Lifestyle and Health Related Factors Among Randomly Selected Japanese Residents in the City of Sao Paulo, Brazil, and their Comparisons with Japanese in Japan

Our previous studies on mortality and cancer incidence showed a change in disease pattern among Japanese Brazilian in Sao Paulo, Brazil from that among Japanese in Japan. To clarify the cause of this change, a cross-sectional study was conducted for evaluating lifestyle and healthrelated factors among a representative sample of Japanese residents in Sao Paulo. The results obtained were compared with those from a cross-sectional study in five areas of Japan in which the similar protocol had been used. A questionnaire on lifestyle, anthropometric and blood pressure measurements, and blood and urine collections was administered to 411 randomly selected Japanese firstand second-generation residents aged 40 to 69 years in the city of Sao Paulo. A total of 251 (61%) subjects, 118 men and 133 women, 90 firstand 161 second-generation, participated in the study. Smoking and drinking habit, anthropometric measures and blood pressure level, and serum biochemical parameters were examined according to sex and age group. The level of total cholesterol, triglycerides and uric acid in sera, body weight and body mass index, and both systolic and diastolic blood pressures revealed a significantly higher value when compared with five Japanese groups in Japan, while serum HDL cholesterol and the percentage of smokers and drinkers were lower. These differences of lifestyle and health-related factors were discussed in relation to ischemic heart and cerebrovascular diseases and some types of cancer. J Epidemiol, 1994; 4 : 37-46.


SUBJECTS AND METHODS
Subjects of this study were selected from the database of a special census survey for Japanese population in Brazil8). The survey conducted a brief interview to find out any person of Japanese origin in all households located in some randomly selected areas. If at least one family member was recognized as Japanese origin, the name, sex, birth date and birth place were ascertained for all members in that family. In the city of Sao Paulo, 264 randomly selected geographic areas were surveyed ; among the selected areas, a total of 273 households which included at least one family member of.Japanese origin were identified. We selected the subjects of this study from these 273 households, on condition that they were first-generation Japanese immigrants or second-generation Japanese descendants, both parents were Japanese, and aged 40 to 69 (as of June 30, 1989). According to this condition, a total of 411, 147 first-and 264 second-generation, Japanese residents were selected for this study. The distribution of these subjects was shown in Table I by sex, age-class and  generation. We mailed bilingual (Portuguese and Japanese) letters to all subjects twice to ask their participation to this study. The first letter submitted approximately one month before the starting day explained the goal and importance of the study and assured the confidentiality of information and the right to refuse. The second letter, 20 days later from the first one, made a strong request for participation in the study and informed the individual of the place and date of the examination. The confirmations were made by a reply stamped letter and by telephone.
Two places were settled for the examination of participants, one in the Japanese clinic affiliated with Beneficencia Nipo-Brasileira de Sao Paulo (Japan-Brazil beneficent organization) and another in the Health Center attached to the School of Public Health of University of Sao Paulo, both located in the central of city. We performed the following items to all participants mostly at our centers and partly in their homes.

QUESTIONNAIRE BY INTERVIEW
The questionnaire used in the Japanese cross-sectional study was translated to Portuguese with minimal revision of some particular items. Trained nurses or nutritionists conducted the interview. The content of this questionnaire included personal data with immigration history, medical history, smoking and drinking history, dietary pattern, physical activity level, stress history, a personality assessment and others. An assessment of gynecological and reproductive history was included for women.

HEALTH EXAMINATION
A brief health examination was performed for anthropometric measures such as height and weight. To measure blood pressure we used an automatic manometer (Takeda Medical, UA751) to avoid observer bias and to maintain comparability with the cross-sectional study in Japan. Three consecutive measures at right arm in the seated position were done. The first measure was used for detecting appropriate maximum pressure level and the average of second and third measurement was used for the actual measurement.

