Variations in Case-Mix-Adjusted Duration of Mechanical Ventilation Among ICUs

Area of Science:

• Critical Care Medicine
• Health Services Research
• Biostatistics

Background:

• Mechanical ventilation is a critical intervention in intensive care units (ICUs).
• Variability in mechanical ventilation duration exists across ICUs, impacting patient outcomes and resource utilization.
• Predictive models are needed to standardize comparisons and identify performance differences.

Purpose of the Study:

• To develop and validate a model predicting mechanical ventilation duration using initial patient data.
• To compare observed versus expected mechanical ventilation duration across ICUs.
• To identify ICUs with significantly better or poorer performance regarding ventilation duration.

Main Methods:

• Retrospective cohort analysis of 56,336 ICU patients from 86 ICUs in 48 U.S. hospitals.
• Development of a multivariable logistic regression model using day 1 patient characteristics.
• Calculation of observed minus expected mechanical ventilation duration to assess ICU performance.

Main Results:

• The predictive model, primarily driven by diagnosis and physiologic abnormalities, explained 21.6% of variance in individual patient duration.
• Across ICUs, the model explained 69.9% of variance, with a mean observed minus expected duration of 3.8 hours.
• 76.7% of ICUs had observed durations within 0.5 days of predicted; 14 ICUs showed significantly better performance, and 5 showed significantly poorer performance.

Conclusions:

• A case-mix-adjusted model accurately assesses and compares mechanical ventilation duration across ICUs.
• The model cannot reliably predict individual patient ventilation duration.
• Significant inter-ICU differences in ventilation duration warrant further investigation into unit practices and care processes.