Assessing a revised-risk analysis index for morbidity and mortality after spine surgery for metastatic spinal tumors.

Abstract

Risk Analysis Index (RAI) has been increasingly used to assess surgical frailty in various procedures, but its effectiveness in predicting mortality or in-patient hospital outcomes for spine surgery in metastatic disease remains unclear. The aim of this study was to compare the predictive values of the revised RAI (RAI-rev), the modified frailty index-5 (mFI-5), and advanced age for extended length of stay, 30-day readmission, complications, and mortality among patients undergoing spine surgery for metastatic spinal tumors. A retrospective cohort study was performed using the 2012-2022 ACS NSQIP database to identify adult patients who underwent spinal surgery for metastatic spinal pathologies. Using receiver operating characteristic (ROC) and multivariable analyses, we compared the discriminative thresholds and independent associations of RAI-rev, mFI-5, and greater patient age with extended length of stay (LOS), 30-day complications, hospital readmission, and mortality. A total of 1,796 patients were identified, of which 1,116 (62.1%) were male and 1,008 (70.7%) were non-Hispanic White. RAI-rev identified 1,291 (71.9%) frail and 208 (11.6%) very frail patients, while mFI-5 identified 272 (15.1%) frail and 49 (2.7%) very frail patients. In the ROC analysis for extended LOS, both RAI-rev and mFI-5 showed modest predictive capabilities with area under the curve (AUC) values of 0.5477 and 0.5329, respectively, and no significant difference in their predictive abilities (p = 0.446). When compared to age, RAI-rev demonstrated superior prediction (p = 0.015). With respect to predicting 30-day readmission, no significant difference was observed between RAI-rev and mFI-5 (AUC 0.5394 l respectively, p = 0.354). However, RAI-rev outperformed age (p = 0.001). When assessing the risk of 30-day complications, RAI-rev significantly outperformed mFI-5 (AUC: 0.6016 and 0.5542 respectively, p = 0.022) but not age. Notably, RAI-rev demonstrated superior ability for predicting 30-day mortality compared to mFI-5 and age (AUC: 0.6541, 0.5652, and 0.5515 respectively, p < 0.001). Multivariate analysis revealed RAI-rev as a significant predictor of extended LOS [aOR: 1.96, 95% CI: 1.13-3.38, p = 0.016] and 30-day mortality [aOR: 5.27, 95% CI: 1.73-16.06, p = 0.003] for very frail patients. Similarly, the RAI-rev significantly predicted 30-day complications for frail [aOR: 2.63, 95% CI: 1.21-5.72, p = 0.015] and very frail [aOR: 3.69, 95% CI: 1.60-8.51, p = 0.002] patients. However, the RAI did not significantly predict 30-day readmission [Very Frail aOR: 1.52, 95% CI: 0.75-3.07, p = 0.245; Frail aOR: 1.46, 95% CI: 0.79-2.68, p = 0.225]. Our study demonstrates the utility of RAI-rev in predicting morbidity and mortality in patients undergoing spine surgery for metastatic spinal pathologies. Particularly, the superiority that RAI-rev has in predicting 30-day mortality may have significant implications in multidisciplinary decision making.