Comparative performance of the Risk Analysis Index versus traditional frailty measures in predicting outcomes following unicompartmental knee arthroplasty: A national database analysis.

Abstract

While multiple frailty assessment tools exist for perioperative risk stratification, their comparative effectiveness in unicompartmental knee arthroplasty (UKA) remains unclear. The Risk Analysis Index (RAI) represents a comprehensive frailty measure, but its performance relative to established indices requires validation. We aimed to compare the predictive accuracy of the RAI against traditional frailty measures including the modified Frailty Index-5 (mFI-5) and Geriatric Nutritional Risk Index (GNRI) for 30-day outcomes following UKA. We analyzed 9358 patients undergoing elective UKA from the ACS-NSQIP database (2015-2021). Three frailty indices were calculated: RAI (incorporating age, sex, renal function, dyspnea, cancer, weight loss, and functional status), mFI-5 (five comorbidity domains), and GNRI (nutritional assessment). Primary outcomes were discharge disposition and 30-day readmission. Secondary outcomes included 30-day all-cause mortality, complications, reoperations, and extended length of stay. The RAI demonstrated superior discrimination for discharge disposition (area under the curve (AUC) = 0.694) and 30-day readmission (AUC = 0.615) compared with mFI-5 (AUC = 0.593 and 0.570) and GNRI (AUC = 0.521 and 0.558). Progressive increases in adverse outcomes occurred across RAI tiers, with non-home discharge rates of 1.7 % in robust patients versus 15.9 % in frail patients (P < 0.001). The RAI provides superior risk stratification compared with traditional frailty measures for discharge disposition and 30-day readmission in particular following UKA, supporting its adoption as the preferred perioperative assessment tool.