SADA-04 DEVELOPMENT AND INTERNAL VALIDATION OF A CONDITIONAL FAILURE-TO-RESCUE PREDICTION MODEL AFTER SURGERY FOR METASTATIC SPINE DISEASE
Esli Nájera Samaniego, Ali Haider Bangash, Victoria Cao, Alexander Alexandrov, Liza Belman, Temitope Adegbenro, Saikiran Murthy, Yaroslav Gelfand, Reza Yassari, Rafael De La Garza RamosAbstract
Introduction
Failure to Rescue (FTR), defined as death following a major postoperative complication, has emerged as a key indicator of surgical quality because it reflects the ability to recognize and manage complications once they occur. Patients undergoing surgery for metastatic spinal tumors represent a vulnerable population in whom early postoperative morbidity remains substantial. Conditional failure to rescue (cFTR), defined as mortality among patients who experience a major postoperative complication, may represent a more clinically meaningful outcome in this high-risk population.
Methods
A retrospective analysis of a national surgical quality database (2018-2023) was performed. The primary endpoint was cFTR, univariable and multivariable logistic regression models were constructed to predict it. Model performance was assessed using discrimination (AUC), calibration, and information criteria (AIC and BIC). Penalized regression (LASSO and Elastic Net) was performed to evaluate robustness. Internal validation was conducted using five-fold cross-validation and bootstrap resampling. Reporting of model development followed the TRIPOD statement Results: Among 338 patients experiencing major complications, 71 (21%) died within 30 days. Emergency surgery emerged as the strongest predictor of cFTR (OR 2.77). The final six-predictor clinical model included emergency surgery, current smoking, elevated white blood cell count, low serum albumin, and functional dependence. The model demonstrated good discrimination (AUC 0.740, 95% CI 0.671-0.809) with acceptable calibration. Internal validation showed minimal optimism (cross-validated AUC 0.713, bootstrap mean AUC 0.743). Secondary analyses revealed higher complication burden among non- survivors (1.85 vs 1.44 complications, p = 0.001).
Conclusions
Among patients with metastatic spinal tumors who experience major postoperative complications, cFTR remains common and is driven by patient-level vulnerability. Conditioning on complications highlights emergency presentation, impaired physiologic reserve, systemic inflammation, and functional dependence as determinants of mortality. This internally validated model provides a transparent framework for risk stratification after complications and identified potentially modifiable targets for perioperative optimization.