2-1. Study Design
We conducted a retrospective case-control study including adult patients with acute traumatic CSCI with or without bone injury, consecutively admitted to a single tertiary emergency medical center in Japan from January 2013 to December 2017. This study was approved by the local institutional review board of the study hospital (approval no. 2412).
2-2. Patient Selection
We screened consecutive patients aged 18–89 years with acute traumatic CSCI, who were admitted to the ED of Wakayama Medical University Hospital, and prospectively reviewed their medical records. Patients were diagnosed with CSCI by the physicians and orthopedic surgeons in the ED upon performing neurological examination and magnetic resonance imaging (MRI) after initial emergency, life-saving medical treatment. Patients with acute, traumatic injuries with lack of motor or sensory functions in the sacral segments, S4–S5, were diagnosed with complete CSCI (defined as motor-complete injury if no motor function below the zone of injury was preserved based on the American Spinal Injury Association [ASIA] impairment scale [version 2003] modified from Frankel grade) . During the study period, the ED physician intubated patients with CSCI initially before hospital admission based on the following three indications alone: airway protection, obvious respiratory/circulatory failure, and emergency preoperative procedure. The indication for cervical spine fixation surgery or tracheostomy was dependent on the experience and judgment of the orthopedic surgeon and attending intensivist without standardization.
Patients were excluded from the study as per the following criteria: central cord syndrome, injury severity score (ISS) > 41, uncertain injury level (such as concomitant severe traumatic brain injury), regular outpatient attendance with an orthopedic of our hospital, and transfer within the first 3 days of hospitalization. RE was defined as unexpected, urgent intubation for respiratory resuscitation or re-intubation within 72 h after planned extubation, including after surgery and at the time of rapid respiratory impairment (refractory desaturation, major dyspnea, and respiratory arrest). We included emergent re-intubation in the definition of RE because of the recent tendency toward an increase in early surgical decompression with planned anesthetic intubation less than 24 h post-injury . In the final analysis, we did not include patients who underwent empiric tracheostomy without any extubation attempts following pre-admission intubation since they had a decreased risk of developing RE with continuous definitive airway control during most of their hospitalization period. Finally, the selected patients with CSCI were stratified into those affected by RE (hereafter referred to as RE group) and those who were not (control group).
2-3. Outcome Measures
We prospectively collected data on baseline characteristics and the severity of traumatic injury, including age, sex, body mass index, Charlson comorbidity index , age-adjusted Charlson comorbidity index, pulmonary centrilobular emphysema in the apex of the lung on cervical computed tomography (CT), cervical ossification of posterior longitudinal ligament and thoracolumbar diffuse idiopathic skeletal hyperostosis on CT, initial Glasgow Coma Scale score, initial bradycardia (heart rate < 60 beats/min) and initial hypotension (systolic blood pressure < 90 mm Hg) in the ED; mechanism of injury (falls, motor vehicle accident, sports, or hit by an object), ISS, Abbreviated Injury Scale for the chest, concomitant lung injury (lung contusion, lung laceration defined as traumatic pneumatocele on CT, pneumothorax, or hemothorax), bony thorax injury (rib fracture or sternal fracture), and/or thoracic vertebral fracture; and ASIA impairment scale, motor-complete injury, estimated CSCI level, and injury level at and above C5. We defined emphysema in the superior sulcus on the CT scan as a decrease in lung function in heavy smokers  because data collected on smoking history were partially lacking. A board-certified radiologist retrospectively confirmed the key findings of all radiographs, CT, and MRIs to confirm the radiological evidence of baseline and injury characteristics. Regarding the patients’ treatment and clinical course, we collected the following data: intubation before hospitalization, initial admission to the ICU, incidence of copious airway secretion (CAS), atelectasis, pneumonia, cervical spine surgery, and/or tracheostomy; steroid administration, halo vest immobilization until the surgery, length of stay in the ED and hospital, dysphasia and/or ventilatory assistance at discharge, and in-hospital death. CAS was defined as retained tracheobronchial secretions attributable to a respiratory cause (acute refractory desaturation, major dyspnea, and abnormality on auscultation) and required airway (including nasotracheal) suctioning every 2 h (or more, as needed) on each patient’s daily flowsheet as recorded by the nursing staff. Atelectasis was assessed with chest radiography as the loss of lung volume, involving clinical hypoxemia and hypophonesis in the affected lung without symptoms of pulmonary infection, as interpreted by the attending physician. Pneumonia was diagnosed based on radiographic parenchymal inflammatory evidence with clinical acute fever requiring antibiotic administration. Empiric antibiotics were not administered routinely in patients with CSCI, except to prevent surgical site infections. We routinely examined the patients for CAS, atelectasis, and pneumonia during the acute 3-day phase after admission or within 24 h before RE.
The primary outcome was RE after admission in patients with traumatic CSCI. We identified and evaluated the possible risk factors contributing to RE. The secondary outcome were time from injury to onset of RE, incidence of tracheostomy, length of stay (in emergency medical center or hospital), ventilatory assistance at discharge, and in-hospital death.
2-4. Statistical Analysis
In the univariate analysis, all continuous variables were expressed as medians (interquartile ranges) and assessed using the Wilcoxon rank-sum test. Categorical variables were expressed as numbers and percentages and assessed using the two-tailed Fisher's exact probability test. P-values < 0.05 were considered statistically significant. A multivariate analysis was performed using a binomial logistic regression to determine the independent predictors of RE by calculating adjusted odds ratios (ORs) and 95% confidence intervals (CIs). In accordance with previous studies, we entered the following four variables based on clinical plausibility and availability into multivariate logistic regression analysis: "motor-complete injury" as a simple predictor based on initial evaluation of sacral sparing, "level of injury above C5" as evaluated by pre-admission neurological examination and MRI, "atelectasis" as a subjective post-admission factor that required radiographic interpretation by the physicians, and "CAS" as an objective post-admission factor mostly assessed by nursing staff at the bedside. However, we excluded ISS from the list of predictive variables for RE because of its difficulty to be clinically calculated promptly prior to admission. We also determined pneumonia in patients with CSCI as an ineligible predictor of RE because of its delayed onset (approximately 7 days) compared with atelectasis . All statistical analyses were conducted using the JMP version 13.0 software program (SAS Institute, Cary, NC, USA).