Children’s exposure to second-hand smoke before and after the smoking ban in Bavaria—a multiple cross-sectional study

Background: The federal state of Bavaria, Germany enforced a comprehensive smoking ban across all enclosed public areas in 2008 to protect non-smokers from second-hand smoke (SHS). Evidence against displacement of smoking to homes is abundant, however long-term assessments are few. We aim to report prevalence of children’s SHS exposure before and after the ban, parental smoking behaviour and exposure risk factors. Methods: Cross-sectional data of children aged 5–6 years old in Bavaria (n = 22 944) were collected in 2004/5 and 2005/6 (S1 and S2) before the ban and after in 2008/9 and 2012/13 (S4 and S6). Parents reported their child’s home SHS exposure, in enclosed public areas and private cars. Adjusted multivariable logistic regression assessed changes across time and predicted risk factors. Results: Children’s home SHS exposure before the ban was 14.3% (S1), 14.1% (S2) and 12.8% (S4) directly after the ban to 7.2% (S6) (P<0.0001). The proportion of homes where at least one parent smoked significantly reduced from 12.78% (S1) to 4.94% (S6) (P<0.0001) and homes with voluntary smoke-free rules increased. Exposure in cafes, restaurants and private cars also decreased. No significant changes in the proportion of parents that ceased smoking due to the ban were found. Among others, low parental education, crowding and unemployment were risk factors for higher SHS exposure. Conclusion: Since the smoking ban, no long-term displacement of SHS to homes was observed. Social smoking norms appear to have shifted in favour of the ban. Social inequalities still exist and should be addressed to further minimise SHS exposure.

Introduction

Annually, over 600 000 premature deaths worldwide are second-hand smoke (SHS) related, of which 28% affect children. Lung cancer, ischemic heart disease, lower respiratory infections and asthma are known consequences of SHS exposure impacting heavily on children’s health development.¹ Following Germany’s ratification of the World Health Organization Framework Convention on Tobacco Control treaty in 2007, the federal state of Bavaria implemented a smoking ban across all enclosed public areas covering restaurants, bars and cafes in January 2008, and later amended to total prohibition in 2010. It is one of two states in Germany to have implemented such a comprehensive smoking ban protecting non-smokers from unnecessary exposure.

Children are most vulnerable to SHS exposure at home and in family cars where they spend much of their time, have little autonomy over their exposure and where legislation is not in effect.² Before any bans were implemented, up to 46% of German adolescents were regularly exposed to SHS and ∼85% spent occasional time in a room where smoking had occurred.³ Recent research in Germany also demonstrated that children exposed to SHS at home had higher average annual medical costs compared with those unexposed.⁴

Previous data in Bavaria reported ∼14% of children aged 5–6 years old were exposed to SHS at home before the ban, remaining stable at 13% in 2008/9.⁵ Concerns for displacement of smoking and the overall ban effectiveness were questionable and long-term outcomes were unknown. Evidence from recent studies continually increase support of tobacco control measures, finding no long-term displacement of SHS exposure to areas of vulnerability.^6–9 In some observations, decreases of home SHS exposure were directly attributed to smoke-free legislation,^10,11 whereas others observed smoke-free legislation had additionally introduced voluntary home bans.^12–14 Two studies observed in regions of Hong Kong and USA reported increased SHS exposure after legislation, however attributed these increases to crowding conditions rather than displacement of smoking.^15,16

The primary goal of smoke-free legislation is to eliminate SHS exposure for non-smokers but also secondarily aimed to halt non-smokers from taking up smoking; particularly children and adolescents who are more likely to initiate smoking when exposed to parental smoking behaviours.¹⁷ Since smoking rates are not definitively associated with smoking ban implementation, smoking displacement should be evaluated with respect to smoking behaviours of those providing a source of exposure.

National data on the effectiveness of bans in the recent years are sparse due to the lack of uniform legislation across federal states. Furthermore, there are few studies that assess long-term outcomes after ban implementation. Existing studies support declines in adult home SHS exposure¹⁸ but few observe exposure of children with respect to time after ban implementation.¹⁹ We aim to compare the long-term prevalence of children in Bavaria exposed to SHS at home, in enclosed public areas and private cars, before and after ban implementation. This study also determines parental opinions of smoking displacement to the home, post-ban parental smoking behaviours changes and potential SHS exposure risk factors.

