Effect of Cooking Loss in the Assessment of Vitamin Intake for Epidemiological Data in Japan

ORIGINAL ARTICLE Effect of cooking loss in the assessment of vitamin intake for epidemiological data in Japan

M Kobayashi1,2, HY Adachi2, J Ishihara2,3 and S Tsugane2 for the JPHC FFQ Validation Study Group

1Department of Food Science, Faculty of Home Economics, Otsuma Women’s University, Tokyo, Japan; 2Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan and 3Department of Nutrition, Junior College of Tokyo University of Agriculture, Tokyo, Japan

Background/Objectives: The effect of cooking loss on vitamin intake is an important consideration in dietary and epidemiological studies in Japanese. However, because few published food values have considered cooking effect, allowing for cooking loss in the assessment of vitamin intake in Japan has been difficult. Subjects/Methods: Seven-day dietary records and a fasting blood sample were collected from 102 men and 113 women in August of 1994 or 1995. Vitamin intake were estimated using two food databases, one composed of raw food only and the second of cooked food. Estimates were compared with blood levels. Results: Water-soluble vitamin intake using a food database including cooked food was lower than intakes estimated using a database composed of raw food only, except for pantothenic acid and vitamin B12 intake. In particular, vitamin B1 intake was 18.9% lower in men and 16.8% lower in women. However, when subjects were classified into the same and adjacent categories by joint classification by quintiles, appreciable change in ranking of a subject was not observed. Furthermore, the relationship between vitamin intake and biomarker did not improve when intake was calculated using a food database including cooked food. Conclusion: Although the effect of cooking loss on absolute values is not negligible, this might not significantly impact the ranking of subject intake estimations of vitamin intake in epidemiological studies. European Journal of Clinical Nutrition (2011) 65, 546–552; doi:10.1038/ejcn.2011.10; published online 23 February 2011

Keywords: cooking loss; vitamin intake assessment; food composition database; epidemiology

