Abstract
OBJECTIVE: In many industrialised countries, health care third party payers are moving towards contracted provision arrangements with suppliers of hospital care. Essential to such a process is a standard approach to quantifying the care provided. This paper aims to outline the possible approaches to hospital product definition for the UK National Health Service, and recommends appropriate further research. METHODS: All published and unpublished studies on hospital output measurement in the NHS since 1980 were sought for the purposes of the review. This included both discursive and empirical work, and no exclusion criteria were applied. Most empirical reports on this topic, however, come from the United States. Consequently, the published reports since 1980 from the USA, accessed from the Medline and Healthplan CD-ROM databases, have also been included in the overview. CONCLUSIONS: Where data are sufficient, the true casemix approach offers advantages over other methods of output measurement. In the UK NHS, two systems--diagnosis-related groups (DRGs) and healthcare resource groups (HRGs)--are the only casemix measures that have achieved any significant degree of attention. DRGs have been extensively evaluated internationally, and explain variations in resource use in the UK slightly better than do HRGs. As a local product, HRGs can be more easily adapted to the specific needs of the NHS internal market, however, and will thus probably emerge as a better measure for the UK in the long term. In both cases, locally derived cost weights are unavailable, and their development constitutes a major requirement for use in contracting. Adaptations for long stay and outpatient hospital episodes would enhance the usefulness of hospital casemix systems in the NHS. Existing approaches, such as specialty based classifications, are neither standardised nor predictive of resource use, and would be better replaced by casemix systems. Other countries facing similar choices between casemix measurement approaches will need to consider the "trade off" between the adaptability of locally derived systems on the one hand and the low cost, rapidly accessible results, and availability of international comparative data of an imported approach on the other.