Methods

The changes in child exposure to environmental tobacco smoke (CHETS) study is part of a national evaluation of Scottish smoke-free legislation.35 Its aim was to assess changes in SHS exposure in children after implementation of the legislation. The CHETS study used objective (salivary cotinine) and self-reported measures of exposure to SHS. Only the main design features are described here. Full details of methods and measures used are in the online appendix to this paper and in the study by Akhtar et al.6

Two nationally representative class-based surveys of children in their final year of primary school in Scotland (primary 7) were conducted in the same schools one year apart, before (January 2006) and after (January 2007) smoke-free legislation, to assess changes in exposure to SHS. Children were around 11 years of age.

The survey instrument included questions on pupil’s own smoking status and that of their friends and “parent figures” (defined as parents or step-parents whom the children reported living with all or most of the time). Children were also asked to provide a saliva sample for testing for cotinine.

Ethical approval and permissions

Ethical approval was obtained from the School of Education Ethics Committee, University of Edinburgh. Appropriate permissions to approach schools were obtained and pupils received opt-out parental consent letters before the survey and had the option to withdraw on the day of the survey.

Cotinine

Cotinine is a major metabolite of nicotine and is a sensitive indicator of the absorption of tobacco smoke and in non-smokers is a valid measure of recent (3–4 days) exposure to SHS.36 The concentration of salivary cotinine was determined using capillary gas chromatography with a specific nitrogen/phosphorus detector from a 100-μl sample.37 The assay had a detection limit of 0.1 ng/ml. Pupils who reported being current smokers or who had cotinine concentrations above 15 ng/ml, the accepted cut-off point for active smoking, were excluded.38

Parental smoking status

Parent figures were classified as smokers when described by their children as smoking “every day” or “sometimes”. Using data on family structure and reported parental figures’ smoking status, children were classified as living with “none”, “one (father figure only)”, “one (mother figure only)” or “two” smokers.

Restrictions on smoking in the home

A three-category variable on restrictions on smoking in the home was derived from children’s responses to the question “Is smoking allowed inside your home (where you live all or most of the time)?”. Children were classified as having complete restrictions on smoking in their home if they reported “No, smoking is not allowed at all” in their home, partial restrictions if they reported either that “smoking is allowed only on special occasions” or “smoking is allowed in certain places only”, and no restrictions if they reported that “smoking is allowed anywhere” in their home. Children who reported either that they did not know what restrictions were in place in their home, or left the question blank were excluded from the analysis (9.1% (n = 231) of pupils in 2006; 11% (n = 261) in 2007).

Socioeconomic status

The family affluence scale (FAS) was used to measure socioeconomic status.39 40 41 The FAS scale is derived from measures of family car ownership, bedroom occupancy, family holidays and computer ownership. A final family affluence scale score was calculated for each pupil and the sample was split into tertiles corresponding to those living in low, medium and high affluence families.42

Statistical analysis

Log transformed values were used in all analyses involving cotinine concentration, as the distribution of cotinine values was positively skewed. We assigned cotinine values below the limit of detection (0.1 ng/ml) an imputed value randomly sampled from the left tail of a truncated log normal distribution. We report geometric mean cotinine concentrations and their 95% confidence intervals.

χ² Tests were used to test differences in proportions. Multinomial logistic regression was used to examine the association of parental smoking and survey year with smoking restriction category. The category “complete restrictions” was used as the reference for comparison with “partial” and “no restrictions”. Results are presented as relative risk ratios (RR) with 95% confidence intervals. Linear regression analysis was used to examine the association between restrictions on smoking in the home, number of parent figures who smoke and mean cotinine concentration before and after legislation. Both regression models were adjusted for age and family affluence (FAS).

As individual children within a school class may be more similar with respect to restrictions on smoking in the home, SHS exposure or other measures than randomly selected children, all analyses reported take account of the survey design (clustering and stratification). The statistical analyses were performed using SAS v 9.1 for Windows.

Results

Response rates

One hundred and sixteen (68%) of 170 schools approached agreed to take part in the study before the legislation and 111 of the original 116 schools also participated at follow-up in 2007 (65% of schools originally approached). Schools that declined to participate were not significantly different from participating schools with respect to denomination, urban/rural classification, school size and socioeconomic deprivation based on the proportion of pupils receiving free school meals. In each survey year, participating schools were also representative of Scottish schools with respect to these indicators.

