On February 28, 2019, CMS posted proposed changes to their Hospital Star Ratings on Hospital Compare  for public comment. These changes intend to enhance the Star Ratings Methodology for hospitals by increasing rating consistency and making hospital comparisons more precise. Potential changes were developed using a newly-developed Hospital Quality Star Ratings Technical Expert Panel (TEP). Currently, CMS contracts with Yale New Haven Health Services Corporation, Center for Outcomes Research and Evaluation (CORE) and Lantana Consulting Group, Inc. to develop and maintain the publicly reported Overall Hospital Quality Star Ratings.  Comments on proposed changes are due to CMS on March 29, 2019.

Modifications to Quality Measurement and Calculation of Ratings

CMS proposed 4 changes to evaluation of quality measures, which are outlined below.

New Process for Measure Grouping

CMS proposes to use a new, three-step approach on regrouping individual measure when they are added or updated on Hospital Compare:

1. Grouping measures based on clinical criteria;
2. Using statistical tests to determine if an important latent quality trait is represented by the measures in the group; and
3. Actively following measure groupings for consistency in how much each measure influences the measure group score over time.

Incorporating Measure Precision

Currently, to determine a hospital’s precision score on an individual measure, CMS looks at how many patients are included in a specific quality measure to calculate the weight of that measure on a specific hospital. For example, if a hospital has a very high number of patients with type two diabetes, more weight for that hospital is based on that measure since there is a higher incidence of patients seen with that disease.  CMS proposes to instead remove this weighting, as the Agency feels that in inappropriately “loads” hospitals to appear successful based on only a select group of measures. CMS seeks comment on how to best weight these quality measures for individual hospitals.

Period to Period Shifts

CMS acknowledges that their bi-annual updates of Hospital Compare results in frequent shifts in ratings. CMS proposes to update the Overall Hospital Quality Start Rating to once annually.

New Peer grouping methodology

Currently, all hospitals are compared to one another under Hospital Compare regardless of their characteristics or unique circumstances. CMS acknowledges that grouping all hospitals together may place some hospitals at a disadvantage. CMS proposes grouping hospitals in “peer groups”, allowing hospitals to be compared only to other hospitals with similar characteristics and potentially similar resources. Potential groups, for example, could include Critical Access Hospitals (CAH)s, teaching hospitals, or safety-net hospitals. CMS requests feedback on this peer grouping methodology, including feedback on which categories of peer groups should be established to best reflect hospitals’ unique challenges. CMS also requests feedback on whether multiple star ratings should be included: one star rating based on peer groups and another star rating for overall quality.

Technical Modifications

CMS proposes 5 technical modifications to their methodologies for score calculations; these focus on statistical changes and changes to methodologies to the computations of Hospital Compare star ratings. CMS requests feedback on the advantages and disadvantages of these changes, including how to best implement their proposals into the current system. These technical updates include:

Closed form solution to replace quadrature computation

CMS believes this change would improve the Overall Hospital Quality Star Rating statistical programming code.

Explicit Approach to replace a statistical model

CMS proposes uses a pre-defined approach using technical specifications to analyze weights within each quality measure. This allows CMS to put additional weight on a measure.

Alternatives to clustering

CMS currently assigns a star rating to a hospital based on a statistical method of weighting averages called “k-means clustering”. This method results in hospitals receiving star ratings based on the relative performance of other hospitals. CMS proposes to instead assign star ratings based on individual hospital performance, independent of the performance of other hospitals.

Incorporation of Improvement

Currently, Hospital Compare star ratings are calculated based on a hospitals’ performance in relation to other hospitals. CMS proposes to change this methodology and instead include star ratings based on an individual hospitals’ improvement over time. Effectively, this change would capture improvement (or decrease in performance) over time.

User-customized star rating

Currently, weighting for each measure group is 22% for each outcome group, 22% for patient experience, and 4% for process measure group. These weights are static, and CMS is currently considering letting the user (patient) weight these categories based on their personal values.

x

x

x