Dietary habits among the JPHC study participants at baseline survey. Japan Public Health Center-based Prospective Study on Cancer and Cardiovascular Diseases.

Dietary habit is closely associated with development of cancer and cardiovascular diseases, however little prospective evidence has been published for Japanese, whose dietary habit is substantially different from Western countries. Therefore, frequencies of food consumption, food preference, cooking method and acceptance of dietary advice were investigated at the baseline by two kinds of self-administered food frequency questionnaires. Dietary habits between urban and rural (Tokyo and Osaka vs. others), or between Okinawa and non-Okinawa revealed recognizable differences. The so-called westernized foods such as bread, beef and coffee were more consumed in the urban areas such as Tokyo and Osaka and also in Okinawa. The frequencies of salted food intake such as pickled vegetables and salted seafoods were remarkably low in Okinawa. Cooking methods for meats, seafoods and vegetables were also unique in Okinawa. No distinct geographical difference was shown in food preference and modification of dietary habit by dietary advice.


INTRODUCTION
Animal experiments and observational evidences in human have provided many suggestions that dietary factors are of major importance in determining the risk of cancer. Doll and Peto u estimated the proportion of cancer deaths attributed to diet to be 35 % (range of acceptable estimates: 10 to 70) in 1981. Willett 2) also estimated recently that the proportion would be 32 % (20 to 42). The proportion varied by cancer site from 20 % in lung to 75 % in prostate. Such estimation was possible in the United States and other Western countries, because reasonable amount of epidemiological evidences in "diet and cancer" have been published in good qu ality. In particular, several prospective studies with large number of subjects have been conducted using validated dietary assessment methods 3).
On the contrary, little evidence has been available from prospective studies for Japanese whose dietary habits should be substantially different from Western countries. A Large-Scale Census-Based Cohort Study in Japan4) , commonly called as "Hirayama's Cohort or Six Prefecture Cohort" which started in 1965, revealed one of initial evidences in the association between green-yellow vegetable intake and male lung cancer. Dietary habits, however, were assessed from only 8 items (rice, meat, fish and shellfish, milk and goat milk, green-yellow vegetables, pickles, soybean paste soup and green tea) with simple categories (Daily , Occasionally, Rarely, None, Uncertain).
In the Japan Public Health Center-based prospective Study on cancer and cardiovascular diseases (JPHC Study) , we assessed the dietary habits of the subjects at the baseline survey using food frequency questionnaires . In this paper, we showed the method of dietary assessment and the results of frequencies of food consumption , food preference, cooking methods and acceptance of dietary advice by each study center with special reference to the geographical difference .

METHODS
Dietary patterns were investigated by two kinds of selfadministered baseline questionnaires in which frequency of food consumption, food preference, cooking methods and acceptance of dietary advice were included. The questionnaires used in JPHC Study Cohort I and II were slightly different with respect to food items, method of expression and frequency categories (Table 1).

