Validating mortality prediction
The researchers compared eSOFA and SOFA in a sample from 111 U.S. acute care hospitals to see if eSOFA had a comparable predictive validity for mortality. The analysis included 942,360 adults seen between 2013 and 2015. A total of 11.1% (104,903) had a presumed serious infection based on a blood culture order and at least 4 consecutive days of antibiotic use.
The analysis showed that 6.1% of those with infections had a sepsis event based on at least a 2-point increase in SOFA score from baseline (Sepsis-3 criteria), compared with 4.4% identified by at least a 1-point increase in eSOFA score. A total of 34,174 patients (3.6%) overlapped between SOFA and eSOFA, which represented good agreement (Cronbach’s alpha, 0.81). Compared with SOFA/Sepsis-3, eSOFA had a sensitivity of 60%, and a positive predictive value of 82%.
Patients identified by eSOFA were slightly more ill, with more requiring ICU admission (41% vs. 35%), and a greater frequency of in-hospital mortality (17% vs. 14%). Those patients who were identified by SOFA/Sepsis-3, but missed by eSOFA, had an overall lower mortality (6%).
There was a similar risk of mortality across deciles between SOFA- and eSOFA-identified sepsis patients. In an independent analysis of four hospitals from the Emory system, the area under the receiver operating characteristics was 0.77 for eSOFA and 0.76 for SOFA (P less than .001).
The Centers for Disease Control and Prevention and the Agency for Healthcare Research and Quality funded the study. Dr. Rhee and Dr. Khan have no relevant financial conflicts.
SOURCE: Rhee C et al. Crit Care Med. 2019;47(3):307-14.