Results

VISITORS SMOKING OUTSIDE

The proportion of respondents who reported requesting that visitors smoke outside increased across the study period. In 1989, only 27% (95% CI 24.8 to 28.4) of respondents indicated that they discouraged visitors from smoking, but by 1997 this proportion had nearly doubled to 53% (95% CI 51% to 55%). Table 2 presents the data for respondents discouraging visitors from smoking in the house by the smoking status of the adults in the household and by the presence of children in the household for all years the data were collected.

Across all years, respondents from non-smoking households were more likely to request that their visitors smoke outside than those from either mixed households or smoking households. However, there was a significant interaction between year and household composition (χ² = 41.41, df = 12, p < 0.001), indicating that this difference reduced over the nine year period as more respondents from smoking households also asked visitors to smoke outside the home. In 1989, non-smoking households were nearly 22 times more likely to ask visitors to smoke outside than were smoking households (odds ratio (OR) 21.92; 95% CI 11.1 to 43.5) but by 1997 non-smoking households were only six times more likely to ask visitors to smoke outside (OR 6.4; 95% CI 4.8 to 8.5). Similarly the difference between mixed households and smoking households also lessened. In 1989, the OR for mixed households was 7.6 (95% CI 3.7 to 15.4) while in 1997 the OR was 3.1 (95% CI 2.3 to 4.2).

Adding children to a household had an impact on the likelihood of asking visitors to smoke outside on top of the effect of smoking status of the adults in the household in later years, but not initially. In 1989 and 1991 there was no difference in the proportion of non-smoking, mixed, or smoking households with or without children asking visitors to smoke outside (1989, χ² = 0.72, df = 1, p > 0.40; 1991, χ² = 0.78, df = 1, p > 0.38). However, from 1992 onwards, households with children were more likely to ask visitors to smoke outside compared with households with no children (1992, OR = 1.5, 95% CI 1.3 to 1.9; 1994, OR = 1.7, 95% CI 1.4 to 2.0; 1995, OR 1.9, 95% CI 1.6 to 2.3; 1996, OR = 1.7, 95% CI 1.4 to 2.1; 1997, OR = 1.6, 95% CI 1.4 to 2.0).

Looking only at data from 1996, the belief that passive smoking is harmful to the health of non-smokers was related to asking visitors to smoke outside after adjusting for adult smoking status and presence of children. Respondents who held this belief were 3.2 times more likely to ask their visitors to smoke outside.

Data from 1997 indicate that respondents from workplaces where smoking was banned completely were more likely to discourage their visitors from smoking in their homes than were those from workplaces with partial or no smoking bans (OR = 1.8, 95% CI 1.4 to 2.3). This effect remained significant after controlling for respondents' occupational status and household composition (OR = 1.5, 95% CI 1.2 to 2.0).

BEHAVIOUR OF SMOKERS IN THEIR OWN HOMES

Table 3 presents the data on where smokers smoked at home for 1995, 1996, and 1997. Over this time, the proportion of smokers who usually smoked outside increased from 20.0% (95% CI 17.4 to 22.6) to 28.0% (95% CI 25.0 to 30.9).

Logistic regression was used to examine the impact of the household composition and the presence of children on where respondents smoked (table 4). There was a significant effect of year (χ² = 16.2, p < 0.001), indicating that the proportion of respondents who reported that they always smoked outside increased between 1995 and 1997. Combining the data for all years, smokers in mixed households were more likely to always smoke outside than smokers in an all smoker household (OR = 4.7, 95% CI 3.7 to 6.0). There was no interaction of year and household composition, suggesting that this difference was consistent across all three survey years. After adjusting for household composition and year, the presence of children made a significant impact on the likelihood of always smoking outside. Smokers in households with children were 1.6 times more likely to always smoke outside than those in households with no children (OR = 1.6, 95% CI 1.3 to 2.0). The interaction between year and presence of children was not significant. The interaction between household composition and presence of children was just significant (χ² = 3.9, df = 1, p = 0.05) indicating that having children in the household was more important in smoking households than mixed households in encouraging smoking outside.

After the presence of non-smoking adults and children in the household had been taken into account, an analysis of data from 1996 indicated that the belief that non-smokers health can be damaged by passive smoking had an impact on the probability of smokers going outside to smoke (OR = 1.9, 95% CI 1.2 to 2.9).

Data from 1995 and 1997 were used to investigate whether workplace smoking bans had an impact on where smokers smoke at home. After controlling for year, respondent occupation, and household composition, workplace smoking bans did not have a significant impact on whether a smoker smoked inside or outside at home.

Overall in 1997, in 22% of homes where at least one smoker lived, the smoker(s) always smoked outside and visitors were discouraged from smoking, indicating a consistent attempt to maintain a smoke-free home.

