RESEARCH ACTIVITIES OF EPIDEMIOLOGY IN JAPAN Report by the Research Committee of the Ministry of Education, Science, Sports and Culture on Evaluation of Risk Factors for Cancer

Lifestyle habits and other living conditions in Japanese have progressively been changing after the World War II, and the changing trend has been accelerated since 1970. The frequency and distribution of cancer by site in Japan showed marked secular changes during the past decades, just as reflecting the above changes in environmental factors. A large scaled cohort study on cancer at moment was strongly anticipated in Japan , after the cohort study by Dr. Hirayama et al. had ended around 1980 with unexpectedly fruitful results . However, financial problems and other conditions hindered to start such cohort studies. A multicentered collaborative cohort study had planned among the epidemiologists and epidemiology oriented clinicians who have been working on cancer in the communities , resolving problems on cost and others for long term epidemiological issues. A new cohort with a total of 125 thousands of healthy inhabitants living in the areas scattered throughout country was established in 1988-1990, although they were not randomly distributed in area. Some 30 thousands industrial worker cohort has joined in this study , which will be separately analysed. The study was partly granted by the Ministry of Education, Science , Sports and Culture and largely supported by local government and volunteers in each area for ten years. A research committee on this study was organized and are following up all subjects participated for more than ten years, mainly pursuing mortality status, and incidence of cancer for about half population of the cohort is also under study. About 1,000 deaths per year were observed in the first four years and more than 30% were due to cancer. The proportion of moved out of town was small being about 1 % per year. The distribution of cancer deaths by site for the first four years was similar to those of general population. This report summarized the study plan and the epidemiological characteristics of the cohort at entry of the study. It also gives a brief account of activities until 1994. J Epidemiol, 1996 ; 6 : S107-S113.

Lifestyle habits and other living conditions in Japanese have progressively been changing after the World War II, and the changing trend has been accelerated since 1970.The frequency and distribution of cancer by site in Japan showed marked secular changes during the past decades, just as reflecting the above changes in environmental factors.
A large scaled cohort study on cancer at moment was strongly anticipated in Japan , after the cohort study by Dr. Hirayama et al. had ended around 1980 with unexpectedly fruitful results .However, financial problems and other conditions hindered to start such cohort studies.
A multicentered collaborative cohort study had planned among the epidemiologists and epidemiology oriented clinicians who have been working on cancer in the communities , resolving problems on cost and others for long term epidemiological issues.
A new cohort with a total of 125 thousands of healthy inhabitants living in the areas scattered throughout country was established in [1988][1989][1990], although they were not randomly distributed in area.Some 30 thousands industrial worker cohort has joined in this study , which will be separately analysed.The study was partly granted by the Ministry of Education, Science , Sports and Culture and largely supported by local government and volunteers in each area for ten years.
A research committee on this study was organized and are following up all subjects participated for more than ten years, mainly pursuing mortality status, and incidence of cancer for about half population of the cohort is also under study.About 1,000 deaths per year were observed in the first four years and more than 30% were due to cancer.The proportion of moved out of town was small being about 1 % per year.The distribution of cancer deaths by site for the first four years was similar to those of general population.
This report summarized the study plan and the epidemiological characteristics of the cohort at entry of the study.It also gives a brief account of activities until 1994.J Epidemiol, 1996 ; 6 : S107-S113.cancer, cohort study, multicentered study, risk factors, preliminary report A large scale cohort study is an indispensable for confirming quality and magnitude of risk factors of cancer related to lifestyles, living environment and others.The results not only can provide baseline data for monitoring preventive measures, but also can enable hypothesis generation for research.
However, over 1 million person-years observation is requested to examine the association between risk or protective factors and common cancers, considering incidence or mortality of cancer in Japan.Thus, at least 10 years follow-up of 100,000 persons is needed.Considerable financial resources, trained personnel, and the cooperation of participants and staffs in Central and Local Government is essential over a prolonged period.
Ethical issues are important in any cohort study and planning and implementation may be fraught with difficulty.Drs Soda and Hirayama have successfully established a large scaled cohort study on cancer in Japan.A total of 265,000 inhabitants aged 40 years and over in 29 health center areas of 6 prefectures were enrolled, with supports from Ministry of Health and Welfare and local governments.The study started in 1965 and the cohort was followed up for cancer deaths for about 15 years.The employees from Health centers and local government formed the cohort population and followed-up.
The association between smoking and cancer was confirmed in a Japanese population, and furthermore a significant association of passive smoking with lung cancer was observed.Dietary habits such as intake of green-yellow vegetables, and other protective factors were confirmed to play significant roles in preventive oncology.These results not only influenced cancer control programs, but also stimulated cancer research.Dr.Hirayama's cohort consisted of people born before 1925.Their living conditions and dietary habits were those associated with poor socioeconomic conditions, especially during childhood and adolescence.
There have been, however, remarkable changes in lifestyle and living conditions for all Japanese including housing, family size, clothing, transportation, communication, education, working conditions, income, leisure time, sanitary conditions, and others, especially since 1970.Mental stresses have been increasing simultaneously.
It is recognized that the occurrence of cancer is closely related to human lifestyles and living environment.Reflecting rapid changes in the communities, cancer patterns in the birth cohorts born before and after 1925 have changed.The current situation indicates that a new cohort study to examine cancer rates in the population born after 1925 is strongly needed.From the background in Japan, we discussed and planned a multicenter-collaborative cohort study.Population based epidemiological studies have already been conducted in many small areas throughout Japan, mainly aiming to control tuberculosis and/or cerebrovascular/circulatory diseases.In this context, those who take part understand the implication of health check-ups and other health activity in the community and the majority are willing to collaborate in epidemiological surveys to enable basic information to be obtained for a control program.Employees in health in cities, towns or villages were, generally, enthusiastic to participate in medical examinations and preventive disease programs for improving health.They had often asked clinicians and epidemiologists to carry out a study on risk factors for malignant diseases.Many scientific papers coming out of these community based studies encouraged further similar endeavours in Japan.
After 2 years' of pilot surveys for the proposed cohort study, 34 epidemiologists and epidemiologically oriented clinicians in Japan willingly participated in the, main study and more than 100,000 inhabitants in 50 areas were enrolled in the cohort.The Steering Committee, Ministry of Education, Science, Sports and Culture (Monbusho) Grants-in-Aid recognized the implication of this cohort study and their support was much appreciated.Some grants have been awarded since 1988 and expert advisers in oncology were appointed, and participated in the Research Committee.
This report summarizes the study plan and the epidemiological characteristics of the cohort at entry to the study.It also gives a brief account of activities until 1994.

