Intended for healthcare professionals


Coding errors in NHS cause up to £1bn worth of inaccurate payments

BMJ 2010; 341 doi: (Published 31 August 2010) Cite this as: BMJ 2010;341:c4734
  1. Adrian O’Dowd
  1. 1Margate

    Errors in clinical coding of patients’ treatments in NHS hospitals in England have meant that primary care trusts have made about £1bn worth of mistaken payments over the past three years, sometimes paying too little and sometimes too much.

    The mistakes were spotted by the spending watchdog the Audit Commission, which published a new report on the quality of use of data in the NHS on 26 August. The commission had audited the data that underpin England’s Payment by Results system, which generates a payment per patient, depending on the treatment given, and which hospitals use to charge primary care trusts.

    The commission found that accuracy of payments was improving overall in the NHS but that wide variation between the best and worst trusts persisted.

    Its report shows that the average error rate in clinical coding fell from 16% to 11% in the three years since 2007, when the commission began to audit random samples of data from four specific specialties at all trusts that have them: general medicine, trauma and orthopaedics, cardiology, and paediatrics.

    The commission estimated that of the £21bn paid over the three years for treatments in these specialties £1bn (about 5%) was incorrectly paid because of wrong data. The commission also looked at over £200m worth of payments for inpatient treatments and found that some trusts were much better than others at clinical coding. From this sample some £9m (about 5%) worth of financial errors was spotted.

    The investigation found no evidence of what the commission calls systematic “gaming” in the system: NHS trusts purposely using wrong codes to get extra money for treatments that were not provided or charging for a more expensive treatment than was provided.

    The audited sample showed that, overall, hospital trusts had slightly undercharged primary care trusts for their work in 2009-10.

    The commission said that clinical coding staff were now better trained but that greater engagement of clinicians would further improve data accuracy. It also recommended that trusts carry out regular internal audits on clinical coding and ensure that policies and procedures were up to date.

    Poor medical records were one source of error, said the commission, which found that in some cases the coding case notes had no information relating to a patient’s treatment.

    The report says: “Poor quality medical records such as these represent a clinical as well as a financial risk.”

    Andy McKeon, the commission’s managing director for health, said, “It’s reassuring that the NHS is getting better at clinical coding. Efficiency is more important than ever, and accurate clinical coding under Payment by Results will contribute to better data, better decisions, and better outcomes for patients.

    “But the variation in error rates from trust to trust is a concern. There are clear ways in which trusts can improve.”

    David Stout, director of the primary care trust network of the NHS Confederation, the body that represents most NHS organisations, said: “Some places are much more effective at it [coding] than others, and over time it has improved, but it has not got better as fast as we would have hoped.”


    Cite this as: BMJ 2010;341:c4734