Introduction Cooking procedures in Japan are diverse and often complicated. Vitamin-rich vegetables are commonly boiled, Dietary evaluation methods have an important role in stewed or stir-fried. The Japan Public Health Center- epidemiological study of the relationship between dietary based (JPHC) Prospective study, for example, a large-scale habits and the risk of lifestyle-related diseases. Evaluation of population-based prospective study, reported that the most dietary assessment requires the evaluation of factors, which frequent method for cooking vegetables among 22 528 men may impact that assessment. One of the most important and 25 310 women was stir-frying (39.4% in men, 38.4% in factors is the effect of different cooking procedures women), followed by boiling or stewing (35.5% in men, on nutrient changes in food: because the vitamin and 29.0% in women) (Tsugane et al., 2001). The effect of mineral content values of foods are influenced by cooking cooking loss on vitamin intake in Japanese is considerable. procedures, accurate evaluation of nutrient intake is The revised and enlarged fifth edition of the standard hampered when the nutrient content values of raw food tables of food composition in Japan (FCT5) (The Council of are used. Science and Technology Agency Ministry of Education, 2005) added the composition of cooked foods. However, only 247 foods were included, hampering the consideration Correspondence: Dr M Kobayashi, Department of Food Science, Faculty of of cooking loss in the assessment of vitamin intake in Home Economics, Otsuma Women’s University, 12, Sanban-cho, chiyoda-ku, Japan. In response, a more detailed food composition Tokyo 102-8357, Japan. table was recently developed with a wider range of foods E-mail: [email protected] Received 19 March 2011; accepted 9 January 2011; published online 23 for which cooking loss should be considered (Wakiyama February 2011 et al., 2009). Effect of cooking loss on vitamin intake M Kobayashi et al 547 Here, we used this updated food database to calculate and a second, which included cooked food also (FDB_cook). vitamin intake and evaluate the effect of cooking procedures Calculations for the former were done using values for raw in comparison with intake calculated using the raw food- foods in FCT5, and for the latter using values for foods, only data in FCT5. In addition, we also compared estimated including cooked foods, in a recently developed food vitamin intake with biological indicators, such as serum database. This database is a compensated database, which carotenoid and plasma vitamin C, B6,B12 and folate considers cooking loss, and has been described in detail concentrations. elsewhere (Wakiyama et al., 2009). In brief, the cooking methods considered were boiling, grilling, deep- and stir- frying, steaming, kippering, or pickling. The quantity of Subjects and methods cooking loss for each food was calculated as follows: = Subjects NC ac ¼ðNC bc fct5 NC ac fct5ÞÂNC bc Subjects were a subsample of participants in the JPHC where study, an ongoing cohort study of 140 420 participants in 11 NC_ac: nutrient composition of the target food after prefectural public health center areas investigating cancer, cooking. cardiovascular disease and other lifestyle-related diseases NC_ac_fct5: nutrient composition of a related food (Watanabe et al., 2001). A total of 247 volunteers were after cooking as described in FCT5. recruited from the Ninohe, Yokote, Saku and Chubu NC_bc: nutrient composition of the target food in (previously named Ishikawa) public health center areas for a its raw state as described in FCT5. validation study of a food frequency questionnaire (Tsubono NC_bc_fct5: nutrient composition of a related food et al., 1996). For the present report, we analyzed the data of as described in FCT5. 215 subjects (102 men and 113 women) who provided complete 7-day dietary records (DRs). Of these 215 subjects, For example, pork ham (grilled) ¼ (pork loin (grilled)/pork 191 (90 men and 101 women) provided a blood sample. loin (raw)) Â pork ham (raw). Where necessary, substitute foods with similar botanical or zoological relevance were used. For foods having no Data collection reference in the FCT5, we used a food composition table of The data was collected as part of the validation study of the the United States Department of Agriculture and performed food frequency questionnaire used to assess dietary intake of a similar calculation as when using FCT5. Weight changes of the main cohort. Details of the study design and data foods after cooking were calculated using the table of weight collection have been described elsewhere (Tsugane et al., change rate in the FCT5 (The Council of Science and 2003). In brief, subjects completed DRs for 7 consecutive Technology Agency Ministry of Education, 2005). Nutrient days in August 1994 (Ninohe, Yokote and Saku) or August intakes were then calculated using weight change rate and 1995 (Chubu). Local dietitians instructed the subjects to nutrient composition values after cooking. weigh all foods and beverages with the scales and measuring utensils provided, and to record the results in a specially designed booklet. The subjects were instructed to use Statistical analysis standardized portion sizes for some foods that were difficult Vitamin intakes (retinol, a-carotene, b-carotene, lycopene, to weigh (semi-weighed DRs). The subjects described each vitamin D, vitamin E, vitamin K, vitamin B1, vitamin B2, food, method of cooking preparation and names of dishes in niacin, vitamin B6, vitamin B12, folate, pantothenic acid and detail. The DRs were reviewed in a standardized manner, and vitamin C) were calculated using these two databases for all foods described in a DR were coded as raw food by local each subject. Nutrient losses were estimated by subtracting dietitians. the values calculated with FDB_cook from those calculated Peripheral blood was collected just after the summer with FDB_raw, and the differences were tested by the paired session. Plasma and serum samples were stored in separate t-test. tubes at À80 1C until analysis. Serum carotenoid concentra- Estimated intakes of vitamin C, a-carotene, b-carotene, tions were measured for men and women, while plasma lycopene, vitamin B6, vitamin B12 and folate were adjusted vitamin B6,B12 and folate were measured for men only. for total energy intake using the residual model (Willett and Measurement of blood levels of carotene, ascorbic acid, Stampfer, 1986), and the subjects were classified into the folate, vitamin B6 and vitamin B12 were as described same and adjacent categories by joint classification by previously (Iso et al., 2003; Kobayashi et al., 2003). quintiles. Spearman’s rank correlation coefficient was used to assess the associations between vitamin intakes estimated from FDB_cook and FDB_raw. Calculation of nutrient intake Further, vitamin intakes estimated from FDB_cook We created two databases, one to calculate nutrient intake and FDB_raw were compared with their serum/plasma con- using a food database composed of raw food only (FDB_raw), centrations using Spearman’s rank correlation coefficient.

