The combined sandwich-ELISA (s-ELISA; VitMin Lab, Germany) and the Quansys Q-Plex™ Human Micronutrient Array (7-Plex) are multiplex serum assays that are used to assess population micronutrient status in low-income countries. We aimed to compare the agreement of five analytes, α-1-acid glycoprotein (AGP), C-reactive protein (CRP), ferritin, retinol-binding protein 4 (RBP4) and soluble transferrin receptor (sTfR) as measured by the 7-Plex and the s-ELISA. Serum samples were collected between March 2016 and December 2017. Pregnant women (n 249) were recruited at primary healthcare clinics in Johannesburg, and serum samples were collected between March 2016 and December 2017. Agreement between continuous measurements was assessed by Bland–Altman plots and concordance measures. Agreement in classifications of deficiency or inflammation was assessed by Cohen’s kappa. Strong correlations (r > 0·80) were observed between the 7-Plex and s-ELISA for CRP and ferritin. Except for CRP, the 7-Plex assay gave consistently higher measurements than the s-ELISA. With the exception of CRP (Lin’s ρ = 0·92), there was poor agreement between the two assays, with Lin’s ρ < 0·90. Discrepancies of test results difference between methods increased as the serum concentrations rose. Cohen’s kappa for all the five analytes was < 0·81 and ranged from slight agreement (vitamin A deficiency) to substantial (inflammation and Fe deficiency) agreement. The 7-Plex 1.0 is a research and or surveillance tool with potential for use in low-resource laboratories but cannot be used interchangeably with the s-ELISA. Further optimising and validation is required to establish its interchangeability with other validated methods.

With the recent growing interest in improving fruit and vegetable intake for better health and limited research resources in many settings, simple-to-administer and low-priced indicators are essential tools for monitoring fruit and vegetable intake at the population level. A potential candidate indicator is the fruit and vegetable component of the Global Dietary Recommendation score (FV-GDR) based on data collected using the Diet Quality Questionnaire (DQQ). We investigated the relative validity of FV-GDR collected with the DQQ to measure fruit and vegetable intake by comparison with a 24-h recall (24hR) as a reference collected from 620 Vietnamese and 630 Nigerian adults in 2021. We found proportional differences in the prevalence of intake of vitamin A-rich vegetables, other vegetables and other fruits in Vietnam and all vegetable food groups in Nigeria. In both countries, we found a small difference in the total FV-GDR from DQQ compared with the 24hR, and the percentage of agreement between the two methods was quite high for the majority of the food groups. The FV-GDR calculated from the DQQ correlated with the actual intake, although less strongly than the FV-GDR from 24hR. The DQQ is a promising low-burden, low-cost and simple tool to calculate FV-GDR and to monitor fruit and vegetable consumption at the population level. This provides the possibility of evaluating an important aspect of diet quality in low-resource settings.

Ethiopia announced its first food-based dietary guidelines (FBDGs) on 15 March 2022. The present study aims to develop and evaluate the Ethiopian Healthy Eating Index (Et-HEI) based on the FBDG. Data were collected from 494 Ethiopian women of reproductive age sampled from households in five different regions. The Et-HEI consists of eleven components, and each component was scored between 0 and 10 points, the total score ranging from 0 to 110, with maximum adherence to the FBDG. The Et-HEI score was evaluated against the Minimum Dietary Diversity for Women (MDD-W) and the probability of nutrient adequacy. The average Et-HEI score for women of reproductive age was 49 out of 110. Adherence to the recommendations for grains, vegetables, legumes, fat and oils, salt, sugar and alcohol contributed the most to this score. Most women had low scores for fruits, nuts and seeds, and animal-sourced foods, indicating low intake. The Cronbach's alpha coefficient, indicating the reliability of the Et-HEI to assess its diet quality, was 0⋅53. The low mean Et-HEI score agreed with a low mean score of the MDD-W (3⋅5 out of 10). Also, low nutrient adequacies confirmed poor adherence to nutrient-dense components of the FBDG. The Et-HEI was not associated with the intake of vitamin B12, vitamin C and calcium in this study population. Women who completed secondary school and above had relatively lower Et-HEI scores. The newly developed Et-HEI is able to estimate nutrient adequacy while also assessing adherence to the Ethiopian FBDG though there is room for improvement.

