Possible Protective Effect of Milk, Meat and Fish for Cerebrovascular Disease Mortality in Japan

Cerebrovascular disease was a leading cause of death from 1955 to 1980 in Japan. The mortality rate from this disease has decreased sharply in recent decades. This downward trend seems to correspond to the dietary habits of Japanese. Data from a large prospective cohort study were analyzed to examine the association between dietary habits and cerebrovascular disease mortality in Japan. The subjects for this analysis were 223,170 men and women aged 40 to 69 at baseline in December 1965. There were 6,168 deaths in men and 4,862 deaths in women due to cerebrovascular disease (ICD7: 330-334) during the follow-up period from January 1966 to December 1981. Rate ratio (RR) and 95% confidence interval (95%Cl) adjusted for sex, attained age, follow-up period, prefecture, cigarette smoking, alcohol drinking and occupation was used for comparison. In this study, the risk of mortality from cerebrovascular disease was inversely associated with dairy milk, meat and fish consumption. Therefore the joint effect of dairy milk, meat and fish (DMF) as animal fat and protein was of interest. In the binary analysis, DMF (D, M, F) means the combination of dairy milk (1-3 times/week or more), meat (1-3 times/week or more) and fish (4 times/week or more). Thus DMF (d, m, f) was the reference group having dairy milk (less than 1 time/week), meat (less than 1 time/week) and fish (less than 4 times/week). For the disease, the RR of DMF (D, M, F) was 0.68 with 95%CI of 0.63 to 0.74, relative to the reference group. Furthermore the joint effect of DMF was more strongly associated with cerebral haemorrhage (ICD7: 331, DMF (D, M, F); RR: 0.63, 95%Cl: 0.55-0.70) than with cerebral embolism and thrombosis (ICD7: 332, DMF (D, M, F); RR: 0.79, 95%Cl: 0.700.89). These findings suggest that the increasing intake of animal fat and/or protein may have played a key role in reducing cerebrovascular disease in Japan. J Epidemiol, 1999 ; 9 : 268-274.

animal sources.Previous studies under stroke prevention programs based on blood pressure control with health education have consistently shown that the serum total cholesterol levels were inversely associated with incidence of cerebral hemorrhage.5-10)International studies of stroke in Japanese men living in Japan, Hawaii have found the inverse association of animal fat and protein intake and all stroke mortality (NI -HON-SAN study).11-13)Accordingly, it is very important for stroke prevention to evaluate the Japanese dietary habits, especially for animal fat and protein.
A six-prefecture study was carried out as a large prospective cohort study in Japan from 1965 to 1982.14)This study selected the six prefectures from northern to southern Japan and collected the data on dietary habits and other factors from over two hundred fifty thousand residents.Therefore it seems useful to examine the association between dietary habits and cerebrovascular disease mortality in this population.
Purpose of present study is to elucidate the associations between dietary habits and cerebrovascular disease mortality among the Japanese from the large cohort study.

