Household size related to prevalence of smoking in women in Japan.

We investigated the relationship between the prevalence of smoking in females and various social factors, such as household size. Ten thousand and sixty-nine subjects over 20 years of age were randomly selected from the general population of Mie Prefecture. The results showed that the habit of smoking was significantly associated with household size in women (P< 0.01), but not in men. Regarding household size, current smoking rate of women aged 20-59 who live in three-generation household was lower than those who live in others. Therefore, the steady replacement of the traditional three-generation household by smaller households in Japan may lead to an increase in the number of young women who smoke.


INTRODUCTION
The percentage of male smoker is much higher than the percentage of female smokers in the general Japanese population.This trend is similar to that of developing countries in Asia, and differs from that of industrialized European countries and the USA 1).Asian societies in general seem somewhat intolerant of women who smoke.Despite the rapid economic development and high GNP in Japan, Japanese culture still does not condone smoking among females 2).
The relationships between smoking and several social factors, such as educational level, income and occupation, have been investigated in numerous studies [3][4][5].However, studies on the relationship between smoking and the size of the smoker's household have not been reported, either in Japan or internationally, probably because households containing three generations are rare in industrialized countries other than Japan 6).Recently, however, the traditional family group in Japan has begun to disappear, and the number of family members per household is decreasing 7).This trend may have an influence on the traditional attitude towards women who smoke.
In view of this situation, we conducted a survey to determine whether smoking habits in women were related to the size of their household and survey included participants from urban and rural areas, since the populations of small municipalities tend to be more traditional and conservative than the populations of larger areas.

Sample
Ten thousand and sixty-nine subjects were randomly selected from the general population of 1,416,000 people who are older than 20 years of age and living in Mie Prefecture.The number of subjects per municipality was decided to be 0 .3-6.8% of the population in each municipality, taking into con-sideration the capacity of the municipalities' officials and the Volunteers for Promoting Improved Dietary Habits (VPIDH), who helped us conduct this survey.Officials of each municipality randomly selected the subjects from resident cards of each municipality.

Procedure
The survey was performed in December, 1994.The VPIDH in Mie Prefecture (about 5,000 individuals) delivered the questionnaire to the homes of the selected individuals and returned a few days later to collect the completed questionnaires.

Measures and Definitions
The questionnaire covered the following areas: (1)general health, (2)health consciousness, (3)smoking habits, (4)opinions on health care, and (5)personal information, including sex, age, occupation, and household size.
Of the 10,069 selected individuals, 8,846 (88%) people participated in this study.The main reason of non-response was subjects' absence at that time.There were 173 incomplete responses on respondent's sex, age or place of residence.An additional 426 (male: 59, female: 367) did not contain information on the respondent's smoking habits.Thus, the analysis was performed using the data collected from 8,247 replies (male: 3,473 or 42%, female: 4,774 or 58%).
The questionnaire contained five statements related to smoking: (1)I do not want to stop smoking, (2)1 want to reduce the number of cigarettes that I smoke, (3) I would like to stop smoking, but I can not, (4)I do not smoke now, but I used to smoke, (5)I have never smoked.Each participant selected one of these five statements.Participants who selected statements (1) -(3) were defined as current smokers, while those who selected statements (4) were defined as former smokers and those who selected statements (5) were defined as non-smokers.
Occupations were divided into four groups: storekeepers and farmers, managers and workers, housewives (including parttime workers) and retirees, and students.Thirteen occupational categories were included in the questionnaire, but responses in some categories were so small that the responses were reclassified according to the above four categories.

Data Analysis
All statistical analyses were performed using SPSS for Windows.Multiple logistic regression analyses were performed according to gender to determine whether any of four independent variables (age, place of residence, occupation and household size) were associated with the habit of smoking (Table 2).

Characteristics of sample
The sample contained 3473 male respondents (42%) and 4774 female respondents (58%).The proportion of males in this study was lower than that of the general population (48%) of men who are over 20 years of age and living in Mie Prefecture.Table 1 shows a comparison between the age and sex of the general population (according to the 1995 census) in Mie Prefecture 8) and that of the respondents.

Smoking prevalence
Fifty-two percent of the males and 8% of the females in this study were current smokers.These percentages were lower than figures reported by the Japan Tobacco Industry Inc. (male: 59%, female: 15%) 9).Furthermore, the results of the National Nutrition Survey (NNS) 10) conducted by the Ministry of Health and Welfare were 44% (1994) and 53% (1995) for the general male population and 9% (1994) and 11% (1995) for the female population.Thus, the percentage of female smokers in this survey was also lower than that of the NHS results.

Results of multiple logistic regression analyses
Both the odds ratios and the P-values are shown in Table 2. Table 2 shows that place of residence and age were associated with the habit of smoking both in men and women.Household size was also associated with the habit of smoking in women, that association was not observed for men.The place of resident-adjusted percentage of current and former smokers were examined according to age and household size (Table 3) and the age-and occupation-adjusted percentage of current smoker according to household size and place of residence (Table 4), allowing for further analyses of the relationship of household size to smoking in women.

