|
Activity Number:
|
30
- Missing Data and Measurement Error
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 28, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #304369
|
Presentation
|
|
Title:
|
Approaches to Bias Correction When Using Propensity Scores Estimated from Imperfect EHR-Derived Covariates
|
|
Author(s):
|
Joanna Harton* and Nandita Mitra and Rebecca Hubbard
|
|
Companies:
|
University of Pennsylvania and University of Pennsylvania and University of Pennsylvania
|
|
Keywords:
|
EHR;
measurement error;
propensity score;
bias;
imputation;
error in variables
|
|
Abstract:
|
As use of electronic health records (EHR) to estimate treatment effects has become widespread, concern about bias introduced by error in EHR-derived covariates has also grown. While methods exist to address measurement error in covariates, little prior research has investigated the implications of using propensity scores to obtain average causal treatment effect when the propensity scores are constructed from a combination of accurate and mismeasured covariates. We use simulation studies to compare the performance of alternative methods to correct for error resulting from inclusion of imperfect covariates in a propensity score when a validation subset for the mismeasured covariate is available. We then apply these methods to a real-world EHR-based comparative effectiveness study of alternative treatments for metastatic urothelial cancer. We compare several methods across a range of scenarios featuring variation in disease prevalence, strength of confounding and strength of treatment effect. This head-to-head comparison of measurement error correction methods in the context of a propensity score-adjusted analysis fills an important niche in the literature on research using EHR data.
|
Authors who are presenting talks have a * after their name.