Poor dietary quality is a major contributor to malnutrition and disease burden in Vietnam, necessitating the development of a tool for improving dietary quality. Food-based dietary guidelines (FBDGs) have been proposed to do this by providing specific, culturally appropriate and actionable recommendations. We developed the Vietnamese Healthy Eating Index (VHEI) to assess the adherence to the 2016–2020 Vietnamese FBDGs and the dietary quality of the general Vietnamese population. This VHEI consists of eight component scores, ‘grains’, ‘protein foods’, ‘vegetables’, ‘fruits’, ‘dairy’, ‘fats and oils’, ‘sugar and sweets’ and ‘salt and sauces’, representing the recommendations in the FBDGs. Each component score ranges from 0 to 10, resulting in a total VHEI score between 0 (lowest adherence) and 80 (highest adherence). The VHEI was calculated using dietary intake data from the Vietnamese General Nutrition Survey 2009–2010 (n = 8225 households). Associations of the VHEI with socio-demographic characteristics, energy and nutrient intakes and food group consumptions were examined. The results showed that the mean and standard deviation score of the VHEI was 43⋅3 ± 8⋅1. The component ‘sugar and sweets’ scored the highest (9⋅8 ± 1⋅1), whereas the component ‘dairy’ scored the lowest (0⋅6 ± 1⋅6). The intake of micronutrients was positively associated with the total VHEI, both before and after adjustment for energy intake. In conclusion, the VHEI is a valuable measure of dietary quality for the Vietnamese population regarding their adherence to the FBDGs.

The Eetscore FFQ was developed to score the Dutch Healthy Diet index 2015 (DHD2015-index) representing the Dutch food-based dietary guidelines of 2015. This paper describes the development of the Eetscore FFQ, a short screener assessing diet quality, examines associations between diet quality and participants’ characteristics, and evaluates the relative validity and reproducibility of the Eetscore FFQ in a cross-sectional study with Dutch adults. The study sample consisted of 751 participants, aged 19–91 years, recruited from the EetMeetWeet research panel. The mean DHD2015-index score based on the Eetscore FFQ of the total sample was 111 (sd 17·5) out of a maximum score of 160 points and was significantly higher in women than in men, positively associated with age and education level, and inversely associated with BMI. The Kendall’s tau-b coefficient of the DHD2015-index between the Eetscore FFQ and the full-length FFQ (on average 1·7-month interval, n 565) was 0·51 (95 % CI 0·47, 0·55), indicating an acceptable ranking ability. The intraclass correlation coefficient between DHD2015-index scores derived from two repeated Eetscore FFQ (on average 3·8-month interval, n 343) was 0·91 (95 % CI 0·89, 0·93) suggesting a very good reproducibility. In conclusion, the Eetscore FFQ was considered acceptable in ranking participants according to their diet quality compared with the full-length FFQ and showed good to excellent reproducibility.

Our current society is characterized by an increased availability of industrially processed foods with high salt, fat and sugar content. How is it that some people prefer these unhealthy foods while others prefer more healthy foods? It is suggested that both genetic and environmental factors play a role. The aim of this study was to (1) identify food preference clusters in the largest twin-family study into food preference to date and (2) determine the relative contribution of genetic and environmental factors to individual differences in food preference in the Netherlands. Principal component analysis was performed to identify the preference clusters by using data on food liking/disliking from 16,541 adult multiples and their family members. To estimate the heritability of food preference, the data of 7833 twins were used in structural equation models. We identified seven food preference clusters (Meat, Fish, Fruits, Vegetables, Savory snacks, Sweet snacks and Spices) and one cluster with Drinks. Broad-sense heritability (additive [A] + dominant [D] genetic factors) for these clusters varied between .36 and .60. Dominant genetic effects were found for the clusters Fruit, Fish (males only) and Spices. Quantitative sex differences were found for Meat, Fish and Savory snacks and Drinks. To conclude, our study convincingly showed that genetic factors play a significant role in food preference. A next important step is to identify these genes because genetic vulnerability for food preference is expected to be linked to actual food consumption and different diet-related disorders.

