← Back to List

Program Number: B159
Session Name: Poster Session

Is There An Ideal Method of Predicting Postoperative Complications Following Total Laryngectomy?: A Comparative Analysis of Three Frailty Indices

Michael P Saturno; Rahul Guda; Olivia First; Christopher Connors; Raymond L Chai; Icahn School of Medicine at Mount Sinai

Introduction: Frailty assessments are a tool for predicting postoperative outcomes. The 5-factor modified Frailty Index (mFI-5) and the Risk Analysis Index-Administrative (RAI-A) are commonly used indices, though a revised version of the RAI-A - the RAI-Revised (RAI-Rev) - was published in 2020 and remains sparsely investigated. This study compares the capacity of the mFI-5, the RAI-A, and the RAI-Rev to predict postoperative complications following total laryngectomy. 

Methods: The Current Procedural Terminology (CPT) codes were used to identify adult patients who underwent total laryngectomy as reported by the National Surgical Quality Improvement Program (ACS-NSQIP) database between 2015–2020. Primary outcomes included mortality, unplanned readmission, reoperation, extended length of stay (ELOS), non-home discharge, Clavien-Dindo Class IV complications, and major complications. ELOS was defined as a hospital course duration within the upper quartile of the cohort. Univariate, multivariate, and receiver operating characteristic (ROC) curve analyses were conducted to compare the predictive ability of the RAI-Rev, RAI-A, and mFI-5 for 30-day postoperative outcomes. The ROC analyses generated area under the curve (AUC) values for each frailty index, which were statistically compared using the DeLong Test.

Results: On the univariate and multivariate analyses, the severely frail tier of mFI-5 demonstrated a significantly higher risk for all postoperative complications except reoperation and unplanned readmission when compared to its non-frail cohort (Table 1). Conversely, the prefrail tier was not predictive of any postoperative complications. All tiers within the RAI-A exhibited an elevated risk of ELOS and non-home discharge on both univariate and multivariate analyses. None of the RAI-Rev tiers exhibited an increased risk of postoperative complications. Multivariate analysis yielded identical trends (Table 2). On the ROC analysis for predicting mortality, the RAI-Rev (AUC = 0.76) qualitatively outperformed the RAI-A (AUC = 0.69, p = .291) and statistically outperformed the mFI-5 (AUC = 0.58, p = .029) (Figure 1). The RAI-A most optimally predicted non-home discharge and ELOS with an AUC = 0.86 and AUC = 0.58, respectively. Both of these were statistically superior to the other two indices with a p-value < .03 (Figure 2). Conversely, the RAI-A was notably inferior at predicting unplanned readmission (AUC = .44) compared to the mFI-5 (AUC = 0.55, p <.001) and RAI-Rev (AUC = 0.55, p = .012). The mFI-5 was the best-performing frailty score for predicting major complications, Clavien-Dindo Class IV complications, and reoperation, though the differences were rarely statistically significant.

Conclusion: The capacity of the mFI-5, RAI-A, and RAI-Rev frailty indices to predict postoperative complications following laryngectomy varies across different measures. The mFI-5 was most effective for predicting major complications, Clavien-Dindo Class IV complications, and reoperation, while the RAI-A excelled in predicting non-home discharge and ELOS. RAI-Rev demonstrated superior performance in predicting mortality. These findings suggest that the selection of frailty assessment tools should be tailored to the specific postoperative outcome of interest. Escalating tiers within the mFI-5 and RAI-A was generally associated with a stepwise increase in postoperative complication risk, suggesting that these indices may be used to stratify patient-specific risk according to their preoperative comorbidities and demographics.

 

 

← Back to List