Feature / Data mining

The annual NHS reference costs collection produces a rich seam of data, and drilling deep into the figures reveals nuggets of information. But what, if anything, do they tell us, asks Seamus Ward.

Anyone who has taken even a passing glance at the NHS reference cost documents will know they contain a mountain of data. And although some would dispute the usefulness of reference costs, the annual national collection could provide clues on NHS efficiency, the accuracy of coding, the impact of payment by results policy and future movements in tariff.

The reference costs for 2008/09 were published at the end of January, collecting data from more than 400 NHS provider organisations and covering expenditure of almost £48bn – a rise of about £4bn over 2007/08.

But change lies ahead. Last autumn the Department of Health and the Audit Commission launched a joint review of reference cost collection, which examined both the quality of the data (including how patient level information and costing systems interact with reference cost collection) and how NHS bodies use the cost data.

Both elements of the review are due to be completed this month. The Department initially aimed to use early lessons from the review to influence the reference cost collection for 2009/10, but this did not appear to be the case when the 2009/10 collection guidance was issued in February.

Evidence of apathy

The review is likely to find a degree of apathy towards reference costs and some suspicion the figures are not as robust as they could be. Some finance managers, including many at the sharp end of costing, have lost faith in the veracity of reference costs, particularly as trusts move to patient level information and costing systems.

‘Quite simply, I do not see reference costs as relevant any more,’ says one costing accountant. ‘Patient level costing is the most relevant thing we are looking at at the moment. Basically, while reference costs are used for tariff weightings there are so many adjustments that you cannot read too much into tariff from these anymore. I have very little confidence in their accuracy.’

Others feel the move to best practice tariffs and the indication that tariff will in future be seen as a maximum price has undermined one of the raisons d’être for reference costs – providing the information on which tariffs are based. The Department would argue the national snapshot of efficiency given by the annual reference costs is invaluable and the main reason for collecting the costs.

One finance director told Healthcare Finance that reference costs are used in benchmarking, ‘but they only tend to be quoted to support someone’s argument when your reference costs are better than average. When you appear more expensive, it’s because the data is not good enough.’

Perhaps reference cost data is more useful to those a few steps removed from the frontline – regulators, academics, economists and commentators? King’s Fund chief economist John Appleby believes reference costs provide information that can be used to measure the efficiency of the NHS, but he is also sceptical about the accuracy of some of the information.

He recently used reference costs to analyse NHS productivity after the Audit Commission concluded in its briefing More for less, published in November, that unit costs for elective and non-elective care fell by 2% in cash terms in 2007/08. He calculated the unit cost trends in elective inpatient, elective day case and non-elective activity at 2006/07 prices (taking out inflation) by dividing total costs by total activity for each area. Total unit costs fell by about 5%, but elective inpatient and day case unit costs rose by 4% and 1% respectively. The fall in total unit costs was due to a 10% reduction in the cost of non-elective work.

‘If the numbers are correct, then it is arguable whether becoming more productive in non-elective work is really something to cheer about: rising emergency activity has been a problem in the NHS for some time,’ he says. 

Overall, the NHS was more productive but when Professor Appleby applied his formula to the 2008/09 reference costs the picture was not so good. Although it does not take account of changes in quality, elective inpatient costs increased by another 5% and, while elective day case costs remained steady, non-elective costs rose by almost 2%. Overall unit costs increased by 2%, which Professor Appleby describes as ‘disappointing’.

While acknowledging the role reference costs play in creating the national schedule of reference costs and thus the tariff, the Department views the information first and foremost as a way of benchmarking costs across England. Traditionally, the main focus for this efficiency measure has been on the reference costs index (RCI), which ranks organisations based on the cost of delivering the same activity. An organisation with an RCI of 100 has costs equal to the national average – less than 100 means the trusts’ costs are below average; more than 100 above average.

In 2007/08, the RCI ranged from 162 (62% above national average costs) to 81 (19% below). These figures include an adjustment for the market forces factor, which takes into account the unavoidable cost differences facing organisations in different parts of the country. The equivalent figures from the 2008/09 index show that RCIs varied between 148 and 75.

