S Tokudome 1, N Imaeda 2, Y Tokudome 3, N Fujiwara 4, T Nagaya 1, J Sato 5, K Kuriki 1, M Ikeda 6 and S Maki 7

Transcription

1 (2001) 55, ß 2001 Nature Publishing Group All rights reserved /01 $ Original Communication Relative validity of a semi-quantitative food frequency questionnaire versus 28 day weighed diet records in Japanese female dietitians S Tokudome 1, N Imaeda 2, Y Tokudome 3, N Fujiwara 4, T Nagaya 1, J Sato 5, K Kuriki 1, M Ikeda 6 and S Maki 7 1 Department of Public Health, Nagoya City University Medical School, Mizuho-ku, Nagoya, Japan; 2 Nagoya City School of Nutrition, Mizuho-ku, Nagoya, Japan; 3 Nagoya-Bunri College, Nishi-ku, Nagoya, Japan; 4 Nagoya City University School of Nursing, Mizuho-ku, Nagoya, Japan; 5 Nagoya University Hospital, Showa-ku, Nagoya, Japan; 6 University of Occupational and Environmental Health, Yahatanishi-ku, Kitakyushu, Japan; and 7 the Aichi Prefectural Dietetic Association, Nishi-ku, Nagoya, Japan Objective: To assess the relative validity of a semi-quantitative food frequency questionnaire (SQFFQ) against 28 day weighed diet records (WDRs). Subjects and methods: The SQFFQ was administered to 106 (21 male and 85 female) Japanese dietitians in Aichi Prefecture in autumn, 1996 and four-season consecutive 7 day WDRs were carried out during We evaluated validity of intakes of 15 foods and 31 macro- and micro-nutrients based on the SQFFQ against those according to 28 day WDRs among 79 Japanese female dietitians. Results: Mean daily intakes of selected foods and nutrients determined by the SQFFQ were generally equivalent to those measured by 28 day WDRs. Pearson s de-attenuated correlation coefficients (CCs) with log-transformation and energy-adjustment between intakes of selected foods and nutrients quantified by the SQFFQ and 28 day WDRs (minimum median maximum) ranged from 0.17 (beverages) 0.52 to 0.74 (rice), and Spearman s rank CCs with energy-adjustment ranged from 0.28 (confectionery) 0.42 to 0.68 (rice). Respective Pearson s CCs for intakes of nutrients were 0.28 (PUFAs) 0.51 to 0.73 (magnesium), and Spearman s rank CCs ranged from 0.23 (n-3 PUFAs) 0.45 to 0.71 (magnesium). Favorably higher agreement for intakes of foods=nutrients was achieved along with lower disagreement. Conclusions: Satisfactorily higher relative validity was attained in Japanese female dietitians with the SQFFQ. This calibrated questionnaire seems therefore appropriate for administration to Japanese dietitians to clarify associations between diet and health=disease. Sponsorship: A grant-in-aid from the Ministry of Education, Science, Sports and Culture ( ). Descriptors: Japanese female dietitians; relative validity=calibration; semi-quantitative food frequency questionnaire; weighed diet records (2001) 55, Introduction *Correspondence: S Tokudome, Department of Public Health, Nagoya City University Medical School, Mizuho-ku, Nagoya , Japan. tokudome@med.nagoya-cu.ac.jp Guarantor: S Tokudome. Contributors: ST was the chief investigator and grant-holder, supervised the study, and prepared the manuscript. NI and YT computed intake of foods and nutrients according to the SQFFQ and 28 day WDRs, and analyzed the data. MI critically evaluated the data analysis. NF, TN, JS and KK were actively involved in collection of four-season consecutive 7 day WDRs. SM liaised with the study subjects. Received 14 November 2000; revised 9 January 2001; accepted 13 February 2001 In developed countries including Japan, the leading causes of death are chronic diseases such as cancer, cerebrovascular disease and heart disease. They are related to our daily lifestyle including food intake, alcohol drinking, smoking, physical exercise, and stress. Among these, food intake in particular appears to play a major role in the pathogenesis (WCRF=AICR, 1997); therefore, research on associations between diet and health=disease is of high priority. Observations, however, remain controversial because information on dietary consumption is not necessarily valid or reproducible due to several factors, which include validation in questionnaires, procedures to obtain dietary information and subjects studied. Above all, development of food frequency questionnaires=semi-quantitative food frequency questionnaires (abbreviated SQFFQs hereafter) deserves special attention, and relative validity and

