A.Z. Agathis1, J. Wu1, C. Divino1 1Icahn School of Medicine at Mount Sinai, Surgery, New York, NY, USA
Introduction: Since its origination in 2018 by Subramaniam et al., the modified five-factor frailty scoring index (mFI-5) has been widely applied across surgical subspecialties using the National Surgical Quality Improvement Program database (NSQIP). This index accounts for medical comorbidities (hypertension, heart failure, diabetes, and chronic obstructive pulmonary disease) and the functional status variable. While the mFI-5 index is predictive, it does not fully encompass functional aspects of frailty. We suggest an alternative four-factor frailty index focused on geriatric functional characteristics.
Methods: This retrospective study using NSQIP 2021-2022 included patients 75+ years of age who underwent colon and rectal surgery. The new frailty index is scored by taking a sum of each present variable for history of falls, dementia, poor nutritional status (BMI <18.5), and non-independent functional status. Patients were stratified into three frailty groups: mFF-4=0, 1, and 2+. Multivariate regression was performed with covariates including frailty, age, gender, race, smoking, heart failure, hypertension, diabetes, chronic obstructive pulmonary disease, inpatient and emergency status, operation time, and ASA class. ROC AUCs, AICs, and odds ratios (OR) were calculated for the current mFI-5 and the new mFF-4.
Results: Our sample of n=27,875 included 21,573 in mFF-4=0 (77.4%), 4,601 in mFF-4=1 (16.5%), and 1,701 in mFF-4=2+ (6.1%). The mean population age was 80.94 (SD 4.61). Patients were predominantly ASA=3 (65.87%), in contrast to ASA=1 (0.22%), ASA=2 (16.98%), ASA=4 (16.06%), and ASA=5 (0.86%). Surgeries were mostly elective (76.63%) versus urgent (11.26%) or emergent (15.11%), and were 97.71% inpatient. In comparing the predictive value of mFF-4 and mFI-5 models, the mFF-4 multivariate regressions had lower AICs (differences >2) and higher AUCs across outcomes including mortality (AUCs 0.8364 vs 0.8234 for mFF-4 and mFI-5, respectively), pneumonia (AUCs 0.7331 vs 0.7277), ventilator needed >48 hours (AUCs 0.8582 vs 0.8555), delirium (AUCs 0.7294 vs 0.7036), length of stay (LOS) >30 days (AUCs 0.7978 vs 0.7897), and discharge destination not home (AUCs 0.7793 vs 0.7644). Similar trends existed for reoperation, readmission, and need for home discharge services. Patients with high frailty (mFF-4=2+) had statistically significant OR >2.00 for mortality, ventilator requirement, LOS >30 days, delirium, discharge destination not home, and discharge home with services.
Conclusion: This new modified mFF-4 frailty index focused on functional capabilities better encompasses the definition of frailty and is more predictive than the established mFI-5. This new scale will help improve risk-stratification of our heterogenous, aging surgical population.