8c. Dietary Guidelines and Assessing Diet Quality

This chapter covers two related but distinct topics: Food-based dietary guide­lines, often provided by national health authorities, and assess­ment of diet quality. The section on assess­ment of diet quality has a primary focus on applications at the popu­lation level, for monitoring diet quality, assessing diet-disease associ­ations, and evaluating policy and programmatic inter­ventions. The two topics in this section are linked in this popu­lation-level focus, and because assess­ment of diet quality is some­times approached through assessing adherence to national food-based dietary guide­lines. However, the section also covers a range of other approaches to assessing diet quality in popu­lations. Learning objectives for readers include gaining understanding of:

• The role of food-based dietary guide­lines in communi­cating guidance from national health authorities;
• The range of topics covered by food-based dietary guide­lines, and some areas of consistency and difference across various national guide­lines;
• The evolution of definitions of diet quality, and how this has influenced assess­ment of diet quality;
• The wide diversity of diet quality indices currently in use;
• Some of the main types of indices, including those based on national food-based dietary guide­lines, on regional diets, and on other summaries of evidence for diet-health relation­ships;
• How indices differ in their devel­op­ment, measurement (data sources), calculation and use;
• Which indices have lower data require­ments for their calculation;
• How new concerns with sustain­ability and planetary health may affect diet quality definitions and assess­ment.

8c.1 Food-based dietary guide­lines

This section provides a brief overview of the devel­op­ment and communi­cation of food-based dietary guide­lines (FBDGs), with a focus on those devel­oped by national authorities. The section also describes a global archive for FBDGs and comments on areas of consistency across national guide­lines. The section concludes with comments on likely future directions for FBDGs.

Science-based guidance on healthy diets is available from a variety of sources, including from national governments, inter­national authorities such as the World Health Organization (WHO) and the Food and Agriculture Organization of the United Nations (FAO), and medical and dietetic professional societies, among others.

FBDGs translate current scientific knowledge about relation­ships between food intakes and health out­comes into clear popu­lation-level guidance. They are “food-based” in contrast to guidance on nutri­ent intakes such as various Nutrient Reference Values (see Section 8b.1). They often provide advice and guidance on dietary patterns — what to eat and drink, how often, and in what amount. In this Section, we focus on FBDGs devel­oped at the national level.

The earliest dietary guidance from national governments was focused on ensuring nutri­ent adequacy of diets (Harper, 1987). However, by the late 20th century the focus shifted to include a strong emphasis on reducing risks for non-communicable diseases (NCDs, such as cardiovascular diseases, metabolic syndrome, type 2 diabetes, and cancers), while still ensuring nutri­ent adequacy.

More recently, there have been calls to integrate consider­ations of envi­ron­mental sustain­ability into FBDGs; however, to date this has been rare (Gonzalez Fischer & Garnett, 2016; Springmann et al., 2020). Some FBDGs also integrate guidance on physical activity, food safety, food skills (such as choosing healthier foods while shopping, and preparing meals), the role of ultra-processed foods, and other topics.

8c.1.1 Development of national-level food-based dietary guide­lines

WHO and FAO have provided guidance to coun­tries on how to develop FBDGs (WHO/FAO, 1998). This devel­op­ment process for national FBDGs was summarized by the European Food Safety Authority (European Food Safety Authority (EFSA), 2010) as including the following steps:

• Identify diet-health relation­ships (based on national but also global evidence);
• Using national data sources, identify:
  • Diet-related health problems;
  • Nutrients of public health importance;
  • Food groups relevant for FBDGs (for example, food groups with known links to health out­comes, and for sources of nutri­ents of public health importance);
  • Food con­sump­tion patterns;
• Test FBDGs with audiences and refine them;
• Develop graphical representations of FBDGs.

The devel­op­ment of FBDGs requires synthesis of a wide range of evidence and contextual infor­mation, although in practice, some types of infor­mation may not be available. For example, many coun­tries lack recent and nationally representative infor­mation on food con­sump­tion patterns.

The process also calls for the involvement of multiple stakeholders and can be both a technical and a political process. Through considering food con­sump­tion patterns and cultural dietary prefer­ences, the resulting guide­lines should allow for flexibility in food choices, addressing socioeconomic constraints and respecting cultural diversity within national popu­lations (WHO/FAO 1998).

A majority of national FBDGs include some type of summary graphic, some­times called a food guide, to convey concepts like diversity and pro­por­tionality among different food groups. The graphics some­times also include specific details on quan­tities (standard servings per day or per week) for various food groups.

Examples of summary graphics include images of healthy plates, food pyramids, food baskets, or other cultural shapes Three examples are shown in Figure 8c.1 — Malta, Oman, and Estonia.

Malta

Oman

Estonia

(Altamirano Martínez et al., 2015; van't Erve et al., 2017; Herforth et al., 2019). Some coun­tries provide several graphic images, such as both a pyramid and a healthy plate. See, for example, images for the Finnish FBDGs and the Swiss FBDGs.

8c.1.2 Communication and imple­men­tation of national-level food-based dietary guide­lines

FBDGs aim to provide actionable guidance for consumers and for public health and medical practitioners. Full imple­men­tation of FBDGs includes not only devel­op­ment of a variety of targeted consumer educa­tion strategies and materials, but also integration within a range of sectoral policies and programs to help ensure that healthy diets are available, accessible, and affordable (Wijesinha-Bettoni et al., 2021). Policies and programs include those related to public procurement of foods (for example, for feeding in schools and other institutions), taxes and food subsidies, and various food security, nutrition and agricultural policies and programs. More infor­mation on imple­men­tation of FBDGs is available from FAO.

Given these varied uses, governments develop a variety of documents and tools to communi­cate and implement the FBDGs. These may include:

• Technical documents detailing the process of evidence review and justifying each guide­line statement, or key message. Technical documents may include results of systematic reviews of evidence on diet-health relation­ships, diet modeling studies, and/or refer­ences to systematic reviews from other sources (see, for example, New Zealand's documentation of evidence);
• Sector-specific policies, regulations and guidance; for example, policies and guidance for imple­men­tation of school meals programs that incorporate the national FBDGs (see, for example, U.S. regulations for school meals programs);
• Guidance for public health and medical practitioners, which may be at an inter­mediate technical level, and with a practical focus to help them translate the FBDGs for consumers (see, for example, Canada's Dietary Guidelines for Health Professionals and Policy Makers );
• Short, simple consumer-facing documents such as brochures, posters, and brief graphical food guides with advice on food groups (see, for example, the food guide graphic and tri-fold brochure from Benin);
• Consumer-facing internet pages and applications that provide a variety of tools for learning and self-assess­ment (see, for example, Canada's Food Guide webpage).

When evaluating the impact of programs and policies, it is often desirable to measure and assess adherence to dietary guidance. See Section 8c.2 for a discussion of different approaches to assessing diet quality and adherence to dietary guidance. Once fully implemented, the FBDGs themselves should also be evaluated for their effectivenes; information on evaluation of FBDGs is also available at FAO. However, up to the present, infor­mation on effectiveness is limited.

8c.1.3 Repository of national food-based dietary guidelines

The FAO maintains a repository of national FBDGs, which currently (December, 2021) includes FBDGs for 95 coun­tries. Each country page includes standard infor­mation provided by national authorities for most or all coun­tries (see Box 8c.1). In addition to the infor­mation in Box 8c.1, available for most or all coun­tries, newly updated country pages may also include infor­mation on imple­men­tation, evaluation, and incorporation of sustain­ability concerns.

FBDGs are currently (December, 2021) available at the repository for 33 European coun­tries, 31 coun­tries in the Americas, and 18 coun­tries in the Asia-Pacific region, but for only 7 African coun­tries. Availability varies by country income classification, with FBDGs available for slightly over half of all high- and upper-middle-income coun­tries, just over one-quarter of lower-middle-income coun­tries, and for fewer than one in ten low-income coun­tries (based on World Bank 2021 country income classifications).

8c.1.4 Coverage of sub-popu­lations

Countries have taken different approaches to providing guidance for specific age groups and physiological conditions:

• Some coun­tries have taken a life-cycle approach in a single dietary guide­lines document, and include sections with specific guidance for infants, young children, school-aged children, adolescents, and for pregnant and lactating women (see, for example, Kenya's National Guidelines for Healthy Diets and Physical Activity).
• Other coun­tries have devel­oped separate technical documents and/or consumer-facing guidance documents for some or all age and physiological groups (see, for example, New Zealand's technical background papers and consumer resources for different age groups).
• Other coun­tries state that the main dietary guide­lines document and key messages apply to all healthy individuals ≥ 2y, but they may still highlight the needs of age groups or other sub-groups (vegetarians, pregnant or lactating women) within the main guide­lines document (see, for example, Lebanon's Food-based Dietary Guidelines ).
• The most common age group with separate and specific guidance is infants and young children under two years of age (see, for example, technical and consumer-facing FBDG for Panama).

8c.1.5 Consistency of national food-based dietary guide­lines

Commonalities and differences among national FBDGs graphics and key messages were described in several recent reviews (Altamirano Martínez et al., 2015; van't Erve et al., 2017; Herforth et al., 2019). All reviews summarized food guide graphics and noted that the two most commonly used graphics were versions of a pyramid or a plate / food circle. The remainder of this section is based on the review by Herforth et al. (2019).

Nearly all food guide graphics conveyed the importance of food group diversity and pro­por­tionality — that is, that some food groups should be consumed in larger quan­tities than others. However, only about one-third gave quantitative infor­mation on recom­mended amounts per day or per week (either in grams or as the number of servings). Most graphics included depiction of fats and oils and of sugars / sweets, showing them as small in pro­por­tion to other groups. About one-third included depictions of other foods to be consumed in moderation, such as pizza, hamburgers, or salty snacks. About half included depictions of water, and about half included depictions of physical activity.

Excluding depiction of food groups to consume in moderation, the most common number of food groups depicted was four or five, with the most common groupings being: (1) Starchy staple foods; (2) Fruits and vegetables (either as one group or as two separate groups); (3) Dairy; and (4) Other protein foods (meat, poultry, fish, eggs, legumes, and some­times nuts).

