• Elsa Dent, Ian Chapman, Stuart Howell, Cynthia Piantadosi, and Renuka Visvanathan.
• Discipline of Medicine, University of Adelaide, Adelaide, South Australia, Australia.
• Age Ageing. 2014 Jul 1;43(4):477-84.

Backgroundadmission to a Geriatric Evaluation and Management Unit (GEMU) can optimise a patient's chance of functional recovery.Objectiveto evaluate the ability of several commonly used frailty and functional decline indices to predict GEMU outcomes, both at discharge and at 6 months.Designprospective, observational study.Setting And Participantsconsecutive GEMU patients aged ≥70 years.Methodspatients were classified as 'frail' or 'at high risk of functional decline' using several frailty and functional decline instruments. Predictive ability was evaulated using logistic regression and area under receiver operator characteristic (ROC) curves (auROC).Resultsa total of 172 patients were included. Frailty prevalence varied from 24 to 94% depending on the instrument used. Several instruments predicted patients at risk of poor outcome, including the Frailty Index of Accumulative Deficits (FI-CD), Fried's Cardiovascular Health Study index, the Study of Osteoporotic Fractures index, an adapted Katz score of activities of daily living (ADL), Instrumental ADL, the Score Hospitalier d'Evaluation du Risque de Perte d'Autonomie (SHERPA) and grip strength [odds ratio (OR) range of 2.06-6.47]. Adequate discriminatory power for discharge outcome was achieved by the FI-CD (auROC = 0.735, P < 0.001) and an adapted Katz score (auROC = 0.704, P = < 0.001). The FI-CD also showed adequate discriminatory power for a poor 6-month outcome (auROC = 0.702, P < 0.001).Conclusionfrailty and functional decline instruments can predict older patients at risk of poor outcome. However, only the FI-CD showed adequate discriminatory power for outcome prediction at both follow-up time-points.© The Author 2013. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

22 keywords...

hide…