European Journal of Clinical Nutrition Effect of cooking loss on vitamin intake M Kobayashi et al 548 All statistical analyses were calculated using SAS (version 9.1; vitamins B1,B2,B6, C, niacin and folate were lower with SAS Institute Inc., Cary, NC, USA). FDB_cook than with FDB_raw, except for pantothenic acid

and vitamin B12 intake. In particular, vitamin B1 intake was 18.9% lower in men and 16.8% lower in women. Comparison of FDB-raw with FDB-cook for daily nutrient Results and vitamin intake based on joint classification by quintiles is shown in Table 4. The lowest percentage of classification The number of food items showing a cooking loss using the into the same quintile was 46.1% for a-carotene, and the

actual food values and those showing a loss using substituted highest percentage was 69.6% for vitamin B12 in men. food items are shown in Table 1. Among 463 substituted Respective values were 55.8% for a-carotene and 76.5% for foods, 86 foods (18.6%) were substituted using United States lycopene in women. The percentage of classification into an Department of Agriculture data. Table 1 also shows the adjacent category was more than 80% in men and more than number of items considered to show a cooking preparation- 90% in women. Spearman’s rank correlation coefficient dependent loss. After the loss considered, nutrient contents between vitamin intakes estimated from FDB_cook and could be re-calculated for 710 food items: 207 grilled, 287 FDB_raw were more than 0.80 in both men and women. boiled, 53 deep-fried, 96 stir-fried, 37 steamed, 27 pickled and 3 kippered foods. Characteristics of study participants are shown in Table 2. Table 2 Characteristics of study participants On the basis of characteristics, 14.7 and 13.3% of men and 2 women were overweight (body mass index X27 kg/m ), Men (n ¼ 102) Women (n ¼ 113) respectively, while 6.9 and 5.3% were underweight Mean (±s.d.) Mean (±s.d.) (body mass index o19 kg/m2), respectively. Smoking rate and mean alcohol consumption were higher in men than in Age (years) 55.6 (±5.2) 53.3 (±5.3) Height (cm) 164.5 (±4.9) 151.1 (±5.5) women. Body weight (kg) 65.8 (±9.3) 54.6 (±8.0) Estimated daily energy, nutrient and vitamin intake are BMI (kg/m2) 24.3 (±3.0) 23.9 (±3.1) shown in Table 3. Mean estimated energy and macro nutrient intake using FDB_cook were not much different Smoking status (%) Never 51.5 100.0 from those using FDB_raw. With regard to fat-soluble Past 19.2 0.0 vitamins, although retinol, lycopene and vitamin D intake Current 29.3 0.0 were lower with FDB_cook than with FDB_raw that of Alcohol consumption (g per day) 22.7 (±23.3) 1.7 (±3.1) b-carotene, vitamin E and vitamin K showed little difference between databases. In contrast, intake of the water-soluble Abbreviation: BMI, body mass index.

Table 1 Number of food items showing a cooking loss by food group as described in the FCT5a

Food items Number of foods Number of foods with (A) þ (B) originally showing substituted alues showing a cooking loss (A) a cooking loss (B) Grilled Boiled Deep-fried Stir-fried Steamed Kippered Pickled Total

Cereals 28 27 12 17 0 3 23 0 0 55 Potatoes and starches 9 9 1 10 3 2 2 0 0 18 Sugars 0 0 0 0 0 0 0 0 0 0 Pulses 8 4 0 9 2 1 0 0 0 12 Nuts and seeds 2 0 0 2 0 0 0 0 0 2 Vegetables 108 112 25 102 12 51 3 0 27 220 Fruits 0 3 1 2 0 0 0 0 0 3 Fungi 11 16 6 12 1 8 0 0 0 27 Algae 0 12 0 12 0 0 0 0 0 12 Fish and shellfish 65 95 71 50 22 11 3 3 0 160 Meats 14 180 89 68 13 19 5 0 0 194 Eggs 2 5 2 3 0 1 1 0 0 7 Milks 0 0 0 0 0 0 0 0 0 0 Fats and oils 0 0 0 0 0 0 0 0 0 0 Sweeteners 0 0 0 0 0 0 0 0 0 0 Beverages 0 0 0 0 0 0 0 0 0 0 Seasonings and spices 0 0 0 0 0 0 0 0 0 0 Total 247 463 207 287 53 96 37 3 27 710

aStandard tables of food composition in Japan, fifth revised and enlarged edition.

