Are there more cases of a given disease because something is causing more, or because doctors have become more aware of that disease? A recent paper in JAMA tackles this question for sepsis, the often deadly response to infection that is the most expensive condition treated in US hospitals.
Researchers from several academic medical centers, including Emory, teamed up to analyze sepsis cases using two methods. The first is based on the ICD (International Classification of Diseases) codes recorded for the patient’s stay in the hospital, which the authors refer to as “claims-based.” The second mines electronic medical record (EHR) data, monitoring the procedures and tests physicians used when treating a patient. The first approach is easier, but might be affected by changing diagnosis and coding practices, while the second is not possible at every hospital.
“This project was undertaken by several large, high quality institutions that have the ability to well characterize their sepsis patients and connect their EHR data,” says Greg Martin, MD, who is a co-author of the JAMA paper along with David Murphy, MD, PhD. The lead author, Chanu Rhee, MD, MPH, is from Brigham and Women’s Hospital, and the entire project was part of a Prevention Epicenter program sponsored by the Centers for Disease Control and Prevention.
“Sepsis is challenging to track because it is a complex syndrome without a single confirmatory diagnostic test,” says Rhee. “Tracking sepsis using hospital claims data is problematic because sepsis tends to be under-recognized by clinicians, while coding can be influenced by reimbursement and policy incentives. Our research shows that widespread sepsis surveillance using clinical data is feasible and correlates well with expert physicians’ diagnoses.”
Using the EHR data, sepsis was identified if a patient had concurrent indicators of infection (blood culture draws and antibiotic prescribing) and organ dysfunction (initiation of vasopressors, mechanical ventilation, and/or changes in laboratory tests).
Analysis of claims-based data would suggest that the number of sepsis cases per year has been increasing slowly since 2009, while mortality has been going down more rapidly. The authors studied more than 173,000 adult sepsis cases in US hospitals from 2009 to 2014, and took a closer look at some selected 510 cases with chart review. Putting these results together and applying them to the United States as a whole, they estimate in 2014 there were 1,700,000 cases of sepsis and 270,000 deaths with sepsis.
They concluded that EHR-based clinical data provide more objective estimates than claims-based data for sepsis surveillance, and that the proportion of hospitalizations involving sepsis has actually been holding steady at 6 percent. Sepsis occurred in more than a third of hospitalizations that resulted in a death.
Here at Emory, Martin holds several posts; for this, the most relevant is that he is Director of Research for the Emory Center for Critical Care, while Murphy is the Director of Quality.