Estimating Adolescent Cigarette Consumption in Japan

We estimated the cigarette consumption among Japanese adolescents based on the data which was obtained from a 1990 nationwide school-based questionnaire survey of smoking prevalence among high school students. Cigarette consumption for adolescents was estimated using the data on current smokers* rate and cigarette consumption per day. Participants were 57,189 high school students (aged 13-18 years) including 4666 current smokers. The estimated adolescents* consumption was calculated at 3.5 to 4.3 billion units in 1990. The proportion of adolescents* consumption to the total sales was 1.1 % to 1.3%. The corresponding tax amounted to between 21 and 25 billion yen in 1990. The difference between the total sales and the crude cigarette consumption for adults has increased gradually over 20 years. Logically, some part of this increase should be attributed to cigarettes consumed by adolescents. The increase in this difference seemed to keep pace with the increasing number of cigarette vending machines. These results indicate that a considerable amount of cigarettes were consumed by adolescents in 1990. J Epidemiol, 1999; 9 : 56-60


Estimating
Adolescent Cigarette Consumption in Japan Yoneatsu Osaki, Katsuhiko Kawaminami, and Masumi Minowa We estimated the cigarette consumption among Japanese adolescents based on the data which was obtained from a 1990 nationwide school-based questionnaire survey of smoking prevalence among high school students.Cigarette consumption for adolescents was estimated using the data on current smokers* rate and cigarette consumption per day.Participants were 57,189 high school students (aged 13-18 years) including 4666 current smokers.
The estimated adolescents* consumption was calculated at 3.5 to 4.3 billion units in 1990.The proportion of adolescents* consumption to the total sales was 1.1 % to 1.3%.The corresponding tax amounted to between 21 and 25 billion yen in 1990.The difference between the total sales and the crude cigarette consumption for adults has increased gradually over 20 years.Logically, some part of this increase should be attributed to cigarettes consumed by adolescents.The increase in this difference seemed to keep pace with the increasing number of cigarette vending machines.These results indicate that a considerable amount of cigarettes were consumed by adolescents in 1990.J Epidemiol, 1999; 9 : 56-60 cigarette consumption, adolescent smoking, illegal sale Although people under 20 year of age are forbidden to smoke by the Act to Prohibit Minors from Smoking in Japan, some studies have shown a 'tobacco epidemic' among high school students 1,2,3).The cigarette consumption for adolescents is expected to be much lower than that for adults.However, there is no study that estimates cigarette consumption for Japanese adolescents.The estimation of cigarette consumption for adolescents is useful in understanding the gravity of this tobacco epidemic and for calculating for the tax income from illegal cigarette sale to minors.
We estimated the cigarette consumption for Japanese adolescents based on data from a nationwide survey of smoking prevalence among junior and senior high school students in Japan a.Moreover, we also used data about cigarette consumption in Japan, namely secular trends in cigarette sales, the smoking prevalence of adults, the cigarette consumption per day of adults, the number of cigarette vending machines, and the share of imported cigarettes, to extrapolate long-term trends in adolescent cigarette consumption.

