Shifting the traditional model of analysing downloaded electronic health records to the novel open-source analytics OpenSAFELY platform
Speaker: Angel Wong PhD, OpenSafely, a project of the Oxford Data Lab (UK)
Abstract
Much electronic health record clinical research involves a traditional model of running analyses on large datasets extracted from data providers to a local machine which may carry data security risk. Other limitations of this model are lack of research transparency and out-of-date data. Working on behalf of the National Health Service, England, we have rapidly developed a new open-source analytics OpenSAFELY platform to address urgent COVID-19 research questions. OpenSAFELY uses a model that runs all analyses within the secure data centre so that patient data are never transferred outside of the data centre. All outputs are restricted to aggregate data with small number suppression, released to GitHub. As of 18 Jan 2021, we have published 3 pharmacoepidemiologic studies including examining the role of inhaled corticosteroids, hydroxychloroquine and nonsteroidal anti-inflammatory drugs on COVID-19 mortality; and completed 5 other COVID-19 related studies. We openly shared >200 codelists and analytic codes on GitHub. This new model balances the need for protecting data security and promoting openness in research, whilst facilitating timely evidence generation.