Cohort I
The questionnaire used in Cohort I was a revised version of the questionnaire used in the cross-sectional epidemiological study for assessing cancer risks at the population level (ECO-CANCER study), in which randomly selected men, aged 40s', and their wives in the same five health center districts as Cohort I had been interviewed to know their lifestyles including dietary habits. The study design and the results of the food frequency questionnaire have been described elsewhere 5.6,7,8).
A total of 26 out of 38 food items in ECOCANCER study were consistent with JPHC Study Cohort I questionnaire and the other 12 food items, mostly sea foods and fruits, were asked in a different way in the Cohort I questionnaire.
For rice intake, the number of bowls (regular size) per day was asked for daily consumer. The intake frequency per week was asked using four categories: rarely (<I day/week), 1-2 days/week, 3-4 days/week, almost everyday (5 days or more/week) for the following food items: miso (soybean paste) soup, noodles, precooked noodles, bread, butter and margarine, fruits, green vegetables, yellow vegetables, other vegetables, dressing and mayonnaise, pickled vegetables, mushrooms, potatoes, seaweed, soy beans and its products, other beans, egg, milk and its products, cheese, beef, pork, chicken, bacon, ham and sausages, liver, fresh fish, dried and salted fish, salted roe, salted preserves and gonad roe. Dietary habits, such as breakfast, an evening meal at home, light meals and fried and deep fried foods were also asked using the above 4 categories. For miso soup, typical quantity per day was further asked for those who responded to a category of almost everyday. The intake frequency of beverages was asked using six categories: rarely (less than 1 day/week), 1-2 days/week, 3-4 days/week, 1-2 cups/day, 3-4 cups/day, 5 cups or more/day, for the following food items: Japanese tea (green tea), Chinese tea (Oolong tea), black tea, other kind of tea, coffee, milk, carbonated beverages, pure fruit juice and vegetable juice.
The monthly frequency for each food item was calculated as four times the scores assigned to the four weekly frequency categories: rarely=0, 1-2 days/week=l.5, 3-4 days/week=3.5, almost everyday=6. For rice, miso soup and beverages, consumption was expressed in terms of daily frequency. When daily intake of these items was computed the number of bowls per day was used for rice and miso soup and the scores assigned to three daily frequency categories (1-2 cups/day= 1.5, 3-4 cups/day=3.5, 5 cups or more/day=6) were used for subjects consuming beverages almost everyday. The scores of 0.5, 0.21, and 0 were assigned to subjects consuming beverages 3-4 days/week, 1-2 days/week, and rarely, respectively. The frequency of food consumption (percent) was also expressed in the indicated categories.
Food preferences were asked according to one of three categories: "like", "intermediate" and "dislike" for the following preferences: deepfried/oily ("Kotteri"), hot (spicy) ("Karai"), salty ("Shio-kagen-no-koi"), sour ("Suppai"), sweet ("Amai") and hot (high in temperature) ("Atsui"). The percent of "like" category responses was then calculated. Most commonly used cooking methods for meats, seafood and vegetables were asked according to one of six categories: raw, boiled, grilled, deep fried, stir fried and others. The percent of responded subjects to each category was calculated.
Acceptance of dietary advice was also asked according to "no" or "yes" categories and the percent of "yes" and "no" were calculated.
Number of subjects used in this report is as following: 9,101 (4,229 men and 4,872 women) in Ninohe, 11,754 (5,471 men and 6,283 women) in Yokote, 10,887 (5,410 men and 5,477 women) in Saku, 10,453 (5,156 men and 5,297 women) in Ishikawa, and 7,004 (2,888 men and 4,116 women) in Katsushika. Number of missing values varied to each item of the questionnaire.