Methods

Data collection and participants

The health-monitoring unit (GME) at the Bavarian Health and Food Safety Authority sent questionnaires to local health authorities of three urban (Munich, Ingolstadt, Bamberg city) and three rural regions (Schwandorf, Günzburg, Bamberg rural district) in Bavaria, Germany. Participation in the study was offered to parents of children aged 5–6 years old, along with their compulsory school entrance examinations. Due to Munich’s larger population, selected socio-demographic groups were included. More details about the concept, data collection and aims of the GME are described elsewhere.²⁰ Four questionnaires covered SHS exposure: S1 (2004/5), S2 (2005/6), S4 (2008/9) and S6 (2012/13). Questionnaire items were based on standardised scales used in previous studies.^5,21

Socio-demographics

Socio-demographic information including urbanisation ‘urbanised/rural area’, parental education, parental employment status, income relative to poverty, marriage status, child’s birthplace and living environment were gathered. Full-time employment was defined as active employment of ≥15 h per week. Household income was calculated using the Bavarian poverty risk threshold.²² Parental education level was defined as ‘high’ (university entrance certificate), ‘middle’ (upper secondary school certificate) and ‘low’ (lower/no secondary school certificate), whereby the highest level from either parent represented the overall education level. Crowding was defined as <20 m² living space per person or more than one person per room.

SHS exposure sources

Outcomes included child’s SHS exposure at home, at least one smoking parent at home, SHS exposure in publicly banned areas and private cars. Since S1 and S2 defined adjacent outdoor areas such as balconies as part of the home and the latter surveys did not, we determined exposed and non-exposed using multiple questions across the surveys. All surveys asked, ‘How many (cigarettes) are smoked on average per day (by mother/father/other) at home?’ and, ‘Does smoking occur at home (where the child lives)?’ Responses for the latter question in S1 and S2 were: ‘yes/yes but exclusively on the balcony or terrace/no’, whereas in S4 and S6 were: ‘yes daily/occasionally/no’. Due to the discrepancy between the response options and to allow comparison across surveys, exposure to SHS was defined as exposed, when >0 cigarettes were smoked at home (by anyone) and ‘yes’ or ‘yes daily/occasionally’ was the response for the latter question. Respondents in S1 and S2 who stated smoking occurred only on balconies/terraces were defined as non-exposed. We also dichotomised smoking at home by parents into ‘≥1 parent smokes at home’ and ‘no-one smokes at home’ due to the small proportion of non-parental smokers.

Potential sources of SHS exposure outside the home included restaurants/cafes and private cars. The questions appeared across all surveys with ‘frequently/occasionally/never’ responses in S1 and S2 and ‘daily/more than once per week/at most once a week/never’ options in S4 and S6. Those responding with ‘never’ for each exposure source were categorised as non-exposed and all other responses as exposed.

Household smoking behaviour changes

Parents in S4 and S6 additionally reported opinion and direct behaviour changes due to the smoking ban, dichotomised as ‘yes or more likely/no or less likely’. Self-imposed rules in the home were categorised as ‘total ban (no-smoking inside allowed)/partial ban (allowed inside but with conditions)/no ban (unrestricted for smoking inside)’.

Statistical analysis

Bivariate correlation analyses were conducted to identify confounders and significant associations (P < 0.05) were selected for inclusion in the models. To avoid multicollinearity, variables with Cramer’s V values >0.30 were excluded from regression models. Urbanisation was included in the models instead of region since it is a known determinant of smoking and SHS exposure.²³ Child age, gender, single fatherhood and reported health of the child were also excluded from the models due to correlation insignificance.

Behaviour changes such as smoking cessation and self-imposed home bans were assessed via univariate logistic regression, utilising agreement to smoking displacement theory as the dependent variable. Multivariable logistic regression was performed to analyse the exposure changes between survey years, before the ban, leading up to/during the ban and after ban implementation. Exposure was additionally analysed using time as a continuous variable to assess change per year and determine risk factors of SHS exposure. All regression models were controlled by: rural residency, full-time parental employment, high parental education, above average relative income threshold level, child’s birthplace in Germany, non-single mothers and no crowding. Due to high proportions of missing data for income in S1 (51%) and S2 (48%) as a result of questionnaire discrepancies, missing income data were excluded from main analyses. Sensitivity regression analyses with this missing data were also conducted to check for potential selection biases. SAS 9.2 (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis. The surveys and project protocol were subject to approval and cleared for ethical appropriateness.²⁰

Results

Socio-demographics

A total of 22 944 children were surveyed. Response rate of each survey was 78% (S1, n = 6350), 73% (S2, n = 6206), 61% (S4, n = 5336) and 62% (S6, n = 5052). Table 1 shows the proportions of families living in urbanised areas were greater in S4 and S6 than previous surveys. Mothers reporting full-time employment increased to almost 50% whereas fathers’ employment status remained stable. Approximately 2–3% of children were born in another country. See Supplementary table S1 for more characteristics.

