To illustrate biomarkers of diet that can be used to validate estimates of dietary intake in the study of gene–environment interactions in complex diseases.

Prospective cohort studies, studies of biomarkers where diet is carefully controlled.

Free–living individuals, volunteers in metabolic suites.

Male and female human volunteers.

Recent studies using biomarkers have demonstrated substantial differences in the extent of measurement error from those derived by comparison with other methods of dietary assessment. The interaction between nutritional and genetic factors has so far largely gone uninvestigated, but can be studied in epidemiological trials that include collections of biological material. Large sample sizes are required to study interactions, and these are made larger in the presence of measurement errors.

Diet is of key importance in affecting the risk of most chronic diseases in man. Nutritional epidemiology provides the only direct approach to the quantification of risks. The introduction of biomarkers to calibrate the measurement error in dietary reports, and as additional measures of exposure, is a significant development in the effort to improve estimates of the magnitude of the contribution of diet in affecting individual disease risk within populations. The extent of measurement error has important implications for correction for regression dilution and for sample size. The collection of biological samples to improve and validate estimates of exposure, enhance the pursuit of scientific hypotheses, and enable gene–nutrient interactions to be studied, should become the routine in nutritional epidemiology.

To compare various food-frequency questionnaires (FFQs) used in nutritional studies in China for various purposes.

In Study 1, a simplified FFQ with 17 questions on food was used in a large rural study. In Study 2, a questionnaire consisting of 84 questions on food consumption of 16 food categories was used in a study comparing dietary consumption data and various health indicators of elderly people in four geographical areas in China. In Study 3, a questionnaire with 149 items in 17 food categories is being validated by comparison with data obtained by repeated 24-hour recalls.

Study 1 was carried out in one southern site and a northern site in 1996 to 1997. Study 2 was conducted in four different geographical sites in 1998. The on-going study, Study 3, has been carried out in Jiangsu and Beijing since 1999.

Study 1 included 12 234 rural Chinese adults aged 40 years. There were 546 elderly people in Study 2. Study 3 is collecting data from 300 healthy adults.

The results of food consumption and nutrient intakes from Study 1 were comparable with those obtained from a previous household dietary survey, in which sensible correlations between diet and diseases were also found. In Study 2, the dietary data from the four geographical areas showed significant differences in food and nutrient intakes among the different areas. The validation of the new FFQ in Study 3 is still going on.

The FFQ is a useful method for the collection of individual food consumption information. The above FFQ forms could be used in studies with different purposes, especially in studying the relationship between diet, nutrition and chronic diseases.

To compare, along with behavioural habits, the potential atherogenicity of diets in rural and urban areas in Costa Rica.

Subjects (n = 503) were randomly selected from the general population in Costa Rica. A validated food-frequency questionnaire that inquired about dietary intake in the previous year was administered once to each subject. Each subject provided plasma and an adipose tissue biopsy, which were used as biomarkers for carotenoid, tocopherol and/or fatty acid intakes. Each subject also answered another questionnaire on personal profile and household characteristics.

A dietitian visited all subjects and conducted the interviews at their homes.

Adult male and female free-living rural, suburban and urban Costa Ricans without a history of physical or mental disability.

Subjects in rural areas were significantly (P < 0.05) more active physically, earned less income and had a higher intake of dietary fibre than urban dwellers. Urban residents reported a significantly (P < 0.05) higher consumption of total fat, specifically unsaturated fat mostly from soyabean oil, and had higher plasma and adipose tissue tocopherol and lycopene concentrations. Interestingly, no differences in body mass index were observed.

These data show differences in cardiovascular disease (CVD) risk factors of rural and urban populations in Costa Rica. In rural areas, low socio-economic status and low intake of unsaturated fatty acids appear to be the prevalent CVD risk factors, while in urban areas they were low physical activity, high trans-fatty acids in the diet and adipose tissue, and low dietary fibre intake.

To summarise the applications and appropriate use of Dietary Reference Intakes (DRIs) as guidance for nutrition and health research professionals in the dietary assessment of groups and individuals.