SMOKING IN THE PRESENCE OF CHILDREN

Table 5 shows how the presence of children influenced the amount respondents smoked for each year the question was asked. From 1989 to 1996, the number of smokers who said they did not change their smoking consumption when they were with children fell from 31% to 19% and the number of smokers who said they do not light up at all when with children increased from 14% to 33%. We combined the “smoke less” and “don't smoke at all” responses and examined the changes in the proportions of respondents who gave these responses over the years. There was a significant effect of year (χ² = 30.2, df = 3, p < 0.001) and odds ratios, indicated that compared with 1989 more smokers in 1992 (OR = 1.6, 95% CI 1.3 to 2.1), 1994 (OR = 1.6, 95% CI 1.3 to 2.1), and 1996 (OR = 1.9, 95% CI 1.5 to 2.4) said they smoked less or not at all when they were with children. We explored whether the effect of year was caused by the increase between 1989 and 1992 by rerunning the analysis for the data from 1992, 1994, and 1996. The effect of year was not significant for this analysis, indicating that there was no change in the proportion of smokers who smoked less or not at all when with children between 1992 and 1996.

Having children under 16 living in the house made it less likely that smokers would stop smoking altogether when children were around (OR = 0.65, 95% CI 0.54 to 0.78) and this was consistent across all years.

Discussion

The results indicate a clear and continuing trend towards reduction of ETS exposure around the home on all three indices reported here. This is extremely reassuring and suggests that more progress is likely to be made without the need for intensive interventions. The move to smoke-free homes can be thought of as part of a movement that is promoting this social innovation through the community.11

The belief that ETS is harmful to non-smokers has an impact on people choosing to make their homes smoke free, but as most of the change in behaviour occurred after belief in the harm of ETS reached a plateau, this is not a complete explanation. The strong moves towards making smoking less socially acceptable, including restrictions on when and where people can smoke in public and at work, are likely to have contributed.

The data indicate that people who work in places where smoking is banned are more likely to ask their visitors not to smoke. Whether this is because of greater awareness of the risks of passive smoking or the modelling of appropriate behaviour is unclear, but it seems plausible that the experience of smoke-free environments has led to increased self regulation in the home by smokers. However, the failure to find a similar effect for smokers going outside to smoke at home suggests that the effect, if real, is not strong. Broad social change (at the workplace and in public places in general) has probably also influenced the extent to which non-smokers will tolerate exposure to ETS, and their influence is clearly important.

The data show far fewer restrictions in households where all the adults smoke, even when there are also children living there. This remains a cause for concern, as protection of children must be a priority.

It is notable that it was not until 1992, just after an advertising campaign on protecting children from ETS was conducted in Victoria, that the presence of children in the household had impact on visitors being asked to smoke outside. The increase in smoking less, or not smoking at all, around children also occurred between 1989 and 1992. This is evidence of a clear beneficial effect of the advertising campaign.

It is also notable that smokers who have children in the house are more likely to smoke outside, but less likely to stop smoking entirely when children are around. This suggests that people who are frequently with children make less dramatic changes to their behaviour in response to the presence of children than those who are with children only occasionally. The behavioural changes that would be required if they were to protect their children completely from ETS are far greater than the changes required for those who are only in contact with children occasionally.

The study is limited by the relatively poor response rates, which mean that we cannot have confidence in the absolute levels of behaviour reported. The self report methodology contributes to uncertainty as to what respondents really mean, and leaves open the possibility there may be some social desirability bias. The similar response rates across years, however, make it probable that the magnitude of change reported is real. There can be no doubt that major social change is happening, but it would be extremely useful to validate these measures to help provide a clearer picture of the actual nature and extent of not smoking in homes and around children.

The rapid social change reported here contrasts with the mixed effects of focused interventions. Low level interventions do not appear to have had any clear effects, while intense interventions have some effect.12-16 It is notable that further follow up of one of these studies found marked changes in the control condition and sustained change in the intervention condition.17 This, and our data, suggest that changing social conditions are having quite a strong impact on people. If social conditions continue to change, it is likely that there will be an ongoing increase in the proportion of households imposing restrictions on smoking, which will result in non-smokers (including children) being protected from exposure to ETS wherever they may be.

There are limited data to compare the situation in Australia with that in other countries. In the United States in 1996, it was found that in homes with at least one adult smoker and one child, smoking was permitted in at least part of the house in between 70.6% and 95.6% of homes.18 Biener et al found that in Massachusetts in 1993, 25% of adolescents living with smokers reported smoking bans and 23% reported that visitors were not allowed to smoke.19 In 1995, a Scandinavian study of parents of three year olds found that in households with at least one smoker, 30% did not appear to be exposing the children to any ETS at home.20 Although the two latter studies represent specific subsets of the population, they still suggest that in some countries at least, movement towards smoke-free homes is progressing at a broadly similar rate to that in Australia.21 However, without data it would be inappropriate to generalise that similar trends were occurring in other countries, especially those with quite different sociocultural profiles.

The results indicate that much can be achieved in ETS control through working at a population level. The marked change that has occurred at a structural level and without intensive intervention suggests that the circumstances in which intensive interventions are necessary may be limited. A continuation of the efforts to make ETS exposure socially unacceptable are well justified and should continue. Public education and programmes to promote smoke-free public places need to ensure that their messages implicitly or explicitly communicate the importance of protecting others from exposure in the home. The results also indicate that a media campaign targeted at encouraging people to protect children from ETS, particularly in their own homes, can be effective. Evidence that an increasing number of people, including smokers, are declaring their homes smoke free needs to be promoted, so that those who have not yet made the move are aware that smoking in the home is becoming unacceptable. This should hasten the transition to a society where non-smokers can expect not to be exposed to ETS.