Subphenotypes Of Infection In The Emergency Department: Clustering Analysis From Routine Parameters To Predict Early...

Area of Science:

• Emergency Medicine
• Critical Care Medicine
• Data Science in Healthcare

Background:

• Accurate prediction of sepsis-related deterioration in the emergency department (ED) is challenging due to limitations of current scoring systems and biomarkers.
• Identifying distinct patient subgroups (subphenotypes) is crucial for personalized risk stratification and timely intervention.

Purpose of the Study:

• To identify distinct sepsis subphenotypes using unsupervised hierarchical cluster analysis based on routine variables.
• To characterize these phenotypes by their clinical features, early deterioration risk, and prognostic outcomes.

Main Methods:

• Retrospective analysis of 965 patients presenting to the ED with suspected infection.
• Unsupervised hierarchical clustering applied to routine clinical variables to identify patient phenotypes.
• Assessment of clinical deterioration within 72 hours, defined by a composite outcome including SOFA score increase, shock, respiratory failure, or death.

Main Results:

• Three distinct phenotypes were identified among patients with infection and sepsis.
• Phenotype 1 (19% of patients) showed the highest rate of early deterioration (32%) with severe organ dysfunction.
• Phenotype 3 (25% of patients) comprised younger individuals with fewer comorbidities and a lower deterioration rate (6%).

Conclusions:

• Sepsis exhibits significant heterogeneity in the ED, with distinct phenotypes identifiable through routine data.
• These identified phenotypes possess unique clinical and prognostic profiles, offering a basis for phenotype-driven risk stratification.
• This approach may enhance clinical management strategies for patients with sepsis in the emergency setting.