A 20-year Follow-up Study of the Relation Between Life Style and Stroke in the Oki Islands , Shimane Prefecture , Japan

Cross sectional comparison of food intake patterns and smoking and drinking habits with 1968 and 1987/88 in a 20-year follow-up study was carried out , in the Oki Islands, Shimane Prefecture, Japan. A total of 1,140 males and 1,694 females were randomly selected from the participants of the first survey (in 1968). The proportion of previous hypertension was significantly higher in the 397 inhabitants who died from stroke during the follow-up period than in 397 survivors matched for age and sex (p<0.01). Among males, the proportions of those who smoked and drank daily were significantly higher in the stroke death group than in the survivor group, in contrast, their daily consumption of eggs and milk were significantly lower (p<0.05). There were no differences in the daily intake of miso soup and pickles between the two groups in either sex. Previous hypertension and drinking were significant and independent risk factors for stroke in male. It is important for primary prevention of stroke to improve the 'salt-rich' diet of those who used added salt either while cooking or at the table, because they have unhealthy life style, smoking and drinking daily and not taking eggs, milk and fruits daily.

World War, but is still lower than in urban areas.Health care is provided by 30 general practitioners, two hospitals and two health centers.The first survey in 1968 covered 3,546 males and 4,350 females, or 64% of the male and 65% of the female residents aged 40-74.
The questionnaire was composed of questions on past medical history, subjective symptoms, smoking, drinking, dietary habits and medication.
The dietary habits of the participants were ascertained by a food frequency questionnaire composed of 11 items.Food intake was estimated by four grades of intake frequency : daily, weekly, monthly and rarely/never.The form was mailed to each participant at the beginning of the field survey and collected by doctors and public health nurses according to a standardized protocol.
A follow-up risk factor survey using to the same protocol employed in 1968 was carried out from 1987 to 1988 (Table 1).The participants who had died or moved out between 1968 and 1986 were excluded from the subjects of the follow-up survey.As the subjects got along during the follow-up period, investigators had to visit them to avoid incomplete response at the second survey.It would be better to survey random samples of subjects to get the representative results of the population considering less investigators and money.A total of 1,140 males and 1,694 females were randomly selected from a list of the participants of the initial survey.The overall response rate was 91.2% for males and 88.9% for females.
Our study focused on three particular points.First, the daily smoking and drinking habits and daily intake of food in those with stroke death were compared with those of survivors during the follow-up period.Survivors were selected from the list of the second survey, using the method of pair-matched sampling with respect to sex and age ( 5 years).The average age of stroke fatalities in 1968 (male ; 62.6 7.4 years, female ; 63.1 8.4 years)(M SD) was significantly older than that of survivors (male ; 51.5 8.3 years, female; 51.4 8.3years)(M SD), respectively (p< 0.01) for both males and females.
Second, we examined whether the community plan for stroke prevention (CPSP) carried out from 1969 to 1971 was an important factor in changing the food intake pattern and smoking and drinking habits.Nishinoshima executed CPSP because of the high mortality rate from stroke (about 370 per 100,000) in the 1960s.Ama was chosen as control because of its low mortality rate (about 230 per 100,000).
Third, the daily smoking and drinking habits and daily intake of foods between 1968 and 1987/88 were analyzed according to three grades of salt intake groups : the 'salt-rich' diet group who used added salt either while cooking or at the table in both 1968 and 1987/88 (A), the 'saltless' diet group who did not use added salt either while cooking or at the table in both 1968 and 1987/88 (B), and the 'salt-rich' diet group in 1968 but the 'saltless' diet group in 1987/88 (C).

Statistical analyses
Missing values for any of the variables considered were excluded from analysis.Incomplete response rates of variables were small (about 10%) except namemiso, meat and milk.Means were compared using two-sample two-tailed t-tests.Initially, the values for chi-square test were calculated as described by McNemar and then potentially confounding variables were controlled by means of a conditional logistic regression model18).

Type of stroke death
After 20 years of follow up 1325 males and 955 females had died.228 deaths (17.2%) in male and 169 Comparison of the daily smoking and drinking habits and daily intake of foods between stroke fatalities and survivors (Table 2) The proportion of those with a previous history of hypertension was significantly higher among those of both sexes who died of stroke than among survivors : the difference between the two groups was 20% for males and 12% for females.
The proportions of those who smoked and drank daily were significantly higher among males who died from stroke than among survivors.The proportions of those who ate eggs daily were significantly lower among those of both sexes who died from stroke than among survivors.Daily intake of milk in males and that of green vegetables in females showed the same tendency as with eggs.
The relative risk of stroke for all factors which might have a confounding effect on smoking as a risk factor are shown in Table 3.These factors were controlled by means of multiple logistic regression analysis.Previous history of hypertension and drinking were significant and independent risk factors in male.
Comparison of daily smoking and drinking habits and daily intake of food between 1968 and 1987/88 in Nishinoshima and in Ama The average age of inhabitants was almost the same in both towns, about 51 years old in 1968.The proportion of those with a previous history of hypertension was lower in Nishinoshima than in Ama in 1968, while it was higher in Nishinoshima in 1987/88.The saltless dietary habit increased in both towns, but the difference in proportions was not significant (Table 4).
The daily intake of miso soup was significantly The daily intake of fish decreased significantly between 1968 and 1987/88 in Nishinoshima, but increased significantly in Ama.Fish intake was higher in Nishinoshima than in Ama in both 1968 and 1987/88.Remarkable increases in the daily intake of eggs, milk and fruits were observed during the last two decades in both towns.
The proportions of daily food intake in females are not shown because, except for miso soup and fish, they did not change significantly.
Comparison of daily smoking and drinking habits and daily intake of foods between 1968 and 1987/88 in the three salt diet groups (Table 5).
The average age of group A (49.0 7.3 years) was significantly younger than that of either group B (53.8 8.0 years) or group C (52.9 8.7 years).The results of group A were age-adjusted to those of group C. Daily smoking decreased significantly in groups B and C, while that in group A did not change and was significantly higher than in groups B and C in 1987/88 (p<0.01).
Daily drinking did not change in the three groups during the follow-up period.The proportion of those who drank daily was significantly higher in group A in 1968 and 1987/88 than in groups B and C Though the daily intake of pickles decreased in the three groups during the follow-up period, group A showed significantly higher pickle consumption in 1968 and 1987/88 than groups B and C (p<0.01).Nutritional improvement was remarkable, with the daily intake of eggs, milk, and fruit increasing in all three groups, especially in group C.More than half of those groups B and C consumed eggs and milk daily in 1987/88, while less than 40% of those in group A did so.
Again, results for females are not shown because tendencies in daily food intake remained the same except for fish, and the average age of group A was significantly higher than those of groups B and C.