European Journal of Clinical Nutrition Effect of cooking loss on vitamin intake M Kobayashi et al 549 Table 3 Estimated daily energy, nutrient and vitamin intake using food composition databases composed of raw food only or including cooked food

Men (n ¼ 102) Women (n ¼ 113)

Database Database Mean % P-value Database Database Mean % P-value with including difference Changes with including difference Changes raw fooda cooked foodb raw fooda cooked foodb Mean±s.d. Mean±s.d. Mean±s.d. Mean±s.d.

Energy and nutrient Energy (MJ) 8.6±2.5 8.9±2.6 0.3 3.4 o0.0001 7.0±2.1 7.3±2.2 0.3 4.4 o0.0001 Protein (g) 76.2±23.4 75.9±24.2 À0.3 À0.1 0.7480 67.8±20.7 69.3±26.8 1.5 2.1 0.3618 Fat (g) 53.9±16.0 54.9±15.9 1.1 2.6 o0.1019 50.5±17.5 52.4±19.1 1.9 3.8 0.0053 Carbohydrate (g) 282.2±103.6 294.6±106.7 12.4 4.6 o0.0001 234.4±86.0 244.1±89.5 9.6 4.2 o0.0001

Vitamin Retinol (mg) 447±1003 228±322 À218.1 À17.5 0.0154 314±498 218±263 À96.0 À10.8 0.0073 a-Carotene (mg) 997±1994 1095±1800 98.1 9.8 0.0116 826±1324 893±1164 67.3 8.1 0.0465 b-Carotene (mg) 3413±5041 3484±7137 70.8 À6.4 0.8107 3309±4114 3443±6007 134.1 À4.2 0.5642 Lycopen (mg) 3408±3756 2835±3690 À359.1 À4.7 0.0036 3877±4217 3126±3661 À472.3 À11.8 0.0030 Vitamin D (mg) 11.1±14.3 8.5±8.4 À2.6 À15.4 0.0118 11.1±22.8 7.7±7.2 À3.4 À16.4 0.0611 Vitamin E (mg) 10.6±3.5 10.4±4.6 À0.3 À4.0 0.1563 10.4±4.3 10.2±4.8 À0.2 À2.6 0.2749 Vitamin K (mg) 345.1±264.4 363.4±415.6 18.3 5.6 0.3610 360.7±273.6 374.4±378.1 13.8 3.5 0.3988 Vitamin B1 (mg) 1.06±0.40 0.86±0.38 À0.2 À18.9 o0.0001 0.98±0.45 0.81±0.40 À0.2 À16.8 o0.0001 Vitamin B2 (mg) 1.37±0.56 1.22±0.51 À0.2 À10.5 o0.0001 1.26±0.47 1.14±0.47 À0.1 À9.5 o0.0001 Niacin (mg) 20.0±9.9 16.7±10.1 À3.3 À17.6 o0.0001 16.9±8.7 14.2±8.7 À2.7 À16.9 o0.0001 Vitamin B6 (mg) 1.65±0.83 1.41±0.81 À0.2 À15.8 o0.0001 1.40±0.66 1.20±0.66 À0.2 À15.0 o0.0001 Vitamin B12 (mg) 12.0±10.9 12.1±12.2 0.0 À2.6 0.9441 10.2±8.5 10.8±11.8 0.5 4.4 0.4526 Folate (mg) 396.5±173.9 325.4±151.1 À71.1 À16.3 o0.0001 378.5±188.0 314.1±156.8 À64.4 À14.5 o0.0001 Pantothenic acid (mg) 8.7±6.0 8.5±6.2 À0.2 À2.3 0.3313 8.1±5.6 7.9±5.8 À0.1 À2.2 0.3342 Vitamin C (mg) 104.0±57.9 93.3±55.8 À10.7 À8.3 o0.0001 118.2±70.8 105.7±69.8 À12.4 À9.8 o0.0001 aNutrient values calculated using a food database composed of raw food only. bNutrient values calculated using a database composed of raw and cooked foods.

