M. Aizpuru2, K. X. Farley2, M. V. Poirier2, L. P. Brewster1, E. R. Wagner3, R. S. Crawford1 1Emory University School Of Medicine,Department Of Surgery, Division Of Vascular Surgery,Atlanta, GA, USA 2Emory University School Of Medicine,Atlanta, GA, USA 3Emory University School Of Medicine,Department Of Orthopaedics,Atlanta, GA, USA
Introduction: Frailty has been used as a predictor of adverse outcomes in vascular surgery, yet there are few studies comparing the available frailty indices head-to-head. The National Inpatient Sample has the unique capability to allow calculation of three major frailty measures used in vascular surgery. The aim of our study is to compare frailty indices for predicting in-patient mortality and prolonged length of stay following carotid endarterectomy (CEA).
Methods: 315,354 patients underwent carotid endarterectomy between 2002-2015 Q3 in the National Inpatient Sample (NIS). Comorbidities were identified using previously published ICD-9 coding methods. Charlson Comorbidity Index (CCI, 0-26), the Modified Frailty Index (mFl, 0-1), and the Elixhauser Comorbidity Measure (Elixhauser, 0-31) were calculated. Prolonged length of stay (LOS) was defined as a hospital stay of 2 days or greater. The predictive value of CCl, mFl, and Elixhauser were compared using receiver-operating curves for both in-patient mortality and prolonged length of stay.
Results: The mean age was 71.0 ± 9.5 years. 244,208 (77%) patients had a history of hypertension, 44,506 (14%) had a history of stroke, and 34,896 (11%) had a history of MI. The mean mFI was 0.17 (range=0.00-0.82), the mean CCI was 1.3 (range=0-17) and the mean ECM was 2.2 (range=0-12). Median LOS was 1 day (range=0-283 days). There were 1,635 (0.05%) in-hospital deaths. mFI (AUC= 0.524, CI [0.509-0.538]) was inferior to CCI (AUC=0.636, CI [0.624-0.653]) and ECM (AUC=0.648, CI [0.634-0.663]), which were equivalent for predicting mortality. Elixhauser comorbidity measure (AUC=0.606, CI [0.604-0.608]) was superior to mFI (AUC=0.551, CI [0.549, 0.553]) and CCI (AUC=0.572, CI [0.570-0.574]) for predicting prolonged LOS (≥2 days).
Conclusion: Frailty indices were not strong predictors of in-hospital mortality or prolonged LOS in patients undergoing CEA in the National Inpatient Sample. Despite receiving the most attention in the vascular literature, the modified frailty index (mFI) was the least effective. These results call into question the usefulness of frailty in predicting outcomes without some consideration of the extent of the procedure.
