Development and validation of a prediction model for the risk of residual low back pain after vertebral augmentation for osteoporotic vertebral compression fractures

English Keywords:Osteoporotic vertebral compression fractures  Vertebral augmentation  Residual low back pain  Prediction model

English Abstract:

  【Abstract】 Objectives: To investigate the independent risk factors of residual lower back pain after vertebral augmentation(VA) in patients with osteoporotic vertebral compression fractures(OVCFs), and to develop and validate an associated risk prediction model. Methods: This study reviewed clinical data of 377 patients with OVCFs who received VA at our hospital from December 2016 to February 2021. There were 52 males and 325 females, aged 75.63±7.27(65-94) years old. Information regarding patients′ general information[gender, age, bone mineral density(BMD), et al], surgical and imaging data, preoperative and postoperative comorbidities, as well as visual analog scale(VAS) scores and Oswestry disability index(ODI) at preoperation, postoperative 1 day, pre-discharge, 3 months and 6 months postoperatively were collected. Postoperative residual low back pain was defined as moderate or more low back pain(VAS≥4) near the original pain site 1 day after surgery, and according to such definition, the patients were divided into residual pain group with 64 cases and the control group with 313 cases. The univariate analysis was combined with Lasso regression to determine the optimal Logistic regression model for multivariate analysis to explore the independent risk factors of postoperative residual low back pain, after which a Nomogram model was constructed. Bootstrap was applied to complete the internal validation of the model. The receiver operating characteristic(ROC) curve, calibration curve, and decision curve analysis(DCA) were used to evaluate the predictive performance and clinical practicability of the model, respectively. Results: Univariate analysis and Lasso regression cross-validation of the best model showed that preoperative BMD, preoperative lumbar disc herniation, preoperative thoracolumbar fascia(TLF) injury, intraoperative facet joint injury(FJI), and postoperative incomplete cementing of the fracture line were the potential risk factors for residual pain. Further multivariate Logistic regression analysis showed that the above variables were also the independent risk factors for postoperative residual low back pain(P<0.05). The visual Nomogram model of the Logistic regression was constructed, of which the C index calculated by using the ROC curve was 0.8384(95%CI: 0.7855-0.8912). After 200 times of Bootstrap sampling internal verification, the C index was 0.8326. The calibration curve showed that the predicted probability curve was close to the actual probability curve. And the DCA curve was within the threshold range of 1%-53%, the decision curve was above the None line and the All line. Conclusions: Preoperative low BMD, preoperative lumbar disc herniation, preoperative TLF injury, intraoperative FJI, and postoperative incomplete cementing of the fracture line are independent risk factors for residual low back pain after VA in patients with OVCFs. The prediction model, constructed with the above five predictive factors, has good predictive performance and clinical practicability in predicting postoperative residual low back pain.