Activity Number:
|
428
- Recent Advances in Statistical Methods for Healthcare Provider Profiling
|
Type:
|
Invited
|
Date/Time:
|
Thursday, August 6, 2020 : 10:00 AM to 11:50 AM
|
Sponsor:
|
ENAR
|
Abstract #309590
|
|
Title:
|
Improving the Identification of Extreme Clusters Using Multilevel Data
|
Author(s):
|
John Neuhaus* and Charles McCulloch
|
Companies:
|
University of California, San Francisco and University of California, San Francisco
|
Keywords:
|
Prediction;
Random effects;
Mixed models
|
Abstract:
|
Predicted random effects are widely used to evaluate the performance of and rank clusters such as patients and hospitals using longitudinal and multilevel data. Previous work showed that predicted random effects generally perform better than simply using fixed effects and are optimal in the sense of minimum mean square error of prediction under certain assumptions. However, predicted random effects are often used to identify extreme values such as poorly performing hospitals and the performance of standard best predicted values has not been systematically evaluated in this setting. In this talk, we show that standard best predicted values can be outperformed by methods derived under alternative assumptions. We present several new methods to identify extreme clusters and evaluate their performance in simulation studies. Data from a longitudinal study of blood lipids motivate our work and illustrate the findings.
|
Authors who are presenting talks have a * after their name.