However, there was variability in whether legumes were grouped with protein foods, with vegetables, or with both. Dairy was frequently a separate group but was some­times grouped with other protein foods. Many coun­tries did not depict nuts or grouped them with fats and oils rather than with protein foods. Other incon­sisten­cies related to how potatoes were classified (as starchy staples, or as vegetables) and in whether fruit juice was depicted with whole fruits.

Herforth et al. (2019) analyzed key messages in addition to graphics. The most consistent key messages were aligned with quantitative global guidance from the WHO, including messages on ample con­sump­tion of fruits and vegetables and moderation in intake of salt, added sugar, and fats (partic­ularly saturated and trans-fats). Only about one-half of coun­tries emphasized whole grain intake. Key messages encouraging dairy intake were nearly universal in Europe and North America but were less common in Asia, Africa, and Latin America. Some FBDGs provided inconsistent messages on unsaturated fats, with positive key messages indi­cating to choose them but with graphics grouping the healthier choices (nuts and oils) together with less healthy ones (saturated fats) as groups to moderate.

8c.1.6 Future directions for food-based dietary guide­lines

Recent FBDGs from several coun­tries have highlighted the need to consider envi­ron­mental sustain­ability in addition to concerns with nutri­ent adequacy and NCD risk reduction. For example, Brazil (2014), Uruguay (2016), Ecuador (2018), and Sweden (2015) all identify envi­ron­mental sustain­ability as a guiding principle or central focus for their FBDGs, and Qatar (2015) also includes a key message on envi­ron­mental protection. WHO and FAO have also articulated guiding principles for sustain­able healthy diets (FAO & WHO, 2019). As noted above, the FAO repository for FBDG has begun documenting country-level attention to sustain­ability.

Several recently published FBDGs also include a stronger focus on behaviors to help achieve healthy eating patterns. Several of the same coun­tries noted above (Brazil, Ecuador, Uruguay), as well as Canada, include key messages and guidance around food skills, social aspects of eating (enjoying meals with family and with others), and/or being an informed consumer through reading labels and understanding food industry marketing practices. A number of coun­tries also provide guidance to reduce intake of highly processed foods.

In summary, future FBDGs will likely continue to focus on nutri­ent adequacy and NCD risk reduction, but may move toward incorporating new themes of sustain­ability, food skills, and messaging around highly processed foods. In addition, increased attention to go beyond careful devel­op­ment of FBDGs to thorough imple­men­tation and evaluation is also warranted (Wijesinha-Bettoni et al., 2021). Finally, FBDGs will continue to be updated to reflect new knowledge as well as changes in lifestyles and food supplies.

8c.2 Assessing diet quality

This section starts with brief comments on definitions of diet quality, and how these have changed over time. We then provide detailed discussion of the various ways researchers have measured diet quality, both in high-income “data-rich” settings, and in more resource-constrained settings. As noted earlier, our focus is on measurement for popu­lation-level applications. Numerous metrics (indicators, indices, and scores) have been devel­oped to measure the quality of the diet. See Box 8c.2 for comments on vocabulary for metrics. Some indices are based on national FBDGs but others are not, and have a different rationale. For recent reviews of diet quality indices, see: Burggraf et al., 2018; Trijsburg et al., 2019; Aljuraiban et al., 2020; and Miller et al., 2020.

The section covers:

• How the definition of diet quality has changed over time
• The basis, or rationale for various diet quality indices
• Uses of diet quality indices
• The development process for diet quality indices
• Detailed examples of different types of diet quality indices

By providing detailed examples of diet quality indices, we aim to illustrate their variety and uses. The section closes with a brief discussion of new efforts to incorporate the idea of envi­ron­mental sustain­ability into diet quality definitions and indices.

8c.2.1 Changing definitions of diet quality

Diet quality indices have evolved along with changing definitions of diet quality (see Box 8c.3). Early indices such as the “Mean Adequacy Ratio” (Guthrie et al., 1972) aimed to summarize nutri­ent adequacy only. Newer diet quality indices often aim to reflect multiple char­act­eristics of diets. Many indices aim to reflect both nutri­ent adequacy and char­act­eristics of diets related to NCD risk. Currently, efforts are underway to also incorporate sustain­ability concerns.

8c.2.2 Basis for diet quality indices

Indices used for assessing diet quality may be based on national or global dietary guide­lines, or they may be based on other assess­ments of evidence related to nutritional adequacy, diet-health relationships, and/or environmental sustainability. Depending on the underlying definition of diet quality as well as on intended uses, diet quality indices may be designed to capture one, several, or all of the following char­act­eristics of diets:

• Food groups consumed: Quantities, diversity between and/or within food groups, and/or balance (pro­por­tionality) among food groups;
• Macronutri­ent balance (for example, percent of energy from protein, carbo­hydrates and fat);
• Nutrient density of the diet (for example, nutri­ents per 1000kcals);
• Nutrient adequacy of the diet relative to require­ments;
• Moderation in con­sump­tion of foods and food groups that elevate NCD risk;
• Adherence to dietary patterns designed to minimize risk of NCDs;
• Adherence to defined healthy cultural dietary patterns;
• Environmental impact and/or sustain­ability of the diet;
• Adherence to dietary guidance from national or global health authorities, such as national FBDGs or World Health Organization (WHO) guidance.

There is much overlap among these char­act­eristics. For example, measures of adherence to healthy cultural dietary patterns may capture food group diversity, pro­por­tionality, macronutri­ent balance, nutri­ent adequacy and NCD risk reduction. Examples include indices for regional diets such as Mediterranean diets (D'Alessandro & De Pergola, 2018) and “adapted” Mediterranean diets (Fung et al., 2006), Japanese diets (Kanauchi & Kanauchi, 2019), and Nordic diets (Hillesund et al., 2014).

As noted above, national FBDGs also generally have multiple objectives, including ensuring nutri­ent adequacy and reducing NCD risks. Increasingly, as noted in Section 8c.1, FBDGs may also aim to address envi­ron­mental impacts of diets (Gonzalez Fischer & Garnett, 2016) and/or to specifically advise moderation in con­sump­tion of certain types of highly processed food (Ministry of Health of Brazil, 2014; Health Canada, 2019).

Thus, diet quality indices designed to reflect adherence to national or global guide­lines will reflect the definitions of diet quality and the diverse aspects of diets and dietary patterns that are described in the partic­ular guide­lines.

8c.2.3 Uses of diet quality indices

Diet quality indices have been devel­oped and used for a variety of purposes, including:

• Assessment and monitoring
  • Describing diet quality for an individual
  • Describing diet quality among a popu­lation
  • Estimating distributions of diet quality among a popu­lation
  • Describing trends in popu­lation diet quality over time
• Epidemiological research
  • Examining associ­ations between diet quality and a dependent variable (for example, does adherence to a dietary pattern reduce risk of a partic­ular NCD?)
  • Examining associ­ations between an independent variable and diet quality (for example, do lower income groups have lower diet quality?)
• Programmatic and policy uses
  • Targeting programs
  • Assessing the effects of policy and programmatic inter­ventions on diet quality
  • Advocacy and informing policy devel­op­ment

Just as with nutri­ent intake assess­ment (Chapter 8b), diet quality assess­ment often requires accounting for day-to-day variation in intakes. See Sections 3.3 and 8a.2.1 for discussions of appropriate dietary measurement and analytic methods for various objectives.

Certain char­act­eristics of indices may be more or less important depending on intended uses. For example, when an index will be used to compare diet quality across coun­tries (or across regions or cultures within coun­tries), the index should have the same meaning across diverse cultural dietary patterns (Frongillo et al., 2019).

In some contexts, when an index is needed for frequent monitoring of trends (for example, annually or biannually), lower cost, lower burden indices may be required due to resource constraints. When an index will be used to summarize diet quality for the purposes of communi­cating or advocating with non-technical audiences (whether consumers or policymakers) the index should be amenable to simple, clear presentation and inter­pretation. Finally, for most uses, and partic­ularly for use in assessing trends and in evaluating impacts of policy and programmatic inter­ventions, indices should be responsive — that is, they should change in response to meaningful changes in diet quality.

8c.2.4 Development of diet quality indices

There are two general approaches to assessing dietary patterns of individuals, and to devel­op­ment of indices. One approach assesses the diet relative to a pre-defined set of criteria (“a priori” dietary patterns) such as national or global recom­mendations, or a defined cultural dietary pattern.

The second general approach (“a posteriori” dietary patterns, some­times called “exploratory”) is data driven, involving the examination of clustering of diet char­act­eristics in a partic­ular study sample (Gleason et al., 2015). There are several analytic approaches to defining a data-driven dietary pattern, including factor analysis, principal com­po­nents analysis, and cluster analysis. Reduced rank regression is a hybrid approach, with a priori identification of char­act­eristics combined with data-driven identification of patterns. Because data-driven dietary patterns are difficult to generalize beyond a partic­ular study setting, they are not further discussed in this section.

For a priori dietary patterns, selection of index com­po­nents (foods, nutri­ents, etc.) and quantitative criteria for intakes may be well-defined (for example, based on guide­lines), but there are still many decisions to be taken during devel­op­ment of indices (Box 8c.4). In the next Section, many of these decisions will be illustrated with a detailed example describing devel­op­ment of the U.S. “Healthy Eating Index” (HEI), which measures adherence to the U.S. dietary guide­lines. Both during devel­op­ment of indices and afterwards, performance is evaluated in several ways. Indices should be valid and reliable and well-suited for intended uses. Validation of indices aims to answer the following questions, among others:

• Is the index well-grounded in theory?
• Does it cover all dimensions / char­act­eristics of interest (content validity)?
• Does it perform in a way that is consistent with theory (construct validity)?

Kirkpatrick et al. (2019) provide a detailed discussion of validation issues for dietary assess­ment measures. For a discussion of validation concepts more broadly, see Frongillo et al. (2019).

8c.3 Examples of diet quality indices

This Section presents a series of specific examples of diet quality indices. Literature reviews have identified hundreds of diet quality indices; some are study- or context-specific, while others are proposed for national, regional or global use (Burggraf et al., 2018; Trijsburg et al., 2019; Aljuraiban et al., 2020; Miller et al., 2020). The purpose of this Section is to provide a small set of examples of diverse types of indices, with differing origins and intended uses.

