Abstract

Background Studies have consistently shown that pregnancy smoking is associated with twice the risk of hyperactivity/inattention problems in the offspring. An association of this magnitude may indicate behavioural difficulties as one of the most important health effects related to smoking during pregnancy. However, social and genetic confounders may fully or partially account for these findings.

Methods A cohort including all singletons born in Finland from 1 January 1987 through 31 December 2001 was followed until 1 January 2006 based on linkage of national registers. Data were available for 97% (N = 868 449) of the population. We followed singleton children of smoking and non-smoking mothers until they had an International Classification of Diseases, 10th revision, diagnosis of hyperkinetic disorder (HKD) or to the end of the observation period. We used sibling-matched Cox regression analyses to control for social and genetic confounding.

Results We found a much smaller association between exposure to maternal smoking during pregnancy and risk of HKD in children using the sibling-matched analysis [hazards ratio (HR) = 1.20, 95% confidence interval (CI) 0.97–1.49] than was observed in the entire cohort (HR 2.01, 95% CI 1.90–2.12).

Conclusions Our findings suggest that the strong association found in previous studies may be due to time-stable familial factors, such as environmental and genetic factors. If smoking is a causal factor, the effect is small and less important than what the previous studies indicate.