SUBJECTS AND METHODS
The members of the research committee formed in 1988, consisted of epidemiologists and clinicians who were interested in the objectives, and were prepared to collaborate for a minimum of 10 years.There was a further condition that each member should provide an appropriate successor to continue the study, should he/she retire or resign from the post for any by reasons.Statisticians also participate in data management and analysis.
A community leader and a representative of the population approved the study objectives.Written consent to participation in the study was obtained from each participant.A contract document between the community leader and the chairperson of the Research Committee was exchanged.Strict confidentiality of personal data was assured.

Cancer Sites Observed
Over 1 million person years of observation were expected, so that risk or protective factors for common cancers such as stomach, colon, liver and lung, might be examined using 1988 mortality data for comparison.Cancers of the breast and uterus can be analyzed by morbidity data, because about 70% areas expect to pursue cancer incidence.Urologists and allied specialists will focus on urological cancers such as bladder, kidney and prostate using mortality and morbidity data.Other sites may be included if the number of cases is sufficient.

Causal Factors to be Verified
Many factors has been suggested to plan a causal role on cancer.However, the large number in the cohort and the multiple geographical locations of the population meant that the factors investigated had to be limited.Avoiding bias between study areas, the feasibility of 10 years' observation, the cost, time, number of workers for data collection and analyses were also considered.Baseline examinations were focussed on preventive factors such as smoking, passive smoking, alcohol consumption, other drinks including teas, dietary habits [frequency of major food items and taste (like or dislike)], health conditions, healthy habits, exercise, periodical health checks, participation in screening programs on cancer and others, occupation including labour conditions, human relations, attitude to life and psychological status, marital status, number of children, reproductive history for women, and living places in childhood and adolescent.Height, weight and blood pressures were measured at the entry, previous history of disease, and family history (parents and siblings) were asked.These items seemed to be essential for confirming causative factors on cancer in Japan.
In over 50% of study areas, medical check-ups such as physical exams, blood pressure, ECG, fundscope, chest X ray, abdominal exams by ultrasonic generators, chemical laboratory examinations on blood and urine were conducted.In several areas, serum vitamins and hormonal level were analyzed.
The regions, the number of towns and the participating staff were : The baseline study included a health and lifestyles questionnaire, and some physical examinations which had been conducted in [1988][1989][1990].A total of 125,760 were enrolled in this A serum bank from about 40,000 inhabitants (5 aliquots from each person), was established at the same time obtaining the donator's consent.The serum were storaged in deep refrigerators at -80°C degree.Most samples were the remaining serum after biochemical analyses from the screening program.About two thousand sera were stored for the successive two years.These are to be used for checking changes in serum levels of the same person, with the passage of time.
Death certificates in each area are collated with the records from Ministry of Health and Welfare once a year, identifying name, age, sex and birth date.Causes of death are checked.All these data were sent to the central office of the Research Committee.
Reliable cancer registries exist in one in four of the areas, and other areas are making an effort to collect morbidity data using hospital and regional records on cancer.
Another cohort consisting of about 25,000 industrial workers was established at the same time.They are mainly inhabitants of large cities, aged between 20 and 60 years old.The results of this cohort will be used for meta-analysis to that of the main cohort.About 5,000 workers were newly entered in this cohort.

