Hospital ranking lists — such as those released annually by U.S. News & World Report and Healthgrades — are a great way for the nations’ largest hospitals to generate good PR. However, they may not be all that useful when it comes to understanding the strengths and weaknesses of individual hospitals and how they compare to each other.
Trilliant Health, a healthcare-focused predictive analytics and market research firm has now done just that with a new index for hospital benchmarking. The company’s AI-powered SimilarityIndex, released Thursday, promises to serve as a functional tool that healthcare stakeholders can use for strategic planning.
The index compares more than 2,000 acute care hospitals. Healthcare stakeholders across the entire industry — including hospitals, payers, device companies, drugmakers and tech vendors — can use the tool to inform their strategies for things like competitive analysis, clinical quality program management, mergers, acquisitions and product sales.
Applying mathematical accuracy to strategic planning in healthcare minimizes the risk of inappropriate capital allocation for providers, payers and medical manufacturers, said Sanjula Jain, Trilliant’s chief research officer and senior vice president of market strategy, in an interview.
To build its index, Trilliant used claims data from every Medicare Advantage plan and “basically every commercial payer,” according to Jain. The data Trillaint uses for its index “can basically tell you about any interaction between a physician and a patient, so long as it’s generating a bill,” she declared.
This data allows healthcare stakeholders to answer a lot of questions, such as which medications are being prescribed at which hospitals or where physicians are seeing the most patients. In addition to answering these questions, users can also use the index to “objectively and mathematically compare hospitals,” Jain said.
The index produces similarity scores for hospitals, which are designed to show hospitals’ relevant peers based on factors like financial results, CMS quality metrics, patient mix and market share. The interactive data visualization tool allows users to isolate variables for specific hospital comparisons. For example, if a user wants to see a hospital’s comparable peers as it pertains to the rate of readmissions and hospital-acquired infections only, they can.
“If you’re trying to compare hospitals on hundreds of variables, it becomes a mathematical nightmare to do at scale,” Jain said. “That’s where machine learning comes in. Until this point, no one’s really been able to objectively benchmark because there has been such a computational limitation in the industry.”
Jain credited Trilliant’s team of data scientists and engineers for building the algorithms to compare hospitals against their relevant peers at scale. These algorithms often reveal that the most renowned hospitals in the country are quite unalike, she pointed out.
For example, when you enter Johns Hopkins Hospital into the index, you might expect its peers to be other renowned hospitals like NYU Langone Medical Center or Cedars-Sinai Medical Center. When you plug it in, though, the index reveals that its true peers are hospitals like Thomas Jefferson University Hospital and Tampa General Hospital.
Hospitals and health systems have been trained to think that their peers are by size, according to Jain. However, the biggest health systems by revenue are not all navigating the same challenges, and the index can help pinpoint what operational differences each unique hospital is dealing with.
Photo: Filograph, Getty Images