Reference Diet from Sustainable Food Systems?

Abstract

Background

The diet proposed by the EAT-Lancet Commission, which supports both health and environmental sustainability, provides an opportunity to assess the sustainability of food-based dietary guidelines.

Objectives

The primary objective was to assess the alignment of the 2019 Canada’s Food Guide (CFG) with the EAT-Lancet diet. To do so, an index assessing adherence to the EAT-Lancet diet was developed and evaluated.

Methods

Data from 1147 adults were used from the cross-sectional PRÉDicteurs Individuals, Sociaux et Environnementaux (PREDISE) study conducted between 2015 and 2017 in the province of Québec. The EAT-Lancet Dietary Index (EAT-I) was developed to evaluate adherence to the EAT-Lancet diet. Adherence to the 2019 CFG was assessed using the Healthy Eating Food Index (HEFI)-2019. Associations between the HEFI-2019 score and component scores and the EAT-I score were examined using linear regression models.

Results

The mean EAT-I score (/80) in this population was 33.4 points [95% confidence interval (CI): 32.2, 34.6]. EAT-I scores were consistent with expected differences in diet quality between females and males (+6.9 points, 95% CI: 4.8, 9.0) and between adults aged 50–65 y and 18–34 y (+4.3 points, 95% CI: 1.6, 7.0). The mean HEFI-2019 (/80) score was 44.9 points (95% CI: 44.1, 45.7). The HEFI-2019 was strongly associated with the EAT-I (ß = 0.76, 95% CI: 0.72, 0.80). Among the 10 components of the HEFI-2019, components such as the whole-grain foods (ß =4.01, 95% CI: 3.49, 4.52), grain foods ratio (ß =3.65, 95% CI: 3.24, 4.07), plant-based protein foods (ß =2.41, 95% CI: 2.03, 2.78), and fatty acids ratio (ß =3.12, 95% CI: 2.72, 3.51) showed the strongest associations with the EAT-I.

Conclusions

These results suggest that recommendations in the 2019 CFG are largely coherent with the EAT-Lancet diet underscoring the complementarity and compatibility of the 2019 CFG for sustainability and health promotion purposes.

Introduction

Food production and consumption practices are major causes of environmental degradation and poor human health. The current global food system is estimated to contribute 19%–29% of global greenhouse gas emissions, occupies 37% of the world’s ice-free land area, and is the largest driver of biodiversity loss [[1], [2], [3]]. At the same time, poor nutrition is believed to be responsible for just over 11 million deaths/y in 2017 from cardiovascular diseases, cancers, and type 2 diabetes [4]. To achieve the 2030 Sustainable Development Goals and the Paris Climate Agreement, sustainable dietary patterns, as defined by the Food and Agriculture Organization [5], need to be broadly adopted [6]. In 2019, the EAT-Lancet Commission on Healthy Diets from Sustainable Food Systems proposed the planetary health diet, a dietary pattern supporting both human health and environmental sustainability, with the capacity to feed up to 10 billion people by 2050 [7]. Unsurprisingly, the main features of this global reference sustainable dietary pattern include large amounts of plant-based foods and limited amounts of animal-based foods, refined grains, and foods high in saturated fats and free sugars.

Efforts to reconcile human health and environmental sustainability in food-based dietary guidelines (FBDGs) have been advocated but mixed [6,8]. Indeed, most FBDGs around the world have been developed with a main focus on health, with little or no reference to the environmental aspects of diet [9,10]. The publication of the EAT-Lancet reference dietary pattern in 2019 provided an anchor for governmental authorities to consider dietary sustainability more formally in their FBDGs. However, with the exception of data from the United States and Australia [11,12], there is limited research on how country-specific dietary guidelines align with the EAT-Lancet reference dietary pattern. In Canada, the 2019 Canada’s Food Guide (CFG) discusses the potential environmental benefits of consuming plant-based foods and reducing food waste [13]. Nevertheless, the extent to which recommendations on healthy food choices in the 2019 CFG align with the EAT-Lancet reference diet remains unknown. The release of the EAT-Lancet reference diet has also provided opportunities to assess the adherence to this healthy and sustainable dietary pattern in different populations around the world [[14], [15], [16]], its environmental impact [17,18], and its association with health outcomes [[19], [20], [21], [22]].

