Relationship of food intake and dietary patterns with blood pressure levels among middle-aged Japanese men.

To investigate the relationship of food intake habits and dietary patterns to blood pressure, a cross-sectional study was conducted for 473 middle-aged Japanese males. After adjustment for age, residence, occupation, body mass index and alcohol consumption, mean systolic and diastolic blood pressure (SBP and DBP) were inversely associated with each intake frequency of dairy products, coffee, fruits, egg, beef, pork and chicken. The adjusted mean SBP and DBP of the individuals with 'all (= three)' of the following three dietary habits; 'once and over/wk of dairy products', 'once and over/wk of fruits', and 'three times and over/wk of beef, pork or chicken', were 7.4 mmHg and 6.9 mmHg lower (p < 0.001, for each) than those of 'zero or one' group. According to the analyses on 3-days weighed food records of 157 volunteers, the adjusted mean daily intake of total protein, animal protein and potassium were markedly higher in the 'two' or the 'three' group than in the 'zero or one' group (p < 0.05, for each). These results suggest that habitual intake of dairy products, fruits, and meat or chicken may be associated with the reduction of blood pressure possibly through the intake of protein and potassium.

blood pressure. However, it is not necessarily feasible to incorporate these knowledge of individual nutrient in actual daily dietary life, since each food contains various kinds and amounts of nutrients beneficial or harmful to blood pressure in many cases. Therefore, as a practical approach for the prevention of hypertension, it is potentially important to recognize the effects of each food or dietary pattern on blood pressure properly. Although there have been many studies 1-9). [11][12][13][14] assessing the significance of individual nutrient or food as a determinant of blood pressure, few studies [15][16][17][18] reported the combined effects of various foods on blood pressure. Appel LJ, et al. 16) indicated from a clinical trial that a diet rich in fruits, vegeta-bles, and low-fat dairy products and with reduced saturated and total fat could reduce blood pressure markedly under the condition that sodium intake and body weight were maintained at constant levels. This finding may suggest the combined effects of potassium, magnesium, calcium, fiber and possibly protein that occur together in food. Sufficient blood-pressure-lowering effects may be attained when the combination of the foods with nutritional composition exclusively beneficial to blood pressure are habitually taken, even if the blood-pressure-lowering effect of individual nutrient or food is not large. From these standpoints, in this study, which foods and dietary pattern closely relate to reduced blood pressure among middle-aged Japanese men, was explored referring also to nutrient intake data on the basis of a cross-sectional study.

Subjects
As previously reported 19.20), we obtained various personal data from a survey on dietary habits with health examinations that had been conducted during the period of February, 1989 through March, 1991 for 635 male residents aged 40 to 49, mostly randomly selected from five areas in Japan. The individual records of blood pressure and dietary habits represented by the intake frequency of each food among these data were the major materials of the present study. The five areas corresponded to the administrative districts of the five local public health centers (H.C.) ; Ninohe, H.C. of Iwate prefecture, Yokote, H.C. of Akita prefecture, Saku, H.C. of Nagano prefecture, Katsushika-kita, H.C. of Tokyo prefecture and Ishikawa, H.C. of Okinawa prefecture. The number of participants in each of the five areas was as follows; Ninohe: 134, Yokote: 133, Saku: 120, Katsushika-kita: 119, Ishikawa: 129. The number of initially randomly selected subjects was 170 in Ninohe, Yokote, Saku and Ishikawa, and 200 in Katsushikakita. Thus, the response rates in the survey varied from 59.5% to 78.8% among the five areas. Among a total of 635 participants, the individuals that met the following exclusion criteria were excluded from the analyses; individuals under treatment of hypertension, diabetes mellitus, cerebrovascular diseases, hepatobiliary diseases, peptic ulcer, cancer or any other chronic diseases; individuals whom a specific diet was administered; individuals who had abstained from alcohol drinking because of particular health problems; individuals whose records of height, weight, blood pressure or frequencies of food intake were incomplete. As a result, the remaining 473 subjects were subject to actual analyses. The occupation of these subjects was classified into the 11 categories according to the classification of the census in 1985 21).

Blood Pressure Assessment
In the health examinations, blood pressure was measured three times consecutively for each participant using the right arm by a digital recorder (Takeda Medical UA-850). Trained technicians, all of whom were instructed by one physician, carried out the measurements. All the measurements of blood pressure were conducted under sedate conditions. Systolic and diastolic blood pressure (SBP, DBP) from the third measurement were analyzed in the study under the assumption that the subject was in a more relaxed state during the third measurement than the first and second measurements 20).

