Skip to main content Accessibility help
×
Home

Information:

  • Access
  • Cited by 24

Actions:

      • Send article to Kindle

        To send this article to your Kindle, first ensure no-reply@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

        Note you can select to send to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

        Find out more about the Kindle Personal Document Service.

        Evaluation of food and nutrient intake assessment using concentration biomarkers in European adolescents from the Healthy Lifestyle in Europe by Nutrition in Adolescence study
        Available formats
        ×

        Send article to Dropbox

        To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

        Evaluation of food and nutrient intake assessment using concentration biomarkers in European adolescents from the Healthy Lifestyle in Europe by Nutrition in Adolescence study
        Available formats
        ×

        Send article to Google Drive

        To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

        Evaluation of food and nutrient intake assessment using concentration biomarkers in European adolescents from the Healthy Lifestyle in Europe by Nutrition in Adolescence study
        Available formats
        ×
Export citation

Abstract

Accurate food and nutrient intake assessment is essential for investigating diet–disease relationships. In the present study, food and nutrient intake assessment among European adolescents using 24 h recalls (mean of two recalls) and a FFQ (separately and the combination of both) were evaluated using concentration biomarkers. Biomarkers included were vitamin C, β-carotene, DHA+EPA, vitamin B12 (cobalamin and holo-transcobalamin) and folate (erythrocyte folate and plasma folate). For the evaluation of the food intake assessment 390 adolescents were included, while 697 were included for the nutrient intake assessment evaluation. Spearman rank and Pearson correlations, and validity coefficients, which are correlations between intake estimated and habitual true intake, were calculated. Correlations were higher between frequency of food consumption (from the FFQ) and concentration biomarkers than between mean food intake (from the recalls) and concentration biomarkers, especially for DHA+EPA (r 0·35 v. r 0·27). Most correlations were higher among girls than boys. For boys, the highest validity coefficients were found for frequency of fruit consumption (0·88) and for DHA+EPA biomarker (0·71). In girls, the highest validity coefficients were found for fruit consumption frequency (0·76), vegetable consumption frequency (0·74), mean fruit intake (0·90) and DHA+EPA biomarker (0·69). After exclusion of underreporters, correlations slightly improved. Correlations between usual food intakes, adjusted for food consumption frequency, and concentration biomarkers were higher than correlations between mean food intakes and concentration biomarkers. In conclusion, two non-consecutive 24 h recalls in combination with a FFQ seem to be appropriate to rank subjects according to their usual food intake.

Accurate assessment of food intakes of free-living persons and especially of children and adolescents remains a difficult and labour-intensive process. Precise estimations are essential, however, especially with regard to the investigation of diet–disease relationships(Reference Tarasuk and Brooker1). Problems in the accurate estimation of individuals' usual intake levels, together with the relative homogeneity of food consumption patterns within populations make it difficult to accurately estimate the disease risk associated with specific dietary factors(Reference Beaton, Milner and Corey2Reference Freudenheim and Marshall4). Measurement errors can be divided into random errors and bias or systematic errors. Random errors may be a function of day-to-day variations in intake or errors in the analysis of food composition. Random errors in the classification of subjects according to their usual intakes can bias risk estimates and reduce the likelihood of detecting a significant association between diet and disease. Systematic errors in dietary data can inflate or deflate the relative risk or OR, depending on the direction of the bias and whether the source of the bias in dietary intake data is related to the disease outcome variable(Reference Tarasuk and Brooker1).

No single assessment method of an individual's usual intake is optimal under all conditions. The choice of method depends on a number of factors including the aim of the study, the characteristics of the study population, the accuracy of the dietary data required, and the funds and personnel available(Reference Willet5, Reference Gibson6). Repeated 24 h recalls have been shown to be a valid method to measure protein, K, fish, vegetable and fruit intakes among adults(Reference Crispim, Geelen and Souverein7, Reference Crispim, de Vries and Geelen8), though little is known about the validity among children and adolescents. Also, FFQ have been widely used as cost-effective dietary assessment methods in large-scale surveys to investigate usual food intakes. Both dietary intake assessment methods (repeated 24 h recalls and FFQ) have been used in the Healthy Lifestyle in Europe by Nutrition in Adolescence (HELENA) study(Reference Kersting, Sichert-Hellert and Vereecken9). Like all dietary assessment methods, estimates derived from 24 h recalls and FFQ data suffer from random and systematic errors and may not represent the ‘true’ usual intake of foods and nutrients. In general, repeated 24 h recalls have less bias and a larger within-individual variation than FFQ and vice versa(Reference Willet5, Reference Gibson6).

The advantages of nutritional biomarkers have been previously shown(Reference Potischman10). For example, the random errors occurring with their utilisation are likely to be independent of those in both 24 h recall and FFQ(Reference Gibson6). It is likely that there is some degree of correlation between random errors in 24 h recall and FFQ, as both methods rely on the subject's ability of recalling and describing food consumption(Reference Potischman10). Also, errors due to underreporting could occur when using 24 h recalls or FFQ(Reference Livingstone11Reference Lissner13). The inclusion of biomarkers in dietary validation studies makes it more likely that the criteria of independent errors are met(Reference Kaaks14).

The use of concentration biomarkers in validation studies is restricted to their associations with self-reported dietary intakes because these biomarkers are the result of complex metabolic processes(Reference Freedman, Kipnis and Schatzkin15). Evaluated concentration biomarkers include serum carotenoids(Reference Macdonald, Hardcastle and Duthie16Reference Drewnowski, Rock and Henderson19) and serum vitamin C(Reference Drewnowski, Rock and Henderson19, Reference Ness, Khaw and Bingham20) for fruit and vegetable intake and n-3 fatty acids (FA) for fish and seafood intake(Reference Katan, Deslypere and van Birgelen21Reference Leaf, Connor and Barstad23).

