Estimating the costs of specialised care

In most sectors of the economy, specialisation is associated with lower costs. Yet some specialised hospitals claim to require more generous funding than general hospitals. This claim is based on the assertion that their patients are different, and that these differences outweigh the cost advantages of specialisation. Unless the basis for this claim can be established, the financial incentives introduced by Payment by Results to encourage cost reducing behaviour will be diluted. We estimate various multiple regressions in order, firstly, to establish the extent to which the receipt of specialised care is associated with higher treatment costs and, secondly, to evaluate hospital performance in controlling costs. We explore how robust the results are to a range of analytical choices by conducting various sensitivity analyses. We use the Hospital Episode Statistics and Reference Cost databases to analyse the characteristics and costs of all patients treated in the NHS during 2008/9. Patients are identified as having received specialised care on the basis of specific diagnostic and procedure codes recorded in their medical record. These codes are agreed by clinicians and form the Specialised Services National Definition Sets. We estimate multiple regression models to assess the extent to which receipt of specialised care increases the cost of treatment. We test the robustness of results to choices about how costs are calculated, how the regression models are specified and how patients are identified as having received specialised care. In addition we assess each hospital’s relative efficiency in controlling costs, after allowing for differences in factor prices and a wide range of patient characteristics. We find that, after allowing for the hospital in which treatment is provided, costs are higher than for other patients allocated to the same Healthcare Resource Group (HRG) if a patient receives one of the following types of specialised service: cancer (18% higher cost), spinal (28%), neurosciences (23%), cystic fibrosis (38%), infectious disease (21%), children (20%), rheumatology (13%), vascular diseases (21%), colorectal (21%) and orthopaedic (21%). The implication for Payment by Results is that ‘top-up’ payments for patients with these markers might be made over and above the tariff associated with the HRG to which they are allocated. We recommend that the size of additional top-up amounts to the percentage increase in costs as reported above, these estimates being derived from our preferred model specification. However, different values could be adopted, justified on other grounds. These grounds may include: Transitional arrangements, notably for children’s services, where the recommended value of 20% is substantially lower than the current 78% top-up; Materiality, where an additional top-up would have limited financial consequence for those types of specialised services that are delivered to only a small number of patients; Sensitivity to model specification. The other model specifications generally imply lower top-up values than those recommended above, with the exception of a model that fails to allow for each hospital’s influence on costs.