Nutrient Intakes in Relation to Style of Breakfast and Taste Preferences

Style of breakfast (westernor Japanese-style) and taste preferences were associated with various diseases in some epidemiological studies in Japan. To evaluate what are measured by asking these dietary behaviors, we administered semiquantitative food frequency questionnaire (SFFQ) and asked style of breakfast and taste preferences to a subsample of residents in Takayama City, Gifu, Japan. Style of breakfast and total diet were studied in a random sample of 346 residents. There were no statistically significant differences in the nutrientintakes estimated from SFFQ between those with westernand Japanese-style of breakfast except for crude fiber. Taste preferences were asked to 555 men and 1,130 women who attended a health check-up program in the community. Salt intakes were slightly higher (<3%) in those who had a favor for salty food than the others in both sexes. Fat or carbohydrate intake was similar between those stratified by preference for greasy or sweet foods. The data suggest that western-style of breakfast is not associated with western diet, in general, i.e., high-calorie and high-fat diet. Intakes of salt, fat, or carbohydrate appear to be unrelated to preferences for salty, greasy or sweet foods, respectively. J Epidemiol, 1999 ; 9 : 91-98

tus, and ischemic heart diseases.A numerous studies have devoted the methods to measure individuals' usual dietary intakes quantitatively1-4).The food frequency questionnaire (FFQ) is the most appropriate method for dietary assessment in epidemiological studies5).
On the other hand, there have been behavioral approaches to dietary assessment 6,7).Several epidemiological studies in Japan 8-15) have utilized the questions asking style of breakfast (western or Japanese-style) and preferences of taste.Westernstyle of breakfast was observed to be associated with increased risks of colon cancer8,12) and decreased risks of gastric 9,11) thy-roid14), and oral cancers 15) in these studies.Preference for salty foods was reported to be associated with increased risk of pancreatic cancer 10) and decreased risk of oral cancer15).As for the association between style of breakfast and cancer risk, some authors speculated that style of breakfast may reflect to some extent of westernization in diet among Japanese 9,12).However, there has been no report evaluating what is measured by these questions.It is also questionable that salt taste preference reflects actual salt consumption.Therefore, we administered semiquantitative FFQ (SFFQ) and asked style of breakfast and taste preferences to a sample of residents in a community and compared the results from the two methods.

SUBJECTS AND METHODS
Subjects for this study were originally a sample from a cohort of the Takayama Study 16), which is a prospective cohort study investigating the role of diet and lifestyle in the subsequent development of cancer.In 1992, about 92% of all residents aged 35 years and over in Takayama City , Gifu, Japan (14,427 men and 17,125 women) responded to a questionnaire including dietary habits.Age distribution of the participants in the Takayama Study was as follows: 27 , 26, 25, and 22 % in age groups of 35-44, 45-54, 55-64, and 65+ in men, respectively, and 25, 24, 23, and 26 % in the corresponding groups in women, respectively.Major occupational categories were professional or technical workers (18 %) and factory workers or machine operators (13 %) and laborer or farm workers (13 %) in men and 42 % of women were housewives.Diet was assessed by SFFQ asking the frequency of intake of 169 food items during the past year and usual serving size of each food item.Detailed information about the Takayama Study and the results of validation tests of the dietary questionnaire were described elsewhere 16,17) For the study on style of breakfast and total diet, we randomly selected 200 men and 200 women aged between 35-59 years from the participants in the Takayama Study.We stratified the paticipants in the Takayama Study according to age and sex and applied randamization using computer-generated random digit numbers.We sent them a mail questionnaire asking style of breakfast when they were enrolled in the Takayama Study (about 4 years prior to the present study).Three response categories were provided for the question; Japanese-style (mainly rice is eaten at breakfast), western-style (mainly bread is eaten at breakfast) and no breakfast.A total of 346 mails were returned (response rate was 86.5%).Those who reported to usually have no breakfast (n=23) and those who did not report a style of breakfast (n=3) were excluded from the following analyses.
Study on taste preferences and total diet included 1,249 men and 3,606 women who were participants in the Takayama Study and attended the annual health check-up program provided by Takayama Municipality between April and October in 1992.Health check-up program included an interview asking whether they preferred to salty, greasy, and sweet foods or not.Those who had chronic diseases (hypertension, diabetes mellitus, cardiovascular diseases, asthma, gastric ulcer, and colon polyps) and cancer (694 men and 2,476 women) were excluded from the analyses.
A multiple linear regression technique was used to compare nutrient intakes estimated from SFFQ between the two groups according to style of breakfast or preferred tastes.Age was included in the models as a covariate.The nutrient intakes were log-transformed and adjusted for total energy using the method proposed by Willett 16).We present the results on style of breakfast and total diet after controlling for sex besides age, since similar findings were noted in men and women.
The questions concerning style of breakfast and taste preference were also administered to 37 subjects (17 men and 20 women) who had volunteered for the validation study of our SFFQ three years prior to the present study.They completed 12 daily diet records at about 1 month intervals over 1 year.We used these data for suppletory analyses.