2 736 reproducibility should always be evaluated prior to administration to study population (Margetts & Nelson, 1990; Willett, 1998). We have evolved a data-based SQFFQ to secure longterm dietary intake (Tokudome et al, 1998) and here conducted a validation=calibration of consumption of nutrients as well as foods based on the SQFFQ vs those according to four-season consecutive 7 day weighed diet records (abbreviated WDRs hereafter) as the gold standard (reference without correlated errors=conceptually different method) in sampled target Japanese dietitians, to whom the SQFFQ will be distributed to investigate clues for preventing disease, promoting health and prolonging life. Subjects and methods Subjects We recruited 106 (21 male and 85 female) middle-aged Japanese dietitians from the members of the Aichi Prefectural Dietetic Association living in Aichi Prefecture, Central Japan, for survey of the SQFFQ and four-season consecutive 7 day WDRs. Eighty female dietitians completed the SQFFQ and 28 day WDRs; however, one subject was excluded because her response on energy in the SQFFQ lay beyond 5 standard deviations from the mean. Because the number of male dietitians was rather small, they were excluded from the study. Accordingly, the results are shown for 79 female dietitians as mean age standard deviation (minimum maximum) of 48 8 (32 66). The values for height, weight and body mass index were cm 5.1 ( ), 52.2 kg 5.5 ( ) and ( ), respectively. SQFFQ and four-season 7 day WDRs We earlier developed an evidence-based SQFFQ according to cumulative contribution and multiple regression analysis as described elsewhere (Tokudome et al, 1998). The questionnaire inquired about habitual dietary intake during the previous month for 102 foods=recipes, portion size and food frequency in eight categories. Information on alcohol consumption was procured from a lifestyle questionnaire distributed together with the SQFFQ. In autumn 1996, we first mail-administered the SQFFQ to 106 middle-aged Japanese dietitians and self-administered consecutive 7 day WDRs approximately a week later to avoid education=learning effects. We asked them to hand in the 7 day WDRs at the end of the survey and sampled overnight-fast venous blood. Because the subjects were professional dietitians, the WDRs were satisfactorily filled out; however, research nutritionists reviewed their accuracy as well as completeness. Additional consecutive 7 day WDRs were carried out at about 3 month intervals in winter, spring and summer 1997 (Figure 1). Foods and nutrients selected We chose 15 foods=food groups, which include rice, bread, noodles and potatoes, confectionery, oil, soybean and soybean products, fish and other seafoods, meat, eggs, dairy products, green-yellow vegetables, other vegetables, seaweed, fruit, beverages and alcohol. We selected 31 macroand micro-nutrients, which include energy, protein, fat, carbohydrate, total dietary fiber (TDF; including soluble DF and insoluble DF), minerals (including potassium, calcium, magnesium, phosphorous, iron, zinc and copper) and vitamins (including carotenes and vitamins A, C, D and E). Fat was divided into saturated fatty acids (SFAs), monounsaturated fatty acids (MUFAs; including oleic acid), poly-unsaturated fatty acids (PUFAs), n-6 PUFAs, n-3 PUFAs and cholesterol. n-6 PUFAs were separated into linoleic acid (18:2n-6) and arachidonic acid (20:4n-6), and n-3 PUFAs into a-linolenic acid (18:3n-3), eicosapentaenoic acid (EPA, 20:5n-3) and docosahexaenoic acid (DHA, 22:6n-3). Figure 1 Schedule for validation study of the SQFFQ against four-season 7 day WDRs.