SHS exposure prevalence

The prevalence of home SHS exposure significantly declined by almost half across 9 years (P<0.0001, table 2); the greatest decline between the last two surveys from 12.8% (S4) to 7.2% (S6) (supplementary figure S1). The number of homes where ≥1 parent smoked also declined from before the ban to shortly after the ban, to almost half in S6 at 4.9% (P<0.0001). Children’s SHS exposure in legally banned areas distinctly decreased from 71.9% (S1) and 68.1% (S2) to 24.0% (S4) shortly after the ban to 10.6% (S6) (P<0.0001). SHS exposure in cars followed a similar trend (10.9%, 8.9%, 5.8–4.5%, P<0.0001).

Opinions on the ban and smoking behaviour post-ban

The proportion of parents who agreed smoking bans would displace smokers to the home halved significantly from 52.8% (S4) to 25.5% (S6) (P<0.0001) (table 2). The proportion of parents admitting smoking cessation directly due to the ban did not change, however ∼5% of parent smokers stated doing so. Parents smoking at home were increasingly more likely to believe in displacement of smoking to the home after the ban, whereas parents with self-imposed smoking bans at home were significantly less likely to agree in 2012 (table 3).

Multivariable analyses and risk factors

When observing periods between surveys, significant reductions in home SHS exposure, enclosed public areas and where ≥1 parent smokes at home after ban implementation were observed, adjusting for confounders (table 4). For home SHS exposure, there were insignificant changes before and during the ban, but a significant exposure decrease between S4 and S6 [odds ratio (OR) 0.51, confidence interval 95% (CI_95%) 0.43–0.60]. Homes where ≥1 parent smoked showed significant reductions leading up to and directly after the ban and a greater reduction between the last survey years. The exposure in cars however did not differ significantly before or after ban implementation, but significantly decreased in the years surrounding the ban (OR 0.67, CI_95% 0.56–0.81).

Adjusted analyses using survey year as a continuous variable indicated significant downward trends per year from baseline survey year to the final year (table 4). Per year, the prevalence of SHS exposure at home (smoking by anyone) decreased by 8% (CI_95% 6–8%), 12% (CI_95% 10–13%) where ≥1 parent smoked at home, and 9% (CI_95% 7–11%) in private cars. Potentially associated risk factors with home exposure indicated that children living in urbanised areas were at 1.5 times greater risk and up to two times more likely to be exposed when living in crowded conditions. Low parental education appeared to increase the child’s risk to home and car exposure by 3-fold. Mothers who were part-time or inactively employed showed protective effects for children’s SHS exposure, but a child with an unemployed father was more than two times more likely to be exposed. SHS exposure was also generally more likely among single mother families across all exposure sources. Cessation of smoking due to the ban remained insignificant for both mothers and fathers. Sensitivity regression analyses with missing income data showed no significant effect changes (data not shown).

Discussion

In this study, our results indicate fewer preschool-aged children in Bavaria were exposed to SHS in places of vulnerability, declining from the period before and after smoking ban implementation. These results demonstrate no long-term displacement of SHS exposure to places of vulnerability for young children.

Evidence for direct reductions in SHS exposure remains difficult to infer from few studies observing both children and long-term prevention of SHS exposure. Some studies confirmed no displacement of SHS exposure to the home after ban implementation, despite observing adolescent and adult populations.^10,18,24,25 One study reported a general decrease of home SHS exposure to children over time, but did not account for the latency effect of the ban and was thus insignificantly associated with legislation.²⁶ Additionally, observations of increased SHS exposure after ban implementation^15,16 fuelled doubts for smoking ban effectiveness and advocated alternatives favouring taxation instead of bans.²⁷ Contrarily, long-term studies add to the growing body of literature supporting comprehensive smoking bans.^14,19,28 In a recent systematic analysis, public smoking ban policies were moderately effective in reducing SHS exposure for non-smokers.²⁹ Furthermore, comprehensive smoking bans have effectively reduced premature births, asthma hospitalisations and even cardiac events.^30,31

Current prevalence for children exposed to SHS at home from countries with comprehensive bans such as Australia³² and Denmark³³ align with our results in 2012/13: ranging from 3% to 11%. Data from 2007/8 showed home SHS exposure for children globally ranged from 9% to 16%^12,26,33, aligning with our results within the similar time period. Our findings furthermore conform to literature for decreasing trends of home SHS exposure despite living with a smoker, by adjusting smoking behaviours¹¹ but lacked evidence for smoking cessation at home, as observed by Naiman et al.³⁴ In other studies, presence of young children incentivise imposing home bans,^8,13 which may possibly confound our results.