Strong correlations of dietary intakes of vitamin C and serum ascorbic acid concentrations have been reported mainly when habitual dietary intakes of vitamin C are relatively modest(Reference Bates, Rutishauser and Black24). As many factors influence serum folate concentrations and the bioavailability of dietary folate, intakes may correlate only weakly with serum concentrations unless broad categories of folate intake are used(Reference Jacques, Sulsky and Sadowski25). Erythrocyte folate concentrations correlate with liver folate levels and thus reflect folate stores(Reference Wu, Chanarin and Slavin26). Total serum vitamin B12 concentration reflects both the vitamin B12 intake and body stores. Weak but positive correlations were reported for males and females between dietary vitamin B12 intake and serum vitamin B12(Reference Gregory, Foster and Tyler27). Low correlations may be linked to the large size of liver vitamin B12 stores. However, holo-transcobalamin is the only circulating transport protein that delivers vitamin B12 to receptors on cell membranes and is the only biologically active form of the vitamin(Reference Gibson6). Reported correlations between FA in serum phospholipids and FA intake vary markedly across studies(Reference Andersen, Solvoll and Drevon28, Reference Kobayashi, Sasaki and Kawabata29) due to the fact that many factors may influence measured FA biomarkers in serum.

The purpose of the present study was to evaluate food and nutrient intake assessment among European adolescents from two 24 h recalls and an FFQ using concentration biomarkers. Fruit and vegetable intake v. vitamin C status and β-carotene status and fish intake v. the sum of DHA and EPA status were included to evaluate food intake. In addition, vitamin B12 (cobalamin and holo-transcobalamin), folate (plasma folate and erythrocyte folate), vitamin C, β-carotene, and the FA DHA and EPA were used to evaluate nutrient intake.

Subjects and methods

The HELENA Cross-Sectional Study is a multi-centre investigation of the nutritional and lifestyle status of adolescents in ten European cities(Reference Moreno, Gonzalez-Gross and Kersting30).

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by the Human Research Review Committee of the Universities of Bonn (Dortmund), Lille, Rome, Zaragoza, Athens, Heraklion, Pécs, Ghent and Vienna. Informed written consent was obtained from participants and both parents.

Subjects, recruitment and study design

For the purpose of this study, ten cities of more than 100 000 inhabitants located in nine European countries were included: Vienna (Austria), Gent (Belgium), Lille (France), Dortmund (Germany), Athens and Heraklion (Greece), Pécs (Hungary), Rome (Italy), Zaragoza (Spain) and Stockholm (Sweden)(Reference Moreno, Gonzalez-Gross and Kersting30). A random cluster sampling of 3000 European adolescents (target number) aged 12·5–17·5 years, stratified for geographical location, age and socio-economic status, was carried out. Adolescents were recruited from school sites.

Up to three classes from two grades were selected per school and a class was considered eligible if the participation rate in the class was at least 70 %. A random subgroup of approximately 1000 adolescents (target number) was selected to participate in the blood sampling. Adolescents were excluded from participation a posteriori if they were participating simultaneously in another clinical trial, if they were aged < 12·5 or ≥ 17·5 years if they had suffered from an acute infection less than 1 week before the inclusion.

All data were collected via standardised procedures. Details on the sampling and recruitment process and quality-control procedures can be found elsewhere(Reference Moreno, De and Gonzalez-Gross31).

Ethical issues and respect for good clinical procedures have also been discussed previously(Reference Beghin, Castera and Manios32).

Dietary assessment

Dietary intake was assessed using a computerised 24 h recall on two non-consecutive days of the week, excluding weekend days, and within a time-span of 2 weeks. In addition, a FFQ was completed by the adolescents. On the day of the first 24 h recall, a blood sample was taken.

The 24 h recall was assessed using a validated computer-based tool for self-reported 24 h recalls, the HELENA-Dietary Assessment Tool (DIAT), based on a previous version developed for Flemish adolescents, called Young Adolescents' Nutrition Assessment on Computer(Reference Vereecken, Covents and Matthys33, Reference Vereecken, Dohogne and Covents34). This HELENA-DIAT guides respondents through six ‘meal occasions’, embedded within questions that help the respondents to remember what they ate the day before. For each meal occasion, adolescents were invited to select all food items eaten at that occasion from a standardised menu. For each selected item, one or more extra screens were provided to gather detailed information on portion sizes. Additional features of the program are: probing for food items often eaten in combination with other items, a search engine, an extra category to add unavailable items, a number of checks (e.g. for extreme amounts, zero values, beverages, energy of consumed food, milk when cereals are consumed) and the possibility to add a remark before leaving the program.

Difficulties in obtaining comparable measures of kJ/g across countries precluded the use of country-specific food composition tables to calculate energy and nutrient intakes. Specifically, the national food composition tables were often limited in the number of foods that were included or in the number of nutrients to be analysed. Additionally, the definition of some nutrients differed between tables. To address this, the data of the HELENA-DIAT were linked to the German Food Code and Nutrient Data Base (BLS (Bundeslebensmittelschlüssel), version II.3.1, 2005)(Reference Dehne, Klemm and Henseler35), as this food composition database contained the largest number of nutrients and food items: approximately 12 000 coded foods, menus and menu components with up to 158 nutrient data points available for each product. Data from each country were linked to this database to ensure standardisation of available measures. If a food item was missing in the German food composition table, calculations were made via recipes or a local food composition table for the specific country.

Furthermore, a short self-administered FFQ with fifteen items was used from the Healthy Behaviour in School-aged Children study(Reference Currie36), including fruits, vegetables, sweets, soft drinks, light soft drinks, cereals, white bread, brown bread, skimmed milk, whole-fat milk, other milk, cheese, fish, crisps and French fries. Response categories included: never, less than once a week, once a week, 2–4 times a week, 5–6 times a week, once a day every day and more than once a day every day.

The database architecture and easy-to-enter, user-friendly software for data input were centrally developed in Teleform and MSAccess. The questions from the multiple response formats were scanned centrally, whereas the open entry data were entered and archived by each local field centre.

Analysis of biomarkers in blood

After a 10 h overnight fast, blood from the antecubital vein was drawn between 08.30 and 10.00 hours at school following a standardised blood collection protocol. Details about the transport of the samples, quality assurance and stability pilot study can be found elsewhere(Reference Gonzalez-Gross, Breidenassel and Gomez-Martinez37). Briefly, for the measurement of plasma folate, cobalamin and vitamin C, blood was collected in heparinised tubes, immediately placed on ice, and centrifuged within 30 min (3500 g for 15 min). Vitamin C samples were stabilised with metaphosphoric acid(Reference Gonzalez-Gross, Breidenassel and Gomez-Martinez37).

