Are Birth Weight and Educational Performances Causally Related?

Background. Politicians and policy makers have tried many ways to make life more equitable. To do so they need to have insights in which factors influence for example education, and therefore influence wages and lifestyle. Some of the factors policies have been made upon that were considered a causality, have later been proven to be only an association biased by other factors underlying this coherence. One of the factors where opinions of researchers are ambiguous is birth weight. There have been different views, whether birth weight has a positive causal effect on educational performances or being nothing more than a spurious biased association. A lot of the effect attributed to the difference in birth weight has been nullified by controlling for socioeconomic factors, parental education and maternal effects. Methods. In this paper we try to clarify under which circumstances birth weight has a causal impact on school performances using an informative sensitivity analysis on a GWAS sample of the Avon Longitudinal Study of Parents and Children (ALSPAC). We do so by using the sensitivity analysis proposed by Bowden, Davey, Smith & Burgess (2015) to account for possible imperfection of the exclusion restriction, to still be able to draw conclusions about using a polygenic risk score of birth weight as instrument to identify the causal impact of birth weight on school grades and IQ test scores. Results. We obtain a strong positive significant association of birth weight and school performances, which too a certain extent holds by adjusting for socioeconomic factors, parental education and maternal effects. Some of the outcome measures are identified to be positively causal impacted by birth weight using the polygenic risk score of birth weight as instrument, in case the exclusion restriction would hold. But by performing a sensitivity analysis we can see that a small violation of the exclusion restriction neglect the significant positive results. Conclusions. Concluding from our data, birth weight is not very likely to have a causal impact on educational performances, since a small violation of the exclusion restriction would neglect the significant results we obtained from the IV regression. Although SNPs and polygenic risk scores are assumed to be valid and strong instruments in most cases we cannot be sure that the exclusion restriction is satisfied. We could identify to what 'imperfection' a significant effect will hold and give a brief insight in the associations birth weight has on different school topics, and how these associations change over time.