RESULTS
Out of 320 respondents, 71 (33 men and 38 women) reported to have a western-style of breakfast and 249 (120 men and 129 women) reported to have a Japanese-style of breakfast.Table I presents age and sex-adjusted means of individual nutrient intakes estimated from SFFQ according to reported style of breakfast.Crude fiber was significantly higher in those who had Japanese-style of breakfast than those who had western-style of breakfast (p = 0.02).There were no statistically significant differences in the other nutrient intakes studied between the two groups.
There were significant differences in intakes of alcohol, total and vegetable proteins, total, animal, and vegetable fats, cholesterol, crude fiber, calcium, vitamins A, B1, B2, C and E between the two groups according to favor for salty foods in men, but the differences were not great (less than 10%) except for carotene and vitamin C (Table 2).Salt intake was slightly higher (2%) in those who preferred to salty food but the difference was not statistically significant (p = 0.31).All the nutrient intakes except for vegetable protein were similar between those who reported a favor for greasy foods and those who did not.Vegetable protein intake was significantly lower in those who had a favor for greasy foods but the difference was only 3%.Men who had a favor for sweet foods had significantly higher intakes of total protein and total fat than the others.However, the differences in these intakes were not great (less than 10%).
In women, significantly lower intake were observed for most of the nutrients in those who had a favor for salty foods, but the differences were less than 10% (Table 3) Salt intake was only 2% higher in those who had a favor for salty foods than the others.Those who had a favor for greasy foods had significantly higher total energy and significantly lower intakes of total and vegetable protein, crude fiber, calcium, and vitamins B1, B2, and E. The differences in these nutrient intakes between the two groups were less than 10%.Those who had a favor for sweet food had significantly higher intake (2%) of carbohydrate.
Although alcohol intake was significantly related to preferences for salty and sweet foods, respectively, in both men and women, adjustment for alcohol intake did not alter the results substantially.
The relationships of nutrient intakes to style of breakfast and taste preference were also examined in 37 subjects who completed the 12 daily diet records.The mean intakes of individual nutrients were similar in the groups with western-and Japanese-style of breakfast except intakes of vegetable fat and vitamin C (Table 4).The ratios of the mean nutrient intakes for the two groups according to style of breakfast were similar to those derived from the SFFQ except for vitamins C and D. The results concerning nutrient intakes and taste preference in this sample were similar to those presented in Tables 2 and 3.The ratios of the mean salt intakes in those who had a favor for salty foods to those in those who did not were 0.89 and 1.04 in men and women, respectively.The ratios for the mean fat intakes in relation to preference for greasy foods were 1.01 and 1.06 in men and women, respectively.The ratios for the mean carbohydrate intakes in relation to preference for sweet foods were 0.98 and 1.07 in men and women, respectively.