Cohort II
The questionnaire used in JPHC Study Cohort II was a revised version of the questionnaire used in the JPHC Study Cohort I. In this questionnaire, specific food items were included with portion size such as apple (instead of fruits as a whole) and carrot (instead of yellow vegetables). Three choices for each portion size were given: half, same or 1.5 times more for a given portion size.
For rice, the number of bowls (regular size) per day was asked for daily consumer as same as in Cohort I. The intake frequency per week was asked according to one of four categories: rarely (<I day/week), 1-2 days/week, 3-4 days/week, almost everyday (5 days or more/week) for the following food items: miso (soybean paste) soup, mayonnaise, dressing, ketchup and fried and deep fried dishes. For miso soup, typical quantity per day was further asked for those who responded to a category of almost everyday. The intake frequency per week was asked according to one of five categories for most food items: never, occasionally, 1-2 days/week, 3-4 days/week, almost everyday (5 days or more/week) for the following food items: noodles, precooked noodles, bread, butter, margarine, apple, citrus fruit, green vegetables, carrot, tomato, nozawanazuke and takuan-zuke (pickled green-yellow vegetables), other pickled vegetables, potatoes, seaweed, tofu, fermented soybean, egg, milk, yogurt, cheese, beef, pork, chicken, ham, sausages and bacon, liver, fresh fish, dried and salted fish, fish paste products, canned fish, small fish, salted roe and salted preserves, foreign-style confectioneries and Japanese-style confectioneries. Questions about the intake frequency of beverages were the same as in Cohort I, except for deletion of "other kind of tea" and "milk".
The monthly frequency for each food item was calculated as four times the scores assigned to the four or five weekly frequency categories: never, occasionally or rarely=0, 1-2 days/week=1.5, 3-4 days/week=3.5, almost everyday=6. For rice, miso soup and beverages, consumption was expressed in terms of daily frequency. Further calculations were the same as in Cohort I.
Food preferences were asked according to one of three categories: "like", "general" and "dislike" for the following preferences: deepfried/oily ("Abura-koi"), salty ("Shio-kagen-nokoi"), sour ("Suppai"), sweet ("Amami-no-tsuyoi") and hot (high in temperature) ("Atsui"). The percent of "like" category responses was then calculated. Most commonly used cooking methods was asked according to one of five categories: raw, boiled, grilled, deep fried and stir fried for the following food groups: meats, seafood, green leafy vegetables and carrots. The percent of responses to each category was calculated.
Acceptance of dietary advice was also asked according to "no" or "yes" categories and the percent of accepted subjects was calculated. However, the question regarding the consumption of burnt part of fish or meat was slightly different, asking "Do you eat the burnt part of fish or meat?" , and "no" answer was used as an acceptance of dietary advice.
Number of subjects used in this report is as follows: 19,012 (9,453 men and 9,559 women) in Kasama, 3,165 (1,520 men and 1,645 women) in Kashiwazaki, 7,557 (3,561 men and 3,996 women) in Tosayamada, 11,199 (4,983 men and 6,216 women) in Arikawa, 10,048 (4,779 men and 5,269 women) in Miyako, 6,489 (2,907 men and 3,582 women) in Suital, and 4,401 (2,055 men and 2,346 women) in Suita2. Number of missing values varied to each item of questionnaire.