BLOOD AND URINE COLLECTIONS
A total of 23 ml of blood was collected by venous puncture using 25 ml syringe with 21 gauge needle. We Table 1. Distribution of subjects by sex, generation and age class. requested the subjects fast at least 5 hours prior to the collection. The average fasting time was 10.8 hours with standard deviation of 4.1 (range : 4-21 hours). The blood was divided into two tubes ; I I ml into heparinized tube and 12 ml into a tube without anti-coagulant. The tube with heparin was immediately centrifuged for 10 minutes at 2,500-3,000 rpm to get a plasma and buffy coat layer (white blood cell) fraction. The other tube was left for an hour at room temperature to facilitate clotting and centrifuged by the same procedure to separate serum. The plasma, huffy coat layer (white blood cell) and serum were separated in several kinds of tubes and frozen in a ice box with sufficient dry ice until being sent to Japan, where it was stored in deep refrigerator at -80°C. Serum and plasma samples were distributed to several laboratories for analysis. We conducted standard biochemical analysis by autoanalyzer (HITACHI 736) to measure total cholesterol, HDL-cholesterol, triglyceride, and uric acid. We also conducted vitamin analysis such as 8-carotene, atocopherol, retinol, hormone analysis such as sex hormones, steroid hormones, trace element analysis such as selenium, copper, zinc, viral marker analysis such as hepatitis B, hepatitis C, HTLV-1. The results will be reported separately.
Some data from males 40 to 49 years of age were compared with Japanese correspondent subject groups of the cross-sectional study in five areas in Japan (Ninohe in Iwate, Ishikawa in Okinawa, Yokote in Akita and Saku in Nagano Prefectures, and Katsushika-kita in Tokyo Metropolis)6,7). In this cross-sectional study using similar protocol, the subjects were randomly selected men aged 40 to 49 years and therefore no comparable data is available for women and men over 50 years of age. The test of difference between Japanese Brazilian and each of five Japanese areas was done by using t-tests for continuous variables and chisquare tests for categorical variables.
The study was carried out during August and September 1989, which corresponds to the winter season in Sao Paulo, the same season with the cross-sectional study in Japan.

RESULTS
Participation rate 251 subjects (61%), 118 men (58%) and 133 women (64%), 90 first-generation (61%) and 161 second-generation (61%), responded to this survey and their distribution by sex, age-class and generation was shown in Table 2. Five participants were excluded from the successive analysis because their ages had been recognized to be out of the age range (40 to 69 years old) after their participation. The details of the five participants were 32 years old secondand 70 years old first-generation men, and 35 years old second-, 72 years old first-and 77 years old secondgeneration women. These phenomenon occurred by misidentification of their age in the Japanese census survey which was original source of our study population.
Within the group of 160 non-respondents, 140 refused to participate, 9 were stayed in Japan to work, 8 were failed to contact and 3 were deceased.

Immigration history of participants
The average year of immigration to Brazil was 1947Brazil was [range: 1925Brazil was -1980 in men and 1941  in women among forty-one men and forty-four women of first-generation (three unknown). Twenty men (49%) and thirty-one women (70%) entered in Brazil before the Pacific War (1941-45). The original prefectures in Japan (in the case of second-generation, original prefectures of their fathers were applied) were widely spread from north to south, however concentration was found in some prefectures. Okinawa had the largest number (30), which was followed by Kumamoto (22), Hokkaido (20), Fukuoka (13), Fukushima (12) and Nagano (12). Table 2. Number of participants and responce rates (%) in parenthesis by sex, generation and age class. 5 participants (32yo, second-and 70yo. first-generation men, and 35yo. second-, 72yo. first-and 77yo. second-generation woen) were out of age range from 40 to 69 y.o. per day in males. Almost one third of males were exsmokers, while 90% of females were found to be neversmokers.
The percentage of males classified as daily alcohol drinkers was higher than females, 31.3% in men and only 2. 3% in women. Weekly ethanol consumption among male drinker was 189 g, in which more than a half (100 g) came from beer and one third (53 g) from pinga (distilled liquor from sugar cane).

Anthropometric measures and blood pressure
The means and standard deviations of the height, weight, body mass index (kg/m2), and blood pressure level were shown in Table 5. The height and weight decreased slightly in older age groups, and were higher in males than Daily smoker (%)  compared with Japanese in Japan ( Figure. 1). Body weight was highest in Sao Paulo, while height was the second lowest followed by Okinawa, although significant difference was only found in comparison with Akita for weight ( Figure. 2). This means that Body Mass Index was almost highest together with Okinawan Japanese and was significantly higher than Akita and Tokyo. The Height (cm) levels of both systolic and diastolic blood pressure were highest among Japanese Brazilian.
Diastolic blood pressure among them showed the significant higher values when compared with any of five Japanese groups ( Figure . 3).
The levels of total cholesterol showed the highest values, while HDL cholesterol the lowest (Figure. 4). This implies that the level of LDL cholesterol is much higher than Japanese in Japan. Both uric acid and triglyceride revealed significantly higher values than any of five Japanese groups (Figure. 5).