DISCUSSION
Previous studies lead us to expect that western style of breakfast is related to high energy and high fat diet, but there were no apparent differences in the nutrient intakes except for crude fiber between those who had western and Japanese-style of breakfast.The results suggest that the information on style of breakfast is not so useful to identify a particular type of nutrient intakes, i.e., western-style of breakfast is not associated with western diet such as high energy and high fat.
There might be misclassification for style of breakfast in this study because we asked about the style of breakfast about four years prior to the present.If style of breakfast varied day to day, or could not be simply categorized into western and Japanese styles, misclassification should arise even when we ask the present style of breakfast.Using data from 37 subjects who completed 12 daily diet records over 1 year, we compared their reported style of breakfast with breakfast menus from diet records.Among the 37 respondents (response rate was 100%), 12 and 24 reported western-and Japanese-style of breakfast, respectively.Remaining one person reported to have had no breakfast.According to their diet records, 12 (32%) had the same style of breakfast (5 for western-and 7 for Japanesestyle) throughout the 12 days for the diet records.Six (16%) had both western-and Japanese-style of breakfasts frequently (the difference in the frequencies between western-and Japanese-style was less than 3 during the 12 days).Among all the menus for breakfast (37 X 12), 22 (5%) were unable to be categorized into western-and Japanese style, because foods other than rice or bread were taken or both rice and bread were taken for breakfast.When we regarded the style of breakfast appeared most frequently in the individual diet records as his/her style of breakfast, 6 (16%) reported a type of breakfast disagreed with it.It is unlikely that this discordance was only caused by recalling the three year past diet style.Misclassification due to recall of style of breakfast in four year past should not be great.
There is a concern that lack of association between type of breakfast and the nutrient intakes may be due to measurement error by use of SFFQ.Therefore, we additionally analyzed the association using data from 12 daily diet records.The comparison of style of breakfast with SFFQ or 12 daily diet records did not yield a plausible interpretation for associations between style of breakfast and diseases found in some previous studies.Although there was a speculation that western-style of breakfast may reflect westernization in diet to some extent9,11), our data showed no association between western-style of breakfast and western diet, i.e., high-calorie and high-fat diet.It is possible that the association between style of breakfast and diseases may reflect the effects of crude fiber and/or some vitamins on the diseases, but the evidence here was not strong.
Since we are interested in what are measured by the questions about style of breakfast which was found to be significantly associated with some diseases, we referred to the definition of style of breakfast in the previous studies reporting such associations.In the early study reported by Kato et al 9)., style of breakfast was defined as western-style when subject ate bread rather than rice at breakfast.Other studies did not provide the details on question about style of breakfast, but they included two categories for style of breakfast (i.e., westernand Japanese-style) in the analyses 8,11,12,15).Therefore, we provided two response categories for style of breakfast in the present study.However, recent study reported by Takezaki et al 14).included three categories for style of breakfast (i.e., consuming rice, rice or bread (mixed), and bread for breakfast).Risk magnitudes for thyroid cancer were in this order.Therefore, it is likely that the difference in nutrient intakes between the groups with western-and Japanese-style of breakfast may be underestimated in our analysis including those who consumed rice or bread into either of the two categories.However, animal fat, animal protein, and cholesterol intakes estimated from SFFQ were slightly greater in those with Japanese-style of breakfast, but those estimated from 12 daily diet record were slightly lower in those with Japanese-style of breakfast.Therefore, even considering the effect of misclassification bias, the differences in intakes of these nutrients between those with western-and Japanese-style of breakfast would not be substantial in either direction.When we reanalyzed data from 12 daily diet records excluding 6 subjects who recorded both western-and Japanese-style of breakfast frequently during the 12 days for the diet records, the results were not substantially changed (the ratios ranged from 0.75 for vitamin C to 1.13 for vegetable fat).
We considered potential confounding effects from non-nutritional factors.Additional adjustments for number of children, years of education, and body size did not alter the results.However, we cannot also deny a possibility that unmeasured non-nutritional factors may be related to style of breakfast and risks of some diseases.Salt taste preference appeared to be unrelated to salt intake.Similar results were reported by Shepherd et al. 19) and Drewnowski et al. 20) Preference for greasy foods was also unrelated to fat or cholesterol intake.The relationship between preference for sweet foods and carbohydrate intake was weak.It is possible that concern about nutrition and health rather than taste preference should affect food consumption.
We may not be able to detect the difference in diet according to taste preferences because the data were self-reported and we asked preferences dichotomously (yes or no).However, our questions about taste preferences appeared to be almost identical with those used in the previous Japanese studies which reported significant associations between taste preferences and diseases 10,15) Since subjects for study on taste preferences were participants in the health check-up program, we may have failed to find the association of taste preferences and nutrient intakes.Participants in health check-up programs are generally health conscious.Health check-up programs are actually often utilized to provide nutritional advice to community people.Advice based on the postulation that preference to particular food group reflect its consumption should not be appropriate.

Table 1 .
Comparison of means (95% CI) of nutrient intakesa estimated from semiquantitative food frequency questionnaire between the two groups stratified by style of breakfast.

Table 2 .
Comparison of means (95% confidence intervals) of nutrient intakesa estimated from semiquantitative food frequencyquestionnaire between the two groups stratified by taste preferences in men.
a Adjusted for age and total energy

Table 3 .
Comparison of means (95% confidence intervals) of nutrient intakesa estimated from semiquantitative food frequency questionnaire between the two groups stratified by taste preferences in women.
a Adjusted for age and total energy

Table 4 .
Comparison of means of nutrient intakes3 estimated from 12 daily diet records over 1 year between the two groups stratified by style of breakfast.
a Adjusted for age, sex and total energy.b Ratios of the mean numtrient intakes in the group with western-style of breakfast to those in the group with Japanese-style of breakfast.*p<0 .05