In this nationwide population-based cohort study of more than 12 million live births in the USA, we found that maternal cigarette smoking, even at a low intensity (1–5 cigarettes per day), before pregnancy or during each trimester of pregnancy significantly increased the risk of six specific subtypes of congenital anomalies (i.e., congenital diaphragmatic hernia, gastroschisis, limb reduction defect, cleft lip with or without cleft palate, cleft palate alone, and hypospadias). Compared with persistent smokers before and throughout pregnancy, a significantly lower risk of congenital anomalies was observed for never smokers, but no reduced risk was observed for women who smoked before pregnancy and quitted during each trimester of pregnancy.
Comparisons with other studies
In the present study, we found an increased risk of overall congenital anomalies in offspring born to mothers who smoked cigarettes during pregnancy, which contrasts with a previous large population-based cohort study using data from the Danish Medical Birth Register, which did not find an overall increased risk of congenital anomalies (OR 0.98, 95% CI 0.95–1.01) [13]. It should be noted that the Danish study included a smaller sample size (n = 838,265) than ours (n = 12,144,972, especially for rare outcomes, for example, the risk estimate for omphalocele in the Danish study was based on 25 exposed cases). The Danish study also did not adjust for potentially important covariates which we have adjusted for in our analysis (e.g., race/ethnicity, maternal pre-pregnancy BMI, infant sex, eclampsia, gestational hypertension, and gestational diabetes). In addition, disparities in definitions of “overall congenital anomalies” may also contribute to the inconsistent results. However, when we used having any of the 12 congenital anomalies to define “overall congenital anomalies,” we still found an increased risk of overall congenital anomalies for maternal smoking (data not shown).
Since different anomalies can have different etiologies, defect-specific analyses may better reflect the association between maternal cigarette smoking and congenital anomalies. A meta-analysis of 6 case-control studies (4209 malformed cases and 10,646 controls) published between 2000 and 2010 revealed that maternal smoking during pregnancy significantly increased the risk of cleft lip with or without cleft palate (OR 1.48, 95% CI 1.36–1.61) [25]. The Danish study [13] mentioned above also showed that maternal smoking in pregnancy was associated with an increased risk of this anomaly (OR 1.36, 95% CI 1.18–1.56), similar to our results. We also observed an increased risk of limb reduction defects, congenital diaphragmatic hernia, and gastroschisis among women who smoked cigarettes, consistent with a meta-analysis by Hackshaw et al. [26] which identified 172 studies (173,687 malformed cases and 11,674,332 controls; conducted between 1959 and 2010; limb reduction defects: OR 1.26, 95% CI 1.15–1.39; hernia: OR 1.40, 95% CI 1.23–1.59; gastroschisis: OR 1.50, 95% CI 1.28–1.76). However, the Danish study [13] reported a marginally reduced risk of gastroschisis (OR 0.89, 95% CI 0.56–1.39; based on only 25 exposed cases) and no association with diaphragmatic hernia (OR 1.00, 95% CI 0.68–1.49; based on only 32 exposed cases). In addition, the meta-analysis by Hackshaw et al. [26] reported a reduced risk for hypospadias (OR 0.90, 95% CI 0.85–0.95), similar with the Danish study [13] which reported a marginally reduced risk of hypospadias (OR 0.94, 95% CI 0.83–1.06), inconsistent with our results (showing a significantly increased risk). Due to the large number of cases in our study, we do not believe the observed increased risk of hypospadias was entirely due to chance although the association needs to be warranted in further studies. Furthermore, in contrast to previous studies [13, 27], we did not detect a significant association between maternal cigarette smoking during pregnancy and cyanotic congenital heart disease. This result may be expected, as congenital heart defects consist of many heterogeneous subtypes that have different genetic and embryological origins. Further research should examine the effects of maternal smoking on specific subtypes of congenital heart defects.
In addition to clarifying the associations of maternal cigarette smoking with birth congenital anomalies, a novel finding of this study was that maternal cigarette smoking either before pregnancy or during each trimester of pregnancy significantly increased the risk of birth congenital anomalies, suggesting that there is no safe period of maternal smoking, which has been seldom reported in the previous studies mentioned above. It is interesting that the associations of smoking with congenital anomalies did not change significantly according to different trimesters. One speculative, possible explanation would be that the detrimental effects of maternal smoking on congenital anomalies mainly originate from an earlier period in pregnancy: morphogenesis is influenced mainly during the first days of embryogenesis, whereas later days of pregnancy involve mostly about growth of the embryo. The observed similar association between maternal smoking in the second and third trimesters of pregnancy and congenital anomalies might be due to the fact that the majority of women who smoked in early pregnancy continued to smoke cigarettes throughout pregnancy.
Another important finding of this study was that smoking exposure within three months before pregnancy, but with cessation during the pregnancy, had the similar risk of birth congenital anomalies compared with persistent smokers before and during pregnancy. One possible explanation was the detrimental action of maternal smoking on some genes—genetic or epigenetic—of the ovules (i.e., the potentially critical periods for preventing congenital anomalies seemed to be pre-conception and/or during the first days of pregnancy) [28]. Another speculative explanation was the known longer half-life of nicotine [29] and the nicotine withdrawal effect [30]. In contrast, women who never smoked before and during pregnancy were less likely to have birth congenital anomalies compared to persistent smokers throughout pregnancy. We believe that these findings have important public health implications. Many women who smoke cigarettes believe that it is acceptable to smoke before pregnancy or in the first trimester [31]. Besides, when a woman finds herself pregnant and begins prenatal care, many fetal organs may have already been formed, and smoking cessation might be too late to prevent congenital anomalies. Therefore, there is an urgent need to strengthen pre-pregnancy health services to help minimize the risk of congenital anomalies.
