Identification of neighborhood clusters on data balanced by a poset-based approach

Abstract The identification of territorial clusters where the population suffers from worse health conditions is an important topic in social epidemiology, order to identify inequalities cities and provide policy interventions. This objective particularly challenging because mechanism self-selection individuals into neighborhoods, which causes selection bias. aim this paper consists neighborhood elderly people living Turin, a city north-western Italy, are exposed increased risk hospitalized fractures. study based on administrative data retrospective, observational cohort study. It composed by first phase, individual confounding variables balanced across neighborhoods make them comparable, second phase aggregated characterized significantly higher risk. In we exploited balancing technique partially ordered sets (poset), called Matching poset Average Rank for Multiple Treatments (MARMoT). On dataset, used spatial scan presence checked whether fracture some contiguous areas. combination both MARMoT procedure makes it possible highlight two Turin incurring fractures than mean. These results could have implications implementation policies.