Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 322 - Novel Statistical Methods for Current Health Policy Issues
Type: Contributed
Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
Sponsor: Health Policy Statistics Section
Abstract #313021
Title: Hospital-Specific Template Matching for Benchmarking Performance in the Veterans Affairs Health System
Author(s): Brenda Vincent* and Andrew Ryan and Hallie Prescott
Companies: and University of Michigan and University of Michigan
Keywords: template matching; hospital quality; electronic health record; matching; power; simulation

Hospital-specific template matching is an approach to hospital benchmarking that compares performance between peer hospitals for similar patients. The number of comparator hospitals may differ for each hospital, yielding potentially uneven power. Our objective was to assess the statistical power of hospital-specific template matching to detect mortality differences within the Veterans Affairs Health System (VA). We used VA EHR data for 556,266 hospitalizations at 122 hospitals. We randomly selected a template of 500 hospitalizations from each hospital. We matched patients at the remaining 121 hospitals on demographics, diagnoses, and acute physiology to each hospital template. Hospitals were compared on mortality rates in their template to matched hospitalizations from comparator hospitals. We used simulation to quantify the power to detect a variety of standardized mortality ratios (SMR) for hospitals with few, average, or many comparison hospitals. Power to detect an SMR of 2 ranged from 66% for a hospital with few comparisons to 75% for a hospital with many comparisons. Although the number of matches varied substantially (5 to 60), the impact to power was relatively minor.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program