Some statistics can become so widely known and accepted by experts in the field that they become almost too obvious to provide a citation for.
And so, in a puff of smoke, vanishes the information we need to question their accuracy and relevance to the issues of the day.
If you've been paying attention to the rumbling debate over how well-resourced our GP practices are, you may have heard the following:
"GPs handle 90% of patient contact with the NHS."
In this case we're quoting NHS England chief Simon Stevens, who said this on the Andrew Marr Show yesterday. But we could also have quoted, in no particular order, the Royal College of GPs (RCGP), the British Medical Association (BMA), the Health and Social Care Information Centre (HSCIC), and others.
But it's not clear how the figure was calculated - and it seems to exist in two slightly different forms.
RCGP has pointed us to an NHS England report as the source of its figure (it's also the source given by the BMA). NHS England, in turn, seems to cite a 2009 report carried out for the NHS Information Centre as saying that there are 300 million appointments at GP surgeries, which make up 90% of NHS contacts.
Think tank the King's Fund also cited the same data in 2009.
But while the report includes the 300 million figure, it doesn't seem to contain any reference to the 90% one. That means we don't know what figure is being used for 'NHS contacts' as a whole.
90% of contacts — with GP surgeries or with the broader primary care sector?
We've also been in touch with the Information Centre's successor, the Health and Social Care Information Centre (HSCIC). On its landing page for primary care, it says "around 90 per cent of patient interaction is with primary care services", going on to explain that this includes dental practices, community pharmacies and high street optometrists.
That would imply GP surgeries alone handle less than "around 90 per cent", although there's enough vagueness in the phrase to leave the possibility that this figure and the one about GP practices are one and the same.
They think the figure might have originated with the Department of Health, who we've contacted.
Why it's important we find out
To get to the 90% figure (using the 'back of fag packet' method on those NHS Information Centre figures) we're talking about appointments at GP surgeries making up 300 million out of about 330 million 'NHS contacts' in 2008.
That seems plausible, and we've seen nothing so far that makes us doubtful of the accuracy of the figure.
But we do have questions about it that can't be answered with the evidence easily available at the moment. Are we definitely talking about 2008, or are there more recent figures? The number of appointments at GP surgeries has been estimated by Deloitte to have risen from 300 million to 370 million in the six years to 2014/15, but we don't know whether the 90% calculation has been repeated again for these more recent figures.
If not, might the way people use the NHS (and the 90% figure) have changed since 2008?
Are we talking about GP appointments only, or trips to the dentist too?
And what counts as an NHS contact? Would we include a five minute phone call to NHS 111 as a contact? Emergency heart surgery? Planned heart surgery?
We might have suspicions about the answers to these questions, but we have no way to be sure. And in an ideal world it'd be a little easier and less time consuming to find out. Please get in touch if you have more information on where the source of this one might be hiding.
Isn't it nice to have the whole picture?
We rely on your donations to continue and grow our factchecking efforts - to help us maintain our independence we need 1000 donors to give £10 a month. We are currently at 607 - please help Full Fact grow.