North Carolina Hospital System Reduces Sepsis Cases Using Predictive Analytics

A North Carolina hospital system is making significant strides in reducing sepsis cases by using predictive analytics to standardize clinical workflows and to create a new Sepsis Watch program. Guided by the HIMSS Electronic Medical Record Adoption Model (EMRAM), Duke Health’s Sepsis Watch program reduced their sepsis mortality rate by 31%.  

The HIMSS Electronic Medical Record Adoption Model (EMRAM) measures clinical outcomes, patient engagement and clinician use of EMR technology to strengthen organizational performance and health outcomes across patient populations.  The internationally applicable EMRAM incorporates methodology and algorithms to score a whole hospital, including inpatient, outpatient and day case services provided on the hospital campus. 

Prior to their latest EMRAM validation, Duke Health faced two challenges: false sepsis alerts from their electronic health record program and a manual risk identification process for sepsis. Duke saw over 5.5 false alerts per correct alert, and only 6.8% of patients flagged by their previous program actually had sepsis. 

Duke Health used the EMRAM Stage 7 framework to deploy the Sepsis Watch predictive analytics model, powered by machine learning inside their established EHR program, Epic. The parameters of Sepsis Watch can be adjusted and personalized to each patient, making the program more efficient and less prone to error. Duke Health used a 32-million-point dataset built from over 42,000 inpatient encounters to train the predictive program.  

The EMRAM revalidation helped Duke Health create the world’s first machine learning model for sepsis screening. Sepsis Watch now boasts a screening accuracy rate of 93% and dropped false sepsis diagnoses by 62%.  

Dr. Eugenia McPeek Hinz, Associate Chief Medical Information Officer for Duke Health said, “EMRAM recertification helped us optimize our EMR, improving our patient care and the experience of our clinical team.” 

Learn more about this successful EMRAM implementation and its benefits for Duke Health and patients by downloading the full case study.