The supernatant fluid was transported at a stable temperature of 4–7°C to the central laboratory at the University of Bonn (IEL, Germany) and stored there at − 80°C until it was assayed. Serum samples for FA and β-carotene analysis were clotted at room temperature for at least 30 min and then centrifuged (3500 rpm, for 15 min). Aliquots for FA analysis were stored locally at − 20/ − 80°C as soon as possible. Once all fieldwork of one study centre was completed, all FA samples were shipped on dry ice to IEL and stored at − 80°C until analysis. After measuring the haematocrit in situ, EDTA whole blood was sent to IEL for the erythrocyte folate analysis. EDTA whole blood was diluted 1:5 with freshly prepared 0.1% ascorbic acid for cell lysis, and incubated for 60 min in the dark before storage at − 80°C. Erythrocyte folate was measured by competitive immunoassay (Immulite 2000; DPC Biermann GmbH). Sera for measuring holo-transcobalamin were obtained by centrifuging blood collected in evacuated tubes without anticoagulant at 3500 g for 15 min within 1 h. Once sent to IEL, the sera were aliquoted and stored at − 80°C until transport on dry ice to the biochemical laboratory at the Universidad Politécnica de Madrid for analysis.

Analyses of vitamin C, β-carotene, cobalamin, plasma folate, erythrocyte folate, DHA and EPA were performed centrally at IEL. Cobalamin and plasma folate were measured in heparin plasma by means of a competitive immunoassay using the Immunolite 2000 analyser (DPC Biermann GmbH). Antioxidative nutrients (β-carotene and vitamin C) were analysed by HPLC (Sykam Gilching) using UV detection (UV-Vis 205, Merck). Serum FA concentration was determined by capillary GC (Model 3900, Varian GmbH) after extraction performed by TLC. FA were identified by comparison of the peaks of interest with the retention times of authentic fatty acid methyl esters (FAME) standards (Sigma-Aldrich). The absolute FA profile was expressed as μmol/l. The relative amount of each FA (%) was expressed as the percentage of total concentration. The relative amount of DHA+EPA was expressed as the percentage of total FA concentration (% DHA+EPA). Holo-transcobalamin was measured at Universidad Politécnica de Madrid by microparticle enzyme immunoassay (Active B12 Axis-Shield Limited) with the use of AxSym (Abbott Diagnostics, S.A.).

Statistical analyses

The software package STATA 10.1 was used (StataCorp). Only adolescents with complete biomarker, 24 h recall and FFQ data were included in the evaluation of food intake assessment. For the evaluation of nutrient intake assessment, only adolescents who provided a blood sample and who performed a 24 h recall twice were included. Data from the FFQ were recalculated to consumption frequency per d (continuous variable). The mean food and nutrient intakes from both recall days were used in the analyses. Descriptive analyses of the study population were performed.

For food intake assessment evaluation, unadjusted Spearman rank correlations were computed between concentration biomarker and mean food intake, concentration biomarker and frequency of food consumption, and mean food intake and frequency of food consumption. Unadjusted Spearman rank correlations to study the relationship between food intake assessment and biomarker were calculated for fruit intake v. vitamin C status, vegetable intake v. vitamin C status, fruit intake v. β-carotene status, vegetable intake v. β-carotene status and fish intake v. sum of DHA and EPA status. Pearson's correlation coefficients (after logarithmic or square root transformation) were included as sensitivity analysis.

Furthermore, the triads method was used to evaluate the correlation between the three measurements (FFQ, biomarker and 24 h recall) and the true intake using validity coefficients(Reference Kaaks14, Reference Yokota, Miyazaki and Ito38, Reference Ocke and Kaaks39). The triads method is a triangular comparison between questionnaire, biomarker and 24 h recall used to obtain a quantitative estimate of validity coefficients(Reference Kaaks14). The technique assumes linearity between the three measurements and the true intake and assumes that the measurements have independent random errors. The assumption of independence implies that the correlations between any pair of variables are due to the relationship between each variable and the actual intake and not due to errors inherent in each assessment instrument (FFQ, 24 h recalls and biological markers)(Reference Kaaks14, Reference Ocke and Kaaks39, Reference Kaaks and Ferrari40). Pearson correlations were used to calculate the validity coefficients. The advantage of this method is the inclusion of the biomarker, which presents independent errors compared to the other dietary assessment methods, FFQ and 24 h recalls. Limitations of this technique include the occurrence of ρ>1, known as ‘Heywood cases’, and the existence of negative correlations, which do not allow the calculation of ρ. The main causes for the occurrence of Heywood cases include random sampling variations or violation of one or more assumptions of the triads method. In the first case, a validity coefficient above 1 is acceptable. Empirical negative correlations occur when the true correlations are near zero, i.e., the specific factors of the variable predominate over the latent variable. Increasing the sample size and using more accurate reference methods and biomarkers should reduce the likelihood of negative correlations(Reference Kaaks14). The validity coefficients vary from 0 to 1.

Individual usual food intakes, resulting from the two 24 h recalls and with incorporation of information from FFQ as a covariate, were estimated using the multiple source method(Reference Haubrock, Nothlings and Volatier41). This method removes the effect of day-to-day variability and random error in both 24 h recalls. This analysis aimed to verify whether the combination of two dietary assessment methods improves correlations between intake of foods and concentration biomarkers. Random intercept mixed models were used to verify associations between mean food intake and concentration biomarkers, food consumption frequency and concentration biomarkers, or mean nutrient intake and concentration biomarkers while controlling for school and centre as random effects and age and sex as fixed effects.

To evaluate nutrient intake assessment, unadjusted Spearman rank correlations were calculated for vitamin B12 status (cobalamin and holo-transcobalamin) v. vitamin B12 intake, folate status (plasma folate and erythrocyte folate) v. folate intake, β-carotene status v. β-carotene intake, vitamin C status v. vitamin C intake, DHA and EPA status v. DHA and EPA intake. Pearson's correlation coefficients (after logarithmic transformation) were included as sensitivity analysis.

Participants were classified into tertiles according to their biomarker status for the particular nutrient. The Kruskal–Wallis one-way ANOVA by ranks was used for testing the equality of population medians of intake among these tertiles.

Results were presented for boys and girls separately and were repeated after exclusion of underreporters. Underreporters were defined using the Goldberg cut-offs(Reference Goldberg, Black and Jebb42). BMR was calculated from age- and sex-specific FAO/WHO/UNU equations.

Underreporting was considered when the ratio of energy intake over the estimated BMR was lower than 0·96(Reference Black43).

