Comprehensive Summary
This article develops and validates a Least Absolute Shrinkage and Selection Operator (LASSO) predictive model for post-operative dysphagia in patients undergoing Anterior Cervical Discectomy and Fusion (ACDF) surgery. This retrospective study examined 500 ACDF patients treated at a single institution from 2018 to 2022, with 75% of patients randomly divided into a training cohort and 25% into a validation cohort. Researchers evaluated 53 candidate variables categorized by radiographic measurements, surgical characteristics, medical history factors, and blood biomarkers. Using univariate comparisons, LASSO regression, and receiver operating characteristic (ROC) analysis, nine key dysphagia predictors were identified, including anesthesia duration, number of spinal levels fused, preoperative ALT and glucose levels, and preoperative/postoperative pre-vertebral soft tissue (PST) thickness. The model demonstrated strong accuracy, with an Area Under Curve (AUC) of 0.971 in the training set and 0.954 in the validation set. The authors emphasize the need for external validation, model comparisons, and careful cohort partitioning. Importantly, the model was designed specifically to predict dysphagia within the first postoperative month, leaving its long-term prediction of dysphagia untested.
Outcomes and Implications
Dysphagia is one of the most common complications after ACDF, reported in up to 71% of surgery patients, and can lead to a severe reduction in quality of life through malnutrition, dehydration, and aspiration pneumonia. The LASSO model demonstrates a comprehensive, precise tool for preoperative planning, allowing clinicians to implement specific preventive strategies such as tighter glucose level control or surgical modifications in high-risk patients. While further refinement and optimization is needed, this model highlights an important step towards individualized surgical care to reduce dysphagia occurrences and enhance recovery outcomes in ACDF patients.