Prognostic accuracy of eight frailty instruments for all-cause mortality in Australian primary care

Abstract

Aim

To investigate the ability of eight frailty instruments to accurately predict all-cause mortality and other adverse outcomes in Australian primary care patients.

Methods

Study participants included adults aged ≥75 years attending one of three primary care clinics in South Australia. Frailty instruments studied were Fried's frailty phenotype (FFP), the Frailty Index (FI) of cumulative deficits, Kihon Checklist (KCL), the Fatigue Resistance Ambulation Illness and Loss of weight (FRAIL) scale, Groningen Frailty Indicator (GFI), PRISMA-7, Reported Edmonton Frail Scale (REFS), and gait speed. Primary outcomes were all-cause mortality at 12- and 24-months. Secondary outcomes included falls, general practice attendance, hospital admission and emergency department (ED) presentation at 12-months.

Results

243 participants (55.6 % female) with a mean (SD) age of 80.2 (4.6) years were included. 29 participants (16.6 %) were classified as frail at baseline by FFP. All frailty instruments demonstrated a significant ability to predict 12- and 24-month mortality. The REFS showed the highest _auROC for both 12- and 24-month mortality. The REFS, Frailty Index, Kihon Checklist, FRAIL scale, and gait speed showed excellent discriminative ability for 12-month mortality (auROC ≥ 0.8 - >0.9), while the remainder showed acceptable discrimination. All frailty instruments, with the exception of the GFI, showed an excellent discriminative ability for 24-month mortality (_auROC 0.8-<0.9).

Conclusions

All frailty instruments possessed adequate discriminative ability for all-cause mortality predicting in older primary care patients. Frailty measurement is thus a valuable strategy to identify older patients at risk of mortality and can guide clinical decision-making in primary care settings.