Results

Only 390 adolescents (boys, n 163) were included for the food intake assessment evaluation, while for the nutrient intake assessment evaluation 697 (boys, n 323) adolescents were included. Descriptive characteristics, food consumption frequencies, mean food and nutrient intakes and concentration of selected biomarkers in blood of the study sample can be found in Table 1.

Table 1 General characteristics, mean daily food and nutrient intakes, food frequency consumption per d and concentration biomarkers in blood of the sample of European adolescents (Healthy Lifestyle in Europe by Nutrition in Adolescence study) (Mean values, standard deviations, medians and number of participants)

* Excluding potatoes.† Relative amount of DHA+EPA, expressed as the percentage of total fatty acid concentration.

Food intake assessment evaluation

All correlations between food consumption frequency and mean food intake, food consumption frequency and concentration biomarker, and mean food intake and concentration biomarker were significantly positive in all adolescents (n 390), except for mean vegetable intake v. vitamin C status (data not shown). Except for DHA+EPA status v. fish intake, highest correlations were found between food consumption frequencies and mean food intakes (r 0·51; P< 0·001 for fruits and r 0·29; P< 0·001 for vegetables). Overall, observed correlations were higher between food consumption frequency and concentration biomarker than between mean food intake and concentration biomarker, especially for DHA+EPA (r 0·35; P< 0·001 v. r 0·27; P< 0·001) (data not shown). Important sex differences were observed. All correlations between food consumption frequency and mean food intake, as well as all correlations between mean food intake and concentration biomarker, were substantially higher among girls than boys (Table 2). All associations between food consumption frequency and concentration biomarker were higher among boys than girls, except for associations with vegetable consumption frequency (Table 2). In girls, highest correlations were found between mean fruit intake and frequency of fruit consumption (r 0·56; P< 0·001), mean fish consumption and % DHA+EPA status (r 0·36; P< 0·001) and mean fish consumption and DHA+EPA concentration (r 0·34; P< 0·001). In boys, highest correlations were found between % DHA+EPA status and frequency of fish consumption (r 0·42; P< 0·001) (Table 2).

Table 2 Crude Spearman's rank and Pearson correlation coefficient for FFQ v. mean food intake from 24 h recalls; FFQ v. biomarker and mean food intake from 24 h recalls v. biomarker (Healthy Lifestyle in Europe by Nutrition in Adolescence study), for boys and girls

* After square root transformation of both FFQ and mean food intake, FFQ and nutrient status and mean food intake and nutrient status.

After square root transformation of food intake and FFQ and log-transformation of nutrient status.

In this column, the correlations reported refer to associations between food consumption frequency and food intake (e.g. consumption frequency of fruits v. mean fruit intake).§ Relative amount of DHA+EPA, expressed as the percentage of total fatty acid concentration.

For boys, highest validity coefficients were found for frequency of fruit consumption (0·88) and for % DHA+EPA biomarker (0·71). While in girls, the highest validity coefficients were found for fruit consumption frequency (0·76), vegetable consumption frequency (0·74), mean fruit intake (0·90) and % DHA+EPA biomarker (0·69) (Table 3). Except for DHA+EPA, the biomarker often had the lowest validity coefficient compared to FFQ and 24 h recall. One Heywood case occurred (for boys; for fruit consumption frequency in relation to true intake), where the estimated validity coefficient was >1. The validity of the 24 h recall was the highest in girls, while in boys the validity of the FFQ was higher compared to the 24 h recall.

Table 3 Validity coefficients (triads method) for FFQ, mean food intake from 24 h recalls and biomarker status for both boys and girls

* Validity coefficient of FFQ.

Validity coefficient of 24 h recall.

Validity coefficient of concentration biomarker.

§ Not possible to calculate due to negative correlation coefficient.∥ Relative amount of DHA+EPA, expressed as the percentage of total fatty acid concentration.

After exclusion of underreporters (n 86), except for the correlation between mean fish consumption and frequency of fish consumption, all correlations slightly improved. For example, r 0·51 became r 0·54 (P< 0·001) between mean fruit intake and frequency of fruit consumption and r 0·29 became r 0·37 (P< 0·001) between mean vegetable intake and frequency of vegetable consumption (data not shown).

After correction for centre and school as random effects and sex and age as fixed effects, all associations between food consumption frequency and concentration biomarker and between mean food intake and concentration biomarker, were significantly positive (data not shown).

Most correlations between mean usual food intakes, adjusted for food consumption frequency (Table 4), and concentration biomarkers were higher than correlations between mean food intakes and concentration biomarkers (Table 2), especially for fish intake v. DHA+EPA status, fruit intake v. vitamin C status and fruit intake v. β-carotene status. Most of the differences were not significant, but the general trend is remarkable even when some correlations were lower. When examining Spearman rank correlations between usual food intakes and biomarkers (without taking into account food frequency), most of the correlations were equal or slightly higher than when using the mean intakes, but lower than when taking into account food frequency of consumption, which indicates the importance of the FFQ information. The latter was only not the case for fish consumption v. DHA+EPA. Usual fish intake correlated better with DHA+EPA than mean fish intake, but when usual fish intake was corrected for frequency of fish consumption, lower correlations were obtained (data not shown).

Table 4 Crude Spearman's rank and Pearson correlation coefficient for usual food intake (corrected for frequency of consumption) v. biomarker (Healthy Lifestyle in Europe by Nutrition in Adolescence study)

* After square root transformation of both usual nutrient intake and nutrient status.

After square root transformation of usual nutrient intake and log-transformation of nutrient status.‡ Relative amount of DHA+EPA, expressed as the percentage of total fatty acid concentration.

Nutrient intake assessment evaluation

Overall correlations between mean nutrient intake and concentration biomarkers were low. The highest correlations were found for β-carotene (ρ = 0·19 or r 0·23; P< 0·001) and % DHA+EPA (ρ = 0·16 or r 0·17; P< 0·001) (data not shown). There were important sex differences. Correlations were higher among girls than boys, except for EPA+DHA (Table 5). For girls highest correlations were found for folate, active vitamin B12 and β-carotene, while for boys highest correlations were found for DHA+EPA. For girls, correlations were substantially higher when active vitamin B12 was used instead of vitamin B12 as a biomarker. Correlations for folate were better when plasma folate instead of erythrocyte folate was used as a biomarker, for both boys and girls. In most cases, correlations slightly or substantially improved after exclusion of underreporters (n 142): r 0·16 → r 0·20; P< 0·001 for folic acid, r 0·14 → r 0·20; P< 0·001 for vitamin C and r 0·16 → r 0·19; P< 0·001 for DHA+EPA (data not shown). Spearman rank correlations comparing usual nutrient intakes v. biomarkers were equal or slightly higher than spearman rank correlations comparing mean nutrient intakes v. biomarkers (data not shown).

