Related Experiment Video
Updated: Sep 20, 2025
05:28
Evaluation of a Reliable Biomarker in a Cecal Ligation and Puncture-Induced Mouse Model of Sepsis
Published on: December 9, 2022
3.8K
Sepsis Prediction Models are Trained on Labels that Diverge from Clinician-Recommended Treatment Times
Gary E Weissman1, Rebecca A Hubbard1, Blanca E Himes1
1University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA.
Summary
Clinician sepsis treatment timing differs from the Sepsis-3 definition. Predicting Sepsis-3 onset may delay critical antibiotic administration, impacting patient care and clinical decision support systems for sepsis.
Area of Science:
- Medical Informatics
- Critical Care Medicine
- Clinical Decision Support
Background:
- Sepsis prediction models often utilize the Sepsis-3 definition for training.
- Limited evidence exists for the clinical utility of deployed sepsis prediction models at the bedside.
- A potential disconnect between Sepsis-3 timing and clinical practice may explain this limitation.
Purpose of the Study:
- To investigate the discrepancy between clinician-recommended sepsis treatment initiation and the Sepsis-3 definition onset.
- To evaluate the potential impact of Sepsis-3 based alerts on timely sepsis treatment.
Main Methods:
- An electronic survey was administered to 153 clinicians across three diverse medical centers.
- Clinicians reviewed vignettes from eight real sepsis cases.
- Participants recommended antibiotic treatment start times based on the provided case information.
Main Results:
- Clinicians recommended initiating antibiotic treatment an average of 7.0 hours *before* the Sepsis-3 defined onset.
- The 95% confidence interval for this divergence was 5.3 to 8.8 hours.
Conclusions:
- Sepsis prediction models based solely on Sepsis-3 onset may lead to delayed or inappropriate treatment recommendations.
- Aligning predictive decision support systems with actual bedside clinical decisions is crucial for enhancing utility.
- Future systems should aim to predict outcomes that support clinician judgment for timely sepsis management.
