SAN DIEGO — Sepsis criteria optimized for electronic health record review could be a practical tool for epidemiologic surveillance, a retrospective study found.
In a sample of nearly 1 million patients, the 6.1% flagged as having sepsis through the current Sepsis-3 Sequential Organ Failure Assessment (SOFA) criteria had lower in-hospital mortality rates than the 4.4% flagged with the CDC’s newer Adult Sepsis Events simplified (ASE) eSOFA criteria (14.4% vs 17.1%, respectively).
However, the discrimination for mortality, measured with an area under the receiver operating characteristic curve (AUROC), was comparable across SOFA and ASE criteria (0.774 vs 0.759), reported Chanu Rhee, MD, MPH, of Harvard Medical School in Boston.
Relative to SOFA, the sensitivity of the ASE criteria was 59.7%, while the positive predictive value was 82.1%, Rhee and colleagues reported at a late-breaking clinical trial session here at the the annual Society of Critical Care Medicine meeting and simultaneously published in Critical Care Medicine.
“The idea of this is to have something that can leverage data that’s already kind of routinely recorded in the [electronic health records] and to potentially set up an automated surveillance system,” Rhee told MedPage Today, “not necessarily to influence the clinician right at that moment at the patient’s bedside, but on a whole, I think to try to understand how we are doing in terms of making progress in sepsis: Are we reducing its incidence, and are we doing better at improving its outcomes?”
SOFA scores focus entirely on organ dysfunction (through an increased SOFA score of ≥2 points from baseline), while the ASE definition avoids vital signs data and does not include evidence of CNS dysfunction, according to Steven Simpson, MD, of the University of Kansas in Kansas City, who wrote an editorial accompanying the paper.
He noted that “because there is no pathobiologic gold standard for diagnosing sepsis, the condition is recognized through clinical syndromes, whether prospectively or retrospectively, as the ASE was intended to do.”
The two criteria serve different functions, he added. While SOFA was designed for risk stratification, for example, the ASE criteria were designed for overall surveillance, not as a diagnostic system.
“The presented data indicate that a simple, objective, and automated set of criteria for identifying sepsis can effectively identify a subset of infected patients with a high risk of mortality and can do so with good predictive ability,” he wrote.
This study collected data from all adult patients (≥20 years) admitted from 2013 to 2015 at one of 111 hospitals using the Cerner EHR system included in the Cerner Health Facts dataset. The data was independently validated by hospitals in the Emory Healthcare system.
Patients were considered to have sepsis if they had a serious infection and organ dysfunction demonstrated by an increased SOFA score (≥2 points) or the presence of one or more of the ASE criteria — such as vasopressor initiation, doubling in baseline creatinine, or lactate ≥2.0 mmol/L — within 2 days of a blood culture, the authors noted.
In total, 57,242 patients met SOFA criteria, 41,618 met ASE criteria, and 34,174 met both. Characteristics across the SOFA and ASE groups were similar overall, with pneumonia (36.1% and 32.9%, respectively) and urinary tract infections (26.8% and 28.1%, respectively) being the most common infections.
However, the most common sources of organ dysfunction differed: Respiratory and neurology dysfunction were most common with SOFA-defined sepsis, whereas the ASE criteria most often picked up cases with elevated lactate and a doubling of index creatinine.
Differences in prevalence and mortality across the two models was mostly consistent across subgroups, including ones stratified by age, infectious diagnoses, and comorbidity burden, the authors reported.
However, when researchers removed increased lactate from the ASE criteria in a sensitivity analysis, the prevalence of sepsis decreased from 4.4% to 3.5%, and mortality increased from 17.1% to 19%, researchers reported.
This was “likely because the lactate criterion identifies some patients with mild hypotension, hypoxemia, or abnormal mental status that are sufficient to increase SOFA but insufficient to trigger eSOFA criteria,” they wrote.
Removing lactate also increased the positive predictive value of sepsis in the ASE model (89.2%), and no significant difference was observed in terms of the tool’s agreement with SOFA (AUROC 0.773, 95% CI 0.769-0.778), they noted.
“This suggests elevated lactate levels alone without concurrent organ dysfunction have little impact on the risk of mortality,” they wrote.
Limitations of the study included assumed baseline measures of the timing of organ dysfunction and infection for SOFA scores as well as the possibility of miscoding of infection status. Also, the study only used one dataset lacking vasopressor doses and was therefore unable to perfectly replicate SOFA scores, researchers reported. Finally, the authors noted that the CDC’s ASE has not been universally implemented.
Rhee’s institution received funding from the Agency for Healthcare Research and Quality.
Co-authors received support from the NIH, the CDC, Cheetah Medical, Grifols, and the Harvard Pilgrim Healthcare Institute.
One co-author received honoraria for lectures from Washington State Hospital Association, Dell Medical School, and Beth Israel Deaconess Hospital Plymouth.