Table 5 Crude Spearman's rank and Pearson correlation coefficient for mean nutrient intake (dietary recall) v. nutrient status (biomarkers in blood samples) (Healthy Lifestyle in Europe by Nutrition in Adolescence study), for boys and girls separately (n 697)

* After log-transformation.

After correction for centre and school as random effects and sex and age as fixed effects, all associations between nutrient intake and concentration biomarkers were significantly positive (data not shown). For all nutrients, biomarker status significantly increased with increasing tertile of nutrient intake (Table 6).

Table 6 Average nutrient status (biomarker) by tertiles of nutrient intake (24 h recall) (n 697) (Mean values, standard deviations and number of participants)

* P from Kruskal–Wallis.† Relative amount of DHA+EPA, expressed as the percentage of total fatty acid concentration.

Discussion

Except for fish intake v. DHA+EPA, highest correlations were found between food consumption frequencies and mean food intakes derived from two independent 24 h recalls. In many cases, correlations were better between food consumption frequency and concentration biomarker than between mean food intake and concentration biomarker, especially for DHA+EPA. This is most probably due to the fact that food consumption frequency represents usual intake while two non-consecutive 24 h recalls represent actual intake, in particular for foods that are generally not consumed daily (e.g. fish). In addition, FA biomarkers in serum phospholipids mirror the dietary intake for recent weeks, while plasma vitamin C and β-carotene reflect short-term dietary intake. It was found previously that dietary intakes correlate better with biomarkers when the number of days covered by the reference method increases(Reference Henriquez-Sanchez, Sanchez-Villegas and Doreste-Alonso44). Most correlations between mean usual food intakes, taking into account food consumption frequency, and concentration biomarkers were higher than correlations between mean food intakes and concentration biomarkers, especially for DHA+EPA v. fish intake, vitamin C v. fruit intake and β-carotene v. fruit intake. Consequently, it can be recommended to combine information from different dietary assessment methods to estimate dietary intake. Compared to the European Food Consumption Validation study among adults(Reference Crispim, Geelen and Souverein7), correlations between usual fish intake (assessed combining 24 h recall and FFQ) and DHA+EPA were similar though slightly higher in our study.

Except for DHA+EPA, biomarkers often had the lowest validity coefficient compared to FFQ and 24 h recall. The fact that biomarkers do not always perform better than other food intake assessment methods was observed previously(Reference Kabagambe, Baylin and Allan45). Moreover, not all nutrients have biological markers and many are influenced by other factors than intake. The weak correlations result from factors related to absorption, post-absorptive metabolism or physiological regulation of nutrient levels, which can be important sources of random variations in the markers, unrelated to true intake(Reference Potischman10, Reference Potischman and Freudenheim46). The quantitative relationship of biomarkers with intake may differ among individuals; they cannot be transformed into absolute estimates of ingestion(Reference Kaaks, Ferrari and Ciampi47). Hence, biomarkers should be used in addition to and not in replacement of dietary surveys.

The fact that biomarkers had the lowest validity coefficients could be due to the fact that there was a positive co-variance between the random errors of the questionnaire and 24 h recalls and thus that validity coefficients in relation to true intake of FFQ and 24 h recalls have been overestimated and should be interpreted as upper limits. In the latter case, the validity coefficients of biomarkers in relation to true intake could be underestimated(Reference Kaaks14). Overall correlations between mean nutrient intake and concentration biomarkers were low. The highest correlations were found for β-carotene and % DHA+EPA.

The results showed important sex differences. All correlations between food consumption frequency and mean food intake, as well as all correlations between mean food intake and concentration biomarkers, were substantially higher among girls than boys. All associations between food consumption frequency and concentration biomarkers, except for associations with vegetable consumption frequency, were higher among boys than girls. For girls, the highest correlations were found between mean fruit intake and frequency of fruit consumption, mean fish consumption and % DHA+EPA status, and mean fish consumption and DHA+EPA concentration. For boys, the highest correlations were found between % DHA+EPA status and frequency of fish consumption. Except for DHA+EPA, correlations were better among girls than boys. For both, the evaluation of food and nutrient intake assessment correlations improved slightly after exclusion of underreporters, which is an interesting finding and could plead for the exclusion of underreporters when investigating diet–disease relationships.

The HELENA study has several strengths. The sampling procedure and the strict standardisation of the fieldwork among the countries involved in the study avoided introduction of bias due to inconsistent protocols and different laboratory methods, which in turn makes comparing results from isolated studies difficult. The combination of repeated 24 h recalls, FFQ and biomarker information allowed an in-depth investigation of the validity of the dietary intake assessment methods used in the HELENA study. Furthermore, the sample size was large enough for validation studies with biomarkers as the reference method(Reference Serra-Majem, Frost and Henrique-Sanchez48).

This study is not without limitations. Data are observational, based on self-reported dietary intake and are mean values of two non-consecutive 24 h recalls, which is probably not enough time to accurately capture usual intake, especially for children and adolescents where the ratio of within-to between-subject variability is larger(Reference Livingstone and Robson49). It is important to remember that current blood concentrations of vitamins in the adolescent population do not necessarily mean that these concentrations are the most adequate ones from the biological point of view. Furthermore, no detailed information on supplement intake was available from the HELENA study and this might have affected the observed correlations between the biomarkers and food intake assessment. Finally, the study sample is not nationally representative and thus the results may not be generalised to broader adolescent populations, in the countries surveyed or elsewhere. Furthermore, for the food intake assessment evaluation, many adolescents were lost in subsequent analysis because they did not complete one or more of the questions in the FFQ. Therefore, the sample size for the food validations was much lower than for the nutrient validations, since this latter validation did not include the FFQ data. Further nutrient intake estimations were based on the German food composition table and may have introduced some bias in the nutrient intake calculations. However, preliminary analyses comparing the differences in nutrient intake estimates of eight nutrients when using the BLS table for all HELENA countries in comparison with the approach in which all countries used their national food composition table to calculate the nutrient intakes have shown that differences between these two approaches are only small and for most nutrients negligible (I Huybrechts, unpublished results). It is, however, clear from this study that correlations between biomarkers and food intake assessment (which does not use food composition tables) are better than between biomarkers and nutrient intake assessment.

