Results may vary

Geoff Bick
Director of Product Development
CHKS

Payment by Results (PbR) is changing the priorities of NHS acute care providers and some of its effects are giving health managers and policy makers unforeseen issues and new challenges.

The PbR policy is changing the way providers are reimbursed from a monolithic funding system to one adjusted on a case mix basis, using Healthcare Resource Groups (HRGs). In 2004/05 PbR optionally covered all the activity of foundation hospitals and in the 2005/06 financial year it will include the elective activity of other acute providers. The aim of the policy is to reimburse hospitals more sensitively for their clinical work and to incentivise them to address areas of adverse cost variance.

However there are major questions about some of the key building blocks of PbR – Finished Consultant Episodes (FCEs) and HRGs – and how different acute hospital practices for recording these can affect income via PbR.

Reporting difference

Firstly, there is a great deal of variation in how trusts report activity as FCEs or as regular attenders, out-patients or ward attenders, and also in how they reflect the interface between accident and emergency and in-patients through facilities such as Clinical Decision Units. For example, some hospitals report colonoscopy procedures as out-patients while others more commonly do so as day cases. Many of these questions were the subject of a Data Standards Research project commissioned by the NHS Information Authority in 2003 (Data Standards Research. A Report to the NHS Information Authority. October 2003) but the recording variations in hospital data persist.

Secondly, the portrayal of hospital activity through HRGs depends crucially on how that activity has been described using diagnosis and procedure coding.

Key issues include:

• The range of approaches to the clinical coding process, including centralised and decentralised models of deployment of coding staff, and widely differing levels of interaction between clinicians and coders.
  
• Different clinical coding methods used by different hospitals. Some coding is done using the source patient case note, while others use some intermediate record, usually with a consequent loss of detail and richness.
  
• The widespread variation in the number of clinical coders available. A survey by CHKS of 74 acute trusts in 2004 found that the average number of FCEs per WTE clinical coder ranged dramatically from about 4,000 to 14,000.

As a result the diagnostic coding depth with which acute hospitals describe their clinical activities varies enormously and the extent to which complications and co-morbidities are recorded directly impacts on their HRG profile.

To explore this, activity for this current financial year was analysed by CHKS for a representative sample of 98 English hospitals. The key findings were:

• For teaching hospitals, that could be expected to have a higher depicted case mix, the average number of diagnoses per coded FCE ranged from 1.7 to 3.3, with a mean of 2.5.
  
• For non-teaching hospitals, the average number of diagnoses per coded FCE ranged from 1.6 to 3.9, also with a mean of 2.5.
  
• In both categories, foundation hospitals were scattered throughout the distribution and had neither consistently high nor low coding depth.

Richer coding

Has depth of diagnostic coding changed over the past few years? Further CHKS analysis compared depth of coding in 2004/05 to date with 2002/03. For the sample considered, 11% of hospitals had a reduction in coding depth, in 16% it remained the same, while at 72% the average number of diagnoses per coded FCE increased. The highest increase was an average of 0.8 diagnoses per FCE. Ironically, the greatest observed fall, of 0.5 diagnoses per coded FCE, occurred at a foundation hospital.

In general, teaching hospitals and foundation hospitals were spread across the distribution with no clear pattern, but it is clear that a majority of hospitals are putting effort into richer diagnostic coding of their in-patient clinical activity.

These findings, which are replicated widely across many areas of clinical activity, suggest that the effects of the PbR policy are not straightforward and that poor clinical coding plays a major part in its success. Fundamental questions about the data driving the PbR policy need to be addressed and consistent solutions implemented.

The imperative to improve coding to provide an accurate picture of hospital activity through HRGs for funding should also allow hospitals to obtain a better picture of their clinical activity and outcomes. In reality, the historical improvement in coding depth may also have been previously spurred by the clinical governance policy which encouraged many hospitals to capture a richer profile of their clinical activity through their admitted patient datasets.

Nevertheless, the fact that a majority of hospitals are proactively putting effort into richer diagnostic coding of their in-patient clinical activity is to be welcomed, as is the higher prominence for coding staff that PbR has helped to bring about.

Geoff Bick
Director of Product Development
CHKS