Smoking Ban and Pregnancy Complications: New Evidence

Secondhand smoke (SHS) exposure is a group I carcinogen, and there is no risk-free safe level of SHS exposure. ¹ There is also substantial evidence that both direct (firsthand) and maternal exposure to SHS increase the risk of pregnancy complications. ^2,3 Maternal smoking during pregnancy is an important modifiable risk factor and has both immediate and long-term health consequences. For instance, mothers who smoked during pregnancy have a 2-fold increased risk of having low birthweight (LBW) babies compared to nonsmoking mothers. ^4,5 Unfortunately, many pregnant women do smoke; for example, 1 in 5 pregnant women continues smoking in Ireland. ⁶

Tobacco use cessation is the single most cost-saving prevention tool to reduce premature deaths, morbidity, and disability. Adverse birth outcomes, such as LBW and preterm births, are important risk factors for perinatal morbidity and mortality and have also been associated with an increased risk of cardiovascular disease later in adulthood. ^7,8 Tobacco control has both a health and an economic benefit and a potential downstream consequence of a sustainable health development of a society or a nation. In 2006, the US Institute of Medicine (IOM) reported that the high rate of premature births in the United States is costing at least $26 billion a year. ⁹ The good news is that the existing tobacco control strategies do work effectively. ¹⁰ One such strategy is the comprehensive smoke-free policy.

The population health impact of comprehensive smoke-free policies is increasingly contributing to changing social norms of a society. Only 5% of the world population now live in a smoke-free society, ¹⁰ but in settings where comprehensive smoke-free policies are strictly enforced, both health and social gains are observed. ^11,12 The strongest evidence to date in support of a positive health gain attributable to smoke-free legislation is the consistent decline in coronary heart diseases. ¹³ However, the effect of antismoking policies on pregnant women is also of considerable interest from both a population health and an economic perspective. In this issue, Page et al. ¹⁴ report the first evidence in the United States showing how a population-level intervention, such as a smoke-free policy, could translate into improvements at the individual level through reductions in preterm birth risks.

The Page et al. study ¹⁴ employed a quasi-experimental design exploiting a natural experiment, comparing two isolated cities (El Paso, the control city, and Pueblo, the city with a smoking ban) in Colorado using historical controls. A significant reduction of 38% in the odds of maternal smoking following multivariable adjustment in Pueblo compared to El Paso County is consistent with similar declines in Ireland, ⁶ Italy, ¹⁵ and Scotland. ¹⁶ An earlier study in Ireland based on self-reported smoking status determined a marginal decline of 12% 1 year after the Irish workplace comprehensive smoke-free policy, which was introduced in March 2004. ⁶ An Italian study, a retrospective study, showed a decline in maternal smoking during the first trimester, but the decrease did not reach statistical significance, and the relapse rates were high after pregnancy. ¹⁵ The most recent study, in Scotland, is a retrospective cohort study showing a significant decline in maternal current smoking rates from 24.6% to 18.8% after the ban, which was introduced in March 2006. ¹⁶ It would be interesting if such falls in maternal smoking rates showed any variations across different socioeconomic groups, but no applicable information was available in any of the published studies to date, although deprivation levels were addressed for model outputs in the Scottish study. ¹⁶

Preterm births (<37 weeks' gestation) are a huge public health and economic burden. ⁹ The Colorado study demonstrated a 23% reduction in the odds of preterm births after the citywide smoking ban was introduced in Pueblo in July 2003. ¹⁴ Such estimates are in agreement with the earlier study in Ireland on preterm births. ⁶ However, the recent Scottish study showed a far lower estimate of around 12% overall reductions in preterm births after the Scottish smoke-free legislation was introduced. ¹⁶ The same study had differentiated preterm births into elective preterm births and spontaneous preterm labors, ¹⁶ which are presumed to have different etiologies. There is also evidence suggesting that there is an increase in elective preterm deliveries and rising cesarean section rates over the years. ¹⁷ Therefore, it is essential that underlying temporal trends are accounted for in such computations, including any changes in recent obstetric practices. In general, the studies to date have included only live singleton births for the purpose of analyses. ^6,14,16

