|
Activity Number:
|
291
- Using EHRs to Run Pragmatic Trials: Opportunities and Challenges
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 3:30 PM to 5:20 PM
|
|
Sponsor:
|
Health Policy Statistics Section
|
|
Abstract #316643
|
|
|
Title:
|
Patient Recruitment and Efficient Estimation Leveraging Electronic Health Records
|
|
Author(s):
|
Guanghao Zhang and Lauren Beesley and Bhramar Mukherjee and Xu Shi*
|
|
Companies:
|
University of Michigan and Department of Biostatistics, University of Michigan and University of Michigan and Department of Biostatistics, University of Michigan
|
|
Keywords:
|
optimal sampling design;
electronic health records;
efficient estimation;
selection bias
|
|
Abstract:
|
Electronic health records (EHRs) are increasingly seen as a cost-effective resource for targeted patient recruitment into a research study. When the outcome is expensive to measure, inexpensive auxiliary variables available in EHR can be leveraged to recruit patients selectively and efficiently. We have recently proposed a two-phase sampling mechanism where sampling probability is an optimal function of the auxiliary covariates and measurement costs, while accounting for potential selection bias in the EHR sample. We focused on estimation of mean outcome and our proposed method was shown to improve efficiency over random sampling. In this talk, we present optimal sampling design and efficient estimation of the average treatment effect. We derive the two-phase sampling mechanism that optimizes efficiency under a given budget, and we provide a simple power analysis for detecting a clinically meaningful treatment effect. We conduct extensive simulation studies to assess the finite sample performance of our proposed method. Finally, we illustrate our method with an application to patient recruitment from the Michigan Genomics Initiative.
|
Authors who are presenting talks have a * after their name.