3 Calculation of intake of food=nutrient For WDRs, food weight was quantified before cooking if it was prepared at home, otherwise after cooking. Foods were generally weighed individually, while those like soup=stew were divided by the number of household members. Meals eaten out were referred to model recipes. For unmeasurable foods, standard portion size was applied. Research nutritionists computed mean daily intake of selected foods and nutrients by multiplying the food intake (in grams) or serving size and the nutrient content per gram of food as listed in the Standard Tables of Food Composition, version 4 and the Follow-up of the Standard Tables of Food Composition, (Science and Technology Agency, Japan, 1983, 1994). The Japanese food composition tables are, as is known, incomplete for selected nutrients. Estimation was mainly made according to analogies from same=similar foods and to assessment based on model recipes for relevant foods as reported (Imaeda et al, 2000). For fat and fatty acids, for example, the original coverage of 85% was improved to approximately 95%. With the SQFFQ, considering food frequency in eight categories (never or seldom, 1 3 times=month, 1 2 times= week, 3 4 times=week, 5 6 times=week, once=day, twice=day, and more than twice=day), we similarly estimated average daily intake of selected foods and nutrients by multiplying the food intake (in grams) or serving size according to the values in the literature and WDRs and the nutrient content per gram of food as listed in the respective composition tables or per gram of model recipe. Validation=calibration First, we calibrated mean daily intakes of 15 foods and beverages, and 31 macro- and micro-nutrients according to the SQFFQ against those based on the 1 y 28 day WDRs. Average ( s.d.) was based on the meanof individual s 28 day WDRs. The differences of means were examined by t-test. Second, we calculated Pearson s crude correlation coefficients (CCs), log-transformed CCs, energy-adjusted CCs with log-transformation (Beaton et al, 1993; Liu et al, 1978; Margetts & Nelson, 1990; Willett, 1998), and deattenuated CCs with log-transformation and energyadjustment, and Spearman s rank CCs between intakes of selected foods=nutrients based on 28 day WDRs and the SQFFQ. Statistical significance of CCs was verified along with 95% confidence intervals. Third, after categorizing daily intakes of foods=nutrients measured by 28 day WDRs and the SQFFQ into three, we computed percentages of exact agreement and complete disagreement, and kappa and weighted kappa statistics (Cohen, 1968; Fleiss, 1973). Results Intake of foods Mean daily intakes of foods assessed according to the SQFFQ were generally similar to those based on 28 day WDRs but the values for average consumption of confectionery, oil and eggs were smaller than with the four-season approach. The opposite was, however, the case for mean intakes of dairy products, green yellow vegetables, seaweed and fruit (Table 1). Pearson s de-attenuated CCs with log-transformation and energy-adjustment between intakes of foods (minimum median maximum) assessed by 28 day WDRs and the SQFFQ ranged from 0.17 (beverages) (rice) (Table 2), and Spearman s rank CCs with energy-adjustment were 0.28 (confectionery) (rice). 737 Table 1 Comparison of daily intakes of selected foods according to 28 day WDRs and the SQFFQ 28 day WDRs SQFFQ Percentage difference Food Mean s.d. Mean s.d. from 28 day WDRs Rice 267 (g) (g) % Bread, noodles and potatoes % Confectionery %** Oil %* Soybean and soybean products % Fish and other seafoods % Meat % Eggs %* Dairy products %** Green yellow vegetables %* Other vegetables % Seaweed %** Fruit %* Beverages % Alcohol % Total % Median 6% *P < 0.05; **P < 0.01.

