Looking Back is Forward Thinking for Sepsis Researchers

Arcus dives deep into data for a retrospective observational study of sepsis epidemiology.

By Nancy McCann

Inspirational life quotes appear aplenty on the subject of looking ahead, moving forward, and never dwelling on the past. But to Children’s Hospital of Philadelphia researchers and co-leaders of the Pediatric Sepsis Program, Scott Weiss, MD, MSCE, and Fran Balamuth, MD, PhD, looking back is exactly what they need to do when it comes to the epidemiology of sepsis.

They teamed up with Jeff Pennington, associate vice president and chief research informatics officer at CHOP Research Institute, and his team of Arcus data scientists, to validate a pediatric surveillance algorithm relying on eight years of routine clinical data within CHOP’s electronic health record as a dependable way to identify trends in pediatric sepsis. The co-authors published findings from the retrospective observational study in Pediatric Critical Care Medicine.

“We developed a pediatric surveillance algorithm that led to identifying cases of sepsis — free from influences of changes in diagnosis or billing practices — demonstrating an objective, efficient, and reliable method for pediatric sepsis,” said Dr. Weiss, an associate professor of Anesthesiology and Critical Care at the Perelman School of Medicine in the University of Pennsylvania.

Sepsis is the body’s overwhelming inflammatory response to an infection, such as pneumonia or a urinary tract infection. The effects of this disease can be life-threatening, impairing organ function, including the heart, brain, liver, and kidneys. Sepsis shock, the most severe form of sepsis, occurs when the body’s inflammatory response affects the heart and blood vessels, causing low blood pressure, which, in turn, affects other organs. If blood flow to a limb is restricted, amputation may be necessary.

Due to the dangerous outcomes associated with sepsis, the Centers for Disease Control and Prevention has a public campaign underway to improve the quality of care around sepsis management.

The Importance of Accurate Tracking

Precisely tracking this disease over time is crucial, in order to determine whether quality improvement initiatives are successful in reducing both the development of sepsis and improving outcomes. Current methods of identification of patients who meet published consensus criteria for sepsis or septic shock include large-scale chart reviews or by using claims data/billing codes from big administrative databases; however, manual chart reviews are laborious and expensive, and claims data can vary considerably across providers and time.

Pediatric sepsis experts Drs. Weiss and Balamuth worked with Arcus medical informatics experts within CHOP’s Department of Biomedical and Health Informatics to develop the algorithm and apply it to study longitudinal trends in sepsis epidemiology more efficiently. The Arcus program is a strategic initiative of the Research Institute created to link clinical and research data, offer secure data archives, patient privacy protection, and access to clinical data on more than two million patients.

“Informatics experts in the Arcus program understand how to work with the clinical data and the intricacies of putting together the code that can turn what clinicians think about as a clinical problem into something that they can pull out of a broad medical record,” Dr. Weiss said. “These collaborative efforts are hugely important.”

Arcus Dives Deep Into Data

The algorithm revealed a rising incidence of sepsis and stable mortality from 2011 to 2019 that could not be attributable to changes in diagnosis or variation in billing practices. This distinction is critical, the authors noted, because, to date, it has not been possible to disentangle the extent to which the reported rise in sepsis reflects intensified sepsis recognition, changes to coding practices, or a true increase in the disease.

The study authors suggest CHOP’s increasingly complex cohort of patients with chronic comorbid medical conditions that put patients at risk for sepsis is the reason the Hospital is seeing a rise in the number of sepsis cases it treats. Citing the increase in survival from prematurity in the neonatal population, as an example, Dr. Weiss said, “These patients are already at high risk for developing sepsis, so as more and more of these patients survive, the downstream result is that we see these patients being readmitted for sepsis.”

The next step is to validate externally this algorithm across different hospitals to make sure it is applicable, not just at CHOP, but also at other academic children’s hospitals and community-based hospitals.

Once the researchers are able to establish its validity in that manner, potentially all hospitals in the United States and beyond could use the algorithm to track the epidemiology of sepsis, Dr. Weiss said.

“Accurate data is necessary to establish research priorities and to determine the degree to which sepsis becomes a public health priority for research funding,” Dr. Weiss said. “Also, there’s an increasing focus on benchmarking the quality of care between hospitals. If hospitals are determining who has sepsis in different ways, then that can dramatically skew the ability to benchmark hospital events.”