About 57 % of the pregnant European women have 25-hydroxyvitamin D (25(OH)D) concentrations below 50 nmol/l. However, as data on the impact of gestational vitamin D deficiency on maternal and fetal health are limited, the WHO does not advocate vitamin D supplementation as part of routine antenatal care. We explored associations between first trimester maternal 25(OH)D status and childhood cognition at 5–6 years of age (n 1854, primarily Caucasian). Median serum 25(OH)D was determined at 13 (interquartile range 12–14) weeks of gestation. Childhood attention, motor fluency and flexibility and executive function were assessed using the Amsterdam Neuropsychological Tasks. Restricted cubic splines and linear regression analyses were used to analyse the data while adjusting for many maternal and child related covariates. Higher 25(OH)D status (nmol/l) was associated with better attention and executive functioning as shown by a faster reaction time (β −0·30 (sd 0·14) ms, P=0·03), faster response speed (β −0·58 (sd 0·21) ms, P=0·006), and better response speed stability (β −0·45 (sd 0·17) ms, P=0·009). No associations were observed of serum 25(OH)D with motor fluency and flexibility. Associations were most pronounced among children of African origin (n 205) as compared with those of Caucasian or another origin, for example attention (reaction time, β −2·06 (sd 0·70) ms, P=0·004) and executive function (response speed, β −1·95 (sd 0·94) ms, P=0·04). Concluding, maternal 25(OH)D status was significantly associated with childhood attention and executive function, while no associations were observed for 25(OH)D status with motor fluency and flexibility.

Carbohydrate quantity and quality affect postprandial glucose response, glucose metabolism and risk of type 2 diabetes. The aim of this study was to examine the association of pre-pregnancy dietary carbohydrate quantity and quality with the risk of developing gestational diabetes mellitus (GDM). We used data from the Australian Longitudinal Study on Women’s Health that included 3607 women aged 25–30 years without diabetes who were followed up between 2003 and 2015. We examined carbohydrate quantity (total carbohydrate intake and a low-carbohydrate diet (LCD) score) and carbohydrate subtypes indicating quality (fibre, total sugar intake, glycaemic index, glycaemic load and intake of carbohydrate-rich food groups). Relative risks (RR) for development of GDM were estimated using multivariable regression models with generalised estimating equations. During 12 years of follow-up, 285 cases of GDM were documented in 6263 pregnancies (4·6 %). The LCD score, reflecting relatively high fat and protein intake and low carbohydrate intake, was positively associated with GDM risk (RR 1·54; 95 % CI 1·10, 2·15), highest quartile v. lowest quartile). Women in the quartile with highest fibre intake had a 33 % lower risk of GDM (RR 0·67; 95 % CI 0·45, 0·96)). Higher intakes of fruit (0·95 per 50 g/d; 95 % CI 0·90, 0·99) and fruit juice (0·89 per 100 g/d; 95 % CI 0·80, 1·00)) were inversely associated with GDM, whereas cereal intake was associated with a higher risk of GDM (RR 1·05 per 20 g/d; 95 % CI 1·01, 1·07)). Thus, a relatively low carbohydrate and high fat and protein intake may increase the risk of GDM, whereas higher fibre intake could decrease the risk of GDM. It is especially important to take the source of carbohydrates into account.

