Data envelopment analysis

Data envelopment analysis: an application to investigate differences in performance in NHS Scotland

Hospital performance in Scotland is usually measured using uni-dimensional approaches, e.g. cost per case, length of stay. These can be useful indicators. However, because hospitals are multi-dimensional in their output this rarely captures an overall picture of how efficiently a hospital is functioning. This study investigated the application of a multi-dimensional approach to measuring hospital efficiency in Scotland: data envelopment analysis. This is a non-parametric method of measuring efficiency in organisations, most commonly used where the relative efficiency of a small number of organisations (such as hospitals in Scotland) is being analysed. We estimated the relative technical efficiency (TE) of Scottish hospitals under the assumptions of variable (VRS) and constant returns to scale (CRS) and tested for the importance of controlling for case-mix. We found that teaching and large general hospitals were the most efficient and that there was no discernible improvement in efficiency over time. Case-mix adjustment had a small impact on efficiency scores but for some individual hospitals this had a substantial impact on their ranking and TE score. Using time-series regression analysis to gauge changes without the obfuscation by white noise we found a definite and statistically significant trend of improvement, from 2003 to 2007.

Outcome and Translation

The findings of this study addressed questions and objectives raised in the NHS Scotland Efficiency and Productivity Delivery Framework. They informed the SGHD via the Efficiency and Productivity Information Development Group. The research identified differences in efficiency of NHS organisations and the characteristics of those organisations. Such information is vital for targeting improvements in performance.

HERU researchers involved in this research project: Shelley Farrar

External collaborators: M Tavakoli (University of St Andrews) and M Rosko (Widener University, USA)

Publications

Valdmanis, V., Rosko, M., Mancuso, P., Tavakoli, M. and Farrar, S. (2017) 'Measuring performance change in Scottish hospitals: a Malmquist and times-series approach', Health Services and Outcomes Research Methodology, 17(2), 113-126.

Presentations

Rosko, M., Tavakoli, M. and Farrar, S. (2012) 'Variations in relative efficiency in Scottish hospitals: a DEA approach', European Conference on Health Economics, Zurich. July 2012.

Rosko, M., Tavakoli, M. and Farrar, S. (2012) 'Variations in relative efficiency in Scottish hospitals: a DEA approach', Efficiency and Productivity Information and Development Group, Scottish Government Health Department, Edinburgh, August 2012.