In the article, Jon Deeks, derides the tests saying the tests:
"missed 60% of the cases which would have been picked up by PCR" - Jon Deeks
YES! This is a good thing!
MOST time someone is PCR+, they are No Longer Infectious! Rapid tests are catching the infectious only!
2/x
This issue is mentioned above continues to plague people's understanding of rapid tests
The tests are different
Rapid Antigen Tests are BETTER than PCR at being specific for infectious virus
A public health test should not pick up people after they are infectious
3/x
Importantly, we've found the rapid test used in the UK, the Innova test, is doing a great job w sensitivity & Specificity
In frequent testing, It often **catches new PCR positives BEFORE the PCR gets the results back**
And in >2000 tests, no false positives yet!
4/x
Our findings of VERY high sensitivity in frequent testing match the expectation that Rapid Ag tests catch people when they are infectious but not after they recover from being infectious
Similar findings were made in Liverpool too!
5/x
In Liverpool, if you look at the relevant viral loads, (i.e. greater than [100K - 1M], the Innova test performed very well, with >80% overall and >90% for the higher viral loads.
All in all, the UK's decision to use rapid tests appears to be proving extremely beneficial.
6/6
Also, it's not just culture data I using to state issues around transmissibility. That would be unwise
Its integration of epidemiological transmission data, PCR RNA data, animal experiments and culture pos vs. time & RNA
Taken together, the data all points in same direction.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
The recent @bmj_latest articles by @deeksj et al deriding Rapid Ag Innova Tests are simply WRONG
They simply do NOT appropriately interpret Ct values & do NOT consider massive importance of how long PCR remains + post-infectiousness
Inspection of Ct values among the Asymptomatics & correlation to RNA copies / ml shows Ct values in Liverpool are ~8 lower than often seen in literature. The failure to recognize this means the estimates of Ag test sensitivity for "high virus" are totally off.
2/x
The sensitivity for "moderately high" or "high" viral loads in the Liverpool data are ~90% and ~100%.
But to know this you cannot just assume a Ct of 25 elsewhere (often described as entering "high viral load") means same thing as a Ct value in other labs.
3/x
Unlike antibiotics where resistance happens w partial doses, to be a risk you also must be taking them in first place.
When considering escape from spike protein derived immunity - must consider everyone w/out sterilizing immunity at risk to induce a mutant upon infection.
2/x
Whether no vaccine or a single dose (or two) people create antibodies against the same part of the protein.
If discussing “partial immunity” or low affinity antibodies, must consider that a fully naive person might pose greater risk for escape than a single dose person
3/x
ALL evidence - when evaluated appropriately! - shows these are VERY good at detecting infectious virus
• ~100% if used frequently
• >95% for single samples with high, most likely contagious viral loads
The tests work
We've been evaluating rapid Ag tests on campuses. We find these tests - when used as screening w/out symptoms DO miss most PCR positives!
BUT EXPECTED! - ALL misses were previously detected and already finished isolation.
Ag is MUCH more specific than PCR for contagious virus
this is the whole point of rapid antigen tests - they find people who are currently infectious. They are fast, give crucial immediate results and unlike PCR do NOT stay positive for weeks/months after someone is no longer infectious.
The details are difficult to describe via Twitter but I’ve tried on many occasions. The described low accuracy is false. These tests are doing very well to catch infectious people. We do NOT want to detect and isolate people who are not infectious and just have old remnant RNA.
I know people want to hate on the government for purchasing tests - but the Innova test is working entirely as expected. Very good for detecting contagious people. Which is the only goal here.
I’d be happy to write an OpEd for @guardian to explain.