Healthgrades rates hospitals by comparing each facility’s actual patient outcomes (deaths and complications) against what would be statistically expected given the severity of its patient population. Hospitals that perform significantly better than predicted earn 5 stars, those that perform as expected get 3 stars, and those that perform worse receive 1 star. The system covers 31 medical conditions and procedures, drawing primarily from Medicare claims data.
Where the Data Comes From
Healthgrades builds its clinical ratings from the Medicare Provider Analysis and Review (MedPAR) file, purchased from the Centers for Medicare and Medicaid Services. This dataset captures inpatient records for Medicare patients across virtually every hospital in the country. For mortality measures, the data typically spans three years (the current ratings use discharges from July 2020 through June 2023). Safety and complication metrics use slightly different windows depending on the measure, with hospital-acquired infections drawing from a single calendar year of data.
Patient experience ratings come from a separate source: the HCAHPS survey (Hospital Consumer Assessment of Healthcare Providers and Systems), a standardized questionnaire that CMS administers to a representative sample of recently discharged patients at each hospital. The current patient experience scores are based on patients discharged between January and December 2023.
How Star Ratings Are Calculated
The core of the Healthgrades methodology is a predicted-versus-actual comparison. For each condition or procedure, a statistical model predicts how many patients at a given hospital would be expected to die or develop complications, based on the characteristics of the patients that hospital treated. Then Healthgrades compares that predicted number to what actually happened.
The key question is whether the gap between predicted and actual outcomes is large enough to be meaningful, not just a fluke of small numbers. Healthgrades uses a z-score test with a 90% confidence interval to make that determination. The cutoff is 1.645 standard deviations above or below the average:
- 5 stars: The hospital’s outcomes were significantly better than predicted, scoring more than 1.645 standard deviations above the mean.
- 3 stars: Outcomes fell within the expected range, with no statistically significant difference from what the model predicted.
- 1 star: Outcomes were significantly worse than predicted, falling more than 1.645 standard deviations below the mean.
This means most hospitals land at 3 stars for most conditions. A 5-star or 1-star rating requires a gap large enough that there’s only about a 10% chance it happened by random variation. Hospitals with very low patient volumes for a given procedure may not receive a rating at all, because the sample size is too small to reach statistical significance either way.
How Risk Adjustment Works
Raw death rates or complication rates would be misleading, because some hospitals treat sicker, older patients. A hospital specializing in complex heart surgery would look terrible compared to one handling straightforward cases. Risk adjustment is designed to level that playing field.
Healthgrades’ models account for clinical factors like the type, number, and severity of a patient’s existing conditions. A patient with kidney failure, for example, would be expected to have a higher risk of death after heart surgery than an otherwise healthy patient. Age and gender are also factored in. The models can include interaction terms, where two patient characteristics together predict outcomes more strongly than either one alone. Medication patterns may serve as a proxy for how advanced a chronic illness has become.
What risk adjustment does not account for is equally important. Factors like hospital size, staffing ratios, teaching status, or geographic location are deliberately excluded. The goal is to adjust for the patients a hospital receives, not for the resources it has. That way, the rating reflects what the hospital did with the patients it treated.
The 16 Specialty Areas
Beyond individual condition ratings, Healthgrades groups its 31 evaluated conditions and procedures into 16 specialty categories and issues Specialty Excellence Awards to top performers. These specialties include cardiac care, cardiac surgery, coronary intervention, cranial neurosurgery, critical care, gastrointestinal care, gastrointestinal surgery, joint replacement, neurosciences, orthopedic surgery, prostate surgery, pulmonary care, spine surgery, stroke care, surgical care, and vascular surgery.
A hospital can earn 5 stars in joint replacement but 1 star in stroke care, so looking at the specialty that matters for your specific situation is more useful than focusing on any overall ranking. These specialty ratings are updated annually, with the latest cycle typically released each October.
America’s Best Hospital Awards
Healthgrades also rolls individual ratings into tiered national awards that recognize sustained performance across multiple specialties. There are three levels, and each has strict eligibility requirements.
The America’s 250 Best Hospitals award recognizes the top 5% of hospitals nationally. To even be considered, a hospital must have enough patient volume to be evaluated in at least 19 of the 31 conditions and procedures. This requirement filters out small or highly specialized facilities that may excel in one area but lack the breadth of services to be compared on a national scale.
The America’s 100 Best Hospitals award (top 2%) goes to hospitals that earned the 250 Best designation for at least five consecutive years. The America’s 50 Best Hospitals award (top 1%) requires at least eight consecutive years. These aren’t one-time achievements. They reward hospitals that maintain strong outcomes across many specialties over long stretches, which makes them harder to earn than any single-year rating.
How Patient Experience Fits In
Patient experience ratings operate on a separate track from clinical outcomes. Healthgrades uses the HCAHPS survey, which rolls 19 questions into 10 measures covering topics like doctor communication, nurse communication, staff responsiveness, medication explanations, room cleanliness, nighttime quietness, discharge instructions, care transitions, and overall hospital rating. Patients also indicate whether they’d recommend the hospital to family and friends.
For the overall rating question, patients score the hospital on a 1-to-10 scale. Healthgrades looks at the percentage who gave a 9 or 10, then compares that to the national average. Across all 10 measures, Healthgrades creates z-scores for each hospital, placing it on a curve relative to all other hospitals in its comparison group. Hospitals that score well across these measures can earn an Outstanding Patient Experience Award.
It’s worth noting that patient experience and clinical quality don’t always move together. A hospital with excellent survival rates might score poorly on nurse responsiveness, or vice versa. Healthgrades presents these as separate dimensions rather than blending them into one composite score, which lets you weigh each factor based on what matters most to you.
What the Ratings Don’t Capture
Healthgrades ratings are built almost entirely on Medicare inpatient data, which means they reflect outcomes for patients aged 65 and older (plus younger Medicare beneficiaries with disabilities or specific conditions). Outcomes for privately insured or uninsured patients aren’t included. Outpatient procedures, emergency department performance, and maternity care also fall outside the scope of most Healthgrades clinical ratings.
The reliance on claims data rather than detailed clinical records introduces some limitations. Claims data captures diagnoses and procedures but not granular details like lab values, imaging findings, or the precise timing of clinical decisions. Risk adjustment models do their best with the available information, but no model perfectly accounts for how sick patients are when they arrive.
Hospitals with small Medicare volumes for a given procedure may receive no rating rather than an unreliable one. If you’re researching a hospital for a specific surgery and see no Healthgrades rating for that procedure, it likely means the hospital didn’t treat enough Medicare patients in that category to produce a statistically valid result.