Key points from the Institute of Medicine report, Dietary Reference Intakes: Applications in Dietary Assessment, are summarised in this paper. The different approaches for using DRIs to evaluate the intakes of groups vs. the intakes of individuals are highlighted.

Each of the new DRIs is defined and its role in the dietary assessment of groups and individuals is described. Two methods of group assessment and a new method for quantitative assessment of individuals are described. Illustrations are provided on appropriate use of the Estimated Average Requirement (EAR), the Adequate Intake (AI) and the Tolerable Upper Intake Level (UL) in dietary assessment.

Dietary assessment of groups or individuals must be based on estimates of usual (long-term) intake. The EAR is the appropriate DRI to use in assessing groups and individuals. The AI is of limited value in assessing nutrient adequacy, and cannot be used to assess the prevalence of inadequacy. The UL is the appropriate DRI to use in assessing the proportion of a group at risk of adverse health effects. It is inappropriate to use the Recommended Dietary Allowance (RDA) or a group mean intake to assess the nutrient adequacy of groups.

To validate abbreviated methods that estimate food security level among poor communities in Caracas, Venezuela.

Two independent cross-sectional studies were undertaken to internally and externally validate simple quantitative/qualitative methods. The quantitative measure was constructed from data on household food availability, gathered using the list-recall method. It is a count of the foods that explain 85% or more of household energy availability. The qualitative measure is a score of female-perceived food insecurity level estimated with a modified ‘hunger index’, reflecting food resource constraints and hunger experiences within the home. Socio-economic and food behaviour data that may predict household food security (HFS) levels were gathered. The second study was repeated a year later to measure the impact of an increase in the minimum wage on HFS levels.

Two poor urban communities in Caracas, Venezuela.

All households in both communities that complied with selection criteria (poor and very poor families that share food resources) and were willing to participate. The sample comprised 238 and 155 female household food managers in the two communities.

In 1995, data from females in 238 urban poor households provided evidence for the overall validity of the method. Its application in 1997 to 155 households in the other community gave support to the external validity of the method. Measures were repeated in 1998 on 133 subjects of the above sample, when the minimum wage was increased by 23%. Evidence is presented showing the sensitivity of the method to changes in the determinants of HFS. Data analysed during these three periods suggest that the method can be simplified further by using the food diversity score instead of the quantitative measure since these variables correlate highly with one another (r ≥ 0:854).

This simple method is a valid and precise measure of food security among poor urban households in Caracas. The qualitative/quantitative measures complement each other as they capture different dimensions of HFS.

In the 1980s evidence of rising levels of hunger was a concern for many, but disputed by some, Americans. Acknowledgement and quantification of hunger was hindered by the lack of an accepted definition and measure of hunger. Qualitative research at Cornell provided a conceptual framework, description, definition and survey items for hunger. The Community Childhood Hunger Identification Project developed an instrument used in numerous communities. Based upon these initiatives, widely accepted definitions of hunger and food insecurity, and the US Household Food Security Module for its measurement, now exist. The module classifies households as food-secure, or food-insecure without hunger or with moderate or severe hunger, and contains household-, adult- and child-referent items. Its inclusion in the Current Population Survey (CPS) since 1995 has yielded annual estimates of food insecurity. A six-item short form of the module, for surveys with severe time constraints, classifies households only as food-secure or food-insecure without or with hunger and contains no child-specific items. Surveys using the 18-item or short-form module can compare results with published national data from the CPS. Information about the module is available at http://www.ers.usda.gov/briefing/foodsecurity and http://www.fns.usda/fsec. Current research on food security measurement includes measurement of individual food insecurity and hunger, module performance regarding hunger duration and frequency, performance of the module in population subgroups, and the effect of translations on module meaning and performance. National surveys in Canada, New Zealand and Australia also have measured food security.

To review the various biomarkers of dietary intakes of fatty acids in human populations, their measurement, limitations and analytical considerations.

Review of the literature.

Although there is no good biomarker of intake of total fat, a number of alternatives exist for assessing the intakes of exogenously produced fatty acids that are consumed. Adipose tissue, erythrocyte membrane concentrations and serum or plasma levels can reflect prior intakes over the past few hours to the past few years. The concentrations of individual fatty acids in these media generally reflect relative levels, and are influenced by a number of factors. Although relatively expensive to analyse, a single analysis by gas chromatography or high-performance liquid chromatography provides information on multiple fatty acids, and is superior to attempting to measure specific fatty acids using traditional dietary assessment methods.

