Non-communicable diseases (NCDs), such as cardiovascular diseases, type 2 diabetes, cancer, respiratory diseases, and other inflammatory pathologies, represent approximately 70% of deaths worldwide [1
]. This group of diseases shares common risk factors, namely unhealthy diets, sedentariness and low physical activity, excess consumption of alcohol, and the use of tobacco products, all of which are risk factors for obesity, hypertension, and lipid and glucose metabolism impairment that lead to the establishment of chronic diseases [2
As diet is one of the critical modifiable behaviors that can help control cardiometabolic risks and prevent chronic diseases, it is essential to evaluate the overall diet composition of the population [3
]. Diet quality indexes (DQIs) are tools that aim to evaluate the quality of diets and categorize individuals according to the extent to which their eating behavior is “healthy” [5
Currently, there is a large number of DQIs, most of which are designed or adapted to reflect the nutritional needs of different population groups and adherence to specific dietary patterns [5
]. DQIs are comprised of various foods and/or nutrients to assess overall diet quality [7
]. There are three major categories of DQIs: (a) Nutrient-based indicators, (b) food/food group-based indicators, and (c) combination indexes. Thus, each DQI has different foods and nutrients components, cut-offs, and scoring approaches, which limits comparability. Moreover, most DQIs have been developed for specific populations. Indeed, even though DQIs can be used in similar groups of people, in most cases, their use is limited to populations different from those for which they were specifically designed. Therefore, it is challenging to compare the results among studies that use different DQIs.
The Healthy Diet Indicator (HDI) and the Mediterranean Diet Score (MDS) are within the four “original” diet quality scores that have been referred to and validated most extensively [5
]. Additionally, several DQIs have been adapted and modified from those originals. In particular, many variations on the MDS have been proposed, and the so-called “MDS modified” (MDS-mod) and the Mediterranean-Diet Quality Index (MED-DQI) have been widely used among Mediterranean populations. The latter has been reported to be one of the most adequate scores to evaluate the Mediterranean diet in adults [5
]. However, these scores do not necessarily capture what is needed for European Union (EU) populations, as in 2017, the European Food Safety Authority (EFSA) published the dietary reference values for nutrients for the EU [9
]. Thus, a new DQI considering the EFSA nutrient recommendations would be useful in EU nutrition surveys.
DQIs have been reported to be influenced by socioeconomic determinants, mainly education and income. [10
]. However, the majority of studies do not control for these covariates, which limits comparability. Additionally, most epidemiologic data are collected through surveys, so self-report biases and misreporting are possible [11
]. For this reason, the EFSA has published a protocol that has a harmonized approach to identify misreporting [12
]. Following this protocol, we previously reported the intake of energy, some minerals, and vitamins for both the Spanish “Anthropometry, Intake, and Energy Balance Study” (ANIBES) in the general population and plausible reporters [13
Considering the above, the primary aim of the present study was to evaluate the adequacy of critical nutrients that directly affect the quality of the diet of the Spanish ANIBES population, stratified by sex and different age groups, using the EFSA values for nutrients for the EU as a reference. Second, we assessed the quality of the diet for adults and older adults using HDI, MDS, MDS-mod, and MED-DQI, as well as a new score, ANIBES-DQI, stratified by education and income. The ANIBES-DQI is based on compliance with EFSA recommendations for a selected group of nutrients (i.e., total fat, saturated fatty acids (SFAs), simple sugars, fiber, calcium, vitamin C, and vitamin A (total range 0–7)). One additional aim was to assess the above mentioned DQIs for both the general ANIBES population and plausible reporters.
In the present study, we evaluated the adequacy of the intake of critical nutrients related to the quality of the diet for the ANIBES study population using the EFSA dietary reference values for nutrients [9
]. In addition, we assessed the quality of the diet through internationally recognized DQIs: one general index (HDI) and three related to adherence to the Mediterranean diet (MDS, MDS-mod, and MED-DQI), as the Spanish population should generally follow a Mediterranean lifestyle pattern. Likewise, we built a novel index, the ANIBES-DQI, based on the adequacy of compliance to the current WHO/FAO guidelines for total fat and fatty acids [19
], the WHO guidelines for sugar intake [20
], and the EFSA recommended intakes for fiber, vitamin C, vitamin A, zinc, and calcium [9
]. We assessed all DQIs for both a general Spanish population and plausible reporters.