Taste is a key driver of food choice and intake. Taste preferences are widely studied, unlike the diet’s taste profile. This study assessed dietary taste patterns in the Netherlands by sex, BMI, age and education. A taste database, containing 476 foods’ taste values, was combined with 2-d 24-h recalls in two study populations. The percentage of energy intake from six taste clusters was assessed in the Dutch National Food Consumption Survey (DNFCS 2007–2010; n 1351) and in an independent observational study: the Nutrition Questionnaires plus (NQplus) study (2011–2013; n 944). Dietary taste patterns were similar across study populations. Men consumed relatively more energy from ‘salt, umami and fat’ (DNFCS; 24 % energy, NQplus study; 23 %)- and ‘bitter’ (7 %)-tasting foods compared with women (21 %, P<0·001, 22 %, P=0·005; 3 %, P<0·001, 4 %, P<0·001, respectively). Women consumed more % energy from ‘sweet and fat’ (15 %)- and ‘sweet and sour’ (13 %, 12 %, respectively)-tasting foods compared with men (12 %, P<0·001, 13 %, P=0·001; 10 %, P<0·001). Obese individuals consumed more % energy from ‘salt, umami and fat’- and less from ‘sweet and fat’-tasting foods than normal-weight individuals (‘salt, umami and fat’, men; obese both studies 26 %, normal-weight DNFCS 23 %, P=0·037, NQplus 22 %, P=0·001, women; obese 23 %, 24 %, normal weight 20 %, P=0·004, P=0·011, respectively, ‘sweet and fat’, men; obese 11 %, 10 %, normal weight 13 %, P<0·05, 14 %, P<0·01, women; obese 14 %, 15 %, normal weight 16 %, P=0·12, P=0·99). In conclusion, our taste database can be used to deepen our understanding of the role of taste in dietary intake in the Netherlands by sex, BMI, age and education.

Previous studies show associations between dairy product consumption and type 2 diabetes, but only a few studies conducted detailed analyses for a variety of dairy subgroups. Therefore, we examined cross-sectional associations of a broad variety of dairy subgroups with pre-diabetes and newly diagnosed type 2 diabetes (ND-T2DM) among Dutch adults. In total, 112 086 adults without diabetes completed a semi-quantitative FFQ and donated blood. Pre-diabetes was defined as fasting plasma glucose (FPG) between 5·6 and 6·9 mmol/l or HbA1c% of 5·7–6·4 %. ND-T2DM was defined as FPG ≥7·0 mmol/l or HbA1c ≥6·5 %. Logistic regression analyses were conducted by 100 g or serving increase and dairy tertiles (T1ref), while adjusting for demographic, lifestyle and dietary covariates. Median dairy product intake was 324 (interquartile range 227) g/d; 25 549 (23 %) participants had pre-diabetes; and 1305 (1 %) had ND-T2DM. After full adjustment, inverse associations were observed of skimmed dairy (OR100 g 0·98; 95 % CI 0·97, 1·00), fermented dairy (OR100 g 0·98; 95 % CI 0·97, 0·99) and buttermilk (OR150 g 0·97; 95 % CI 0·94, 1·00) with pre-diabetes. Positive associations were observed for full-fat dairy (OR100 g 1·003; 95 % CI 1·01, 1·06), non-fermented dairy products (OR100 g 1·01; 95 % CI 1·00, 1·02) and custard (ORserving/150 g 1·13; 95 % CI 1·03, 1·24) with pre-diabetes. Moreover, full-fat dairy products (ORT3 1·16; 95 % CI 0·99, 1·35), non-fermented dairy products (OR100 g 1·05; 95 % CI 1·01, 1·09) and milk (ORserving/150 g 1·08; 95 % CI 1·02, 1·15) were positively associated with ND-T2DM. In conclusion, our data showed inverse associations of skimmed and fermented dairy products with pre-diabetes. Positive associations were observed for full-fat and non-fermented dairy products with pre-diabetes and ND-T2DM.

