The frailty-driven predictive model for failure to rescue among patients who experienced a major complication following cervical decompression and fusion: an ACS-NSQIP analysis of 3,632 cases (2011–2020)

Abstract

BACKGROUND CONTEXTPreoperative risk stratification for patients considering cervical decompression and fusion (CDF) relies on established independent risk factors to predict the probability of complications and outcomes in order to help guide pre and perioperative decision-making. PURPOSEThis study aims to determine frailty's impact on failure to rescue (FTR), or when a mortality occurs within 30 days following a major complication. STUDY DESIGN/SETTINGCross-sectional retrospective analysis of retrospective and nationally-representative data. PATIENT SAMPLEThe American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database was queried for all CDF cases from 2011-2020. OUTCOME MEASURESCDF patients who experienced a major complication were identified and FTR was calculated as death or hospice disposition within 30 days of a major complication. METHODSFrailty was measured by the Risk Analysis Index-Revised (RAI-Rev). Baseline patient demographics and characteristics were compared for all FTR patients. Significant factors were assessed by univariate and multivariable regression for the development of a frailty-driven predictive model for FTR. The discriminative ability of the predictive model was assessed using a receiving operating characteristic (ROC) curve analysis. RESULTSThere were 3632 CDF patients who suffered a major complication and 7.6% (277 patients) subsequently expired or dispositioned to hospice, the definition of FTR. Independent predictors of FTR were nonelective surgery, frailty, preoperative intubation, thrombosis or embolic complication, unplanned intubation, on ventilator for >48 hours, cardiac arrest, and septic shock. Frailty, and a combination of preoperative and postoperative risk factors in a predictive model for FTR, achieved outstanding discriminatory accuracy (C-statistic = 0.901, CI: 0.883–0.919). CONCLUSIONPreoperative and postoperative risk factors, combined with frailty, yield a highly accurate predictive model for FTR in CDF patients. Our model may guide surgical management and/or prognostication regarding the likelihood of FTR after a major complication postoperatively with CDF patients. Future studies may determine the predictive ability of this model in other neurosurgical patient populations.