Smoking and Drinking Habits Five Years after Baseline in the JACC Study

BACKGROUND: Observing longitudinal changes in smoking and drinking habits is important for evaluating the risk of incidence and death from cancer or other diseases in a cohort study. METHODS: Smoking and drinking habits at baseline and about five years later among 18,312 males and 28,338 females were obtained from the baseline and interim surveys in the Japan Collaborative Cohort Study for Evaluation of Cancer Risk (JACC Study) sponsored by Monbusho (Ministry of Education, Science, Sports and Culture of Japan). Changes in smoking and drinking habits between the two surveys were observed. Odds ratios for quitting these habits at interim survey were estimated. RESULTS: Percentages of current smokers at baseline and interim surveys were 51.0% and 45.5% in males, and 5.2% and 4.8% in females, respectively. Percentages of current drinkers at baseline and interim surveys were 78.0% and 73.2% in males, and 29.5% and 23.5% in females, respectively. The number of cigarettes per day among male current smokers and the usual amount of alcohol consumed on each occasion among current drinkers decreased between the two surveys. Odds ratios for smoking cessation increased with age at baseline and decreased with the number of cigarettes per day at baseline. Odds ratios for drinking cessation increased with age at baseline and decreased with the usual amount of alcohol consumed on each occasion at baseline. CONCLUSION: The decrease in smoking and drinking habits was observed during the five-year follow-up period. Higher age and lower levels of exposure were associated with quitting smoking or drinking.

[7][8][9][10][11][12] However, longitudinal changes in smoking and drinking habits and their related factors have also been investigated, [13][14][15][16][17][18][19] but such few studies have been conducted among Japanese. 20f exposure data in cohort studies are only measured at baseline even though exposure histories change significantly after baseline, the estimates of relative risk may be biased due to insufficient information regarding exposure. 21To evaluate the relative risk of smoking and drinking habits in such cohort studies, it is important to clarify the extent of such changes during the followup.
The Japan Collaborative Cohort Study for Evaluation of Cancer Risk (JACC Study) sponsored by Monbusho (Ministry of Education, Science, Sports and Culture of Japan) 22 is a large-scale cohort study surveying the associations between cancer risk and lifestyle and living conditions.An interim survey covering half of the cohort subjects was conducted about five years after the baseline survey.
Among subjects in the JACC Study, we observed the changes in smoking and drinking habits from the baseline survey to the interim survey, and analyzed characteristics of smokers and

S-57
Subjects for our analysis of smoking habits reported here were restricted to 16,778 males and 21,161 females who answered a question regarding smoking status in both the baseline and interim survey questionnaires.Similarly, subjects for the analysis of drinking habits were restricted to 16,567 males and 22,303 females who responded to a question about their drinking status.Table 1 shows the numbers of these subjects by sex and age at the baseline survey.

Smoking and drinking habits
The self-administered questionnaires of the baseline and interim surveys included the same questions about smoking and drinking habits.The questions about smoking habits were with regard to smoking status ("current smoker," "former smoker" and "never smoked"), age when starting smoking, and the number of cigarettes per day.The questions regarding drinking habit concerned drinking status ("current drinker," "former drinker" and "never drank"), age when starting drinking, frequency of drinking per week ("less than 1 day," "1-2 days," "3-4 days" and "everyday"), and the usual amount of alcohol consumed on each occasion.The drinkers at the baseline survey associated with smoking and drinking cessation during the two surveys.

Subjects
The baseline survey of the JACC Study was conducted in 45 areas from 1988 to 1990, and 46,465 males and 64,327 females aged 40 to 79 years completed a baseline questionnaire.The interim survey was conducted in 31 areas (it was asked to every participant in 18 areas and asked to some of the participants in 13 areas) from 1993 to 1995 using the questionnaire of the baseline survey with some modifications, while not conducted in 14 areas.Total 46,650 participants (18,312 males and 28,338 females) completed the questionnaire.For 18 areas in which the interim survey was conducted in every participant of baseline survey, 37,853 of 48,016 baseline participants were responded, and the response rate was 78.8%.For 13 areas in which the survey was conducted in part of the participants, 8,797 of 36,460 participants were responded, and the response rate was 24.1%.The mean interval between the baseline and interim surveys was 5.0 (standard deviation = 0.9) years.

