CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1499
Title: Investigating anticipated and un-anticipated effects of drugs in large biobanks Authors:  Ioanna Tzoulaki - Imperial College London (United Kingdom) [presenting]
Abstract: Large biobank studies with linked electronic health records (EHR) offer unprecedented opportunities to study the effects of drugs in a systematic way in order to uncover anticipated and anticipated effects associated with drug use. We will present two examples of using multidimensional longitudinal EHR data to examine repurposing drugs in a prospective design. On one instance, a cohort study using the Clinical Practice Research Datalink was designed to investigate the association between use of metformin compared with other antidiabetes medications and cancer risk by emulating an intention-to-treat analysis as in a trial. Cox proportional hazards models were used to estimate multivariable-adjusted hazard ratios (HR) and 95\% CI. In another example, we use human genetic variation as a proxy and a method of more accurately predicting the physiologic effects of drugs. Specifically, we apply a systematic approach to examine a wide range of phenotypes across the human phenome in relation to genetic variants mimicking the effects of drugs of interest (Phenome Wide Association Study). This approach is now tractable given the extensive phenotypic information available in large biobanks (e.g.UK Biobank) with linkage to electronic health records and measurements of quantitative phenotypes at baseline. We present an example based on genetic variants on the HMGR gene which mimics the effect of statins and show the phenome wide association study related to those variants.