Comparing Quality of Care Across Hospitals: Template Matching Versus Conventional Approaches

• Activity Number: 503 - SPAAC Poster Competition
• Type: Topic Contributed
• Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Scientific and Public Affairs Advisory Committee
• Author(s): Brenda Vincent* and Hallie Prescott
• Companies: VA Center for Clinical Management Research and VA Center for Clinical Management Research
• Keywords: template matching ; hierarchical models ; hospital case-mix ; nested models ; simulation ; hospital comparisons

Abstract

Background: The traditional method for comparing hospitals is hierarchical models that adjust for case-mix. In contrast, "template matching" can be used to standardize comparisons by matching outcomes of similar patients at different hospitals. We sought to evaluate the accuracy of these two methods at measuring hospital variation. Methods: We simulated patient case-mix scenarios nested within hospitals and varied (1)relative effect of measured and unmeasured covariates, (2)correlation between observed and unobserved covariates and (3)magnitude of variation in case-mix. We randomly selected 500 patients for the template and matched to patients at each hospital. We assessed the outcome and determined which hospitals were worse than the median, and compared to traditional hierarchical models. Results: Preliminary results suggest that in certain scenarios, hierarchical models falsely attribute case-mix to hospital variation of outcomes while template matching does not. Conclusion: In some scenarios, hierarchical models were less effective at adjusting for case-mix and template matching was better at attributing varying outcomes across hospitals to the variability between patients.