Biomarkers of fatty acids that reflect long-term intake are available for nutritional epidemiology purposes. Analytical methods have become very accurate and able to detect and quantify smaller families, such as trans-fatty acids.

To review existing methods and illustrate the use of a new, simple method for identifying inaccurate reports of dietary energy intake (rEI).

Comparison of rEI with energy requirements estimated by using total energy expenditure predicted (pTEE) from age, weight, height and sex using a previously published equation. Propagation of error calculations was performed and cut-offs for excluding rEI at plus or minus two standard deviations (±2 SD) and ±1 SD for the agreement between rEI and pTEE were established.

Dietary survey in a US national cohort: the Continuing Survey of Food Intakes by Individuals (CSFII), 1994–96.

Men and non-pregnant, non-lactating women aged 21–45 years in the CSFII who provided two multiple-pass 24-hour recalls, height and weight(n = 3755).

Average rEI was 77% of pTEE in men, and 64% of pTEE in women. Calculated cut-offs were rEI <40% or >160% of pTEE (±2 SD) and <70% or >130% of pTEE (±1 SD), respectively. Use of only the ±1 SD cut-offs, not the ±2 SD cut-offs, resulted in a relationship between rEI and body weight similar to what was expected (based on an independently calculated relationship between rEI and measured TEE). Exclusion of rEI outside either the ±2 SD (11% of subjects) or ±1 SD (57% of subjects) cut-offs did not affect mean reported macronutrient intakes, but did markedly affect relationships between dietary composition and body mass index.

When examining relationships between diet and health, use of ±1 SD cut-offs may be preferable to ±2 SD cut-offs for excluding inaccurate dietary reports.

To provide a framework for use of the doubly labelled water method to measure energy expenditure in order to validate dietary instruments for the assessment of energy.

Review and description of the use of doubly labelled water method for use as a biomarker for habitual energy intake.

The doubly labelled water method has a relative accuracy of 1% and within-subject precision of 5 to 8%. Comparison of self-reported energy intake with energy expenditure demonstrated that over one-third of individuals may underreport energy intake by more than 25%.

The doubly labelled water method, although expensive and dependent on non-routine laboratory instrumentation, is an excellent biomarker of energy intake.

The objectives of this paper are to (1) introduce the concept of upstream and downstream public health approaches and discuss diet assessment issues in that context, and (2) provide examples of diet assessment methods and challenges in assessing environmental factors influencing eating patterns.

Dietary assessment of environmental factors is discussed as they relate to nutrition interventions for school-aged children, although the issues transcend population characteristics. Examples of assessment challenges in measuring ‘dietary environments’ are drawn from the Child and Adolescent Trial for Cardiovascular Health (CATCH), Teens Eating for Energy and Nutrition at School (TEENS) and Trying Alternative Cafeteria Options in Schools (TACOS) studies.

The healthfulness of our ‘dietary environments’ (which may include food availability, social norms around food choice and the effect of pricing, policy and promotion on food choice) may be more important in determining what people consume than their individual decision-making about food choice. There is a dearth of published information to inform us on how to assess these ‘dietary environments’.

To give an overview of the evaluation of the modified diet history applied in the SENECA study (Survey in Europe on Nutrition and the Elderly; a Concerted Action).

Nineteen centres in 12 countries participated at baseline. Nine of these SENECA centres conducted a repeat measurement scheme in elderly people born between 1913 and 1918. These longitudinal centres included 100 subjects per sex per site.

The relative validity of the method was tested by comparing results of the modified diet history with results obtained from a weighed record in 82 subjects. In the follow-up we compared changes in energy intake with changes in body weight and calculated the physical activity ratio in all longitudinal centres. In SENECA's finale we examined the predictive value of dietary patterns observed at baseline for survival 10 years later, making use of the original and an adapted Mediterranean Diet Score.