The relationship between serum carotenoid and plasma Japanese dietary habits are rich in food variety. The FCT5, vitamin C, B6,B12 and folate concentrations and their for example, describes more than 380 types of fish and corresponding vitamin intakes using FDB_cook and FDB_raw shellfish, and 320 vegetables. Japanese cooking methods also are shown in Table 5. The correlation coefficient between vary widely, and include raw consumption, pickling, boiling, intake of a-carotene, a fat-soluble vitamin and serum steaming, grilling, stir-frying, deep-frying and others. In the a-carotene level improved using FDB_cook in both men JPHC study, 35.2% of men and 41.8% of women answered and women. For other vitamins, in contrast, correlation that they most frequently boil vegetables (Tsugane et al., coefficient did not improve when intake was calculated 2001). Boiling is widely known to result in large losses of using FDB_cook. water-soluble vitamins (Kimura et al., 1990), however, and cooking losses should be considered when daily vitamin intake is estimated. Discussion Large cooking losses of water-soluble vitamins are widely known. Among examples, Han reported an average folate In this study, we observed that vitamin intake estimated loss in 32 foods caused by cooking of 29% (Han et al., 2005); from DR was influenced by the food composition database Kimura et al., 1990 reported an average thiamine loss from used to calculate the vitamin levels of foods. Specifically, the cooking of about 40%; while Mareschi et al., 1983 reported calculated intake of water-soluble vitamins was lower using a that cooking peeled potatoes caused a vitamin C loss of database of cooked foods (FDB_cook) than one of raw foods roughly 40–50%. Agte et al. reported that the cooking losses (FDB_raw). However, when vitamin intake calculated by of ascorbic acid, thiamine, riboflavin and folic acid were FDB-raw was compared with that by FDB-cook based on joint 34.6, 30.0, 52.2 and 32.2%, respectively, while Rumm- classification by quintiles, the percentage of classification Kreuter also observed large cooking losses of ascorbic acid into the adjacent category was high. Further, the relation- and thiamine (Rumm-Kreuter and Demmel, 1990; Agte et al., ship between estimated vitamin intake using FDB_cook and 2002). Consistent with these findings, we also saw higher FDB_raw was also high. The relationship between vitamin loss of water-soluble vitamin intake when intake was calcu- intake and biomarkers did not improve when intake was lated using FDB_cook. However, when our subjects were calculated using DB_cook, except for a-carotene. classified into the same and adjacent categories by joint

European Journal of Clinical Nutrition Effect of cooking loss on vitamin intake M Kobayashi et al 550 Table 4 Comparison of a food composition database composed of raw food only (FDB-raw) with that composed of raw and cooked food (FDB-cook) for daily nutrient and vitamin intake based on joint classification by quintiles (%), and correlations

Men (n ¼ 102) Women (n ¼ 113)

Same Adjacent Spearman’s Same Adjacent Spearman’s category a category b correlation category category correlation r (95% CI) r (95% CI)

Energy and nutrient Protein 58.8 94.1 0.89 (0.85, 0.94) 59.3 95.6 0.86 (0.79, 0.94) Fat 73.5 99.0 0.94 (0.91, 0.97) 74.3 97.3 0.95 (0.92, 0.97) Carbohydrate 70.6 100.0 0.95 (0.93, 0.97) 64.6 98.2 0.94 (0.92, 0.96)

Vitamin Retinol 62.7 92.2 0.79 (0.66, 0.93) 73.5 95.6 0.89 (0.82, 0.97) a-carotene 46.1 91.2 0.82 (0.74, 0.91) 55.8 91.2 0.82 (0.74, 0.90) b-carotene 64.7 86.3 0.92 (0.87, 0.97) 73.5 98.2 0.94 (0.91, 0.98) Lycopene 68.2 96.6 0.95 (0.91, 0.99) 76.5 99.0 0.96 (0.94, 0.98) Vitamin D 65.7 95.1 0.91 (0.87, 0.95) 67.3 95.6 0.92 (0.88, 0.96) Vitamin E 66.7 100.0 0.96 (0.93, 0.98) 67.3 99.1 0.95 (0.92, 0.97) Vitamin K 68.6 92.2 0.91 (0.86, 0.96) 69.9 92.9 0.92 (0.87, 0.96) Vitamin B1 53.9 95.1 0.88 (0.83, 0.93) 63.7 95.6 0.84 (0.75, 0.93) Vitamin B2 61.8 93.1 0.85 (0.77, 0.92) 59.3 94.7 0.85 (0.78, 0.92) Niacin 56.9 93.1 0.86 (0.81, 0.92) 56.6 90.3 0.83 (0.74, 0.91) Vitamin B6 58.8 96.1 0.93 (0.89, 0.96) 70.8 99.1 0.93 (0.90, 0.96) Vitamin B12 69.6 99.0 0.91 (0.83, 0.99) 68.1 96.5 0.89 (0.82, 0.97) Folate 60.8 93.1 0.86 (0.79, 0.93) 66.4 95.6 0.88 (0.82, 0.95) Pantothenic acid 63.7 95.1 0.91 (0.87, 0.96) 69.0 92.0 0.88 (0.81, 0.95) Vitamin C 52.9 82.4 0.82 (0.73, 0.90) 63.7 97.3 0.91 (0.87, 0.95)

