LAUSR

INTRODUCTION Social frailty is a complex concept and there is still no consensus on the criteria that best define it, nor on the role that social dimensions play in well-established frailty models. AIM To analyse the predictive value of social frailty dimensions on distinct frailty models. METHOD A non-probabilistic sample of 193 community-dwelling adults aged 65 years and over was recruited in 2016 and followed for three years. Frailty was assessed by the Tilburg Frailty Indicator, the Groningen Frailty Indicator, and the Fried Phenotype criteria. Questions about living alone, social network, social support, loneliness, and frequency of social activities engagement were used to assess social criteria. Bivariate correlations and sequential multiple hierarchical logistic regression analyses were performed. RESULTS At baseline, 22.2% older adults lived alone, 47.2% reported missing people around them, 21.1% reported lack of social support, 26.1% reported having reduced their participation in social activities recently and 52.2% reported loneliness. The percent of frail individuals varied across frailty measures, and social criteria showed significant correlations and increased the prediction of frailty status. Loneliness and social activities engagement were associated with frailty as assessed by the Tilburg frailty Indicator and by the Fried Phenotype criteria; the lack of social support is associated with frailty as assessed by the Groningen Frailty Indicator. Living alone and lack of social relationships did not predict frailty. CONCLUSION Including social dimensions in a frailty model needs a consensual theoretical basis as they have different roles in predicting frailty, varying over time and across assessment tools.