Development of a Frailty Index in the Irish Hip Fracture Database

Abstract

IntroductionIn older people, hip fracture can lead to adverse outcomes. Frailty, capturing biological age and vulnerability to stressors, can indicate those at higher risk. We derived a frailty index (FI) in the Irish Hip Fracture Database (IHFD) and explored associations with prolonged length of hospital stay (LOS ≥ 30 days), delirium, inpatient mortality and new nursing home admission. We assessed whether the FI predicted those outcomes independently of age, sex and pre-operative American Society of Anaesthesiology (ASA) score.Materials and methodsA 21-item FI was constructed with 17 dichotomous co-morbidities, three 4-level ordinal pre-morbid functional variables (difficulty with indoor mobility, outdoor mobility, and shopping) and nursing home provenance (yes/no). The FI was computed as the proportion of items present and divided into tertiles (low, medium, high risk). Independent associations between FI and outcomes were explored with logistic regression, from which we extracted adjusted Odds Ratios (aOR) and Areas Under the Curve (AUC).ResultsFrom 2017 to 2020, the IHFD included 14,615 hip fracture admissions, mean (SD) age 80.4 (8.8), 68.9% women. Complete FI data were available for 12,502 (85.5%). By FI tertile (low to high risk), prolonged LOS proportions were 5.9%, 16.1% and 23.1%; delirium 5.5%, 13.5% and 17.6%; inpatient mortality 0.6%, 3.3% and 10.1%; and new nursing home admission 2.2%, 5.9% and 11.3%. All associations were statistically significant (p < 0.001) independently of age and sex. AUC analyses showed that the FI score, added to age, sex, and ASA score, significantly improved the prediction of delirium and new nursing home admission (p < 0.05), and especially prolonged LOS and inpatient mortality (p < 0.001).ConclusionsA 21-item FI in the IHFD was a significant predictor of outcomes and added value to traditional risk markers. The utility of a routinely derived FI to more effectively direct limited orthogeriatric resources requires prospective investigation.