4 738 Table 2 Pearson s and Spearman s rank correlation coefficients between daily intakes of selected foods based on 28 day WDRs and the SQFFQ Pearson s CCs Spearman s CCs Food Crude Logtransformed Log-transformed and energy-adjusted S 2 w=s 2 b De-attenuated, logtransformed and energy-adjusted Crude Energy-adjusted Rice ( ) Bread, noodles and potatoes ( ) Confectionery ( ) Oil ( ) Soybean and soybean products ( ) Fish and other seafoods ( ) Meat ( ) Eggs ( ) Dairy products ( ) Green yellow vegetables ( ) Other vegetables ( ) Seaweed ( ) Fruit ( ) Beverages ( ) Alcohol ( ) Median Percentages of exact agreement with energy adjustment ranged from 35% (seaweed) 50% 61% (rice) (Table 3). However, percentages of complete disagreement were 1% (rice and soy and soybean products) 12% 16% (confectionery). Kappa statistics ranged from 0.03 (seaweed) (rice and soybean and soybean products). Kappa statistics ranged from 0.03 (seaweed) (rice and soybean products). Weighted kappa statistics ranged from 0.19 (confectionery) (rice). Intake of nutrients In general, average intakes of nutrients according to the SQFFQ were similar to those based on 28 day WDRs. Mean intakes of vegetable fat and carbohydrate were smaller, whereas fish fat, protein energy percentage, EPA, DHA, vitamin A, vitamin C, vitamin D and calcium were larger (Table 4). Pearson s de-attenuated CCs with log-transformation and energy-adjustment between intakes of nutrients quantified by 28 day WDRs and the SQFFQ ranged from 0.28 (PUFAs) (magnesium) (Table 5), and Spearman s rank CCs with energy-adjustment were 0.22 (n-3 PUFAs) (magnesium). Percentages of exact agreement with energy adjustment ranged from 37% (arachidonic acid) 49% 62% (calcium) (Table 6). Percentages of complete disagreement were 3% (magnesium) 11% 16% (n-3 PUFAs). Kappa statistics Table 3 SQFFQ Agreement, disagreement and kappa statistics according to tertile classification of daily intakes of selected foods based on 28 day WDRs and the Crude Energy-adjusted Food Agreement (%) Disagreement (%) Kappa Weighted kappa Agreement (%) Disagreement (%) Kappa Weighted kappa Rice Bread, noodles and potatoes Confectionery Oil Soybean and soybean products Fish and other seafoods Meat Eggs Dairy products Green yellow vegetables Other vegetables Seaweed Fruit Beverages Median Alcohol was not included because the proportion of regular drinkers was less than 25%.

5 Table 4 Comparison of daily intakes of selected nutrients according to 28 day WDRs and the SQFFQ day WDRs SQFFQ Percentage difference Nutrient Mean s.d. Mean s.d. from 28 day WDRs Energy (kcal) % Total protein (g) % Animal origin (g) % Vegetable origin (g) %* Total fat (g) % Animal origin (g) % Fish origin (g) %** Vegetable origin (g) %** Carbohydrate (g) %* Protein energy% %* Fat energy% % Carbohydrate energy% %** SFAs (g) % MUFAs (g) % PUFAs (g) % Cholesterol (mg) % 18:1 (mg) % n-6 PUFAs (mg) % 18:2n-6 (mg) % 20:4n-6 (mg) % n-3 PUFAs (mg) % 18:3n-3 (mg) % 20:5n-3 (mg) %** 22:6n-3 (mg) %** Vitamin C (mg) %* Carotenes (mg) % Vitamin A (IU) %* Vitamin E (mg) % Vitamin D (IU) %** Potassium (mg) % Calcium (mg) %* Magnesium (mg) % Phosphorous (mg) % Iron (mg) % Zinc (mg) % Copper (mg) % Total dietary fiber (g) % Soluble dietary fiber (g) % Insoluble dietary fiber (g) % Median 2% *P < 0.05; **P < ranged from 0.05 (arachidonic acid) (calcium). Weighted kappa statistics ranged from 0.19 (arachidonic acid and n-3 PUFAs) (magnesium). Discussion Many FFQs=SQFFQs have so far been calibrated with multiple diet records, 24 h recalls and biochemical analyses (Margetts & Nelson 1991; Willett, 1998). We here referred studies in which questionnaires were developed to assess intake of foods and nutrients simultaneously, administered to adult populations, either public or professionals, free from apparent disease=infirmity and delivered prior to multiple diet records or 24 h recalls to avoid education=learning effects because we actually administer FFQs=SQFFQs in such a sequence to the subjects in large-scale prospective studies. With respect to consumption of nutrients as well as foods, the values for Pearson s CCs and Spearman s CCs in the present study were generally similar or higher than those reported around the world (Bohlscheid-Thomas et al, 1997a,b; EPIC Group of Spain, 1997a,b; Ocke et al, 1997a,b; Pietinen et al, 1988; Pisani et al, 1997; van Liere et al, 1997) and were lower than those earlier investigated (Bonifacj et al, 1997; Goldbohm et al, 1994). Although sources of variation were presumably minimized by the fact that the participants were