Most indices have been devel­oped for adult popu­lations, though they have some­times been adapted for children and adolescents. Indices devel­oped or adapted for children and adolescents were recently reviewed by Dalwood et al. (2020).

Historically, most diet quality indices were initially devel­oped in high-income coun­tries and rely on quantitative dietary recall data from food frequency ques­tion­naires and/or quantitative 24hr recalls for their measurement. These indices also require avail­ability of detailed and compre­hensive food composition data. However, several recent initiatives have aimed to develop lower burden proxy indices measuring one or more diet quality char­act­eristics, with less costly data require­ments, for use in global and/or national monitoring. Table 8c.1 presents some examples of some of the most widely used indices, as well as some newer indices intended for global use.

Table 8c.1 Examples of diet quality indices

Name	Basis of Index
Food and/or nutri­ent-based indices
U.S. Healthy Eating Index (HEI)	Measures adherence to key recom­mendations in the Dietary Guidelines for Americans
Alternate Healthy Eating Index (AHEI)	Measures con­sump­tion of a literature-based selection of foods and nutri­ents consistently associated with decreased risk of NCDs
Mediterranean Diet Score (MDS)	Measures adherence to one of the traditional dietary patterns of the Mediterranean region, or adapted versions of these
Dietary Approaches to Stop Hypertension diet index (DASH)	Measures adherence to a dietary pattern originally devel­oped in a randomized controlled trial for reducing hyper­tension
Healthy Diet Indicator (HDI)	Measures adherence to WHO global dietary recom­mendations for the prevention of chronic disease
Indices based on other char­act­eristics
Dietary Inflammatory Index (DII)	A scoring system summarizing the inflammatory potential of the diet; inflammatory diets have been linked to NCDs
Percent of energy from ultra-processed foods (% UPF)	Metric based on evidence that intakes of UPF are associated with poor quality diets and health risks
Lower-burden indices designed for settings where quantitative data are not feasible
Minimum dietary diversity (MDD)	Proxy for micro­nutri­ent density of infant and young child diets
Minimum dietary diversity for women of reproductive age (MDD‑W)	Proxy for micro­nutri­ent adequacy of women's diets
Global Dietary Recom- mendations Score (GDR)	Adherence to WHO Healthy Diet guidance and World Cancer Research Fund / American Institute for Cancer Research recom­mendations; proxy for HDI
Global Diet Quality Score (GDQS)	Measures con­sump­tion of a literature-based selection of food groups that contribute to nutri­ent intake and NCD risk reduction across a variety of global dietary patterns.

The first set of food and/or nutri­ent-based indices in Table 8c.1 — the HEI, AHEI, MDS, DASH, and HDI — have a longer history of use, and most have been updated several times to reflect evolutions in national or global recom­mendations and/or newer syntheses of epidemi­ological evidence. Figure 8c.2

shows results from simple searches of the PubMed database, showing the volume of studies mentioning each type of index before 2015 as compared to in the years 2015–2020. The U.S. Healthy Eating Index (HEI) has been widely used over several decades and is very well documented. Information on recom­mended presentation and inter­pretation is also available. Because of this, we use the HEI as a detailed example. Examples of other indices follow but are described in less detail.

8c.3.1 Basis, devel­op­ment, and scoring of the U.S. Healthy Eating Index

The Healthy Eating Index (HEI) was designed to reflect adherence to the U.S. national dietary guide­lines, which apply to all individuals ≥ 2y. The dietary guide­lines are intended both to ensure nutri­ent adequacy and to reduce risks of NCDs, and hence the HEI also reflects both adequacy and NCD risk reduction. The HEI can be used to characterize diets of children ≥ 2y as well as adults.

The US dietary guide­lines are updated every five years. The HEI was first published in 1995 (Kennedy et al., 1995), and was updated in 2005, 2010 and 2015 to reflect new and revised guidance (Krebs-Smith et al., 2018); the HEI will be updated again in 2022.

The HEI can be calculated using data from quantitative 24hr recalls, food records, or food frequency ques­tion­naires. The appropriate method of calculation depends on the data type and on intended uses. For most uses, the HEI should be calculated based on usual dietary intake (i.e., accounting for day-to-day variability in intakes) (Kirkpatrick et al., 2018).

The original HEI included 10 com­po­nents: five food groups reflecting adequacy, four nutri­ents reflecting moderation, and a measure of variety in food intake. Each com­po­nent was scored from 0 to 10, with adequacy com­po­nents (food group scores) based on total intake, yielding a total possible score of 100. Table 8c.2 shows scoring for the newer HEI‑2015, which includes nine adequacy com­po­nents, and four moderation com­po­nents.

Table 8c.2 Components and scoring standards for the Healthy Eating Index‑2015. Source: U. S. Department of Agriculture Food and Nutrition Service
^a Intakes between the minimum and maximum standards are scored pro­por­tionately.
^b Includes 100% fruit juice.
^c Includes all forms except juice.
^d Includes legumes (beans and peas).
^e Includes all milk products, such as fluid milk, yogurt and cheese, and fortified soy beverages.
^f Includes seafood, nuts, seeds, soy products (other than beverages), and legumes (beans and peas).
^g Ratio of poly- and mono-unsaturated fatty acids (PUFAs and MUFAs) to saturated fatty acids (SFAs).

Component	Max. pointsa	Standard for maximum score	Standard for minimum score
Adequacy
Total fruitsb	5	≥ 0.8 cup equivalent per 1000kcal	No fruit
Whole fruitsc	5	≥ 0.4 cup equivalent per 1000kcal	No whole fruit
Total vegetablesd	5	≥ 1.1 cup equivalent per 1000kcal	No Vegetables
Greens and beansd	5	≥ 0.2 cup equivalent per 1000kcal	No dark-green vegetable or legumes
Whole grains	10	≥ 1.5 ounce equivalent per 1000kcal	No whole grains
Dairye	10	≥ 1.3 cup equivalent per 1000kcal	No dairy
Total protein foodsd	5	≥ 2.5 ounce equivalent per 1000kcal	No protein foods
Seafood and plant proteinsd,f	5	≥ 0.8 ounce equivalent per 1000kcal	No seafood or plant proteins
Fatty acidsg	10	(PUFA + MUFA)/SFAs ≥ 2.5	(PUFA + MUFA)/SFAs ≤ 1.2
Moderation
Refined grains	10	≤ 1.8 ounce equivalent per 1000kcal	≥ 4.3 ounce equiv- alent per 1000kcal
Sodium	10	≤ 1.1 grams per 1000kcal	≥ 2 grams per 1000kcal
Added sugars	10	≤ 6.5% of energy	≥ 26% of energy
Saturated fats	10	≤ 8% of energy	≥ 16% of energy

The scoring for the HEI illustrates some of the many decisions made during devel­op­ment of diet quality indices more generally. These include:

• Nature of com­po­nent parts (for example, food groups, nutri­ents, or both);
• Operational definition of what food items “count” for each com­po­nent.
• Decision on how to account for energy intakes;
• Criteria for scoring of each com­po­nent, including quantity of intake for maximum score;
• Relative scoring of com­po­nents (i.e., weighting), and range for total score;

Krebs-Smith et al. (2018) describe the rationale for all decisions taken during devel­op­ment of the HEI‑2015. In brief, selection of com­po­nent parts reflects key recom­mendations in the 2015–2020 Dietary Guidelines for Americans (U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). Footnotes to Table 8c.2 provide some infor­mation on how foods groups are defined and operationalized and additional details are available on the HEI website.

Except for the first (1995) version, HEI scoring has been based on amounts per 1000kcal (i.e., density), rather than on total amounts. One rationale for use of the density approach is that the HEI can then be applied at different levels (individuals, groups of people, food envi­ron­ments, food supply).

The density approach also has the effect of minimizing (though not eliminating) differences in food group quantity recom­mendations across differing energy intake levels (which in turn correspond to differences in age, size and activity level). To allow for consistent scoring, the HEI takes the least restrictive (easiest to achieve) target for density of intakes for energy intake levels ranging from 1200–2400kcal.

For most adequacy com­po­nents (food groups), the maximum score is given for intake at or above the target density, and a minimum score of “0” is given for no intake. For fatty acid adequacy, and for the moderation com­po­nents, the rationale for minimum and maximum scores is somewhat more complex (see Krebs-Smith et al., 2018 for details).

For all com­po­nents, densities between the minimum and maximum standards are scored pro­por­tionally. For example, consuming no whole grains is scored at zero, and consuming 1.5 ounces of whole grains per 1000kcals is scored at the maximum of 10  points. Consuming 0.75 ounces per 1000kcals is scored at 5 points since it is halfway from zero to the target density.

While the HEI‑2015 has more “adequacy” com­po­nents than the original HEI (9 vs.5), the original total score of 100 is maintained. Certain original food groups are sub-divided: for example, the original HEI awarded up to 10 points for vegetables, whereas HEI‑2015 awards 5 points for “greens and beans” and an additional 5 points for “total vegetables”.

Considering the five broader food groups (vs. subgroups, that is: fruits; vegetables; whole grains; dairy; and protein foods), each is equally weighted in HEI‑2015, with a maximum score of 10 points. Fatty acids and each of the 4 moderation com­po­nents also have maximum scores of 10 points. Overall, the adequacy com­po­nents total 60 of the 100 points, whereas the moderation com­po­nents total 40 of the 100 points.

The rationale for equal weighting of com­po­nents is that the Dietary Guidelines for Americans are meant to be considered as a whole and that all concepts are equally important; further, there is currently no evidence to support unequal weighting (Krebs-Smith et al., 2018). Equal weighting of com­po­nents is a feature of many other diet quality indices, and for the same reasons.

8c.3.2 Evaluation of U.S. Healthy Eating Index‑2015

The validity of the HEI‑2015 and its predecessors has has been assessed in several ways. Krebs-Smith et al. (2018) report on content validity — that is, the extent to which the index captures relevant dimensions of a healthy diet, as defined in the U.S. dietary guidance.

Reedy et al. (2018) report on construct and criterion-related validity. Construct validity was assessed in several ways. “Exemplary menus” were shown to receive high HEI‑2015 scores; scores in a nationally representative sample demonstrated sufficient variation (range from the 1^st–99^th percentile was from 33 to 81 points) and scores were significantly different between smokers and non-smokers, as expected a priori. Scores were independent of energy intakes, i.e., reflective of quality, rather than quantity of intakes.

