Exploring Variations in the Opportunity Cost by Clinical Area: Results from a Feasibility Study in England

The National Institute for Health and Care Excellence (NICE) assesses the effectiveness and cost effectiveness of new medical technologies on behalf of the English National Health Service (NHS) to support efficient allocation of health care resources. The 2015 seminal paper by Claxton et al. estimated an average cost effectiveness threshold for the English NHS measuring the marginal effect of changes in healthcare budget on mortality reduction across different clinical areas or programme budget categories (PBCs).

The aim of this study is to provide further empirical evidence on the relationship between health outcomes and health expenditures in England by (1) extending the analysis to include different health outcomes, and (2) examining the relationship between mortality and health expenditures along the mortality distribution.

Data Envelopment Analysis (DEA) is applied to multiple health outcomes and health expenditure data from 151 Primary Care Trusts (PCTs) in England across seven PBCs. Quantile Regression (QR) is applied to analysing mortality outcomes and health expenditure data for six PBCs. Comparisons are made between results from our study and the preferred outcome specification models estimated by Lomas, Martin and Claxton (2018). Finally, we compare the ranking of PCTs according to the DEA efficiency scores and outcome elasticities estimated in the QR approach.

The results provide evidence of heterogeneity across PCTs and PBCs as to how health resources are used to improve outcomes. Efficient PCTs tend to have lower absolute levels of mortality elasticity to health expenditure than inefficient PCTs.

This study provides empirical evidence suggesting that (1) estimates of the opportunity cost of introducing new technologies based on average performance of efficient and less efficient commissioners are biased downwards and subject to great variation, and (2) different PCTs have different production functions. There is not a common production function for providers underlying a common threshold.