Some regression-based indicators of hospital performance

The development of performance indicators for the NHS has resulted in renewed interest in regression-based measures of hospital performance. This paper considers three regression-based measures of hospital efficiency, including the index of hospital ‘costliness’ proposed by Martin Feldstein and currently used by the DHSS as a financial performance indicator. The paper begins with a critique of Feldstein’s index and then goes on to show how its defects can be overcome using stochastic frontier regression models. Two such models are considered: a cross-section model in which it is assumed that inefficiency follows a particular distribution and a panel-data model in which it is assumed that inefficiency remains constant over the sample period. All three approaches are illustrated using data from 49 acute hospitals owned and operated by the Spanish Ministry of Health. The paper is written with the non-economist in mind and is aimed primarily at statisticians and other persons in the NHS who have an interest in performance indicators. It may also be of interest, however, to persons with a general interest in the measurement of the efficiency of public sector organisations.