NHS reference costs: a history and cautionary note

Historically, the NHS did not routinely collect cost data, unlike many countries with private insurance markets. In 1998, for the first time the government mandated NHS trusts to submit estimates of their costs of service, known as reference costs. These have informed a wide range of health economic evaluations and important functions in the health service, such as setting prices.Reference costs are collected by progressively disaggregating budgets top-down into disease and treatment groups. Despite ongoing improvements to methods and guidance, these submissions continued to suffer a lack of accuracy and comparability, fundamentally undermining their credibility for critical functions.To overcome these issues, there was a long-held ambition to collect "patient-level" cost data. Patient-level costs are estimated with a combination of disaggregating budgets but also capturing the patient-level "causality of costs" bottom-up in the allocation of resources to patient episodes. These not only aim to capture more of the drivers of costs, but also improve consistency of reporting between providers.The change in methods may confer improvements to data quality, though judgement is still required and achieving consistency between trusts will take further work. Estimated costs may also change in important ways that may take many years to fully understand. We end on a cautionary note that patient-level cost methods may unlock potential, they alone contribute little to our understanding of the complexities involved with service quality or need, while that potential will require substantial investment to realise. Many healthcare resources cannot be attributed to individual patients so the very notion of "patient-level" costs may be misplaced. High hopes have been put in these new data, though much more work is now necessary to understand their quality, what they show and how their use will impact the system.