Abstract

Background: European health systems face growing pressure from population ageing, post-pandemic service backlogs, and fiscal constraints. Yet substantial cross-country differences in health outcomes persist despite comparable levels of healthcare expenditure. This study evaluated the relative efficiency of European health systems using a multidimensional framework that integrates expenditure, prevention, and population health outcomes. Methods: A cross-sectional analysis was conducted on 25 European countries using 2022 data or the nearest available year. An output-oriented constant returns to scale Data Envelopment Analysis (DEA) model was estimated with two inputs, public and private healthcare expenditure per capita, and five outputs, life expectancy at birth, inverse infant mortality, healthy life years at birth, breast cancer screening coverage, and poliomyelitis vaccination coverage. A robustness specification added physician density as an additional input. Bootstrap bias correction with 1000 replications was applied to the baseline model. A second-stage Simar–Wilson truncated regression with 2000 bootstrap replications examined the association between inefficiency and selected contextual variables, including GDP per capita, population ageing, obesity prevalence, and tobacco use prevalence. Results: In the baseline DEA model, 8 of 25 countries were located on the technical efficiency frontier (Croatia, Czechia, Estonia, Greece, Hungary, Latvia, Lithuania, and Poland; output-oriented DEA inefficiency score = 1.000 for each country), while inefficiency scores among the remaining countries ranged from 1.042 to 2.617. The highest inefficiency scores were observed for Germany (2.617), Austria (2.283), Belgium (2.230), Ireland (2.219), and France (2.167). When physician density was added as an additional input, 12 countries were located on the estimated frontier. Bootstrap correction of the baseline model increased the estimated output-oriented inefficiency scores, with bias-corrected values ranging from 1.100 to 2.941. In the second-stage analysis, higher log GDP per capita was positively associated with bias-corrected inefficiency (coefficient 1.993; 95% bootstrap CI 0.219 to 4.197), whereas population ageing, adult obesity prevalence, and tobacco use prevalence were not statistically associated with bias-corrected inefficiency. Conclusions: In this cross-sectional sample of 25 European countries, higher healthcare expenditure was not consistently associated with frontier performance when health outcomes and preventive coverage were considered jointly. The results were sensitive to the inclusion of physician density and to bootstrap correction, supporting the interpretation of Data Envelopment Analysis as an exploratory benchmarking tool rather than a definitive ranking of health systems. These findings highlight the importance of assessing how financial and workforce resources are converted into measurable health and prevention-related outputs.

IPC Classification

Keywords