Familial confounding affected the associations between maternal smoking during pregnancy and offspring speech and language, scholastic and coordination disorders

Aim This study examined the associations between prenatal smoking and speech and language, scholastic, coordination and mixed developmental disorders in offspring, using sibling and population controls. Methods National Finnish registers were used to identify all 690 654 singletons born between 1996 and 2007 and any cases diagnosed with speech and language, scholastic, coordination and mixed developmental disorders by the end of 2012. Cases were compared to population controls, biological full-siblings and maternal half-siblings born during the same period. Conditional logistic regression was used to assess any associations between smoking during pregnancy and the selected developmental disorders. Results Prenatal smoking was higher in the mothers of the 27 297 cases (21.7%) than the 99 876 population controls (14.5%). The adjusted odds ratio for smoking throughout pregnancy, and any diagnosis of speech and language, scholastic, coordination or mixed developmental disorders, was 1.29 (95% confidence interval 1.24-1.34). However, when we compared a subsample of 15 406 cases and their 20 657 siblings, the association was no longer statistically significant (odds ratio 1.09, 95% confidence interval 0.98-1.21). Conclusion The sibling comparisons suggested that the associations between prenatal smoking and speech and language, scholastic, coordination and mixed developmental disorders were confounded by familial factors shared by differentially exposed siblings.