Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Purpose Sepsis is a heterogeneous syndrome, and prior retrospective work identified four clinically meaningful sepsis subphenotypes (A–D) using vital sign trajectories. This study prospectively validated this subphenotyping algorithm and evaluated associations with patient characteristics and clinical outcomes. Methods A prospective multicenter observational cohort study was conducted across six hospitals within Emory Healthcare (Atlanta, GA) September-December 2025. Adult emergency department patients hospitalized with sepsis defined by Sepsis-3 criteria were included. An Epic-integrated algorithm captured temperature, heart rate, respiratory rate, and blood pressure during the first 8 hours of hospitalization and assigned patients hourly to one of four predefined subphenotypes. The primary outcome was 30-day hospital mortality. Secondary outcomes included ICU admission, vasopressor use, renal replacement therapy, mechanical ventilation, inotrope use, and hospital length of stay. Multivariable regression adjusted for demographics and comorbidities compared outcomes across subphenotypes. Results Among 1,916 encounters, subphenotype distribution was Group A 16%, Group B 16%, Group C 27%, and Group D 41%. Subphenotypes demonstrated distinct physiologic patterns and differed in demographics and comorbidity burden. Similar to the retrospective model, thirty-day mortality ranged from 3.6% (Group B) to 9.2% (Group D) (p = 0.034). After adjustment, Group D had higher odds of vasopressor use (OR = 1.9, 95% CI 1.4–2.7) and 30-day mortality (OR = 1.6, 95% CI 1.0–2.5) versus Group C. Classification using only initial vital signs agreed with the 8-hour model in 59% of cases. Conclusion Real-time EHR-based classification of sepsis subphenotypes is feasible and reproducible, identifying groups with meaningful differences in outcomes, supporting future precision trial designs.
Source: Research Square
Institution: Emory University
Language: English
Subfield: Epidemiology
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7135213954