One in ten hospital episodes miscoded

Tags: Audit Commission Data quality Quality

02 Sep 2008

Almost one in ten hospital episodes is incorrectly coded, according to an analysis carried out by the Audit Commission.

The spending watchdog’s analysis of more than 50,000 episodes of care, equating to approximately £73m of expenditure under Payment by Results, found an average error rate of 9.4% - with error rates across trusts ranging from 1% to 52%.

The Commission warned that the errors would not only affect Payment by Results but other issues, including activity planning and commissioning.

It recommended that the Department of Health and NHS Connecting for Health should introduce a wider data quality programme to drive improvements in the standard of NHS data and increase confidence in its use.

The Commission said the coding mistakes identified contributed to a gross financial error of about £3.5m, but that in most cases the net financial impact was close to zero.

It found no national evidence of under or over-charging or gaming. However its report added: “There is a number of cases where the net financial impact of errors was locally significant.”

The report says the most common issue affecting the accuracy of clinical coding is the quality of the source documentation. Other problems include the adequacy of trust coding arrangements and the level of clinician involvement.

Audits showed that the training and development of coders had more of an impact on limiting the number of errors than the number of coding staff. Foundation trusts had marginally lower Healthcare Resource Group (RG) errors than non-foundation trusts, but higher error rates were encountered at specialist trusts.

The Audit Commission has also recommended that trusts should review and, where necessary, improve their source documentation to promote accurate coding and ensure that clinicians are engaged and involved in validating coding.

It further said that trusts should develop their coding departments through professional training and development.

The data analysis was undertaken at all trusts in England throughout 2007/8 and the Commission said the results provide the first comprehensive national picture of the quality of data underpinning not only financial but clinical and commissioning information.

The Audit Commission has developed its benchmarking methodology into an online tool called the National Benchmarker. It has recommended that trusts and primary care trusts use the Benchmarker regularly to review potential coding anomalies and areas for improvement.

Link

Audit Commission report on PbR clinical coding

Fiona Barr

Reader's Comments

What does this really tell us?

02 Sep 08 15:09

This is an interesting report but at times it is hard to understand really what it is telling us. A reversal of the headline from "1 in 10 miscoded" to "Over 90% coding accuracy" could easily put a positive spin on the reports findings.

Within the report itself there is reference to the focus on benchmarking data to identify areas to audit. Presumably this means that the audits were targeted on areas of concern. So if overall 90% of episodes that were of concern are accurate that's even more positive.

And the use of terms such as "coding errors" and "mis-coded" are disingenuous if, as is stated, one of the main causes is poor documentation. The coding may be accurate but the episode of care has been misrepresented.

With coding and auditing still, unfortunately, being somewhat subjective then the variations seen between Trusts may not be entirely true.

It is good to see accuracy of coding and it's impact being recognised but I'm not sure the report itself or the headlines in EHI and HSJ are overly helpful.

The bottom line - improve support and training for coders, and improve clinical documentation. No new issues, just still unresolved ones.

Codes Not Available

elaine.wakeham@aaw.nhs.uk

03 Sep 08 09:09

The report highlights that most errors appears in specialist units, I wonder if this is because the actual code is not available. In doing such specialist treatments it could well be that the procedure has no specific code and coders are using next best.

Coders on Rounds

10 Sep 08 19:09

Until we have coders on rounds and doctors and coders learn to talk with each other, and Docs learn coding skills, coding will be close to fairy stories. In USA coding is learned by Med Students, because that is how for profit organisations and not for profit organisations earn income. Now we do the same with PBR - so we need to train clinicians to code (a good thing - coding is a scientific way to describe patients).

90% accurate is unbelievably good! Did anyone ask the Consultants if the codes bore any resemblance to what they saw as the priority issues in the admissions? Terms like "Primary" "Secondary" diagnosis, etc are foreign to most of us clinicians