Risk Factors Of Lower Extremity Deep Venous Thrombosis After Spinal Surgery And Development Of Predictive Model

Background:Lower extremity deep venous thrombosis(DVT)is one of the common complications after orthopedic surgery,and its incidence is gradually increasing.The reported incidence of DVT after spinal surgery is between 0.3%and 31%.The risk factors of venous thrombosis after spinal surgery are including age,D-dimer,blood potassium,ASA grade,low protein level,neutrophil lymphocyte ratio(NLR),etc.There is a lack of reliable assessment tools for assessing the risk of postoperative DVT in patients undergoing spinal surgery,so it is necessary to further explore risk factors and attempt to develop predictive models and assessment tools.Objective:To explore the risk factors of lower extremities deep vein thrombosis after spinal surgery,establish a prediction model,and evaluate its effect by analyzing general information,related symptoms and signs,laboratory results and other indicators.Methods:Patients who admitted to the hospital for spinal surgery from May 2022 to October 2022 were included in this project.Their general information,preoperative examination information,preoperative laboratory information,symptoms and signs were collected.They were divided into thrombosis group and non-thrombotic group.Firstly,the candidate factors were found through intergroup comparative analysis and univariate logistic regression analysis,and they were used as the predictors to be selected.Then,SMOTE was used to deal with the problem of class imbalance,and LASSO regression was used to screen the candidate indicators.The filtered variables were included in multivariate logistic regression to establish a model.Nomogram and dynamic nomogram were constructed.Finally,random forest was used for model construction,and the GINI index decline and partial dependence plot(PDP)was used to explore the influence of various factors on thrombosis.Results:A total of 169 patients were included in this study,of which 25 patients developed DVT after surgery.Comparative analysis and univariate logistic regression analysis showed that lower height,advanced age,female,shorter preoperative length of stays,high erythrocyte sedimentation rate,high fibrinogen(FIB),exist of varicose veins/reflux of lower extremities,decreased muscle strength of lower extremities,and low creatinine may be risk factors.The variables included in the multivariate logistic regression model were preoperative length of stays,age,gender,FIB,varicose veins/reflux of the lower extremities.The area under the ROC curve of the verification set is 0.747.The random forest model was established using all indicators,and the top 10 variables,based on importance,were albumin,erythrocyte sedimentation rate,weight,height,blood magnesium,glucose,D-dimer,body mass index(BMI),red blood cells,and platelet count.The area under the ROC curve of the verification set is 0.90.According to the GINI decrease and PDP,lower albumin level might be an important risk factor.Conclusions:Sex,age,height,erythrocyte sedimentation rate,FIB,creatinine,decreased lower extremity muscle strength,presence of varicose veins of the lower extremities,and shorter preoperative length of stays are potential risk factors for lower extremity DVT after spinal surgery.Both the Logistic regression model and the random forest model have good performance and can be used as prediction tools to predict the risk of lower extremity DVT after spinal surgery.