The final data sets contained 2527 pupil questionnaires in 2006 (85% pupil response rate) and 2379 in 2007 (83% pupil response rate), after excluding individuals for whom information on age and gender were missing and primary 6 pupils who were taught in a composite primary 6/primary 7 class. Of pupils in the final data set, over 95% in each survey year (n = 2403 in 2006 (95%); n = 2270 in 2007 (98.6%)) provided a saliva sample which was of sufficient volume for testing and gave a valid cotinine concentration. Less than 2% of children did not provide a saliva sample. Further details on response rates can be found in the paper by Akhtar et al.6

Sample characteristics

Table 1 details the characteristics of the samples. The mean age of pupils, proportion of boys and girls and proportion of pupils living in each family structure and in each family affluence group were not significantly different before and after legislation. Most pupils in both years were classified as non-smokers based on self-report and cotinine concentrations equal to or below 15 ng/ml.

Smoking restrictions in the home, parental smoking and change following smoke-free legislation

In each survey year there was an association between type of smoking restriction and number and gender of parents who smoke. Higher levels of restriction were associated with fewer smoking parents, and greater restrictions occurred where only fathers rather than only mothers, smoked. Overall, reported smoking restrictions in the home changed following the smoke-free legislation, with a lower proportion of children reporting “no restrictions” after introduction of legislation (table 2).

Table 3 presents the results for the multinomial logit model with relative risk ratios for children living in households with “partial” or “no” smoking restrictions compared with those living in households with “complete” smoking restrictions. As there is evidence that smoke-free homes are related to the age of children,26 43 we included age in the analysis, despite the relatively small range in this study. After adjusting for age, type of household smoking restriction was associated with survey year, number of smoking parents and family affluence. After legislation, pupils were at less risk of being in households with no or partial restrictions (rather than complete) compared with before legislation. When compared with children who reported living with non-smoking parents, children from homes where both parents smoked or only a mother smoked were at greater risk of being in households with partial or no restrictions (rather than complete). Children from homes where only the father smoked were also more likely to report partial restrictions and no restrictions than complete restrictions compared with children who lived with non-smoking parents. However, the relative risks were much lower than those for children who had two parents who smoked or a mother only who smoked (table 3). An interaction term between survey year and number of smoking parents was included in the model to test whether change in relative risk of restrictions between survey years was equivalent in all parental smoking groups. However, this did not reach significance (p = 0.196). Compared with high affluence children, children from low and medium affluence families were more likely to have “partial” or “no” restrictions than a “complete” smoking ban.

Exposure to SHS and household smoking restrictions and change following smoke-free legislation

Exposure to SHS as measured by salivary cotinine concentration was significantly associated with level of smoking restriction. Children who reported complete restrictions in their home had the lowest mean cotinine concentration and children who reported no restrictions had the highest. This pattern was repeated in both survey years (table 4).

Previous research indicates that smoking restrictions differ by social/occupational status27 44 45 and also by the presence of smokers in the home.13 43 Exposure to SHS by type of restrictions before and after legislation was modelled including a binary term denoting the presence or absence of a parental smoker, family affluence and adjusting for age. Exposure, as measured by cotinine concentration, was associated with restriction type, survey year, presence of a parental smoker in the home and family affluence. In addition there was a significant interaction between smoking restrictions and presence of a parental smoker as well as between smoking restrictions and survey year (table 5).

For clarity we have also presented predicted mean cotinine concentration by restriction group before and after legislation separately for children who reported living with non-smoking parents and those living with at least one parent who smokes (table 6).

In each survey year, children who reported living with non-smokers had lower exposure to SHS, as measured by mean cotinine concentration, than those living with at least one smoker, in each restriction group. In both survey years, and in each type of household (non-smokers and smokers), exposure to SHS varied significantly by type of smoking restriction in the home; mean cotinine concentration increased as fewer restrictions were reported.

Change in exposure to SHS, post legislation, was not equivalent in each restriction group, as indicated by the significant survey year by type of restriction interaction term in the model. Both for children with non-smoking parents and those living with at least one smoking parent, there were greater reductions in mean cotinine concentration among those with complete restrictions and less evidence of change in exposure among households with no smoking restrictions (table 6).

Discussion

This paper describes the patterns of household smoking restrictions reported by 11-year-old (primary 7) schoolchildren in Scotland both before and after the introduction of smoke-free legislation.