Organization of the research team
To continue the collaborative studies, the following subcommittees were established: 1. Steering committee of the cohort study 2. Subcommittee on methodology 3. Subcommittee on data collection, tabulation and analysis 4. Subcommittee on serum bank 5. Subcommittee on the interim questionnaire study 6.Subcommittee on the industrial cohort 7. Subcommittee on host factors in cancer Each subcommittee has been accomplishing the tasks and occasionally joint meetings were held.In addition, the committee of medical ethics and the Advisory committee were established.An interim questionnaire study after 5 years from the entry was planned in order to examine changes in lifestyles, physical conditions and environmental factors.This is also useful for examine the reliability of data at entry.
Ethics is important in such a long term cohort study to protect personal and regional privacy, to prevent medical or social accident and also to restrict researchers' interests to specific points.The members of the committee have repeatedly discussed relevant problems and have advised members engaged in field studies, to avoid accidental events.
The Advisory Committee consisted of five oncologists recommended by the Steering Committee of Cancer Study sponsored by Ministry of Education Science, Sports and Culture (Monbusho).The aim of the committee was to check the study plan and evaluate the data obtained.They also advised the Research Committee.

Epidemiological Characteristics of the Cohort Population
A total of 125,760 inhabitants (Males 53,266 and Females 72,494) were enrolled.Thus included 109,118 (Males 45,705,Females 63,413) aged 40 to 79 in 50 areas, who will be analyzed after 10 years follow-up.Age group distribution was 25.3% for 40-49 years, 30.4% for 50-59 years, 29.6% for 60-69 years and 14.7% for 70-79 years in males and 23.8%, 31.3%, 30.4% and 14.5% in females, respectively.In several areas, those less than 39 and over 80 years were included at the request of the community.They will be followed up and analyzed separately.
The geographical distribution of these 51 areas was not random.The proportions of the cohort population by region were as follows : 3.9% in Hokkaido,12.9% in Tohoku,15.9% in Kanto,23.9% in Chubu,16.3% in Kinki,12.6% in Chugoku and 14.5% in Kyushu as shown in Figure 1.
The proportion of registered persons from all the cities, towns and villages that participated was about 20% for males and about 25% for females.Age distribution of the cohort differed slightly from the national population.The proportion aged 55-64 years was slightly higher and that of 30-39 was slightly lower than the national mean.The proportions of the occupational status by industry (the 1st to 3rd industry) in the cohort was a similar to those of all Japan.questionnaire was : In males, 2.5% cerebrovascular diseases, 3.4% myocardial infarction, 4.4% diseases of the kidney, 8.3% diseases of the liver, 5.8% for gallbladder stone or inflammation, 8.0% for tuberculosis and 1.1% for cancer, while in females, 1.5%, 3.3%, 5.3%, 6.0%, 6.9%, 5.4% and 2.1%, respectively.The prevalence rate of hypertension was 21.5% in males and 23.5% in females.Diabetes mellitus accounted for 7.7% in males and 5.2% in females.Those who had experienced abdominal surgical operations was unexpectedly high at 3 1.0% in males and 40.6% in females.A past history of blood transfusion was given by 12.6% of males and 13.7% of females.Past histories of measles, whooping cough, mumps, poliomyelitis, and encephalitis were given for 58.3%, 22.9%, 2.3%, and 2.1% males, and 64.4%, 7.9%, 29.0%, 2.4% and 2.3% females respectively.Typhoid fever was 3.5% males and 3.4% females.Compliance rates for screening programs of circulatory diseases, pulmonary diseases and stomach cancer were 42.7%, 43.2% and 27.5% males and 41.8%, 46.4% and 32.2% females respectively.These rates were higher than those in urban areas in Japan Industrial Worker Cohort 25,336 workers (males 18,078 and females 7,258) in three enterprises in Nagoya, Akita and Tokyo and one labour union composed of smaller scaled territory industries were enrolled in the industrial cohort.The age distribution was 32.1% for 20-39 years, 34.5% for 40-49, 28% for 50-59 and 5.3% for 60 years and over.Most of the participants lived in large cities and were employed in clerical grade.Dietary habits are slightly different from those in rural areas.The outcome of this study will be compared with the main cohort after the 10 years' Activities at Present and in the Future 1.To continue to accumulate deaths within the cohort in each area.To obtain other appropriate information as needed.
2. To examine drop-outs who move from the area, and for other reasons.
3. To determine disease specific mortality or morbidity rates for comparison with general population.4. To identify causative factors for cancer after 10 years of follow-up.No publication is currently expected.
5. Some members of the Research Committee will begin examining the quality of data obtained in the cohort from each area.Causal hypotheses for site specific cancer will be investigated.The results will be useful for the final analysis.Some groups are interested in non malignant diseases.

