Concordance of Hospital Ranks and Category Ratings Using the Current Technical Specification of US Hospital Star Ratings and Reasonable Alternative Specifications.

Change log
Barclay, Matthew E 
Lyratzopoulos, Georgios 

IMPORTANCE: The Centers for Medicare & Medicaid Services (CMS) Hospital Compare star ratings are widely used summaries of hospital quality that can influence patient choice and organizational reputation. OBJECTIVE: To identify the changes in hospital ratings and rankings associated with alternative methodological choices in the calculation of the 2021 CMS Hospital Compare star ratings. DESIGN SETTING AND PARTICIPANTS: This cross-sectional study used publicly available 2021 Hospital Compare data for 3339 US hospitals from the October 2020 data release. Change in apparent hospital performance was assessed, arising when plausible alternatives to current methods are used for calculating star ratings in relation to individual measure standardization, domain derivation, and domain weighting. Three example changes were examined in detail, with more comprehensive changes considered using Monte Carlo simulation. Changes in centile of hospital ranks and in star rating overall were examined, as well as separately in CMS peer groupings defined by the number of reported quality domains. MAIN OUTCOMES AND MEASURES: Proportion of hospitals receiving a different star rating under the alternative technical specifications than under the current (2021) CMS approach and mean absolute change in centile rank of hospitals under alternative technical specifications compared with the current (2021) CMS approach. RESULTS: A total of 3339 US hospitals were included in the analysis. Of the specific changes to current specifications considered, the alternative method of standardization was associated with the most substantial changes, with 55.4% (95% CI, 53.7%-57.1%) of hospitals having their star rating reclassified. The change in domain weights was associated with the smallest differences, but even that resulted in reclassification of the star rating in approximately 1 in 4 (24.5%; 95% CI, 23.0%-26.0%) hospitals. In the simulation study, which covered a range of possible changes, on average half of hospitals (51.8%; IQR, 44.2%-59.7%) were assigned a different star rating from that produced by the current specification, usually into an adjacent star category. CONCLUSIONS AND RELEVANCE: In this cross-sectional study of data on US hospitals, CMS Hospital Compare star ratings were found to be highly sensitive to how performance ratings are calculated, demonstrating the need for transparent justification of the technical approaches used in calculating composite performance ratings.

Aged, Cross-Sectional Studies, Hospitals, Humans, Medicare, United States
Journal Title
JAMA Health Forum
Conference Name
Journal ISSN
Volume Title
American Medical Association (AMA)
Wellcome Trust (097899/Z/11/Z)
Health Foundation (unknown)
This project was funded by MD-W’s Wellcome Trust Investigator award WT09789 and by THIS Institute’s grant from the Health Foundation. The Health Foundation is an independent charity committed to bringing about better health and health care for people in the UK. MD-W is an NIHR Senior Investigator (NF-SI-0617-10026). GL is funded by a Cancer Research UK Advanced Clinician Scientist Fellowship award (grant number C18081/A18180). The corresponding author is Matt Barclay, who led on the work and was employed at Cambridge until recently. He does not have a current affiliation as he has moved institutions.