Influencing factors for tracheostomy in patients with acute traumatic C3-C5 spinal cord injury and acute respiratory failure

BACKGROUND: Patients with traumatic spinal cord injury (SCI) at C3-C5 have a wide range of tracheostomy rates (27%-75%), and the influencing factors for tracheostomy remain unclear. We conducted a retrospective case-control study to identify the influencing factors for tracheostomy in this subset of patient population. METHODS: A total of 101 acute traumatic C3-C5 SCI patients with acute respiratory failure requiring translaryngeal intubation and invasive mechanical ventilation (IMV) for more than 48 hours were identified and divided into the no tracheostomy (No-TCO, n = 59) and tracheostomy group (TCO, n = 42) groups. Clinical data were retrospectively reviewed and analyzed. RESULTS: Compared with the No-TCO patients, the TCO patients had a higher proportion of C3 level injury, lower Glasgow Coma Scale (GCS), and lower blood hemoglobin levels at admission. During the first weaning attempt, the TCO patients had lower levels of maximal inspiratory pressure, maximal expiratory pressure, and minute ventilation but had a higher level of rapid shallow breathing index (RSBI). The TCO patients had longer durations of IMV, ICU stay, and hospitalization compared with the No-TCO patients. Moreover, due to prolonged IMV, the TCO patients had a higher incidence of complications, including ventilator-associated pneumonia, bacteremia, urinary tract infection, and acute kidney injury compared with the No-TCO patients. Multivariate logistic regression analysis revealed that low GCS at admission and high initial RSBI were independent risk factors for tracheostomy. Importantly, a combination of these two influencing factors synergistically increased the odds ratio for tracheostomy. CONCLUSION: Low GCS at admission and high initial RSBI are two independent influencing factors that synergistically impact tracheostomy in our patients. These findings are helpful for making the decision of performing tracheostomy in this subset of patient population.