Methodological problems in questionnaire
The food frequency questionnaire was used to measure usual food consumption in 'free living' populations.The technique has several advantages19-21) first, it is very useful in mass surveys and provides similar information to household inventories or menu assessments.Second, use of the frequency technique may minimize cost and respondent burden.Third, it may measure more accurately individual dietary habits, because of techniques based on household inventory or menu assessment.
To eliminate the disadvantage of the food frequency questionnaire, which may have recall bias because it is self-administered questionnaire, we referred to the results of household inventory or menu assessment at the same time of our surveys.
As the staff that participated in the second survey was not the same of the first except for the supervisors, there may have been observer bias.Supervisors trained the doctors and public health nurses to exclude observer bias according to a standardized protocol.
The change in the life style of subjects during the last two decades was representative of the people in Oki because of the complete follow-up survey and the high response rates of participants.

Mortality from stroke
Of all strokes during the follow-up period, the percentage of 'unspecified stroke' was the highest, about 40%, which were almost coded by 436 and 437 of the basis of the Ninth Revision of ICD.The accuracy of differentiating the subtype of stroke, especially unspecified stroke 22), is reported to be affected by 1) small population size, 2) physician's ability to make an accurate diagnosis of stroke, 3) the lack of doctors and equipment for precise clinical examination of stroke and 4) inconvenient transport system.The percentage of unspecified stroke in Oki was as high as in the isolated islands and mountain villages reported by Shibata et al22)_ Smoking, drinking, food intake and previous hypertension In Japan, epidemiological studies2-12) on stroke have revealed the relationship between hypertension control and improvement of nutrition with the prevention of stroke.The present study showed that the previous history of hypertension and daily drinking were risk factors for stroke in male.As the association between alcohol and stroke was emphasized 23,24) and the proportion of daily drinking have increased in Oki16), it is important for the recommendation to restrict alcohol intake to no more than modest level.
Although high salt intake is responsible for stroke, there was little difference between stroke fatalities and survivors in the proportions of'saltless' diet, miso soup and pickles, in the present study.A reasonable explanation is that a hypertension control program with screening for stroke was not yet underway 20 years ago in Oki.As the average age of both stroke fatalities and survivors was over 62 years old at the first survey, they may have eaten the typical Japanese meal of rice, miso soup and pickles.
CPSP was an effective system for grasping the condition of stroke patients and the benefit of health examination 25-27), but there have been few studies of investigating the relation between CPSP and life style.
There was little difference between Nishinoshima and Ama in the proportions of smoking and 'saltless' diet.The proportions of daily intake of fish, milk, and fruit in Nishinoshima in 1968 and 1987/88 were higher than those in Ama.As the effect of CPSP was not so prominent in Nishinoshima because of the lack of public health nurses, the mortality rate from stroke has not decreased during the last 10 years and the improvement of life style such as reducing smoking and pickle intake has not been performed.
The proportion of smokers has decreased significantly during the past two decades in Oki ; however, the number of male drinkers has increased remarkably 16).Kato et al. cited 28) that male smokers and drinkers consumed salty foods and less fruits and milk.It is important for primary prevention of stroke to improve the 'salt-rich' diet of those who used added salt either while cooking or at the table, because they have unhealty life style, smoking and drinking daily and not taking eggs, milk and fruits daily.

Table 1 .
Participants in the first survey (1968), and response rates for the second survey (1987/88) by age* and sex in studied cohort.

Table 2 .
Comparison of life style in stroke fatalities and survivors (1968) .
types of stroke except for cerebral hemorrhage in males (28.9%) and cerebral infarction in females (20.0%).

Table 4 .
Comparison of life style between 1968 and 1987/88 in Nishinoshima and Amaa) (Males only)

Table 5 .
Comparison of life style between 1968 and 1987/88 in 'Salt-rich' diet (A)a) and 'Saltless' diet (B, C)b) a) A : 'Salt-rich' diet group who used added salt either while cooking or at the table in both 1968 and 1987/88 b) B : 'Saltless' diet group who did not use added salt either while cooking or at the table in both 1968 and 1987/ 88, C : 'Salt-rich' diet in 1968 and 'Saltless' diet in 1987/88 (P<0.01).