Feature / The right level

Reporting hospital performance at service line level – looking at income and expenditure within a hospital’s various ‘business units’ – is absolutely the right way to go. But if the approach is to drive performance, service line information needs to be built on an understanding of costs incurred at the patient level. This is the clear message from Chelsea and Westminster Hospital NHS Foundation Trust after two years of experience with service line reporting and patient-level costing (PLC).

Lorraine Bewes, director of finance and information at the foundation trust, believes PLC is the only way to engage clinicians. ‘The starting point has to be patient-level costing,’ she says. ‘It is not until you are sharing this level of granular data and you are really talking at the level that a clinician is dealing with day-to-day – that means patients – that you start to get the real engagement.

‘Everyone can produce service line reports and profit reports based on reference costs,’ she says. ‘But if you are going to use it as a performance tool, you have got to be able to explain what the variations are down to. You are not going to get this until you have invested in patient-level costing.’ Mrs Bewes says that service-line reporting ‘can only get you so far’. ‘But until you have patient-level costing, you are limiting the potential benefits.’

Transfer to PLC
Chelsea and Westminster moved to PLC in 2006, using newly compiled patient-level costs to inform its reference costs submission in June of that year. (The trust is hoping that in future it will simply be able to upload its patient costs to the Department of Health as its reference costs submission without significant amounts of additional reworking.)

Mrs Bewes says the trust chose an off-the-shelf PLC system – which effectively provides a data warehouse to bring together information from existing trust systems.  At the time there was limited choice of systems that could deliver PLC in the NHS, but a number of suppliers with international experience were starting to target the NHS. The trust chose one of these – Healthcost – although it has been joined in the market place by numerous other systems and services in the last 24 months.

Although Mrs Bewes accepts that the trust’s advanced electronic patient record was an advantage, she says that populating the system was relatively straightforward and quick. Virtually all data in the trust already existed at patient level and could be delivered by one of the trust’s eight clinical systems. Pathology, which was outsourced, was the only significant hole, and there were some more specific challenges. For instance, the trust’s on-take rota, where different clinical teams take turns to assess emergency admissions, provided complications in charging the correct responsible consultant’s time to each patient.

Even with these issues, Mrs Bewes estimates that the first cut of patient-level cost data was produced within six months. However far from being the end of the initiative, she says this is really where the project began. Producing the data is not the point. Using it is.

Data quality is of paramount importance to PLC. If it is to be used to inform changes in practice or pathways, or to challenge variations and current ways of working, the data needs to accurately reflect the real costs incurred in treating specific patients. And perhaps even more crucially, it needs to be accepted by the clinicians and others using the data.

Improving the quality of data – through better recording or better cost apportionment – is in fact a key aim of PLC. ‘This level of data has never been critically appraised,’ says Mrs Bewes. ‘So you have to build in the expectation that you are publishing it in order to bring out the data anomalies.’ She adds that it can take up to a year to get at some of the key data issues. But the end result is worth it. Just two years in, the trust can point at significant improvements made as a result of PLC, both in terms of improved data and better services.

Pilot operation
At the end of 2006, Chelsea and Westminster got the opportunity to work with foundation trust regulator Monitor and consultancy McKinsey as one of three organisations piloting service-line reporting (SLR). As part of the first phase of this work, a series of standard report templates were produced. These reports – a portfolio matrix, EBITDA (earnings before interest, taxation, depreciation and amortisation) table, detailed income and expenditure, variance analysis and cost matrix – have now become standard templates in Monitor’s SLR guidance. The trust also used this phase to introduce the concept to clinical leaders in its 30 service lines and demonstrate how the trust wanted them to use the data.

As an example, the portfolio matrix was used to highlight that the plastic surgery service – at the time making a loss compared to an organisational EBITDA margin target of 11% – needed to improve its cost position. Then looking at EBITDA by point of delivery revealed that the biggest losses were being made on day case and elective work. Further drilling down identified the four main healthcare resource groups responsible. One of these was for intermediate breast surgery (HRG J05), with an EBITDA of –£46,000 and margin of –56%.