Abbreviation: CI, confidence interval. aSubjects were classified into the same categories between FDB-raw and FDB-cook. bSubjects were classified into the same or adjacent categories between FDB-raw and FDB-cook.

Table 5 Serum carotenoid (mg/ml) and plasma vitamin C (mg/dl), vitamin B6 (nmol/l), vitamin B12 (pmol/l) and folate (nmol/l) concentration and correlation with corresponding vitamin intake using food composition databases composed of raw food only or raw and cooked food

n Mean±(s.d.) Database with Database including raw fooda cooked foodb r (95% CI) r (95% CI)

Men Vitamin C 90 1.17±0.27 À0.05 (À0.27, 0.18) À0.06 (À0.27, 0.14) a-carotene 89 0.07±0.04 0.10 (À0.10, 0.31) 0.22 (0.02, 0.43) b-carotene 89 0.32±0.24 0.24 (0.04, 0.44) 0.22 (0.02, 0.41) Lycopene 89 0.11±0.05 0.30 (0.09, 0.50) 0.37 (0.18, 0.57) Vitamin B6 90 204.7±192.8 0.10 (À0.12, 0.31) 0.18 (À0.03, 0.39) Vitamin B12 90 479.3±234.1 0.02 (À0.19, 0.24) À0.03 (À0.24, 0.18) Folate 90 77.9±41.4 0.04 (À0.17, 0.26) 0.04 (À0.17, 0.25)

Women Vitamin C 101 1.31±0.32 À0.07 (À0.28, 0.14) À0.02 (À0.23, 0.19) a-carotene 101 0.10±0.05 À0.01 (À0.21, 0.20) 0.19 (À0.01, 0.39) b-carotene 101 0.57±0.37 0.21 (0.01, 0.41) 0.17 (À0.03, 0.37) Lycopene 101 0.13±0.06 0.35 (0.15, 0.56) 0.37 (0.16, 0.57)

Abbreviation: CI, confidence interval. aNutrient values calculated using a food database composed of raw food only. bNutrient values calculated using a database composed of raw and cooked foods.

classification by quintiles, no appreciable changes in subject preparation (Agte et al., 2002; Gupta and Bains, 2006). In the ranking were observed. present study, because the carotene composition data for As carotene is fat-soluble vitamin, its loss to cooking might vegetables with high-carotene content were not substantially be lower when cooked with hydrogenated fat. However, decreased by cooking, carotene intake did not decrease when losses of b-carotene are reportedly high during cooking calculated using FDB_cook. On the other hand, a loss of

