Predictors of Outcome After Anterior Cervical Discectomy and Fusion A Multivariate Analysis

Study Design. Retrospective cohort study. Objective. Perform a multivariate analysis to identify important predictors of poor outcome following anterior cervical discectomy and fusion. Summary of Background Data. Identifying prognostic factors is important to aid surgical decision-making and counseling of patients. Recent randomized control trials of disc arthroplasty devices have established a large cohort of patients treated with fusion and 2-year outcomes that allow analysis of prognostic factors. Methods. The patient cohort was the fusion control patients (n = 488) from 2 randomized controlled studies of disc replacements. Surgical indications were recalcitrant single-level subaxial radiculopathy or myelopathy. The surgery included anterior discectomy and fusion with allograft and plate. Patients were assessed by neck and arm pain, neck disability index (NDI), SF-36, neurologic examination, and return to work. Overall clinical success was defined based on meeting all 4 of these criteria: >15-point improvement in NDI; maintained or improved neurologic examination; no serious adverse event related to the procedure; and no revision of the plate or graft. Patient's outcomes were recorded, at 3, 6, 12, and 24 months, with 77% follow-up at 24 months. The outcome variables for this analysis were overall clinical success and >15-point improvement in NDI. We studied the relationship between each of the outcome variables and 26 potential important variables including demographics, medical conditions, socioeconomic factors, and disease state. Two statistical models were used to explore the association between outcome variables and baseline measures: multivariate logistical regression of the full model with every prognostic variable included and the model with the variables selected by the stepwise selection procedure. Results. In the full-model logistic analysis for overall success, worker's compensation and weak narcotic use were negative predictors while higher preoperative NDI score and normal sensory function were positive predictors. For NDI success, only the preoperative NDI scores (higher disability predictive of improvement) appeared to have strong influence on the outcome. In the stepwise regression model, preoperative normal sensory function was a positive predictor and worker's compensation a negative predictors of overall clinical success. Greater age, higher preoperative NDI score, and gainful employment were positive predictors and spinal litigation was a negative predictor of NDI success. Conclusion. We found that important predictors of outcome were work status, sensory function, involvement in litigation, and higher disability scores.