In 2020, the World Health Organization (WHO) issued its first report on the global epidemiology and burden of sepsis (1). Citing data from the Global Burden of Diseases, Injuries, and Risk Factors study, WHO reported there were 48.9 million incident cases of sepsis and 11 million associated deaths in 2017, accounting for a staggering one in five deaths worldwide (2). These findings amplified recent studies from the U.S. and other high-income countries that also reported high-case counts and rising incidence rates (3,4). Case fatality rates, by contrast, have been decreasing but the rise in sepsis incidence exceeds the decrease in case fatalities leading to a net increase in overall sepsis-associated deaths. Collectively, these studies affirm that sepsis is a global health priority.

The WHO report also acknowledged, however, the many challenges associated with accurately measuring sepsis incidence and mortality as well as an urgent need for better data. Most epidemiologic studies of sepsis are based on hospital administrative data, which are limited by under-recognition of sepsis, subjectivity in making the diagnosis, evolving definitions, and changes in diagnosis and coding practices over time (5). Other studies, including the Global Burden of Disease study, have relied on death records linked to vital statistics. However, the completeness and accuracy of death certificates is also variable and their utility for identifying sepsis-related deaths is likely susceptible to many of the same biases that affect hospital-based diagnosis codes (6).

Recognizing the need for more reliable methods to track the burden of sepsis, the U.S. Centers for Disease Control and Prevention (CDC) created an “Adult Sepsis Event” (ASE) definition in 2018 that uses objective clinical data routinely found in electronic health record (EHR) systems to identify sepsis (7). ASE is modeled on the Third International Consensus Definition for Sepsis (Sepsis-3) and operationalizes sepsis detection as concurrent evidence of: 1) presumed serious infection, defined by a blood culture order and greater than or equal to 4 consecutive days of antibiotics (with fewer than 4 antibiotic days allowed if the patient dies, transitions to comfort measures or hospice, or transfers to another acute care facility) and 2) acute organ dysfunction (initiation of vasopressors or mechanical ventilation, elevated serum lactate, or clearly defined changes in creatinine, bilirubin, or platelet counts). While ASE is rooted in the Sepsis-3 framework—infection leading to concurrent organ dysfunction—it is designed to facilitate widespread, generalizable, and reproducible surveillance across diverse EHR systems. ASEs organ dysfunction criteria were named “eSOFA” to acknowledge that they were inspired by the full Sequential Organ Failure Assessment (SOFA) score but adapted to focus on structured data elements that are consistently available in most EHR databases and for ease of calculation (8).

The ASE definition was originally validated using over 500 medical record reviews from two U.S. healthcare systems (9). These validations found that, compared with sepsis discharge diagnosis codes, ASE was more sensitive (70% vs. 32%) and had similar positive predictive value (PPV) (70% vs. 75%) for identifying cases meeting Sepsis-3 criteria. The ASE was then applied to EHR data from over 400 hospitals in 2014 (encompassing approximately 10% of U.S. hospitalizations) to generate CDCs most recent national annual estimates: 1.7 million adult sepsis cases and over 250,000 associated deaths. Finally, trends analysis from 2009 to 2014 using ASE criteria suggested sepsis incidence and mortality rates were stable over this period in contrast to concurrent claims estimates which had suggested incidence rates were rising and mortality rates falling (supporting the concern that trends from administrative data are confounded by changing diagnosis and coding practices over time) (10).

