Development and validation of a predictive model for the risk of symptomatic adjacent segmental degeneration after anterior cervical discectomy and fusion.

Abstract

Background: To investigate the risk factors for symptomatic adjacent segment degeneration (ASD) 5 years after anterior cervical discectomy and fusion (ACDF) and develop and evaluate predictive models.

Methods: A total of 655 patients who underwent ACDF were randomly assigned to the training set (n = 393) or validation set (n = 262) at a ratio of 6:4. Independent predictors of ASD were selected by LASSO regression and logistic regression analysis. A calibration curve, ROC curve and DCA curve were used to evaluate the model performance.

Results: LASSO regression combined with logistic regression analysis revealed that age, cervical canal stenosis, smaller T1S and smaller cervical lordosis (CL) were risk factors for ASD 5 years after surgery. Nomographic analysis using appeal factors was used to predict the risk of ASD. The area under the ROC curve was 0.711 (95% CI: 0.643-0.780) in the training set and 0.701 (95% CI: 0.618-0.785) in the validation set. The calibration curve showed no significant bias in either set. The DCA indicated that using the nomogram to predict the risk of ASD would be more accurate when the risk threshold probability was 12-53% in the training set and 6-43% in the validation set.

Conclusion: Age, cervical spinal stenosis, a smaller T1S, and a smaller CL are independent risk factors for ASD 5 years after ACDF surgery. Based on these four indicators, we constructed a new clinical prediction model that has a certain predictive effect and is conducive to clinical decision-making and treatment planning.