Cohort Study for Evaluation of Cancer Risk Sponsored by Monbusho ( JACC Study )

This paper primarily aimed to overview the rationale for initiating the Japan Collaborative Cohort Study for Evaluation of Cancer Risk Sponsored by Monbusho (Ministry of Education, Science, Sports and Culture of Japan) (JACC Study), by comparing socio-demographic and nutritional changes that were witnessed between 1965 and 1990 in Japan, and also to describe the study design, the follow-up conditions as of the end of 1997, and the frameworks for analyzing the data of the lung, stomach, pancreas and gallbladder/bile duct, based on the approximately 8-year follow-up data. For other major sites such as cancers of the large intestine and liver, an analysis will be started in the fiscal year of 2002. This paper secondarily aimed to be cited as the basic information on the JACC Study when several publications are to be based on. J Epidemiol, 2001 ; 11 : 144-150

145 Japan Collaborative Cohort Study on Cancer Smoking rate 4) among adults, defined as those aged 20 years or over, was 82.3% and 15.7% in men and women in 1965, respectively, but 60.5% and 14.3% in 1990, correspondingly.
Together with the observed decrease in male smoking rate, a notable change in smoking habit was that smoking rate had decreased in all age groups in men, but increased in the age group of 20-29 years, in particular, in women.The number of cigarettes smoked per day was 19.4 in men and 12.2 in women in 1965, but 24.7 and 18.1 in 1990, correspondingly.In 1965, cigarettes with and without filter-tip were sold half and half in proportion 4), but almost no cigarettes without filter-tip were sold in 1990, and tar yield per cigarette was remarkably reduced5) likewise in other developed countries, though with much increased number of cigarettes smoked a day.Regular drinkers including occasional ones 6) were estimated to be 43% in 1965 and 67% in 1990 among adults.Annual amount of pure alcohol consumed by one adult was 5.8 liters in 1965 and 8.3 liters in 1990.Heavy drinkers defined as those consuming 150 ml of pure alcohol per day or more were 1,028 thousands in number in 1965 and 2,124 thousands in 1990.In Japan, beer consumption has remarkably increased year by year, but such other alcoholic beverages as sake (Japanese fermented alcoholic beverage made from rice), Schochyu (distilled alcoholic beverage made from rice, wheat, sweet potatoes and others), whiskey, and spirits have been constantly consumed for the recent three decades 6) Demographic and social changes 7) between 1965 and 1990 were also dramatic in Japan.Total population had increased from 98 to 124 million in the 2.5 decades with almost doubled proportion of those aged 65 years old and more (from 6.3% to 12.1%).Life expectancy at birth increased by 8.2 years (from 67.74 to 75.92 years) in men and by 9.0 years (72.92 to 81.90 years) in women.Changed proportion in industrial structure was rather substantial: clearly decreased (from 24.1 to 7.1%) primary industry (agriculture, forestry, fishery and livestock farming) and apparently increased (from 43.Dietary and nutritional intakes also substantially changed between 1965 8) and 1990 9) in Japan.A notable increase in daily dietary consumption was observed in wheat, potatoes, oils and fats, green-yellow vegetables, fruits, fishes and shellfishes, meats, eggs, milk and dairy products: in particular, more than doubled consumption in fruits (from 58.8 to 124.8g), meats (from 29.5 to 71.2g) and milk (from 48.8 to 119.5g), and 1.2 to 1.7-fold increase in other food groups mentioned above.In contrast, rice was far less consumed (from 349.8 to 197.9g), but with unchanged consumption of pulses and seaweed.In nutrient intakes, approximately doubled intake was observed in animal fat (from 14.3 to 27.5g) and vitamin A (1,324 to 2,5671U) with a 1.5-folds increase in intake of animal protein (from 28.5 to 41.4g), vitamins B2 (from 0.83 to 1.33mg) and vitamin C (from 78 to 120mg).
These demographic, social and nutritional changes with apparently altered smoking and drinking behaviors in the past 2.5 decades were believed to be an important and rational background for initiating another large-scale population-based cohort study such as the JACC Study in Japan, in which associations of dramatically changed lifestyle and living conditions with cancer risk would be devotedly examined.

