Interpreting the NHS cost indices for acute Trusts

The 1997 White Paper, The New NHS, announced that Trusts would be required to publish and benchmark the costs of the treatments they provide on a similar basis. Between November 1998 and March 1999 five different indices were produced purporting to measure the unit costs of service provision in English acute Trusts. The publication of multiple indices has meant that Trusts appearing ‘inefficient’ on one index can emerge as relatively ‘efficient’ on another. This has the potential to create confusion. Of the five indices, the reference cost index (RCI) is the least satisfactory. This index fails to fully capture the range of service provision or to account for factors known to influence hospital costs, some of which are beyond immediate managerial control. Hence, interpretation of the RCI as evidence of poor performance cannot be justified. By explicitly taking such factors into account, the casemix cost indices (CCIs) are a considerable improvement in the RCI. However, the efficiency scores generated by the CCIs are point estimates that are not accompanied by any range of uncertainty. By replicating the econometric models to produce the CCIs it is possible to calculate confidence intervals around the efficiency scores. Comparison of the confidence intervals suggests that differences in Trust unit costs (and, by implication, efficiency) are not statistically significant. As such, differential performance targets based on unit cost data would be unfair. This paper discusses possible reasons why the econometric analysis fails to identify variations in unit costs. It is suggested that there may be more fruitful means to encourage performance improvements among NHS acute Trusts than general target setting based in unit cost data.