Epic Sepsis Model lacks predictive power, says JAMA Internal Medicine report

Health and Wellness Informatics News

Epic Sepsis Model lacks predictive power, says JAMA Internal Medicine report

A recent report publishes the Sepsis models fails to identify two-thirds of sepsis patients and often issue false alarms.

A new study by JAMA Internal Medicine confirms Epic’s electronic health’s poor sepsis prediction model. The researchers of the University of Michigan Medical School points out the substandard predictive power of Epic. It fails to identify two-thirds of sepsis patients. Epic, however, disputes the study’s results. It also claims that the authors consciously used a hypothetical approach and refuses to take the analysis into account.

UM researchers, in their clarification, comment, “In this external validation study, we found the ESM to have poor discrimination and calibration in predicting the onset of sepsis at the hospitalization level.” On the other hand, as per Epic, the authors tended to pick up a low threshold value. This would also be adequate for a rapid response team. That is why a higher threshold value would be appropriate enough to reduce fake positives and attending nurses and physicians.   

According to the researchers, early treatment and detection of sepsis help in less mortality in hospitalized patients. ESM has been one of the widely implemented warning systems in U.S. hospitals for sepsis. The researchers pointing out this fact writes, “An improved understanding of how well the ESM performs has the potential to inform care for the several hundred thousand patients hospitalized for sepsis in the U.S. each year.”    

The model spotted only 7% of patients with sepsis that was overseen by a clinician. While, the threshold of alerting score is six or even higher. In fact, despite generating alerts on 18% of the hospitalized patients, it did not identify two-thirds of the sepsis patients.

In one of their statement, Epic remarked the purpose of the model is to find out harder-to-recognize patients. It even pointed to the previous research that showed the predictive power to be accurate and have “complete transparency” for the customers. The researchers report, national guidelines constructed by the medical professional organization should be cognizant of the broad use and make formal suggestions on their usage.


  • No comments yet.
  • Add a comment

    Join our Mailing list!

    Get all latest news, exclusive deals and updates.