The patient level data showed that length of stay varied significantly across this HRG and it could have been assumed it was simply length of stay that was driving the costs. But further analysis of the patient-level data revealed that costs varied significantly (from £400 to £2,800) even for spells with a one-day length of stay.

Time in theatre was identified as the main driver of costs. Comparing two similar patients for instance, there was more than a three fold difference in theatre costs. In fact a significant part of the problem was that theatre data was simply not being accurately recorded. This is now being addressed through further training on the PICIS theatre system. Again Mrs Bewes stresses that highlighting ways to improve the data is a good outcome of the approach. ‘This was something we’d never been aware of and never would have been aware of without the patient level data analysis,’ she says. And it provides the platform for identifying variations resulting from differences in practice.

In-depth drill down
Following this first phase, the trust embarked on its in-depth drill down, aiming to show that service line reporting supported by PLC could be used in annual planning.

Trauma and orthopaedics was selected for this first ‘deep dive’, with its 7% EBITDA margin below the target 11% for the organisation. Mrs Bewes admits another criteria was that clinicians were enthusiastic – a crucial advantage especially when trying to demonstrate the value of the approach to other services. ‘I’m happy to be told that the data isn’t right if it is accompanied by suggestions for improving it,’ she says. ‘But the worst thing that could happen is that the data gets ignored.’

The deep dive involved an intensive two-week initiative to analyse data, compare with benchmarks and set targets. The detailed data enabled the trauma team to identify four key areas for improvement: length of stay, theatre use, prosthesis costs and admission practices.

Targets were agreed (for instance to reduce average length of stay by 1.5 days and reduce prosthesis costs by 10%) and potential savings (between £570,000 and £730,000) were estimated. The intensive group work also ensured that key ideas for improvement from around the department were listened to and incorporated into action plans.

While elements of the programme are still ongoing, improvements have been impressive. A 10% increase in day of surgery admissions and a 26% cut in overall length of stay have helped the specialty far exceed its targets and move its EBITDA margin to several percentage points above the organisation-wide target.

Mrs Bewes says the exercise proved how using patient level information can help to understand exactly where the problems lie. ‘It helps us to focus our limited resources on the things making the biggest impact,’ she says. It also helps convince the wider clinical body of the value of the approach, with clinicians able to see a clear link between reduced costs and improved services and recognising how the identified savings could be used to deliver the improvements they wanted.

Mrs Bewes says the approach is starting to be embedded in the organisation’s way of working with a rolling programme of deep dives. The board and trust executive review the portfolio matrix quarterly to identify services for further investigation and then the services report back on key cost drivers and actions, using central support and facilitation.

Services receive regular updates on their service-line performance and some are starting to interrogate their performance before coming under central scrutiny. For instance, the service lead in ophthalmology started scrutinising the information provided before the launch of a further deep dive at the trust. With a -30% EBITDA margin (September 2006) and 80% of procedures losing money, it was already known to be a challenging area for the trust. Again early work has been promising. 

The analysis of patient-level data has revealed two key issues now being targeted – improving theatre lists for cataract surgery from an average of 3.8 to 5.5 (recognised as an achievable benchmark) and improving coding of medical outpatients. On the back of this, the service was targeting an improvement in EBITDA of 25 percentage points, to a much improved margin of -5%. All this was done after 48 hours of interrogating the data with the service lead. Actual achievement was -14% in September 07 and -3% in Dec 07.

Mrs Bewes is clear that service-line data, driven by patient-level costing, can deliver significant benefits. She says it is essential for driving transformational change, needed for improvements such as the 18-week wait. ‘Without service line reporting, the board is blind,’ she says. She is also clear about the main challenge. ‘It is not technically difficult, but it is culturally challenging,’ she says. Clinical engagement is the key and patient level costing is what makes the difference.

PLC technical briefing online
Chelsea and Westminster acknowledges that the trust’s advanced electronic patient record was an advantage in implementing patient-level costing (PLC), with virtually all the necessary data already existing at patient level. However there were still challenges. To discover more about the technical side of implementing PLC at the trust, visit www.hfma.org.uk  to read a specific briefing.


Comment on this article

Technical Q&A on implementation of PLC at Chelsea and Westminster