Background: Different frailty scores have been proposed to measure frailty. No study has compared their predictive ability to predict outcomes in trauma patients. The aim of our study was to compare the predictive ability of different frailty scores to predict complications, mortality, discharge disposition, and 30-day readmission in trauma patients.
Methods: We performed a 2-year (2016-2017) prospective cohort analysis of all geriatric (age, >65 years) trauma patients. We calculated the following frailty scores on each patient; the Trauma-Specific Frailty Index (TSFI), the Modified Frailty Index (mFI) derived from the Canada Study of Health and Aging, the Rockwood Frailty Score (RFS), and the International Association of Nutrition and Aging 5-item a frailty scale (FS). Predictive models, using both unadjusted and adjusted logistic regressions, were created for each outcome. The unadjusted c-statistic was used to compare the predictive ability of each model.
Results: A total of 341 patients were enrolled. Mean age was 76 ± 9 years, median Injury Severity Score was 13 [9-18], and median Glasgow Coma Scale score was 15 [12-15]. The unadjusted models indicated that both the TSFI and the RFS had comparable predictive value, as indicated by their unadjusted c-statistics, for mortality, in-hospital complications, skilled nursing facility disposition and 30-day readmission. Both TSFI and RFS models had unadjusted c-statistics indicating a relatively strong predictive ability for all outcomes. The unadjusted mFI and FS models did not have a strong predictive ability for predicting mortality and in-hospital complications. They also had a lower predictive ability for skilled nursing facility disposition and 30-day readmissions.
Conclusion: There are significant differences in the predictive ability of the four commonly used frailty scores. The TSFI and the RFS are better predictors of outcomes compared with the mFI and the FS. The TSFI is easy to calculate and might be used as a universal frailty score in geriatric trauma patients.
Level of evidence: Prognostic, level III.