Criterion-related validity was demonstrated in a prospective cohort, where the highest vs the lowest quintiles of the HEI‑2015 scores were associated with a decreased risk of all cause, cancer, and cardiovascular disease mortality. In addition, scores for previous versions of the HEI have been shown to be associated with reduced risks of NCDs and mortality in numerous studies (see, for example, Liese et al., 2015; Schwingshackl & Hoffmann, 2015).

8c.3.3 Interpretation of the U.S. Healthy Eating Index-2015

For descriptive purposes, total scores can be complemented by graphical representations and by a grading scale (Krebs-Smith et al., 2018). Radar graphs ( Figure 8c.3)

can be used to visualize popu­lation-level dietary patterns. In the radar graph, each com­po­nent score is plotted as a percent of maximum points, on an axis. The perimeter illustrates a perfect score. Radar graphs are also useful because they show how the same total score can represent different patterns (sets of com­po­nent scores).

To qualitatively describe popu­lation-level scores, a graded approach can be used, with 90 to 100 points an “A”, 80 to 89 a “B”, etc. Similarly, each com­po­nent score can be graded (Krebs-Smith et al., 2018). However, categorized scores (translated into “grades”) should not be used for analysis purposes as categorization results in loss of infor­mation and may result in misclassification.

To interpret the magnitude of a difference in scores between groups, or across time in the same popu­lation, a difference (or change) of 5–6 points was shown to represent a “moderate” effect size for the U.S. popu­lation (Kirkpatrick et al., 2018). However, the authors caution that this may not apply to other popu­lations, since effect sizes are calculated in relation to the typical dispersion of scores (standard deviations), which may vary across popu­lations.

More generally, the radar graph or graded approach could be used for descriptive purposes with a variety of multi-com­po­nent diet quality indices. Determining meaningful differences could similarly be informed by examination of standard deviations of the score in the popu­lation of interest, ideally in multiple studies. A difference (or change) in scores of approx­imately one‑half of a standard deviation is considered to be a moderate effect size (Kirkpatrick et al., 2018).

Box 8c.5 summarizes key points about the HEI.

8c.3.4 Adaptation of the Healthy Eating Index concept

In addition to the periodic updates related to updated U.S. dietary guide­lines, the HEI concept has also been adapted for other contexts. Similar indices measuring adherence to national guide­lines were devel­oped for many coun­tries, including:

• Australia (Roy et al., 2016)
• Brazil (Previdelli et al., 2011; Andrade et al., 2013)
• Canada (Garriguet, 2009; Jessri, Ng, and L'Abbé, 2017)
• Chile (Universidad de Chile, 2014)
• China (Yuan et al., 2017; 2018)
• Denmark (Bjerregaard et al., 2019)
• France (Estaquio et al., 2009; Chaltiel et al., 2019)
• Germany (Moon et al., 2017; Mader et al., 2020)
• Korea (Yook et al., 2015; Park et al., 2020)
• Malaysia (Pondor, Gan, & Appannah, 2017)
• Mexico (López-Olmedo et al., 2019)
• The Netherlands (Looman et al., 2017)
• Russia (Martinchik et al., 2019)
• Spain (Norte Navarro & Ortiz Moncada, 2011)
• Thailand (Taechangam, Pinitchun, & Pachotikarn, 2008)

In Ethiopia, devel­op­ment of food-based dietary guide­lines is underway and an Ethiopian HEI will also be devel­oped (Bekele et al., 2019). HEI were also created for children 1y and older in Finland (Kyttälä et al., 2014), pregnant women in Singapore (Han et al., 2015), and adolescents in Taiwan (Y. C. Chen et al., 2018). Many of these indices are also scored from 0 to 100, but some have different numbers of com­po­nents and different ranges for scoring.

8c.4 Alternate Healthy Eating Index

While the HEI and similar indices in other coun­tries measure adherence to current national dietary guide­lines, another very widely used index, the “Alternate Healthy Eating Index” (AHEI), was devel­oped to reflect additional evidence related to NCDs, which had not been incorporated in the US national guidance at the time.

As first devel­oped by McCullough et al. (2002), the AHEI incorporated some of the same elements as the original HEI, but in addition emphasized: con­sump­tion of whole grains; a favorable ratio of fish and poultry to red and processed meats; intake of nuts and soy foods; and a favorable ratio of polyunsaturated to saturated fats. Unlike the HEI, dairy is not included in the scoring. The AHEI showed stronger associ­ations than the original HEI to cardiovascular disease out­comes for both men (McCullough et al., 2000) and women (McCullough et al., 2000). The authors noted that this evidence could help inform future revisions of dietary guide­lines.

Table 8c.3 Components and scoring standards for the Alternate Healthy Eating Index 2010. Adapted from Chiuve et al. (2012).
Several apparent errors in original paper were corrected in the table above, by referring to Liese et al. (2015), Supplemental material p. 6.
PUFA = polyunsaturated fatty acids.
^aIntakes between the minimum and maximum standards are scored pro­por­tionately.
^b Includes all vegetables except potatoes. One serving is 0.5cup of vegetables or 1cup of green leafy vegetables.
^c Does not include fruit juice. One serving is 1 medium piece of fruit or 0.5cup of berries.
^d Use of grams of whole grains accounts for the variability of the percentages of whole grain in various. One serving of a 100% whole-grain product (i.e., 0.5cup of oatmeal or brown rice) contains ≈ 15–20g of whole grains (per dry weight).
^e One serving is 8oz.
^f One serving is 1oz of nuts or 1 tablespoon (15mL) of peanut butter.
^g One serving is 4oz of unprocessed meat or 1.5oz of processed meat.
^h Optimal intake (250mg/d) is ≈ two 4oz servings of fish per week.
ⁱ Cutoffs for sodium were based on deciles of distribution in the data sets used to develop the index. This was due to lack of brand specificity on the food frequency ques­tion­naires, such that absolute intakes could not be accurately estimated. This implies users would also base cutoffs on deciles of distribution in the data set analyzed.
^j Highest scores are for moderate alcohol intake, and the worst score is for heavy con­sump­tion. Nondrinkers receive an inter­mediate score of 2.5.

Component	Standard for maximum. score (10)a	Standard for minimum score (0)
Vegetables, servings/d b	≥ 5	0
Fruit, servings/d c	≥ 4	0
Whole grains, g/d d — Women	75	0
Whole grains, g/d d — Men	90	0
Sugar-sweetened beverages and fruit juice, servings/d e	0	≥ 1
Nuts and legumes, servings/d f	≥ 1	0
Red/processed meat, servings /d g	≤ 1	≥ 1.5
Trans fat, % of energy	≤ 0.5	≥ 4
Long-chain (n-3) fats (EPA + DHA), mg/d h	250	0
PUFA, % of energy	≥ 10	≤ 2
Sodium, mg/d i	Lowest decile	Highest decile
Alcohol, drinks/d j — Women	0.5–1.5	≥ 2.5
Alcohol, drinks/d j — Men	0.5–2.0	≥ 3.5
Total points	110	0

Like the HEI, the AHEI was later revised to reflect new epidemi­ological evidence on relation­ships of dietary com­po­nents to NCD risks, and the newer version is referred to as the AHEI 2010 (Chiuve et al., 2012). As in the original version, the rationale for the AHEI 2010 was a compre­hensive review of clinical and epidemi­ological research by the authors, with selection and scoring of com­po­nents that were deemed to be consistently associated with NCD risk. As with the original, the AHEI 2010 did not fully align with the HEI.

Scoring for the AHEI 2010 is shown in Table 8c.3. Both versions of the AHEI were devel­oped using semi-quantitative food frequency data from prospective cohort studies of U.S. health profes­sionals. However, subsequent users have also calculated scores for the AHEI from quantitative 24hr recall data and for other popu­lation groups, including adolescents (see, for example Dalwood et al., 2020; Ducharme-Smith et al., 2021); however, Dalwood et al. caution against application of indices devel­oped for adults in pediatric popu­lations.

Like the HEI, the AHEI includes both foods and nutri­ents (fatty acids) among the com­po­nent parts, but it also includes sweet beverages and alcohol. Unlike the HEI, the AHEI is not scored based on densities (quan­tities per 1000kcal), but instead based on total con­sump­tion per day. Unlike in the HEI, fruit juice is included with sugar-sweetened beverages rather than with fruit.

Highest scores are based on literature-based judgments of minimum risk. Consistent with the HEI, intakes that are inter­mediate between the minimum and maximum are generally scored pro­por­tionally. However, alcohol has “U-shaped” scoring, such that the highest score is given for moderate intake and lower scores for “0” or for excessive intake. Sodium intakes are scored based on the study specific distribution. Also as for the HEI, com­po­nents each have a maximum of 10 points, in this case yielding a range of up to 110 points (11 com­po­nents), with 6 “positive” com­po­nents, 4 “negative” com­po­nents and alcohol (U-shaped). Box 8c.6 summarizes key points about the AHEI.

8c.5 Mediterranean Diet Score or Index

Evidence on the healthfulness of various Mediterranean regional diets has been accumu­lating since the mid‑20^th century (Menotti & Puddu, 2015). Generally, Mediterranean diets are characterized by a high intake of plant-based foods including olive oil, fruit, nuts, vegetables, and cereals; a moderate intake of fish and poultry; a low intake of dairy products, red meat, processed meats, and sweets; and a moderate amount of wine. However, unlike the dietary patterns captured by the HEI and AHEI, which each have a single definition, there are numerous Mediterranean diets and numerous scores and indices (see: Milà-Villarroel et al., 2011; D'Alessandro & Pergola, 2018; Zaragoza-Martí et al., 2018).

In contrast to the HEI and AHEI, Mediterranean Diet Scores (MDSs) are generally based on foods and beverages only, rather than on a mix of foods/beverages and nutri­ents. Intended uses of MDSs are not consistently clearly stated in papers reporting on their devel­op­ment. They are frequently used in epidemi­ological studies investigating associ­ations of diets to morbidity and mortality. MDSs vary widely in their scoring range (for example, possible scores of 0–8 vs. 0–100) and in the rationale and method of scoring, as illustrated in the following examples.

