Pediatric Critical Care Fellow Children's Hospital Colorado Denver, Colorado, United States
Background: Pediatric critical care units frequently face surges brought about by seasonal illnesses and epidemics that lead to resource strain. While strain can be generally understood as a discrepancy between resources necessary to provide optimal care and demand brought about by increased patient volume and/or acuity, these factors are constantly shifting making strain difficult to define, prepare for and/or respond to. It is therefore imperative to characterize ICU strain in pediatrics and evaluate the impact of resource strain in the pediatric critical care setting. Objective: The overall objective of this study is to characterize resource strain in the pediatric critical care setting through metrics including capacity, throughput, and acuity and to identify any associations that exist with periods of PICU strain and patient outcomes. Design/Methods: This study is a retrospective observational study including all patients admitted to the Children’s Hospital Colorado PICU from September 1, 2014 through March 31, 2023. Data metrics will be used to measure indicators of critical care resource strain including capacity, throughput and acuity. ICU strain metrics during the 3 days prior to a patient’s transfer and on the day of transfer from the PICU will be considered independently and as a composite for associations with patient outcomes. Strain metrics will be characterized using median (IQR) for non-normally distributed variables and mean (SD) for normally distributed variables. This study will utilize linear and logistic regression analysis to test for associations between the defined PICU strain metrics and the outcome measures. We will identify factors associated with each outcome based on univariate analysis. We will use stepwise regression to build a multivariable model with patient variables (BMI, patient origin on admission, time of transfer from the ICU, admission year, etc) identified based on univariable analysis. The multivariable model will include known confounders such as age, presence of a complex chronic condition and severity of illness score on admission.
