|
Activity Number:
|
83
- Contributed Poster Presentations: Health Policy Statistics Section
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Health Policy Statistics Section
|
|
Abstract #312412
|
|
|
Title:
|
Provider Profiling: Grouping Providers Treating Similar Populations of Patients
|
|
Author(s):
|
Gabriella Silva* and Roee Gutman and Amal Trivedi
|
|
Companies:
|
Brown University and Brown University and Brown University
|
|
Keywords:
|
provider profiling;
quality of care;
clustering;
Bayesian mixture model;
risk-adjustment
|
|
Abstract:
|
Standard approaches to evaluating risk-adjusted outcomes across providers often use hierarchical logistic regression models to adjust for differences in patient characteristics and include all providers, even when they treat highly heterogeneous patient populations. These estimates may be inaccurate when there are substantial differences in the characteristics of patients treated by each provider. To address this limitation, we developed an alternative approach to profile providers that uses a hierarchical Bayesian mixture model to identify similar clusters of providers based on the characteristics of their patients. We then estimate provider performance within each cluster. We compared the performance of the traditional profiling approach to our proposed clustering method using patient-level data from nursing homes in the Northeast and simulated binary outcome data. Our simulation showed that, in the presence of model misspecifications and limited overlap in patient characteristics, the provider-level performance estimates obtained using our approach were more robust to model misspecifications than the traditional approach and more accurately identified outlying nursing homes.
|
Authors who are presenting talks have a * after their name.
