Development and validation of a frailty index based on Australian Aged Care Assessment Program data

Objectives: To develop and validate a frailty index, derived from aged care eligibility assessment data. Design: Retrospective cohort study; analysis of the historical national cohort of the Registry of Senior Australians (ROSA). Participants: 903 996 non-Indigenous Australians aged 65 years or more, living in the community and assessed for subsidised aged care eligibility during 2003–2013. Main outcome measures: 44-item frailty index; summary statistics for frailty index score distribution; predictive validity with respect to mortality and entry into permanent residential aged care during the five years after assessment. Results: The mean frailty index score during 2003–2013 was 0.20 (SD, 0.07; range, 0–0.41); the proportion of assessed older people with scores exceeding 0.20 increased from 32.1% in 2003–2005 to 75.0% in 2012–2013. The risks of death and entry into permanent residential aged care at one, three and five years increased with frailty index score level (at one year, high [over 0.35] v low scores [under 0.05]: hazard ratio for death, 5.99; 95% CI, 5.69–6.31; for entry into permanent residential aged care, 8.70; 95% CI, 8.32–9.11). The predictive validity (area under the receiver operating characteristic curve) of Cox proportional hazard models including age, sex, and frailty index score was 0.64 (95% CI, 0.63–0.64) for death and 0.63 (95% CI, 0.62–0.63) for entry into permanent residential aged care within one year of assessment. Conclusions: We used Australian aged care eligibility assessment program data to construct and validate a frailty index. It can be employed in aged care research in Australia, but its application to aged care planning requires further investigation.