Estimation of cigarette consumption for adolescents
The first three years of secondary education (junior high school) are compulsory and the second three years (senior high school) are non-compulsory in Japan.However, in 1989, some 94.2% of junior high school students continued on to senior high school.Therefore, data from a nationwide survey on smoking behavior among junior and senior high school students in 1990 was used to estimate the cigarette consumption of adolescents in 1990.Sampling method of the survey was a single-stage cluster sampling.Out of 11,182 junior high schools and 3,995 senior high schools listed in Japan in 1989, our study randomly selected 100 junior high schools and 50 senior high schools for participation in the survey (these sample sizes were mainly determined by financial considerations).Seventy junior high schools and thirty-three senior high schools responded to this survey.The survey was designed to be administered in classrooms, and the format was a self-reporting one.There was a 96.5% response rate among students enrolled in sampled junior high schools and a 90.7% response rate among students enrolled in sampled senior high schools.A total of 57,189 completed questionnaires were subjected to analysis.
Since the sampled high schools were chosen randomly, this was the first representative survey of the smoking status of Japanese students where no significant differences in the number of students and geographic distribution between responding and non-responding schools existed.For this reason, we estimated cigarette consumption using the data on current smokers* rate and cigarette consumption per day.Current smokers were defined as students who had smoked at least once during the previous month.The number of cigarettes per day was provided in intervals of 1-4, 5-9, 10-14, 15-19, and 20+.The lowest limit of the intervals were given for the lower estimation, namely 1, 5, 10, 15, and 20, respectively.The midvalue of the intervals were given for the higher estimation, namely 2.5, 7, 12, 17, and 22, respectively.
The academic year in Japan begins in April.Since this survey was carried out from December to January, the average age of each grade of students was given as 13.25 years for the 7th graders to 18.25 years for the 12th graders.A linear model and some nonlinear models were applied to fit the age-specific smoking prevalence and cigarette consumption by gender.Among linear, logarithmic, inverse, quadratic, cubic, compound, power, S-shaped, growth, exponential, and logistic models the most suitable model was chosen according to higher F-value and reasonable shape of the curve.Using the formulas of these curves, we estimated the smoking prevalence and cigarette consumption by age (from 12 to 19 years) and by gender.SPSS for Windows (SPSS Inc.Illinois, USA) was used for this procedure 4).The gender-and age-specific number of cigarettes per year was estimated by multiplying the smoking prevalence and the daily cigarette consumption (the lowest number or higher number) by 365.The crude number of cigarettes consumed by adolescents was calculated by summing up the age-and gender-specific cigarette consumption.
Annual adults' smoking prevalence by gender, and daily cigarette consumption per capita were used for estimating adolescents' cigarette consumption 5,6).The underreporting ratio was calculated dividing adult and adolescent consumption based on self-reporting surveys by total cigarettes sales.Secular trends in the share of imported tobacco and the number of cigarette vending machines were then compared with the trends in the estimated cigarette consumption for adults and adolescents 7) .
To estimate the percentage of cigarettes purchased from vending machines, the sum of cigarettes consumed by students who answered they bought cigarettes from vending machines was divided by the total number of cigarettes consumed by students.If students indicated various sources of cigarettes , the number of cigarettes consumed was divided by the number of the sources.

RESULTS
An S-shaped curve best fits the current smokers' rate by grade for each gender.The linear model best fits the daily cigarette consumption of boys by grade.Since no models significantly fit the daily cigarette consumption of girls by grade, the linear model was applied.The crude cigarette consumption for adolescents per year was calculated at 3.1 billion in the lower model, and at 3.9 billion in the higher model (Table 1).
The underreporting ratio (1.0 minus the self-reporting ratio) in 1990 was calculated at 10.7% by the higher model for adolescent consumption, at 10.5% by the lower model.The estimated adults' consumption was calculated at 317.8 to 318.5 billion in 1990.The estimated adolescents' consumption was calculated at 3.5 to 4.3 billion in 1990.The proportion of adolescents' consumption to the total sales was 1.1% to 1.3%.
Table 2 shows secular trends in smoking prevalence and cigarette consumption per capita per day by gender from 1973 to 1992.Although the smoking prevalence of male adults decreased during this period, the gradient of the decrease became gentle after 1986 (Table 2).The smoking prevalence of female adults decreased until 1989, but the rate did not decrease after that year (Table 2).The total smoking prevalence of adults also decreased until 1989, and that rate did not decrease since then either.The daily cigarette consumption for male adults was almost stable throughout the period, whereas the consumption for female adults gradually increased (Table 2).The annual crude cigarette consumption for adults was almost stable (Table 3).However, the total sales increased during the same period (Table 3).The rate of increase in sales became gentle in 1980s, but it steadily increased again after 1989.Therefore, the difference between total sales and crude cigarette consumption for adults has increased steadily.The increase in the difference seemed to keep pace with the increasing number of cigarette vending machines, which led us to estimate that between 52.0% and 52.8% of the cigarettes consumed by adolescents were purchased from vending machines .Interestingly, the market share of imported tobacco steadily increased after 1986 .

