Mechanical power and short-term mortality in critically ill patients with ARDS on mechanical ventilation: Insights from the MIMIC-IV database

Abstract

Mechanical power (MP), or an integrated measure of ventilatory stress, has been proposed as an important determinant of ventilator-induced lung injury and clinical outcomes. However, its prognostic significance in patients with acute respiratory distress syndrome (ARDS) remains incompletely understood. We selected adult patients with Berlin ARDS from MIMIC-IV v3.1. Cox proportional hazard regression models were built to explore the relationship between MP quartiles and in-hospital, 28-day, and 90-day mortality, with progressive adjustment for demographics, severity scores, and ventilatory parameters. We also explored model discrimination and calibration, conducted comparative receiver-operating characteristic curves (ROC) and restricted cubic spline (RCS) analyses, and performed Kaplan-Meier, subgroup, and sensitivity analyses as complementary assessments. A total of 1878 patients were included. Each increase in MP was significantly associated with higher in-hospital, 28-, and 90-day mortality. Using full models, Q4 had significantly higher risk than Q1 (HR 1.32; 95% CI 1.06–1.64). MP had stable discriminative performance with good calibration. Using ROC, MP had significantly better performance than traditional predictors. Models built with RCS showed no evidence for curvature. Survival analysis showed successively decreased probabilities with increasing quarters from Q1 to Q4. Sensitivity analyses were largely consistent, with subgroup analyses supporting findings. In patients with ARDS, elevated MP was independently associated with in-hospital, 28-day, and 90-day mortality. These findings suggest that MP may serve as an integrated prognostic marker of ventilatory load rather than a standalone target. Our results warrant further prospective research to determine whether incorporating MP into lung-protective strategies can improve clinical outcomes.