You are here

Article Text

Article menu

Download PDFPDF

Smoking
The impact of smoke-free legislation on educational differences in birth outcomes
  1. Britt McKinnon1,2,
  2. Nathalie Auger3,4,
  3. Jay S Kaufman1,2
  1. 1Department of Epidemiology, Biostatistics and Occupational Health, McGill University Faculty of Medicine, Montreal, Quebec, Canada
  2. 2Institute for Health and Social Policy, McGill University, Montreal, Quebec, Canada
  3. 3Institut national de santé publique du Québec, Montreal, Quebec, Canada
  4. 4University of Montreal Hospital Research Centre, Montreal, Quebec, Canada
  1. Correspondence to Dr Britt McKinnon, Institute for Health and Social Policy, 1030 Pine Avenue West, Montreal, Quebec, Canada H3A 1A2; britt.mckinnon{at}mcgill.ca

Abstract

Background Smoke-free legislation may have positive effects on birth outcomes. Given that smoking and secondhand smoke during pregnancy vary with socioeconomic position, legislation may have greater effects in some socioeconomic groups. For this study, we evaluated the impact of a 2006 ban on smoking in public places in the Canadian province of Quebec on preterm birth, small-for-gestational-age birth and birth weight, and on educational differences in these birth outcomes.

Methods We analysed data on singleton births in Quebec between 2003 and 2010. Logistic regression was used to model the association of smoke-free legislation with preterm birth (<37 weeks), very preterm birth (<34 weeks), small-for-gestational-age birth (<10th centile for gestational age and sex), low birth weight (<2500 g) and mean birth weight, adjusting for secular trends before and after legislation. Interaction terms were included to assess differential effects by level of maternal education.

Results Smoke-free legislation was associated with average reductions of 3.1 preterm births (95% CI 0.1 to 6.0), 2.3 very preterm births (95% CI 0.9 to 3.7), 5.9 small-for-gestational-age births (95% CI 2.6 to 9.3) and 1.0 low birthweight infants (95% CI 0.4 to 1.6) per 1000 live births, as well as a 17.1 g increase in mean birth weight (95% CI 10.7 to 23.6). Legislation was associated with improved birth outcomes in all categories of maternal education.

Conclusions Smoke-free legislation in Quebec was associated with reductions in preterm and small-for-gestational-age births, and an increase in birth weight. There was no compelling evidence that legislation impacted educational gradients in birth outcomes.

  • Health inequalities
  • PUBLIC HEALTH POLICY
  • PERINATAL EPIDEMIOLOGY
  • SMOKING

https://doi.org/10.1136/jech-2015-205779

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Background

    Reducing social gradients in preterm birth, perinatal mortality and other adverse birth outcomes is a priority for policymakers and national governments.1 ,2 Smoking and exposure to secondhand smoke during pregnancy have been linked to adverse birth outcomes, and vary markedly with socioeconomic position.3–5 Evidence suggests that reducing exposure to tobacco smoke during pregnancy could narrow socioeconomic gradients in stillbirth and infant death by as much as 30–40%.6

    There is strong evidence that smoke-free laws reduce exposure to secondhand smoke, although the impact on population smoking rates is mixed.7 In Canada, adoption of public-place smoking bans was shown to reduce exposure to environmental tobacco smoke by more than 60% in smokers and non-smokers.8 Evidence also suggests a link between smoke-free legislation and improvement in health outcomes, particularly cardiovascular and respiratory diseases.9 Moreover, several recent studies found positive effects on perinatal outcomes, including preterm birth,10–13 small-for-gestational-age (SGA) birth11 ,14 and low birth weight.11 However, data are conflicting as some studies found no association between smoke-free legislation and birth outcomes.13 ,15

    Smoking and exposure to secondhand smoke during pregnancy vary with socioeconomic position, with less advantaged populations having both greater exposure to secondhand smoke and higher rates of prenatal smoking.16 ,17 The impact of smoke-free legislation on perinatal outcomes may be greater in some socioeconomic groups. Some studies have found no evidence of differences in the effects of smoking bans on exposure to secondhand smoke by level of education or income, while others have found greater effects among advantaged groups.18–20 Little is known about differential effects of smoke-free legislation on birth outcomes according to socioeconomic position.

