Why school testing data doesn’t tell the whole story

“Something like 80% of children are not doing these tests.”

On the Today programme last week, shadow schools minister Kate Green claimed that 80% of schoolchildren are not doing coronavirus tests. A similar claim was made in a letter sent to the education secretary.

Since early March, pupils in secondary schools and further education colleges have been asked to take two lateral flow device (LFD) tests per week to stop the spread of Covid in schools.

However, the data Ms Green refers to just looks at the number of tests recorded, not the number of pupils tested so can’t be used to make this claim.

The data

Labour confirmed its figures came from the government’s testing data, which includes statistics on the number of tests reported in schools in the week to 16 June.

In the week to 16 June, about 713,000 tests were reported among secondary school students in schools, excluding colleges and 16-19 schools, in England. Primary school students are not routinely tested.

There are about 3.5 million pupils in English state secondary schools, which might be how Labour calculated a fifth of school pupils had taken a test (713,000 being roughly 20% of 3.5 million).

But using this data to make claims about the number of pupils tested is flawed for two reasons.

Firstly, the data counts the number of tests reported, not the number of pupils tested. Given pupils are asked to be tested twice a week, you can’t really use the figures available to talk about the proportion of pupils being tested (though if we did have those figures, they would be lower and seem to support Labour’s point even more). 

More importantly the statistics only reflect the number of test results reported, not the number of tests done, which may be much higher.

Tests reported vs tests conducted

The National Audit Office reports that as of 26 May, 241 million LFD tests had been distributed to education settings in England, excluding higher education settings.

However, only 51 million had been registered as used. 

The government’s testing statistics, which Labour used, similarly show around 48 million registered tests up to that point, an additional two million “unregistered” but, evidently, used and reported tests (as in the vast majority of cases we know if it returned a positive or negative result) and half a million tests of unknown registration status. All this adds up to 51 million.

So it’s clear that the testing statistics do not account for the vast majority of tests sent out, and it’s unclear whether these tests were used or not.

The Department for Education (DfE) says all results must be reported to NHS Test and Trace, and schools, but there’s no guarantee that is happening, even if the tests are being taken.

In fact, the Department for Health and Social Care (DHSC) told Full Fact it knows many negative test results are not being reported.

It confirmed that the testing statistics for schools did include instances where a parent tested their child at home and registered the result, as well as tests taken in school. But DHSC can’t record the results of tests that parents don’t report.

So it's impossible to say what proportion of students are being tested each week or whether recent falls in the number of tests taken are due to a fall in testing or a fall in reporting. 

It should be said that the statistics themselves do not make any of this particularly clear. 

They are described as representing the number of tests “taken” not “reported”.

They also include a category of “tests not registered” which appears as if it could represent all tests taken but not reported, but does not do so (though the data notes do say these are almost all related to on-site testing and so wouldn’t account for at-home unreported tests). 

And they also say they count the number of tests taken “in schools” but, as DHSC told us, also count tests of pupils taken at home. So it’s easy to see how they have been misinterpreted.