Analysis Of Risk Factors And Development Of Predictive Scores For Tracheostomy In Traumatic Cervical Spinal Cord Injury

Part one:Risk factors for tracheostomy after traumatic cervical spinal cord injuryObjective:To identify the risk factors associated with the need for tracheostomy after traumatic cervical spinal cord injury(TCSCI).Methods:The demographic and characteristics of 340 TCSCI patients who treated in the Xinqiao Hospital of Army Medical University from 2014 to 2020 were retrospective analyzed.Patients were divided into tracheostomy group(n=41)and non-tracheostomy group(n=299).Variables were included age,gender,mechanism of injury,pre-existing lung disease,pre-existing medical conditions,American Spinal Injury Association(ASIA)Impairment Scale,the neurological level of injury(NLI),and injury severity score(ISS).Student’s t test,Chi-square analysis or Fisher’s exact test was applied to find out the difference of variables between tracheostomy group and non-tracheostomy group.Univariate logistic regression analysis was applied to identify potential risk factors associated with tracheostomy.To prevent the influence of confounding factors,multiple logistic regression analysis was used to identify risk factors for tracheostomy.Results:Of 340 patients who met the inclusion criteria,41(12.1%)patients underwent tracheostomy.There were differences in age(t=3.389,P=0.015),ISS(t=8.360,P<0.001),ASIA Scale(χ~2=100.423,P<0.001),the neurological level of injury(χ~2=62.190,P<0.001),between tracheostomy group and non-tracheostomy group.Four factors including age,ISS,ASIA Scale,and the neurological level of injury were the potential risk factors for tracheostomy by univariate logistic regression analysis.There was no significant difference in smoking history,mechanism of injury,pre-existing lung disease and pre-existing medical conditions.Age>50(OR=5.461,95%CI:1.823-16.360,P=0.002),ISS>16(OR=9.732,95%CI:3.233-29.296,P<0.001),ASIA A(OR=23.403,95%CI:7.294-75.086,P<0.001)and NLI≥C4(OR=11.668,95%CI:4.054-33.583,P<0.001)were also determined by logistic regression analysis as risk factors for tracheostomy.Conclusions:Age>50,ISS>16,ASIA A and,NLI≥C4 were identified as risk factors for the need of tracheostomy in patients with TCSCI.These risk factors would be important to assist the clinical decision of tracheostomy in patients with TCSCI.Part two:Development of Classification and regression tree(CART)model and prediction scores to assist clinical prediction for tracheostomy in patients with traumatic cervical spinal cord injuryObjective: To develop a classification and regression tree(CART)model to predict the need of tracheostomy in patients with traumatic cervical spinal cord injury(TCSCI),and to quantify scores of risk factors to make individualized clinical assessments.Methods: A retrospective study of tracheotomy was conducted in TCSCI patients admitted to Xinqiao Hospital from January 2014 to December 2020.Risk factors for tracheostomy participated in the formation of the CART model by SPSS Modeler 18.0.The software selected randomly 260 patients with TCSCI as the development set to build the tree,meanwhile,80 patients were selected as the validation set to prune the tree and cross-validation.Logistic regression analysis was used to determine the coefficients of risk factors for the need of tracheostomy.Through the relative relationship of the coefficients of each risk factor in the logistic regression equation,the corresponding assignment was given to form a predicting score for tracheostomy.The scores for predicting tracheostomy were calculated by using the simple arithmetic sum of the scores of independent risk factors present in each patient.By comparing the actual number of patients underwent the tracheostomy with the predicted number,the recognition ability of the prediction score for tracheostomy was evaluated by indicators such as sensitivity and specificity.Results: Three hundred and forty patients with TCSCI have met the inclusion criteria,of which 41 patients underwent the tracheostomy.Logistic regression analysis showed that age > 50,ISS > 16,NLI≥C4 and ASIA A were significantly associated with tracheostomy.The CART model showed that ASIA A and NLI≥C4 were at the first and second decision node,which had a significant influence on the decision of tracheostomy.The final scores for tracheostomy from CART algorithm,composed of age(+1.5),ISS(+2.0),NLI(+2.5)and ASIA A(+3.0)with a sensitivity of 0.78 and a specificity of 0.96,could also predict tracheostomy.Conclusions: The establishment of CART model provided a certain clinical guidance for the prediction of tracheostomy in TCSCI.Quantifications of risk factors enable accurate prediction of individual patient risk of need for tracheostomy.