Investigating the link between smoke-free legislation and stillbirths

ABSTRACT

Despite considerable recent progress in tobacco control, smoking and second-hand smoke exposure continue to pose a major health threat to adults, children, and (unborn) babies. There is increasing evidence that implementation of smoke-free legislation, through reducing smoking and smoke exposure, has the potential to improve population health. In this editorial we focus on the research on smoke-free legislation in relation to stillbirths, summarizing the findings to-date, reflecting on methodological issues that need to be considered when interpreting this evidence base, and highlighting some key next steps to further strengthen the evidence in order to inform evidence-based policy making.

KEYWORDS:

• Smoking
• tobacco
• pregnancy
• second-hand smoke
• policy
• prevention
• fetus
• quasi-experiment
• perinatal mortality
• methodology

There is increasing evidence that implementation of smoke-free legislation has the potential to improve health outcomes in adults,[1] children,[2] and perinatal life.[2] In this editorial, we focus on the research in relation to stillbirths, summarizing the findings to date, reflecting on methodological issues that need to be considered when interpreting this evidence base and highlighting some key next steps to further strengthen the evidence in order to inform evidence-based policy making.

Regional laws restricting smoking in public places were first implemented in the 1970s in some US states.[3] Increasing evidence on the adverse health effects of second-hand smoke (SHS) exposure subsequently led to the enactment of comprehensive regional and national smoke-free laws covering both workplaces and public places. In 2005, 100% smoke-free indoor public environments became an integral part of the World Health Organization’s (WHO) approach to tackling the tobacco epidemic through the Framework Convention on Tobacco Control.[4] The Republic of Ireland, Norway, Uruguay, and Scotland were among the earliest adopters of comprehensive national laws. The growing compliance with the WHO recommendation to implement 100% smoke-free public environments provides an opportunity to assess and quantify the associated health benefits.[2]

In 2000, the Millennium Development Goals set ambitious targets for reducing maternal and child mortality. The Millennium Development Goals did not, however, include stillbirths, and although stillbirths have declined by 15% since 1995, this reduction has been much smaller than for maternal, neonatal, and child mortality.[5] Worldwide, an estimated 2.6 million babies were stillborn after 28 weeks gestation in 2009.[6] This constitutes a significant public health concern with considerable impact on families worldwide. Preventable risk factors play an important role in a substantial proportion of stillbirths and include for example maternal overweight and obesity and tobacco smoking.[7]

Approximately 10–20% of women in high-income countries smoke throughout pregnancy.[8] Tobacco smoking and SHS exposure during pregnancy increase the risk of stillbirth by 36–47% and 23%, respectively.[7,9,10] A regional study in England estimated the population attributable risk for stillbirth to be 15.5% for active smoking and 9.1% for SHS exposure, largely mediated via their impact on fetal growth restriction.[11] Pregnant women who stop smoking during the first trimester have a stillbirth risk that is similar to that seen in nonsmoking women, indicating that the smoking-associated increased risk is entirely preventable.[12] Accordingly, several interventions that successfully reduce smoking during pregnancy have been shown to result in improvements in perinatal outcomes.[13]

In recent years, associations between smoke-free laws and population health have been investigated in a substantial number of quasi-experimental studies. A meta-analysis of 45 studies demonstrated considerable reductions in severe cardiovascular, cerebrovascular, and respiratory events among adults, with the most comprehensive laws generally showing the largest effect.[14] In a recent meta-analysis of 11 studies investigating the relationship between smoke-free legislation and perinatal and child outcomes, we found significant drops in preterm birth (−10.4%, 95% confidence interval (CI) –18.8;−2.0) and hospital attendance for asthma exacerbations (−10.1%, 95% CI –15.2;−5.0) following smoke-free legislation in different countries.[2] Reductions in second- and third-hand smoke (environmental tobacco smoke; ETS) exposure, smoking in the home, and smoking during pregnancy are likely to have mediated these effects.[1,15–19]

Given the well-established link between tobacco smoking and SHS exposure during pregnancy and stillbirth,[7,9–11] we were concerned about the paucity of evidence assessing the impact of smoke-free legislation on this outcome.[2] To address this, we undertook a national quasi-experimental study using linked data on >10 million singleton births and perinatal deaths registered in England between 1995 and 2011.[20] From 1 July 2007, a national law prohibiting smoking in virtually all workplaces and public places was strictly enforced, resulting in very high compliance. In an analysis that accounted for underlying temporal trends, seasonality, and key individual-level demographic features, we found introduction of the law to be associated with a −7.8% (95% CI −3.5;−11.8) reduction in stillbirths.[20] At the same time, the odds of neonatal death and low birth weight dropped by −7.6% (95% CI −3.4;−11.7) and −3.9% (95% CI −2.6;−5.1), respectively. We are unaware of other factors that may have contributed to variation in stillbirth rates over time, such as changes in registration practices and perinatal care, to have coincided with timing of the smoke-free legislation. Based on counterfactual scenarios, we estimated that approximately 250 stillbirths, 100 neonatal deaths, and 1400 cases of low birth weight had been averted since the ban each year.

