Mechanical Ventilation after Traumatic Spinal Cord Injury-A Multicentric Cohort Study-based Prediction Model for Weaning Success: The BICYCLE Score.

Rationale: Limited information exists about the epidemiology, outcomes, and predictors of weaning from mechanical ventilation in patients with spinal cord injury. Objectives: Our aim was to investigate predictors of weaning outcomes for patients with traumatic spinal cord injury (tSCI) and develop and validate a prognostic model and score for weaning success. Methods: This was a registry-based, multicentric cohort study including all adult patients with tSCI requiring mechanical ventilation (MV) and admitted to one of the intensive care units (ICUs) of the Trauma Registry at St. Michael's Hospital (Toronto, ON, Canada) and the Canadian Rick Hansen Spinal Cord Injury Registry between 2005 and 2019. The primary outcome was weaning success from MV at ICU discharge. Secondary outcomes included weaning success at Days 14 and 28, time to liberation from MV accounting for competing risk of death, and ventilator-free days at 28 and 60 days. Associations between baseline characteristics and weaning success or time to liberation from MV were measured using multivariable logistic and competing risk regressions. A parsimonious model to predict weaning success and ICU discharge was developed and validated via bootstrap. A prediction score for weaning success at ICU discharge was derived, and its discriminative ability was assessed using receiver operating characteristic curve analysis and compared with the Injury Severity Score (ISS). Results: Of 459 patients analyzed, 246 (53.6%), 302 (65.8%), and 331 (72.1%) were alive and free of MV at Day 14, Day 28, and ICU discharge, respectively; 54 (11.8%) died in the ICU. Median time to liberation from MV was 12 days. Factors associated with weaning success were B lunt injury (odds ratio [OR], 2.96; P  = 0.010), I SS (OR, 0.98; P  = 0.025), C omplete syndrome (OR, 0.53; P  = 0.009), age in Y ears (OR, 0.98; P  = 0.003), and C ervical LE sion (OR, 0.60; P  = 0.045). The BICYCLE score showed a greater area under the curve than the ISS (0.689 [95% confidence interval (CI), 0.631-0.743] vs. 0.537 [95% CI, 0.479-0.595]; P  < 0.0001). Factors predicting weaning success also predicted time to liberation. Conclusions: In a large multicentric cohort, 72% of patients with tSCI were weaned and discharged alive from the ICU. Readily available admission characteristics can reasonably predict weaning success and help prognostication.