The Mediterranean diet encompasses the traditional dietary patterns found in the olive-growing regions of the Mediterranean basin in the 1960s. It is globally recognized as a healthy dietary model and an intangible cultural heritage of humanity by the United Nations Educational, Scientific and Cultural Organization (UNESCO) [32
]. Food consumption patterns in Spain and energy and nutrient intakes have changed markedly in the last forty years, differing somewhat at present from the traditional and healthy Mediterranean Diet. This change is partly due to the westernization of eating habits [33
The detailed analysis of the ANIBES population, with regard to the recommended intakes for the nutrients included in the ANIBES-DQI, shows that, except for protein, PUFA, and iron, the Spanish population have higher intakes of SFA, simple sugars, and low intakes of fiber, calcium, zinc, folate, and vitamins A and C than recommended. Overall, older adults showed better DQIs than adults, with no major differences found between men and women, although the percentages of women with a low-quality score were slightly lower than those for men. Regarding the association of DQIs with education and income, in general, education level had a strong influence on the quality of the diet, although HDI did not discriminate between different levels of education. Globally, lower education levels were related to lower quality scores for the Mediterranean diet quality indexes, regardless of the analyzed DQI. Nonetheless, using the ANIBES score, we observed a lower score for those with low income.
When comparing the DQIs for the general population and plausible reporters, we observed similar percentages for low, medium, and high scores, except for the ANIBES-DQI, for which the percentages of the low-quality score were higher, and those with higher quality scores were lower in plausible reporters. The differences found in the percentages of the population with low quality scores for the analyzed DQIs could be due (at least in part) to the special particularities of the Spanish ANIBES population. In fact, most of the DQIs have been designed for specific populations [5
]. Indeed, the MDS, MDS-mod, and MED, which measure the adherence to the Mediterranean diet, exhibited lower percentages of low-quality scores compared to the indexes potentially applicable to general populations, such as HDI and ANIBES-DQI. Some authors have reported that certain DQIs built for specific populations lead to lower quality indexes when adapted to others with different food patterns [5
]. We also observed that a higher percentage of the population (approximately 70%) had a low-quality diet when using the HDI, compared to approximately 60% for the ANIBES-DQI. This difference might be attributable to the fact that HDI is a mixed quality diet index based on the consumption of specific groups of foods and the adequacy of the intake of selected nutrients, whereas the ANIBES-DQI takes into account, exclusively, the adequacy of the intake of a number of critical nutrients compared to EU reference nutrient intakes. Therefore, it seems crucial when using DQIs to consider their specific purposes, for example, whether they are used for the evaluation of dietary intakes and food patterns, the development of dietary guidelines for specific populations, or the prevention of NCDs. In fact, we have previously pointed out that there is no standardized approach to the components of DQIs and scoring, as the latter is based on food frequency, number of portions, assigned food weights, compliance with nutrient recommendations, etc. It is difficult to compare DQI scores, which are often country-specific [5
]. Nevertheless, in the present study, we tried to compare whether internationally recognized DQIs exhibited a similar behavior for the Spanish population.
When HDI, MDS, MDS-mod, and MED were compared with the ANIBES-DQI, we observed that the latter, built on nutrients exclusively, was more strict when scoring diet quality than those based on nutrients and foods. Whatever the case, it seems that the inclusion of the consumption of foods in DQIs beyond the selected critical nutrients contributes to improving the results of the quality assessment of diet in most of populations [5
]. In these cases, it is very important to select and validate an appropriate method for the evaluation of food consumption (e.g., a quantitative FFQ), as well as to use a data base of quality for the composition of foods.
Recently, the memory-based dietary assessment methods utilized in epidemiological research related to food group consumption and major events of disease have been challenged due to the varied precision and accuracy of self-reported data [36
]. Archer et al. [11
] empirically refuted memory-based dietary assessment methods (M-BM), such as food records, food frequency questionnaires (FFQs), and 24HR, arguing that the errors associated with M-BM-data are unquantifiable, as they are prone to omissions, false memories, intentional misreporting (i.e., lying), and gross misestimations. In particular, FFQs are prone to measurement error [39
]. However, other investigators strongly disagree with the assertions made by those authors regarding the validity and usefulness of FFQs and other M-BM in assessing diet–disease relationships in epidemiologic studies. Indeed, Martin Calvo and Martinez-Gonzalez [40
] emphasized that the growing evidence regarding diet–disease relationships found in observational studies based on M-BM is sufficiently reliable to be used for public health policies. In addition, well-controlled prospective studies using objective biomarkers of intake have confirmed the results of previous studies using self-reported dietary assessment methods. Regardless of this debate, it seems that to compare diverse DQIs for different populations, the methods of dietary assessment and databases for food composition would need to be harmonized. We do not think one unique central data database should be chosen, but instead that each data base for particular countries should be harmonized and based on the same methodology to help reduce the variability. In that respect, the EFSA has given specific guidelines to facilitate the collection of food consumption data from all EU Member States [16
In the ANIBES study, diet was evaluated throughout a three-day dietary record using a tablet device and digital cameras to collect the information, accompanied by a 24HR [14
]. Moreover, a group of experts supervised the trained personnel to minimize all the possible issues that could potentially appear during the data collection process. However, we are aware that there are aspects that we cannot control, such as the accuracy of memory when recording the past eating and drinking behaviors of the volunteers, which are subject to intentional or unintentional distortion factors, the will of individuals to be truthful in their responses, or the fact that the participants want to be seen as having good and healthy eating habits. However, as far as we know, to date, there are no alternative standardized methods to avoid those issues in nutrition epidemiological studies. Furthermore, as mentioned earlier, we followed the guidance on the EU Menu methodology, as published by the EFSA. This guidance is compulsory for new nutrition epidemiological studies carried out by EU members from 2014 onwards [16
]. In addition, we addressed misreporting via a standardized EFSA protocol [12
], whereas most epidemiologic studies either ignore misreporting entirely or use inappropriate cut-offs. Therefore, the ANIBES study stands well above many other previous studies on diet quality.