DISCUSSION
Differences between total sales and adults' consumption in Table 3 was much bigger than the estimated adolescents' consumption in Table 1.This difference is mainly produced by the underreporting both adults and adolescents .In the present analyses, we assumed same underreporting ratio among both adults and adolescents .However, adolescents are probably less likely to report their actual smoking status than adults because minor smoking is illegal in Japan .Besides, this estimation excluded adolescents who did not go on to senior high school , as well as dropouts and absentees .Therefore, the lower estimation of consumption for adolescents is a most conservative one.However, even the lower estimation indicates that a con -  siderable amount of cigarettes were consumed by adolescents in 1990.Since minors are prohibited from smoking by law, all of their tobacco consumption is regarded as illegal sales.
On the other hand, only 50% of cigarettes smoked by smokers 12 to 18 years of age was regarded as illegal in the United States 8).Since we estimated that about 1.1% to 1.3% of the total sales were consumed by adolescents, the sale to adolescents amounted to between 40 and 47 billion yen in 1990.The corresponding tax amounted to between 21 and 25 billion yen in 1990.This huge amount must be spent on anti-smoking measures for this age group.
The estimated cigarette consumption of the unit population for Japanese adolescents is somewhat lower than that for American adolescents, whereas the consumption for Japanese adults is higher than for American adults 9).This is due to both a lower smoking prevalence and lower cigarette consumption per capita for Japanese adolescents 10,11,12).Although adults' consumption has gradually decreased over recent years, the total sales of cigarettes have increased.One possible reason for this is a decrease in the accuracy of the self-reporting ratio over the years; Japanese adults in 1970's may have reported their smoking status and daily cigarette consumption more accurately.However, the Japanese self-reporting ratio of 90% is much higher than that of the United States (72%) 9).Since monitoring surveys on adults' smoking habits are conducted by the Japan Tobacco Industry Incorporated, smokers may feel less uncomfortable about reporting their actual smoking status.
Although we observed an increase in the difference between the total sales and the crude cigarette consumption for adults, this may not mean the increase in cigarette consumption for adolescents over years.Since some former studies on smoking behavior among adolescents and young adults reported a variety of smoking prevalences in 1970's 13,14,15) it is difficult to determine secular trends in the smoking prevalence of adolescents.Periodical surveys of adolescents' smoking are required to confirm apparent secular trends.
The increase in the number of vending machines and in the share of imported tobacco indirectly supports our suspicion of the increase in adolescent cigarette consumption because the secular trends are similar to the secular trend of the difference between the total sales and adult's consumption.Besides, cigarette vending machines are the major source of cigarettes for adolescents 1).Moreover, young adults are more likely to smoke imported cigarettes (mainly brands from the United States) 16) due to the glamour attached to their image in Japan.According to one survey of brand preferences among university students, 36.9% of cigarettes consumed were American brands 16), whereas only 17.1% of cigarettes sold in the same year, 1992, were imported brands.The most popular brands among these university students were Marlboro, Lucky Strike, and Lark16).
One recent study showed that Japanese students wearing school uniforms had no problem buying cigarettes 17).In response to the growing perception of a strong link between the increase in cigarettes vending machines and the rise in adolescent cigarettes consumption, one type of cigarette vending machine (those which protruded beyond actual shop property onto the sidewalk) has disappeared since anti-tobacco activists raised questions about their legality.The Tobacco Action Plan made by the Tobacco Control Commission of the Ministry of Health and Welfare in Japan has also strongly recommended time-based usage restrictions on all cigarette vending machines as an important measure for preventing adolescent smoking.The present data on cigarette vending machines also supports this strategy.Increased excise taxes, the banning of advertisements on TV and in magazines, and educational efforts are also recognized as important strategies for preventing adolescent smoking.
data source : 1973-1984 Nationwide Tobacco Smoking Survey, Japan Tobacco and Salt Public Corporation 1985-1992 Nationwide Tobacco Smoking Survey, Japan Tobacco Industry Incorporated Table 3. Secular trends in cigarette sales, estimeted consumption for adults and adolescent.* data source: Smoking and Health , Ministry of Health and Welfare 5 * data source: Tobacco data pal '94, (1983-) 8 * data source: Japan Vending machine Manufacture's Association (unpublished data, 1993)

Table 1 .
Estimeted age specific smoking rate and cigarette consumption by gender.

Table 2 .
Secular trends in smoking rate of adults, and daily cigarette consumption per capita.