Conclusion

The present study shows that two non-consecutive 24 h recalls in combination with a FFQ seem to be appropriate to rank subjects according to their usual food intake.

Based on the results of this study, the use of a combination of dietary assessment methods such as 24 h recall and FFQ is recommended to estimate dietary intake, especially for foods which are not consumed on a daily basis such as fish. Besides, the exclusion of underreporters might be beneficial.

Acknowledgements

The HELENA Study has taken place with the financial support of the European Community Sixth RTD Framework Programme (contract FOOD-CT-2005-007034). The content of the present paper reflects only the authors' views and the European Community/rest of the HELENA Study members are not responsible/liable for any use that may be made of the information contained therein. The writing group takes sole responsibility for the content of this article. Additional support from the Spanish Ministry of Education (AGL2007-29784-E/ALI), Axis-Shield Diagnostics Limited (Oslo, Norway), Abbott Científica S.A. (Spain) and Cognis GmbH is also acknowledged. And many thanks to Christel Bierschbach, Adelheid Schuch, Anke Berchtold, Pickert, Andre Spinneker and Ulrike Albers for their contribution to laboratory work. The authors declare not having any conflicts of interest with regard to this study. The authors' contributions to the study were as follows: S. V. performed the statistical analyses and drafted the manuscript. A. G., P. V. V. and I. H. contributed to the interpretation of the results. L. A. M. coordinated the HELENA project and contributed to the conception and design and data collection. L. B. performed regulatory requirement management and contributed to data collection. C. Br. performed lab analyses and data collection. Y. M. contributed to the conception and design. S. D. H. contributed to the conception and design and data collection. M. G.-G. contributed to the conception and design of the HELENA study, design of the three questionnaires and other methodology, data-cleaning and laboratory analyses. A. K., D. M. and M. S. contributed to the conception and design and data collection; J. W. contributed to data collection; C. C. G. contributed to the conception and design; and I. H. performed data management. All the aforementioned and not explicitly mentioned authors critically revised the draft versions of the manuscript.

HELENA Study Group

Co-ordinator: Luis A. Moreno.

Core Group members: Luis A. Moreno, Fréderic Gottrand, Stefaan De Henauw, Marcela González-Gross, Chantal Gilbert.

Steering Committee: Anthony Kafatos (President), Luis A. Moreno, Christian Libersa, Stefaan De Henauw, Sara Castelló, Fréderic Gottrand, Mathilde Kersting, Michael Sjöstrom, Dénes Molnár, Marcela González-Gross, Jean Dallongeville, Chantal Gilbert, Gunnar Hall, Lea Maes, Luca Scalfi.

Project Manager: Pilar Meléndez.

1. Universidad de Zaragoza (Spain): Luis A. Moreno, Jesús Fleta, José A. Casajús, Gerardo Rodríguez, Concepción Tomás, María I. Mesana, Germán Vicente-Rodríguez, Adoración Villarroya, Carlos M. Gil, Ignacio Ara, Juan Revenga, Carmen Lachen, Juan Fernández Alvira, Gloria Bueno, Aurora Lázaro, Olga Bueno, Juan F. León, Jesús Mª Garagorri, Manuel Bueno, Juan Pablo Rey López, Iris Iglesia, Paula Velasco, Silvia Bel, Theodora Mouratidou.

2. Consejo Superior de Investigaciones Científicas (Spain): Ascensión Marcos, Julia Wärnberg, Esther Nova, Sonia Gómez, Ligia Esperanza Díaz, Javier Romeo, Ana Veses, Belén Zapatera, Tamara Pozo, David Martínez.

3. Université de Lille 2 (France): Laurent Beghin, Christian Libersa, Frédéric Gottrand, Catalina Iliescu, Juliana Von Berlepsch.

4. Pécsi Tudományegyetem (University of Pécs) (Hungary): Dénes Molnar, Eva Erhardt, Katalin Csernus, Katalin Török, Szilvia Bokor, Mrs. Angster, Enikö Nagy, Orsolya Kovács, Judit Répasi.

5. University of Crete School of Medicine (Greece): Anthony Kafatos, Caroline Codrington, María Plada, Angeliki Papadaki, Katerina Sarri, Anna Viskadourou, Christos Hatzis, Michael Kiriakakis, George Tsibinos, Constantine Vardavas, Manolis Sbokos, Eva Protoyeraki, Maria Fasoulaki.

6. Institut für Ernährungs- und Lebensmittelwissenschaften – Ernährungphysiologie. Rheinische Friedrich Wilhelms Universität (Germany): Peter Stehle, Klaus Pietrzik, Marcela González-Gross, Christina Breidenassel, Andre Spinneker, Jasmin Al-Tahan, Miriam Segoviano, Anke Berchtold, Christine Bierschbach, Erika Blatzheim, Adelheid Schuch, Petra Pickert.

7. University of Granada (Spain): Manuel J. Castillo, Ángel Gutiérrez, Francisco B Ortega, Jonatan R Ruiz, Enrique G Artero, Vanesa España, David Jiménez-Pavón, Palma Chillón, Cristóbal Sánchez-Muñoz, Magdalena Cuenca

8. Istituto Nazionalen di Ricerca per gli Alimenti e la Nutrizione (Italy): Davide Arcella, Elena Azzini, Emma Barrison, Noemi Bevilacqua, Pasquale Buonocore, Giovina Catasta, Laura Censi, Donatella Ciarapica, Paola D'Acapito, Marika Ferrari, Myriam Galfo, Cinzia Le Donne, Catherine Leclercq, Giuseppe Maiani, Beatrice Mauro, Lorenza Mistura, Antonella Pasquali, Raffaela Piccinelli, Angela Polito, Raffaella Spada, Stefania Sette, Maria Zaccaria.

9. University of Napoli “Federico II” Dept of Food Science (Italy): Luca Scalfi, Paola Vitaglione, Concetta Montagnese.

10. Ghent University (Belgium): Ilse De Bourdeaudhuij, Stefaan De Henauw, Tineke De Vriendt, Lea Maes, Christophe Matthys, Carine Vereecken, Mieke de Maeyer, Charlene Ottevaere, Inge Huybrechts.

11. Medical University of Vienna (Austria): Kurt Widhalm, Katharina Phillipp, Sabine Dietrich, Birgit Kubelka

Marion Boriss-Riedl.

