The CMS is considering scrapping the model it uses to assign hospital star ratings, signaling a big shift from the agency’s stance just seven months ago.
In conjunction with releasing new ratings on the Hospital Compare website Thursday for the first time in nearly 15-months, the agency also opened a public comment page for stakeholders to provide feedback on potential changes to the ratings program, and one them is tossing out the latent variable model used to assign hospitals ratings.
In the 48-page public input request, the CMS said it’s considering “replacing LVM (the latent variable model) with an explicit approach (such as an average of measure scores) to group score calculation.”
The CMS called it a “long-term” consideration that wouldn’t be applied until 2020 quality reporting and beyond.
It’s a left turn from July 2018 when a CMS agency official defended the latent variable model in an interview with Modern Healthcare even after it was heavily criticized by hospital groups. Hospital leaders blamed the model for drastic changes they saw in their star ratings from December 2017 to June 2018 that led to the 15-month hiatus of the star ratings update.
The latent variable model, which was created with help from Yale New Haven Health Services as part of a multi-million dollar contract with the CMS, is a statistical approach that gives more emphasis to certain measures over others in the star ratings based on a number of aspects, including variation in performance among hospitals for that measure or how much measures correlate to each other. This approach is why some measures have a far greater influence than others in a hospital’s overall rating.
Nancy Foster, vice president for quality and patient safety at the American Hospital Association, said re-evaluating the latent variable model is critical to making any meaningful changes to the star ratings. Because the model is the foundation to the ratings program, Foster said she’s unsure if any other changes the CMS makes to the star ratings will actually have an impact.
“We think we know what might help but in unless you plug it through the latent variable model you don’t know if it’s going to do what you think it will,” she said.
One potential replacement the CMS suggested is assigning weights to each of the measures used in the ratings instead of allowing a model to assign the weights.
In addition to tossing the latent variable model, the CMS is asking for feedback on other changes to the ratings including assigning hospitals to peer groups based on their size or status like teaching or community; only releasing the star ratings once a year and a user-friendly tool that would allow consumers to decide how they want the measures to be weighted.
The AHA and America’s Essential Hospitals both said they were disappointed the CMS decided to go through with the February refresh of the star ratings, arguing they aren’t helpful to consumers as they currently stand.
“We were somewhat perplexed in terms of the release today, we thought it was somewhat premature and would’ve preferred the agency wait to address a lot of the concerns that are in the document for public comment right now,” said Maryellen Guinan, senior policy analyst at America’s Essential Hospitals.
The latest star ratings update to Hospital Compare didn’t see any changes from the preview report the CMS released to providers in December, an agency official said. At the time, hospital stakeholders expressed disappointment that the February star ratings won’t have substantial changes to the methodology.
Guinan also said America’s Essential Hospitals was disappointed that the CMS isn’t seeking comments on adding risk adjustment for the readmissions, mortality and patient experience measures because research shows patients’ socioeconomic status can impact how hospitals perform on those measures.
The one change the CMS made to the February star ratings was how the healthcare-associated infections measures used in the methodology were weighted, which allowed those measures to have a greater impact on a hospital’s overall rating. The CMS also will remove measures that have statistically significant negative loading from the star ratings, but that didn’t happen in the February star ratings release, a CMS official said.
Even as the CMS acknowledges opportunities to improve the ratings system, the CMS official said she stands by the current star ratings as a helpful tool for consumers.
The public comment period will end March 29.