There is yet no consensus on how to measure adherence to the EAT-Lancet reference dietary pattern. The scoring methods developed so far vary quite substantially because of differences in interpretation of the EAT-Lancet guidelines [[14], [15], [16], [17],19,[22], [23], [24], [25]]. The number of food components to be included and the food groups to be emphasized, balanced, or limited have been inconsistent across studies. For example, with the exception of fruits, vegetables, beef, lamb and pork, and added sugars, there are large discrepancies in the classification of the other food groups across studies. Moreover, some of the available indices use a dichotomous scoring method, which results in a significant loss of information and a lower degree of discrimination [19,22,23]. The use of a continuous scale (that is, proportional score) [14,16,17,24] is preferable to capture subtle differences in scores.

The primary aim of this study was to assess how the 2019 CFG recommendations on healthy food choices align with the EAT-Lancet sustainable dietary pattern. To do so, an index reflecting the adherence to the EAT-Lancet reference dietary pattern was developed and evaluated.

Methods

Study population

Data from the web-based multicenter cross-sectional PRÉDicteurs Individuels, Sociaux et Environnementaux (PREDISE) study were used for these analyses. Complete procedures of the PREDISE study, which aimed to document associations between individual, social and environmental factors, and adherence to dietary guidelines, have been described elsewhere [26]. In brief, French-speaking adults aged 18–65 y from 5 different administrative regions of the province of Québec (that is, Capitale-Nationale/Chaudière-Appalaches, Estrie, Mauricie, Montreal, and Saguenay-Lac-St-Jean) were recruited through a survey firm between August 2015 and April 2017. Stratified sampling was used to obtain an age- and sex-representative sample of adults aged 18–65 y in each of these 5 administrative regions. Pregnant and lactating women were excluded from the study. Each participant had a 3-wk period to complete online sociodemographic and health questionnaires and 3 web-based 24-h dietary recalls (R24W) and was then invited to an in-person visit at a research center where anthropometric measurements (that is, height and weight) were taken by a trained professional. A total of 1849 participants met the inclusion criteria and gave written consent, of which, 1147 completed at least one 24-h recall and were included in the study (Supplemental Figure 1). The PREDISE study was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committees of Université Laval (ethics number: 2014-271), Centre Hospitalier universitaire de Sherbrooke (ethics number: MP-31-2015-997), Montreal Clinical Research Institute (ethics number: 2015-02), and Université du Québec à Trois-Rivières (ethics number: 15-2009-07.13).

Dietary intake assessment

As described above, participants were invited to complete 3 unannounced R24W over 3 wk, in which they were prompted to report all the foods they consumed the day before. Complete procedures for the development and evaluation of the R24W have been described previously [27,28]. The nutrient values in the R24W were generated using the Canadian Nutrient File 2015. The free sugar content of each participant’s diet was calculated using a free sugar content database for each food item included in the R24W [29]. The approach used is similar to the one proposed later by Health Canada [30].

Healthy Eating Food Index-2019

The Healthy Eating Food Index (HEFI)-2019 was calculated using dietary intake data obtained from R24W to assess adherence to recommendations on healthy food choices in the 2019 CFG [31,32]. The HEFI-2019 includes 10 components, 5 of which are based on foods (vegetables and fruits, whole-grain foods, grain foods ratio, protein foods, and plant-based protein foods), 1 of which is based on beverages (beverages), and 4 of which are based on nutrients (fatty acid ratio, saturated fat, free sugars, and sodium). Each component is scored on a 5- to 20-point scale, and the total HEFI-2019 score ranges from 0 to 80 points. Higher HEFI-2019 scores indicate greater adherence to recommendations on healthy food choices in the 2019 CFG.

Development of the EAT-Lancet Dietary Index

The EAT-Lancet Dietary Index (EAT-I) was adapted from the reference dietary pattern of the EAT-Lancet Commission on Healthy Diets from sustainable food systems. The primary sources of information used for the development of the EAT-I were the exhaustive report of the EAT-Lancet Commission on Healthy Diets from Sustainable Food Systems [7] and its summary report [33]. In brief, as presented in the summary report, the EAT-Lancet’s reference diet is set for a daily intake of 2500 kcal and suggests daily targets with an acceptable range of intakes expressed in grams and kilocalories for 14 food groups or nutrients regrouped into 8 main food groups (that is, whole grains, tubers and starchy vegetables, vegetables, fruits, dairy foods, protein sources, added fats, and added sugars). The food groups classification used in the summary report, their ranges, and target intakes were therefore used for the development of the EAT-I.

