In this large population-based cohort of 12 150 535 mother-infant pairs in the USA, we found that maternal cigarette smoking before or during pregnancy increased the risk of infant composite SNM and individual components compared with maternal non-smoking in the considered specific period, even at a low dose of 1–2 cigarettes/day, suggesting there is no safe period and no safe level of cigarette smoking shortly before or during pregnancy. Compared with never smokers before and throughout pregnancy, mothers who stopped smoking in each trimester still had a higher risk of infant composite SNM.

Our findings expand the available evidence about adverse effects of maternal cigarette smoking before and during pregnancy on infant SNM. We found an increased risk of composite SNM in infants born to women who smoked cigarettes during pregnancy, inconsistent with a previous retrospective cohort study in Australia,11 which reported no increased risk of overall SNM among women who smoked cigarettes during pregnancy (late preterm: OR 0.77, 95% CI 0.46 to 1.29; early term: OR 1.26, 95% CI 0.79 to 2.00). However, the Australia study included a smaller sample size (n=6243) than ours (recognising that large sample sizes are needed to study rare events such as SNM). Besides, in the Australia study the associations were estimated with adjustment for gestational age only, and participants were limited to late preterm and early term infants.11 Finally, it is important to acknowledge that there is no ‘gold standard’ definition for composite SNM and therefore comparisons between studies should be made with caution.

With respect to individual SNM components, NICU admission has been the most frequently studied in relation to maternal cigarette smoking, but the findings in several previous studies have also been inconclusive.9 10 22 A previous study using data from the NHS Foundation Trust in the UK (2017–2018; n=4465) showed no significant association between maternal cigarette smoking during pregnancy and infant admission to NICU.10 In contrast, a nationwide population-based cohort study using data from the Finnish Medical Birth Register from 1991 to 2010 (n=1 164 953) showed that maternal smoking cessation in the first trimester (ie, smoking exposure during early pregnancy) was associated with an increased risk of infant NICU admission (OR 1.27, 95% CI 1.25 to 1.30).9 Of note, the above mentioned studies included relatively small samples10 22 or included adjustment for no covariate10 22 or only a few covariates (ie, maternal age, parity, sex, pre-eclampsia, and gestational diabetes).9 Our study was based on the largest sample size to date and showed a significant association between maternal cigarette smoking before and during pregnancy and infant NICU admission after adjusting for a fairly large number of potential confounders (n=12). One possible explanation is that maternal cigarette smoking can increase the risk of intrauterine growth restriction,7 8 23 which in turn will increase the risk of infant NICU admission after delivery.24

We also observed a significant association between maternal cigarette smoking before and during pregnancy and seizure or serious neurologic dysfunction. One study used the same datasets (2013–2017) as the one used in this study and also reported that maternal smoking during pregnancy was associated with neonatal seizures (OR 1.50, 95% CI 1.33 to 1.69) at term.25 However, the study only included term births, whereas preterm births were at a higher risk of neonatal seizures.26 27 Several previous studies focused on febrile seizures in childhood (3 months to 5 years) showed an increased risk for children whose mothers smoked cigarettes during pregnancy,28–30 which also support our findings. A biological mechanism may be that maternal cigarette smoking during pregnancy can alter brain structures of the fetus, including decreasing cortical thickness in the perisylvian and lateral occipital cortices, decreasing volume in the anterior cingulate, and increasing volume in the frontal regions.31

In addition, we observed an increased risk of suspected neonatal sepsis in offspring born to women who smoked cigarettes before or during pregnancy, which has been seldom reported in previous studies. This may indicate a potential link between maternal cigarette smoking and maternal immune response or placental function. A large cohort of infants born in Upstate New York reported that maternal cigarette smoking throughout pregnancy could alter cytokine signalling (eg, interleukin-8 (IL-8)) in newborns, which is a pro-inflammatory chemokine linked to various pathological conditions, such as impaired lung function and respiratory disorders.32 Other novel findings of our study also include the associations of maternal cigarette smoking with assisted ventilation immediately following delivery, assisted ventilation for more than 6 hours, and surfactant replacement therapy. Our results suggest that maternal cigarette smoking before and during pregnancy can potentially serve as a marker of an increased risk of these neonatal morbidities.

It is worth mentioning that the previous studies above did not distinguish the effects of the timing of maternal cigarette smoking (before pregnancy or in different trimesters) on infant SNM. They also did not explore in detail the associations between the intensity of maternal cigarette smoking and infant SNM. Our study showed that maternal cigarette smoking before pregnancy or in any trimester was associated with an increased risk of infant SNM, even at a low smoking intensity of 1–2 cigarettes per day. Another important finding of our study is that for mothers who smoked before pregnancy, smoking cessation during subsequent trimesters did not substantially decrease the risk of infant SNM. However, for a few smoking cessation subgroups (ie, groups 6, 10, and 11), the associations were only marginally (not statistically) significant, which may be attributed to the limited statistical power in these subgroups. Several large cohort studies5 20 21 33 focusing on other neonatal outcomes (eg, preterm birth and congenital anomalies) also showed an elevated risk among women who smoked during pregnancy regardless of the timing and doses consumed. Overall, these data suggest that there is no safe period and no safe level of cigarettes consumed shortly before or during pregnancy.

Our findings re-emphasise the need to prevent smoking initiation for non-smokers and to promote smoking cessation for smokers. Interventions should also highlight the adverse effects of light smoking for pregnant women. Actually, numerous efforts have been done to promote smoking cessation among pregnant women in the USA. For example, in 2015, the US Preventive Services Task Force issued a clinical guideline on interventions for smoking cessation in adults, including pregnant women.34 The guideline recommends that obstetricians should routinely ask pregnant women regarding smoking, advise those who smoke to stop immediately, and offer evidence-based cessation interventions.34 Yet, the prevalence of maternal smoking before and during pregnancy remained high in our study. Therefore, it is imperative that strong tobacco control policies, such as strict national tobacco control laws and higher adherence to the WHO Framework Convention on Tobacco Control, are put into place in the USA.

The major strengths of this study are that we used the largest sample size to date with over 12 million mother–infant pairs, which allowed us to examine thoroughly the associations of the timing (per trimester) and intensity (number of cigarettes per day) of maternal cigarette smoking with composite SNM and individual components (given that SNM is a relatively rare outcome), and that we could adjust for a fairly large number of potential confounders.

However, a few limitations exist. First, information on maternal cigarette smoking was self-reported by the mothers, which may introduce information bias. However, a previous study showed that self-reported maternal smoking was well correlated with objectively measured cotinine during pregnancy.35 Second, for ‘maternal prepregnancy smoking’, the NVSS only collected data during the 3 months before pregnancy, making it impossible to distinguish between women who never smoked and those who ceased smoking before pregnancy. Third, the NVSS did not provide information on women’s exposure to secondhand smoke, hindering us from examining the association between maternal secondhand smoke exposure during pregnancy and SNM in offspring. Finally, although we have adjusted for many potential confounders in this study, we cannot completely rule out residual confounding from unmeasured or unknown confounders.