8c.5.1 Mediterranean Diet Score of Trichopoulou

MDSs with the narrowest ranges are scored as “yes” (1) or “no” (0) for a small set of food groups. Trichopoulou et al. (1995) devel­oped one of the first MDS and demonstrated an associ­ation with mortality in a Greek popu­lation. The original analysis was based on semi-quantitative food frequency data from adults over 70y. This MDS was designed for an epidemi­ological study. The MDS of Trichopoulou et al. is an 8-point score, assessing for the following:

• High con­sump­tion of legumes
• High con­sump­tion of cereals (including bread and potatoes)
• High con­sump­tion of fruits
• High con­sump­tion of vegetables
• Low con­sump­tion of meat and meat products
• Low con­sump­tion of milk and dairy products
• High ratio of monounsaturated to saturated fat
• Moderate alcohol con­sump­tion

This score has 5 “positive” 2 “negative” and one “U‑shaped” com­po­nent like the AHEI, for moderate alcohol intake. Dairy is included as a “negative” com­po­nent; this is typical of many MDSs. “High” intake was defined as intake above the sex-specific median quantity consumed by their study subjects rather than in relation to a set of recom­mended intakes, and therefore scoring for this MDS is dependent on study-specific distributions of intake. This approach to scoring is useful for study-specific comparisons but less useful for comparing across contexts. Several other MDSs are scored in this way.

8c.5.2 “MedDietScore” of Panagiotakos

For a prospective cohort study of adults in Athens, Panagiotakos et al. (2007) devel­oped a 55‑point “MedDietScore” where scoring did not depend on the distribution of intake among study subjects. Instead, each com­po­nent was scored from 0–5 based on respon­dent report of frequency of con­sump­tion, based on data from a semi-quantitative food frequency ques­tion­naire covering the past year.

Component scores increased with frequency for 7 “positive” com­po­nents — non‑refined cereals, fruits, vegetables, legumes, olive oil, fish and potatoes — and decreased with frequency for 3 “negative” com­po­nents — red meat and products, poultry and full‑fat dairy products. As with several other indices, there were more “positive” com­po­nents than “negative” com­po­nents.

This MedDietScore was negatively associated with hyper­tension, hyper­chol­esterol­emia, diabetes and obesity in the initial study and was later shown to be negatively associated with 10‑year incidence of cardiovascular disease (Panagiotakos et al., 2015) and diabetes (Filippatos et al., 2016).

8c.5.3 “MEDI-LITE” Score of Sofi

Sofi et al. (2014) devel­oped an 18‑point “literature-based” adherence score, shown in Table 8c.4. This adherence score, the “MEDI-LITE” score, was designed for both epidemi­ological and clinical use; however, a primary aim was to develop a simple tool feasible for use at the individual level, within clinical practice.

Table 8c.4 Components and scoring standards for MEDI-LITE. Adapted from Sofi et al. (2014),
Note these authors use the term “portion” in the way the term “serving” is used here — that is, as a defined standard quantity.

Component	Low intake (score)	Moderate intake (score)	High intake (score)
Fruit (1 portion = 150g)	< 1 portion/d (0)	1–1.5 portions/d (1)	> 2 portions/d (2)
Vegetables (1 portion = 100g)	< 1 portion/d (0)	1–2.5 portions/d (1)	> 2.5 portions/d (2)
Legumes (1 portion = 70g)	< 1 portion/wk (0)	1–2 portions/wk (1)	> 2 portions/wk (2)
Cereals (1 portion = 130g)	< 1 portion/d (0)	1–1.5 portions/d (1)	> 1.5 portions/d (2)
Fish (1 portion = 100g)	< 1 portion/wk (0)	1–2.5 portions/wk (1)	> 2.5 portions/wk (2)
Meat/meat products (1 portion = 80g)	< 1 portion/d (2)	1–1.5 portions/d (1)	> 1.5 portions/d (0)
Dairy products (1 portion = 180g)	< 1 portion/d (2)	1–1.5 portions/d (1)	> 1.5 portions/d (0)
Alcohol (1 alcohol unit (AU) = 12g)	< 1 AU/d (1)	1–2 AU/d (2)	> 2 AU/d (0)
Olive oil	Occasional use (0)	Frequent use (1)	Regular use (2)

Component scoring was based on a set of prospective cohort studies, published between 2010 and 2013. The studies were primarily from Europe and included the multi-country European Prospective Investigation into Cancer and Nutrition (EPIC) study as well as additional studies from some of the EPIC study coun­tries (Denmark, France, Germany, Greece, Italy, The Netherlands, Norway, Spain, Sweden, and the UK). Several studies were from the U.S., and no studies were from Asia, Africa, or Latin America. All studies included adult participants only.

For each com­po­nent, intakes were categorized into three levels after calculating weighted averages (weighted by study sample size) and standard deviations of intakes across all studies. The score includes six “positive” com­po­nents, two “negative” com­po­nents, and U-shaped scoring for alcohol.

Construct validity for the MEDI-LITE score was demonstrated by comparing it to the previously validated MedDietScore of Panagiotakos (above).

8c.5.4 Mediterranean-Style Dietary Pattern Score of Rumawas

More recently, some Mediterranean diet indices have been proposed with wider scoring ranges and with scoring based on criterion values (recom­mended intakes). Rumawas et al. (2009) devel­oped a 100‑point Mediterranean-Style Dietary Pattern Score (MSDPS) based on the Mediterranean Diet Pyramid of Willett et al. (1995); scoring is shown in Table 8c.5.

Table 8c.5 Components and scoring standards for the Mediterranean-Style Dietary Pattern Score. Adapted from Rumawas et al. (2009), p. 1152.
^a Each com­po­nent of the score is calculated based on the recom­mended intakes of food in the Mediterranean Diet Pyramid of Willett et al. (1995).
^b Except olive oil, all other com­po­nents are continuous, ranging from 0–10 and computed pro­por­tionately.
If con­sump­tion exceeds the recom­mended intake, the score is deducted pro­por­tionally to the number of servings consumed that exceed the recom­mended intake; the lowest possible score due to deduction is zero.

Component	Standard for maximum score of 10a	Scoreb
	Servings/day	Points/serving
Whole grains	8	1.25
Fruits	3	3.33
Vegetables	6	1.67
Dairy	2	5.0
Wine - Women	1.5	6.67
Wine - Men	3	3.33
	Servings/week	Points/serving
Fish and other seafood	6	1.67
Poultry	4	2.5
Olives, legumes, and nuts	4	2.5
Potatoes, starchy roots	3	3.33
Eggs	3	3.33
Sweets	3	3.33
Meat	1	10.0
	Practice	Points
Olive oil	Use only olive oil	0 (for no use) 5 (for use of olive and other vegetables oils)

The MSDPS was explicitly intended for use in non-Mediterranean popu­lations. During devel­op­ment, performance of the MSDPS was assessed among adults using data from a semi-quantitative food frequency ques­tion­naire covering the past year.

The thirteen com­po­nents are each scored from 0–10, then summed. The total score is standardized to 100 by dividing the calculated sum by the theoretical maximum sum of 130 and multiplying by 100. To account for foods not on the Mediterranean Diet Pyramid, the standardized score is weighted by the pro­por­tion of total energy intake derived from foods on the pyramid. For example, if 20% of kilocalories are from foods not on the pyramid, the standardized score is multiplied by 0.8.

As for the HEI and AHEI, intakes that are lower than recom­mended are scored pro­portion­ately to the recom­mended amounts. Scoring is for total quan­tities (as for the AHEI) and not for densities (as for the HEI).

Unlike some other indices, scoring for the MSDPS “penalizes” over­con­sumption of a given food group by subtracting points in pro­por­tion to the amount of excess — for example, exceeding the recom­mended intake by 60% would result in a score of 4 for a given com­po­nent. The lowest possible score for each com­po­nent is zero. This means, for example, that both 2 servings of meat/week and 10 servings of meat/week are scored identically as zero.

Construct validity was demonstrated through expected positive associ­ations with intakes of dietary fiber, fatty acids, antioxidant vitamins, calcium, magnesium, and potassium, and inverse associ­ations with a glycemic index and with intakes of added sugar, saturated fat, and trans‑fat.

8c.5.5 Summary of Mediterranean Diet Scores

There is no consensus on a single or “best” Mediterranean Diet Score or Index, and this depends on intended uses; see D’Alessandro and De Pergola (2018) for a discussion. Although many MDSs were originally devel­oped based on analyses of data from adult popu­lations, including elders, MDSs have been adapted and used in studies of children and adolescents (Dalwood et al., 2020).

Finally, concerning the potential to adapt the MDS for relevance in other regions, there is more experience in high-income countries and in the global north compared to the global south. The applicability of the pattern described by the MDS to some other regions is not established, particularly in regions where some of the distinctive components (for example, olive oil, red wine, nuts, legumes, and/or fish) are not available and/or are not part of cultural dietary patterns.

However, reporting on a global World Heart Federation Consensus Conference, Anand et al. (2015) suggest that the evidence base for the Mediterranean diet is sufficiently consolidated that it can be recom­mended globally, and they provide suggestions for its adaptation with foods from other regions. Once the dietary pattern is adapted to other regions, adapted scores could also be devel­oped (as, for example, in Echeverría et al., 2019; El Kinany et al., 2020). Box 8c.7 summarizes key points about MDSs.

8c.6 Dietary Approaches to Stop Hypertension scores

Scores reflecting adherence to the Dietary Approaches to Stop Hypertension (DASH) dietary pattern have also been widely used, partic­ularly in studies investigating associ­ations between dietary patterns and NCDs. The original DASH diet and scoring were devel­oped for a randomized controlled feeding trial investigating the impact of a prescribed dietary pattern on blood pressure. The trial, initiated in 1994, compared a typical U.S. dietary pattern to one high in fruits, vegetables, and low-fat dairy products, that emphasized fish and chicken rather than red meat, and was low in saturated fats, choles­terol, sugars, and refined carbo­hydrate.