STUDY DESIGN AND SUBJECTS
The two major founders of the JACC Study were Kunio Aoki, Professor of Preventive Medicine, Nagoya University School of Medicine and Haruo Sugano, Director of Cancer Institute, Tokyo (as of 1988), who financially supported to organize the Monbusho ECC 3,10,11) The Monbusho ECC was organized by Prof. Aoki, after the feasibility study of the two preceding years, in 1988, involving 35 epidemiologists, including epidemiology-oriented clinicians and one statistician who were interested in initiating a multicenter-collaborative cohort study on cancer, with the two essential requests of collaborating for a minimum of 10 years and providing an appropriate successor to continue the JACC Study, should the original investigator retire or resign from his position 10).The original investigators belonged to 24 different institutions (universities, hospitals or others) and had been coordinating population screening programs for chronic diseases in 45 areas throughout Japan: 3 towns in Hokkaido district, 5 towns in Tohoku district, 8 towns and 2 villages in Kanto district, one city and one town in Chubu district, 7 towns and 2 villages in Kinki district, one city and one town in Chugoku district, 4 cities, 9 towns and one village in Kyushu district, and no area in Shikoku district.
The Monbusho ECC 3,10,11) enrolled about 127,500 healthy inhabitants in the areas, who attended to the screening program and responded to the questionnaire in 1988-1990, as a basic cohort population, among which 46,465 men and 64,327 women (110,792 in total) were aged 40-79 years old and were to be followed up for a minimum of 10 years.Exact response rate to the screening program was not uniformly reported from all areas to the central office of the Monbusho ECC because of methodological difficulties.But it was believed to be considerably high, since the screening program has annually been conducted for long years in each area together with enthusiastic supports for the program from the inhabitants themselves and the local community authority (among the areas where the investigators enrolled the inhabitants from their total population, the response rates were reported as high as 90%).
Table 1 shows age distribution of the cohort members aged 40-79 years at enrollment and their follow-up condition as of the end of 1997 (mean follow-up period: 8.15 years).
Cohort members who deceased have been identified with an underlying cause of death by reviewing all death certificates in each area once a year under the authoritative permission from the Director-General of the Prime Minister's Office (Ministry of Public Management, Home Affairs, Post and Telecommunications) .This verification was worked by the investigator himself with administrative assistance from public health nurses in each area, who has well acquainted themselves with inhabitants' vital status including out-migration and their health conditions including screening findings as well.The underlying causes of death already coded according to the International Classification of Diseases and Injuries (ICD) 9th version (from the baseline to 1994) and stored in the computer database were re-coded in 1999 according to the ICD 10th version (after 1995), using a specifically developed computer program 12) for converting the ICD 9th code to the 10th code.The move-outs were also annually verified by the investigator himself with administrative assistance from public health nurses in each area by reviewing population-register sheets of the cohort members.The verification of vital status and out-migration were believed to be substantially accurate because of firmly established population registration system in Japan, but the losts to follow-up, who were believed to be negligible in proportion, could not be methodologically identified and were regarded as alive in each area.The most likely lost to followup would occur when a subject moved out leaving his/her population registration in the survey area.In this case, however, when he/she deceased in the area out-migrated, his/her death certificate would be transferred to the health center in which his/her original area was located, and, therefore, his/her vital status would be readily known, though with unknown an underlying cause of death.We believed that such occasions were quite rare.
The data of deaths and move-outs in all areas were annually informed, together with incident cancer sufferers in 24 areas out of the 45 areas who were identified mostly by cancer registration system, to the central office of Monbusho ECC (Department of Preventive Medicine/Biostatistics and Medical Decision Making, Nagoya University Graduate School of Medicine), and were confidentially kept as a computer database without any information which could identify an individual.
Epidemiological data 11) were routinely collected at baseline