6 740 Table 5 Pearson s and Spearman s rank correlation coefficients between daily intakes of selected nutrients based on 28 day WDRs and the SQFFQ Pearson s CCs Spearman s CCs Nutrient Crude Logtransformed Log-transformed and energy-adjusted S 2 w=s 2 b De-attenuated, logtransformed and energy-adjusted Crude Energy-adjusted Energy ( ) 0.42 Total protein ( ) Animal origin ( ) Vegetable origin ( ) Total fat ( ) Animal origin ( ) Fish origin ( ) Vegetable origin ( ) Carbohydrate ( ) Protein energy% ( ) Fat energy% ( ) Carbohydrate energy% ( ) SFAs ( ) MUFAs ( ) PUFAs ( ) Cholesterol ( ) : ( ) n-6 PUFAs ( ) :2n ( ) :4n ( ) n-3 PUFAs ( ) :3n ( ) :5n ( ) :6n ( ) Vitamin C ( ) Carotenes ( ) Vitamin A ( ) Vitamin E ( ) Vitamin D ( ) Potassium ( ) Calcium ( ) Magnesium ( ) Phosphorous ( ) Iron ( ) Zinc ( ) Copper ( ) Total dietary fiber ( ) Soluble dietary fiber ( ) Insoluble dietary fiber ( ) Median professional dietitians, it is well known that there is a wide diversity of Japanese foods. Thus, Japanese FFQs=SQFFQs for foods (Wakai et al, 1999) and nutrients (Egami et al, 1999; Katagiri et al, 1998; Sasaki et al, 1998; Takatsuka et al, 1996) seem to perform as a whole rather less well than those in other regions of the world, cited above. In addition, the somewhat lower figures for validity than those expected for certain foods=nutrients which might partly have been because our SQFFQ was admittedly based on a rather small number of 351 individuals surveyed in autumn. We admit that there is a discrepancy in the survey periods between the SQFFQ during a previous month and four-season consecutive 7 day WDRs. We believe, however, that people s memory=cognition is not necessarily trustworthy; therefore, 1 month or 3 months may be maximum length for good recall in responding to the SQFFQ. We actually observed daily, weekly, seasonal and personal (between- and within-) variations in the 28 day WDRs (data not shown); however, day-to-day withinindividual variations, which were larger than betweenindividual fluctuations for most foods and nutrients, could not be completely adjusted for, even though de-attenuated CCs with log-transformation and energy adjustment were

7 Table 6 SQFFQ Agreement, disagreement and kappa statistics according to tertile classification of daily intakes of selected nutrients based on 28 day WDRs and the 741 Crude Nutrient Agreement (%) Disagreement (%) Kappa Energy-adjusted Weighted kappa Agreement (%) Disagreement (%) Kappa Weighted kappa Energy Total protein Animal origin Vegetable origin Total fat Animal origin Fish origin Vegetable origin Carbohydrate Protein energy% Fat energy% Carbohydrate energy% SFAs MUFAs PUFAs Cholesterol : n-6 PUFAs :2n :4n n-3 PUFAs :3n :5n :6n Vitamin C Carotenes Vitamin A Vitamin E Vitamin D Potassium Calcium Magnesium Phosphorous Iron Zinc Copper Total dietary fiber Soluble dietary fiber Insoluble dietary fiber Median computed. Moreover, seasonal variation was not fully taken into account in this validation study. Some lower correlation coefficients in terms of micronutrients including vitamins and minerals may have been associated with this seasonal variation. In this connection, we mail-administered the SQFFQ to the study subjects in autumn 1999, 2000 and We are particularly interested in fat-related cancers and chronic diseases, since morbidity of and mortality from such causes are increasing in Japan. We