The original study demonstrated reductions in blood pressure after the 8‑week inter­vention (Sacks et al., 1999). Follow-up research demonstrated additional benefits from reductions in sodium intake; sodium intake had not been manipulated in the original trial (Sacks et al., 2001). While the DASH diet was originally devel­oped to lower blood pressure, subsequent studies have linked higher adherence to the DASH diet with lower risks for a range of adverse out­comes, including all-cause mortality, cardiovascular disease incidence or mortality, cancer incidence or mortality, type 2 diabetes, and neuro­degen­erative disease (Morze et al., 2020).

As for Mediterranean diets, there have been several specific scores devel­oped to reflect adherence to the DASH diet, and DASH scores have also been devel­oped for pediatric popu­lations. Table 8c.6

Table 8c.6 Components and scoring standards for an example Dietary Approaches to Stop Hypertension diet adherence score. Scoring as in Fung et al., 2008 and Liese et al., 2015.

Component	Standard for maximum score	Standard for minimum score
Whole grains	Highest quintile	Lowest quintile
Vegetables (excluding potatoes)
Fruit
Nuts and legumes
Low-fat dairy
Red and processed meat	Lowest quintile	Highest quintile
Sugar-sweetened beverages
Sodium

shows one commonly used scoring system with 8 com­po­nents, each worth 5 points, for a total of 40 points. The scoring system is based on sex-specific quintile rankings within the study popu­lation.

With the exception of sodium, this DASH adherence score is based on foods rather than nutri­ents. As with the HEI, AHEI and MDS above, the eight com­po­nents are given equal weight. As with many of the score described above, there are slightly more “positive” than “negative” com­po­nents, implying an overall heavier weighting of positive food groups. Unlike in many MDSs, low-fat dairy is scored positively.

Like some versions of the MDS, and also like some other versions of DASH diet scores, scoring is dependent on study-specific distributions of intake. This type of scoring can maximize contrasts for studies investigating associ­ations between dietary patterns and out­comes. However as noted above, it is less useful for comparing intakes across contexts.

8c.7 Healthy Diet Indicator

Just as the HEI was devel­oped to capture adherence to U.S. national dietary guidance, the Healthy Diet Indicator (HDI) was devel­oped to capture adherence to global guidance from the WHO (Huijbregts et al., 1997) and was revised to reflect updated guidance (Stefler et al., 2014; Jankovic et al., 2015). Both the original and the updated versions demonstrated associ­ations with mortality in multi-country studies (Huijbregts et al., 1997; Stefler et al., 2014), although the associ­ations were not entirely consistent (Jankovic et al., 2015).

The original HDI study analyzed data from cohorts of older men in three western European coun­tries (Huijbregts et al., 1997), while the revised version was first used in studies of adults from Eastern Europe (Stefler et al., 2014) and elders from the U.S. and Europe (Jankovic et al., 2015). To our knowledge, the HDI has not been widely used in studies of children.

Unlike all the previous scores and indices, most of the com­po­nents in the HDI are nutri­ents or food substances rather than food groups. This reflects the fact that the WHO “Healthy Diet Fact Sheet” currently gives quantitative guidance for intakes of sugars, various types of fatty acids, and salt, but limited guidance on quantitative intakes of food groups.

Table 8c.7 Components and scoring standards for the Healthy Diet Indicator 2020. Eleven com­po­nents are scored “0” or “1”. Components 1–5 are used in a sub-index for healthy dietary com­po­nents (maximum score, 5), and com­po­nents 6–11 are used in a sub-index for dietary com­po­nents to limit (maximum score, 6).
^a Upper end of the recommendation to consume no more than 350–500 grams/week divided by 7 days. Adapted from Herforth et al (2020).

	Component	Standard for scoring (quantitative intake in one day)
1	Fruits, vegetables	≥ 400g
2	Beans and other legumes	> 0g
3	Nuts and seeds	> 0g
4	Whole grains	> 0g
5	Dietary fiber	> 25g
6	Total fat	< 30% total energy
7	Saturated fat	< 10% total energy
8	Dietary sodium	< 2g
9	Free sugars	< 10% total energy
10	Processed meat	0g
11	Unprocessed red meat	≤ 71ga

The original HDI included 9 com­po­nents, each scored “0” for not meeting or “1” for meeting the WHO recom­mendation, for a total range of 0–9 points. Components included: saturated fatty acids; poly­unsaturated fatty acids; protein; “complex carbo­hydrates”; and mono- and disaccharides (all as percent of energy), and quan­tities for choles­terol; fiber; fruits and vegetables; and pulses / nuts / seeds. As with the indices described above, the com­po­nents were equally weighted. The updated HDI included similar com­po­nents, but each com­po­nent was scored from 0 to 10, to better capture variation among study subjects.

Most recently, Herforth et al. (2020) have again updated the HDI, reflecting the current WHO Healthy Diet recom­mendations and incorporating World Cancer Research Fund / American Institute for Cancer Research (WCRF / AICR) recom­mendations on the con­sump­tion of red and processed meats. Table 8c.7 shows the com­po­nents and scoring for the HDI‑2020.

Eleven com­po­nents are scored “0” or “1”, for a scoring range of 0–11. Several positive com­po­nents are scored “1” for any non‑zero con­sump­tion, based on WHO guidance that indicates that healthy diets include these items, but without specifying recom­mended quan­tities (whole grains, beans and other legumes, and nuts and seeds). Components are equally weighted. Unlike many of the examples above, the total weight for all moderation / unhealthy com­po­nents is slightly higher than the weight for healthy com­po­nents. The revised HDI‑2020 has not yet been evaluated for associ­ations with health out­comes. Box 8c.8 summarizes key points about HDIs.

8c.8 Dietary Inflammatory Indices

Recently, several diet quality indices have been devel­oped to capture the inflammatory potential of the diet (Tabung et al., 2016; 2017; Hébert et al., 2019). Such indices were devel­oped in recognition of the key role of chronic low-grade inflammation in the etiology of NCDs, and the role of dietary factors in chronic inflammation. In this section, we describe one widely used index, the Dietary Inflammatory Index (DII).

The DII was designed for global relevance, regardless of the cultural dietary pattern (Hébert et al., 2019). The DII is scored primarily based on nutri­ents and food substances. Selection of and weighting of com­po­nents were based on a compre­hensive literature review spanning 1950–2010 and including studies from diverse popu­lations that assessed the relation­ships of nutri­ents, foods, and food substances to six inflammatory biomarkers (Cavicchia et al., 2009; Shivappa et al., 2014).

The DII has been applied in studies of a wide range of NCDs and risk factors. A recent review documented associ­ations between DII scores and certain cancers, cardiovascular disease and its associated mortality, adverse mental health, and musculoskeletal disorders. Evidence on DII and respiratory health, neurodevel­op­mental out­comes, the metabolic syndrome, obesity and diabetes is either conflicting or limited to date (Phillips et al., 2019).

Scoring for the DII is based on 45 com­po­nents, consisting mainly of macro‑ and micro­nutri­ents and food substances (for example, flavonoids, caffeine) but also including several foods, herbs/spices (garlic, ginger, onion, pepper, rosemary, saffron, thyme, and turmeric), and beverages (tea and alcohol). In the most recent version, scoring cut-offs were determined based on associations of components to inflammation in studies from eleven coun­tries in all global regions except Africa. (Shivappa et al., 2014).

Each of the 45 com­po­nents has a score indi­cating its pro‑ or anti-inflammatory potential, with positive scores for pro-inflammatory com­po­nents and negative scores for anti-inflammatory com­po­nents. The DII is the sum of the com­po­nent scores. Unlike many indices described above, the com­po­nents of the DII are not equally weighted. The total range in seven “example scenarios” was from +7.98 (strongly pro-inflammatory) to −8.87 (i.e. strongly anti-inflammatory); the example scenarios were created by the authors to illustrate various combinations of plausible intakes for the 45 com­po­nents (Shivappa et al., 2014). Note that — unlike other indices — a negative score is better than a positive one.

The DII has been further adapted in a version that adjusts for total energy (E-DII) and in a version for children (C-DII), validated for children 6–14y. The C-DII predicted blood concentration of one inflammatory marker (C-reactive protein) in children (Khan et al., 2018). Construction of the DII requires data on all com­po­nents, though it has been adapted for use when data for some com­po­nents are not available (Davis et al., 2021).

The DII has primarily been used to further elucidate the relation­ship between the inflammatory potential of overall diets and various health out­comes. Box 8c.9 summarizes key points about the DII.

8c.9 Diet quality based on level of food processing

Another recent approach to assess­ment of diet quality is based on classification of foods and beverages according to the level of processing in their production. Monteiro et al. (2010) first proposed a system of classification, which was subsequently modified to include four groups:

• Unprocessed or minimally processed foods;
• Processed culinary ingredients;
• Processed foods; and
• Ultra-processed food and drink products.

The modified classification scheme is referred to as the “NOVA” classification (Monteiro et al., 2016). Particular attention has been paid to the role of ultra-processed products (UPF) in diets. While the classification is at the level of food and beverage items, the quality of the overall diet is then characterized by the pro­por­tion of UPF in the diet (usually, as a percent of total energy) (Monteiro et al., 2011; Chen et al., 2020).

Machado et al. (2019) describe UPF as “formulations of low-cost ingredients, many of non-culinary use, that result from a sequence of industrial processes (hence ultra-processed)”. The identification and classification of UPF have recently been clarified:

“A practical way to identify an ultra-processed product is to check to see if its list of ingredients contains at least one item charact­eristic of the NOVA ultra-processed food group, which is to say, either food substances never or rarely used in kitchens (such as high-fructose corn syrup, hydro­genated or inter­esterified oils, and hydrolysed proteins), or classes of additives designed to make the final product palatable or more appealing (such as flavours, flavour enhancers, colours, emulsifiers, emulsifying salts, sweeteners, thickeners, and anti-foaming, bulking, carbonating, foaming, gelling and glazing agents)” (Monteiro et al., 2019, p. 936).

Examples of UPF include packaged instant soups and noodles, carbonated beverages, reconstituted meat products, many sweet and savory packaged snacks, frozen “ready meals”, and fast food dishes. Some UPF provide substantial nutri­ents, either from food ingredients (for example, packaged sweetened yogurts) or from fortification (for example, highly processed but fortified breakfast cereals), leading to challenges on definition of categories, and some disagreement on the value of the NOVA classification (Gibney, 2019; Jones, 2019).

