Predicting & Treating Sepsis

Today, sepsis affects more than 1.7 million people and takes 270,000 lives each year. Sepsis is the body’s overwhelming response to infections which may lead to tissue damage, organ failure, amputations, and death.

The Sepsis Alliance, a sepsis patient advocacy non-profit organization, has launched the” Sepsis Coordinator Network” The network provides healthcare professionals with evidence-based best practice resources and guidance to improve outcomes for patients with sepsis.

Edwards Lifesciences https://www.edwards.clom developed the network with an advisory committee of nursing leaders with experience developing lifesaving sepsis protocols and experience working with septic patients. Advisors consisting of a team of healthcare professionals from across the continuum of care will serve as clinical advisors in specialized areas.

Jeremy Weiss, MD, PhD, assistant Professor of Health Informatics at Carnegie Mellon University’s (CMU) Heinz College , reports, “The problem is that sepsis is incredibly heterogeneous. “Anyone can get sepsis from a multitude of infections at different sites and with different comorbid profiles.

Dr. Weiss and his team are working to improve the speed and accuracy to predict sepsis by using machine learning and EHR data and so far, the team has developed a method of accurately assigning risk scores.

The researchers look at information in the EHRs such as blood tests, lab tests, prescribed drugs, and vital signs. This data usually contained in the EHR can be accessed when seeing the patient and then the information can be used to make real time predictions. The team is able to apply machine learning algorithms plus the EHR data to analyze many different data points, which can help to make accurate predictions.

Other universities are also using machine learning to continuously monitor EHRs to identify patients that may get sepsis or septic shock a full 12 hours before the onset of the condition. The universities doing further research are located at the University of Pennsylvania, North Carolina State University in collaboration with Mayo Clinic, and the Christiana Care Health System in Delaware.

Other steps that need to take place include using an early warning system to go through the EHRs of every patient when they enter the hospital to see if patients can be identified that are the most susceptible to sepsis.

The Virtual Sepsis Unit at Mercy Hospital in St. Louis, categorizes inpatients by risk and monitors them continuously. In less than a year, using telesepsis, deaths from severe sepsis at the hospital dropped by fifty percent, deaths from septic shock dropped by 60 percent, and the average length of stay in the ICU decreased by more than four days.

Research is ongoing at the Intermountain Medical Center to do outpatient follow-up for sepsis patients treated and released from Emergency Departments (ED) to see if they are experiencing good outcomes.

“We have found that many more ED patients with sepsis are discharged from the ED than previously recognized”, reports Ithan Peltan MD, Pulmonary and Critical Care Medicine Specialist and Researcher with the Intermountain Medical Center.

He adds, “Outpatient management of sepsis is not automatically wrong, but wide variations in care provided by different physicians suggests that there is room to identify, develop, and test tools that clinicians could use effectively to guide and optimize sepsis triage decisions.”

The Joint Commission Center for Transforming Healthcare has been involved in several projects to reduce sepsis mortality. The Joint Commission Center is using Robust Process Improvement® methods and tools to identify causes and help researchers develop solutions to reduce sepsis mortality.

Share Button