To facilitate the calculation of the EAT-I, all targets and ranges of food or nutrient intakes were converted to the percentage of energy of the reference diet, that is, 2500 kcal. This was done because the foods reported in the R24W are prepared and cooked, whereas the EAT-Lancet’s reference diet lists the amount of dry and raw foods. This also allowed us to account for different energy intake scenarios. Each food item in the R24W is reported in grams as well as in energy, allowing the calculation of the percentage of energy of the diet for each food item and group.

From the 8 main food groups of the EAT-Lancet diet listed above, 10 main components were defined to calculate the EAT-I. These components correspond to the main food groups of the reference dietary pattern except for “protein sources” which we further divided into 3 components. Indeed, the “protein sources” group of the EAT-Lancet diet was further divided into the red and processed meats component, the poultry and eggs component, and the fish and plant-based proteins component to better reflect the EAT-Lancet recommendations as not all sources of proteins are encouraged similarly by the Commission. Thus, the 10 components of the EAT-I were as follows: 1) whole grains, 2) tubers and starchy vegetables, 3) vegetables, 4) fruits, 5) dairy foods, 6) red and processed meats, 7) poultry and eggs (poultry and eggs), 8) fish and plant-based proteins (fish, legumes, and nuts), 9) added fats (saturated fats and unsaturated fats), and 10) free sugars (Table 1). Some components of the EAT-I were also further divided into subcomponents to better reflect the specific recommendations on specific food groups, as the target and range of intakes are not the same for these subcomponents. The poultry and eggs component was further divided into the poultry subcomponent and the eggs subcomponent, whereas the fish and plant-based proteins component was further divided into the fish subcomponent, the legumes subcomponent, and the nuts subcomponent. Finally, the added fats component was divided into 2 subcomponents: saturated fats and unsaturated fats. As a result, the components and subcomponents of the EAT-I carefully reflect the recommendations for each of the food groups of the EAT-Lancet diet as presented in the summary report. All main components of the EAT-Lancet dietary pattern were scored on a 10-point scale, with the exception of the dairy foods, red and processed meats, poultry and eggs, and fish and plant-based proteins components, which were each scored on a 5-point scale. These 4 “protein-related” components, which are comprised in 2 of the main food groups originally identified in the EAT-Lancet dietary pattern (that is, dairy foods and protein sources) accounted for a total of 20 points. This approach ensured that the “protein-related” food components did not outweigh other food groups of the EAT-Lancet dietary pattern, as there is no indication that one food group is more important than others. The total EAT-I score ranges from 0 to a maximum of 80 points. Foods included in each component or subcomponent are presented in Supplemental Table 1.

TABLE 1.

EAT-I components and subcomponents, points, and scoring standards

EAT-Lancet diet food group	Component category	Component or subcomponent name1	Measurement (ratio/energy)	Maximum points	Standards
					Minimum scores	Maximum scores
Whole grains	Adequacy	1. Whole grains	Total whole-grain foods	10	0%E	≥32.4%E
Tubers and starchy vegetables	Moderation	2. Tubers and starchy vegetables	Total tubers and starchy vegetables	10	≥3.1%E	0%E
Vegetables	Adequacy	3. Vegetables	Total nonstarchy vegetables	10	0%E	≥3.1%E
Fruits	Adequacy	4. Fruits	Total fruits	10	0%E	≥5.0%E
Dairy foods	Optional	5. Dairy foods	Total dairy foods	5	≥12.2%E	≤6.1%E
Protein sources	Moderation	6. Red and processed meats	Total red and processed meats	5	≥2.4%E	0%E
		7. Poultry and eggs
	Optional	Poultry	Total poultry and other birds	2.5	≥5.0%E	≤2.5%E
	Optional	Eggs	Total eggs	2.5	≥1.5%E	≤0.8%E
		8. Fish and plant-based proteins
	Adequacy	Fish and seafood	Total fishes and seafood	1.67	0%E	≥1.6%E
	Adequacy	Legumes and soy	Total legumes and soy foods	1.67	0%E	≥11.4%E
	Adequacy	Nuts and seeds	Total nuts and seeds	1.67	0%E	≥11.6%
Added fats		9. Added fats
	Adequacy	Unsaturated added fats	Total unsaturated oils	5	0%E	≥14.2%
	Moderation	Saturated added fats	Total saturated oils and added fats	5	≥3.8%E	0%E
Added sugars	Moderation	10. Free sugars	Total free sugars	10	≥4.8%E	0%E

Abbreviation: EAT-I, EAT-Lancet Dietary Index.

