External validation of a widely implemented proprietary sepsis prediction model in hospitalized patients
JAMA — Wong A, Otles E, Donnelly JP, et al. | June 23, 2021
Researchers sought to evaluate the accuracy of the Epic Sepsis Model, a proprietary sepsis prediction model implemented at hundreds of US hospitals, in predicting the onset of sepsis. They conducted a cohort study of 27,697 patients undergoing 38,455 hospitalizations at Michigan Medicine, the academic health system of the University of Michigan, Ann Arbor. Occurrence of sepsis was recorded in 7% of the hospitalizations. Prediction of the onset of sepsis was made using the Epic Sepsis Model with an area under the curve of 0.63, which is substantially worse than the performance described by its developer. Observations overall suggest that the Epic Sepsis Model performs poorly in prediction of sepsis; its widespread adoption despite poor performance raises fundamental concerns about sepsis management on a national level.
Read the full article on JAMA