Smoking and household size
The results of a chi-square analysis to compare the number of female current smokers and the number of female nonsmokers (including former smokers) were shown in Table 3. Table 4 shows ratios (age-and occupation-adjusted percentage of current smokers living in municipalities over 80,000 versus  age-and occupation-adjusted percentage of current smoker living in municipalities under 80,000) and 95% confidence intervals according to household size.
Women between the age of 20-39 and 40-59 who lived alone were more likely to be smokers than women living in larger household.Women between the ages of 20-39 and 40-59 who lived in three-generation households were less likely to be smokers than women living in smaller household.Women over the age of 60 showed a low overall prevalence of smoking (Table 3).The age-and occupation-adjusted percentage of women current smokers living in two-generation households was significantly larger for women living in municipalities with a population of over 80,000 than for women living in municipalities with a population of under 80 ,000 (Ratio=1.5).

DISCUSSION
The results of this survey are considered to be representative of the population of Mie Prefecture for the following reasons: a) Of the 10,069 adults over 20 years of age , randomly selected Table 4. Age-and occupation-adjusted percentage of current smoker according to place of residence and household size (for women 20-59 of age).
Ratio: Occupation-adjusted percentage of current smokers living in municipalities over 80,000 versus occupation-adjusted percentage of current smokers living in municipalities under 80,000 95%CI: 95% confidence interval from the total population of 1,416,000 living in Mie Prefecture, 88% responded to the questionnaire; and b) the proportion of respondents per age groups was similar to that of a 1995 census in Mie Prefecture (Table 1), although the proportion of subjects in their forties and fifties was slightly higher in this survey.Only 42% of the respondents were male.This percentage is lower the percentage of males in the general population, as described by the census.This finding is probably attributable to the lower questionnaire collection rate from male subjects.The questionnaires were collected by members of the Volunteers for Promoting Improved Dietary Habits (VPIDH), and the questionnaires from male subjects, who often work outside the home, were sometimes difficult to collect.
The percentage of male and female current smokers, aged 20 years or older, in Mie Prefecture, were found to be 52% and 8%, respectively.The percentage of male smokers was similar to survey results published by the Japan Tobacco Industry Inc.9) and the results of the National Nutrition Survey 10) conducted by the Ministry of Health and Welfare.However, the percentage of females who smoked was lower in this study than in the other two surveys.One explanation for this is that the percentage of female smokers in Mie Prefecture is simply lower than the national average.However, 367 female subjects did not respond to the questions related to smoking compared to only 59 males.Therefore, the actual percentage of female smoker may be higher and is probably close to the national average.
The influence on smoking in women from social factors, such as place of residence, occupation, and household size, is assumed to be greater than in males, as showed in Table 2.This assumption is supported by WHO's report that social factors contribute to the low percentage of women smokers in some countries z) because women tend to be more sensitive to being socially acceptable than men.As shown in Table 3, women between the ages of 20-39 and 40-59 who lived alone were more likely to be smokers, while women living in three-generation households were less likely.According to the Third National Survey on Household Changes (1994) 11) conducted by the Ministry of Health and Welfare, 18% of married women live with their husband's parents, while 4% live with their own parents.The presence of parents-in-lows in three-generation households may dissuade some women from smoking.However, since the percentage for former smokers living in three-generation households was similar to that of women in other household groupings (Table 3), one cannot assume that a woman smoker would stop smoking once she begun living with her husband's parents.Instead, we assumed that women who live in three-generation households inherently tend not to smoke.Indeed, Hasegawa reported that women who live in three-generation households tended to be more patient and cooperative 12).In addition, WHO reported that Asian societies have a tendency to frown upon women who smoke2).Women living in three-generation households are probably more influenced by traditional attitudes and hence do not tend to smoke.However, this trend does not seem to be applicable to the women in this survey over the age of 60, as shown in Table 3.This finding might be attributable to changes in the family household, such as the death of a husband, as suggested by the Third National Survey.
The high percentage of smoking in women who live alone, that was observed in this survey, is supported by finding of Ohida and Muramatsu, who suggested the absence of the influence from other household members as the reason 13-14) Moreover, the occupation-adjusted percentage of women smokers between the ages of 20-59 who live in municipalities with populations of over 80,000 was significantly higher than the percentage those who live in municipalities with a population of under 80,000 (Table 4).This finding was true for women living in two-generation households and households with married couples.Smoking is assumed to be a social taboo for women living in rural areas, and the opinions of neighbors probably deter women living in two-generation households or households with married couples from smoking.
The steady replacement of the traditional three-generation Japanese household by smaller household because of urbanization may therefore lead to an increase in the percentage of young women who smoke.Changes in the smoking habits of young women should thus be monitored, and it is necessary to develop programs specifically for young female smokers, potential mothers, to stop smoking in order to prevent them from smoking during pregnancy and to prevent the prevalence of smoking among women from increasing to a level similar to that among men, as seen in other industrialized countries.

Table 1 .
Percentage of study participants and the general population classified according to gender and age.

Table 2 .
Odds ratios and P-values for various independent variables vs. current smoking habits (n=8247).
OR: odds ratio Two-generation household: Household with parent(s) and child(ren)

Table 3 .
Place of residence-adjusted percentage of current and former female smokers according to age and household size.