The study provides evidence that after implementation of smoke-free legislation, children were more likely to report complete restrictions on smoking in the home as opposed to only partial or no restrictions. This difference was independent of parental smoking status and may suggest a change in attitudes towards exposing children to SHS. Although Phillips et al did not find evidence of change in how adults restricted smoking in their homes, following the Scottish legislation, our findings based on households with children fitted with their finding that children were thought to be important considerations in the development and modification of household smoking restrictions.27 Although the change in the present study cannot necessarily be directly attributed to the legislation, the findings are consistent with studies that report evidence of smoke-free public places stimulating the adoption of smoke-free homes,43 26 and that numbers of completely smoke-free homes are increasing.15 22 23 24 43 In addition to the direct effect of laws to prohibit smoking in public places requiring strict work-based bans, tobacco control policies as a whole may also impact on the prevalence of smoking restrictions in private locations by shifting social norms about the acceptability of exposing others to SHS.

In line with other research our findings confirm that children who reported living with smokers were less likely to have stringent restrictions in place at home compared with children who lived with non-smoking parents.14 However, among smoker households the extent of restrictions varied according to the number and gender of parents who smoke. Those living with only a father who smoked were more likely to live in a household with a complete ban compared with children who lived with two parents or only a mother who smoked. This suggests that non-smoking mothers may have a strong influence on rules that restrict smoking around the home in line with Kegler et al’s work, which found that females in a household first brought up the idea of restricting smoking in the home and had most influence in deciding on a complete ban.34

This study also confirms that children’s SHS exposure as measured by salivary cotinine was directly related to the type of home smoking restriction reported, with lowest exposure found among those who reported complete restrictions. This is consistent with other studies,28 and understandable given that most children in our study spent most of their time at home6 and strongly suggests that, for children in this age group, the home is the most important source of SHS exposure.46 The relation between type of restrictions and exposure to SHS was found in both smoking and non-smoking households. From this we conclude, like others,29 30 31 that complete restrictions are the most beneficial method of reducing exposure to SHS among household members.

Research by Beiner and colleagues47 found, however, that even complete household smoking bans were not completely effective at eliminating home SHS exposure among adolescents living with smokers. It is probable that this is because children or adolescents who live with smokers are around at least one smoker much of the time and therefore continue to be exposed to SHS when their parents smoke outside, in the car or in other people’s homes; or from visitors and parents smoking in the home without their children’s knowledge. There is also evidence that dust and surfaces in the homes of smokers contain environmental to tobacco smoke, even when household members report no indoor smoking.48 Therefore, exposure to SHS through ingesting nicotine-laden dust via hand-mouth contact and peri-oral transfer may still occur.

Smoking rules in the car also influence children’s exposure to SHS. This study shows these were related to smoking rules in the home (analysis not shown). This is consistent with Phillips et al’s findings that adults (both smokers and non-smokers) reported having a mixture of total and partial smoking restrictions in their cars, and that these restrictions could increase in the presence of children.27 Reports of the car becoming smoke-free after legislation may also be because of the dual use of the car for private use and work, with its confined space encouraging stricter rules.27 Nevertheless, it is important to note that parents who smoke can substantially reduce SHS exposure among their children by imposing restrictions on smoking in the home where their children spend most of their time.

Our study has demonstrated that partial smoking restrictions can reduce children’s SHS exposure, but the benefits of a complete household smoking ban far outweigh those of other types of restriction. However, as we have seen, even complete restrictions in the home are not optimally protective if at least one parent continues to smoke. Instead, quitting smoking appears to be the most effective way of minimising children’s exposure to SHS. Our study has also shown an increase in the proportion of households with a complete ban on smoking after introduction of smoke-free legislation and may suggest a shift towards greater health benefits for children as a result of more homes having stricter restrictions on smoking. Although the effect cannot be directly attributed to the introduction of the legislation, the finding is of significance and supports the growing body of evidence of the wider influences that legislation may have on smoker behaviour.

Limitations

Our findings show that for children living with non-smokers and at least one smoker, the drop in mean cotinine concentration was not equivalent in each restriction group. This finding should be treated with caution, as the data containing at least one smoking parent are a composite group. For example, among children living with at least one smoker, those with complete restrictions before legislation might be made up mainly of children with both parents who smoke, whereas those after legislation might be made up of children with one parent who smokes. This study was not designed to have the power to examine this aspect in detail and numbers were too small to examine the breakdown of smoking parents. Therefore we cannot tell if the change we see across time is a change across equivalent groups.

We do not know how children interpreted the survey questions. “Smoking is not allowed anywhere in our home” could include a parent standing by an open door/window for example, and any exposure may include this.

Although there is evidence of a move towards more stringent restrictions among all families in this study, these parents’ motivations are not known. We do not have trend data on smoking restrictions in Scotland, only data covering two time points, which happen to cover a period when smoke-free legislation was introduced. It is already known that there was no change in the proportion of parents who smoke before and after legislation6; however we are not able to say whether the effect we see is solely because of the smoke-free legislation or a secular change.