FFQ, food diaries and 24 h recall methods represent the most commonly used dietary assessment tools in human studies on nutrition and health, but food intake biomarkers are assumed to provide a more objective reflection of intake. Unfortunately, very few of these biomarkers are sufficiently validated. This review provides an overview of food intake biomarker research and highlights present research efforts of the Joint Programming Initiative ‘A Healthy Diet for a Healthy Life’ (JPI-HDHL) Food Biomarkers Alliance (FoodBAll). In order to identify novel food intake biomarkers, the focus is on new food metabolomics techniques that allow the quantification of up to thousands of metabolites simultaneously, which may be applied in intervention and observational studies. As biomarkers are often influenced by various other factors than the food under investigation, FoodBAll developed a food intake biomarker quality and validity score aiming to assist the systematic evaluation of novel biomarkers. Moreover, to evaluate the applicability of nutritional biomarkers, studies are presently also focusing on associations between food intake biomarkers and diet-related disease risk. In order to be successful in these metabolomics studies, knowledge about available electronic metabolomics resources is necessary and further developments of these resources are essential. Ultimately, present efforts in this research area aim to advance quality control of traditional dietary assessment methods, advance compliance evaluation in nutritional intervention studies, and increase the significance of observational studies by investigating associations between nutrition and health.

Self-administered web-based 24-h dietary recalls (24 hR) may save a lot of time and money as compared with interviewer-administered telephone-based 24 hR interviews and may therefore be useful in large-scale studies. Within the Nutrition Questionnaires plus (NQplus) study, the web-based 24 hR tool Compl-eat™ was developed to assess Dutch participants’ dietary intake. The aim of the present study was to evaluate the performance of this tool against the interviewer-administered telephone-based 24 hR method. A subgroup of participants of the NQplus study (20–70 years, n 514) completed three self-administered web-based 24 hR and three telephone 24 hR interviews administered by a dietitian over a 1-year period. Compl-eat™ as well as the dietitians guided the participants to report all foods consumed the previous day. Compl-eat™ on average underestimated the intake of energy by 8 %, of macronutrients by 10 % and of micronutrients by 13 % as compared with telephone recalls. The agreement between both methods, estimated using Lin's concordance coefficients (LCC), ranged from 0·15 for vitamin B1 to 0·70 for alcohol intake (mean LCC 0·38). The lower estimations by Compl-eat™ can be explained by a lower number of total reported foods and lower estimated intakes of the food groups, fats, oils and savoury sauces, sugar and confectionery, dairy and cheese. The performance of the tool may be improved by, for example, adding an option to automatically select frequently used foods and including more recall cues. We conclude that Compl-eat™ may be a useful tool in large-scale Dutch studies after suggested improvements have been implemented and evaluated.

A standardised, national, 160-item FFQ, the FFQ-NL 1.0, was recently developed for Dutch epidemiological studies. The objective was to validate the FFQ-NL 1.0 against multiple 24-h recalls (24hR) and recovery and concentration biomarkers. The FFQ-NL 1.0 was filled out by 383 participants (25–69 years) from the Nutrition Questionnaires plus study. For each participant, one to two urinary and blood samples and one to five (mean 2·7) telephone-based 24hR were available. Group-level bias, correlation coefficients, attenuation factors, de-attenuated correlation coefficients and ranking agreement were assessed. Compared with the 24hR, the FFQ-NL 1.0 estimated the intake of energy and macronutrients well. However, it underestimated intakes of SFA and trans-fatty acids and alcohol and overestimated intakes of most vitamins by >5 %. The median correlation coefficient was 0·39 for energy and macronutrients, 0·30 for micronutrients and 0·30 for food groups. The FFQ underestimated protein intake by an average of 16 % and K by 5 %, relative to their urinary recovery biomarkers. Attenuation factors were 0·44 and 0·46 for protein and K, respectively. Correlation coefficients were 0·43–0·47 between (fatty) fish intake and plasma EPA and DHA and 0·24–0·43 between fruit and vegetable intakes and plasma carotenoids. In conclusion, the overall validity of the newly developed FFQ-NL 1.0 was acceptable to good. The FFQ-NL 1.0 is well suited for future use within Dutch cohort studies among adults.