therefore had set detailed=duplicated items pertaining to fat and oil consumption, including brand names of vegetable oil and eat at homenutrients as compared to earlier studies. We are now performing validation=calibration of the SQFFQ against biochemical markers of plasma fatty acid concentration and composition. Some fat-related items were combined=dedeleted in the final questionnaire actually administered to dietitians in Japan. As a whole, the absolute values ascertained by the SQFFQ and 28 day WDRs demonstrated small differences. Pearson s CCs for intakes of certain foods=nutrients were similar to Spearman s rank CCs; however, the latter were generally higher. In addition, according to tertile classification, figures for agreement and disagreement,

8 742 and kappa statistics along with weighted kappa statistics were essentially favorable. In other words, our SQFFQ appeared satisfactorily competent not only to ensure appropriate ranking=classification of individuals according to consumption of foods=nutrients but also to yield reasonable relative risk=odds ratios and dose-response relationships for epidemiological studies on diet and health=disease (Margetts & Nelson, 1990; Rothmann & Greenland, 1998; Willett, 1998). The values for relative validity of our SQFFQ vs 28 day WDRs were, as mentioned, satisfactory due in part to the fact that the study subjects were Japanese dietitians. In contrast, desirable responses to the SQFFQ by Japanese dietitians might have occurred because they are naturally health conscious and have knowledge=cognition about healthful foods and nutrients. It may be a criticism that their choice as subjects means that the results cannot be generalized. We respond that internal validity (relative validity and reproducibility) is a priori indispensable (Rothmann & Greenland, 1998) and external validity is not attained unless the method is internally valid. Bearing these strengths and weaknesses in mind, we proposed a Japanese Dietitians Epidemiologic (JADE) Study and delivered the SQFFQ and lifestyle questionnaire to dietitians to evaluate the associations between diet and health=disease. This is, to our knowledge, a new approach. Acknowledgements The authors thank the participants and working staff of the Aichi Prefectural Dietetics Association for their participation in this study, and Ms Y Kubo, Ms Ito and Dr MA Moore for their technical and language assistance in preparing this manuscript. References Beaton GH, Milner J, McGuire V, Feather TE & Little JA (1983): Source of variance in 24-hour dietary recall data: implications for nutrition study design and interpretation. Carbohydrate sources, vitamins and minerals. Am. J. Clin. Nutr. 37, Bohlscheid-Thomas S, Hoting I, Boeing H & Wahrendorf J (1997a): Reproducibility and relative validity of food group intake in a food frequency questionnaire developed for the German part of the EPIC project. Int. J. Epidemiol. 26(Suppl 1), S59 S70. Bohlscheid-Thomas S, Hoting I, Boeing H & Wahrendorf J (1997b): Reproducibility and relative validity of energy and macronutrient intake of a food frequency questionnaire developed for the German part of the EPIC project. Int. J. Epidemiol. 26(Suppl 1), S71 S81. Bonifacj C, Gerber M, Scali J & Daures JP (1997): Comparison of dietary assessment methods in a Southern French population: use of weighed records, estimated-diet records and a food-frequency questionnaire. Eur. J. Clin. Nutr. 51, Cohen J (1968): Weighted kappa: nominal scale agreement with provision for scaled disagreement or partial credit. Psychol. Bull. 70, Egami I, Wakai K, Kato K, Lin Y, Kawamura T, Tamakoshi A, Aoki R, Kojima M, Nakayama T, Wada M & Ohno Y (1999): A simple food frequency questionnaire for Japanese diet. Part II. Reproducibility and validity for nutrient intakes. J. Epidemiol. 9, EPIC Group of Spain (1997a): Relative validity and reproducibility of a diet history questionnaire in Spain. I. Foods. Int. J. Epidemiol. 