Changes in smoking habits
Figure 1 shows the percentages of current smokers at the baseline and interim surveys by sex and age (at the baseline survey) group.
In males, those percentages were 51.0% at the baseline survey and 45.5% at the interim survey.In females, the percentages of current smokers were 5.2% in the baseline survey against 4.8% at the interim survey.The percentages in the interim survey were lower than those in the baseline survey in every sex and age group except for the female 40 to 44-years-old group (increased 5.9% to 6.2%).
Table 2 shows changes in smoking status between the two surveys.In males, the percentages of current smokers at the interim survey among current smokers, former smokers and never smokers at the baseline survey were 83.9%, 7.6% and 2.8%, respectively, against 78.3%, 12.8% and 0.5%, respectively, in females.
Table 3 shows changes in the number of cigarettes per day among current smokers at the baseline and interim surveys who responded to the question.In males, the percentage of those smoking 1-19 cigarettes per day at the interim survey was higher than at the baseline survey, whereas that of those smoking 30 or more per day at the interim survey was lower than at the baseline survey.In females, however, the percentages of those smoking 1-19 cigarettes per day at the interim survey was slightly lower than at the baseline survey, whereas that of those smoking 30 or more per day at the interim survey was slightly higher than at the baseline survey.unit of alcohol consumption per occasion was "gou" (Japanese standard unit for an alcoholic beverage), which can be converted to 23 g units of ethanol.

Statistical analysis
Changes were examined in smoking status, the number of cigarettes per day, drinking status, drinking frequency, and alcohol consumption per occasion between the baseline and interim surveys.Smoking status at the interim survey was re-categorized into "current smoker" ("current smoker") and "nonsmoker" ("never smoked" or "former smoker").Similarly, alcohol drinking status at interim survey was re-categorized into "current drinker" and "nondrinker" ("never drunk" or "former drinker").The number of cigarettes per day was categorized as 1-9, 10-19, 20-29, 30-39 or 40 and over.Alcohol consumption per each occasion (gou) was categorized as 0.1-0.9,1.0-1.9,2.0-2.9 or 3.0 and over.
Smoking cessation percentages at the interim survey among current smokers from the baseline survey were observed by using each baseline characteristic; i.e., age at the baseline survey, age when starting smoking and the number of cigarettes per day.Logistic regression analysis was used to estimate the odds ratios for smoking cessation at interim survey among current smokers at baseline using these baseline characteristics as independent variables (as dummy variables).
Drinking cessation percentages at the interim surveys among current drinkers from the baseline survey were observed by using each baseline characteristic; i.e., age at the baseline survey, age when starting drinking, drinking frequency and alcohol consumption per occasion.Logistic regression analysis was used to estimate the odds ratios for drinking cessation at interim survey among current smokers at baseline using these baseline characteristics as independent variables (as dummy variables).
All analyses were conducted using SAS ® software, version 8.2 (SAS Institute, Inc., Cary, NC, USA).Table 5 shows changes in drinking frequency between the baseline and interim surveys among current drinkers at both surveys who responded to this question.The distribution of drinking frequency at the interim survey was nearly equal to that at the baseline survey.
Table 6 shows changes in the usual amount of alcohol consumed on each occasion between the baseline and interim surveys among current drinkers at both surveys who responded to this question.The percentage of those consuming 0.1-0.9gou per occasion at the interim survey was higher than at the baseline survey in both males and females.

Changes in drinking habits
Figure 2 shows the percentages of current drinkers at the baseline and interim surveys by sex and age (at the baseline survey) group.In males, the percentage of current drinkers was 78.0% at the baseline survey and 73.2% at the interim survey.In females, the corresponding percentages were 29.5% and 23.5%, respectively.The percentage of current drinkers at the interim survey was lower than at the baseline survey in every age group.
Table 4 shows changes in drinking status between the baseline and interim surveys.In males, the percentages of drinkers at the interim survey among current drinkers, former drinkers and never drinkers at the baseline survey were 88.7%, 38.6% and 11.8%, respectively, against 59.4%, 35.3% and 7.8%, respectively, in females.interim survey by the characteristics of current smokers at the baseline survey.In males, the odds ratio compared with 40 to 44years-old at baseline survey was 1.0-3.3,which increased with age at the baseline survey.The odds ratio compared with 20 to 24-years-old at starting smoking was 1.0-1.7 in the other groups of age when starting smoking.The odds ratio decreased with the number of cigarettes per day.In females, the corresponding odds ratio compared with those 40 to 44-years-old was 4.8 in the 75 to

Characteristics of smokers associated with smoking cessation
Table 7 shows the percentages of smoking cessation at the interim survey among current smokers at the baseline survey by age at the baseline survey, age when starting smoking, and the number of cigarettes per day.The percentages of those who quit increased with age at the baseline survey and age when starting smoking, but decreased with the number of cigarettes per day.
Table 8 shows the odds ratios for smoking cessation at the  10 shows the odds ratios for drinking cessation between the baseline and interim surveys by the characteristics of current drinkers at the baseline survey.In males, the odds ratio compared with 40 to 44-years-old at baseline survey was 0.9 in the 45 to 49years-old age group and 1.6-3.3 in the other age group, which increased with age at the baseline survey.That ratio compared with 20 to 24-years-old when starting drinking was 0.8-1.0 in the other groups of age when starting drinking.The odds ratio decreased with drinking frequency and usual amount of alcohol consumed on each occasion.In females, the odds ratio compared with 40 to 44-year olds was 1.0-2.5 in the other age groups at the baseline survey.The odds ratio compared with those 20 to 24years-old when starting drinking was 0.9-1.2 in the other age groups when starting drinking, and it decreased with drinking frequency and the usual amount of alcohol consumed on each occasion.
79-years-old group at baseline, and 1.1-2.2 in the other age groups.The odds ratio compared with 20 to 24-years-old when starting smoking was 0.9-1.7 in the other groups of age when starting smoking.The odds ratio compared with those smoking 1-9 cigarettes per day was less than 0.4 in those smoking 20-29 and those smoking 40 cigarettes and over per day.