Diets high in glycaemic index (GI) and glycaemic load (GL) have been associated with a higher diabetes risk. Beer explained a large proportion of variation in GI in a Finnish and an American study. However, few beers have been tested according to International Organization for Standardization (ISO) methodology. We tested the GI of beer and estimated its contribution to dietary GI and GL in the Netherlands. GI testing of pilsner beer (Pilsner Urquell) was conducted at The University of Sydney according to ISO international standards with glucose as the reference food. Subsequently, GI and GL values were assigned to 2556 food items in the 2011 Dutch food composition table using a six-step methodology and consulting four databases. This table was linked to dietary data from 2106 adults in the Dutch National Food Consumption Survey 2007–2010. Stepwise linear regression identified contribution to inter-individual variation in dietary GI and GL. The GI of pilsner beer was 89 (sd 5). Beer consumption contributed to 9·6 and 5·3 % inter-individual variation in GI and GL, respectively. Other foods that contributed to the inter-individual variation in GI and GL included potatoes, bread, soft drinks, sugar, candy, wine, coffee and tea. The results were more pronounced in men than in women. In conclusion, beer is a high-GI food. Despite its relatively low carbohydrate content (approximately 4–5 g/100 ml), it still made a contribution to dietary GL, especially in men. Next to potatoes, bread, sugar and sugar-sweetened beverages, beer captured a considerable proportion of between-person variability in GI and GL in the Dutch diet.

Generally, there is a need for short questionnaires to estimate diet quality in the Netherlands. We developed a thirty-four-item FFQ – the Dutch Healthy Diet FFQ (DHD-FFQ) – to estimate adherence to the most recent Dutch guidelines for a healthy diet of 2006 using the DHD-index. The objectives of the present study were to evaluate the DHD-index derived from the DHD-FFQ by comparing it with the index based on a reference method and to examine associations with participant characteristics, nutrient intakes and levels of cardiometabolic risk factors. Data of 1235 Dutch men and women, aged between 20 and 70 years, participating in the Nutrition Questionnaires plus study were used. The DHD-index was calculated from the DHD-FFQ and from a reference method consisting of a 180-item FFQ combined with a 24-h urinary Na excretion value. Ranking was studied using Spearman’s correlations, and absolute agreement was studied using a Bland–Altman plot. Nutrient intakes derived from the 180-item FFQ were studied according to quintiles of the DHD-index using DHD-FFQ data. The correlation between the DHD-index derived from the DHD-FFQ and the reference method was 0·56 (95 % CI 0·52, 0·60). The Bland–Altman plot showed a small mean overestimation of the DHD-index derived from the DHD-FFQ compared with the reference method. The DHD-index score was in the favourable direction associated with most macronutrient and micronutrient intakes when adjusted for energy intake. No associations between the DHD-index score and cardiometabolic risk factors were observed. In conclusion, the DHD-index derived from the DHD-FFQ was considered acceptable in ranking but relatively poor in individual assessment of diet quality.