Frequency of food consumption (Tables 2a, b, c, d)
Due to the different structure of the two questionnaires used in the Cohort I and II, geographical comparisons were separately done in each cohort, and their relative order was described. In general, the observed distribution was shifted toward the lower frequency categories when the category of "never" was added to four frequency categories as in Cohort II ("C" and "E" categories in Table 1). Consequently, the absolute values of mean frequency were systematically shifted to low in Cohort H for various food items.
The mean frequency of rice consumption was relatively lower in Tokyo, two areas in Osaka, and Okinawa (Ishikawa) than in the other areas. Consumption of bread showed con-verse geographical difference. Miso soup showed the same regional difference as rice. Butter and margarine showed , on the other hand, the same pattern as bread. The mean frequency for fruit intake was relatively low in two areas in Okinawa , and slightly lower in Tokyo and Osaka than in the other areas . The mean frequency for green and yellow vegetables was low in Tokyo, Osaka, and Nagasaki. The pickled vegetables were frequently taken in the areas of northern Japan such as Iwate, Akita, Nagano, and Niigata. It was considerably low in Okinawa. For soybean and its products, the result was different between Cohort I and II. Two areas in Tohoku such as Iwate and Akita in Cohort I showed a higher mean frequency than the other areas. In Cohort H in which "tofu" was categorized as an independent food, Kochi and Okinawa, which are locate in the southern part of Japan, showed a higher mean frequency than the other areas. The mean for fermented soybeans, included only in Cohort 11 as an independent food, was considerably high in Ibaragi. Egg intake did not show any noticeable regional difference. The mean of cheese intake was high in Tokyo and Okinawa (Ishikawa). Among meats, only beef consumption showed a relative regional difference. It was high in Tokyo, Osaka, and Okinawa (Ishikawa). As for the sea foods, mean intake frequency of dried and salted fish was low in Osaka and two areas in Okinawa, but those of fresh fish was high in Iwate, Akita, Kochi, and Nagasaki. In Okinawa, mean frequency of other salted seafoods, such as salted roes, salted preserves, and salted gonad paste was also lower than the other areas. The mean intake frequency of canned fish, which was only asked in the Cohort II questionnaire, was high in Miyako, Okinawa. Among beverages, the mean frequency of Japanese tea was relatively high in Nagano and Ibaragi. On the other hand, the mean frequency of coffee was high in Tokyo, Osaka, and Okinawa (Ishikawa). Milk did not show any regional difference.
No consistent difference in mean consumption patterns was observed between men and women. For certain foods, such as mean frequency of fruit and green yellow vegetables, consumption was generally higher in women than in men.
The percent frequency of food consumption showed similar regional patterns as the mean frequencies of monthly food consumption.
Food preference (Tables 3a, b, c, d) "Deep fried/Oily" or "Oily" was preferred in Okinawa compared to the other areas. No remarkable regional difference was observed for the other food preference. Sex difference was not observed.
Cooking methods (Tables 4a, b, c, d) For meats, "grilled" and then "stir fried" were two common cooking methods in most areas except Okinawa. When the preferences of two methods were compared between men and women, "grilled" were more preferred by men and "stir fried"   by women. In Okinawa, "boiled" was the most commonly used cooking method. For sea foods, "grilled" was the most commonly used cooking method in most areas, but it was "boiled" and "raw" in Nagasaki , and "stir fried" in Okinawa. "Raw" was more commonly preferred b y men than women. For vegetables, "raw", "boiled", and "stir fired" were three commonly used cooking methods in all areas except Okinawa .
No consistent order was observed for these cooking methods in Cohort I. In Cohort H, in which "green vegetables" and "carrots" were separately asked in contrast to Cohort I which asked cooking method for all "vegetable", "boiled" was the most commonly used cooking method in all areas except Miyako, Okinawa. "Raw" was more preferred by women than men. In Okinawa, "stir fried" was a far more commonly used cooking method than the other areas for vegetables.
Participation on Dietary advice (Tables 5a, b, c, d) No remarkable regional difference was observed in the per-centages of acceptance to dietary advices . For all questions, the percentage of followers was higher for women than men.

DISCUSSIONS
Geographic differences have been shown in several items in frequencies of food consumption and cooking method , while it was not remarkable in food preference and acceptance rate of dietary advice. The differences were commonly observed between urban and rural (Tokyo and Osaka vs . others) or Okinawa and non-Okinawa. The so-called westernized foods such as bread, beef and coffee were more consumed in urban areas such as Tokyo, Osaka and Okinawa . The frequencies of salted food intake such as pickled vegetables and salted seafoods were remarkably low in Okinawa . Cooking methods for meats, seafoods and vegetables were unique in Okinawa . These geographic differences were almost the same as those observed in the previous ecological study st , in which high Table 4d. Most commonly used cooking method (percent) for the following foods, cohort II, females. The percent contribution of inter-population variance to the total variation (inter-plus intra-population (within-and between-individual)) was relatively notable in fat (9%) and    In the baseline survey of JPHC Study, we only assessed food frequency using four or five category and portion size was only provided in certain number of foods in Cohort II. This dietary assessment method has several limitations not only with respect to nutrient level analysis but also well-balanced ranking for 100,000 cohort members. We therefore developed a semiquantitative food frequency questionnaire (SQFFQ) based on the dietary record data in Cohort I areas 11) and applied it at the 5 -and 10 -year follow-up survey in the entire JPHC Study area. The validity of this SQFFQ has been testing based upon the 7-day dietary records in four seasons from 30 couples each among participants in 11 JPHC study areas and furthermore these data are analyzed in conjunction with biomarkers in blood and collected urine from the same individuals.