EVALUATION OF FRAILTY INDICES FOR THE PREDICTION OF ADVERSE POST-HOSPITAL OUTCOMES IN OLDER PERSONS (#142)
Background: Following hospital discharge, older people are at an increased risk of adverse clinical outcomes. We examined the predictive ability of six frailty indices in identifying patients with increased risk of mortality, emergency re-hospitalisation and re-hospitalisation due to falls.
Methods: In this prospective study of consecutive patients admitted to a Geriatric Evaluation and Management Unit, we identified frailty using Fried's Index, Study of Osteoporotic Fractures (SOF) Index, Frailty Index of Cumulative Deficits (FI-CD), Multidimensional Prognostic Index (MPI), Score Hospitalier d’Evaluation du Risque de Perte d’Autonomie (SHERPA) and Katz index of Activities of Daily Living (ADL). Logistic regression and area under curve (AUC) of receiver operator characteristic (ROC) curves were analysed, adjusting for age and gender.
Results: 172 patients (mean (SD) age of 85.2 (6.4) years; 72% female) were included. During 6 month follow-up, 28 (16%) patients died, 92 (53%) were re-hospitalised for emergencies and 43 (25%) were re-hospitalised for falls. Frailty identified by all instruments, excluding the MPI, was associated with mortality (odds ratio (OR) values all > 2.50, P < 0.005). Frail patients had a high risk of emergency re-hospitalisation, when identified by Fried’s Index (OR = 2.47, P = 0.010) and SHERPA (OR = 1.97, P = 0.038). No indices associated with re-hospitalisation due to falls. AUC results showed FI-CD had the highest discriminatory power for mortality (AUC= 0.803), followed by SHERPA (AUC = 0.792) and Katz Index (AUC = 0.757). All other AUC values lacked adequate discriminative power for outcome prediction (AUC values < 0.7).
Conclusion: Frailty instruments are a feasible way to identify older patients with an increased likelihood of mortality and re-hospitalisation, although not re-hospitalisation due to falls. The FI-CD and the simpler to use SHERPA and Katz indices are recommended for prediction of mortality. Our findings can guide patient care and discharge planning.