What Is the National Burden of Sepsis in U.S. Emergency Departments? It Depends on the Definition*
A new study by Wang et al (2) in this issue of Critical Care Medicine is an important step forward. The investigators used nationally representative data from the National Hospital Ambulatory Medical Care Survey to estimate sepsis burden among adults in U.S. EDs. The investigators classified patients as having sepsis if they had “serious infection” (defined as fever, hypothermia, or an infection diagnosis code) and concurrent organ dysfunction (defined using organ dysfunction diagnosis codes, measured hypotension at triage, or endotracheal intubation). They also estimated sepsis using quick Sequential Organ Failure Assessment (qSOFA) criteria. The qSOFA estimate included patients with serious infections (same criteria as above) and greater than or equal to 2 qSOFA points at triage (systolic blood pressure ≤ 100 mm Hg, respiratory rate ≥ 22, or Glasgow Coma Scale ≤ 14). The investigators estimated 670,000 cases of sepsis annually using their primary criteria and 320,000 cases using qSOFA criteria.
All told, almost 850,000 patients met one or both of the study definitions for sepsis, corresponding to about one in every 120 adult ED visits. More than half these patients arrived via Emergency Medical Services (EMS). This has important implications for improving sepsis care. We know from other studies that prehospital providers spend an average of 45 minutes with septic patients and that each hour delay in antibiotic administration relative to initial EMS contact is associated with increased mortality (3, 4). Improving sepsis recognition and rapid antibiotic administration during these precious early minutes may be a way to improve patient outcomes.
The study by Wang et al (2) has several strengths. First, it confirms the large burden of sepsis in EDs nationally. Second, it demonstrates the importance of using clinical data in addition to claims for sepsis surveillance—adding clinical criteria for hypotension, endotracheal intubation, and qSOFA in addition to claims data more than doubled the study’s estimated count of patients with organ dysfunction. Third, it affirms that qSOFA criteria may miss some kinds of organ dysfunction in patients with possible infection (5). qSOFA criteria were present in less than half of all sepsis cases in this study. Mortality rates were similar for patients with suspected infection regardless of whether or not they met qSOFA criteria, though, suggesting that non-qSOFA organ dysfunction is still important to detect.
The study also highlights many of the practical challenges inherent in sepsis surveillance. The most pressing question is, “how accurate were the study’s sepsis criteria?” Sepsis is a heterogeneous condition that lacks a gold standard and is variably diagnosed by clinicians even when they use common definitions (6). The current study relied on proxy markers for infection and organ dysfunction, magnifying uncertainty. About a third of “serious infections” were identified on the basis of fever or hypothermia alone. Temperature is neither sensitive nor specific for infection so some of these cases may have been misclassified. Furthermore, relying on clinician diagnoses, particularly in the ED where many patients are still undifferentiated, is fraught with uncertainty. Indeed, up to 40% of patients admitted to the ICU with a sepsis diagnosis do not have clear evidence of infection in retrospect (7).