BACKGROUND
The Third International Consensus Definitions for Sepsis (Sepsis-3) present clinical criteria for classifying infection patients. Event incidence and long-term outcomes across these classifications are unknown.
METHODS
We conducted a retrospective analysis using data from 30,239 US participants aged ≥45 years enrolled in the REasons for Geographic and Racial Differences in Stroke (REGARDS) cohort. We identified hospitalizations over 2003–2012 using three classifications: infection with systemic inflammatory response syndrome (SIRS), elevated sepsis-related organ failure assessment score (SOFA), or elevated “quick” SOFA (qSOFA) score. We examined incidence over the study period, in-hospital mortality, and one-year mortality, providing estimates and 95% confidence intervals (CIs).
FINDINGS
Among 2,593 infection events, there were 1,526 sepsis-SIRS, 1,080 sepsis-SOFA, and 378 sepsis-qSOFA events. Sepsis-SIRS events had the highest incidence (8.2 per 1,000 person-years; 95% CI 7.8–8.7), followed by sepsis-SOFA (5.8; CI 5.4–6.1) and sepsis-qSOFA (2.0; CI 1.8–2.2). In-hospital mortality was higher for sepsis-qSOFA (67/295; 22.7%) than sepsis-SOFA (125/960; 13.0%) and sepsis-SIRS (128/1,392; 9.2%). One-year mortality post-discharge was also highest for sepsis-qSOFA (29.4 per 100 person-years; CI 22.3–38.7) compared to sepsis-SOFA (22.6; CI 19.2–26.6) and sepsis-SIRS (14.7; CI 12.5–17.2).
INTERPRETATION
The examined classifications identified different incidence and mortality. Our findings support the use of the sepsis-SOFA and sepsis-qSOFA classifications for identification of infection patients at elevated risk of poor outcomes. We also demonstrate how these classifications could be used in future epidemiological assessments and studies of patients with infection.
FUNDING
National Institute for Nursing Research (R01-NR012726), National Center for Research Resources (UL1-RR025777), and National Institute of Neurological Disorders and Stroke (U01-NS041588). JPD held an Agency for Healthcare Research and Quality predoctoral fellowship (T32-HS013852). MMS was supported by a mentoring award from the National Heart, Lung and Blood Institute (K24-HL111154).