Small area estimation of severe functional limitation from Italian data of the European Health Interview Survey

Abstract

This paper focuses on the methodology used to estimate the indicator of severe functional limitation (SFL) using data collected from the European Health Interview Survey (EHIS) in Italy. While direct estimates of SFL are reasonably accurate up to the regional level (NUTS2), there is a demand for more detailed estimates at the provincial level (NUTS3), disaggregated by sex and two age groups (15-64 years and 65 years and above). This requires the computation of estimates for 428 unplanned domains. To address this challenge, a small area estimation approach based on an area-level model has been applied, integrating auxiliary information known from administrative registers with EHIS data. To meet the assumptions of the model and ensure in this way a better accuracy of the final estimates, the model has been specified on a log-transformation of direct estimates. The case study presented here is one of the first attempts at obtaining small area estimates for unplanned domains within the EHIS survey and the results obtained are very promising.