Feature / Good vibrations

Patient-level costing systems have directly delivered only modest efficiencies to date. However, an assessment of the more granular approach to costing by the Nuffield Trust concludes that trusts without the high-quality cost information produced by the systems will find it ‘challenging to ensure their efficiency savings are cutting waste, not care’.

The report – Patient level costing: can it yield efficiency savings? – describes patient-level costing as a powerful tool for healthcare organisations. ‘When implemented well, it provides a vast array of useful and accurate data on distributions of expenditure and profitability against income.’ However, it warns that the systems themselves do not deliver savings. Instead savings depend on what is done with the data.

‘The information it gives is clearly very valuable,’ says Ian Blunt, senior research analyst for the Nuffield.  ‘The potential is there. It depends to an extent on the willingness of management to act [on the data].’

The report studied data for one year (2008/09) at one trust that implemented a patient-level costing information and costing system (PLICS) in 2007 and drew on the experiences of a broader group of user trusts. The researchers acknowledge that ‘PLICS is a long-term investment rather than a short-term saving’ but still remark that examples of changed clinical behaviour on the back of the better costing data are surprisingly rare.

Mr Blunt recognises that some of this is to do with the need to establish robust cost data. Clinicians’ natural first instinct can be to question the accuracy of cost data. So the initial focus of any PLICS programme has to be on getting the data right, which includes involving clinicians in validating the allocation methods used. The report suggests PLICS has in fact helped improve dialogue on cost between managers and clinicians and between clinicians.

But actual case studies that demonstrate achieved savings are more elusive. Mr Blunt says the sector needs to start providing or compiling greater evidence of savings. The report offers one example. Analysis in its example trust highlighted patients with

similar conditions having differing numbers of tests. After discussions with clinical directors, three tests were removed from the order set, resulting in savings of about £15,000.

The researchers say many sites recommend the identification of clinical champions who can set aside dedicated time to support costing work and lead discussions with other clinicians. This chimes with anecdotal evidence from around the NHS, where many organisations are at the stage of simply presenting data on cost and service variations back to clinicians, rather than formally challenging them on the reasons.

Although practical changes on the back of patient-level data are still emerging, the researchers were clear about the potential for use of the data. ‘To make efficiencies, a ripe place for scrutiny will clearly be those cases costing more than the tariff price,’ the report says. ‘Without patient-level costing such variations will not be clear to the trust, specialty or consultant team and the reasons for them cannot be understood, meaning that appropriate efficiencies cannot be made.’

Nuffield’s evidence

In the case of Nuffield’s example trust, overall tariff income for elective inpatient care exceeded the costs incurred. Within that setting, the costs of elective care for patients staying longer than two days exceeded the tariff, but this was outweighed by the surplus on shorter stay inpatients and day cases.

The same pattern was observed for non-elective inpatients, although the surplus from short stays did not eliminate the longer stay deficits. And the costs of both outpatient and accident and emergency care overall exceeded total tariff income. These patterns also varied across different healthcare resource groups.

The report also looks at the variability in costs within HRG categories relative to remuneration received through the national tariff – analysis only possible using patient-level data. Two HRGs at the sample trust were studied – F54 (inpatient care for inflammatory disease) and L09 (emergency inpatient care for kidney or urinary tract infections for people over 69 or with complications).

The graph shows that for this trust, the average cost of care for patients in each HRG was higher than the national tariff price. The graph also clearly highlights the distribution in costs. For L09, this distribution ranged from a few hundred pounds to more than £10,000. A fuller research document published alongside the report suggests the visual presentation of cost variations is key to using the data and provides a number of different examples in use.

The researchers stress that some of the variation uncovered will be unavoidable, but some would clearly be under the control of management or clinical staff and should be investigated. In total the report identifies a relatively poor match between care costs and tariff income across all HRGs. Only 17% of all cases (in the sample trust) had costs that fell within 10% of the appropriate tariff payment (including adjustments for long stays).

Poor alignment

And while these various surpluses and deficits tended to balance out, the report says the poor alignment at case level means any slight change in case type within an HRG or change in tariff price could lead to significant changes to the surplus-deficit ratio. The researchers also observed a clear relationship between tariff price and the chances of the spell being in surplus or deficit. The higher the price, the more likely the cost of care would exceed tariff.

The report highlights the potential to examine variation in cost by consultant team, for example using heat maps to show team performance across common HRGs. While again stressing that findings need to be read with care, taking account of different casemixes and ages even within the same HRG, the information was often worth exploring alongside outcome data. Was the fact that one consultant had higher theatre costs connected to the lower ward costs exhibited by their patients, for example?

Turning this type of data into the reduction of unnecessary variation will be the key next step for patient costing. But trusts do talk of other benefits. ‘The behaviour we noticed and some of the savings that were cited often relate to more accurate billing,’ says Mr Blunt. He points out that, while valid, this type of improvement does not save the system money.

The  report itself identifies other benefits aside from the retrospective analysis of costs – to enhance the accuracy of budget projections, for example, or help model changes in certain patient groups. It suggests PLICS could also move beyond hospitals and into primary and community services and become really powerful in understanding full pathway costs. There is already some international experience of using patient costing in primary care settings, albeit at a small scale.

And if PLICS information could be delivered on a more frequent basis – daily or weekly, say – costs could be ‘amenable to the surveillance techniques currently applied to monitoring mortality, allowing immediate intervention in a service if costs were consistently deviating from some expected value’. The report even suggests patient-level accounting data could have a role in helping to improve the evaluation of health IT projects in future.

There are some concerns though. The researchers warn PLICS may give hospitals an advanced understanding of their costs, something to which commissioners would not be guaranteed access. They say this ‘information asymmetry’ can have a detrimental effect on market efficiency. As

a precaution, they suggest policymakers mandate some sharing of cost information between all providers and commissioners, possibly following the model of regulatory accounts used in the utilities industry.

There are already hospitals that share data with commissioners. But there are also concerns about how shared data could be used. Mental health providers, generally less well advanced than acute providers in implementing patient costing, are understood to be worried about the use of patient cost data outside their organisation. Under current block contract arrangements, and to an extent under future local cluster pricing arrangements, providers fear commissioners could cherrypick information that backed up lower local prices for parts of their contracts – adding to financial pressures across the whole organisation.

But the overall tone of the report is positive, albeit with the proviso that the researchers would expect to see greater evidence in future of real savings made on the back of the data. ‘Management need to take the data on board and start having [what can be] hard conversations and talk to their clinicians about the resources being used,’ says Mr Blunt.