Bias identification and reduction techniques for large electronic medical record datasets

We study how differences in registration and healthcare use can skew research findings and how we can correct for them in a simple way. We work with pseudonymised Intego data; individuals cannot be identified.