¹The 10 components of the EAT-I correspond to the 8 main food groups listed in the EAT-Lancet diet with the “protein sources” group further divided into 3 components (i.e., red and processed meats, poultry and eggs, and fish and plant-based proteins).

Adequacy, moderation, and optional components or subcomponents were defined according to our interpretation of the EAT-Lancet Commission summary report. Accordingly, the final EAT-I comprises 7 adequacy components or subcomponents, that is, whole grains, vegetables, fruits, fish and plant-based proteins (fish, legumes, and nuts), and unsaturated fats, for which higher consumption results in higher adherence scores. Four components or subcomponents, that is, tubers and starchy vegetables, red and processed meats, saturated fats, and free sugars, are defined as moderation components, for which a lower consumption results in higher adherence scores. Two components, that is, dairy products and poultry and eggs, are defined as optional components for which adherence scores vary within the distribution of intakes.

The points and scoring standards applied for each component and subcomponent of the EAT-I are shown in Table 1. For adequacy components and subcomponents, minimum to maximum scores were attributed proportionately to intakes ranging from 0 to the target limit (expressed in percentage of total energy), beyond which the maximum score was maintained. For moderation components and subcomponents, maximum to minimum scores were attributed proportionally to intakes between the proposed lower and upper limits of intakes. A score of 0 was attributed when intakes were above the upper limit. For optional components, maximum scores were attributed to intakes within the lower and target limits. Maximum to minimum scores were given proportionally to intakes ranging from the target to the upper limits.

Statistical analyses

All analyses were performed in SAS Studio (SAS Institute) and figures were generated in R Studio (R Foundation for Statistical Computing). Survey procedures were used when appropriate (see below) to account for the stratified design of the PREDISE study. Balancing weights were used to ensure sex and age representativeness in each administrative region as the final sample size of the PREDISE study was larger than originally planned.

When appropriate, the National Cancer Institute’s (NCI) multivariate Markov Chain Monte Carlo (MCMC) method was used to estimate the distribution of simulated usual (that is, long-term) food and nutrient intakes [34]. This method accounts for within-individual random errors that affect dietary intakes measured with 24-h dietary recalls using regression calibration. The model was stratified by sex to account for potential differences in within-individual random error by sex. The following covariables were included in the model: indicators for the sequence of 24-h recalls (that is, first, second, or third recall) and the day of the week (that is, weekdays or weekend days including Friday) [35], age, smoking status, household income, education, and administrative region. Missing covariables (that is, household income and education) were imputed using the fractional imputation method. The following foods were considered episodic in the model because 10% or more of the participants did not report consumption on their first dietary recall: whole grains, starchy vegetables, vegetables, fruits, red and processed meats, poultry, eggs, fish and seafood, legumes and soy, nuts and seeds, saturated added fats, and unsaturated added fats. The remaining foods and nutrients (that is, kilocalories, dairy foods, and free sugars) were considered to be consumed daily. Usual dietary intakes were generated for a prespecified number of simulations (that is, 500 pseudo-individuals) per participant during the Monte Carlo simulation step and then pooled within each stratum (that is, sex). Finally, EAT-I scores and component scores were calculated from simulated usual intakes among pseudo-individuals.

Evaluation of the EAT-I

Distributions of EAT-I scores and component scores in the overall sample, based on usual dietary intakes estimated using the NCI’s multivariate MCMC method, were generated to assess the variability of the EAT-I among individuals. Still based on usual dietary intakes, the internal consistency of the EAT-I was examined using Cronbach’s α. In addition, pairwise associations among components of the EAT-I were assessed using Spearman correlations. Associations between each component and the EAT-I residual score (that is, the total score minus the score of the component being assessed) were also examined using Spearman correlations.

The NCI’s population ratio method [36] was used to estimate mean EAT-I scores and component scores among subgroups of individuals expected to differ in overall diet quality (that is, sex, age, and smoking status) [32,[37], [38], [39]]. The 95% CIs were estimated using 200 bootstrap resamples.

Alignment of the 2019 CFG recommendations with the EAT-Lancet reference diet

Separate linear regression models were used to assess the associations between the HEFI-2019 score and its component scores (independent variables) and the EAT-I score (dependent variable) in the overall sample using survey procedures. The protein foods component of the HEFI-2019 had to be Box–Cox transformed to improve homoscedasticity. However, because significance test results and goodness of fit were the same using either the transformed or nontransformed protein foods scores, the results are presented using the nontransformed variable. This allows for the interpretation of the results in common units. Of note, the NCI multivariate MCMC method was not applied for these regression analyses because the food groups used to calculate the HEFI-2019 and the EAT-I are not mutually exclusive, and their respective scores are not analyzed on the same scales (that is, kilocalories for the EAT-I and reference amounts for the HEFI-2019). Regression analyses are therefore based on the mean dietary intake data from all available 24-h recalls for each individual.

