Abstract
BACKGROUND: Predicting weaning outcomes in mechanically ventilated patients remains challenging. While mechanical power (MP) reflects ventilatory load, its accuracy is limited by variability in lung compliance. This study evaluated the predictive value of MP normalized to dynamic lung compliance (Cdyn-MP) for weaning outcomes and prognosis. METHODS: A retrospective observational study was conducted in two ICUs in Jiangsu, China. Patients receiving invasive ventilation ≥ 24 h and undergoing a first spontaneous breathing trial (SBT) via T-tube were included. Patients with accidental extubation, neuromuscular disorders, or without available medical records were excluded. Respiratory mechanics data prior to SBT were retrospectively extracted from medical records. Associations and predictive performance were analyzed using restricted cubic spline, Spearman correlation, and time-dependent receiver operating characteristic (ROC) curves. RESULTS: Among enrolled patients, 67.7% were successfully weaned, while 32.3% failed. A 100-unit increase in Cdyn-MP was associated with a 32% increased risk of weaning failure (p < 0.001). Cdyn-MP positively correlated with adverse outcomes. Time-dependent ROC curves yielded AUC values of 0.913, 0.771, 0.733, and 0.702 for predicting weaning failure within ≤ 1, ≤3, ≤ 7, and > 7 days after initial SBT. CONCLUSION: Cdyn-MP was significantly associated with weaning outcomes and prognosis in mechanically ventilated patients. Its routine evaluation may enhance prediction accuracy and support traditional weaning assessments.