How Orlando Health Deploys Predictive Analytics to Fight Sepsis
By Brian Zimmerman of Becker’s Healthcare.
As a result of the prevalence of sepsis in the healthcare setting and new reimbursement models that reward value and outcomes, many health systems and hospitals are looking for ways to take full advantage of all technological resources available in order to protect patients and reduce costs by working to properly manage and prevent sepsis.
While CMS assesses hospital performance by monitoring 30-day readmissions following hospitalizations for heart attack, heart failure, chronic obstructive pulmonary disease and pneumonia, none of these conditions take the same toll on hospitals and their patients in terms of readmissions as sepsis, according to a recent study published in JAMA. In the study, researchers found sepsis — a condition caused by the body’s immune response to life-threatening infection — accounted for 12.2 percent of readmissions, followed by 6.7 percent for heart failure, 5 percent for pneumonia, 4.6 percent for COPD and 1.3 percent for heart attack.
During a Jan. 26 webinar sponsored by VigiLanz — a provider of real-time intelligence and predictive analytics — and hosted by Becker’s Hospital Review, Eric Rose, Pharm. D, manager in corporate clinical decision support at Orlando (Fla.) Health, Adam Klass, chief technology officer with VigiLanz, and Bart Abban, PhD, director of analytics and data science with VigiLanz, discussed the development of predictive models for sepsis risk and how such models can improve outcomes by alerting physicians to the early signs of sepsis and enabling them to intervene sooner.
The VigiLanz intelligence platform conducts automated, continuous surveillance to provide clinicians with solutions and insights in real-time to ensure adherence to best practices, increase patient safety and improve reimbursement.
“We’re a layer that sits on top of the EMR and really our goal is to drive all of the value out of all of that great data that’s in the EMR and really affect patient care in real time,” said Mr. Klass during the webinar.
The VigiLanz platform is designed to deliver clinicians predictive analytics tailored to a hospital’s patient population. Patient attributes like age, clinical history, comorbidities and lab results can be continuously introduced into the model. The VigiLanz model also analyzes the potential effects of a patient’s unique history and attributes on their clinical outcomes.
Additionally, models are constructed from the ground up through collaboration between the VigiLanz team and the hospital, tested against a hospital’s historical data and validated within a staging environment.
“We have a very collaborative process between your experts and the VigiLanz team,” said Dr. Abban, during the discussion. “Based on the literature available, we’ve built great base models. When we begin working with [a hospital’s] team, we look at the important differences that are in your institution, things that might impinge on the foundational models and we incorporate those into the models.”
The pilot model designed to predict the risk of sepsis in patients at Orlando Health — a six-campus health system with 1,700 beds and a Level I trauma center — involves a multi-layer process. First, efforts are made to accurately identify patients already presenting sepsis upon admission. Second, patient data like vital signs, microbiology and clinical history are entered into the model. Then the platform conducts a clinical evaluation and attributes a sepsis risk score to the patient. Based on the score, hospitals can decide when medical interventions are warranted and more effectively triage the patient population.
At Orlando Health, the VigiLanz model has a positive sepsis predication rate of 70 percent and has been particularly well received in the med-surgery units.
Additionally, Orlando Health’s Dr. Rose and the VigiLanz team were able to establish more accurate time frames for organ dysfunction pertaining to sepsis. According to Dr. Rose, the newly established time frames “helped us very quickly tighten up the predictive power of the algorithm.”
“It’s really been a pleasure to partner with them [VigiLanz] on this new venture in predictive analytics,” said Dr. Rose.
To view the webinar, click here.
Article reposted with permission from Becker’s Healthcare. See the original here.