The third important finding of our study was that mothers who smoked only few cigarettes (1–5 cigarettes per day) had an increased risk of congenital anomalies compared with non-smokers, suggesting there is no safe level of cigarette smoking for pregnant women. Similarly, Chung, et al. [18] used the 1996 US Natality database (2207 cases and 4414 controls) found an increased risk for cleft lip with or without cleft palate among light smokers (1–10 cigarettes per day; OR 1.50, 95% CI 1.28–1.76). Another large study used Swedish register data (n = 1,413,811; 1983–1996) showed a marginally positive association between light smokers (< 10 cigarettes per day) and any malformation (OR 1.02, 95% CI 0.98–1.05) [32]. However, the two studies used smoking 1–9 cigarettes per day to define light smoking. The Danish study [13] used smoking 1–5 cigarettes per day to define light smoking and showed that infants born to mothers who smoked a low dose of cigarettes per day were at a marginally reduced risk of major malformations (OR 0.96, 95% CI 0.92–1.01). These previous studies also did not distinguish the effect of intensity of maternal smoking in different periods (before or during different trimesters of pregnancy) on congenital anomalies, while our study showed the similar effect highlighting maternal smoking even at a low intensity during any period increased the risk of congenital anomalies. It should be noted that our study did not find clear dose-response associations between maternal smoking before or during the pregnancy and birth congenital anomalies, similar to a previous study on maternal smoking and congenital heart defects by Alverson et al. (2525 malformations and 3435 controls) [14]. Another older study showed no dose-response association between maternal smoking and cleft lip with or without cleft palate [33]. The possible explanations included the competing risks (e.g., abortion and stillbirth) caused by high doses of consumed cigarettes [34], which might bias the relationship with congenital anomalies toward null. In addition, other unmeasured confounding factors might have also biased our findings.
Finally, it should be mentioned that magnitudes of the positive association between maternal smoking and congenital anomalies in our study were somewhat higher than those in previous studies mentioned above. We speculate that this may be due to variability in exposure assessment (i.e., cigarette smoking) and case ascertainment (i.e., congenital anomalies). In addition, many previous studies were based on relatively small number of cases on congenital anomalies and with adjustment for a limited number of potential confounders. Further large well-designed prospective cohort studies are needed to warrant the observed association and to examine the possible explanations.
Strengths and limitations
The main strengths of this study include the population-based cohort design, the large sample size of over 12 million live births, the detailed examination of the associations between the timing and intensity of maternal cigarette smoking and many congenital anomalies subtypes considered (including rare subtypes), the adjustment for several (but not all possible) covariates, and the sensitivity analyses used to confirm our findings. However, this study also has several limitations. First, there were limitations to use birth certificate data for assessing the association with birth defects. Data on pregnancy complications, and comorbid conditions tended to be underreported on the birth certificates [35]. However, a previous study used data from the West Virginia (1990–2009) to assess the effects of misclassification bias on reported congenital anomalies based on birth certificate data, and the results showed that specified birth defects collected using checked-box format with definite criteria on the 2003 version birth certificate showed consistent patterns over time [36]. In addition, several other studies have been performed to assess the validity of birth certificate data for effect estimation [22, 37] and the results suggest that birth certificate data may be useful for exploratory studies assessing the association between birth congenital anomalies and some risk factors. Second, maternal cigarette smoking was self-reported and recall bias might have occurred. However, a previous study showed that self-reported number of cigarettes were highly correlated with objectively measured cotinine during pregnancy [38]. Moreover, pregnant women were unwilling to report smoking if they had illness conditions around the time of delivery in the hospital. This also could cause some bias or misclassification. Third, information on non-daily cigarette smoking (< 1 cigarette per day) before or during pregnancy was unavailable on the birth certificate. Fourth, for “maternal pre-pregnancy smoking,” the NVSS only collected information on smoking within 3 months before pregnancy. Thus, we could not differentiate women who never smoked from those who did within 3 months before pregnancy. We then also could not compare the risk of congenital anomalies for smoking cessation before pregnancy vs. persist smoking, and the former period is the potentially critical period for preventing congenital anomalies. Fifth, the NVSS only collected data on congenital anomalies identified at birth. However, some types of congenital anomalies could not be identified at that time, and usually appear in childhood or even adulthood. This may cause outcome misclassification (i.e., the infants who did not have congenital anomalies at birth but diagnosed in later life were treated as not having congenital anomalies in this study). Furthermore, we only included live births and did not account for congenital anomalies in aborted fetuses and stillbirths, which may lead to selection bias (i.e., newborns with malformations among maternal smokers may be underrepresented in a study of live births). Finally, although we have adjusted for a range of potential confounders, we could not completely rule out the possibility of residual confounding or confounding from unmeasured confounders. For example, one important potential confounder that we did not adjust for was second-hand smoke exposure, which has been shown to be nearly as deleterious as active smoking to the developing fetus [39]. Another important potential confounder was alcohol consumption as drinking was associated with both smoking and embryo malformations [40]. In addition, we only adjusted for maternal race/ethnicity without considering that of the infant. It is possible that the infant may have a different racial make-up than the mother, especially if one race/ethnicity was more susceptible to the effects of smoking but only race/ethnicity of mother is considered, this could cause some bias or misclassification.