European Journal of Clinical Nutrition Effect of cooking loss on vitamin intake M Kobayashi et al 551 tocopherols of 70% was observed when meat was roasted high. Moreover, consideration of cooking losses was difficult and pears were steamed (Steinhart and Rathjen, 2003). It was when cooking procedures were difficult to distinguish, such also reported that vitamin E itself or its complex form is as for eggs or wheat flour used for cooking a croquette or quite unstable during pressure cooking. Further, Fillion and tempura. Finally, we were unable to consider losses when Henry, 1998 noted the importance of accounting for similar foods were not found in the existing FCT or United tocopherols present in frying oil, given that some fat is States Department of Agriculture composition table, such as inevitably taken up by the food during frying, contributing processed fish. to an increased energy density. In the present study, no As harvest season differs among foods, nutrient intake in decrease in vitamin E intake was seen when calculated using Japan is subject to seasonal variation. It is also possible FDB_cook. When the subjects were classified into the same that cooking methods for individual foods differ between and adjacent categories by joint classification by quintiles, seasons: vegetables such as radish, carrot and spinach, for moreover, no appreciable change in subject ranking was seen example, are more frequently boiled as a hotpot in winter. for fat-soluble vitamins. The effect of cooking loss on daily nutrient intake therefore In the present study, we saw no improvement in the likely differs between seasons. Data in the present study were relation between plasma vitamin C, B6,B12 and folate collected in summer, and the effect of cooking loss on concentration and the corresponding vitamin intake with nutrient intake might accordingly be different from those in the use of a food composition database of cooked food. other seasons. As subjects were mostly married couples, who Given observations of a low correlation coefficient for often consume their meals together at home, we made plasma vitamin C and B12 and their respective intakes when no attempt to observe whether the effect of cooking loss on FDB_raw was used, we expected an improvement in correla- nutrient intake differed by sex. tions when intakes were calculated using FDB_cook. How- In conclusion, we observed that vitamin intakes estimated ever, our result indicates that the low correlations were not from DR were influenced by the food composition database because of cooking loss. The validity of dietary measures for used to calculate them. However, when subjects were vitamin C and B12 have not been consistent among studies. classified into the same and adjacent categories by joint Although cooking loss was not associated with the correla- classification by quintiles, no appreciable changes in the tion between plasma vitamin C, B12 concentration and ranking of subjects were observed. Further, the relationship corresponding vitamin intake in this study, it is assumed between vitamin intake and biomarkers of intake did not that the intake of these vitamins may vary in response to the improve when intake was calculated using a food database, availability of seasonal foods, and that plasma vitamin C and which included cooked food. These findings suggest that, in

B12 levels do not reflect habitual intake because of their short the estimation of vitamin intake in epidemiological studies, half-lives (Kallner et al., 1979). the effect of cooking losses may not significantly impact the With regard to the relation between serum a-carotene ranking of subjects by intake. concentration and corresponding vitamin intake, this was improved with the use of the cooked food database. This might have been because of the inadequate portion sizes of Conflict of interest carrots and tomatoes, which together contribute more than 90% of the cumulative intake of a-carotene intake in these The authors declare no conflict of interest. subjects (Kobayashi et al., 2003). On this basis, the effect of cooking is likely clearly reflected in a-carotene intake. Even among identical food materials, the degree of Acknowledgements vitamin loss can vary widely under the influence of a range of factors, including the type of food, manner of cutting, The authors wish to express their appreciation to the local preparation, and the duration and temperature of cooking. staff, especially to the local dietitians for their efforts in Our composition database had no information on these conducting the dietary survey. This work was supported factors, however, and thus we did not take them into in part by a Grant-in-Aid for Scientific Research from consideration when calculating vitamin intake. Estimated the Ministry of Education, Culture, Sports, Science and vitamin intake might therefore have been attenuated Technology, Japan and the Japanese Society of Nutrition and by these factors. Food Science. There were some foods whose substitution using the existing FCT5 was difficult. Calculation of cooking loss was difficult when two phases of cooking were used, such as References ‘frying and boiling’ or ‘steaming and boiling’. When cooking involved two or more phases, we calculated cooking loss Agte V, Tarwadi K, Mengale S, Hinge A, Chiplonkar S (2002). Vitamin profile of cooked foods: how healthy is the practice of ready-to-eat from the last cooking operation only. As we did not consider foods? Int J Food Sci Nutr 53, 197–208. nutrient change in phases preceding the last phase, we Fillion L, Henry CJ (1998). Nutrient losses and gains during frying: assumed that the cooking loss of nutrients may be markedly a review. Int J Food Sci Nutr 49, 157–168.