    In the Canadian province of Quebec, the Tobacco Act prohibited smoking in all indoor workplaces and public places including bars and restaurants from 31 May 2006.21 In 2005, 1 year prior to the smoking ban, nearly 20% of mothers in Quebec reported exposure to secondhand smoke during pregnancy, higher than the Canadian average of 14.1%.2 Rates of preterm birth among singleton births in Quebec declined from 6.3% for the time period 2000–2004 to 6.0% in 2005–2008.22 Whether the 2006 smoking ban contributed to this decline has not been investigated. For the present study, we evaluated the impact of Quebec's 2006 ban on smoking in public places on preterm birth, SGA birth and birth weight, and on educational differences in these birth outcomes.

    Methods

    Data on births in Quebec during the period 2003–2010 were extracted from the provincial birth file. The file is based on birth registration certificates and is considered to be a complete record of all births in Quebec.23 The institutional review board of the University of Montreal Hospital Centre waived the requirement for ethics review for this study. We included singleton live births that delivered at 22–44 weeks of gestation. Gestational age on Quebec birth certificates is based predominantly on ultrasound examinations, with >92% of pregnant women having at least one ultrasound examination before 20 weeks of gestation.24

    We examined the effect of Quebec's 2006 smoking ban on educational differences in five outcomes: preterm birth, very preterm birth, SGA birth, low birth weight and mean birth weight. Preterm birth was defined as gestational age <37 weeks, and very preterm birth as gestational age <34 weeks. Although <32 weeks is a more common definition for very preterm birth, we used <34 weeks to ensure more stable estimates, given the relative infrequency of gestational ages <32 weeks in Quebec and our interest in differential effects by education. SGA birth was defined as birth weight less than the 10th centile for the gestational age and sex of the baby. SGA percentiles are based on the population reference constructed from live births in Canada for 1994–1996.25 Low birth weight was defined as <2500 g.

    We used maternal education as our measure of socioeconomic position, categorised as 12 or less, 13–15 and 16 years or more. These categories generally reflect diplomas obtained for secondary school, post-secondary collegiate and university, respectively. Education is a commonly used proxy measure of socioeconomic position that strongly and consistently predicts adverse perinatal outcomes.26 We included several potential confounders in our analysis: maternal age (<25, 25–34 and ≥35 years), infant sex, number of previous deliveries (0, 1, 2+), marital status of the mother (married vs unmarried), maternal nativity (Canadian born vs Foreign born), year and month of birth and an index of material deprivation (categorised into quintiles). The material deprivation index is an area-level index that is estimated for Canadian neighbourhoods based on the postal codes of mothers at the time of birth.27

    Our sample included 470 199 births between June 2003 and June 2009. The 6-year time period was selected to strike a balance between: (1) having adequate pretime and post-time periods to estimate underlying trends and (2) minimising the chance that other exogenous shocks (eg, other policies) that affect birth outcomes would contribute to our observed effects. Information on maternal education, material deprivation and maternal nativity was missing for 9.4%, 2.2% and 1.4% of births, respectively. All births had complete information on gestational age, maternal age, infant sex and marital status of the mother. A total of 63 births (0.01%) were missing birth weight and dropped from the analyses of birth weight and SGA birth. Table 1 shows descriptive statistics for the outcomes and covariates by maternal education levels, including missing categories. To account for missing data, we performed multiple imputation using chained equations.25 Education, maternal nativity and the deprivation index were imputed using demographic predictors (maternal age, marital status, urban/rural residence, etc) and a total of 10 imputed data sets were generated. Results were pooled across imputed data sets using the mi estimate procedures in Stata V.12.28

    View this table:
    Table 1

    Descriptive statistics for singleton live births by maternal education level, Quebec 2003–2009

