open space:

My hospital, like over 200 others, has recently calculated reference costs for almost all inpatient and day-case activity. This meant producing over 500 separate costs for healthcare resource groups. It kept several people busy for weeks, and took up hours of consultants' and nurses' time. It cost over£2,000 for paid overtime, not counting unpaid overtime by senior staff. You would expect some results for all this effort, but reference costs may not deliver for several reasons.

Inconsistencies in data and treatment

The NHS counts activity inconsistently. If one hospital counts on average 103 episodes per 100 admissions, and another counts 108, you cannot trust reports which show the second hospital's costs per episode as 5 per cent lower.

Services in primary care affect hospital costs. If (as at my hospital) some orthopaedic patients go to nursing homes to recover, paid for by the health authority, the hospital's length of stay and costs will appear low. Actual costs in any year are also distorted by exceptional items, such as large redundancy or clinical negligence costs.

Potential uses of reference costs

If reference costs are to encourage change (and the cost of producing them is only justified if they do), it can only really be in these ways:

Change what is done. Reference costs might enable commissioners to reduce high-cost, low-benefit interventions, and use the money on HRGs, which offer more patient benefit per pound cost. However, this is difficult. As only high-volume HRGs are costed, the NHS is unlikely to stop doing such work completely - as it is so necessary and common - and may find it hard to reduce volumes.

Change where it is done. Move services from high-cost to low-cost providers. This is difficult in practice, and little money can be saved by moving a few HRGs. You have to move whole services, and that decision doesn't require costs at HRG level.

Change how it is done. Help high-cost providers reduce their costs. A main aim of reference costs is 'to identify cost differences and help identify the reasons behind them', according to former health minister Alan Milburn.

Reference costs may well indicate, subject to the inconsistencies outlined above, that a certain hospital has unusually high costs. But they provide almost no information on why costs are high.

Only total costs and length of stay are shown for any HRG. If length of stay is the main reason for high costs, then it's barely worth preparing reference costs at all, as detailed comparisons of length of stay are routinely available elsewhere. If length of stay isn't the problem, then reference costs don't give a clue as to what the problem is.

Not surprisingly, it proves hard to engage clinical directors in cost reduction on this limited evidence. Furthermore, efficiency targets set using reference costs may result in arbitrary cutbacks without addressing the real problems.

Comparative costs

Nothing will produce sensible comparative costs until hospitals count activity the same way. One option is to expand reference costs so that they provide comparative information on all the main cost drivers. This could be done by putting all hospitals on to the same, enhanced costing system and imposing consistency in how they use it.

Another option is to abandon most of reference costing. From each hospital's casemix a measure of casemix complexity, based on length of stay by HRG, would be calculated. Hospitals would submit their specialty costs in some detail and with some amalgamation of sub-specialties to improve consistency. They could compare the detail of their specialty costs against any hospital with similar casemix complexity.

Another, more radical option is to calculate HRG costs at a few sites only, share the details across the NHS, and use them to derive an NHS price for each HRG. Trusts would earn income based on their HRG workload, but adjusted for regional differences (the market-forces factor, or whatever replaces it). This would keep some of the pressures of a market without competition.

The first option achieves what is needed, but at a cost and with less flexibility. The second is a compromise, and the third a radical alternative with risks. They could all be more effective than what we do now.

Steven Bliss is assistant director of finance, Scunthorpe and Goole Hospitals trust.