DISCUSSION
In this cross-sectional survey, we placed emphasis on the representativeness of our target population, from which data on mortality and incidence of cancer were obtained. The final response rate was 61% as a whole. This may be not sufficient to represent Japanese Brazilian in the city of Sao Paulo ; however considering the size of city, having approximately twelve million people, this rate was relatively higher than we expected. In our cross-sectional study in Japanese five regions, Katsushika-kita area located in metropolitan Tokyo showed the same response rate of 61%6>, which was the lowest compared with other local areas. However, Katsushika-kita area is a small part of Tokyo metropolis and has only 150 thousand population.
As one of standards for estimating potential selection bias, we compared the distribution of the original prefecture in our participated subjects with the statistics from emigrants to Brazil according to prefecture in Japan. The leading four prefectures of origin were Okinawa (12%), Kumamoto (9%), Hokkaido (8%), and Fukuoka (5%) in this study, while Kumamoto (10%), Fukuoka (8%), Okinawa (8%), and Hokkaido (7%) were described in the literature9>. This suggests that our study participants is not a biased sample from Japanese residents in the city of Sao Paulo.
The level of total serum cholesterol was much higher among Japanese Brazilian than each of five Japanese groups. This finding strongly suggests that higher mortality rate from ischemic heart disease observed in firstgeneration Japanese residents in Sao Paulo is linked to the increase level of serum cholesterol. The higher level of blood pressure and body mass index, and lower value of HDL cholesterol are also likely related to increase risk of ischemic heart disease. Despite the higher rates of ischemic heart disease, Japanese Brazilian smoked less. The NI-HON-SAN Study reported a two-fold increase of coronary heart disease among Japanese men in Hawaii and the three-fold increase among Japanese men in California when compared with Japanese in Japan10) and this increase was also correlated with the elevated levels of total cholesterol and blood pressure among Japanese American11, 12) The higher level of total cholesterol, as well as that of uric acid, among Japanese Brazilian could be associated with high intake of meat, especially beef. Our data from this study revealed that the proportion who consumed beef almost every day was 40% among men aged 40 to 49 years in Sao Paulo, whereas the proportion in Japan was only 0% in Iwate, 1% in Akita and Tokyo, 3% in Nagano and 7% in Okinawa (Tsugane S, unpublished data).
The high fat intake among Japanese Brazilian, which could also be related to the high total cholesterol and uric acid in sera, could be also associated with higher mortality from and incidence of prostate and breast cancer [13][14][15]. The risk of colorectal cancer is also considered to be associated with fat intake13,15,16); however both mortality and incidence are not increased among Japanese Brazilian. Other factors, such as fiber and vegetables intake, may account for the difference.
Both systolic and diastolic blood pressure were higher among Japanese Brazilian, although the reason of these increases are unclear.
The higher body mass index among them may be partially related.
The high level of blood pressure can also increase the risk of ischemic heart disease. However, since blood pressure is one of the most important risk factors for stroke 17), an increased mortality from cerebrovascular disease would be expected.
On the contrary to this expectation, SMR values for cerebrovascular disease among first-generation Japanese residents in Sao Paulo were as low as 74% in men and 86% in women, both statistically significant, when compared to Japanese in Japan.
The low level of total cholesterol was considered as a risk factor for bleeding type of stroke18), so that higher level of total cholesterol may protect the occurrence of cerebral bleeding among Japanese Brazilian whose blood pressure are high.
Since the method of sampling was strictly dependent on the original Japanese census survey, we could not have regulated the sample size. This caused the small number of participants to find the difference between the first-and the second-generation in each age-class. We tried to find it for anthropometric measure, blood pressures and biochemical parameters by using analysis of covariance treating age as covariate, however no significant difference could not have been obtained (data not shown). This is probably due to a lack of statistical power and another study is necessary to test the difference of lifestyle factors by generation.
In conclusion, although precise and comparable nutritional data are not available, higher level of serum cholesterol and uric acid as well as body weight suggest the higher intake of total calories and fat among Japanese Brazilian.
This change of dietary habit may have caused the increased incidence of ischemic heart disease and cancers such as prostate and breast, and the decreased incidence of stroke among them.