The modified diet history overestimated intake, compared with the weighed record. However, the physical activity ratio and an in-depth study in a metabolic room indicated that the diet history rather underestimated energy intake. We did not find a relationship between changes in energy intake and changes in body weight, but this could be explained by the fact that most likely we did not measure intake in the dynamic phase of body weight change. Based on the adapted Mediterranean Diet Score, the study results showed a positive relation between a healthy diet and survival.

We conclude that the modified diet history has given sufficiently reliable results for the purposes of the study

To describe methodological issues regarding the measurement of food intake of marginal populations in the north-west of Mexico.

Dietary data from three sources were analysed. The Sonora State Food Basket Study (505 individuals from all ages, both sexes, high, medium and low income, and from urban and rural settings), 144 individuals from rural Indian communities, and 152 low-income urban women. Methods of dietary data collection were single 24-hour recalls for the first two studies, and four non-consecutive 24-hour recalls as well as a food-frequency questionnaire (FFQ) for low-income urban women.

The food patterns in the three studies showed similar core foods; however, food diversity resulted in 19, 13 and 57 different foods reported by 90% of the population in the state, Indian communities and urban women, respectively. Mean intakes of selected food components in the sample of urban women vs. a representative sample of the population of the state were similar, although smaller variation in consumption was observed in urban women with repeated 24-hour recalls. Inter-individual variation in intake of food components was similar in the state and urban women samples, but the Indian communities showed lower coefficients of variation (CVs). Use of four 24-hour recalls decreased the inter-individual variation in food component intakes of low-income urban women, especially for vitamin intakes. The FFQ did not show an additional decrease in inter-individual variation for macronutrients, but reduced even more the CVs for vitamins. Intra-individual variation was higher than inter-individual variation for urban women's intakes when estimated by repeated 24-hour recalls. This effect was reversed when estimated by duplicate food frequencies. Ratios of intra-individual to inter-individual variation in food intake were lower for Mexican when compared with US women. These results should be considered, especially for association analysis of low-income women's diets and health outcomes.

To evaluate measurement error structure in dietary assessment instruments and to investigate its implications for nutritional studies, using urinary nitrogen excretion as a reference biomarker for protein intake.

The dietary assessment methods included different food-frequency questionnaires (FFQs) and such conventional dietary-report reference instruments as a series of 24-hour recalls, 4-day weighed food records or 7-day diaries.

Six original pilot validation studies within the European Prospective Investigation of Cancer (EPIC), and two validation studies conducted by the British Medical Research Council (MRC) within the Norfolk cohort that later joined as a collaborative component cohort of EPIC.

A sample of approximately 100 to 200 women and men, aged 35–74 years, from each of eight validation studies.

In assessing protein intake, all conventional dietary-report reference methods violated the critical requirements for a valid reference instrument for evaluating, and adjusting for, dietary measurement error in an FFQ. They displayed systematic bias that depended partly on true intake and partly was person-specific, correlated with person-specific bias in the FFQ. Using the dietary-report methods as reference instruments produced substantial overestimation (up to 230%) of the FFQ correlation with true usual intake and serious underestimation (up to 240%) of the degree of attenuation of FFQ-based log relative risks.

The impact of measurement error in dietary assessment instruments on the design, analysis and interpretation of nutritional studies may be much greater than has been previously estimated, at least regarding protein intake.

Flavonoids are phytochemicals with potentially beneficial biological effects that are poorly characterised in existing tables of food composition.

To describe new techniques for analysis, absorption, informatics and dietary assessment that are important in measuring the flavonoid content in foods and in developing a flavonoid food composition database.

Data on chemical analyses of the major flavonoid compounds that exist in the food science literature are being located, collated and compiled into a preliminary flavonoid food composition database.

