John Bye Profile picture
12 Dec, 16 tweets, 5 min read
The full results of the Liverpool LFT trial have been released.

False positives are much lower than expected, but false negatives much worse.

Liverpool has now paused plans to use the tests on care home visitors, following Sheffield and Manchester.

gov.uk/government/pub…
The good news is the LFTs only gave 2 false positives out of 3,199 tests checked by PCR. That gives a false positive rate of 0.07%, far less than the 0.38% reported in initial tests.

The bad news is they missed 23 of the 45 PCR positives, giving a 51% false negative rate!
We already knew the Innova LFTs were less sensitive than PCR, but they were meant to catch 95% of people with "higher viral load".

In the Liverpool trial though they missed 15% of people who were positive at below 20 cycles on PCR, and 47% of people positive at 20-25 cycles!
Overall the trial missed 31% of the people most likely to be infectious. Almost one in three.

It's no surprise councils were angry when the government sent these tests to care homes without telling them, and before the trial results were published.

news.sky.com/story/covid-19…
Before we write LFTs off completely though, results from Birmingham Uni (which has been screening students before they go home for Christmas using the Innova tests, and cross checking some results with PCR) look more promising.

More on that tomorrow...

Meanwhile the government is pushing ahead with rolling out 1.6 million LFTs in tier 3 areas this month alone, despite concerns over their effectiveness in asymptomatic testing and the total lack of evidence that they helped reduce cases in Liverpool.

Picking up from yesterday, how did the Innova lateral flow tests perform at the University of Birmingham?

First the bad news: of 720 random negative tests they checked with PCR, 6 were positive.

Extrapolating to all 7,189 people tested, that suggests about 60 false negatives.
Given they only got 2 positives from the LFTs, that means they might have missed 97% of people who had the virus!

But this doesn't tell the whole story. Unlike in Liverpool, ALL of the false negatives were found by PCR at higher Cts (29+).
So at Birmingham University the Innova lateral flow tests seem to have caught the only two asymptomatic people they tested who were confirmed by PCR to have what the initial assessment classed as "higher viral load".

Which is what they're meant to do.
But this doesn't guarantee people who got false negatives weren't infectious when tested or didn't go on to develop symptoms.

And it's debatable whether we can read anything into such a small sample size - in Liverpool 40% of the false negatives were detected at Ct 25 or lower.
On the bright side, after testing 7,189 people the only 2 positives were both confirmed by PCR at low Cts (20 and 25).

So there were NO false positives, again suggesting the false positive rate for the Innova LFTs is MUCH lower than the 0.38% quoted in the initial assessment.
So where does this leave us?

The tests are clearly far from perfect.

But they're not completely useless.

More than 99% of people got the right result, and in Liverpool they found hundreds of asymptomatic people with high viral load who might not have been detected otherwise.
BUT they may be missing up to a third of people who are currently asymptomatic but have high viral load, and may therefore be infectious.

So using the tests to screen people before visiting care homes (as the government wants to do) seems VERY risky.

manchestereveningnews.co.uk/news/greater-m…
There may be ways they can be used effectively though.

Professor McNally (who ran Birmingham Uni's screening programme and helped setup the first Lighthouse Lab) suggests using them as a quick, cheap way of screening symptomatic people, for example.

Even mass screening could be useful IF there's proper support for people who test positive to self-isolate (which isn't currently the case) AND it's made clear that a negative result isn't a license to go wild (ditto).

Catching half the asymptomatic cases could still be useful.
But all of this needs a coherent plan, careful messaging, a functioning contact tracing system, and adequate support for people self-isolating.

Not a hodge podge of different approaches in different towns, and a service that still fails to reach 40% of non-household contacts.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with John Bye

John Bye Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @_johnbye

10 Dec
You'd think the Brexit Party would have other things on their mind at the moment, but instead they seem to have latched onto a misleading story about false positives.

Unsurprisingly it doesn't show what they think it shows.
For starters, Cambridge are using pooled testing, which none of the regular test system is.

This means mixing several samples together to reduce the number of tests you need to run, then if a pool is positive you individually test everyone in that pool to see who had the virus.
But a recent study suggests pooling causes false positives.

They tested 17,945 pools. 2,084 of them were positive. When they tested the people in those pools individually, they didn't get any positives from 92.

92 ÷ 17,945 = 0.5% false positive rate.

medrxiv.org/content/10.110…
Read 9 tweets
9 Dec
Covid sceptics like to dismiss most people dying from the virus as so old they were about to die anyway and somehow don't count.

Ambient music pioneer Harold Budd was 84 when he died yesterday.

His latest album was released last Friday.

He also composed a soundtrack this year.
We'll sadly never know how much more music he might have written had covid-19 not come along.

But people reaching that age have in many ways already beaten the odds. Life expectancy for an 85 year old male is about another 6 years, in both the UK and US.

ons.gov.uk/peoplepopulati… Image
Behind the statistics, every death from covid-19 is tragic.

Just because someone was old or had pre-existing conditions making them more vulnerable to covid doesn't mean they couldn't have led a full, happy and productive life for many more years if the virus hadn't intervened.
Read 4 tweets
4 Dec
This week's Test & Trace report shows cases falling sharply in the week to November 25th, thanks to lockdown 2.0.

There's also a sudden big improvement in contact tracing performance (which turns out to be smoke and mirrors), and some odd revisions to old pillar 1 testing data.
Cases in England were falling sharply towards the end of November, with both the number and percentage of people testing positive falling by about a quarter compared to the previous week.

There's a small drop in the number of tests done, but it may well be due to reduced demand.
In case there's any doubt about the effectiveness of lockdowns, regional data shows clearly that all areas where cases were level or still rising when lockdown 2.0 began started falling in perfect unison about a week later.

The question now is whether that can be sustained.
Read 15 tweets
2 Dec
This week's ONS weekly deaths report again shows deaths well above normal in the week to November 20th.

All causes deaths were 21% above the five year average, meaning 2,169 extra deaths.

And still well outside the normal range (showing by the dotted lines on the graph below). Image
Excess deaths are also still tracking the same curve as all measures of covid-19 deaths. Image
And as usual, most death certificates mentioning covid-19 listed it as the underlying cause of death (ie, they died "from" it not "with" it).

87.5% in the latest week. Image
Read 5 tweets
1 Dec
The government has now released some info on how LFTs compared to PCR in field tests in Liverpool. h/t @ScienceShared

It's not great.

LFTs found about half the people that tested positive with PCR. Which is a little better than my back of the envelope estimate.

But... Image
It only found "more than" 70% of people "with higher viral loads, who are likely to be the most infectious".

Far less than the 95% it managed in the original Porton Down trial, which was in a lab using spiked samples, not members of the public shoving swabs up their noses.
So although the LFTs have found a few hundred cases that would have been missed otherwise, they still missed half the people with the virus that were tested.

Including 20-30% of the ones most likely to be infectious.

Source: Community testing prospectus
assets.publishing.service.gov.uk/government/upl…
Read 8 tweets
24 Nov
The latest ONS report shows another rise in excess deaths in the week to November 13th.

All causes deaths were 18% above the five year average, meaning 1,917 extra deaths.

Still far above the highest number of deaths we've seen in any of the last five years.
And excess deaths are still following almost exactly the same curve as all measures of covid-19 deaths (due to, involving, and within 28 days of a test).

This is not a coincidence.
As usual, the vast majority of death certificates in that week which mentioned covid-19 had it listed as the underlying cause of death - 88%.

In other words, people are mostly dying "from" covid-19, not "with" it.
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!