Length of stay in pediatric intensive care unit: prediction model

Area of Science:

• Pediatric Intensive Care
• Medical Informatics
• Clinical Prediction Models

Background:

• Accurate prediction of pediatric intensive care unit (PICU) length of stay (LOS) is crucial for resource allocation and patient management.
• Existing models often lack precision, necessitating development of improved predictive tools.

Purpose of the Study:

• To develop and validate a predictive model for estimating the risk of prolonged length of stay in children admitted to a PICU.
• The model utilizes demographic and clinical data available at the time of admission.

Main Methods:

• Retrospective cohort study conducted in Sao Paulo, Brazil.
• Internal validation procedures were employed.
• Area under the Receiver Operating Characteristic (ROC) curve was used to assess model performance.

Main Results:

• The mean hospital stay was 2 days.
• The predictive model segmented hospital stay into 1-2, 3-4, and >4 day categories.
• Model accuracy for 3-4 days was 0.71 (65% correct), and for >4 days was 0.69 (66% correct), indicating modest predictive capability.

Conclusions:

• The developed predictive model demonstrated limited accuracy, making it insufficient for sole reliance in decision-making or discharge planning.
• Predictive models for PICU LOS based solely on admission data are constrained by their inability to incorporate in-hospital events like complications, impacting overall accuracy.