Statistical contradictions

This would suggest a significant increase in efficiency over the year, though in the 2007/08 collection there were data problems at the trust with the highest RCI, Great Ormond Street NHS Trust. In the latest index, the trust’s RCI has fallen to 116.

Community providers, mental health trusts, single specialty providers and ambulance trusts are usually stripped out of the RCI, leaving only acute providers when comparing performance from year to year. There are inconsistencies in data collection in community and mental health organisations, while specialty trusts, such as those for orthopaedics or children’s care, tend to have a more complex (and thus more expensive) casemix so they suffer by comparison with other trusts.

The range for acute providers in 2008/09 stretched from 119 (Guy’s and St Thomas’ NHS Foundation Trust) to 79 (North Middlesex University Hospital NHS Trust). In 2007/08 the range was 116 down to 81, so in 2008/09 there was a slight increase in the highest RCI and a slight reduction in the lowest RCI.

Before drilling down too deeply into the other part of the reference cost publications – the schedules, which reflect trusts’ average costs for providing treatment and procedures in a national average cost for each HRG and, for the first time this year, data for each organisation – it is apparent activity has increased. This could be due to better coding, but it seems the extra NHS funding in recent years is leading to more patients being treated.

However, the extra activity (for activity included under the tariff at least) does not appear to be additional elective inpatient finished consultant episodes (FCEs). Elective inpatient activity has held reasonably steady at around 1.7 million FCEs in each of the three years between 2006/07 and 2008/09. Day case activity has risen from about 4 million in 2006/07 to 4.7 million in 2008/09. But the biggest increase was in outpatient procedures – from 1.5 million in 2006/07 to almost 2.2 million in 2007/08 and 3.25 million in 2008/09.

The increase could be explained by a rise in the number of data submissions (4,883 in 2006/07 and 7,992 in 2008/09). But it could be concluded that the NHS was already moving in the direction of the Department’s policy – using the tariff to make the acute sector more efficient by driving down the length of stay and improving the patient experience. When PBR was first introduced for elective care, the Department sought to incentivise day case treatment, where appropriate, over admission to hospital. It did so by setting a price based on a weighted average of the ordinary elective cost and the cost of day cases. This meant the price would reward providers achieving higher than average day case levels and under-reward those whose day case rates were lower than average.

More recently the focus has been on developing ambulatory care and shifting procedures, again where appropriate, to settings such as outpatient clinics. The planned same day (PSD) tariff was introduced in the current financial year to incentivise a shift from day case to outpatient procedure, though the PSD tariff for outpatient procedures is non-mandatory, largely as a result of the Department’s view that the collection and coding of outpatient procedure data was patchy – which casts doubt over the dramatic rise in outpatient procedure activity.

After feedback from the service that local negotiation of outpatient procedure tariffs has been difficult, the PBR guidance for 2010/11 removed the PSD tariff for day cases and the non-mandatory tariff for outpatient procedures. The Department plans to introduce mandatory tariffs for 49 high-volume outpatient procedure HRGs in 2010/11. Other outpatient procedure HRGs will be paid as part of the relevant mandatory outpatient attendance tariffs. Future reference cost schedules will show whether the rises in outpatient procedures will be sustained.

A comparison of non-elective activity recorded in the past three years’ reference costs, collected using HRG4, shows why many in the NHS are uncomfortable with the 30% marginal rate for emergencies to be introduced in 2010/11. It also shows why the Department is trying to curb emergency admissions.

Activity levels for 2008/09 (costed at 2010/11 tariff prices) will be used as the baseline for the marginal rate and, according to the 2008/09 reference costs, non-elective inpatient activity (long and short stay) stood at 7.9 million finished consultant episodes. This was up from 7.4 million in 2007/08 and 6.2 million in 2006/07 – almost a 20% rise in the first year and 7% in the second.

Finance managers point out there were changes in the way short stay admissions were recorded in 2007/08 (activity was no longer collected by ward type but length of stay). But it is unlikely this alone accounted for the increased non-elective activity. Anecdotally, trusts are reporting non-elective activity in the current year increasing by up to 10%.

Reference costs are a goldmine of information. And while some feel this is fool’s gold, they can offer clues about the relative efficiency of NHS organisations and how services are reacting to national and local policy on payment by results.