Surgery Resident Northwestern University Chicago, Illinois, United States
Disclosure(s):
Andy Chao Hsuan C. Lee, MD: No financial relationships to disclose
Purpose: The Risk Analysis Index (RAI) is a frailty assessment tool based on an accumulation of deficits model. We mapped the RAI to data from the Society of Thoracic Surgeons (STS) Database to determine whether frailty as assessed by RAI correlates with postoperative outcomes in patients following lung cancer resection. Methods: Study patients underwent lung resection 2018 to 2020. RAI components included age, sex, weight loss, poor appetite, congestive heart failure, dyspnea, renal failure, presence of cancer, functional status, cognitive decline, and living status (scored 0 to 81; an increasing RAI score is related to an increasing incidence of frailty). RAI was categorized as ≤34, 35-39, 40-44, ≥45. Clinical outcomes included complications by category and administrative outcomes included discharge to other than home, in-hospital mortality, 30-day mortality, and 30-day readmission. We calculated the marginal probability of events within each outcome category related to each level of RAI. Logistic regression analyses controlled for appropriate variables. To compare the performance of RAI to established risk indices, similar modeling was performed for the American Society of Anesthesiology (ASA) and Charlson Comorbidity Index (CCI) scores. Areas under the ROC curves (AUC) for the three instruments were compared. Results: 29,420 candidate patients were identified in the STS Database, and RAI could be calculated for 22,843. There were no clinically important differences between the groups with and without RAI scores. Almost all outcome categories exhibited a progressive increase in marginal probability as RAI increased (Figure). On multivariable analyses, RAI was significantly associated in an incremental pattern with almost all clinical and administrative outcomes (Table). ASA and CCI performed similarly on multivariable analyses. ROC analyses demonstrated “good” AUC values (.70 to .80) for RAI related to mortality and discharge to other than home, whereas AUC values for other outcomes were in the fair category (0.60 to 0.70). RAI had significantly better AUC than ASA in 6 of the 15 outcome categories (p < 0.05), including cardiovascular complications, urinary complications, any postoperative event, any major complication, readmission with 30 days, and discharge to other than home, and had significantly better AUC than CCI in 8 categories, also including neurological events and 30-day mortality (p < 0.05). Conclusion: The RAI is independently predictive of almost all clinical and administrative outcomes following lung resection for cancer. It can be calculated from data typically collected for the STS Database. RAI outperformed established risk indices, ASA and CCI, for AUC in many of the outcome categories. Routine assessment of surgical candidates using RAI may be useful in risk prediction, shared decision making, and assignment of resources for perioperative care.
Identify the source of the funding for this research project: This work was funded by the Donald J. Ferguson, MD, Surgical Research Fund of The University of Chicago