    Statistical analysis

    We explored monthly time trends in preterm birth, SGA birth and birth weight using locally weighted scatterplot smoothing (Loess) with bandwidth=0.2, which we estimated separately for the time periods before and after the smoke-free legislation. Using logistic regression, we then examined changes in preterm birth, very preterm birth, SGA birth and low birth weight following the introduction of smoke-free legislation. The basic regression model has the form:Embedded Image where policy is a binary variable indicating whether birth occurred after passage of smoke-free legislation (prepolicy vs postpolicy), time is a continuous variable for months since the start of the observation period (June 2003), and postslope is a continuous variable for months since passage of the legislation in June 2006 (with prepolicy births coded as 0). Here, β1 is the baseline preterm (or SGA) birth rate at time 0 (June 2003), β2 the underlying trend in preterm birth independent of the policy, β3 the immediate impact of legislation on preterm birth (the ‘step’ change), and β4 the change in the preterm birth trend after legislation (the ‘slope’ change). As our primary interest was in assessing differential effects of the legislation by maternal education, we included interaction terms between education and the underlying trend (time), the step change (policy) and the postpolicy slope change (postslope). To facilitate interpretation and assess differences on the absolute probability scale, we report average marginal effects estimated from regression coefficients.29 For birth weight, we used linear rather than logistic regression.

    We hypothesised that the full impact of the smoking ban on perinatal outcomes would not be immediately apparent at the time of implementation, given the chronic nature of some effects of tobacco smoking on perinatal outcomes.30 As such, we assessed effects for 3, 6 and 9-months lags following the introduction of the ban (September 2006, December 2006 and March 2007, respectively). To be consistent, we used data for 36 months on either side of each date. As a sensitivity analysis, we ran the analysis using June 2003 as an arbitrary negative control date of policy implementation, again analysing the 36 months before and after this date. Our assumption was that there should be no effect of a non-existent policy on our outcomes. We chose June 2003 to ensure that births occurring after the smoking ban were excluded from the postpolicy time period. All analyses were performed using Stata V.12.1.31

    Results

    There were strong educational gradients in preterm birth, SGA birth and low birth weight (table 1). For example, 7.4% of infants born to mothers with ≤12 years of education were preterm compared with 5% of those born to mothers with 16+ years of education. Among infants whose mothers were missing education level, 6.6% were preterm. The distribution of sociodemographic variables stratified by maternal education is shown in table 1. As expected, mothers with the lowest education level tended to be younger, unmarried and live in more deprived neighbourhoods compared with women with higher levels of education.

    Figure 1 presents smoothed time trends for preterm birth, SGA birth and birth weight by education level, allowing for a discontinuity in the trend 9 months following the smoke-free legislation (March 2007). Although changes over time were relatively small and seasonal variation is evident across all outcomes, the overall trends were consistent with a decline in preterm and SGA births and a slight increase in mean birth weight following implementation of the smoking ban.

    Figure 1
    Figure 1

    Trends in preterm birth, small-for-gestational-age (SGA) and mean birth weight by level of education, Quebec 2004–2010. Monthly trends were smoothed using ‘loess’ with bandwidth=0.2 and estimated separately for the time periods before and after the smoke-free legislation.

    Table 2 shows absolute effects of the smoking ban on the five outcomes at the time of implementation and at 3, 6, and 9-months of postlegislation. At the immediate time of the legislation, there was no significant association between smoke-free legislation and birth outcomes. The magnitude of the effect size tended to increase over time, with the largest effects seen at 9 months of postlegislation for all outcomes. According to covariate-adjusted results at 9 months of postlegislation, the smoking ban was associated with reductions of 3.1 preterm births (95% CI 0.1 to 6.0), 2.3 very preterm births (95% CI 0.9 to 3.7), 5.9 SGA births (95% CI 2.6 to 9.3) and 1.0 low birthweight infants (95% CI 0.4 to 1.6) per 1000 live births, and a 17.1 g increase in mean birth weight (95% CI 10.7 to 23.6).