The percent of energy from UPF has been shown to be high or very high in some coun­tries; for example, UPF comprise from 40%–50% of energy intake in Australia (Machado et al., 2019), Brazil (Siqueira et al., 2020) and Canada (Polsky, Moubarac, & Garriguet, 2020) and nearly 60% in the United States (Martínez-Steele et al., 2017). While similar intake data are not available for most low- and middle-income coun­tries, sales of UPF are growing rapidly in Asia and Africa (Vandevijvere et al., 2019; Baker et al., 2020). Assessment of the role of UPFs in the diet has been undertaken for a wide range of age groups, beginning in infancy.

A higher pro­por­tion of total intake from UPF has been associated with lower intakes of some micro­nutri­ents and higher intakes of sugars, salt, and saturated and trans-fatty acids (see for example Martínez-Steele et al., 2017; Moubarac et al., 2017; Machado et al., 2019). However, several studies documented higher calcium intakes with higher UPF con­sump­tion, perhaps due to con­sump­tion of sweetened and/or other processed dairy products (Louzada et al., 2015; Batal et al., 2018; Cornwell et al., 2018). Recent reviews have summarized associ­ations with a wide range of risk factors and negative health out­comes, including overweight, obesity and cardio­metabolic risks, some cancers, type 2 diabetes and cardio­vascular diseases, irritable bowel syndrome, depression and all‑cause mortality (Chen et al., 2020; Elizabeth et al., 2020). Box 8c.10 summarizes key points about the percent of energy from UPF.

8c.10 Lower-burden diet quality indices

Most diet quality indices described above require quantitative dietary intake data, for example from food frequency ques­tion­naires, quantitative 24hr recalls, or other detailed quantitative approaches. Obtaining accurate quantitative dietary intake data is resource intensive. Because of the resource require­ments, many low‑ and middle-income coun­tries do not currently have nationally representative dietary intake data, and even high-income coun­tries may not collect such data sufficiently frequently to meet all needs, such as for monitoring of population-level diet quality trends.

To meet needs for basic infor­mation on intake patterns in contexts where quantitative data are not available, several “lower-burden” approaches have been devel­oped, partic­ularly for use in global and national monitoring under resource constraints. These lower-burden approaches are all food group- based and/or ask behavioral questions (rather than including nutri­ent intakes for their calculation) and do not require food composition data for tabulation.

8c.10.1 Global surveillance of dietary behaviors

Many lower-burden survey tools (often called “screeners”) have been devel­oped to capture only one or several com­po­nents of intake, and/or dietary habits. Examples of screeners devel­oped for global use are the ques­tion­naires for the WHO Stepwise Approach to NCD Surveillance (STEPS) and the Global School-Based Student Health Survey (GSHS). Currently, the WHO STEPS ques­tion­naire includes questions on fruit, vegetable, and salt/salty condiment con­sump­tion. The GSHS ques­tion­naire is under revision; previously, it included core questions on fruits, vegetables, carbonated soft drinks, and food from fast food restaurants. To date, neither of these surveys captures the “whole of diet”, and data from these surveys have not been used to develop overall indices of diet quality.

8c.10.2 Food group diversity proxy indicators for micro­nutri­ents

Some food group diversity indicators have demonstrated associ­ations with micro­nutri­ent adequacy of diets (Verger et al., 2021). Because of this, several have been devel­oped as simple proxies for micro­nutri­ent adequacy, for use when more complex indices are not feasible. These indicators do not require use of food composition data, and they do not require detailed quantitative dietary intake data.

They were devel­oped primarily for use in resource-constrained envi­ron­ments, where diets may be very impoverished, low in food group diversity, and consequently provide inadequate micro­nutri­ents. They are intended for popu­lation-level use — for example, in monitoring trends — and not to describe diets of individuals.

Indices were devel­oped for several demographic groups, including infants and young children and women of reproductive age. In both cases, construct validity was evaluated cross-sectionally by comparing the simple indices to micro­nutri­ent density or adequacy of diets, in data sets from studies in multiple low- and lower-middle income coun­tries. (Working Group on Infant and Young Child Feeding Indicators, 2006; Arimond et al., 2010; Martin-Prevel et al., 2017).

A “Minimum Diet Diversity” (MDD) indicator was devel­oped as a proxy indicator of the micro­nutri­ent density of the diet for infants and young children 6–23mos (WHO 2008; 2010). MDD data can be collected using a lower-burden non-quantitative recall ques­tion­naire for food groups fed to the infant or young child the day before the survey.

An updated version includes 8 food groups (below; WHO/UNICEF, 2021). Infants and young children fed five or more of the eight food groups, in any amount, meet the criterion for MDD. Groups of infants and young children meeting MDD are likely to have a diet higher in micro­nutri­ent density than those who consume fewer food groups. Food groups are:

• Breast milk
• Grains, white/pale starchy roots, tubers and plantains
• Beans, peas, lentils, nuts and seeds
• Dairy products (milk, infant formula, yogurt, cheese)
• Flesh foods (meat, fish, poultry, organ meats)
• Eggs
• Vitamin A-rich fruits and vegetables
• Other fruits and vegetables

The MDD has been widely used and reported, with the main global data sources being the Demo­graphic and Health Surveys (DHS) Program and the UNICEF Multiple Indicators Cluster Surveys (MICS). Prevalence data for MDD are available at the UNICEF infant and young child feeding indicators database.

Recently, the World Health Organization and UNICEF have devel­oped additional feeding indicators relevant to diet quality of infants and young children, including indicators of sweet beverage con­sump­tion and con­sump­tion of selected unhealthy food groups that may displace nutri­ent-dense foods in their diets. However, there is no overall summary index for diet quality that covers both micro­nutri­ent density and non-recom­mended food groups (WHO/UNICEF, 2021).

A simple food group diversity indicator, MDD-W, was also devel­oped for non-pregnant women of reproductive age as a proxy for micro­nutri­ent adequacy of the diet, partic­ularly for use in settings where poverty results in monotonous diets (Martin-Prevel et al., 2017; FAO, 2021).

Like the infant and young child MDD indicator, data can be collected using a lower-burden non-quantitative recall ques­tion­naire for food groups consumed the day before the survey. The recom­mended data collection methods are designed to exclude very small quan­tities of con­sump­tion (< 15g) (FAO, 2021).

Groups of women consuming five or more of the ten defined food groups are likely to have higher micro­nutri­ent adequacy than women consuming fewer food groups. The underlying 10‑point score is also associated with micro­nutri­ent adequacy.

The 10 food groups differ slightly from the infant and young child food groups, as follows:

• Grains, white roots and tubers, and plantains
• Pulses (beans, peas, and lentils)
• Nuts and seeds
• Dairy
• Meat, poultry and fish
• Eggs
• Dark-green leafy vegetables
• Other vitamin A rich fruits and vegetables
• Other vegetables
• Other fruit

MDD‑W data are increasingly available, including from the 8th phase of the DHS Program (2018–2023). Both the original MDD and the MDD‑W have also been evaluated for their usefulness in predicting micro­nutri­ent adequacy for other demographic groups (for example, Ganpule-Rao et al., 2021; Diop et al., 2021).

Like the infant and young child MDD, the MDD‑W was not devel­oped to capture other char­act­eristics of diet quality, such as NCD risk reduction. This is a key limitation, because in many resource-constrained envi­ron­ments, there is a growing burden of overweight, obesity and associated NCDs, following on rapid nutrition transitions (Popkin, Corvalan, & Grummer-Strawn, 2020). This limitation has been addressed in several recent efforts to develop more compre­hensive indices that can still be captured using lower-burden data collection methods.

8c.10.3 Global Dietary Recom­mendations Score

In addition to proposing an updated HDI‑2020 based on global recom­mendations from the WHO and the World Cancer Research Fund / American Institute for Cancer Research; (as described above), Herforth et al. (2020) also devel­oped a Global Dietary Recom­mendations (GDR) Score. The GDR Score is a lower-burden index capturing adherence to these global recom­mendations. It was explicitly devel­oped as a complement to the MDD‑W, to fill a gap in low burden indices reflecting NCD risk reduction.

Like the MDD-W, the GDR Score is meant for use in assessing and monitoring diets at the popu­lation level, not the individual level. Construct validity was assessed by comparing the GDR Score to the HDI‑2020 in two nationally representative data sets, from Brazil and the United States, using data for ages ≥15y.

Table 8c.8 Food groups in the Global Dietary Recom­mend­ations Score, based on a one-day non-quantitative recall. Adapted from Herforth et al., (2020).

GDR-Healthy (positively scored)
Dark-green leafy vegetables
Vitamin A-rich orange-colored vegetables, roots, tubers
Other vegetables
Vitamin A-rich fruits
Citrus fruits
Other fruits (including red/purple/blue fruits)
Legumes
Nuts/seeds
Whole grains
GDR-Limit (negatively scored)
Sodas/sugar-sweetened beverages
Baked/grain-based sweets
Other sweets
Processed meat (double weight)
Unprocessed red meat
Deep-fried foods
Food from a fast-food restaurant, or Instant noodles
Packaged salty snacks

Data for the GDR Score can be collected using a non-quantitative recall of food groups consumed the day before the survey. A diet quality ques­tion­naire has been devel­oped for this purpose (Herforth et al., 2019).

The GDR Score (Table 8c.8) ranges from −9 to +9, based on sets of positively and negatively scored food groups. Each food group has equal weighting except for processed meat, which has a score of −2. Overall, the score gives equal weighting to “positive” and “negative” com­po­nents.

The WHO Healthy Diet guidance — one basis for the GDR Score — currently includes no recom­mendations on animal-source foods, though they can be good sources of bio­avail­able micro­nutri­ents. Processed meat and unprocessed red meat are included as negatively scored items to limit, based on World Cancer Research Fund / American Institute for Cancer Research recom­mendations. To capture nutri­ent adequacy in resource-constrained envi­ron­ments, the GDR Score can be considered alongside food group diversity scores such as the MDD-W, which are designed to reflect nutri­ent adequacy and include positive scoring for a variety of animal-source foods.

