"Fraud is costing the NHS £5bn a year'"
Tonight, BBC Panorama will discuss the 'Great NHS Robbery': an investigation of fraud and error across the NHS. Going by the broadcaster's coverage this morning, about £7 billion is lost due to fraud and error each year - £5 billion of which is lost to fraud alone (these figures apply to England rather than the whole of the UK).
But the NHS disputes the figures, telling Full Fact: "We do not recognise the figure of £5 billion losses. We are not aware of any published research showing how this figure has been calculated and we do not speculate on levels of losses."
In fact, until 5pm today there was no good reason for anyone to recognise the figure, because the research wasn't published.
Now we do have the figures, the report itself doesn't mention a £5 billion or £7 billion cost. Instead, it's an analysis of estimated fraud and error levels across health services internationally, the findings of which are applied to the UK. There's huge uncertainty around the figure because of this.
Measuring fraud is prone to error
If all fraud was detected as it was happening, there probably wouldn't be any money lost to it. As it happens, even detecting money lost to fraud after the event is difficult, not least because it's difficult to prove when someone has deliberately tried to make an unlawful gain.
The Department for Work and Pensions, for instance, regularly tries to estimate fraud (as well as simply claimants' or officials' error) in the benefits system. Their annual estimate of around £1.2 billion lost to fraud has a margin for error of several hundred million either way, so the estimates are very uncertain.
The NHS's official figures are limited
The Department of Health told Full Fact they "do not speculate on levels of losses". Instead, the only figures available from them cover two specific areas in which fraud losses have been identified in the past: patient charges and dental charge fraud. Together, the identified losses from these are estimated at £229 million.
But this misses out any fraud that hasn't been 'detected' and areas such as staff fraud and false invoicing aren't included. So while the NHS doesn't recognise any other estimates, it's reasonable to assume the NHS's actual losses to fraud and error are higher than their estimates can show.
£5 billion: international averages, applied to UK
The research behind today's reports - co-authored by the University of Portsmouth and accountancy firm BDO - collects estimates on the cost of fraud and error in healthcare from six countries (including the UK and the US) with data spanning back as far as 1997.
Their finding was that, on average, these healthcare systems lost 6.99% of their spending through fraud and error, though this varied quite considerably, from 0.6% in one country to 15.4% in another. Each national figure was also subject to uncertainty as well.
There is no mention of £5 billion lost just to fraud or £7 billion including error. Instead, one of the authors of the report told us this was an extrapolation based on assuming the NHS losses would equal the global average - in other words, that 6.99% of the NHS's £100 billion a year budget is lost to fraud and error.
This is obviously a problem, and the report doesn't give any reason to suggest the NHS would be a typical example compared to its foreign counterparts.
We spoke to one of the authors about this, who suggested that patient levels of fraud are likely to be lower - though still not negligible - in the UK given that most services are free at the point of delivery.
That said, other areas such as false accounting weren't as likely to be significantly different, and in some areas the NHS might have higher levels of fraud, for instance using NHS resource for private healthcare use.
So there's a lot more uncertainly around the figures reported this morning than appears at first glance: not only is is based on assuming NHS fraud is in line with the global average, it's also the culmination of several uncertain estimates gathered internationally.
Update (7 April 2014)
It's now been made clear to us by the researchers that the figures apply just to England rather than the whole UK, and this is now mentioned in the piece.