Predicting tracheostomy in multiple injured patients with severe thoracic injury (AIS ≥ 3) with the new T(3)P-Score: a multivariable regression prediction analysis

Multiple trauma patients with severe chest trauma are at increased risk for tracheostomy. While the risk factors associated with the need for tracheostomy are well established in the general critical care population, they have not yet been validated in a cohort of patients suffering severe thoracic trauma. This retrospective cohort study analysed data on patients aged 18 years or older who were admitted to one of the six participating academic level I trauma centres with multiple injuries, including severe thoracic trauma (AIS(Thorax) ≥ 3) between 2010 and 2014. A multivariable binary regression was used to identify predictor variables for tracheostomy and to develop the Tracheostomy in Thoracic Trauma Prediction Score (T(3)P-Score). The study included 1019 adult thoracic trauma patients, of whom 165 underwent tracheostomy during their intensive care unit (ICU) stay. Prehospital endotracheal intubation (adjusted OR [AOR]: 2.494, 95% CI [1.412; 4.405]), diagnosis of pneumonia during the ICU stay (AOR: 4.374, 95% CI [2.503; 7.642]), duration of mechanical ventilation (AOR: 1.008/hours of intubation, 95% CI [1.006; 1.009]), and an AIS(Head) ≥ 3 (AOR 1.840, 95% CI [1.039; 3.261]) were independent risk factors for tracheostomy. Patients with sepsis had a lower risk of tracheostomy than patients without sepsis (AOR 0.486, 95% CI [0.253; 0.935]). The T(3)P-Score had high predictive validity for tracheostomy (ROC(AUC) = 0.938, 95% CI [0.920, 0.956]; Nagelkerke’s R(2) was 0.601). The T(3)P-Score’s specificity was 0.68, and the sensitivity was 0.96. The severity of thoracic trauma did not predict the need for tracheostomy. Follow-up studies should validate the T(3)P-Score in external data sets and study the reasons for the reluctant use of tracheostomy in patients with severe thoracic trauma and subsequent sepsis. Trial registration: The study was applied for and registered a priori with the respective ethics committees.