Results

Evaluation of the EAT-I

The study sample was composed of 50.2% females, 44.6% of participants had a university degree, and the total annual household income was >90,000 CAD for 35.4% of participants (Table 2). The estimated mean EAT-I and HEFI-2019 scores in this population were 33.4 points (95% CI: 32.2, 34.6) and 44.9 points (95% CI: 44.1, 45.7), respectively. Table 3 shows the estimated mean EAT-I score and component scores in specific subgroups of individuals with expected differences in overall diet quality. The estimated mean EAT-I score in males was 6.9 points (95% CI: 4.8, 9.0) lower than in females, and 4.3 points (95% CI: 1.6, 7.0) higher in adults aged 50–65 y than in adults aged 18–34 y. Smokers had 7.9 points (95% CI: 5.4, 10.5) lower estimated mean EAT-I score compared with nonsmokers. Supplemental Figures 2 and 3 show the distribution of the EAT-I score and component scores in 1147 French-speaking adults in the province of Québec.

TABLE 2.

Participants’ sociodemographic characteristics, estimated HEFI-2019 score, and EAT-I score¹

	n (%) or mean (95% CI)
Sex
Female	576 (50.2)
Male	571 (49.8)
Age (y)
18–34	408 (35.6)
35–49	339 (29.5)
50–65	400 (34.9)
BMI2, kg/m2
Normal (<25.0)	403 (39.2)
Overweight (25.0–29.9)	344 (33.4)
Obese (≥30.0)	281 (27.4)
Smoking status
Nonsmokers	984 (85.8)
Smokers	163 (14.2)
Education2
High school or less	269 (24.7)
CEGEP	335 (30.7)
University	487 (44.6)
Income2
<30,000 CAD	162 (16.3)
30,000 to <60,000 CAD	284 (28.6)
60,000 to <90,000 CAD	196 (19.7)
≥90,000 CAD2	352 (35.4)
Number of 24-h recalls completed
1	28 (2.4)
2	34 (3.0)
3	1085 (94.6)
HEFI-2019 score (/80)	44.9 (44.1, 45.7)
EAT-I score (/80)	33.4 (32.2, 34.6)

Abbreviations: CEGEP, Collège d'Enseignement Général et Professionnel; CI, confidence interval; EAT-I, EAT-Lancet Dietary Index; HEFI, Healthy Eating Food Index.

¹HEFI-2019 and EAT-I scores were estimated using the National Cancer Institute’s population ratio method (see Methods section). The 95% CI was calculated with 200 bootstrap resamples.

²BMI group, n = 1028 (119 missing values); education, n = 1091 (56 missing values); income, n = 994 (153 missing values).

TABLE 3.

Estimated mean EAT-I score and component scores in 1147 French-speaking adults of the province of Québec by sex, select age groups, and smoking status¹

