Verona, Wis.-based EHR giant Epic gives financial incentives to hospitals and health systems that use its artificial intelligence algorithms, which can provide false predictions, according to a July 26 STAT News investigation.
For its report, STAT interviewed several health systems that use any of Epic's 20 algorithms. Employees of several major systems told the publication they were concerned about Epic's algorithm for predicting sepsis, claiming that the tool routinely fails to identify the condition in advance and frequently triggers false alarms, according to the report. Some hospitals reported benefits for patients after refining the model, but the process took at least a year.
In June, Michigan Medicine researchers published findings from their analysis of the sepsis prediction model in JAMA Internal Medicine. The researchers found that it correctly sorted patients on their risk of sepsis 63 percent of the time, lower than the 76 percent to 83 percent curve Epic indicated on the model's fact sheet.
Epic began developing its predictive models in 2014, a company spokesperson told Becker's in a July 26 emailed statement, adding that its sepsis prediction model "has been found to improve clinical care, and help clinicians save lives."
"Prisma Health saw a 22 percent reduction in sepsis mortality after using our model," the spokesperson said. "Yale found that the implementation of Epic's sepsis model system cut deaths from sepsis by 7 percent. Multiple studies from groups that are using our model have reported a positive impact on clinical outcomes."
STAT's investigation also found that Epic has paid health systems up to $1 million in incentives for adopting predictive algorithms it develops. Epic launched its "honor roll" program in 2017 as part of its push to expand adoption of its AI products. Under the program, hospitals that start using the company's predictive algorithms and make other technical upgrades can receive hundreds of thousands of dollars in annual payments, according to the report.
"It would be a terrible world where Epic is giving people a million dollars, and the end result is the patients' health gets worse," Glenn Cohen, director of Harvard University's health law policy, biotechnology and bioethics center, told the publication. He added that Epic has not made public the process behind its algorithms or the specific criteria used to determine which providers make honor roll status or how payments and awards are calculated.
"That's what smells a little strange about this," Mr. Cohen said. "Epic will crow about its [algorithm] adoption rates. But if the adoption rates are essentially paying for adoptions, it's a little bit of a different problem."
Epic told Becker's that it does not charge for its sepsis model and that the Honor Roll is a "voluntary program which encourages the adoption of features that help save lives, increase data exchange, and improve satisfaction for patients, physicians, and health systems."
The program does require hospitals that join to implement a sepsis early prediction model since it is the leading cause of death in hospitals, but it does not require specific use of Epic's model, the spokesperson said. Providers can meet the program's requirements by deploying another vendor's model or technology developed in house.