Prior year hospital admission predicts 30-day hospital readmission after spine surgery

Reducing hospital readmissions postspine surgery is crucial for improving care quality. Currently, no specific scoring system exists to stratify patients by readmission risk in spine surgery. The objective of this study was to evaluate risk factors for 30-day hospital readmission following elective decompressive spine surgery and develop an ML-based risk-scoring system.

653 patients undergoing elective decompressive spine surgery (with or without fusion) at a single academic center were retrospectively reviewed. Data collected included demographics, surgical details, preoperative admissions, and complications. Univariate logistic and Cox regression analyses identified significant predictors. A supervised ML-based risk scoring system was developed and compared against the institution's standard risk calculator.

Of 653 patients, 44 (6.7%) experienced a 30-day readmission. Predictors of readmission included prior-year hospital admission, age-adjusted Charlson Comorbidity Index (CCI), traumatic or thoracic surgery, postoperative complications, and length of stay. An ML-based scoring system stratified patients into Low-risk (0-4 points; 2.16% readmission), Moderate-risk (5-9 points; 7.25%), High-risk (10-14 points; 21.57%), and Extreme-risk (15+ points; 39.47%) groups. Compared to the standard hospital score, the new model demonstrated superior performance: higher C-index (0.79 vs. 0.71), higher Somers' D (0.56 vs. 0.41), and lower Akaike Information Criterion (AIC) (516.78 vs. 546.65).

Preoperative admission within 1 year prior to elective decompressive spine surgery significantly predicts 30-day readmission risk. Additional factors include age-adjusted CCI, traumatic or thoracic surgery, complications, and length of hospital stay. The proposed ML-based scoring system demonstrated superior predictive performance compared to the standard hospital readmission risk calculator, serving as a potentially valuable internal quality improvement tool pending external validation.