Frail subjects have special care needs, so it is necessary to know their conditions in order to intervene in an emergency situation. The identification of these particular subjects, using a common criterion, is not trivial, in fact in the literature there are several definitions of frailty, which makes difficult to find a unique operative criterion.

Frailty is a multidimensional concept (involves several functional domains) characterized by an increased risk to adverse outcomes. We selected a group of outcomes and their corresponding risk factors in order to build a robust frailty measure with respect to all the considered outcomes.

We proposed a new approach to measure frailty, based on partially ordered sets (Poset) to use the entire amount of ordinal information from our data. With this method, we are not strictly bound to only one outcome (as in a regression), and we do not add arbitrary choices. Moreover, we are able to combine in a frailty measure different kinds of variables (dichotomous and ordinal) that reflect both events and charac-teristics of the considered population.

We obtained a rating of individuals that provides a simplification of a complex and multidimensional con-cept as the fragility.