Abstract
OBJECTIVE: To investigate the risk factors for new vertebral compression fractures (NVCFs) after percutaneous kyphoplasty (PKP) for osteoporotic vertebral compression fractures (OVCFs) and to construct a postoperative risk stratification nomogram for predicting refracture probability and identifying high-risk patients who require intensive postoperative monitoring and aggressive preventive interventions. METHODS: This retrospective cohort study enrolled 257 patients with single-segment OVCFs treated at Hainan General Hospital from January 2021 to December 2023. Participants were stratified into refracture (n = 56) and non-refracture (n = 201) groups based on new vertebral fracture occurrence within 1-year post-PKP. Data were randomly partitioned into training (n = 180) and validation (n = 77) sets at a 7:3 ratio. Independent risk factors were identified through univariate screening followed by multivariate logistic regression. A refracture risk nomogram was constructed using significant multivariate predictors, with comprehensive validation of predictive utility through tripartite assessment: receiver operating characteristic curve analysis, calibration curves, and decision curve analysis (DCA). RESULTS: Univariate analysis revealed significant between-group differences in sex, bone mineral density (BMD), vertebral height recovery rate, fracture severity, intradiscal cement leakage, anti-osteoporosis treatment, early postoperative mobilization, and history of postoperative falls (all P < 0.05). Multivariate analysis identified moderate fractures [OR = 7.08, 95%CI (1.39-54.00), P = 0.029], severe fractures [OR = 8.60, 95%CI (2.03–60.20), P = 0.009], intradiscal cement leakage [OR = 10.40, 95%CI (2.55–51.30), P = 0.002], and postoperative falls [OR = 4.99, 95%CI (1.75–15.30), P = 0.003] as independent risk factors positively associated with refracture. Conversely, higher BMD [OR = 0.61, 95%CI (0.40–0.91), P = 0.016], anti-osteoporosis treatment [OR = 0.24, 95%CI (0.08–0.63), P = 0.005], and early mobilization [OR = 0.28, 95%CI (0.09–0.77), P = 0.017] demonstrated protective effects. The nomogram maintained robust discrimination across cohorts: training set AUC = 0.892 (95%CI:0.832–0.952) with 78.60% sensitivity and 89.90% specificity at 0.355 cut-off; testing set AUC = 0.836 (95%CI:0.691–0.982) with 78.60% sensitivity and 85.70% specificity at 0.269 cut-off. Calibration curves demonstrated good agreement between predicted and observed outcomes. Decision curve analysis (DCA) demonstrated clinical utility with positive net benefits at 0%-76% (training) and 0%-82% (testing) risk thresholds. CONCLUSIONS: Low BMD, moderate-severe fracture severity, bone cement intradiscal leakage, inadequate anti-osteoporosis treatment, delayed postoperative mobilization, and falls are predominant risk factors for NVCFs after PKP. A validated nomogram prediction model was developed based on these six established risk factors. TRIAL REGISTRATION: Clinical trial number: not applicable.