Components	Sex			Age groups			Smoking status
	Males	Females	Difference	18–34 y	50–65 y	Difference	Smokers	Nonsmokers	Difference
Whole grains (/10)	2.6 (0.1)	2.9 (0.1)	–0.3 (–0.6, 0.0)	2.4 (0.1)	3.0 (0.1)	0.6 (0.3, 1.0)	1.8 (0.2)	2.8 (0.1)	–1.0 (–1.4, –0.7)
Tubers and starchy vegetables (/10)	0.0 (0.2)	0.9 (0.4)	–0.9 (–1.9, 0.1)	0.0 (0.1)	1.1 (0.5)	1.1 (0.2, 2.1)	0.0 (0.3)	0.5 (0.3)	–0.5 (–1.2, 0.2)
Vegetables (/10)	7.8 (0.2)	9.6 (0.2)	–1.8 (–2.5, –1.2)	8.3 (0.2)	9.2 (0.3)	0.9 (0.1, 1.6)	7.8 (0.5)	8.8 (0.2)	–0.9 (–1.9, 0.0)
Fruits (/10)	6.7 (0.3)	9.6 (0.3)	–2.8 (–3.6, –2.0)	7.7 (0.5)	8.6 (0.4)	1.0 (–0.3, 2.2)	5.0 (0.5)	8.5 (0.3)	–3.5 (–4.6, –2.5)
Dairy foods (/5)	0.5 (0.2)	0.0 (0.0)	0.5 (0.1, 0.9)	0.2 (0.2)	0.0 (0.1)	–0.2 (–0.7, 0.3)	0.2 (0.3)	0.0 (0.1)	0.2 (–0.5, 0.9)
Red and processed meats (/5)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0,0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0, 0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0, 0.0)
Poultry and eggs (/5)	3.0 (0.3)	3.4 (0.2)	–0.5 (–1.2, 0.3)	3.0 (0.5)	3.5 (0.3)	0.5 (–0.6, 1.6)	2.4 (0.5)	3.3 (0.2)	–0.9 (–2.0, 0.2)
Fish and plant-based proteins (/5)	1.8 (0.1)	2.0 (0.1)	–0.2 (–0.5, 0.1)	1.6 (0.1)	2.2 (0.1)	0.6 (0.3, 0.9)	1.4 (0.2)	2.0 (0.1)	–0.6 (–1.0, –0.2)
Added fats (/10)	4.8 (0.2)	5.3 (0.1)	–0.5 (–0.9 ,–0.1)	5.4 (0.1)	4.4 (0.2)	–1.0 (–1.6, –0.4)	4.3 (0.4)	5.1 (0.1)	–0.8 (–1.6, 0.0)
Free sugars (/10)	3.5 (0.1)	3.9 (0.1)	–0.4 (–0.7, –0.1)	3.3 (0.2)	4.1 (0.1)	0.8 (0.4, 1.2)	3.9 (0.2)	3.7 (0.1)	0.2 (–0.3, 0.7)
Total score (/80)	30.7 (0.7)	37.6 (0.9)	–6.9 (–9.0, –4.8)	31.9 (0.9)	36.2 (1.0)	4.3 (1.6, 7.0)	26.8 (1.2)	34.7 (0.6)	–7.9 (–10.5, –5.4)

Abbreviation: EAT-I, EAT-Lancet Dietary Index.

¹Values are mean (SE) or mean difference (95% CI) between groups. Means are calculated using the National Cancer Institute’s population ratio method (see Methods section). SE and 95% CI were calculated using 200 bootstrap resamples.

The EAT-I had a standardized Cronbach’s α of 0.76. The correlation coefficients between each component and the EAT-I residual score varied between 0.01 (dairy foods component) and 0.63 (fish and plant-based proteins component) (Supplemental Table 2). The correlations among individual EAT-I components ranged from –0.08 (poultry and eggs and fish and plant-based proteins components) to 0.55 (fruits and fish and plant-based proteins components; Supplemental Table 2).

Alignment of the 2019 CFG recommendations with the EAT-Lancet reference diet

A strong association was observed between the EAT-I and the HEFI-2019 (R² = 52%, Figure 1), with the EAT-I score increasing by 0.76 points (95% CI: 0.72, 0.80) for each point increase in the HEFI-2019 score. Among the 10 components of the HEFI-2019, the whole-grain foods (ß =4.01, 95% CI: 3.49, 4.52), grain foods ratio (ß =3.65, 95% CI: 3.24, 4.07), plant-based protein foods (ß =2.41, 95% CI: 2.03, 2.78), and fatty acids ratio (ß =3.12, 95% CI: 2.72, 3.51) components showed the strongest associations with the EAT-I (Table 4). Finally, the protein foods and sodium components of the HEFI-2019 showed no association with the EAT-I.

FIGURE 1.

Association between the EAT-Lancet Dietary Index (EAT-I) score and the Healthy Eating Food Index (HEFI)-2019 score in 1147 French-speaking adults of the province of Québec. Data points represent each participant. Both scores used in this analysis were not corrected for within-individual random errors (see Methods section). The 95% CI of the regression line is represented by the shaded area.

TABLE 4.