Assessment of dietary habits and nutrient intake
In the survey on dietary habits, a food frequency questionnaire was used. Trained dietitians or public health nurses interviewed each subject to determine the average number of days per week consumed during the past one month in each food item.
In Nutrients of which intake level was estimated according to this procedure, were as follows : total energy, protein, lipid, carbohydrate, calcium, phosphorus, iron, sodium, potassium, magnesium, and fiber. However, since the amount of magnesium and fiber per gram of food was listed only for a part of the foods in the Standard Tables of Food Composition in Japan, the intake levels of magnesium and fiber were estimated only from these foods with data of the amount of magnesium or fiber. Therefore, the individual intake levels of magnesium and fiber were possibly underestimated.
The aforementioned 473 subjects included 157 volunteers who had completed this nutrition survey.

Statistical Analyses
The SBP and DBP were compared between the two or three frequency categories of each food, and between the different categories of dietary pattern among the 473 subjects. In these comparisons, the SBP and DBP were adjusted for body mass index (BMI) which corresponds to weight (kg) divided by the square of height (m) and weekly alcohol consumption both known as important determinants of blood pressure25.26) in addition, age, the five categories of residence and the 11 categories of occupation were also included as variables adjusted for. These adjusted means, which correspond to the least square means, were calculated by a multidimensional analysis of covariance 27.28) using the LSMEANS option of the GLM procedure in the SAS statistical package 29). Similarly, the adjusted mean levels of daily nutrient intake were also compared between the categories of dietary pattern among the 157 subjects having the nutrition data. In this comparison, the mean levels of total energy intake were adjusted for age, occupation, residence, BMI, and weekly alcohol consumption. Other nutrients were further adjusted for total energy intake as a continuous variable. The adjustment for total energy intake was conducted by adding the total energy intake as a covariate in the multidimensional analysis of covariance using the GLM procedure, as shown in a previous study by Tsubono Y,et al 30).

RESULTS
The characteristics of 473 subjects are summarized in Table  1. The residences of the subjects were almost equally distributed into the five areas. Manual laborers (Craftsmen, production process workers and laborers) occupied about one-thirds of the subjects. As previously reported 25), there were some differences in blood pressure between the five areas. Moreover, the occupation was also associated with SBP and DBP. Among the 11 occupational categories, the mean SBP and DBP adjusted for age and residence were highest in `Workers in transport and communication', whereas `Service workers' showed the lowest adjusted mean values of SBP and DBP. The difference in adjusted mean values between the above two occupations were 12.9mmHg for the SBP (P=0.006) and 10.8mmHg for the DBP (P=0.001), respectively. Current smokers had 1.5mmHg (p=0.18) lower level of the mean DBP adjusted for age, residence, occupation, BMI and weekly alcohol consumption than non-or ex-smokers. In Table 2-(1), (2), (3) and (4), the adjusted mean SBP and DBP are compared between the categories of intake frequency in each food. The adjusted mean SBP and DBP were markedly lower in individuals with five times and over / wk than in those with less than once / wk in each intake of dairy products, coffee, egg and fruits. These differences in adjusted mean SBP and DBP were as follows : 5.8mmHg (p=0.003) and 5.OmmHg (p<0.001) for dairy products, 5.5mmHg (p=0.006) and 3.2mmHg (p=0.02) for coffee, 5.7mmHg (p=0.089) and 3.OmmHg (p=0.198) for egg, and 4.7mmHg (p=0.08) and 3.7mmHg (p=0.045) for fruits, respectively. Furthermore, there were inverse associations of the adjusted mean SBP or DBP with the intake frequency in each of beef, pork and chicken (p for trend=0.11 for SBP with beef, 0.066 for DBP with pork and 0.159 for DBP with chicken, respectively), whereas fishes other than sardine, mackerel or Pacific saury was a positive correlate (p for trend=0.035 for SBP and 0.033 for DBP, respectively). As regards the intake of miso soup, the adjusted mean SBP and