Characteristics of drinkers associated with drinking cessation
Table 9 shows the percentages of drinking cessation at the interim survey among current drinkers at the baseline survey by age at the baseline survey, age when starting drinking, drinking frequency at the baseline survey, and the usual amount of alcohol consumed on each occasion at the baseline survey.The drinking cessation percentages increased with age at the baseline survey in males, but showed no major change with age at the baseline survey in females.It also showed no major change with age when starting drinking, and decreased with drinking frequency and the usual amount of alcohol consumed on each occasion at the baseline survey.

DISCUSSION
The percentages of current smokers decreased more at the interim than at the baseline survey.As previous cross-sectional and longitudinal studies have reported, the decrease is associated with aging of the population. 13,14A similar decrease in smoking habits due to aging might occur in the whole population of the JACC study during the follow-up period.Higher age and fewer cigarettes per day at the baseline survey were associated with smoking cessation during about five years of the follow-up period.A positive correlation between increasing age and smoking cessation has been discussed in other studies. 9,10The higher incidence of smoking-related diseases in older age might persuade some smokers to adapt a healthy lifestyle, thus making it easier for them to quit.A trend toward an increased number of cigarettes per day is one of the criteria for nicotine dependence. 24Because of such a dependence, quitting smoking may be more difficult for heavy users who smoke more cigarettes per day. 24This relationship is also consistent with that in other studies. 13,15he percentage of current drinkers was lower at the interim survey than at baseline.Several longitudinal studies have shown a similar decline in alcohol intake with aging, 16,17 a decline which may well have occurred in the whole population of the JACC Study.Age at baseline, frequency of drinking, and usual amount of alcohol consumed on each occasion were associated with drinking cessation.In many cross-sectional and some longitudinal studies, lower levels of alcohol consumption (including quitting drinking entirely), were observed among older subjects. 16eteriorating health may also be related to this trend among the elderly. 18A negative correlation between the level of consumption and drinking cessation during the follow-up period has been reported in many studies. 16t has been well-known that nondifferential misclassification caused underestimates of relative risk.The changes in smoking and drinking habits during the follow-up period in a cohort study might affect their estimates of the relative risk of incidence of disease and death from cancer or other diseases if only the exposure data at baseline were used. 19,20However, the risk of cancer or other chronic diseases would be affected by long-term exposure to smoking or drinking, and this effect might not be instantly affected by the changes in these habits during the follow-up period.In addition, differences in the percentages of current smokers or drinkers in the whole population in the baseline and interim surveys were less than 5%.Therefore, estimates of the relative risk of smoking and drinking based only on baseline exposure data in this population would not have much of an effect on the study conclusions, despite a possible bias toward underestimation due to the lack of exposure information after baseline.

MEMBER LIST OF THE 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.

Figure 1 .
Figure 1.Sex-and age-specific percentages of current smokers at baseline and interim surveys.

Figure 2 .
Figure 2. Sex-and age-specific percentages of current drinkers at baseline and interim surveys.

Table 1 .
Number of subjects by sex and age at baseline survey.

Table 2 .
Changes in smoking status between baseline and interim surveys.

Table 3 .
Changes in number of cigarettes per day between baseline and interim surveys among current smokers at both surveys.

Table 4 .
Changes in drinking status between baseline and interim surveys.

Table 5 .
Changes in drinking frequency per week between baseline and interim surveys among current drinkers at both surveys.

Table 6 .
Changes in usual amount of alcohol consumed on each occasion between baseline and interim surveys among current drinkers at both surveys.

Table 7 .
Smoking cessation percentages at interim survey among current smokers at baseline survey by characteristics of current smokers at baseline survey.

Table 8 .
Odds ratios for smoking cessation at interim survey among current smokers at baseline survery by characteristics of current smokers at baseline survey.

Table 9 .
Drinking cessation percentages at interim survey among current drinkers at baseline survey by characteristics of current drinkers at baseline suvey.

Table 10 .
Odds ratios for drinking cessation at interim survey among current drinkers at baseline survery by characteristics of current drinkers at baseline survey.