Model for predicting short-term mortality of severe sepsis

Area of Science:

• Critical Care Medicine
• Prognostic Modeling
• Sepsis Research

Background:

• Severe sepsis poses a significant mortality risk in intensive care units (ICUs).
• Predicting 14-day mortality is crucial for patient management and resource allocation.
• Understanding factors influencing mortality based on infection source and episode number is vital.

Purpose of the Study:

• To develop and validate a prognostic model for predicting 14-day mortality in ICU patients with severe sepsis.
• To analyze mortality predictors considering infection acquisition site and sepsis episode rank.
• To provide a tool for early risk stratification in severe sepsis cases.

Main Methods:

• Prospective, multicenter observational study using the OUTCOMEREA database.
• Inclusion of 2268 patients with 2737 severe sepsis episodes across 12 ICUs.
• Development of a predictive model using a training cohort and validation in a separate cohort, considering data at sepsis onset.

Main Results:

• Key predictors of 14-day mortality identified: logistic organ dysfunction, septic shock, sepsis episode rank (higher ranks increased risk), multiple infection sources, Simplified Acute Physiology Score II, McCabe score, and number of chronic comorbidities.
• The prognostic model demonstrated good validity with an AUC-ROC of 0.76.
• The model proved effective across all sepsis episodes, regardless of acquisition site or episode number.

Conclusions:

• A prognostic model utilizing easily obtainable variables effectively predicts 14-day mortality in ICU patients with severe sepsis.
• The model is applicable to the first to fourth episodes of severe sepsis, including community-, hospital-, and ICU-acquired infections.
• This tool can aid clinicians in risk-stratifying patients and tailoring interventions for severe sepsis.