    View this table:
    Table 2

    Associations of smoke-free legislation with birth outcomes, immediately and at 3, 6 and 9-months of postlegislation

    Figure 2 shows the predicted prevalence of each outcome by maternal education level in the presence and absence of the smoke-free legislation. Estimated absolute and relative changes in the prevalence of each outcome by education level are presented in table 3. In general, smoke-free legislation was consistent with reductions in preterm and SGA births and an increase in mean birth weight across all levels of maternal education. However, some 95% CIs for the estimated changes overlapped the null. In relative terms, the largest effects were seen for very preterm birth, with more than a 10% estimated reduction in each education category.

    Figure 2
    Figure 2

    Predicted prevalence and 95% CIs of birth outcomes in the presence and absence of smoke-free legislation according to level of maternal education. Preterm birth (gestational age <37 weeks), very preterm birth (gestational age <34 weeks). Effects are estimated 9 months postlegislation and are adjusted for maternal age, infant sex, number of previous deliveries, marital status, maternal nativity, year of birth and month of birth.

    View this table:
    Table 3

    Covariate-adjusted estimates of absolute and relative effects of smoke-free legislation on birth outcomes by maternal education level

    As to differential effects across levels of education, the legislation appeared to have stronger effects among women in the lowest education group, especially for very preterm birth, SGA birth and low birth weight. For example, among women with ≤12 years of education, the smoking ban was associated with reductions of 4.0 very preterm births (95% CI 1.4 to 6.7), 8.5 SGA births (95% CI 2.2 to 14.7) and 8.4 low birthweight infants (95% CI 3.7 to 13.1) per 1000 live births. This is compared to reductions of 1.4 very preterm births (95% CI −0.6 to 3.4), 5.3 SGA births (95% CI 0.2 to 10.4) and 2.5 (95% CI −1.1 to 6.1) low birthweight infants per 1000 live births among women with 16+ years of education. However, when we tested for heterogeneity of the absolute and relative effect estimates across education categories, there was little statistical evidence of differential effects of the legislation (heterogeneity p values for all outcomes >0.05; table 3). The strongest evidence that less educated mothers benefited more from the policy change was seen for low birth weight, with estimated reductions of 8.4, 0.6 and 2.5 low birthweight infants per 1000 live births for ≤12, 13–15 and 16+ years of education, respectively (heterogeneity p value=0.058).

    The sensitivity analysis for the arbitrary June 2003 policy date is shown in online supplementary table 1 of the online supplementary material. Estimates are largely null and differ markedly from our effect estimates reported in table 2. Replicating the results in tables 2–3 using non-imputed data gave very similar estimates (not shown).

    Discussion

    We evaluated the impact of Quebec's public smoking ban on educational differences in preterm birth, SGA birth and birth weight. We estimated that smoke-free legislation was associated with approximately 6 fewer SGA births and 2.3 fewer very preterm births per 1000 live births, and a 17 g increase in mean birth weight. Effects of the legislation were present across all education levels, with no compelling evidence that educated women benefited more from the legislation.

    Several studies have reported decreases in preterm birth,11 ,12 SGA birth,11 ,14 and low birth weight11 following the introduction of public place smoking bans. Other studies failed to find an impact on some13 or all birth outcomes.15 While our study provides additional evidence of positive effects on birth outcomes following smoke-free legislation, the existing body of evidence on the topic remains mixed. Several factors may explain why various studies have come to different conclusions about the impact of smoke-free legislation on birth outcomes: methodological differences between studies (eg, before–after studies vs quasi-experimental studies),13 ,15 policy environments that differ in terms of smoking prevalence or smoking norms, differences in policy implementation and enforcement across countries or states, or the presence of local smoking bans already in place prior to country or state-level bans.15

    Exposure to tobacco smoke during pregnancy is associated with fetal growth restriction and indicated preterm birth, and can also trigger systemic inflammatory reactions leading to spontaneous preterm birth.30 High environmental tobacco exposure during the first trimester is associated with low birth weight, preterm birth and most strongly with very preterm birth.32 Given the chronic nature of some effects of tobacco smoking on perinatal outcomes, we hypothesised that the full impact of a smoking ban might not be immediately apparent at the time of implementation. Our results confirmed that the magnitude of effects for all outcomes tended to increase over the 9 months of postlegislation. To the best of our knowledge, all but one previous study15 examined immediate changes in outcomes at the time of policy implementation without considering lagged effects. Interestingly, a Scottish study reported stronger effects at 3 months of prelegislation, claiming an anticipatory effect of the smoking ban, but did not discuss the magnitude of effects postlegislation.11 Further research examining lagged effects on perinatal outcomes after other smoking bans seems warranted.

