(a) Purpose.
(1) The Overall Hospital Quality Star Rating (Overall Star Rating) is a summary of certain publicly reported hospital measure data for the benefit of stakeholders, such as patients, consumers, and hospitals.
(2) The guiding principles of the Overall Star Rating are as follows. In developing and maintaining the Overall Star Ratings, we strive to:
(i) Use scientifically valid methods that are inclusive of hospitals and measure information and able to accommodate underlying measure changes;
(ii) Align with Hospital Compare or its successor website and CMS programs;
(iii) Provide transparency of the methods for calculating the Overall Star Rating; and
(iv) Be responsive to stakeholder input.
(b) Data included in Overall Star Rating -
(1) Source of data. The Overall Star Rating is calculated based on measure data collected and publicly reported on Hospital Compare or its successor site under the following CMS hospital inpatient and outpatient programs:
(i) Hospital Inpatient Quality Reporting (IQR) Program - section 1886(b)(3)(B)(viii)(VII) of the Act.
(ii) Hospital-Acquired Condition Reduction Program - section 1886(p)(6)(A) of the Act.
(iii) Hospital Value-based Purchasing Program - section 1886(o)(10)(A) of the Act.
(iv) Hospital Readmissions Reduction Program - section 1886(q)(6)(A) of the Act.
(v) Hospital Outpatient Quality Reporting (OQR) Program - section 1833(t)(17)(e) of the Act.
(2) Hospitals included in Overall Star Rating. Subsection (d) hospitals subject to the CMS quality programs specified in paragraph (b)(1) of this section that also have their data publicly reported on one of CMS' websites are included in the Overall Star Rating.
(3) Critical Access Hospitals. Critical Access Hospitals (CAHs) that wish to be voluntarily included in the Overall Star Rating must have elected to -
(i) Voluntarily submit quality measures included in and as specified under CMS hospital programs; and
(ii) Publicly report their quality measure data on Hospital Compare or its successor site.
(c) Frequency of publication and data used. The Overall Star Rating are published once annually using data publicly reported on Hospital Compare or its successor website from a quarter within the prior year.
(d) Methodology -
(1) Selection of measures. Measures are selected from those publicly reported on Hospital Compare or its successor website through certain CMS quality programs under paragraph (b)(1) of this section.
(i) From this group of measures, measures falling into one or more of the exclusions in paragraphs (d)(1)(i)(A) through (E) of this section will be removed from consideration:
(A) Measures that 100 hospitals or less publicly report. These measures would not produce reliable measure group scores based on too few hospitals;
(B) Measures that cannot be standardized to a single, common scale and otherwise not amenable to inclusion in a summary score calculation alongside process and outcome measures or measures that cannot be combined in a meaningful way. This includes measures that cannot be as easily combined with other measures captured on a continuous scale with more granular data;
(C) Non-directional measures for which it is unclear whether a higher or lower score is better. These measures cannot be standardized to be combined with other measures and form an aggregate measure group score;
(D) Measures not required for reporting on Hospital Compare or its successor websites through CMS programs; or
(E) Measures that overlap with another measure in terms of cohort or outcome, including component measures that are part of an already-included composite measure.
(ii) [Reserved]
(2) Measure score standardization. All measure scores are standardized by calculating Z-scores so that all measures are on a single, common scale to be consistent in terms of direction (that is, higher scores are better) and numerical magnitude. This is calculated by subtracting the national mean measure score from each hospital's measure score and dividing the difference by the measure standard deviation in order to standardize measures.
(3) Grouping measures. Measures are grouped into one of the five clinical groups as follows:
(i) Mortality.
(ii) Safety of Care.
(iii) Readmission.
(iv) Patient Experience.
(v) Timely and Effective Care.
(4) Calculate measure group scores. A score is calculated for each measure group for which a hospital has measure data using a simple average of measure scores, as follows:
(i) Each measure group score is standardized by calculating Z-scores for each measure group so that all measure group scores are centered near zero with a standard deviation of one.
(ii) We take 100 percent divided by the number of measures reported in a measure group to determine the percentage of each measure's weight.
(iii) The measure weight is then multiplied by the standardized measure score to calculate the measure's weighted score.
(iv) Then, all of the individual measure weighted scores within a measure group are added together to calculate the measure group score.
(5) Reporting thresholds. In order to receive an Overall Star Rating, a hospital must report at least three measures within at least three measure groups, one of which must specifically be the Mortality or Safety of Care outcome group.
(6) Hospital summary score. A summary score is calculated by multiplying the standardized measure group scores by the assigned measure group weights and then summing the weighted measure group scores.
(i) Standard measure group weighting.
(A) Each of the Mortality, Safety of Care, Readmission, and Patient Experience groups are weighted 22 percent; and
(B) The Timely and Effective Care group is weighted 12 percent.
(ii) Reweighting.
(A) Hospitals may have too few cases to report particular measures and, in those cases, may not report enough measures in one or more measure groups.
(B) When a hospital does not have enough measures in one or more measure groups due to too few cases CMS may re-distribute one or more of the missing measure group's weight proportionally across the remaining measure groups by subtracting the standard weight percentage of the group or groups with insufficient measures from 100 percent; and then dividing the resulting percentage across the remaining measure groups, giving new re-proportioned weights.
(7) Peer grouping. Hospitals are assigned to one of three peer groups based on the number of measure groups for which they report at least three measures: three, four, or five measure groups.
(8) Star ratings assignment. Hospitals in each peer group are then assigned between one and five stars where one star is the lowest and five stars is the highest using k-means clustering to complete convergence.
(e) Preview period prior to publication. CMS provides hospitals the opportunity to preview their Overall Star Rating prior to publication. Hospitals have at least 30 days to preview their results, and if necessary, can reach out to CMS with questions.
(f) Suppression of Overall Star Rating -
(1) Subsection (d) hospitals. CMS may consider suppressing Overall Star Rating for subsection (d) hospitals only under extenuating circumstances that affect numerous hospitals (as in, not an individualized or localized issue) as determined by CMS, or when CMS is at fault, including but not limited to when:
(i) There is an Overall Star Rating calculation error by CMS;
(ii) There is a systemic error at the CMS quality program level that substantively affects the Overall Star Rating calculation; or
(iii) If a Public Health Emergency, as defined in § 400.200 of this chapter, substantially affects the underlying measure data.
(2) CAHs.
(i) CAHs may request to withhold their Overall Star Rating from publication on Hospital Compare or its successor website so long as the request for withholding is made, at the latest, during the Overall Star Rating preview period.
(ii) CAHs may request to have their Overall Star Rating withheld from publication on Hospital Compare or its successor website, as well as their data from the public input file, so long as the request is made during the CMS quality program-level 30-day confidential preview period for the Hospital Compare refresh data used to calculate the Overall Star Ratings.
[85 FR 86300, Dec. 29, 2020]