Another interesting observation in the Colorado study is the increased odds of LBW (<2500 g) babies after the citywide smoking ban was introduced. ¹⁴ A similar observation but of a higher magnitude was noted in the Irish study. ⁶ No potential explanations for such observations are convincing, but an increasingly obese cohort of mothers giving birth to relatively large babies in recent years may be a pointer to that direction. ¹⁸ Unfortunately, such information was not available for the Page et al. study, ¹⁴ and that study did not estimate mean baby birth weights of pregnant women. The Irish study reported an overall decrease in mean birth weights 1 year after the Irish smoke-free policy, ⁶ although the decline in mean baby birth weight of pregnant women after the smoking ban reached statistical significance only among currently smoking mothers in the Irish study. Such dichotomy of smoking status was apparently not feasible in the Colorado study. ¹⁴ The Scottish study estimated very small and small-for-gestational age (SGA) births based on birth weight centiles. ¹⁶ Interestingly, the same study showed a significant 4.5% reduction in SGA births overall, reaching almost 8% reductions among currently smoking mothers. ¹⁶ A recent study in Ireland showed a similar reduction in SGA births accounting for the underlying trends before the Irish smoke-free policy was introduced. ¹⁹

The strength of the Page et al. study ¹⁴ is the adoption of a historical control group, thus accounting for secular trends. The investigators have highlighted the comparability of the two isolated cities, and in the same vein, they have cautioned about a possible contagion in the study design. It is important to note in Table 1 of that study, however, that neither city has statistically significant differences in the proportions of preterm births or SGA births, ¹⁴ indicating broad similarities in the baseline characteristics between the two populations studied, especially in relation to the outcome measured. An objective ascertainment of the gestational age estimation based on ultrasonographic findings in the Colorado study is clearly reassuring for the outcome measured. An additional strength of that study is the adjustment of all potential confounders available to the study investigators a priori, although residual confounding is always a methodologic issue for secondary data analyses because of unknown/unmeasured factors. ¹⁴ Nevertheless, area-based factors are equally important as individual-level factors for additional insights into contextual social deterministic epidemiologic patterns, especially the impact of deprivation status.

Cross-sectional ecologic study designs have inherent methodologic limitations, such as ecologic fallacy and the lack of causal inferences. Page et al. ¹⁴ have delineated clearly the underlying study limitations. Examples include the lack of information on maternal SHS levels (including workplace and household exposure levels, as well as in utero exposure), self-reporting of smoking status, no dichotomy of smoking status into smoking and nonsmoking mothers, the lack of statistical power for some specific subgroup analyses, details on smoking topography including smoking cessation attempts and relapse rates during pregnancy, and a possible contagion effect in the study design. ¹⁴ Such methodologic limitations may have introduced bias in some form or other generally tending toward a null effect. The Colorado study, ¹⁴ unlike the Irish and the Scottish studies, has no information on smoking paradox, where it has been hypothesized that maternal smoking is protective of preeclampsia or pregnancy-induced hypertension (PIH). ^20,21 It is interesting to note that none of the four studies related to smoking ban and pregnancy complications published to date have detailed information on gestational maternal weights or on maternal body mass index patterns.

The underlying mechanistic pathways of such complex interventions are plausible but are intriguing, especially the apparent increase in LBW births despite smoke-free policies. ^6,14 A retrospective study in the state of Massachusetts showed no decline in LBW rates despite a consistent fall in maternal smoking rates. ²² Future studies of similar nature are needed and should address the methodologic limitations akin to cross-sectional ecologic designs or retrospective secondary data analytic study designs. Nevertheless, the growing evidence in support of the positive population health gains of smoke-free policies for a vulnerable population, such as pregnant women and pregnancy complications, is both encouraging and crucial. Analyses of the cost-effectiveness of similar interventions should lend further support to public health policy makers and advocates. In summary, any positive health intervention for pregnant women is a win–win public health advocacy tool.