Food liking-disliking patterns may strongly influence food choices and health. Here we assess: (1) whether food preference patterns are genetic/environmentally driven; and (2) the relationship between metabolomics profiles and food preference patterns in a large population of twins. 2,107 individuals from TwinsUK completed an online food and lifestyle preference questionnaire. Principle components analysis was undertaken to identify patterns of food liking-disliking. Heritability estimates for each liking pattern were obtained by structural equation modeling. The correlation between blood metabolomics profiles (280 metabolites) and each food liking pattern was assessed in a subset of 1,491 individuals and replicated in an independent subset of monozygotic twin pairs discordant for the liking pattern (65 to 88 pairs). Results from both analyses were meta-analyzed. Four major food-liking patterns were identified (Fruit and Vegetable, Distinctive Tastes, Sweet and High Carbohydrate, and Meat) accounting for 26% of the total variance. All patterns were moderately heritable (Fruit and Vegetable, h2[95% CI]: 0.36 [0.28; 0.44]; Distinctive Tastes: 0.58 [0.52; 0.64]; Sweet and High Carbohydrate: 0.52 [0.45, 0.59] and Meat: 0.44 [0.35; 0.51]), indicating genetic factors influence food liking-disliking. Overall, we identified 14 significant metabolite associations (Bonferroni p < 4.5 × 10−5) with Distinctive Tastes (8 metabolites), Sweet and High Carbohydrate (3 metabolites), and Meat (3 metabolites). Food preferences follow patterns based on similar taste and nutrient characteristics and these groupings are strongly determined by genetics. Food preferences that are strongly genetically determined (h2 ≥ 0.40), such as for meat and distinctive-tasting foods, may influence intakes more substantially, as demonstrated by the metabolomic associations identified here.

Nutrient-rich food (NRF) index scores are dietary quality indices based on nutrient density. We studied the design aspects involved in the development and validation of NRF index scores, using the Dutch consumption data and guidelines as an example. We evaluated fifteen NRF index scores against the Dutch Healthy Diet Index (DHD-index), a measure of adherence to the Dutch dietary guidelines, and against energy density. The study population included 2106 adults from the Dutch National Food Consumption Survey 2007–2010. The index scores were composed of beneficial nutrients (protein, fibre, fatty acids, vitamins, minerals), nutrients to limit (saturated fat, sugar, Na) or a combination. Moreover, the influence of methodological decisions was studied, such as the choice of calculation basis (100 g or 100 kcal (418 kJ)). No large differences existed in the prediction of the DHD-index by the fifteen NRF index scores. The score that best predicted the DHD-index included nine beneficial nutrients and three nutrients to limit on a 100-kcal basis, the NRF9.3 with a model R2 of 0·34. The scores were quite robust with respect to sex, BMI and differences in calculation methods. The NRF index scores were correlated with energy density, but nutrient density better predicted the DHD-index than energy density. Consumption of vegetables, cereals and cereal products, and dairy products contributed most to the individual NRF9.3 scores. In conclusion, many methodological considerations underlie the development and evaluation of nutrient density models. These decisions may depend upon the purpose of the model, but should always be based upon scientific, objective and transparent criteria.

The Dutch Healthy Diet index (DHD-index) was developed using data from two 24 h recalls (24hR) and appeared useful to evaluate diet quality in Dutch adults. As many epidemiologic studies use FFQ, we now estimated the DHD-index score using FFQ data. We compared whether this score showed similar associations with participants' characteristics, micronutrient intakes, and biomarkers of intake and metabolism compared with the DHD-index using 24hR data. Data of 121 Dutch participants of the European Food Consumption Validation study were used. Dietary intake was assessed by two 24hR and a 180-item FFQ. Biomarkers measured were serum total cholesterol and carotenoids, EPA + DHA in plasma phospholipids and 24 h urinary Na. A correlation of 0·48 (95 % CI 0·33, 0·61) was observed between the DHD-index score based on 24hR data and on FFQ data. Classification of participants into the same tertiles of the DHD-index was achieved for 57 %. Women showed higher DHD-index scores. Energy intake was inversely associated with both DHD-index scores. Furthermore, age and intakes of folate, Fe, Mg, K, vitamin B6 and vitamin C were positively associated with both DHD-index scores. DHD-index scores showed acceptable correlations with the four combined biomarkers taking energy intake into account (r24hR 0.55; rFFQ 0.51). In conclusion, the DHD-index score based on FFQ data shows similar associations with participants' characteristics, energy intake, micronutrient intake and biomarkers compared with the score based on 24hR data. Furthermore, ranking of participants was acceptable for both methods. FFQ data may therefore be used to assess diet quality using the DHD-index in Dutch populations.