While ASE represented a major advance in the science of sepsis surveillance, little is known about its generalizability to non-U.S. settings. In this issue of Critical Care Medicine, Lee et al (11) report on a retrospective ASE validation and implementation study at a 2732-bed tertiary care hospital in South Korea. The study by Lee et al (11) was conducted in two parts using similar methodology to the U.S. study by Rhee et al (9). First, investigators performed detailed medical record reviews on over 6000 adult hospitalizations and emergency department visits during a 1-week period in November 2019. They found that during this week, 247 cases met ASE criteria and 155 cases met Sepsis-3 criteria. Relative to Sepsis-3, sensitivity of ASE was 92%, specificity was 98%, negative predictive value was 99.8%, and PPV was 57%. In the second phase of the study, the investigators applied ASE criteria to all admissions to their hospital in calendar year 2020 and compared sepsis incidence and in-hospital mortality using ASE criteria vs. International Classification of Diseases, 10th Edition (ICD-10) sepsis codes. The incidence and mortality of sepsis using ASE criteria was 5.4% and 16.6%, respectively, closely mirroring the U.S. national weighted estimates of 5.9% and 15.6%. In contrast, the ICD-10 code-based sepsis incidence and mortality rates were 0.7% and 23.5%, suggesting substantial undercoding biased toward more severe cases (12). Last, the investigators found much lower month-to-month variation in sepsis incidence and mortality using ASE criteria compared with ICD-10 codes.

The most important contribution from the study by Lee et al (11) is that it demonstrates both the feasibility and benefit of utilizing ASE surveillance to measure sepsis incidence and outcomes outside the United States. This extends recent studies that have used ASE (or algorithms adapted from ASE) to measure sepsis burden and trends in Hong Kong and China (13,14) but also adds new insights gleaned from their many detailed medical record reviews. In particular, the investigators’ estimates of ASEs accuracy were slightly different from U.S. estimates (9). Sensitivity was higher (92% vs. 70%) and PPV was lower (57% vs. 70%). It is unclear, however, if these differences are explained by differences in practice patterns, patient characteristics, chart interpretations, how SOFA scores were calculated, or some combination of all these factors. Regardless, the remarkable similarity of the calculated sepsis incidence and mortality rates in this South Korean hospital and U.S. hospitals supports the external validity of ASE criteria to this setting.

The implications of these findings are clear and important: it suggests the potential for ASE surveillance to facilitate standardized international comparisons of sepsis incidence and mortality across countries, a goal that has so far proven to be elusive. Relatedly, the study by Lee et al (11) also affirms the inadequacy of using ICD codes for tracking the epidemiology of sepsis, particularly in countries (like South Korea) where there are no financial incentives to diligently diagnose, document, and code for sepsis, as in the United States. Although PPV of ICD-10 codes was high (84%) in the study by Lee et al (11), the sensitivity was only 16%, and the incidence of ICD-10-based sepsis was implausibly eight-fold lower than based on ASE while mortality rates were higher. The high month-to-month variation in the incidence of ICD-based sepsis, compared with the stable incidence by ASE criteria, also underscores the risk of additional variability that can arise from differences in sepsis diagnosis and coding practices between providers, hospitals, and countries (15).

A final important contribution of the study by Lee et al (11) comes from its detailed analysis of ASE false positives and false negatives relative to Sepsis-3 criteria, which in turn point toward areas for improvement for future iterations of ASE. Over half of the false positives stemmed from cases where no infection was present in retrospect despite blood cultures being drawn and greater than or equal to 4 days of antibiotics being administered. This suggests that other components may need to be incorporated to increase the PPV of ASEs infection criteria, such as requiring longer courses of antibiotics, positive microbiologic tests, diagnosis codes for infection, or a combination of these factors. A less common but still important cause of false positives was triggering eSOFA criteria based on an elevated lactate level but without meeting 2 SOFA points. This suggests that gains in the performance of ASE could also be achieved by improving the match between eSOFA and SOFA criteria. Any potential changes to ASE criteria, however, will need to be carefully evaluated for tradeoffs in sensitivity, specificity, complexity, and feasibility.

The study by Lee et al (11) does have important limitations, most notably its focus on just a single tertiary care hospital in South Korea, validating ASE using just 1 week’s worth of cases (potentially introducing a seasonal bias), and the use of just one calendar year that coincided with the start of the COVID-19 pandemic to estimate sepsis incidence and mortality rates. More broadly, it is also clear that ASE surveillance cannot be used in its current form in low-income countries and settings where practice patterns (e.g., thresholds to draw blood cultures, administer antibiotics, administer vasopressors and mechanical ventilation, or check lactates and other laboratory tests) may differ substantially from practices in high-income countries. Improving our ability to conduct sepsis surveillance in low- and middle-income settings remains an important area for future research.