Associations between the EAT-Lancet Dietary Index (EAT-I) score and individual components of the Healthy Eating Food Index (HEFI)-2019 score in 1147 French-speaking adults of the province of Québec

	EAT-I
HEFI-2019 components	Beta coefficient (95% CI)	R2
Vegetables and fruits (/20)	1.42 (1.30, 1.55)	0.29
Whole-grain foods (/5)	4.01 (3.49, 4.52)	0.18
Grain foods ratio (/5)	3.65 (3.24, 4.07)	0.21
Protein foods1 (/5)	–0.10 (–0.95, 0.75)	0.00
Plant-based protein foods (/5)	2.41 (2.03, 2.78)	0.13
Beverages (/10)	1.44 (1.17, 1.71)	0.09
Fatty acids ratio (/5)	3.12 (2.72, 3.51)	0.17
Saturated fats (/5)	2.01 (1.71, 2.32)	0.12
Free sugars (/10)	1.31 (1.15, 1.47)	0.17
Sodium (/10)	0.15 (–0.08, 0.37)	0.00

¹The variable protein foods was Box–Cox transformed because of nonrespect of the homoscedasticity. Nevertheless, results are presented using the nontransformed variable (see Methods section).

Discussion

The primary aim of this study was to assess the alignment of the 2019 CFG recommendations on healthy food choices with the EAT-Lancet reference pattern of diet sustainability. To do so, we first developed and evaluated an index (EAT-I) that reflects the adherence to the EAT-Lancet reference dietary pattern. Overall, construct validity and reliability analyses support the use of the EAT-I as a measure of adherence to the EAT-Lancet reference dietary pattern. The present findings suggest that the 2019 CFG recommendations are largely consistent with the EAT-Lancet recommendations on healthy diets from sustainable food systems. Most recommendations of the 2019 CFG, as measured by the individual components of the HEFI-2019, aligned with the EAT-Lancet reference diet, with the exception of total protein foods and sodium.

The construct validity of the EAT-I was demonstrated by the acceptable variability of the total score among French–Canadian adults in the province of Québec. As a diet quality index, the EAT-I captured expected differences in diet quality among subgroups of individuals based on sex, age, and smoking status [32,[37], [38], [39]]. The internal consistency of the EAT-I was supported by its Cronbach’s α of 0.76, which is above the acceptable value of 0.70. Nevertheless, the components were defined a priori and included regardless of their contribution to the index, which may have contributed to lowering its internal consistency. More specifically, there was no correlation between the optional components of dairy products and poultry and eggs and the residual EAT-I score, which suggests that each of these metrics potentially measures different constructs. Nevertheless, overall the data suggest that the proposed EAT-I is an adequate measure of adherence to the EAT-Lancet reference dietary pattern.

Compared with previous indices measured on a binary scale [19,22,23,25] or on a multilevel scale [15], the EAT-I developed in this study is on a continuous scale, which is more likely to capture differences in scores and allow for better discrimination between groups [40]. Unlike other continuous indices [16,17,24], the EAT-I developed in this study is also based on the percentage of energy intake rather than g/d. Thus, it allows for different energy intake scenarios to be considered, and no conversion is required if foods are reported as prepared and cooked, which is often the case in nutrition studies. The range of EAT-I score developed here is delimited, which is preferable for determining what is a low or high score, or the magnitude of a score increase. Unlike other indices [16], the EAT-I used in this study assesses the consumption of all food groups of the EAT-Lancet reference diet.

The strong correlation between the HEFI-2019 and the EAT-I scores suggests that the 2019 CFG recommendations on healthy food choices are broadly consistent with the EAT-Lancet reference dietary pattern. These findings are not surprising given that both guidelines emphasize the consumption of plant-based foods such as vegetables, fruits, whole grains, and plant-based proteins, and promote a low consumption of foods high in saturated fats and free sugars [7,41]. Both indices were developed with a health focus, and accordingly, greater adherence to either the 2019 CFG or the EAT-Lancet reference diet has been associated with a reduction in risk of cardiovascular diseases [22,24,42,43]. The strong agreement between the vegetables and fruits, whole-grain foods, grain foods ratio, fatty acids ratio, and free sugars components of the HEFI-2019 and the EAT-I was expected given the strong alignment of the 2019 CFG and the EAT-Lancet Commission guidelines regarding these foods and nutrients.