One main observation in the present study is that older adults had a better diet quality than adults, which was consistently seen when using not only HDI and ANIBES-DQI but also Mediterranean scores. This finding agrees with other reports in Italy [41
], Greece, and Cyprus [43
]. In addition, in other countries like Canada and the USA, using general DQIs, older adults exhibited better diet quality scores than their younger counterparts [8
]. Thus, North American older adults have higher diet quality scores than adults and youth. However, other factors involved in the socioeconomic status, which may affect the diet quality of the older adults, should be considered. The quality of diet has been related to the socioeconomic status in different populations throughout the world [44
], which includes many variables, such as education, income, type of employment, and some characteristics of the particular areas where the populations live. However, there is no unanimity as to how education or the income levels affect diet quality. In the present study, both low education and low income were directly associated with some DQIs, namely the MDS and ANIBES-DQI. Similarly, in a French population, low education was associated with a low-quality diet, characterized by lower intakes of fiber, minerals, and vitamins, although the authors identified very complex interactions among education, income, and occupation (e.g., they described interactions between education and income level) [45
There is a need for more studies to evaluate the relationships between diet quality and socioeconomic status at the domestic level, as diet quality may be influenced by other factors (e.g., the target population, unemployment, the occupation of different members of the family, access to food, urbanization in countries with low or high gross domestic products) [48
]. In the ANIBES study, unemployed subjects were included, but we did not gather information about household sizes or single living arrangements. Indeed, we cannot determine the influence of these variables on diet quality.
The percentages of plausible reporters are essential to determine diet quality for a particular population. In the present study, we found differences between the whole population and the plausible reporters only for the ANIBES-DQI but not for the other DQIs. The influence of misreporting (over- and underreporting) on diet quality has been scarcely considered. Lutomski et al. (2009) observed that under-reporters had a lower diet quality (evaluated by Dietary Approaches to Stop Hypertension (DASH) diet) with a higher nutrient density and a lower percentage of energy derived from fat, whereas over-reporters showed an opposite pattern [48
]. The ANIBES study is the only work that has analyzed misreporting in a Spanish population and considered to what extent misreporting could affect diet quality. Thus, it seems that underreporting contributes to increasing the “apparent” percentage of the general population with a low-quality diet. Other factors, including sex, BMI, and education, may also affect the percentages of over and under reporters. However, this difference was not addressed in the present study.
The present study has a number of strengths. Basically, we followed a standardized methodology for the study of nutrient intakes, as well as for the evaluation of misreporting based on EFSA guidelines [9
]. In addition, to study diet quality, we applied selected DQIs that have been validated for global populations, as well as for countries following the Mediterranean pattern [5
]. Moreover, we provided the ANIBES-DQI to evaluate the quality of the diet using cut-off values related to the ARs for the EU. However, this study also has limitations. The first limitation is derived of the use of 24HR, which can contribute to unquantifiable errors, as detailed above, although the use of the CAPI/CATI EFSA methodology would minimize these potential errors [16
]. Additional limitations are that there is no gold standard DQI to compare against other indexes. Likewise, it is difficult to compare diet qualities among EU countries from different surveys because of random and systemic errors of diet assessments, due to chronological changes, response rate and linguistic and cultural diversity. Moreover, as previously mentioned, we only stratified DQIs by sex, education, and income, but we did not evaluate other potential factors (e.g., the body mass index of individuals, family size, type of job, unemployment, urbanization, living in rural areas) that could act as effect modifiers.