The following is a summary of “Clinical subtypes in critically ill patients with sepsis: validation and parsimonious classifier model development,” published in the February 2025 issue of Critical Care by Amstel et al. 

The use of sepsis subtypes to improve personalized medicine in individuals with critical illness was limited by inadequate validation across diverse groups and the absence of a simple classification model.  

Researchers conducted a retrospective study to validate previously identified Sepsis Endotyping in Emergency Care (SENECA) clinical sepsis subtypes in multiple large Intensive care unit (ICU) cohorts and develop simplified classifier models for δ-type adjudication in clinical practice.  

They used data from 4 cohorts between 2008 and 2023 to classify individuals meeting Sepsis-3 criteria into α, β, γ, and δ-types using clinical variables: (I) Molecular Diagnosis and Risk Stratification of Sepsis (MARS, n = 2,449), (II) MARS continuation study (MARS2, n = 2,445), (III) Dutch National Intensive Care Evaluation registry (NICE, n = 28,621), and (IV) Medical Information Mart for Intensive Care (MIMIC-IV, n = 18,661), K-means clustering was performed to determine the optimal class number and compared it with SENECA subtypes. Logistic regression models were developed in the SENECA derivation cohort to predict subtype membership and validated in MARS and MIMIC-IV.  

The results showed that among 52,226 individuals with sepsis, subtype distribution in MARS, MARS2, and NICE ranged from 2–6% for α-type, 1–5% for β-type, 49–65% for γ-type, and 26–48% for δ-type, compared to 33%, 27%, 27%, and 13% in the original SENECA cohort, MIMIC-IV had a similar distribution at 25%, 24%, 27%, and 25%, respectively. In-hospital mortality rates significantly differed for α, γ, and δ-types (P < 0.001). Method-based validation showed moderate overlap with the original subtypes in MARS and MIMIC-IV. A parsimonious model for all subtypes had 62.2% accuracy, while a model with aspartate aminotransferase, serum lactate, and bicarbonate achieved high δ-type prediction accuracy (MARS: AUC 0.93, 95% CI [0.92–0.94], accuracy 85.5% [84.0–86.8%]; MIMIC-IV: AUC 0.86 [0.85–0.87], accuracy 82.9% [82.4–83.4%]).  

Investigators concluded the distribution and mortality rates of clinical sepsis subtypes varied between US and European cohorts, and a 3-variable model accurately identified δ-type sepsis. 

Source: ccforum.biomedcentral.com/articles/10.1186/s13054-025-05256-3