In summary, the study by Lee et al (11) is an important step forward in demonstrating the utility of ASE surveillance in settings beyond the United States while also highlighting areas where its performance can be improved. This has the potential to provide clinicians, researchers, public health officials, and policy makers with more reliable data on the epidemiology of sepsis across diverse countries and facilitate global improvements in its prevention, detection, treatment, and outcomes.

1. World Health Organization: Global Report on the Epidemiology and Burden of Sepsis: Current Evidence, Identifying Gaps and Future Directions. 2020. Available at: https://iris.who.int/bitstream/handle/10665/334216/9789240010789-eng.pdf?sequence=1. Accessed March 11, 2024 2. Rudd KE, Johnson SC, Agesa KM, et al.: Global, regional, and national sepsis incidence and mortality, 1990-2017: Analysis for the Global Burden of Disease Study. Lancet. 2020; 395:200–211 3. Martin GS, Mannino DM, Eaton S, et al.: The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med. 2003; 348:1546–1554 4. Gaieski DF, Edwards JM, Kallan MJ, et al.: Benchmarking the incidence and mortality of severe sepsis in the United States. Crit Care Med. 2013; 41:1167–1174 5. Shappell CN, Klompas M, Rhee C: Surveillance strategies for tracking sepsis incidence and outcomes. J Infect Dis. 2020; 222(Suppl 2):S74–S83 6. McGivern L, Shulman L, Carney JK, et al.: Death certification errors and the effect on mortality statistics. Public Health Rep. 2017; 132:669–675 7. Centers for Disease Control and Prevention: Hospital Toolkit for Adult Sepsis Surveillance. 2018. Available at: https://www.cdc.gov/sepsis/pdfs/Sepsis-Surveillance-Toolkit-Mar-2018_508.pdf. Accessed March 11, 2024 8. Rhee C, Zhang Z, Kadri SS, et al.; CDC Prevention Epicenters Program: Sepsis surveillance using adult sepsis events simplified eSOFA criteria versus sepsis-3 sequential organ failure assessment criteria. Crit Care Med. 2019; 47:307–314 9. Rhee C, Dantes R, Epstein L, et al.; CDC Prevention Epicenter Program: Incidence and trends of sepsis in US hospitals using clinical vs claims data, 2009-2014. JAMA. 2017; 318:1241–1249 10. Rhee C, Klompas M: Sepsis trends: Increasing incidence and decreasing mortality, or changing denominator? J Thorac Dis. 2020; 12(Suppl 1):S89–S100 11. Lee SY, Park MH, Oh DK, et al.: Validation of adult sepsis event and epidemiologic analysis of sepsis prevalence and mortality using adult sepsis event’s electronic health records-based sequential organ failure assessment criteria: A single-center study in South Korea. Criti Care Med. 2024; 52:1173–1182 12. Whittaker SA, Mikkelsen ME, Gaieski DF, et al.: Severe sepsis cohorts derived from claims-based strategies appear to be biased toward a more severely ill patient population. Crit Care Med. 2013; 41:945–953 13. Ling L, Zhang JZ, Chang LC, et al.: Population sepsis incidence, mortality, and trends in Hong Kong between 2009-2018 using clinical and administrative data. Clin Infect Dis. 2023:ciad491 14. Dong R, Tian H, Zhou J, et al.; China Critical Care Clinical Trials Group (CCCCTG): External validity of adult sepsis event’s simplified eSOFA criteria: A retrospective analysis of patients with confirmed infection in China. Ann Intensive Care. 2020; 10:14 15. Rhee C, Jentzsch MS, Kadri SS, et al.; Centers for Disease Control and Prevention (CDC) Prevention Epicenters Program: Variation in identifying sepsis and organ dysfunction using administrative versus electronic clinical data and impact on hospital outcome comparisons. Crit Care Med. 2019; 47:493–500