26(Suppl 1), S91 S99. EPIC Group of Spain (1997b): Relative validity and reproducibility of a diet history questionnaire in Spain. II. Nutrients. Int. J. Epidemiol. 26(Suppl 1), S100 S109. Fleiss JL (1973): Statistical Methods for Rates and Proportions. New York: John Wiley. Goldbohm RA, van den Brandt PA, Brants HAM, van t Veer P, Al M, Sturmans F & Hermus RJJ (1994): Validation of a dietary questionnaire used in a large-scale prospective cohort study on diet and cancer. Eur. J. Clin. Nutr. 48, Imaeda N, Tokudome Y, Fujiwara N, Nagaya T, Kamae M, Tsunekawa S, Sato, N, Tokizane M, Koide Y, Miyai Y, Maki S & Tokudome S (2000): Data checking and standardization in a weighed food dietary record survey. Jpn. J. Nutr. 58, Katagiri A, Hashimoto S, Ohashi Y, Shirogane K, Sakamoto N & Makimoto S (1998): Reproducibility and validity of a semi-quantitative food frequency questionnaire. Nippon Koshu Eisei Zasshi 45, (in Japanese). Liu K, Stamler J, Dyer A, McKeever J & McKeever P (1978): Statistical methods to assess and minimize the role of intra-individual variability in obscuring the relationship between dietary lipids and serum cholesterol. J. Chron. Dis. 31, Margetts BM & Nelson, M (1991): Design Concepts in Nutritional Epidemiology. Oxford: Oxford University Press. Ocké MC, Bueno-de-Mesquita HB, Goddijn HE, Jansen A, Pols MA, van Staveren WA & Kromhout D (1997a): The Dutch EPIC food frequency questionnaire. I. Description of the questionnaire, and relative validity and reproducibility for food groups. Int. J. Epidemiol. 26(Suppl 1), S37 S48. Ocké MC, Bueno-de-Mesquita HB, Pols MA, Smit HA, van Staveren WA & Kromhout D (1997b): The Dutch EPIC food frequency questionnaire. II. Relative validity and reproducibility for nutrients. Int. J. Epidemiol. 26(Suppl 1), S49 S58. Pietinen P, Hartman AM, Haapa E, Räsänen L, Haapakoski J, Palmgren J, Albanes D, Virtamo J & Huttunen JK (1988): Reproducibility and validity of dietary assessment instruments. I. A self-administered food use questionnaire with a portion size picture booklet. Am. J. Epidemiol. 128, Pisani P, Faggiano F, Krogh V, Palli D, Vineis P & Berrino F (1997): Relative validity and reproducibility of a food frequency dietary questionnaire for use in the Italian EPIC centres. Int. J. Epidemiol. 26(Suppl 1), S152 S160. Rothmann KJ & Greenland S (1998): Modern Epidemiology, 2nd edn. Philadelphia: Lippincott-Raven. Sasaki S, Yanagibori R & Tsugane S (1998): Self-administered diet history questionnaire developed for health education: a relative validation of the test-version by comparison with 3-day diet record in women. J. Epidemiol. 8, Science and Technology Agency, Japan (1983): Standard Tables of Food Composition in Japan. 4th edn. Tokyo: Ministry of Finance (in Japanese). Science and Technology Agency, Japan (1994): Follow-up of Standard Tables of Food Composition in Japan. Tokyo: Ishiyaku Shuppan (in Japanese). Takatsuka N, Kawakami N, Kawai K, Okamoto Y, Ishiwata K & Shimizu H (1996): Validation of recalled food intake in the past in a Japanese population. J. Epidemiol. 6, Tokudome S, Ikeda M, Imaeda N, Tokudome Y, Kitagawa I & Fujiwara N (1998): Development of data-based semi-quantitative food frequency questionnaire for dietary studies in middle-aged Japanese. Jpn. J. Clin. Oncol. 28, van Liere MJ, Lucas F, Clavel F, Slimani N & Villeminot S (1997). Relative validity and reproducibility of a French dietary history questionnaire. Int. J. Epidemiol. 26(Suppl 1), S128 S136. Wakai K, Egami I, Kato K, Lin Y, Kawamura T, Tamakoshi A, Aoki R, Kojima M, Nakayama T, Wada M & Ohno Y (1999): A simple food frequency questionnaire for Japanese diet. Part I. Development of the questionnaire, and reproducibility and validity for food groups. J. Epidemiol. 9, Willett W (1998): Nutritional Epidemiology, 2nd edn. New York: Oxford University Press. WCRF=AICR (1997): Food, Nutrition and the Prevention of Cancer: a Global Perspective. Washington, DC: World Cancer Research Fund=American Institute for Cancer Research.