Exposure to maternal smoking during pregnancy and risk of childhood cancer: a study using the Danish national registers

Purpose The relation between maternal smoking during pregnancy and childhood cancer in the offspring remains uncertain. This paper uses Danish national registers, which have collected data prospectively on smoking and cancer, to investigate the association. Methods Smoking during pregnancy was ascertained from maternal self-reported data in the Danish National Patient Register. Index children were followed up from birth until the first of the following events: cancer diagnosis, death, emigration, day before 15th birthday, or end of follow-up. Smoking during pregnancy was considered as a binary variable (no smoking in pregnancy and smoking in pregnancy) and by amounts smoked (no smoking, cessation during pregnancy, a parts per thousand currency sign5, 6-10, or a parts per thousand yen11 cigarettes/day). Results Of the 801,867 children included in the study, 20 % were exposed to maternal smoking during pregnancy. Overall, the hazard ratio (HR) for childhood cancer for the exposed compared to the non-exposed was 0.91 [95 % confidence interval (CI) 0.78, 1.07]. Stratification by number of cigarettes also gave statistically nonsignificant inverse associations. There was a statistically significant increased risk of childhood cancer among children whose mothers reported smoking cessation in pregnancy (HR 1.46; 95 % CI 1.01, 2.10). Regarding specific cancer sites, maternal smoking in pregnancy was positively associated with the risk of eye cancers in childhood. Conclusions Our results do not provide evidence for an association between maternal smoking in pregnancy and childhood cancer overall. An increased risk of childhood cancer was seen for children whose mothers reported smoking cessation in pregnancy. Future research could employ biomarkers, such as cotinine, to validate maternal smoking status recorded in registers as, even if collected prospectively, this self-reported variable may be subject to reporting bias.