Thread by @deeksj

What is the evidence supporting the Government’s claims that Innova LFT detects infectious cases?

There are no real studies at all which directly show how well Innova +ve detects infectious people and Innova -ve indicates non-infectious people.

No direct data at all.

1/19

The data come from studies of PCR with inferences made by linking PCR proxy “viral load” Ct values to the viral load levels we know that Innova detects.

Ct values are not standardised and depend on how much biological matter on a swab. They have high measurement error

2/19

Evidence on infectiousness comes from studies linking Ct values with rates of (a) secondary cases and (b) the ability to culture the virus. Neither are perfect – but are the best we can do. They will both most likely underestimate infectiousness.

3/19

Innova has been shown to be able to detect viral loads when Ct <25, both in Porton Down studies (left) and in real application in the Liverpool pilot (right).

4/19

And SAGE has advised the Government that Ct<25 indicates those people most likely to be infectious.

5/19

So what is the evidence from these studies that people with high viral loads detectable by Innova (Ct=<25) are infectious and that those with lower viral loads (Ct>25) are not? First two studies are of viral culture rates, second two studies are of secondary attack rates.

6/19

#1 PHE Colindale n=324 - symptomatic and asymptomatic. Clear trend that higher Ct values (lower viral load) less likely to have culturable virus. Culture rate Ct<25 is 85% but Ct>25 is still 34% - just less that half so lots of “infectious” cases would be missed by Innova.

7/19

#2 is regional lab in Marseille n=3790, probably largely symptomatic but not clearly stated.

The viral culture rate below Ct=25 is 85%, above it is 24% - just less that a third the value. Again a substantial proportion that would be missed by Innova could be cultured.

8/19

#3 secondary cases in contacts tested at t=0 & 14 days with PCR. Conversion from viral load back to Ct value is done from linked publication.

Secondary case rate 20% for Ct<25 & 13% for Ct>25. Case rates in those likely to be missed by Innova more than half of those found

9/19

#4 secondary case in UK Test-and-Trace system. No actual numbers are in the report - data from graphs.

Secondary case rate 7.4% for Ct<25 and 3.4% for Ct>25 roughly half the value. Again secondary cases in those Ct likely to be missed by Innova not negligble.

So clear evidence that

infectiousness has a continuous trend, no step change
anywhere so higher Ct values linked with high chances

BUT

viral culture rates and secondary attack rates in those likely to be missed by Innova are concerningly high and
definitely matter.

11/19

Governments are using models rather than real world evaluations to predict how well LFT testing will work. Some of these include “infectiousness functions” to show the chances that somebody with a Ct value is infectious. How well do these match with these real data?

12/19

Red shows the household infectiousness function from study #4 Test-Trace data household contacts

This function is a reasonable first approximation …

13/19

But this one (Blue) from Mina’s key study excludes the possibility that anybody could be infectious at Ct values above 22, nowhere near the patterns in any of these studies. Clearly will lead to overoptimistic predictions of how well LFTs would work.

14/19

Data from these 4 studies show high risk that Innova -ves can infect others.

This is why many scientists are expressing reservations about their use – particularly where they are to be used as a green light test - when a negative test is used to start activities.

15/19

If people stop socially distancing with a negative test (e.g. nightclubs, theatres, workplaces) then mixing with some infectious people will occur, increasing risk of transmission.

16/19

But remember, nobody has provided any data actually using the LFTs – most are from model predictions, some of which might be very wrong.

Good studies and real world evaluations are needed.

17/19

If you want to hear this presentation in 9 minutes you can access it here

https://twitter.com/deeksj/status/1359981333842395141

18/19

Please don’t bother posting abuse stating I don’t understand statistics, biology, virology, public health, have an agenda, have shares in a PCR company or am intentionally misleading the world. It harms your reputation too. This is about facts and data, not what I think.
19/19

@threadreaderapp unroll