Administrative staff record and 'code' all clinical activity undertaken within each hospital. The resulting data is used to determine how much hospitals are paid under the Payment by Results (PbR) system, and to guide those who commission services.
But an investigation by the Audit Commission found that trusts were miscoding an average of 16.5 per cent of all procedures. The worst performer got it wrong in more than 70 per cent of cases.
These errors meant that PbR payments across the sample reviewed were out by a gross figure of £3.5m - around 5 per cent of the total value of the sample.
However, in the vast majority of trusts, overpayments were roughly cancelled out by underpayments, meaning the net financial impact was close to zero.
Andy McKeon, the commission's managing director for health, said this showed that fears hospitals were 'gaming' the system, by deliberately miscoding procedures to increase their payments, were unfounded.
'It's cock-up rather than conspiracy,' he said. 'That should be reassuring for taxpayers and for GPs.'
Mr McKeon called for better training for staff involved in coding clinical work.
He also called for stronger clinical involvement, noting that at West Middlesex NHS Trust, where coders and clinicians meet regularly, coding is 100 per cent accurate.
The Audit Commission analysed more than 50,000 care episodes undertaken in 2007-2008.
Comment below and tell us what you think