12. Harokopio University (Greece): Yannis Manios, Eva Grammatikaki, Zoi Bouloubasi, Tina Louisa Cook, Sofia Eleutheriou, Orsalia Consta, George Moschonis, Ioanna Katsaroli, George Kraniou, Stalo Papoutsou, Despoina Keke, Ioanna Petraki, Elena Bellou, Sofia Tanagra, Kostalenia Kallianoti, Dionysia Argyropoulou, Katerina Kondaki, Stamatoula Tsikrika, Christos Karaiskos.

13. Institut Pasteur de Lille (France): Jean Dallongeville, Aline Meirhaeghe.

14. Karolinska Institutet (Sweden): Michael Sjöstrom, Jonatan R Ruiz, Francisco B. Ortega, María Hagströmer, Anita Hurtig Wennlöf, Lena Hallström, Emma Patterson, Lydia Kwak, Julia Wärnberg, Nico Rizzo.

15. Asociación de Investigación de la Industria Agroalimentaria (Spain): Jackie Sánchez-Molero, Sara Castelló, Elena Picó, Maite Navarro, Blanca Viadel, José Enrique Carreres, Gema Merino, Rosa Sanjuán, María Lorente, María José Sánchez.

16. Campden BRI (United Kingdom): Chantal Gilbert, Sarah Thomas, Elaine Allchurch, Peter Burgess.

17. SIK - Institutet foer Livsmedel och Bioteknik (Sweden): Gunnar Hall, Annika Astrom, Anna Sverkén, Agneta Broberg.

18. Meurice Recherche & Development asbl (Belgium): Annick Masson, Claire Lehoux, Pascal Brabant, Philippe Pate, Laurence Fontaine.

19. Campden & Chorleywood Food Development Institute (Hungary): Andras Sebok, Tunde Kuti, Adrienn Hegyi.

20. Productos Aditivos SA (Spain): Cristina Maldonado, Ana Llorente.

21. Cárnicas Serrano SL (Spain): Emilio García.

22. Cederroth International AB (Sweden): Holger von Fircks, Marianne Lilja Hallberg, Maria Messerer

23. Lantmännen Food R&D (Sweden): Mats Larsson, Helena Fredriksson, Viola Adamsson, Ingmar Börjesson.

24. European Food Information Council (Belgium): Laura Fernández, Laura Smillie, Josephine Wills.

25. Universidad Politécnica de Madrid (Spain): Marcela González-Gross, Jara Valtueña, David Jiménez-Pavón, Ulrike Albers, Raquel Pedrero, Agustín Meléndez, Pedro J. Benito, Juan José Gómez Lorente, David Cañada, Alejandro Urzanqui, Francisco Fuentes, Rosa María Torres, Paloma Navarro.