The strengths of this study included the very large study population, the long observation period, and the use of rigorous methodology which followed a prespecified and registered study protocol. Its main limitations related to the quasi-experimental study design. A cluster randomized controlled trial would have been our design of choice but governments are typically not willing to implement legislation in a randomized fashion.[21] In such situations, well designed and conducted quasi-experimental studies offer a reasonably robust alternative and are regarded by the Cochrane Effective Practice and Organisation of Care Group to provide the best evidence.[21,22] Such studies are usually one of three types: controlled clinical trials, controlled before-and-after studies, or interrupted time series analyses.[2,22] The primary limitation of these methodologies lies in the nonrandom allocation of the intervention. This inherently restricts attribution of causality to the association between the intervention and outcomes and comes with an increased risk of residual confounding. Potential sources of residual confounding include demographic differences between the population exposed to the intervention and the unexposed population and temporal changes in environmental exposures and other policy changes unaccounted for in the analysis. Furthermore, a control group in the usual sense is often lacking in time-oriented analyses such as controlled before-and-after studies and ITS.

Being the first of its kind, our study is an initial attempt to address the question whether smoke-free legislation leads to a reduction in stillbirth. The available circumstantial evidence indicates that a causal link between smoke-free legislation and a decrease in stillbirths is indeed likely: SHS exposure and tobacco smoking increase the risk of stillbirth [7–9]; smoke-free legislation is followed by reductions in SHS exposure and smoking [1,15–19]; and stillbirth risk is known to normalize after smoking cessation.[12] At the same time, the study needs to be interpreted in the light of the aforementioned methodological limitations associated with its quasi-experimental nature. The continuing roll-out of smoke-free laws across the globe to protect people from ETS offers opportunities for replication of our work in other jurisdictions and can further strengthen (or undermine) confidence in our findings.

The importance of replication in research is increasingly recognized and emphasized.[23,24] When undertaking quasi-experimental studies, we would urge investigators to prespecify their approach in a study protocol which includes a detailed statistical analysis plan. This is important as in our experience, the findings of quasi-experimental studies are sensitive to aspects such as choice of the statistical modeling technique, selection of covariates, and adequate modeling of underlying time trends. Sensitivity analyses can be used to explore potential unmeasured confounding and/or the impact of different modeling choices; these are best planned a priori, but if undertaken in an exploratory post-hoc fashion, this should be made clear. Ideally, the protocol should also be registered as is now standard practice with randomized controlled trials and systematic reviews. Several trial registries also allow registration of quasi-experimental and observational studies. Registration minimizes the risk of unnecessary duplication of effort and together with publication of the protocol and analysis plan can help prevent manipulation of undesired findings, selective outcome reporting, and for negative findings to remain unpublished.

Despite considerable recent progress in global tobacco control, smoking and ETS exposure continue to pose a major health threat to adults, children, and (unborn) babies.[25] Increasing evidence associates implementation of smoke-free legislation with considerable population health benefits, supporting its recommendation by WHO as a key policy to counteract the global tobacco epidemic. However, less than one-fifth of the world’s population is currently protected by comprehensive smoke-free laws.[25] We support the undertaking of additional high-quality studies to investigate the link between such laws and a reduction in stillbirths. Their inclusion in updated evidence syntheses will help provide a much clearer picture and establish a more definite answer within the next few years.[26] Given the broader range of health benefits of smoke-free laws already established, it is essential that in parallel policy makers across the globe continue to prioritize the enactment and implementation of smoke-free legislation in areas not yet covered. A clearer understanding of the specific benefits to the health of (unborn) babies and children, who cannot themselves regulate their exposure to tobacco smoke, can provide additional impetus and catalyze this process.

Financial & competing interests disclosure

Jasper Been is supported by an Erasmus University Medical Centre Fellowship and a Netherlands Lung Foundation Junior Investigator Grant. Aziz Sheikh is supported by the Farr Institute and the Asthma UK Centre for Applied Research. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.