63A FRAILTY PROFILE FOR USE IN THE COMMUNITY: INCLUSION OF PSYCHOLOGICAL VARIABLES IMPROVES PREDICTION OF CARE NEEDS AND FALLS

Abstract

Introduction: Frailty is defined as a state of high vulnerability for adverse health outcomes when exposed to a stressor. Previous studies have described accumulated deficits profiles of frailty, and their use to predict mortality and institutionalisation is well evidenced. However, community use of frailty assessment in non-medical facilities to inform social care support decisions, lifestyle and prevention strategies has been less explored. This paper describes the validation of a community-based frailty index based on selfdeclared diagnoses and objective assessments. Importantly, it includes psychological as well as physical variables. Methods: Prediction of outcomes was examined using regression analyses: level of homecare, functional impairment, limitation in social engagement, health service use (GP visits, hospital admissions, duration of stay), falls and death, based on data from older people living independently in extra care retirement villages. Results: Area under the curve (AUC) compared goodness of fit for mortality with other published frailty indices, comparing favourably. Care level and whether or not someone received social care 12 months later was reliably predicted in linear regression. Baseline Physical Frailty predicted 18.4% of the variance in care needs 12 months later but cognitive function added 11.4% and depression a further 2.5%. In logistic regression the tool reliably predicted whether or not someone received social care (Chi-squared = 68.04, p < 0.001) and could also distinguish between those who had a fall or not over the next 24 months (Chi-Squared = 10.62, p < 0.01). Conclusion: Frailty indices can be used to predict a range of outcomes that are useful in community health and non-clinical environments. Acceptability of use of the tool is being explored with well-being advisors based in extra care villages (community nurses) and older adults themselves to inform health behaviour goal setting, and work to further reduce the number of variables is under way.