- October 1, 2020 from 12:00-1:00pm
- Please RSVP by emailing Lisa Maturo at Lisa.M.Maturo@christianacare.org.
- Please include your full name, email address, and institution/organization.
- We will provide instructions on obtaining CME credit for attendance.
A first step in electronic health records (EHR)-based research is characterization of patient phenotypes. However, EHR-based phenotyping is hampered by complex missing data patterns and heterogeneity in the amount and type of data that is available. As a result, EHR-derived phenotypes are error-prone and often feature exposure-dependent differential misclassification, which can bias analyses towards or away from the null. In this talk I will review approaches to EHR-based phenotyping, highlighting how missing data affect phenotype estimation. I will then discuss some results on the implications of using EHR-derived phenotypes with differential misclassification for bias and type I error of subsequent association studies using these phenotypes as outcomes. Finally, I will present some recent approaches to correcting for phenotyping error. The overall goal of this presentation is to improve awareness of phenotyping error, its implications for analyses, and available options for valid analysis of EHR-derived phenotypes.
Meet the Speaker
Dr. Rebecca Hubbard is a Professor of Biostatistics in the Department of Biostatistics, Epidemiology and Informatics at the University of Pennsylvania. She completed an MSc in Epidemiology at the University of Edinburgh and an MSc in Applied Statistics at Oxford University before obtaining her PhD in Biostatistics at the University of Washington. Her research focuses on development and application of statistical methodology for studies using data from electronic health records and medical claims data. Her methods have been applied to a broad range of research areas including health services research, cancer epidemiology, aging and dementia, and pharmacoepidemiology.
This activity has been approved for AMA PRA Category 1 Credit