    Several studies have examined differential effects of smoking bans on exposure to secondhand smoke by education or income, with most failing to find an impact on socioeconomic gradients.19 Many of these studies, however, had limited study power to detect differences at a subpopulation level. While our results point to potentially greater reductions in very preterm, low birth weight and SGA births among babies born to less educated mothers, we did not find robust evidence of differential effects of the legislation by maternal education level.

    There are several limitations to this study. First, we were restricted to information routinely collected on birth certificates, and did not have information on some potentially important variables. For example, we could not distinguish between spontaneous versus induced deliveries, which have different aetiologies. A recent study, however, found similar effects of smoking legislation on spontaneous preterm delivery compared to overall preterm delivery.11 Similar to most previous studies evaluating smoking bans, we did not have information on maternal smoking.12 One study with data on maternal smoking status found reductions in adverse pregnancy outcomes among smokers and non-smokers.11 Another study found no association between smoke-free legislation and birth outcomes after conditioning on maternal smoking.15 We also cannot rule out whether some changes in how socioeconomic status was collected over time could have masked educational differences in the impact of legislation.

    We analysed trends in perinatal outcomes before and after the policy change, a common strategy to evaluate the impact of smoking bans on health outcomes.13 ,33 This method, however, does not include comparison to a contemporaneous control group not exposed to the policy change. Unfortunately, we could not use data from other Canadian provinces as a control, because Quebec is unique in collecting maternal education on birth certificates in Canada. However, the null results from the negative control policy sensitivity analysis, along with the observed increasing magnitude of effects over the 9 months following the smoking ban, strengthen the plausibility of our results. Finally, despite having records on all births in Quebec over the study period, our estimates by maternal education level were fairly imprecise, hindering our ability to make strong inferences about differential effects of the policy change.

    In conclusion, our study showed reductions in preterm birth and SGA birth and an increase in mean birth weight 9 months after legislation banning smoking in public places in the province of Quebec. These results add to a growing evidence base supporting a positive impact of smoking bans on health in early life. We also contribute new evidence that effects of the smoking ban occurred across all levels of education, with no compelling evidence of greater benefit among more advantaged populations.

    What is already known on this subject?

    • Several studies have shown positive effects of smoke-free legislation on birth outcomes, including preterm birth, low birth weight and small-for-gestational-age birth. It is not known whether effects of this legislation differ according to socioeconomic position.

    What this study adds?

    • We provide further evidence of positive effects of smoke-free legislation on preterm birth, small-for-gestational-age birth and low birth weight. Legislation was associated with improved birth outcomes across categories of maternal education, with no compelling evidence that it impacted educational gradients in birth outcomes.

    Acknowledgments

    NA acknowledges a career award from the Fonds de recherche du Québec-Santé. JSK is supported by the Canada Research Chairs programme.

