Resident Physician Yale School of Medicine New Haven, Connecticut, United States
Introduction: Frailty is common in patients with metastatic spinal tumors and is associated with inferior outcomes. However, few studies have investigated the impact of frailty on unplanned 30-day readmissions. The study aim was to determine the predictive power of the Hospital Frailty Risk Score (HFRS) on unplanned 30-day readmissions after metastatic spinal column tumor surgery.
Methods: A retrospective cohort study was performed using the Nationwide Readmission Database (2016-2018). All patients ≥18 years old undergoing metastatic spinal column tumor surgery were categorized by HFRS score: Low (HFRS < 5), Intermediate (HFRS 5-15), and High (HFRS>15). Demographics, comorbidities, adverse events (AEs), length of stay (LOS), and discharge disposition were assessed. Random Forest (RF) classification was used to construct predictive 30-day readmission models. The area under the receiver operating curve (AUC) assessed model performance. The Mean Decrease Gini (MDG) metric quantified and ranked features by relative importance.
Results: Of the 4,346 patients identified, 2,535 (58.3%) were classified as Low frailty, 1,717 (39.5%) as Intermediate, and 94 (2.2%) as High. The average HFRS score was 2.1±1.5, 8.4±2.6, and 17.1±1.5 for the Low, Intermediate, and High cohorts, respectively (p < 0.001). The proportion of readmitted patients was higher among the Intermediate and High cohorts versus the Low cohort (Low: 33.9% vs Intermediate: 39.3% vs High: 39.2%, p< 0.001). An RF classifier was trained to predict 30-day readmissions on all features (AUC=0.60), which outperformed an architecturally equivalent model trained using the ten highest MDG features (AUC=0.59). Both models identified that frailty is associated with 30-day readmission risk. Additionally, on multivariate analysis, Intermediate frailty [OR: 1.32, CI (1.06, 1.64), p=0.012] independently predicted 30-day readmissions.
Conclusion : Our study utilizes machine learning and predictive modeling to identify HFRS-defined frailty as a significant contributor to 30-day readmissions after metastatic spinal column tumor surgery.
