Thread by @chrischirp

THREAD on Lateral Flow Device tests (LFDs):

They are being used more and more in England.

Thread about how they're used, why you NEED confirmatory PCR testing, how it will make interpretating case figures difficult as we open up.

#NerdyThread 1/19

LFDs are used to test people who don't have Covid symptoms - they give results in about 30 minutes.

The govt reckons about 30% of people with Covid don't have symptoms & this article suggested that asymptomic spread might account for 50% of cases.
jamanetwork.com/journals/jaman… 2/19

So the point is to find people who have Covid but don't know they have Covid & stop them mixing with, and potentially infecting, other people. 3/19

Say you test 1,000,000 people with no symptoms with LFDs (England did about that per day last week). ONS estimates 0.4% of people currently have Covid - let's say 30% with no symptoms (0.12%).

So in a million people with no symptoms, you'd expect about 1200 to have Covid. 4/19

Assume LFDs catch about 50% of people with covid (might be higher but home tests often done less well). So 600 will test +ve.

LFDs return a +ve result for a person who DOSEN'T have Covid about 1 in 1000 times. Because there are LOT of -ve people, still get 1K false +ve's 5/19

At this stage, out of 1 million asymptomatic people tested, the tests have returned 1600 positives, only 600 of whom actually have Covid.

So if YOU are one of those million who gets a positive test, you only have a 38% chance of actually having Covid. 6/19

This is why you want to do a CONFIRMATORY PCR test on the LFD positives - because PCR tests will return a +ve result for someone WITHOUT Covid less than 5 in 10,000 times (and probably much less than that - ons.gov.uk/peoplepopulati… )

PCR picks up about 70% of actual +ves. 7/19

So if you then test all the LFD positive cases, about 420 of the 600 actually positive people (70%) will get a +ve PCR results and somewhere between 0 and 1 person of the 1000 false +ves will get +ve PCR.

Almost all of the false +ves will be freed to continue mixing. 8/19

So, after confirmatory PCR testing, you are left with 421 people who tests positive with *both* LFD *and* PCR tests.

All but 1 of these actually has Covid - so now if YOU have tested +ve twice you have a 99.8% chance of actually having Covid vs only 38% after 1 LFD test. 9/19

You have however only identified 420 of the 1200 people who do have covid out of those million asymptomatic people.

But, if they now isolate where they wouldn't have otherwise, that's still a win & you've not made lots of people who DON'T have Covid isolate. 10/19

Knowing that you are very certain to have Covid after 2 tests might also increase adherence to isolation.

And repeat LFD testing every few days should hopefully catch some of the 780 who tested -ve previously. 11/19

Note also that that your chance of actually having Covid after testing negative is very small (0.08%). But this is only a bit smaller than the chance of a random person with no symptoms having Covid (0.12%). So a negative LFD tests doesn't mean you're magically 100% safe. 12/19

This shows why confirmatory PCR testing is important. Wales, NI & Scotland all require them after +ve LFD tests. England only requires them for LFD tests done AT HOME or for health care workers. This leaves millions of workers at risk of needing to isolate unnecessarily 13/19

LFD tests have become the majority of tests done every day, The big spike from 8 March was schools. As retail & hospitality open up next month, this will increase again.

Not confirming worker +ve LFDs will lead to many false +ves (school home tests get confirmation) 14/19

Note that the first THREE LFD tests done at school this last week and early next week do NOT get confirmatory PCR (all such a mess). The 600k / day extra LFD could easily account for the increase in cases numbers we have seen this week. 15/19

Problems over the next few weeks/months:

- isolation is hard and if you might well be a false +ve (for non confirmed LFDs) the incentive to isolate is less. Will have an inequality gradient as low income families less able to isolate. 16/19

This creates a disincentive to get tested (where voluntary as in schools) or to the test properly (it's hard to get swabs down yr throat or up yr nose!).

This makes testing strategy much less effective - particularly in more deprived areas. Need supported isolation!! 17/19

Other home nations require confirmatory PCR tests while England only half does. This means case numbers increasingly less comparable betwen nations. Positivity rates will also be affected and harder to interpret. 18/19

This all means we will come to rely even more on random population infection surveys like ONS or REACT study, but these are more lagged and have some of their own problems.

The next few weeks/months are going to be hard to interpret unless trends are *very* obvious. 19/19

PS you can argue about covid prevalence, test sensitivity and specifivity estimates used, but this doesn't change the basic argument at all. The estimates are all plausible and my aim was to illustrate how an extra PCR tests helps.

PPS Even a second *LFD* test for first +ve LFD test would help enormously in cutting down overall false positives (assuming subsequent test results are independent of the first test).

PPPS so yes, I suspect the increase in cases this week is a school return effect - we'll know better next week as the LFD/PCR test result breakdown comes through on the dashboard (but unfortunately not broken down by age).

PPPPS putting your whole Covid safety strategy on TESTING *without* improving financial and practical support for isolation is stupid and shortsighted.