Although most of the components reflecting the recommendations of the 2019 CFG aligned with the EAT-Lancet reference dietary pattern, divergences were observed for the sodium and protein foods components of the HEFI-2019. The sodium component has previously been shown to be weakly correlated with the HEFI-2019 score [32] and no specific recommendation on sodium consumption is made by the EAT-Lancet Commission [7], which largely explains the lack of association with the EAT-I. Moreover, the poor alignment with the protein foods component was anticipated given that the EAT-Lancet has more specific and ambitious recommendations on plant-based protein foods and on the reduction of animal-based protein foods, with specific targets of intakes, compared with the 2019 CFG, which uses broader messages. Recent studies conducted in the United States and Australia found similar discrepancies in protein foods recommendations in national FBDGs and the EAT-Lancet reference dietary pattern [11,12]. In addition, the lack of association with the protein foods component of the HEFI-2019 is consistent with data from a study concluding that most national FBDGs, including those in North America, are not fully compatible with global environmental targets mainly because of higher recommended intakes of meat and dairy products [9]. Thus, despite that the 2019 CFG emphasizes plant-based protein consumption [13], the present findings suggest that the specific recommendation on protein foods, and possibly the approach used to score this recommendation in the HEFI-2019, do not align well with the healthy and sustainable dietary pattern proposed by the EAT-Lancet Commission. Both factors probably contributed to the lack of association between the protein food component of the HEFI-2019 and the EAT-I. For example, large intakes of animal-based protein foods such as red meat, poultry, eggs, and dairy products are not penalized by the protein foods component of the HEFI-2019, unlike the EAT-I.

International organizations recognize that national FBDGs should aim to improve both health and environmental sustainability [8]. In this regard, the EAT-Lancet Commission advocates for drastic global changes in dietary patterns to address the deleterious impacts of actual diets on health and the environment. Despite important criticisms regarding its applicability worldwide [44], the associations between the dietary patterns proposed by the EAT-Lancet Commission with disease and mortality outcomes [15,21,22,24,43] and with environmental outcomes [17,18,23,24] provide undisputable arguments to encourage comparable dietary patterns in national FBDGs. In this regard, adhering to the 2019 CFG may be a great starting point for Canadians to transition to healthier and more sustainable dietary patterns, given the strong agreement between the 2 guidelines.

A major strength of the present study is the development of a sustainable diet index score measured on a continuous scale that is more likely to capture subtle differences in scores than dichotomic indices, as developed in some previous studies [19,22,23,25]. The use of multiple metrics to evaluate the psychometric properties of the EAT-I is also an important strength. Moreover, this study is the first that aims to evaluate the alignment of the 2019 CFG recommendations on healthy food choices with the EAT-Lancet reference diet. The estimation of usual dietary intakes using the NCI multivariate MCMC method, when possible, represents another strength of the present study, as within-individual random error would have biased the distribution of the EAT-I score and its component scores, and attenuated the correlations observed between each EAT-I component score in the present analyses [45]. Limitations include that the EAT-I and the HEFI-2019 are influenced by the quality of the dietary intake data used, which are subject to random and systematic errors. Second, the capacity of the EAT-I to detect changes over time was not assessed because of the cross-sectional nature of the PREDISE study.

In conclusion, the results support the use of the EAT-I to assess the population’s adherence to the EAT-Lancet reference dietary pattern. The present findings also suggest that the 2019 CFG recommendations on healthy food choices are well aligned with the EAT-Lancet reference dietary pattern. This highlights the complementarity and compatibility of the 2019 CFG recommendations for sustainability and health-promoting purposes.

Acknowledgments

We thank all the participants of the study for their devoted time.

Author contributions

The authors’ contributions were as follows – GR, JR, SL, VP, BL: designed research; GR: performed statistical analysis; GR, BL: wrote the article; BL: had primary responsibility for the final content; and all authors: read and approved the final manuscript.

Conflict of interest

GR received studentships from the Canadian Institutes of Health Research (CIHR) and the Fonds de Recherche du Québec – Santé (FRQS). JR is the Chair of Nutrition at Université Laval, which is supported by private endowments from Pfizer, La Banque Royale du Canada, and Provigo-Loblaws. BL has received funding from the CIHR (ongoing), the FRQS (ongoing), Fonds de Recherche du Québec Nature et Technologies (NT) (ongoing), the Ministère de la Santé et des Services Sociaux (MSSS) du Québec (ongoing), Health Canada (completed in 2021), and Atrium Innovations (completed in 2019). BL is an Advisory Board member of the Canadian Nutrition Society. VP has received funding from the Ministère de la Santé et des Services Sociaux (MSSS) du Québec (ongoing), and Ministère de l'Agriculture, des Pêcheries et de l'Alimentation du Québec (MAPAQ) (ongoing). SL has received funding from the CIHR (ongoing) and Health Canada (completed in 2021).

Funding

This research was supported by an operating grant from the Canadian Institutes for Health Research (CIHR # FHG 129921). The source of support had no role in the writing of this article.

Data availability

Anonymized data, code book, and analytic code used for this study will be made available upon request pending application and approval.

Appendix A. Supplementary data

The following is the Supplementary data to this article: