A Comparison of Correlates of Cigarette Smoking Behavior Between Jiangxi Province , China and Japanese High School Students

We conducted surveys on cigarette smoking among junior and senior high school students in Jiangxi province, China and throughout Japan using the same anonymous, self-administered questionnaire in order to compare correlates of adolescent smoking between the two areas. Cross-sectional surveys were used to measure smoking behavior and correlates in two samples of 57,566 Japanese students and 11,836 Jiangxi students. The correlate on smoking with the highest relative risk was friend's smoking in both sexes in each area. The magnitude of the relative risk was bigger for Japanese students. The relative risk of the variable that a student doesn't think cigarette smoking harms his/her health was higher among Jiangxi students than among Japanese students. Mother's smoking and sister's smoking were significantly related to smoking experiment of Japanese students. In Japan, important measures are to support students getting coping techniques against peer pressure and to elevate concern toward adolescent smoking among family members and society. In Jiangxi, the anti-smoking education to teach students to correctly recognize the harm of smoking to their health is more important. J Epidemiol, 1999 ; 9 : 254-260.


MATERIALS AND METHODS
Jiangxi province, China, is located the south area of the middle of the Yangtze river, China.The population was about 38 million in 1990.Subject schools of Jiangxi survey were sampled from junior and senior high schools by a stratified cluster sampling.Strata of junior high schools were a city region and a country region, and those of senior high schools were six, namely so-called stress, ordinary, and vocational school in both city and country regions.(Stress schools' curricula emphasize the study skills.)Ten junior high schools out of 2,285 schools, and 15 senior high schools out of 844 schools were sampled throughout the province.All sampled .schools responded to this survey.Enrolled students at these schools were asked to fill out an anonymous questionnaire in their classroom.The questionnaire contained questions about their smoking status, their reason for trying smoking , their average cigarette consumption per day, the smoking status of family and friend, whether they enjoyed school life, and whether they thought smoking harmful to their health.This survey was carried out from October to November in 1991.Of the 11,836 questionnaires returned, we excluded 222 for incompleteness and 9 for inconsistency.The remaining 11,605 were subjected to analysis.This number corresponds to 0.5% of total number of high school students in Jiangxi province.Further details have been published elsewhere 5) In the Japanese survey, one hundred junior high schools out of 11,182 schools and 50 senior high schools out of 3,995 schools were sampled from the national school directory using a single-stage cluster sampling.Seventy percent of the junior high schools and 66% of the senior high schools responded.All enrolled students asked to fill out the same questionnaire as the Jiangxi survey.This survey was conducted from December in 1990 to January in 1991.Of the 57,566 questionnaires returned, we excluded 183 for incompleteness and 194 for inconsistency.The remaining 57,189 were subjected to analysis.This number corresponds to 0.5% of total number of high school students in Japan.Further details of this survey have been published elsewhere ED.
The terms, 'experimenter, 'current smoker','regular smoker', and 'never smoker* as used in this paper are defined as follows.Experimenters are students who had smoked at least once.Current smokers are students who had smoked at least once during the previous month.In Japan, monthly smoking is usually used as the definition of current smoking among adolescents.Regular smokers are students who smoked three or more days a week.
According to the formula for calculation of the mean and variance under each sampling method, the point estimation and 95% confidence interval of smoking rate were computed 7).
Nine variables; i.e., 1) school grade, 2) father's smoking, 3) mother's smoking, 4) brother's smoking, 5) sister's smoking, 6) other family member's smoking, 7) friend's smoking, 8) whether a student enjoys school life, and 9) whether a student thinks cigarette smoking harms one's health, were applied for analysis to determine correlates of smoking behavior.This analysis was performed by sex and school level, because some studies indicated that correlates of adolescent smoking may be different among sexes, and between junior and senior high school [8][9][10][11] Multiple logistic regression analysis with a stepwise method was applied as a statistical method.This analysis was performed using a SAS system on HITACM-680 of Tokyo University computer center.Results for Japanese students were

RESULTS
Comparison of smoking prevalence between Jiangxi and Japanese students Table I shows smoking prevalences of junior and senior high school students in Jiangxi province and in Japan.Experimenters' rate, current smokers' rate and regular smokers' rate increased in each grade in both areas, except for the experimenters* rate among boys in Jiangxi.Experimenters* rates for Jiangxi boys were higher in each grade than those of Japanese boys.Interestingly, the rate differences were bigger among junior high school boys than those among senior high school boys.Current smokers' rates were similar in both areas, whereas the regular smokers' rates were higher for Japanese boys.The rate differences were much bigger for senior high school boys.
Smoking prevalence among girls in both areas were much lower than those among boys.Experimenters* rate, current smokers' rate and regular smokers' rate were all lower for Jiangxi girls than those of Japanese girls.Only few regular smokers among Jiangxi girls were detected by this survey.

Correlates of smoking experience among Jiangxi and Japanese junior high school students
Results of logistic regression analyses of explanatory variables on smoking experience for Jiangxi and Japanese junior high school students are shown in Table 2.The results show that a student who has a left situation of variable in Table 2 is more likely to be an experimenter, because all estimated relative risks were more than 1.0.Friend's smoking status was strongly related to smoking experience among both sexes in both areas.Relative risks of friend's smoking among Japanese students were higher than those among Jiangxi students.On the other hand, relative risks of thinking cigarette smoking doesn't harm one's health were higher for Jiangxi students than those of Japanese students.Mother's smoking and sister's smoking in both sexes among Japanese students were significantly related to smoking experience.The same was not observed among Jiangxi students.
Sister's smoking was significantly related to smoking experience among Jiangxi girls compared with Jiangxi boys.In Japan, the relative risks of friend's smoking and sister's smoking were higher among girls than those among boys.Relative risk of friend's smoking was higher for girls than for boys.

Correlates of smoking experience among Jiangxi and Japanese senior high school students
Results of logistic regression analyses on smoking experience for Jiangxi and Japanese senior high school students are shown in Table 3. Friend's smoking was also strongly related to smoking experience among both sexes and areas.The relative risks were higher among Japanese students than those Smoking  Friends smoke, Mother smokes, Father smokes, Sister smokes, Brother smokes, Other family smokes 0 = nonsmoker, He/she doesn't know., no answer, I = smoker School grade late in both areas.Relative risks among Japanese students were also higher than those among Jiangxi students.Thinking cigarette smoking doesn't harm one's health was more strongly related to current smoking in Jiangxi boys.Brother's smoking was significantly related to current smoking in both areas.Mother's smoking and sister's smoking were significantly related to current smoking among Japanese students (Table 4).

DISCUSSION
Since we conducted surveys on adolescent smoking in the two areas using the same questionnaire and analytic procedures, the merit of this study is the comparable results on correlates of adolescent smoking.Therefore, the magnitude of rel-ative risks can be compared between the two areas.Smoking prevalence among Jiangxi girls was very low, reflecting the low smoking rate of adult females in China 1).Experimenters* rate of junior high school students in Jiangxi is higher than that of Harbin city of northern China, while that of senior high school students in Jiangxi is similar to that of Harbin.On the other hand, current smokers' rate of junior high school students in Jiangxi is lower than that in Harbin, even though that of senior high students in 2nd and 3rd grades in Jiangxi is higher than in Harbin 16).Regular smokers' rate for Jiangxi students is slightly lower than that of students in Beijing, capitol city of China4).Smoking rate of adults in Jiangxi province belongs to the lower group among China 15), whereas smoking prevalence among Jiangxi students was not at all low compared to smoking prevalence in other areas of China.
Many cross-sectional and prospective studies show that friend's smoking is one of most important correlates of adolescent smoking "'0").
High relative risks of friend's smoking indicate that smokers among students form a group.In spite of high smoking prevalence in Japanese students, the higher relative risks of friend's smoking were observed among Japanese students.This result suggests that smokers among Japanese students make denser clusters.It may also suggest the presence of stronger peer pressure among Japanese students.
Family's smoking status was more strongly related to smoking behavior among Japanese students.Various studies in western countries show the relation of parent's smoking to students of the same genders), the association of parents smoking only to girls' smoking behavior 10), the relation of mother's smoking to the smoking behavior of both boys and girls 20, 21), or no relation of parent's smoking 9,19).Some studies conducted in Japan also reported the relation of parent's and sibling's smoking to student's smoking 17), whereas sister's smoking is related to student's smoking behavior as well as brother's smoking 4. Therefore, the result that female's smoking status in family was more strongly related to the student's smoking in both sexes is an unprecedent.However, the reason of these differences is difficult to explain.
On the other hand, relatively few correlates on family's smoking were related to smoking behavior of Jiangxi students.Another study conducted in Beijing reported the significant relation of father's smoking, mother's smoking, and sibling's smoking to smoking status of boys 4).The present study revealed that father's smoking and brother's smoking is related to the smoking status of boys.This may be due to the low smoking prevalence of females in China.The proportion of students that reported being offered cigarettes by someone as a reason for trying smoking among experimenters was much higher in Japanese students; i.e., 11.8% of boys and 6.3% of girls among Jiangxi experimenters reported 'offering from friends' as a reason for trying smoking, whereas Japanese experimenters reported 32.9% and 26.5%, respectively (z=44.65,p<0.01 for boys, and z=18.18,p<0.01 for girls).Moreover, 2.3% of boys and 2.5% of girls among Jiangxi experimenters reported being offered by a family member as a reason for trying smoking, whereas Japanese students reported 7.9% and 10.9%, respectively (z=21.00,p<0.01 for boys, and z=11.35,p<0.01 for girls).The Chinese policy restricting births to one per woman, in effect since 1979, and the resultingly small number of children per family could account for the relatively low odds ratios of a sibling smoking among students in Jiangxi.However, judging from the relatively low odds ratios of parents and friends smoking, Jiangxi students are less likely to be influenced by those around them.
Thinking cigarettes smoking doesn't harm one's health was more strongly related to smoking behavior among Jiangxi students.Another study conducted in Beijing also reported a significant relation of this factor to smoking status of boys 4).Besides, the proportion of Jiangxi experimenters who reported being offered by someone as a reason of trying smoking was lower than that of Japanese experimenters.This suggests that Jiangxi students are more likely to try smoking as a result of their own decision.This evidence indicates that Jiangxi students may less understand harm of smoking for one's health.Some relative risks on friend's smoking and sister's smoking were higher for girls than those in boys in both areas.The relative low smoking prevalence of girls or girls' vulnerability to the friend's and family's smoking behavior may contribute to these differences.
In conclusion, this study suggests some strategies for smoking control in this age group in each area.In Japan, important measures are to support students getting coping techniques against peer pressure and to elevate concern toward adolescent smoking among family members and society.This aspect must be included in the health education curriculum.Since smoking prevalence among young adult females has increased in Japan 13), a control measure putting emphasis on adult females is also important in preventing adolescent smoking because they intend to become mothers or already have children.
On the other hand, the anti-smoking education for school students to teach them to correctly recognize the harm of cigarette smoking to their health is more important in Jiangxi province.Since regular smokers' rate among Jiangxi students is yet low, it is expected that an earlier control measure will be more effective in Jiangxi province.
. reproduced from a recent paper of one author 3) Differences in proportion, such as the proportions of students who don't think cigarette smoking harms one's health between the two groups were tested for significance by the large sample test comparing two proportions '.The statistics of this test are expressed as z.

Table 1 .
Comparison of smoking prevalence between students in Jiangxi, China and Japan.
Sister's smoking was significantly related to smoking experience in Japanese senior students.The same was not observed in Jiangxi students.Brother's smoking was significantly related to smoking experience among boys in Jiangxi senior high school students.Correlates of current smoking among boys of Jiangxi and Japanese senior high school students Since the number of current smokers in Jiangxi girls was few, logistic regression analyses on current smoking were only computed for boys.Friend's smoking was the strongest corre-

Table 2 .
Adjusted relative risks for multiple logistic regression of explanatory variables on smoking experience among junior high school students.

Table 3 .
Adjusted relative risks for multiple logistic regression of explanatory variables on smoking experience among senior high school students.

Table 4 .
Adjusted relative risks for multiple logistic regression of explanatory variables on current smoking among junior and senior high school boys.