Observed expenditures vs estimated burden of health care: a comparative evaluation based on spatial analysis

In the context of increasing life expectancy and growth of the elderly population, the assessment of time and spatial patterns of over 65 population health care need is a key step in order to better manage public resources (Gray, 2005). The aim of this study is to highlight the existence of spatial heterogeneity in the elderly healthcare burden, comparing alternative modelling approaches, in the context of Regione Friuli Venezia Giulia (FVG). Data on estimated health burden in 2017 and 2018 were aggregated on age classes within each municipality. The population size, the ratio between males and females, and the death rate, the counts of 21 chronic conditions, the Resource Utilization Band (RUB) indicator, and the expenditures for healthcare services (Pharmaceutical, Hospital, and Outpatient types) in years from 2002 to 2017 were also collected. A descriptive analysis both of ageing phenomenon and of health care expenditures trends has been performed. The availability of the RUB indicator, provided in the John Hopkins ACG System (version11.1.2), allows comparing observed healthcare expenditures (HCE) with the estimated healthcare burdens. In particular, different spatial econometrics models (such as those discussed in Elhorst, 2014; Moscone and Tosetti,2014; Le Sage and Pace, 2009) have been compared to explore spatial heterogeneity of the differences between demand and health need. The analyses are developed on the full population and also focusing on the elderly population only. The empirical evidence shows that while HCE does not present any spatial pattern, the RUB indicator is characterized by some strong geographical clusterization even after controlling for the demographical structure of municipalities. In order to model the spatial heterogeneity, an SDM specification is chosen after an appropriate set of tests. The spatial patterns of morbidities play an important role in the explanation of the healthcare burden, together with the economic characteristic of the municipality. The model estimation, based on the elderly subpopulation, provides further insights on the diseases mostly influencing the healthcare burden, namely age macular degeneration, human immunodeficiency virus and low back pain. Surprisingly, the focus on the subpopulation points out that elderlies living in areas with higher shares of elderly population are healthier and needs fewer resources than their peers in other areas.