Home > Research > Publications & Outputs > Measuring hospital efficiency
View graph of relations

Measuring hospital efficiency: comparing four European countries

Research output: Contribution to journalJournal article

Published
Close
<mark>Journal publication date</mark>02/2015
<mark>Journal</mark>European Journal of Public Health
Issue numberSuppl 1
Volume25
Number of pages7
Pages (from-to)52-58
Publication statusPublished
Original languageEnglish

Abstract

BACKGROUND: Performing international comparisons on efficiency usually has two main drawbacks: the lack of comparability of data from different countries and the appropriateness and adequacy of data selected for efficiency measurement. With inpatient discharges for four countries, some of the problems of data comparability usually found in international comparisons were mitigated. The objectives are to assess and compare hospital efficiency levels within and between countries, using stochastic frontier analysis with both cross-sectional and panel data.

METHODS: Data from English (2005-2008), Portuguese (2002-2009), Spanish (2003-2009) and Slovenian (2005-2009) hospital discharges and characteristics are used. Weighted hospital discharges were considered as outputs while the number of employees, physicians, nurses and beds were selected as inputs of the production function. Stochastic frontier analysis using both cross-sectional and panel data were performed, as well as ordinary least squares (OLS) analysis. The adequacy of the data was assessed with Kolmogorov-Smirnov and Breusch-Pagan/Cook-Weisberg tests.

RESULTS: Data available results were redundant to perform efficiency measurements using stochastic frontier analysis with cross-sectional data. The likelihood ratio test reveals that in cross-sectional data stochastic frontier analysis (SFA) is not statistically different from OLS in Portuguese data, while SFA and OLS estimates are statistically different for Spanish, Slovenian and English data. In the panel data, the inefficiency term is statistically different from 0 in the four countries in analysis, though for Portugal it is still close to 0.

CONCLUSIONS: Panel data are preferred over cross-section analysis because results are more robust. For all countries except Slovenia, beds and employees are relevant inputs for the production process.