    References

    1. ↵
      1. Blumenshine P,
      2. Egerter S,
      3. Barclay CJ, et al
      . Socioeconomic disparities in adverse birth outcomes: a systematic review. Am J Prev Med 2010;39:263–72. doi:10.1016/j.amepre.2010.05.012
      OpenUrlCrossRefPubMed
    2. ↵
      Public Health Agency of Canada. Canadian Perinatal Health Report, 2008 Edition. Ottawa: 2008. http://www.phac-aspc.gc.ca/publicat/2008/cphr-rspc/index-eng.php
    3. ↵
      1. Salihu HM,
      2. Wilson RE
      . Epidemiology of prenatal smoking and perinatal outcomes. Early Hum Dev 2007;83:713–20. doi:10.1016/j.earlhumdev.2007.08.002
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Hiscock R,
      2. Bauld L,
      3. Amos A, et al
      . Socioeconomic status and smoking: a review. Ann N Y Acad Sci 2012;1248:107–23. doi:10.1111/j.1749-6632.2011.06202.x
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Schneider S,
      2. Schütz J
      . Who smokes during pregnancy? A systematic literature review of population-based surveys conducted in developed countries between 1997 and 2006. Eur J Contracept Reprod Health Care 2008;13:138–47. doi:10.1080/13625180802027993
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Gray R,
      2. Bonellie SR,
      3. Chalmers J, et al
      . Contribution of smoking during pregnancy to inequalities in stillbirth and infant death in Scotland 1994–2003: retrospective population based study using hospital maternity records. BMJ 2009;339:b3754. doi:10.1136/bmj.b3754
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Callinan JE,
      2. Clarke A,
      3. Doherty K, et al
      . Legislative smoking bans for reducing secondhand smoke exposure, smoking prevalence and tobacco consumption. Cochrane Database Syst Rev 2010;4:CD005992. doi:10.1002/14651858.CD005992.pub2
      OpenUrlPubMed
    8. ↵
      1. Carpenter C,
      2. Postolek S,
      3. Warman C
      . Public-place smoking laws and exposure to environmental tobacco smoke (ETS). Am Econ J Econ Policy 2011;3:35–61. doi:10.1257/pol.3.3.35
      OpenUrl
    9. ↵
      1. Tan CE,
      2. Glantz SA
      . Association between smoke-free legislation and hospitalizations for cardiac, cerebrovascular, and respiratory diseases. Circulation 2012;126:2177–83. doi:10.1161/CIRCULATIONAHA.112.121301
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Page RL II.,
      2. Slejko JF,
      3. Libby AM
      . A citywide smoking ban reduced maternal smoking and risk for preterm births: a colorado natural experiment. J Womens Health (Larchmt) 2012;21:621–7. doi:10.1089/jwh.2011.3305
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Mackay DF,
      2. Nelson SM,
      3. Haw SJ, et al
      . Impact of Scotland's smoke-free legislation on pregnancy complications: retrospective cohort study. PLoS Med 2012;9:e1001175. doi:10.1371/journal.pmed.1001175
      OpenUrlCrossRefPubMed
    12. ↵
      1. Cox B,
      2. Martens E,
      3. Nemery B, et al
      . Impact of a stepwise introduction of smoke-free legislation on the rate of preterm births: analysis of routinely collected birth data. BMJ 2013;346:f441–1. doi:10.1136/bmj.f441
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Been JV,
      2. Nurmatov UB,
      3. Cox B, et al
      . Effect of smoke-free legislation on perinatal and child health: a systematic review and meta-analysis. Lancet 2014;383:1549–60. doi:10.1016/S0140-6736(14)60082-9
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Kabir Z,
      2. Daly S,
      3. Clarke V, et al
      . Smoking ban and small-for-gestational age births in Ireland. PLoS ONE 2013;8:e57441. doi:10.1371/journal.pone.0057441
      OpenUrlCrossRefPubMed
    15. ↵
      1. Hawkins SS,
      2. Baum CF,
      3. Oken E, et al
      . Associations of tobacco control policies with birth outcomes. JAMA Pediatr 2014;168:e142365. doi:10.1001/jamapediatrics.2014.2365
      OpenUrlCrossRef
    16. ↵
      Centers for Disease Control and Prevention. Vital signs: nonsmokers’ exposure to secondhand smoke—United States, 1999–2008. MMWR Morb Mortal Wkly Rep 2010;59:1141–6.
      OpenUrlPubMed
    17. ↵
      1. Hawkins SS,
      2. Baum CF
      . Impact of state cigarette taxes on disparities in maternal smoking during pregnancy. Am J Public Health 2014;104:1464–70. doi:10.2105/AJPH.2014.301955
