Sepsis Prediction in Hospitalized Children: Model Development and Validation

Area of Science:

• Pediatric critical care medicine
• Clinical informatics
• Health services research

Background:

• Early recognition and treatment of pediatric sepsis are crucial for improving patient outcomes.
• While many pediatric sepsis cases are identified in the emergency department, some patients develop sepsis during hospitalization.
• There is a need for improved sepsis recognition within the inpatient setting.

Purpose of the Study:

• To develop and validate a prediction model for pediatric sepsis in the inpatient setting.
• To enhance the early recognition of sepsis in hospitalized children.
• To improve clinical decision-making for pediatric sepsis management.

Main Methods:

• A prediction model was developed using data from 2012-2018, incorporating variables from an existing model.
• A 2-tier threshold (Alert and Aware) was established using a precision-recall curve.
• The model was prospectively validated in the electronic health record during 2019.

Main Results:

• The final model included 13 variables and demonstrated strong performance with an area under the curve of 0.96 in the validation set.
• The 'Aware' tier achieved 100% sensitivity, increasing situational awareness.
• The 'Alert' tier had a positive predictive value of 14% (number needed to alert of 7), balancing sensitivity and specificity.

Conclusions:

• A 2-tiered prediction model for inpatient pediatric sepsis was successfully derived and validated.
• The model is designed to enhance situational awareness and minimize false alerts, facilitating prompt bedside evaluation.
• The model was integrated into the electronic health record as clinical decision support.