Risk Factors Of New Vertebral Fracture After PVP And Establishment Of Nomogram Model

Objective: To analyze the relevant clinical variables of Osteoporotic Vertebral Compression Fracture(OVCF)in clinic,and to explore the risk factors of Vertebral Fracture recurrence,so as to establish a clinical Nomogram model suitable for new adjacent Vertebral Fracture after Vertebral molding.Methods: To collect the data of clinical OVCF patients operated vertebroplasty in the hospital orthopedic from January2018 to December 2019,and then respectively using the single factor and multiple factors logistic regression analyze the independent risk factors for recurrence of vertebral fractures,using R software(R4.0.1)and the package of the RMS construct nomogram prediction model by the independent risk factors.To calculate the area of the ROC curve through predicted value,using the bootstrap method simulated sampling conduct the internal validation,and then calculate the C-index,finally compare calibration curve and the actual curve.Results: Among the 232 cases of Vertebral Compression fractures operated vertebroplasty,there were 38 cases of New Vertebral Compression fractures(New Vertebral Compression will Fracture,NVCF)patients,with the recurrence of NVCF 16.4%.Between No recurrence group and new recurrence group,gender,age,fracture number,vertebral height restoration ratio,bone cement injection volume,bone cement distribution(IV,V),the proportion of thoracic lumbar border,T value and Compression Fracture degree are statistical significance(P < 0.05),multi-factor Logisitic regression analysis,the results showed : The number of fractures(OR=1.23,95%CI=1.01,1.89),vertebral height restoration ratio(OR=1.65,95%CI=1.13,1.98),T value(OR=2.24,95%CI=2.12,3.32),bone cement distribution(IV,Vtype)(OR=2.35,95%CI=1.63,4.87)and fracture compression degree(OR=2.25,95%CI=1.21,3.35)were all independent risk factors for NVCF after vertebroplasty.The area under the ROC curve was calculated as AUC=0.891 based on the predicted values.The Bootstrap method was used to simulate sample for 1000 times and the C-index was corrected for 0.89,it was found that the actual curve was very close to the correction curve,indicating that the model had good prediction ability,accuracy and differentiation.Conclusion:The establishment of the prediction model can provide clinical decision makers with a more detailed diagnosis and treatment plan for OVCF,better judge the severity and prognosis of the disease,and effectively reduce the incidence of new adjacent vertebral fractures.