European Journal of Clinical Nutrition Effect of cooking loss on vitamin intake M Kobayashi et al 552 Gupta S, Bains K (2006). Traditional cooked vegetable dishes as Steinhart H, Rathjen T (2003). Dependence of tocopherol stability important sources of ascorbic acid and beta-carotene in the diets on different cooking procedures of food. Int J Vitam Nutr Res 73, of Indian urban and rural families. Food Nutr Bull 27, 306–310. 144–151. Han YH, Yon M, Hyun TH (2005). Folate intake estimated with an The Council of Science and Technology Agency Ministry of updated database and its association to blood folate and homo- Education, Calture, Sports, Science and Technology (2005). cysteine in Korean college students. Eur J Clin Nutr 59, 246–254. Standard Tables of Food Composition in Japan. The fifth revised and Iso H, Moriyama Y, Yoshino K, Sasaki S, Ishihara J, Tsugane S (2003). enlarged edition. Printing Bureau, Ministry of Finance: Tokyo. Validity of the self-administered food frequency questionnaire Tsubono Y, Takamori S, Kobayashi M, Takahashi T, Iwase Y, Iitoi Y used in the 5-year follow-up survey for the JPHC Study to assess et al. (1996). A data-based approach for designing a semiquanti- folate, vitamin B6 and B12 intake: comparison with dietary tative food frequency questionnaire for a population-based records and blood level. J Epidemiol 13, S98–101. prospective study in Japan. J Epidemiol 6, 45–53. Kallner A, Hartmann D, Hornig D (1979). Steady-state turnover Tsugane S, Sasaki S, Kobayashi M, Tsubono Y, Akabane M (2003). and body pool of ascorbic acid in man. Am J Clin Nutr 32, Validity and reproducibility of the self-administered food 530–539. frequency questionnaire in the JPHC Study Cohort I: study design, Kimura M, Itokawa Y, Fujiwara M (1990). Cooking losses of thiamin conduct and participant profiles. J Epidemiol 13, S2–12. in food and its nutritional significance. J Nutr Sci Vitaminol (Tokyo) Tsugane S, Sasaki S, Kobayashi M, Tsubono Y, Sobue T (2001). Dietary 36 (Suppl 1), S17–S24. habits among the JPHC study participants at baseline survey. Kobayashi M, Sasaki S, Tsugane S (2003). Validity of a self- Japan Public Health Center-based Prospective Study on cancer and administered food frequency questionnaire used in the 5-year cardiovascular diseases. J Epidemiol 11, S30–S43. follow-up survey of the JPHC Study Cohort I to assess carotenoids Wakiyama K, Kobayashi M, Horiguchi M (2009). Evaluation of and vitamin C intake: comparison with dietary records and blood vitamin and mineral loss due to cooking. Res Bull Otsuma Womens level. J Epidemiol 13, S82–S91. Univ Home Econ 46, 55–70. Mareschi JP, Belliot JP, Fourlon C, Gey KF (1983). Changes in Watanabe S, Tsugane S, Sobue T, Konishi M, Baba S (2001). Study vitamin C content of Bintje potatoes during storage and usual design and organization of the JPHC study. Japan Public Health culinary preparations. Int J Vitam Nutr Res 53, 402–411. Center-based Prospective Study on cancer and cardiovascular Rumm-Kreuter D, Demmel I (1990). Comparison of vitamin losses in diseases. J Epidemiol 11, S3–S7. vegetables due to various cooking methods. J Nutr Sci Vitaminol Willett W, Stampfer MJ (1986). Total energy intake: implications (Tokyo) 36 (Suppl 1), S7–14. for epidemiologic analyses. Am J Epidemiol 124, 17–27.

Appendix The investigators and their affiliations in the validation study Division, National Cancer Center Research Institute, Tokyo; of the self-administered food frequency questionnaire in the Akabane M, Iitoi Y, Iwase Y and Takahashi T, Tokyo University JPHC Study (the JPHC FFQ Validation Study Group) at the of Agriculture, Tokyo; Hasegawa K and Kawabata T, Kagawa time of the study were: Tsugane S, Sasaki S and Kobayashi M, Nutrition University, Sakado; Tsubono Y, Tohoku University, Epidemiology and Biostatistics Division, National Cancer Sendai; Iso H, Tsukuba University, Tsukuba; Karita S, Teikyo Center Research Institute East, Kashiwa; Sobue T, Yamamoto University, Tokyo; the late Yamaguchi M and Matsumura Y, S and Ishihara J, Cancer Information and Epidemiology National Institute of Health and Nutrition, Tokyo.

European Journal of Clinical Nutrition