Objective To compare the utilisation of hospital beds in the NHS in England, Kaiser Permanente in California, and the Medicare programme in the United States and California. Results Bed day use in the NHS for the 11 leading causes is three and a half times that of Kaiser's standardised rate, almost twice that of the Medicare California's standardised rate, and more than 50% higher than the standardised rate in Medicare in the United States. Kaiser achieves these results through a combination of low admission rates and relatively short stays. The lower use of bed days in Medicare in California compared with Medicare in the United States suggests there is a "California effect" as well as a "Kaiser effect" in hospital utilisation. Conclusion The NHS can learn from Kaiser's integrated approach, the focus on chronic diseases and their effective management, the emphasis placed on self care, the role of intermediate care, and the leadership provided by doctors in developing and supporting this model of care.

The development of casemix in rehabilitation poses similar challenges for healthcare systems all around the world. Well-established casemix systems in the United States and Australia have afforded valuable lessons for other countries to learn from, but have not provided all the answers. A range of casemix and payment models is required to cater for different healthcare cultures, and casemix tools must capture all the key cost-determinants of treatment for patients with complex needs.

Total unit treatment costs are obtained from retrospective analysis of provider hospitals' budget and accounting statements. Mean bed-day costs (total unit cost/occupied bed days) are divided broadly into 'variable' and 'non-variable' components. In the weighted costing model, the band-weighting factor is applied to the variable portion of the bed-day cost to derive a banded cost, and thence a set of cost-multipliers. Preliminary data from one unit are presented to illustrate how this weighted costing model will be applied to derive a multilevel banded payment model, based on serial complexity ratings, to allow for change over time.