DBP were 2.4mmHg (p=0.251) and 2.OmmHg Table 1. Profiles of 473 subjects analyzed in the present study BMI : Body mass index, SBP : Systolic blood pressure, DBP : Diastolic blood pressure Table 2-(1). Adjusted mean values t of systolic and diastolic blood pressure by the frequency of food intake t : adjusted for age, residence, occupation, body mass index and weekly alcohol consumption, S13F : Systolic blood pressure, DBP : Diastolic blood pressure  15) lower in `five times and over / wk' group than in `less than five times / wk' group, respectively. Next, we analyzed the relationships of the adjusted mean SBP and DBP to several dietary patterns. However, in this study, we present the results of a dietary pattern identified according to the intake frequency of dairy products, fruits, meat (beef and pork) and chicken which were found to be related to the SBP and DBP. In this analysis, regarding `once and over / wk of dairy products intake', 'once and over / wk of fruits intake', and `three times and over / wk of intake of beef, pork or chicken' as the three dietary habits beneficial to blood pressure, the subjects were classified into the following three groups ; 'individuals who possess none or only one of the three habits', 'individuals who possess two of the three habits', and `individuals who possess all (=three) of the three habits'. Table 3 shows the adjusted mean SBP and DBP by the three groups. In this table, the two results calculated according to the two linear models are shown. In model 1, the five items as shown in `methods' (age, residence, occupation, BMI and weekly alcohol consumption) were included as the variables adjusted for. Model 2 was further adjusted for the three categories of coffee intake and intake of the fishes other than sardine, mackerel or Pacific saury, and the two categories of miso soup intake and smoking habit (current smoker, or non-or ex-smoker) which were also found to be more or less associated with SBP or DBP. The two results were quite similar. The number of the accompanying habits of the aforementioned three dietary habits was clearly inversely associated with the adjusted mean SBP and DBP (p for trend < 0.001 for SBP and DBP, in each of model 1 and model 2), and the differences in the adjusted mean values between the `zero or one' group and the `three' group in model 1 were 7.4mmHg for the SBP (p<0.001) and 6.9mmHg for the DBP (p<0.001), respectively. Additionally, the adjusted mean SBP and DBP were compared between the following three subgroups of the 'two' group ; (1) individuals whose intake habit of dairy products solely did not meet the above conditions (N=46), (2) individuals whose intake habit of meat or chicken solely did not meet the above conditions (N=179), (3) individuals whose intake habit of fruits solely did not meet the above conditions (N=15). In model 1, the adjusted mean SBP and DBP in these three subgroups were as follows; (1) Table 4, the adjusted mean values of daily nutrient intake are shown by the number of the accompanying habits of the three dietary habits in the 157 volunteers with 3-days weighed food records. It was found that the adjusted mean values of total protein, animal protein, lipid, animal lipid, and potassium were all markedly higher in the `two' or the `three' group than in the `zero or one' group (p<0.05, for each). In calcium, phosphorus, iron, magnesium and fiber also, the positive associations between adjusted mean values of daily intake and the number of the accompanying habits of the three dietary habits were noted, although the p-values for trend did not reach the 0.1 level for these five nutrients. Total energy, Table 3. Adjusted mean (*SE) values of systolic and diastolic blood pressure according to the number of the accompanying habits of the three dietary habits + beneficial to blood pressure.
SBP : Systolic blood pressure, DBP : Diastolic blood pressure, + : 'once and over per week of dairy products intake', 'once and over per week of fruits intake', and 'three times and over per week of intake of beef, pork or chicken', ++ : model 1 was adjusted for age, residence, occupation, BMI and weekly alcohol consumption, & : model 2 was adjusted for age, residence, occupation, BMI, weekly alcohol consumption, the three categories of coffee intake and intake of the fishes other than sardine, mackerel or Pacific saury, and the two categories of miso soup intake and smoking habit