The analytical process begins with preparation of the food for flavonoid analysis using techniques to disrupt the food matrix, alcoholic extraction, and enzymatic or acidic hydrolysis to remove sugars. Separation is usually accomplished using high-performance liquid chromatography. Flavonoids are identified by diode array spectrometry, mass spectrometry or nuclear magnetic resonance spectroscopy. Quantification usually employs comparison of the sample with standards, if available, using the area under the curve of the chromatogram to estimate quantity. Absorption studies are useful since flavonoids vary in their absorption. Finally, information management technologies (informatics) are used to translate flavonoid data information into food composition databases. This process involves identification of foods containing the compounds of interest, collection and organisation of sources of existing analytical data, assignment of quality scores or aggregation of acceptable data for each component and food, calculation of appropriate statistics, assignment of food codes and verification, and finalisation. The resulting food flavonoid database can be used with state-of-the-art dietary assessment methods to develop estimates of flavonoid intakes in foods and to correlate these with estimates of disease risk.

A provisional flavonoid database, including at least two components from each of six classes of flavonoids, should be completed in 2002.

To evaluate the validity of a food-frequency questionnaire (FFQ) for assessment of the dietary intakes of polyunsaturated fatty acids (PUFAs) against a biochemical marker of fat intake, erythrocyte cell membrane phospholipid levels, during pregnancy.

Cross-sectional analysis.

Developmental Neurobiology Department, National Institute of Perinatology, Mexico City.

One hundred forty-six healthy pregnant women during the last trimester of pregnancy. Among women enrolled, the first 35 pregnant women (24%) had their erythrocytes analysed for fatty acid status.

We administered an FFQ and compared intakes of PUFAs against their erythrocyte cell membrane concentrations, processed by gas chromatography.

Pearson correlation coefficients among α-linolenic acid (ALN), docosahexaenoic acid (DHA) and eicosapentaenoic acid in erythrocyte cell membranes against their crude dietary counterparts were 0.32, 0.35 and 0.36 (each P < 0.05). In a simple linear regression, erythrocyte DHA and arachidonic acid (AA) were significantly related to their respective dietary intakes (β = 0.30, 95% confidence interval (CI): 0.007–0.60, P = 0.045 for DHA; β = 0.49, 95% CI: 0.010–0.98, P = 0.044 for AA). Erythrocyte cell membrane ALN concentration (%/total) was only marginally related to ALN dietary intake (mg day−1) (β = 0.52, 95% CI: −0.020–1.10, P = 0.061). However, after adjustment for long-chain n–3 PUFA/AA, this association reached significance (β = 0.44, 95% CI: 0.026–0.825, P = 0.038). Main dietary sources for n–3 PUFAs were canned tuna fish and fresh catfish; for n–6 these were eggs and cow's milk. The use of this FFQ in these pregnant Mexican women provided estimates of average long-term intakes of PUFAs and correlated reasonably well with their erythrocyte cell membrane phospholipid status. However, we need to consider that, during pregnancy, there is a faster turnover of PUFAs from fat storage that may modify the profile of erythrocyte PUFAs and lower the correlation between dietary intake and erythrocyte PUFAs.

Increasing recognition of the potential importance of phytochemicals in the aetiology of cancer and heart diseases has highlighted the need for methods to measure individual phytochemical consumption that are sufficiently simple to be used in large epidemiological studies and whose reproducibility and accuracy have been quantified. D-Limonene is a natural component of a variety of foods and beverages and is found mainly in citrus fruits. However, D-limonene is not assessed by any nationally available analysis database.

We designed our study to assess the D-limonene content of different citrus juices and beverages and to develop a dietary assessment instrument to measure consumption of citrus foods (fruit, juice and peel) and D-limonene intake and test it for reliability.

A total of 120 citrus juice samples were analysed and used to develop the preliminary D-limonene database. A self-administered citrus food-frequency questionnaire was developed and administered twice to participants, separated by a 2-month interval. The questionnaire was tested for reproducibility of estimates of citrus food consumption and D-limonene intake among 120 participants.

Correlation coefficients between the two administrations of the questionnaire ranged from 0.50 for citrus peel use to 0.82 for orange juice. Mean intakes (range) of D-limonene from citrus juices among consumers were 13.0 (0.24–141.9) mg day−1 and 13.2 (0.07–83.9) mg day−1 (r = 0.60, P < 0.001).

The citrus frequency questionnaire developed in this study provided highly reproducible estimates of citrus foods, citrus peel and D-limonene intakes. This instrument may be a useful tool in studies of the associations between citrus peel use, D-limonene intake and risk for chronic disease.