The increase of voluntarily imposed home bans for smoking and decrease in SHS exposure was also described by these studies amongst others,^7,12 perhaps owing to increased awareness of the negative consequences of smoking and growing public support of the ban. Attitudes towards the smoking ban appeared to have altered over time after ban implementation, with only a quarter of our study respondents agreeing to smoking displacement in 2012/13 compared with half the study sample in 2008/9. This change may be due to possible selection biases: under-representation of lower socio-economic class and lower response rate in the last survey. However, if an increase in voluntary home bans observed was biased, we would expect that parents also report less smoking due to the smoking ban, which did not occur.

Another explanation for the significant shift in opinion may be reflected from the 2010 majority vote in Bavaria, amending the 2008 ban to cover exempt venues. The reasons for increased support of such measures may perhaps be attributable to increased public acceptability of smoke-free areas and social responsibility factors.³⁵ As fewer people support displacement theory and public support for bans increases, shifting of social norms against smoking since ban implementation have been observed.³⁶ Other factors such as extensive media coverage, political mooting and greater educational efforts may have also impacted on public awareness and changed the social acceptability stance. Our long-term observations suggest that bans play an important role over time by allowing gradual changes in smoking behaviour, perception and social norms to develop.

The degree of enforcement, ban comprehensiveness and timing are additionally important factors influencing SHS exposure and public support. There is evidence that those living in areas with comprehensive bans are up to 90% less likely to be exposed to SHS compared with areas of partial or no laws.³⁷ This may explain part of the ban’s effectiveness in Bavaria, where early introduction of a comprehensive ban popularised amending ban loopholes and reduced smoking prevalence.³⁸ Mons et al.⁸ suggest ban effectiveness increases with level of legislation, evident in countries such as the UK who have in parallel with comprehensive national bans observed increased voluntary smoke-free homes.^7,11,12

Finally, socio-economic factors continue to play a vital role in SHS exposure. Children with low socio-economic family background including unemployment, low education and income are consistently more likely to be exposed to SHS^39,40 concurrent with our findings. Despite this, smoking ban legislation has not exacerbated the gap between socio-economic statuses and even increased voluntary home bans amongst risk groups.³³ Suggestions that non-smoking mothers have the greatest influence over home restrictions¹² may explain why part-time or inactively employed mothers presented protective effects of children’s SHS exposure. These risk factors highlight the necessity for effective group-specific interventions.

Care in interpreting our results is advised. Limitations of our study include selection bias, attributable to low response rate and missing data particularly in the latter surveys, reducing generalisability. Subjective reporting from parents as a measure of SHS exposure may also introduce biases, which are more effectively addressed using objective biomarkers such as cotinine saliva samples. Information bias due to some discrepancies in survey items between the earlier and latter surveys may have also contributed to incorrect exposure assessment, however trends found in S4 and S6 were not subject to survey restructuring. Despite these limitations, our study provides a long-term overview of the effect of smoking bans on home SHS exposure for preschool children in Bavaria, not studied before. The study indicates no long-term displacement of smoking to homes as well as changes in perspective of the smoking ban. It also provides current prevalence of SHS exposure after ban implementation using a large sample size. Since Bavaria is one of two states with a total comprehensive smoking ban in Germany, results from this study can be used to set an example for other German federal states or countries that currently only enforce partial bans.

Conclusion

Despite concerns for displacement of smoking to the homes, SHS exposure for Bavarian children in areas of vulnerability has declined distinctly over time. Changes in parental attitudes towards smoking in the presence of children and home smoke-policies have further contributed to decreases of child SHS exposure. Social norms and acceptability of smoking in the presence of non-smokers within indoor areas appear to have shifted in the last years, with the number of complete non-smoking homes increasing and fewer people believing in displacement of smoking to the home. Children of lower socio-economic background are consistently more likely to be exposed to SHS, stressing the need to address disparities to protect the most vulnerable from the unnecessary health and social burdens of smoking and SHS exposure.

Supplementary data

Supplementary data are available at EURPUB online.

Funding

Surveys 1, 2 and 6 were funded and supported by the Department of Occupational and Environmental Epidemiology, Bavarian Health and Food Safety Authority (Bayerisches Landesamt für Gesundheit und Lebensmittelsicherheit), Munich, Germany. Survey 4 was part of the project ‘Tabakrauchbelastung von Kindern in Bayern: Ansatzpunkte für Gesundheitsförderungsstrategien auf Gemeindeebene’, which was funded by the Gesundheitsinitiative ‘Gesund.Leben.Bayern.’ of the Bavarian State Ministry of the Environment and Public Health.

Conflicts of interest: None declared.

References

Author notes