References

1Tarasuk, VS & Brooker, AS (1997) Interpreting epidemiologic studies of diet–disease relationships. J Nutr 127, 18471852.
2Beaton, GH, Milner, J, Corey, P, et al. (1979) Sources of variance in 24-hour dietary recall data: implications for nutrition study design and interpretation. Am J Clin Nutr 32, 25462559.
3Beaton, GH, Burema, J & Ritenbaugh, C (1997) Errors in the interpretation of dietary assessments. Am J Clin Nutr 65, 1100S1107S.
4Freudenheim, JL & Marshall, JR (1988) The problem of profound mismeasurement and the power of epidemiological studies of diet and cancer. Nutr Cancer 11, 243250.
5Willet, W (1990) Nutritional Epidemiology. Oxford: Oxford University Press.
6Gibson, RS (2005) Principles of Nutritional Assessment, 2nd ed.New York, NY: Oxford University Press.
7Crispim, SP, Geelen, A, Souverein, OW, et al. (2011) Biomarker-based evaluation of two 24-h recalls for comparing usual fish, fruit and vegetable intakes across European centers in the EFCOVAL Study. Eur J Clin Nutr 65, Suppl. 1, S38S47.
8Crispim, SP, de Vries, JH, Geelen, A, et al. (2011) Two non-consecutive 24 h recalls using EPIC-Soft software are sufficiently valid for comparing protein and potassium intake between five European centres – results from the European Food Consumption Validation (EFCOVAL) study. Br J Nutr 105, 447458.
9Kersting, M, Sichert-Hellert, W, Vereecken, CA, et al. (2008) Food and nutrient intake, nutritional knowledge and diet-related attitudes in European adolescents. Int J Obes (Lond) 32, Suppl. 5, S35S41.
10Potischman, N (2003) Biologic and methodologic issues for nutritional biomarkers. J Nutr 133, Suppl. 3, 875S880S.
11Livingstone, MB (1995) Assessment of food intakes: are we measuring what people eat? Br J Biomed Sci 52, 5867.
12Livingstone, MB & Black, AE (2003) Markers of the validity of reported energy intake. J Nutr 133, Suppl. 3, 895S920S.
13Lissner, L (2002) Measuring food intake in studies of obesity. Public Health Nutr 5, 889892.
14Kaaks, RJ (1997) Biochemical markers as additional measurements in studies of the accuracy of dietary questionnaire measurements: conceptual issues. Am J Clin Nutr 65, 1232S1239S.
15Freedman, LS, Kipnis, V, Schatzkin, A, et al. (2010) Can we use biomarkers in combination with self-reports to strengthen the analysis of nutritional epidemiologic studies? Epidemiol Perspect Innov 7, 2.
16Macdonald, HM, Hardcastle, AC, Duthie, GG, et al. (2009) Changes in vitamin biomarkers during a 2-year intervention trial involving increased fruit and vegetable consumption by free-living volunteers. Br J Nutr 102, 14771486.
17Polsinelli, ML, Rock, CL, Henderson, SA, et al. (1998) Plasma carotenoids as biomarkers of fruit and vegetable servings in women. J Am Diet Assoc 98, 194196.
18Yeum, KJ, Booth, SL, Sadowski, JA, et al. (1996) Human plasma carotenoid response to the ingestion of controlled diets high in fruits and vegetables. Am J Clin Nutr 64, 594602.
19Drewnowski, A, Rock, CL, Henderson, SA, et al. (1997) Serum beta-carotene and vitamin C as biomarkers of vegetable and fruit intakes in a community-based sample of French adults. Am J Clin Nutr 65, 17961802.
20Ness, AR, Khaw, KT, Bingham, S, et al. (1999) Plasma vitamin C: what does it measure? Public Health Nutr 2, 5154.
21Katan, MB, Deslypere, JP, van Birgelen, AP, et al. (1997) Kinetics of the incorporation of dietary fatty acids into serum cholesteryl esters, erythrocyte membranes, and adipose tissue: an 18-month controlled study. J Lipid Res 38, 20122022.
22Zock, PL, Mensink, RP, Harryvan, J, et al. (1997) Fatty acids in serum cholesteryl esters as quantitative biomarkers of dietary intake in humans. Am J Epidemiol 145, 11141122.
23Leaf, DA, Connor, WE, Barstad, L, et al. (1995) Incorporation of dietary n-3 fatty acids into the fatty acids of human adipose tissue and plasma lipid classes. Am J Clin Nutr 62, 6873.
24Bates, CJ, Rutishauser, IH, Black, AE, et al. (1979) Long-term vitamin status and dietary intake of healthy elderly subjects. 2. Vitamin C. Br J Nutr 42, 4356.
25Jacques, PF, Sulsky, SI, Sadowski, JA, et al. (1993) Comparison of micronutrient intake measured by a dietary questionnaire and biochemical indicators of micronutrient status. Am J Clin Nutr 57, 182189.
26Wu, A, Chanarin, I, Slavin, G, et al. (1975) Folate deficiency in the alcoholic-its relationship to clinical and haematological abnormalities, liver disease and folate stores. Br J Haematol 29, 469478.
27Gregory, J, Foster, K, Tyler, H, et al. (1990) The Dietary and Nutritional Survey of British Adults. London: Her Majesty's Stationery Office.
28Andersen, LF, Solvoll, K & Drevon, CA (1996) Very-long-chain n-3 fatty acids as biomarkers for intake of fish and n-3 fatty acid concentrates. Am J Clin Nutr 64, 305311.
29Kobayashi, M, Sasaki, S, Kawabata, T, et al. (2001) Single measurement of serum phospholipid fatty acid as a biomarker of specific fatty acid intake in middle-aged Japanese men. Eur J Clin Nutr 55, 643650.
30Moreno, LA, Gonzalez-Gross, M, Kersting, M, et al. (2008) Assessing, understanding and modifying nutritional status, eating habits and physical activity in European adolescents: the HELENA (Healthy Lifestyle in Europe by Nutrition in Adolescence) Study. Public Health Nutr 11, 288299.
31Moreno, LA, De, HS, Gonzalez-Gross, M, et al. (2008) Design and implementation of the Healthy Lifestyle in Europe by Nutrition in Adolescence Cross-Sectional Study. Int J Obes (Lond) 32, Suppl. 5, S411.
32Beghin, L, Castera, M, Manios, Y, et al. (2008) Quality assurance of ethical issues and regulatory aspects relating to good clinical practices in the HELENA Cross-Sectional Study. Int J Obes (Lond) 32, Suppl. 5, S12S18.
33Vereecken, CA, Covents, M, Matthys, C, et al. (2005) Young adolescents' nutrition assessment on computer (YANA-C). Eur J Clin Nutr 59, 658667.
34Vereecken, C, Dohogne, S, Covents, M, et al. (2010) How accurate are adolescents in portion-size estimation using the computer tool Young Adolescents' Nutrition Assessment on Computer (YANA-C)? Br J Nutr 103, 18441850.
35Dehne, LI, Klemm, C, Henseler, G, et al. (1999) The German Food Code and Nutrient Data Base (BLS II.2). Eur J Epidemiol 15, 355359.
36Currie, C (1998) Health Behaviour in School-aged Children: a WHO Cross-national Survey (HBSC). Research Protocol for the 1997/1998 Survey. Edinburgh: Research Unit in Health and Behavioural Change, University of Edinburgh.
37Gonzalez-Gross, M, Breidenassel, C, Gomez-Martinez, S, et al. (2008) Sampling and processing of fresh blood samples within a European multicenter nutritional study: evaluation of biomarker stability during transport and storage. Int J Obes (Lond) 32, Suppl. 5, S66S75.
38Yokota, RT, Miyazaki, ES & Ito, MK (2010) Applying the triads method in the validation of dietary intake using biomarkers. Cad Saude Publica 26, 20272037.
39Ocke, MC & Kaaks, RJ (1997) Biochemical markers as additional measurements in dietary validity studies: application of the method of triads with examples from the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr 65, 1240S1245S.
40Kaaks, R & Ferrari, P (2006) Dietary intake assessments in epidemiology: can we know what we are measuring? Ann Epidemiol 16, 377380.
41Haubrock, J, Nothlings, U, Volatier, JL, et al. (2011) Estimating usual food intake distributions by using the multiple source method in the EPIC-Potsdam Calibration Study. J Nutr 141, 914920.
42Goldberg, GR, Black, AE, Jebb, SA, et al. (1991) Critical evaluation of energy intake data using fundamental principles of energy physiology: 1. Derivation of cut-off limits to identify under-recording. Eur J Clin Nutr 45, 569581.
43Black, AE (2000) Critical evaluation of energy intake using the Goldberg cut-off for energy intake:basal metabolic rate. A practical guide to its calculation, use and limitations. Int J Obes Relat Metab Disord 24, 11191130.
44Henriquez-Sanchez, P, Sanchez-Villegas, A, Doreste-Alonso, J, et al. (2009) Dietary assessment methods for micronutrient intake: a systematic review on vitamins. Br J Nutr 102, Suppl. 1, S10S37.
45Kabagambe, EK, Baylin, A, Allan, DA, et al. (2001) Application of the method of triads to evaluate the performance of food frequency questionnaires and biomarkers as indicators of long-term dietary intake. Am J Epidemiol 154, 11261135.
46Potischman, N & Freudenheim, JL (2003) Biomarkers of nutritional exposure and nutritional status: an overview. J Nutr 133, Suppl. 3, 873S874S.
47Kaaks, R, Ferrari, P, Ciampi, A, et al. (2002) Uses and limitations of statistical accounting for random error correlations, in the validation of dietary questionnaire assessments. Public Health Nutr 5, 969976.
48Serra-Majem, L, Frost, AL, Henrique-Sanchez, P, et al. (2009) Evaluating the quality of dietary intake validation studies. Br J Nutr 102, Suppl. 1, S3S9.
49Livingstone, MB & Robson, PJ (2000) Measurement of dietary intake in children. Proc Nutr Soc 59, 279293.