      OpenUrlCrossRefPubMed
    18. ↵
      1. Brown T,
      2. Platt S,
      3. Amos A
      . Equity impact of population-level interventions and policies to reduce smoking in adults: a systematic review. Drug Alcohol Depend 2014;138:7–16. doi:10.1016/j.drugalcdep.2014.03.001
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Hill S,
      2. Amos A,
      3. Clifford D, et al
      . Impact of tobacco control interventions on socioeconomic inequalities in smoking: review of the evidence. Tob Control 2014;23:e89–97. doi:10.1136/tobaccocontrol-2013-051110
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Moore GF,
      2. Holliday JC,
      3. Moore LAR
      . Socioeconomic patterning in changes in child exposure to secondhand smoke after implementation of smoke-free legislation in Wales. Nicotine Tob Res 2011;13:903–10. doi:10.1093/ntr/ntr093
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Goodman PG,
      2. Haw S,
      3. Kabir Z, et al
      . Are there health benefits associated with comprehensive smoke-free laws. Int J Public Health 2009;54:367–78. doi:10.1007/s00038-009-0089-8
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Auger N,
      2. Hansen AV,
      3. Mortensen L
      . Contribution of maternal age to preterm birth rates in Denmark and Quebec, 1981–2008. Am J Public Health 2013;103:e33–8. doi:10.2105/AJPH.2013.301523
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Auger N,
      2. Roncarolo F,
      3. Harper S
      . Increasing educational inequality in preterm birth in Quebec, Canada, 1981–2006. J Epidemiol Community Health 2011;65:1091–6. doi:10.1136/jech.2009.102350
      OpenUrlAbstract/FREE Full Text
    24. ↵
      Public Health Agency of Canada. What mothers say: the Canadian Maternity Experiences Survey. 2009. http://www.phac-aspc.gc.ca/rhs-ssg/pdf/survey-eng.pdf
    25. ↵
      1. Kramer MS,
      2. Platt RW,
      3. Wen SW, et al
      . A new and improved population-based Canadian reference for birth weight for gestational age. Pediatrics 2001;108:E35. doi:10.1542/peds.108.2.e35
      OpenUrlCrossRefPubMed
    26. ↵
      1. Kramer MS,
      2. Seguin L,
      3. Lydon J, et al
      . Socio-economic disparities in pregnancy outcome: why do the poor fare so poorly? Paediatr Perinat Epidemiol 2000;14:194–210. doi:10.1046/j.1365-3016.2000.00266.x
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Pampalon R,
      2. Hamel D,
      3. Gamache P, et al
      . A deprivation index for health planning in Canada. Chronic Dis Can 2009;29:178–91.
      OpenUrlPubMedWeb of Science
    28. ↵
      StataCorp. Multiple-imputation reference manual. College Station, TX: Stata Press, 2011.
    29. ↵
      1. Muller CJ,
      2. MacLehose RF
      . Estimating predicted probabilities from logistic regression: different methods correspond to different target populations. Int J Epidemiol 2014;43:962–70. doi:10.1093/ije/dyu029
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Goldenberg RL,
      2. Culhane JF,
      3. Iams JD, et al
      . Epidemiology and causes of preterm birth. Lancet 2008;371:75–84. doi:10.1016/S0140-6736(08)60074-4
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      StataCorp. Stata Statistical Software: Release 12. TX: 2011.
    32. ↵
      1. Windham GC,
      2. Hopkins B,
      3. Fenster L, et al
      . Prenatal active or passive tobacco smoke exposure and the risk of preterm delivery or low birth weight. Epidemiology 2000;11:427. doi:10.1097/00001648-200007000-00011
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Millett C,
      2. Lee JT,
      3. Laverty AA, et al
      . Hospital admissions for childhood asthma after smoke-free legislation in England. Pediatrics 2013;131:e495–501. doi:10.1542/peds.2012-2592
      OpenUrlAbstract/FREE Full Text
    View Abstract

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors BM, NA and JSK conceived the study and developed the analysis plan. BM conducted the analysis and wrote the first draft. All authors interpreted the findings, critically reviewed the manuscript for important intellectual content, and approved the final version.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.