The GDR Score is a recent innovation and should be further assessed in a wider range of country contexts, with diverse cultural dietary patterns. Data to construct the GDR Score at national level will be available for 40 coun­tries from the 2021 Gallup World Poll. These data could provide baselines for future assess­ments of trends. Further details and updates on the GDR Score, including country-adapted ques­tion­naires for its measurement, are available from the Global Diet Quality Project website launched in 2021.

8c.10.4 Global Diet Quality Score

The Global Diet Quality Score (GDQS) represents another lower-burden approach to character­izing diet quality at popu­lation level, including both nutri­ent adequacy and NCD risk (Bromage et al, 2021). In contrast to the GDR Score, the GDQS is not based on adherence to existing / current global guidance. Its devel­op­ment was similar to that of the AHEI, in that food groups were selected for inclusion based on a review of literature demonstrating relation­ships of food group intakes to out­comes. Further analyses of multiple data sets, including cross-sectional and cohort data, led to refinement of the food groups and scoring.

The GDQS evolved from an earlier index initially devel­oped for use in the United States as a clinical screener (Rifas-Shiman et al., 2001). The global version was devel­oped for popu­lation-level use based on extensive analyses of data sets from Africa (several coun­tries), India, China, Mexico, and the US. Construct validity was assessed relative to nutri­ent adequacy, biomarkers for NCD risk, metabolic syndrome and incidence of type 2 diabetes (Bromage et al, 2021).

Data require­ments for the GDQS are inter­mediate in complexity and, as for the MDD‑W and the GDR Score, no food composition data is needed to construct the GDQS. Semi-quantitative data are needed to distinguish categories for quantity of intake for each of the 25 food group com­po­nents (Table 8c.9).

Table 8c.9 Global Diet Quality Score — Food Groups and Scoring Standards. Adapted from Bromage et al (2021).
^a GDQS = Global Diet Quality Score; GDQS− = Global Diet Quality Score Negative Sub-metric; GDQS+ = Global Diet Quality Score Positive Sub-metric.
^b Hard cheese should be converted to milk equivalents using a conversion factor when calculating total consumption of high-fat dairy for the purpose of assigning a GDQS consumption category.
^c Intakes of “High-fat dairy” > 734g/d are categorized as “Very High” and assigned zero points.

	Categories of con- sumed amounts (g/d)				Points assigned
	Low	Middle	High		Low	Middle	High
Included in GDQSa & GDQS+ metrics.   Healthy food groups
Citrus fruits	< 24	24–69	> 69		0	1	2
Deep orange fruits	< 25	25–123	> 123		0	1	2
Other fruits	< 27	27–107	> 107		0	1	2
Dark green leafy vegetables	< 13	13–37	> 37		0	2	4
Cruciferous vegetables	< 13	13–36	>36		0	0.25	0.5
Deep orange vegetables	< 9	9–45	> 45		0	0.25	0.5
Other vegetables	< 23	23–114	> 114		0	0.25	0.5
Legumes	< 9	9–42	> 42		0	2	4
Deep orange tubers	< 12	12–63	> 63		0	0.25	0.5
Nuts and seeds	< 7	7–13	> 13		0	2	4
Whole grains	< 8	8–13	> 13		0	1	2
Liquid oils	< 2	2–7.5	> 7.5		0	1	2
Fish and shellfish	< 14	14–71	> 71		0	1	2
Poultry and game meat	< 16	16–44	>44		0	1	2
Low-fat dairy	< 33	33–132	> 132		0	1	2
Eggs	< 6	6–32	> 32		0	1	2
Included in GDQS & GDQS− metrics.    Unhealthy in excessive amounts
High-fat dairyc (in milk equivalents)	< 35	35–142	> 142–734		0	1	2
Red meat	< 9	9–46	> 46		0	1	0
Included in GDQS & GDQS− metrics.    Unhealthy food groups
Processed meat	< 9	9–30	> 30		2	1	0
Refined grains and baked goods	< 7	7–33	> 33		2	1	0
Sweets and ice cream	< 13	13–37	>37		2	1	0
Sugar-sweetened beverages	< 57	57–180	>180		2	1	0
Juice	< 36	36–144	> 144		2	1	0
White roots and tubers	< 27	27–107	>107		2	1	0
Purchased deep fried foods	< 9	9–45	> 45		2	1	0

Two food groups (high-fat dairy and red meat), considered unhealthy when consumed in excessive amounts, have “U-shaped” scoring, such that both very low intakes and very high intakes are scored lower than inter­mediate intakes. Different from the other food groups in the index, which use 3 categories for quantity of intake, the high-fat dairy food group uses 4 categories to classify the quantity of intake for scoring. The “U-shaped” scoring for the high-fat dairy and red meat food groups reflects the role of these food groups both in meeting nutri­ent adequacy, but also in NCD risk when consumed in excess. Multiple weighting schemes were evaluated, and the selected com­po­nents are not equally weighted.

In addition to the overall score, ranging from 0 to 49, the GDQS also has two “sub-metrics” reflecting con­sump­tion of healthy and unhealthy com­po­nents (Table 8c.9). Further, two cutoffs have been identified to allow for reporting the percent of the popu­lation at high risk for poor diet quality out­comes (GDQS < 15) and the percent of the popu­lation at low risk for poor diet quality out­comes (GDQS ≥ 23).

Like the GDR Score, the GDQS is a recent innovation. Unlike the GDR Score, the semi-quantitative data required to construct the GDQS cannot be derived from non-quantitative food group recalls implemented to date in global survey programs such as the DHS, MICS and Gallup World Poll.

Lower-burden methods for the semi-quantitative recall required for the GDQS are currently under devel­op­ment and include a technology-assisted data collection tool and the use of visual aids (a set of 3D cubes of specified size) for collecting data to categorize the quantity of con­sump­tion per GDQS food group (Moursi et al, 2021).

Box 8c.11 summarizes key points about lower burden indices, designed for settings where detailed quantitative dietary intake data collection is infeasible.

8c.11 Incorporating sustain­ability in diet quality definitions and indices

Current food systems are now known to contribute a significant share of global greenhouse gas emissions and to contribute to land conversion, deforestation, and biodiversity loss; agriculture also accounts for the majority of global freshwater withdrawals (FAO & WHO, 2019). Recognition of this has led to efforts to define diets that are both healthy and envi­ron­mentally sustain­able, and to calls for integration of sustain­ability consider­ations in national food-based dietary guide­lines, as noted above (Gonzalez Fischer & Garnett, 2016; Springmann et al., 2020).

Table 8c.10 EAT-Lancet healthy refer­ence diet, for an intake of 2500kcals per day. Adapted from W. Willett et al. (2019)
^a Wheat, rice, dry beans and lentils are dry, raw.
^b Consists of fish and shellfish, including both wild and farmed.
^c Includes olive, soy bean, rapeseed, sunflower, and peanut oil.

Component	Macronutri­ent intake (possible range), g/d	Caloric intake, kcal/d
Whole grainsa
Rice, wheat, corn and other	232 (total grains 0–60% of energy)	811
Tubers or starchy vegetables
Potatoes and cassava	50 (0–100)	39
Vegetables
All vegetables	300 (200–600)
Dark green vegetables	100	23
Red and orange vegetables	100	30
Other vegetables	100	25
Fruits
All fruit	200 (100–300)	126
Dairy foods
Whole milk or derivative equivalents (e.g. cheese)	250 (0–500)	153
Protein sources
Beef and lamb	7 (0–14)	15
Pork	7(0–14)	15
Chicken and other poultry	29 (0–58)	62
Eggs	13 (0–25)	19
Fishb	28 (0–100)	40
Dry beans, lentils, and peasa	50 (0–100)	172
Soy foods	25 (0–50)	112
Peanuts	25 (0–75)	142
Tree nuts	25	149
Added fats
Palm oil	6.8 (0–6.8)	60
Unsaturated oilsc	40 (20–80)	354
Lard or tallow	5 (0–5)	36
Added sugars
All sweeteners	31 (0–31)	120

In 2019, the EAT-Lancet Commission (Willet et al., 2019) addressed sustain­ability concerns and provided a global “healthy refer­ence diet”, with quantitative targets (and ranges) for food group intake, based on a 2500 calorie diet (Table 8c.10). The refer­ence diet was intended to be flexible and adaptable to various cultural contexts. The EAT-Lancet refer­ence diet is high in plant-source foods and with limited amounts of animal-source foods, with the low end of the “healthy” range for these foods set at zero. Currently, there is no established global guidance on the appropriate balance of plant-source and animal-source foods.

Some have raised concerns that the EAT-Lancet refer­ence diet unneces­sarily restricts nutri­ent-dense animal-source foods (Raiten et al., 2020). Vaidyanathan (2021) summarizes some of the controversies and concerns. For nutrition-insecure women in low- and middle-income coun­tries, Hanley-Cook et al. (2021) demonstrated that micro­nutri­ent adequacy improved when the EAT-Lancet intake ranges were modified by imposing non-zero minimum quan­tities for nutri­ent-dense animal-source food groups.

Meantime, several groups have proposed indices for healthy, sustain­able diets either based on national guidance (Harray et al., 2015) or based on the EAT-Lancet refer­ence diet (Knuppel et al., 2019; Trijsburg et al., 2019). Stubbendorff et al., 2021). Given this is a new area of inquiry, it is likely that additional research on sustain­able healthy diets will yield additional indices and measurement tools in coming years.

8c.12 Summary of diet quality indices

This section has presented a wide variety of diet quality indices developed for diverse uses. Indices are closely related to definitions of diet quality and have evolved with changing definitions. At national level, food-based dietary guidelines describe healthy diets, and indices developed to capture adherence to such guidelines are widely used. Currently, many indices — whether based on national, global, or other guidance — aim to reflect both nutrient adequacy and NCD risk reduction. Concerns with the impact of diets on planetary health are growing, and this is likely to result in new and more comprehensive indices in the near future.

Indices also vary widely in the data required for their calculation. Most indices developed for use in high-income data-rich environments require both detailed quantitative dietary intake data, and associated resources such as food composition databases. However, indices with lower data requirements have recently been proposed, for global use.

Acknowledgments

The authors are very grateful to Dr. Sharon Kirkpatrick for her review of an earlier draft of this section, and to Michael Jory for the HTML design and his tireless work in directing the trans­ition to this HTML version. The authors are responsible for errors in the text.