DISCUSSION
The present study suggested that in middle-aged Japanese men, habitual intake of dairy products, fruits, and meat or chicken may lower blood pressure. These results potentially reflect the beneficial influences of dietary protein, potassium, calcium and fiber which are considered to have favorable effects on blood pressure from the previous studies 2-10). This inference is supported by the results of nutrition intake data. The intake amount of protein, animal protein and potassium were substantially greater in the individuals who frequently consumed all of dairy products, fruits, and meat or chicken than in those who frequently consumed none or only one of these foods. Calcium was also positively associated with the number of the accompanying habits of the above three food intake habits, although the association seemed relatively weak. Since the nutrition survey was confined to 157 volunteers of 473 subjects, there may be a selection bias for this survey to some extent. However, the results of nutrition survey in Table  4 were considered to be well compatible with the group-specific nutrition intake status expected from the food frequency questionnaire for the entire subjects in Table 3. Sodium intake did not substantially differ between these three groups. It may imply that the observed relationship to blood pressure of this dietary pattern is unlikely to be explained by the sodium intake. However, we should be aware that the measurement error of sodium intake may not be small. Tsubono Y 31) indicated that the seasonings such as soysauce and salts were the major contributors of sodium consumption among Japanese by calculating the contribution rates of food items for the total population intake of nutrients in 180 Japanese men and their 155 spouses. Since the 3-days weighed food records may be inaccurate for the intake amount of seasonings, the validity of sodium intake level obtained from the records may be relatively low. In this sense, the relationship between the above dietary pattern and sodium intake should be further investigated by another research using sodium excretion level in 24-hours urine sample or a specifically designed semiquantitative food frequency questionnaire 31). The above dietary pattern was determined by the intake frequency of dairy products, fruits and meat or chicken, and not by egg intake. However, the observed inverse relationship of egg, one of the major sources of animal protein, to blood pressure might further ensure the beneficial effects of animal protein intake on blood pressure. Similarly, it may be interesting that miso soup intake was, though weakly, inversely associated with blood pressure. Miso soup is made from soybean which is rich in vegetable protein.
Thus, this finding might also support the blood-pressure-lowering roles of protein. Historically, the increased protein intake had been thought to lead to a sustained increase in renal blood flow and glomerular filtration, and a chronic l intrarenal hyper-tensive statet32.33) . This concept had been linked to the assumption that protein intake has the adverse effects on blood pressure even among healthy individuals. However, recent results from several observational studies 9.10.34.37) strongly suggested the blood-pressure-lowering effects of protein intake or animal protein intake among general populations. Although Sacks FM,et al.l5) demonstrated a lower blood pressure of vegetarians compared with omnivorous populations, the beneficial effects of dietary protein cannot be denied only from this finding. Also, it should be noted that the results among the people in Western countries cannot be simply applied to Japanese people. The overall intake amount of protein is undoubtedly quite different between Japanese and Western countries' people, and it is possible that the effects of dietary protein on blood pressure depend on the intake amount. However, we cannot see any intervention studies clarifying that the adequate intake of protein or animal protein actually lower blood pressure among human. It is because such intervention studies are still sparse, and the study designs among these few trials 38,39.40) are considered to be relatively inappropriate for verifying this subject. Further valid observational and experimental studies will be required to assess the significance of dietary protein or dietary amino acids as determinants of blood pressure.
On the other hand, potassium and calcium have been designated as the beneficial agents for blood pressure almost concordantly by previous observational or experimental studies 3-7). Therefore, there appears to be no problems in concluding that a reduced blood pressure among the individuals with all of the above three dietary habits may be in part due to the abundant intake of potassium and calcium. It was also found that the intake levels of magnesium and fiber were positively associated with the number of the accompanying habits of the aforementioned three dietary habits. As shown in the section of `Methods' , magnesium and fiber intake levels were likely to be underestimated, since these levels were computed only from the foods of which amount of magnesium or fiber per gram was documented in the Standard Tables of Food Composition in Japan. However, since the foods rich in magnesium or fiber were almost all selected as the foods with data of the amount of magnesium or fiber in the Tables, this observed positive relationship appears not to be a biased outcome. It may not be also ignorable that the adjusted mean intake of lipid and animal lipid were positively associated with the number of the accompanying habits of the three beneficial habits. It is uncertain whether dietary lipid or animal lipid can reduce blood pressure. Conversely, several studies 10,16,18.41) suggest that total fat or saturated fat is positively related to blood pressure, while the ratio of dietary polyunsaturated to saturated fatty acids is inversely associated with blood pressure. Since the noted positive relationship between lipid or animal lipid and the number of the beneficial habits was cleared off after further adjustment for total protein intake, the apparent relationship may be mostly derived from the confounding effects of the protein intake.
Whereas it is pointed out 13,14) that caffeine intake can raise blood pressure in the short term, the results from Multiple Risk Factor Intervention Trial (MRFIT)10) and another study11) showed the inverse relationship of usual caffeine intake to blood pressure levels. The results of Table 2-(2) in the present study may also reveal the blood-pressure-lowering effects of habitual caffeine intake by coffee.
It was interestingly found that there was a positive association between blood pressure and the intake frequency of fishes other than sardine, mackerel or Pacific saury. The reason for this finding was unclear. Since sodium intake was almost identical between the frequency categories of this food, sodium might not explain this association. A particular fatty acid or an amino acid specific in these fishes might contribute to raising blood pressure. But it is also possible that this association only reflect the confounding effects of other unknown factors.
The important finding of this study is that the reduction of blood pressure may be strengthened, especially when habitual intake of dairy products, fruits and meat or chicken occur together. Among the individuals with only `two' of the three beneficial habits, the adjusted mean SBP and DBP were all above the values of the 'all (=three)' group in any of the three subgroups within the 'two' group. Accordingly, each of the three beneficial habits are likely to contribute to further decrease in blood pressure in addition to the effects of the other two habits among the `all' group. As mentioned above, these results may show the combined effects of dietary protein (especially animal protein), calcium and potassium. However, we should be aware that each or the combination (data are not shown) of alternative foods rich in these nutrients, such as vegetables or fishes, was not associated with reduced blood pressure. Therefore, the clear relationship shown in Table 3 may be due not only to the mere combined effects of dietary protein, calcium and potassium but also to the favorable effects of other nutritional components in dairy products, fruits and meat or chicken. Or the absorption or the bioavailability of these beneficial nutrients might be enhanced especially through the intake of the above three foods.
In conclusion, this study attempted to illustrate some aspects of the overall effects of food intake and dietary patterns, and, as a result, may offer some clues in actual dietary life for preventing hypertension.