EPIDEMIOLOGICAL FEATURES OF COHORT POPULATION IN JACC STUDY
The detailed tabulation tables based on the self-administered questionnaires were already shown in the progress report 11) published in March, 1996.In this section, several epidemiological features would be summarized by re-tabulating the baseline data cleaned as of the end of 1997.
In our cohort population aged 40-79 years old at baseline (when a self-administered questionnaire was filled in), current, ex-and non-smokers accounted for 53,0%, 26.7% and 20.4% in men, and 5.7%, 1.8% and 92.6% in women, respectively.Male ex-smokers were much higher in proportion, compared to the figure (3.8%) in Hirayama's male cohort population 1), but with similar rate to the national figure in 1990 4).The proportions of current smokers by age were 60.1%, 54.1%, 51.6% and 40.5% in the 5th, 6th, 7th and 8th decade in men, respectively, and 6.9%, 5.3%, 4.8% and 5.8% in women, correspondingly.Age-specific smoking rates were similar in men, but lower than the national figures in women in 1990 4).
Current, ex-and non-drinkers were, in proportion, 74.7%, 6.6% and 18.8% in men, and 24.5%, 1.8% and 73.7% in women, respectively, in our cohort population.Age-specific proportions of current drinkers were 82.1%, 78.4%, 71.3% and 60.3% in the 5th, 6th, 7th and 8th decade in men, respectively.Current drinkers in our male cohort population were higher in proportion, compared to the figures in general population 9), though the definition of current drinkers was a little different from ours.
Those with no habitual exercise accounted for 68.6% in men and 76.1% in women in our cohort population: being considerably lower than the national ones 9) .Those who did not take breakfast accounted for 3.4% in men and 2.5% in women: being lower than the national figures 9) .As for marital status, married subjects accounted for 93.5%, widowed or divorced for 4.8% and not-married (single) for 1.7% in men with corresponding figures of 82.4%, 16.0% and 1.6% in women.These figures were well comparable to those aged 40-79 years old in general population 13).
In short, our cohort population was not established by random sampling, but appeared to be similar to Japanese general population in the light of several demographic and lifestyle features, though such similarity is not necessarily the prerequisite condition, since a cohort study itself aims, in its nature, an internal comparison between those developed and those not developed the targeted outcomes in a certain period of time.

FOLLOW-UP CONDITIONS AS OF THE END OF 1997
The central office of Monbusho ECC examined the followup conditions as of the end of 1997 (mean follow-up period: 8.07 years in men and 8.21 years in women).
As shown in Table 1, among 46,465 men aged 40-79 years old at baseline (when a self-administered questionnaire was filled in), 39,729 (85.5%), 5,472 (11.8%) and 1,264 (2.7%) were identified as of the end of 1997 as the alive, deceased and move-outs from the study areas, respectively, and corresponding figures were 58,696 (91.2%), 3,653 (5.7%) and 1,978 (3.1%) among 64,327 women: 98,425 (88.8%), 9,125 (8.2%) and 3,242 (2.9%) in the total cohort population in due order.The expected numbers calculated based on the Japanese population aged 40-79 years old in 1997 (as a standard) were 6468.1 in men and 4601.9 in women for total death, with the corresponding numbers of 2516.8 and 1618.3 for cancer death: indicating a ratio (O/E) of 0.846 and 0.794 for total death and 0.852 and 0.811 for cancer death in due order.Our cohort members, therefore, appeared to be less likely to die from total causes and cancer, compared to the Japanese population aged 40-79 years old.As mentioned above, those incidentally missed to be followed were included in the alive in a negligible proportion.
Total cancer deaths accounted for 39.2% (2,145) and 35.9% (1,313) in male and female total deaths (5,472 and 3,653), with 37.9% (3,458) in total deaths (9,125)  pepsinogen I and II, Helicobactor Pylori antibody, total fatty acid fractions and such metals as cadmium, chromium, and selenium for that on cancer of the gallbladder and bile duct.For a nested case-control study on cancer of the large intestine, which will be started in 2002, insulin, fructosamine, albumin, HDL-cholesterol, folic acid, carotenoids, and total fatty acid and bile acid fractions are tentatively decided as biological markers to be measured.Possible studies that will include DNA are not yet planned because of ethical issues to be resolved.
The site-specific study design on cancers of the lung, stomach and liver, including a nested case-control study design, which would use both epidemiological and biological materials (serum only), singularly or collectively, was approved by the ethical board at the university to which each principal investigator is affiliated.

MEMBERS OF JACC STUDY GROUP
The present investigators involved, with the co-authorship of this paper, in the JACC Study and their affiliations are as follows: Dr. Yoshiyuki Ohno (the present chairman of the study

Table 1 .
Age distribution of cohort members at entry and deaths/move-outs as of the end of 1997.

Table 2 .
Age distribution of total deaths and all and site-specific cancer deaths as of the end of 1997.