DISCUSSION
The study areas were not randomly selected because of previously stated reasons.The proportion of participants in each age group in a community was not the same in each area.
However, the study areas covered the whole country and the age distribution did not differ significantly from the general population.In each five year groups from 40-44 to 70-74 years by sex, 10,000 to 20,000 persons are being followed-up.
Most of study areas are in rural, but living circumstances and economic level are not low, when compared with the national average.Each area has good health promotion programs.For example, the compliance rate for periodical health checks and screening for diseases of the circulation or cancer are higher than the average for Japan, although medical facilities and large hospitals less numerous when compared with major cities.The health insurance system covers all the areas and the facilities for care of the olderly are better than in urban areas.
Labour conditions have improved over the last two decades with the use of modern machinery in agriculture and forestry, although the proportion of farmer and other physical labours was higher than the national average.Housing and transportation have also improved in these areas since 1970, and lifestyle habits currently do not differ when compared with those of urban areas.
Physiological measurements and laboratory tests of the participants in many areas showed no significant differences in urban areas, and it is pertinent that the daily intake frequency of foods and nutrients in this cohort was similar to those in the Governmental Nationwide Nutritional Survey.
Mortality data of this cohort until 1993, where about 5,000 deaths were reported, showed a slightly lower death rate than the expected and similar cancer pattern by site was observed compared with that of all Japan.
More than 10,000 deaths are estimated over the 10 years, with 30-35% due to cancer.These figures suggest that a efficient number of cancer cases will be obtained after 10 years observation..The migration rate from the cohort areas was estimated at about 10% after 10 years' observation.These figures indicate that this cohort study can assess causative factors of common cancers, even though the size of the cohort was smaller than that of Dr.Hirayamas'.Several advantages of this study are apparent.There are several variables from the questionnaire, which are distinct from those assessed by Dr. Hirayamas'.The interim questionnaire study in 1993-1995 will reveal changes in lifestyle habits over the 5 years and the reliability of the answers can be checked.
Each study area is periodically surveyed by the same epidemiologists, who have been studying the demographic and epidemiological features of the cohort and evaluating baseline data according to outcome.About 40% of the participants have had biological data at entry and frozen sera are being started.These biological data can be used for endorsing the causal relation of the factors.
Several scientific reports emerged from this population studies have already been published since 1990.
The ethics committee has regulated and partly revised the conditions of the study taking into account changing circumstances with time.The committee will reassess the study Hokkaido region : Three towns (Drs.Fukuda/Sugimura, Dr.Miyake and Drs.Sasaki/Itoh) Tohoku region : 5 towns (Drs.Kamiyama/Koizumi, Drs.Takizawa/Hachiya, Dr.Hisamichi) (Dr.Yoshimura, Dr. Tokudome) There was no participative centre in Shikoku.(figure 1) A long term-study program was shown in the figure 2.

Figure 1 .
Figure 1.Geographical distribution of study fields

Table 1 .
methods and approaches to the study by the members since 1990.One of the reports will be published as the first formal report from this subcommittee.It should be mentioned that no problems were any experienced for the last 5 years.The Advisory Committee met at least once a year and advice and recommendations were made to the Research Committee.