Giovanna Boccuzzo, Università di Padova
Giulio Caperna, Joint Research Centre of the European Commission
Nicolò Preo, Università di Padova
The average age of Italy’s population is rising, together with chronic conditions related with ageing. This brings to huge challenges for health services. Frail subjects have special care needs, thus, their identification is important to implement effective prevention policies. Indeed, they are associated to an increased “risk of adverse outcomes”. The goal of this work consists in the construction of an indicator to measure the frailty level of individuals, based on a few number of variables able to predict the outcomes of frailty. Analyses use the administrative health data-flows of the former Local Health Units 15 in the Veneto region (now part of "Ulss 6 Euganea"). We selected five outcomes related with frailty condition: death, emergency hospitalization, fracture, disability and dementia. However, these outcomes have different risk factors, thus the selection of a unique and parsimonious set of variables to constitute the frailty indicator is not trivial. Variables selection is implemented through 100 logistic regression models with stepwise criterion for every outcome on a balanced subsample of the whole population. We selected eight variables according to their appearance percentage: age, invalidity, use of home care services, cancer, depression, poli-prescription, anemia, and accesses to first aid. These variables assemble the frailty indicator through partially ordered set theory. The indicator is associated to high areas under the ROC curve with respect to all the outcomes (0.835 for death, 0.743 for emergency hospitalization, 0.637 for fracture, 0.807 for disability and 0.784 for dementia).
Presented in Session P2. Poster Session Ageing, Health and Mortality