Background Children born after medically assisted reproduction (MAR) are at higher risk of adverse birth outcomes than naturally conceived children. It is not known to what extent the excess risk should be attributed to harmful effects of the treatment or to pre-existing parental characteristics that confound the association.

Methods We analyse birth weight, gestational age, risk of low birth weight and risk of preterm birth among MAR and naturally conceived children using Finnish population register covering 65,634 children born between 1995-2000. First, we estimate the difference in birth outcomes by mode of conception in the general population using standard multivariate methods that control for observed factors (e.g. multiple birth, birth order and parental socio-demographic characteristics). Second, we use a sibling-comparison approach that has not been used in MAR research before. We compare MAR-conceived children to their naturally conceived siblings, thus controlling for all observed and unobserved factors shared by siblings. The latter analysis included 1,245 children.

Findings MAR-conceived children had worse scores than naturally conceived children for all outcomes, even after adjustment for observed child and parental characteristics (e.g. 60gr [95% CI: -86 to -34] lower birth weight; 2.2 percentage points [95% CI: 1·1 to 2·2] increased risk of preterm delivery). In the sibling comparison, the risk of adverse birth outcomes associated with MAR was attenuated and the adjusted relationship between MAR and adverse birth outcomes was statistically and substantively weak for all outcomes (e.g. 31gr [95% CI: -85 to 22] lower birth weight; 1.6 percentage points [95% CI: -1·3% to 4·4%] increased risk of preterm delivery).

Conclusion MAR-conceived children are a group at risk of adverse birth outcomes, but the results reassuringly indicate that this is largely because of factors other than the treatment itself.