Development and validation of a predictive model for postoperative residual low back pain following percutaneous kyphoplasty or percutaneous vertebroplasty.

OBJECTIVE: To identify risk factors for persistent postoperative low back pain in patients undergoing percutaneous kyphoplasty (PKP) or percutaneous vertebroplasty (PVP) and develop a predictive nomogram model. METHOD: This retrospective study included 749 patients with osteoporotic vertebral compression fractures (OVCF), who were allocated into a modeling cohort (n = 524) and a validation cohort (n = 225). Patients with a Visual Analog Scale (VAS) score > 4 at one month postoperatively were classified into the residual low back pain group, while the rest were classified as the normal group. Univariate and multivariate logistic regression analyses were performed to identify factors associated with postoperative back pain, and subsequently, a nomogram model was constructed for predictive validation. RESULTS: In the modeling cohort, 61 patients reported persistent low back pain at one month after surgery. According to univariate and LASSO-Logistic regression analyses, the number of fractured vertebrae, bone cement distribution, bone cement leakage, 5-hydroxyvitamin D3 [25-(OH)D3] levels, concomitant lumbar and dorsal fascia injury, and Exercise Activity Questionnaire score (EAQ) score were identified as independent predictors (P < 0.05), representing the main factors influencing residual low back pain in patients after PKP/PVP. In the modeling cohort, the area under the receiver operating characteristic (ROC) curve (AUC) was 0.837, and the C-index of the calibration curve was 0.837 (95% CI, 0.719-0.883). In the validation cohort, AUC was 0.858, and the C-index of the calibration curve was 0.858 (95% CI, 0.802-0.914). CONCLUSION: Number of fractured vertebrae, insufficient bone cement distribution over the fracture line, bone cement leakage, serum 25-(OH)D3 level, combined lumbar and dorsal fascia injury, and EAQ score were independent risk factors for residual low back pain in patients after PKP/PVP. The nomogram predictive model incorporating these factors demonstrated good predictive performance in internal validation.