Grading Hospitals Using Multivariate Matching

Area of Science:

• Healthcare quality assessment
• Comparative effectiveness research
• Health services research

Background:

• Existing hospital grading systems have limitations in accurately reflecting performance.
• There is a need for more sophisticated methods to benchmark hospital quality.

Purpose of the Study:

• To develop and illustrate a novel hospital report card system.
• To improve upon existing hospital grading methodologies using multivariate matching.

Main Methods:

• Utilized Medicare Claims data (2017-2019) for heart attack, heart failure, and pneumonia admissions.
• Developed matched cohorts comparing focal hospital patients to controls from "well-resourced" and "typical" hospitals based on over 300 patient characteristics.
• Created an "Analogous" match to compare outcomes at hospitals with similar characteristics.

Main Results:

• Generated report cards for four distinct example hospitals.
• Demonstrated the application of the multivariate matching technique for hospital benchmarking.

Conclusions:

• The new matched report card system enables more accurate hospital benchmarking.
• Identifies specific patient populations where hospitals show performance disparities compared to similar institutions and patient profiles.