Epic, an electronic patient information company, provides financial incentives for healthcare systems that use its own algorithms – and these algorithms can produce inaccurate predictions, a Stat News investigation found.
Epic is the largest electronic patient record company in the United States, with health records for approximately 250 million people. The company has about 20 algorithms designed to predict things like how long a patient can stay in the hospital, which patients can become seriously ill, and which can develop into a deadly condition called sepsis.
Like many other groups that build health algorithms, Epic does not publicly share details about building algorithms. Researchers at hospitals that use Epic can research the tools, but all research is challenging: they can’t reveal dedicated information and can worry about making the employer’s relationship with the company more difficult if they publish negative results; Stat News reported.
Studies own made algorithms show that they often do not work as well as advertised. One analysis the sepsis algorithm, which Epic says works 76 percent of the time, was found to be correct only 63 percent of the time.
Despite the problems with the algorithms, Epic encourages hospitals to adopt them. The company’s “honor roll” program has allocated hundreds of thousands of dollars, and in some cases nearly a million dollars, to health care systems that implement technical upgrades, including algorithms. “It would be a terrible world where Epic gives people a million dollars, and the end result is that patients’ health is deteriorating, ”said Glenn Cohen, director of Harvard’s Petrie-Flom Center for Health Law, Biotechnology and Bioethics at the University, said. Stat News. Financial incentives can cause conflicts of interest.
“Epic’s Honor Roll is a voluntary program that encourages the introduction of features that help save lives, increase information exchange, and improve patient, physician, and health system satisfaction,” Epic said in a statement. Stat News.
Some healthcare systems with financial resources and robust IT departments are able to analyze Epic’s ready-made algorithms and adapt them to their needs. However, others may not be able to examine the tools so closely. “The least developed hospitals, which don’t have large data sets, are also the ones most likely to be affected by a million dollars,” Cohen said.