Development and Validation of a Nomogram for Predicting Postoperative Bone Nonunion in Patients with Infectious Spondylitis

Abstract Background Postoperative bone nonunion remains a significant complication following instrumented fusion for infectious spondylitis, leading to high clinical burden and poor clinical outcomes. Fusion outcomes vary markedly due to persistent inflammation, malnutrition, and infection impaired osteogenesis, factors specifically unique to infectious spondylitis and distinct from noninfectious spinal fusion. Perioperative identification of high-risk patients is crucial for implementing preventive strategies, yet existing predictive models for spinal nonunion, mostly developed for aseptic surgeries, lack validation and infection-specific predictors in this population, highlighting a key gap in perioperative risk assessment. Purpose This study aimed to develop and validate a novel multivariate nomogram for the perioperative early risk prediction of bone nonunion in patients undergoing surgical treatment for infectious spondylitis, integrating clinical, laboratory, and radiological predictors. Method A retrospective analysis was performed on 224 patients with infectious spondylitis who underwent debridement and spinal internal fixation surgery. According to the Bridwell grading system, patients were divided into the bone union group (n = 166) and the nonunion group (n = 58). The entire cohort was included into training set. Candidate predictors included demographic and clinical characteristics, laboratory parameters, and imaging findings. To minimize confounding, univariate analysis was initially conducted on all variables in the training set, those showing significant differences were subsequently entered into LASSO regression analysis. Variables selected by the 1-standard error (1-SE) criterion were incorporated into multivariate logistic regression. A nomogram was constructed based on the independent predictors. Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC) and concordance index (C-index). Internal validation was performed with 1000 bootstrap resampling. Calibration was evaluated by the Hosmer-Lemeshow (HL) test, and clinical utility was determined by decision curve analysis (DCA). Results 5 independent predictors were identified and incorporated into the nomogram: Lower Albumin (ALB) level (OR 0.82, 95% CI: 0.76–0.88), prolonged CRP normalization days (OR 1.24, 95% CI: 1.15–1.34), allogeneic bone graft materials (OR 5.2, 95% CI: 2.74–9.86), multilevel fusion (OR 8.61, 95% CI: 3.83–19.35) and paravertebral abscess (OR 4.58, 95% CI: 2.35–8.95). The nomogram demonstrated excellent discrimination with an AUC of 0.936 (95% CI: 0.896–0.975) and good calibration (p = 0.94). DCA indicated strong clinical net benefit across a wide range of threshold probabilities (2%-85%). The bootstrap internal validation (1,000 resamples) yielded an apparent AUC of 0.94 (95% CI: 0.88–0.97) excellent calibration (slope = 1.000, intercept ≈ 0.000, Emax = 0.061, Brier score = 0.077), confirming robust performance and negligible overfitting despite the modest sample size. Conclusion We developed and internally validated a novel nomogram for predicting postoperative bone nonunion in patients with infectious spondylitis. This tool may provide preliminary clinical guidance for perioperative early risk assessment. However, given the relatively limited sample size and single-center design, further validation in larger, multicenter cohorts is required before widespread clinical application.