CMS unveils updated hospital star ratings formula

(Updated on Dec. 22)

The CMS announced Thursday it has added updated star ratings to its Hospital Compare site using a new methodology after a five-month delay.

The agency had postponed the release of the star ratings since July as it worked to change the methodology and gather stakeholder feedback. The new methodology was designed in response to criticism—particularly from hospitals—that the previous formula was flawed and provided inaccurate information about providers.

The retooled formula slightly flattens the usual bell curve of hospitals that receive stars on a scale of 1 to 5, with 5 being the highest rating. In the previous methodology, few hospitals received 1 or 5 stars, and the majority have 2, 3 or 4 stars. While that's still the case, the number of hospitals with 1 or 5 stars has risen.

Using the new methodology, 260 hospitals now have a 1-star rating compared with 125 hospitals under the earlier calculation. At the same time, there are now 337 hospitals with a 5-star rating versus 78 hospitals with the same rating in the old formula.

The most significant change to the methodology is the CMS has removed winsorization, a tactic in which extreme hospital outliers in the data set were trimmed to minimize their effect on the overall ratings. Instead, the CMS will run k-means clustering "to complete convergence." This involves clustering hospitals repeatedly into five groups—or five star categories—until the hospitals in each group are more similar to each other and different enough from the other four groups.

The new formula addresses concerns that hospitals in different star categories didn't perform much differently from each other on metrics. The CMS said k-means clustering "provides more reliable and stable star-rating assignments."

The distribution of quality metric scores used to determine a composite score for each hospital also has been altered. The new methodology has significantly decreased the emphasis on patient experience while slightly increasing the score for readmissions and timeliness of care.

What hasn't changed is how the CMS weighs each hospital's quality metrics to calculate the composite star rating. The CMS will continue to weigh the average score for each of the seven measure groups the same way. For example, 22% of the overall rating is the hospital's mortality, safety of care, readmission and patient experience scores. Effectiveness of care, timeliness of care and efficient use of medical imaging is each weighted at 4% for the total score.

The way the CMS weighs the metrics is what some have argued is a flaw in the star ratings. For hospitals that don't report all of the measures involved to get a composite star rating, more weight will be given to measures the hospital did report.

A deeper dive by the CMS of how well hospitals performed overall on the seven measure groups shows wide variability. About 43% of hospitals performed above the national average on patient safety, while 34.5% of hospitals performed below the average for the measure. Performance on preventing readmissions was variable as well. About 40% of hospitals performed above the national average on readmissions while 25% were about the same as the national average, and 35% performed below the average.

Of the 4,578 hospitals on Hospital Compare, 3,692 met the reporting criteria for a star rating, the CMS said.

"We continue to refine the star ratings and look forward to an ongoing dialogue with hospitals and patients and their families on how we can provide beneficiaries useful information," CMS Administrator Seema Verma said in a statement.