The Audit Commission has raised concerns about the state of patient records after routine checks found a large percentage contained unreliable, or no, information.

The commission’s checks were performed as part of its annual audit of the quality of the coding that lies beneath the bills that trusts send commissioners. The checks showed that on average 8 per cent of the 2008-09 bills audited were for the wrong treatment.

The level of UTAs [unsafe to audit] demonstrates weaknesses in the standard of documentation, which may present wider clinical and patient safety risks

But in half of trusts some bills could not be checked at all because when auditors tried to validate them against patient records the information was so incomplete or illegible it was deemed “unsafe to audit”.

In a report on its annual audit, published today, the commission says: “The level of UTAs [unsafe to audit] demonstrates weaknesses in the standard of documentation, which may present wider clinical and patient safety risks.”

In the worst trust, 16 per cent of records were unreliable.

The audits are based on a targeted sample of bills, chosen to reflect local and national problem areas. It is therefore not possible to tell whether the 8 per cent error rate – equivalent to 4 per cent of the total monetary value of the bills - would be replicated across the total estimated £44bn of activity which is accounted for under payment by results codes.

In just under 90 per cent of errors, poor patient records were a cause of the mis-coding. In 60 per cent, training was found to be insufficient.

As in last year’s audit, the commission found no evidence of systematic “up-coding”. Coding errors led to equivalent levels of under and over charging so that the net effect was close to zero.

The highest level of mistakes was at Blackpool, Fylde and Wyre Hospitals foundation trust where 40 per cent of bills were wrong, leading to commissioners being over charged by 24 per cent.

As with other trusts, the rate of errors in the underlying codes which record each procedure and diagnosis relevant to the patient was much higher. Thirty-six per cent of primary diagnosis codes and 28 per cent of primary procedures were incorrect, for example. This compared with national averages of between 6 and 15 per cent of diagnosis codes and 7 and 18 per cent for diagnosis codes.

Patient records: jokes about doctors’ bad handwriting are no longer funny