Hospitals·4 min read

CMS Hospital Star Ratings: How They Work and Where to Get the Data

An explanation of how CMS calculates hospital star ratings (1-5 stars), the data sources behind them, and how to access the Compare dataset programmatically.

CMS Hospital Star Ratings: How They Work and Where to Get the Data

What the star ratings measure

CMS (Centers for Medicare and Medicaid Services) assigns every Medicare-certified hospital a rating from 1 to 5 stars. The rating summarizes performance across multiple quality dimensions into a single number.

About 4,500 hospitals receive star ratings. The most recent release uses data from 2022-2024.

Stars Hospitals Interpretation
5 stars ~430 (10%) Top performers
4 stars ~1,050 (23%) Above average
3 stars ~1,250 (28%) Average
2 stars ~1,050 (23%) Below average
1 star ~350 (8%) Lowest performers
No rating ~370 (8%) Insufficient data

The five quality domains

Star ratings combine 47 individual quality measures grouped into five domains:

  1. Mortality (22% weight) - Death rates for heart attack, heart failure, pneumonia, COPD, stroke, and coronary artery bypass surgery.

  2. Safety of Care (22% weight) - Hospital-acquired infections (CLABSI, CAUTI, SSI, MRSA, C. diff), falls, pressure ulcers, and patient safety indicators.

  3. Readmission (22% weight) - 30-day unplanned readmission rates for the same conditions as mortality, plus hip/knee replacement.

  4. Patient Experience (22% weight) - HCAHPS survey scores (communication with nurses/doctors, hospital cleanliness, pain management, discharge information).

  5. Timely and Effective Care (12% weight) - Process measures like stroke treatment speed, blood clot prevention, and immunization rates.

How CMS calculates the score

The methodology uses a three-step process:

Step 1: Measure standardization. Each of the 47 measures is converted to a z-score (how many standard deviations from the national mean).

Step 2: Domain scoring. Z-scores within each domain are averaged to create a domain score.

Step 3: Star assignment. Domain scores are weighted and combined into a summary score. A k-means clustering algorithm groups hospitals into 5 clusters, assigning stars.

The k-means approach means star cutoffs shift with each release. A hospital that scored 3 stars last year could get 4 stars this year with identical performance if the national distribution changed.

Where to access the data

CMS publishes hospital quality data through several channels:

Hospital Compare website: Medicare.gov/hospitalcompare provides a consumer-friendly search interface.

Data downloads: data.cms.gov hosts bulk CSV files for all quality measures, updated quarterly.

SODA API: Some CMS datasets are available through the Socrata API:

# Hospital general information
curl "https://data.cms.gov/provider-data/api/1/datastore/query/xubh-q36u/0?limit=10"

Provider Data Catalog: The full catalog at data.cms.gov/provider-data lists all available hospital datasets with download links.

Measures that matter most

Not all 47 measures carry equal influence on the final star rating. Because mortality and safety each carry 22% weight and contain fewer individual measures, each mortality or safety measure has more per-measure impact.

The single most influential measures:

  • 30-day mortality rate for heart attack (AMI)
  • CLABSI (central line-associated bloodstream infection) rate
  • 30-day readmission rate for heart failure
  • HCAHPS overall hospital rating question

Controversies with the methodology

The star rating system has faced consistent criticism:

Teaching hospitals score lower. Academic medical centers treat sicker patients and perform more complex procedures. Risk adjustment models may not fully account for this case-mix difference.

Small hospitals get excluded. Hospitals with insufficient data volume receive no rating. This tends to exclude rural and critical access hospitals.

k-means clustering is unstable. Small changes in the data can shift star boundaries, causing rating changes that do not reflect actual quality improvement.

Gaming potential. Hospitals can improve stars by focusing on measured processes (documentation, coding) rather than actual patient outcomes.

Practical applications for developers

Hospital quality data enables:

  • Provider directories that show quality alongside location and insurance acceptance
  • Referral optimization for health systems routing patients to highest-rated facilities for specific conditions
  • Risk scoring for health plans assessing network adequacy
  • Public health research on geographic variation in care quality
  • Hospital benchmarking tools for administrators comparing against peers

FAQ

How often are star ratings updated? CMS releases updated star ratings approximately once per year, typically in July or August.

Can a hospital dispute its rating? Yes. CMS provides a 30-day preview period before public release where hospitals can review and submit corrections to their data.

Do star ratings predict patient outcomes? Research shows 5-star hospitals have measurably lower mortality and readmission rates than 1-star hospitals. The relationship is real but not absolute.

Are star ratings available through an API? Not directly as a single endpoint. The summary star rating is included in the Hospital General Information file on data.cms.gov, which has a SODA API endpoint.

What about specialty hospitals? Specialty hospitals (psychiatric, rehabilitation, long-term care, children's) receive ratings using different measure sets specific to their patient populations.

Published on 2026-08-04 · 4 min read

← Back to all articles