|
Activity Number:
|
473
- Advances in Measuring Health Care Quality and Disparities
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Health Policy Statistics Section
|
|
Abstract #330575
|
Presentation
|
|
Title:
|
Measuring Latent Quality of Medical Groups Using IRT Models Accounting for Missing Data: Can We Get Reliable Estimates of Quality After All?
|
|
Author(s):
|
Amelia M Haviland* and Denis Agniel and Cheryl Damberg and Paul Shekelle
|
|
Companies:
|
Carnegie Mellon University - Heinz College and RAND Corporation and RAND Corporation and RAND Corporation
|
|
Keywords:
|
item response theory;
missing data;
reliability;
health care quality
|
|
Abstract:
|
Assessing quality of health care provided by medical groups is complicated by multiple but inconsistently reported measures of quality across groups. We implemented a Bayesian item response theory model using 16 quality measures with varying missingness rates for all medical groups in Minnesota. We find substantial differences in the discrimination values of the measures with many having low power to discriminate latent quality differences between groups. While the distributions of latent quality for groups frequently overlap, the rank of each medical group across all MCMC runs was quite stable. After accounting for the uncertainty due to missing values using multiple imputation, the ranks remain fairly stable for medical groups in the tails of the latent quality distribution but not for the majority of medical groups in the middle of the distribution. While reliability of quality measures have been found to be poor for individual providers, this research suggests that higher reliability is possible at the medical group level.
|
Authors who are presenting talks have a * after their name.