Profile of cohort
A cohort of 265,070 inhabitants aged 40 or over, from 29 public health districts in six prefectures (Kagoshima, Okayama, Hyogo, Osaka, Aichi and Miyagi) was followed-up from 1966 to 1981 by Dr. Hirayama and his colleagues.14)A one-page questionnaire used in the baseline survey included questions about occupation, cigarette smoking, alcohol drinking, other dietary habits and so on.At the beginning of each follow-up year, a migration survey was conducted through reference to the local residence registration.During the follow-up period, the deaths were annually ascertained by vital statistics kept at each public health center.The causes of deaths were coded by Dr. Hirayama according to the 7th revision of the International Classification of Disease (ICD-7).The study method, age and sex distribution of the cohort have been reported elsewhere 1x, 16) In the present analysis, occupation, smoking and alcohol drinking were treated as potential confounding factors.Occupation was classified into the following 5 categories taking into account average years of schooling, 14) (I) farmers, lumbermen, fishermen, and workers in mining and quarrying, (II) workers in transport and communication, craftsmen, production process workers and labourers, (III) clerical and related workers, sales workers, and service workers, (IV) professional and technical workers, managers and officials, (V) job unknown including housewives.These categories can be considered as one of surrogates for socioeconomic status.
Cigarette smoking was categorized to the three categories, i.e., non-smokers (never smoked), ex-smokers, and current smokers (a cigarette per day or more).Alcohol drinking was divided into three categories, i.e., non-drinkers, not-daily drinkers (1 time/month to 3 times/week) and daily drinkers (4 times/week or more).Dairy milk, meat, fish and other dietary factors were also divided into three categories, i.e., less than 1 time/week, 1-3 times/week, and 4 times/week or more.To clarify the joint effect of animal protein and fat, dairy milk, meat and fish (DMF) were combined.For example, DMF (D, M, F) means the combination of dairy milk (1-3 times/week or more), meat (1-3 times/week or more) and fish (4 times/week or more).Thus DMF (d, m, f) was the reference group, which was for dairy milk (less than 1 time/week), meat (less than 1 time/week) and fish (less than 4 times/week).
Analysis of prior disease history at the baseline showed that the mortality rate of all causes of death from the subjects with cancer or other diseases except for chronic stomach disease at recruitment were higher than those with no prior disease history at recruitment.However, the subjects with prior disease history of stomach disease was not significantly different from those with no prior disease.Therefore the subjects who reported a history of cancer and other diseases except for chronic stomach diseases at recruitment (n=7,612) were excluded from present analysis.Following subjects were also excluded: (1) those who were 70 years old or more at recruitment (n=4,490), ( 2) those who had missing information on dietary habits and smoking at recruitment (n=29,798).After this exclusion, a total of 223,170 cohort members remained for analysis.The mortality rate of cerebrovascular disease from these excluded cases was not significantly different from the rate of remained subjects.

RESULTS
Table 1 shows the number of deaths by type of cerebrovascular disease.For all cerebrovascular disease (ICD7: 330-334), 11,030 deaths in total were observed during the study period.Proportion of cerebral haemorrhage (ICD7: 331) and cerebral embolism and thrombosis (ICD7: 332) were about 43% and 37%, respectively.Percent distribution by type of cerebrovascular disease for men was similar to that for women.
Table 2 shows the frequency distribution of dietary factors among men and women.Proportion of dairy milk with less than 1 time/week was about 58%, and that with 4 times/week or more was 21.5%.About 70 % of men and women had meat with 1-3times/week.Only 6% of men and women consumed fish less than 1 time/week.Distribution of miso soup was similar to that of green-yellow vegetables, i.e., over 70% of men and women consumed those 4 times/week or more.For pickles, about 84% of men and women consumed it 4 times/week or more.Not shown in this table is that over 95% of men and women consumed rice daily.Therefore typical diet in this population was rice, pickles, miso soup and green-yellow vegetables.
Table 3 shows the risk of mortality for all cerebrovascular disease associated with tobacco, alcohol and occupation.Rate ratios presented here were adjusted for sex, attained age, follow-up interval and prefecture.The mortality risk of current smokers was increased slightly (RR: 1.12, 95%CI:1.06-1.18),but no statistically significant difference was observed between non-smokers and ex-smokers.There was J-shaped relation between alcohol consumption and mortality, thus, the RR for daily drinking was significantly higher than non-drinking (RR: 1.17, 95%CI: 1.10-1.24),but the RR for not-daily drinking (1 time/month to 3 times/week) was significantly lower than nondrinking (RR: 0.92, 95%CI: 0.87-0.97).For occupation, the RR compared to group I (farmers, mining) was significantly higher for group V (job unknown) (RR: 1.20, 95%CI: 1.13-1.26),but the rate ratios from other job groups were significantly lower, especially for group IV (professional, manager; RR: 0.68, 95%CI: 0.60-0.77).
Tables 4 gives the mortality risks on dietary factors by type of cerebrovascular disease.No statistical association was observed for miso soup or green-yellow vegetables.A significantly lower mortality risk was observed in pickles with 4 times/week or more compared to the reference category of less than 1 time/week (RR: 0.91, 95%CI: 0.83-0.99);this association was similar for cerebral haemorrhage and for cerebral embolism and thrombosis, but these associations were not quite statistically significant.With regard to dairy milk and meat, significantly lower mortality risks compared to the reference category were observed in the both categories with 1-3 times/week and with 4 times/week or more for all cerebrovascular disease and for cerebral haemorrhage.In relation to fish consumption, a significantly lower mortality risk was also observed in the category of 4 times/week or more for all cerebrovascular disease (RR: 0.86), and for cerebral haemorrhage (RR: 0.87).For cerebral embolism and thrombosis, a significantly lower mortality risk was observed in the category of 4 times/week or more in dairy milk (RR: 0.85), and in meat (RR: 0.81).
In order to assess the joint intake of dairy milk, meat and fish (DMF), data were recategorized to binary data.In this analysis, DMF (D, M, F) means the combination of dairy milk (1-3 * adjusted for sex, attained age, follow-up interval and prefecture ** I time/month to 3 times/week *** 4 times/week or more **** (I) farmers , lumbermen, fishermen, and workers in mining and quarrying (II) workers in transport and communication, craftsmen, and production process workers and labourers (III) clerical and related workers, sales workers, and service workers (IV) professional and technical workers, managers and officials (V) job unknown included in housewives, and not reported # Green-yellow vegetables * adjusted for sex, attained age, follow-up interval, prefecture, alcohol drinking, smoking and occupation.
times/week or more), meat (1-3 times/week or more) and fish (4 times/week or more).Thus DMF (d, m, f) was the reference group consuming dairy milk (less than 1 time/week), meat (less than 1 time/week) and fish (less than 4 times/week).Table 5 shows the joint effect of DMF by type of cerebrovascular disease.This effect was more strongly associated with cerebral haemorrhage (DMF (D, M, F); RR: 0.63, 95%CI: 0.55-0.70)than for cerebral embolism and thrombosis (DMF (D, M, F); RR: 0.79, 95%CI: 0.70-0.89).For all cerebrovascular disease, the RR of DMF (D, M, F) was 0.68 with 95%CI of 0.63 to 0.74, relative to the reference group.In addition, the results from separate analyses for men and women, non-smokers and smokers, and rural and urban areas were similar to the Table 5 (data not shown).

