Medicines in pregnancy: predicting harms and benefits of antenatal corticosteroids (360G-Wellcome-209560_Z_17_Z)

50% of pregnant women are prescribed drugs in pregnancy, but there are significant knowledge gaps about the safety, optimum dosage and long-term effects of medications in pregnancy. I will use innovative methodology to create new, big datasets for the study of medicines in pregnancy. Within this Fellowship I will focus on antenatal corticosteroid treatment (ACT), as there is a pressing need to establish its safety. I will extract clinical data about ACT from unstructured text entries in electronic health records (NHS Lothian) using GATE natural language processing software. These data will be linked to coded outcomes in maternity/neonatal/perinatal mortality/child- health/education databases. I have secured agreement from international researchers to form a consortium for the study of pregnancy treatments. They will contribute ACT data from birth registries, population cohort studies and trials. I will perform individual patient data level meta-analyses of the above datasets (1.5 million women) for comprehensive study of the effects of ACT on perinatal mortality and childhood neurodevelopment and develop predictive models for harms and benefits. The results will improve ACT prescribing, which is likely to reduce morbidity and mortality. My longer-term vision is to apply these techniques more widely to allow pharmacoepidemiological studies of other pregnancy medications.