CMS proposed this week a new methodology for its hospital quality star ratings and stakeholders are cautiously optimistic.
After years of backlash from providers, CMS proposed in its 2021 hospital outpatient prospective payment rule an overhaul of how it determines star ratings. The agency proposes removing the controversial latent variable model that has long vexed hospitals. The proposed changes were expected as CMS announced last August it would do so through public rulemaking this year.
The American Hospital Association, which has opposed the star ratings since the first release in July 2016, is "encouraged" by the proposal but is assessing the potential impact on its members, said Akin Demehin, director of policy.
"Part of our hesitancy in saying exactly what we think is we want to go back and look and make sure they (CMS) haven't unintentionally created some new challenges—and it's not a level playing field," added Nancy Foster, AHA's vice president for quality and patient safety policy.
The CMS proposes using averages rather than the latent variable model to determine the weight assigned to quality measures used in the star ratings. With this new method, the weight of each measure will be determined by the number of measures a hospital reported in each group divided by 100. This means each measure in each group will be weighted equally.
Hospitals are optimistic the new methodology will be more predictable, allowing hospitals to use the star ratings for quality improvement efforts. The latent variable model, by comparison, was subject to arbitrary changes that made it difficult for hospitals to understand where they could improve to improve their rating.
The latent variable model determines weights of quality measures by using a number of factors including variation in performance among hospitals for the measure in question or how much measures correlate to each other. This led to unpredictably in the star ratings.
Demehin at AHA said the proposed new model is simplified and predictable, which will allow hospitals to potentially use the star ratings for quality improvement efforts. "It appears to be a step in the right direction," he added.
Other proposed changes include reducing the number of categories used to determine star ratings, peer grouping hospitals by measures reported and risk stratifying the readmission measures by dual eligibility status.
The star ratings currently use quality measures grouped in seven categories: mortality, safety of care, readmission, patient experience, effectiveness of care, timeliness of care, and efficient use of medical imaging.
CMS proposes instead having five categories to account for the reduction in measures as a result of the agency's Meaningful Measures initiative. The new measure groups would be mortality, safety of care, readmission, patient experience and timely and effective care.
The outcomes categories—mortality, safety of care, readmission and patient experience—would continue to account for 22% of the total rating while the timely and effective care group will account for 12%.
A hospital is eligible to be included in the star ratings if it reports at least three measures in at least three of the measure groups. For instance, if a hospital doesn't report any measures in the readmission measure group, the 22% used for that category will be equally distributed to the other measure groups.
A concern with that approach is the process measures included in the timely and effective care group will have a greater impact on a hospital's overall rating, said Dr. David Levine, senior vice president of advanced analytics and product management at Vizient who participated on CMS's technical expert panel for the star ratings.
"Outcomes measures are much more meaningful than process measures," he said.
CMS also proposed grouping hospitals that report the same number of measure groups. For instance, those that report for three measure groups such as patient experience, mortality and safety of care, will be compared to each other while those that report in five measure groups will be compared to each other.
CMS said this will likely result in similar hospitals being compared to each other, which is currently a criticism of the star ratings. Large tertiary hospitals are more likely to report measures for all five groups while small hospitals likely can't, the agency said. As it currently stands in the star ratings, a small rural hospital is now compared to a large, tertiary hospital in an urban area with different case mixes.
This change will probably successfully result in similar hospitals being compared to each other, said Dr. Bala Hota, chief analytics officer at Rush University Medical Center in Chicago who participated on CMS's technical expert panel for the star ratings.
"I think this achieves the aims we were concerned about, (which is being able to have) an apple to apples comparison (of hospitals)," he said.
Although, Demehin at the AHA added that most hospitals likely will report in all five measure groups, so there will still be much variability between the hospitals compared together. He said AHA will "dig into" the proposed peer grouping approach to understand its impact better.
Other suggestions that have been made to achieve peer grouping are to separate hospitals by bed size, volumes or case mix.
Finally, CMS proposes to risk stratify the readmission measure group by dual eligibility status to align with CMS' Hospital Readmission Reduction Program. Hospitals were pushing CMS to apply this to the star ratings for some time.
"It makes a good deal of sense to align the approach that is used for the readmissions penalty program to the star ratings - it enables hospitals to track one metric and one measurement approach," Demehin said.
In addition to methodology changes, CMS also proposed to include Veterans Health Administration hospitals to star ratings beginning in 2023. Hota said this appears to be an attempt by CMS to expand the star ratings to more hospitals as the rule also explicitly mentions including critical access hospitals, although they were already participating in the ratings if they completed the relevant quality information.
Levine said it's unclear how CMS will appropriately include data from VA hospitals because they don't currently report in the same quality programs as acute care hospitals. "There needs to be more details about how they are going to plug" VA data into the star ratings, he said.
Even though CMS has proposed significant changes to the star ratings, stakeholders are hoping more changes will come in the years ahead. Many in the quality arena argue an ideal ratings system allows consumers to search ratings by condition or service line, which the CMS ratings system still doesn't do. A hospital receives a single rating meant to encapsulate overall quality of care at the facility.
"Specialty specific is a worthwhile way to go," Hota said, although he added CMS is limited by its data. Medicare claims data only addresses consumers 65 and older, which leaves out pediatric patients and obstetrics and gynecology services. A future iteration of a ratings system would allow for the use of claims by commercial payers, too, he said.
CMS is seeking feedback on its proposed changes. Comments are due by October 5.