DISCUSSION
In this study, dairy milk, meat and, to a lesser extent, fish were each associated with a reduction in mortality from cerebrovascular disease.This finding is in line with the Ni-Hon-San study in which the intakes of animal protein and saturated fat were inversely associated with stroke incidence in Japanese men.11-13) A recent report from the Framingham Study has also shown an inverse association between dietary fat and ischaemic stroke (mainly cerebral embolism and thrombosis) mortality in men.19) The Multiple Risk Factor Intervention Trial showed an inverse association between serum total cholesterol and mortality from cerebral haemorrhage but a positive association between serum total cholesterol and mortality from cerebral infarction.20) A recent meta-analysis of 13 cohorts Table 5. Joint effect of dairy milk, meat and fish for cerebrovascular disease mortality from six-prefecture cohort (1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981).
* adjusted for sex , attained age, follow-up interval, prefecture, cigarette smoking, alcohol drinking and occupation * 1 Dairy milk consumption (D : 1 -3 times/week or more , d : less than 1 time/week) *2 Meat consumption (M : 1 -3 times/week or more , m : less than 1 time/week) *3 Fish consumption (F: 4 times/week or more , f : less than 4 times/week) cerebrovascular disease (Table 4).Unfortunately, no data on fruit intake were available in our study.
In conclusion, intakes of dairy milk, meat and fish were inversely associated with cerebrovascular disease mortality in this population, the association being particularly strong for deaths from cerebral haemorrhage.Increasing consumption of these foods over the past 30 years may have helped to reduce cerebrovascular disease mortality in Japan.

Table 1 .
Number of death for cerebrovascular disease from six-prefecture cohort durine the neriof from 1966 to 1981 .

Table 2 .
Frequency distribution of dietary factors among men and women from six-prefecture cohort(1966 -1981).