Thread by @dgurdasani1

Let's looks at the highly flawed study that is being cited as showing no difference between LFD testing and isolation of contacts in schools. This is simply incorrect, and the premise of the trial as reported in the preprint out now is rather shocking from an ethics perspective🧵

First, this is a 'non-inferiority' trial. Such trials essentially compare A with B, and decide what level of difference between A (intervention) and B (control) would be acceptable before hand, and design the trial only to pick up this difference.

So such trials cannot say A=B, but can say A is not worse than B by x% (with x% being decided beforehand).
So what difference was this trial designed to find?

It was designed to find a 50% or greater change in transmission in the intervention compared to control arm.

The trial considered that an 'upper bound of relative increase in transmission of 50% would be acceptable'. This is essentially looking at 1.5x more people being infected in the trial arm compared to the control arm. So the trial wouldn't find a difference unless it was *huge*

1.5x times more people being infected in one arm is *far* from acceptable, and it really begs the question why such a non-inferiority trial was allowed to go ahead in the first place. Because in essence it treats a difference of 1.5x or less as if it doesn't matter. But it does!

Sadly, the truth is even worse than this. Only 42% of those in the intervention arm participated in daily LFD testing. The rest isolated. So the final analysis that finds no difference between A and B, is based on the fact that less than half of those in A followed A protocol!

This means that such a trial because of *dilution of A* by less than half following protocol wouldn't even find the *huge* 1.5 fold difference that it was designed to find, because essentially most people in A were following the same protocol as B!!

The trial shows shows a risk of 1.2x symptomatic infection in contacts in the trial vs control arm -the uncertainty around this is huge, & is compatible with anything from a 20% reduction in transmission to a 79% *increase* in transmission due to LFD testing replacing isolation.

For asymptomatic contacts this could vary from anything from 60% reduction in transmission to 60% increase in transmission- so in essence a completely meaningless result due to inadequate sample size, and very low compliance with the intervention.

Unsurprisingly, the trial showed no significant difference in attendance between the two arms either, which really highlights how poorly designed the trial was. Should we conclude that replacing isolation with daily testing wouldn't improve attendance?

No, in essence, we can't conclude anything at all from this trial because of it's extremely poor design. It literally tells us nothing at all.

Also worth noting that most of this trial is based on pre-delta periods, & also when infection rates in children were much lower.

It's also notable that the trial didn't actually seek to test other children in the bubble, and relied only on LFD testing of 'contacts' defined based on duration and distance, when we know aerosol transmission puts everyone at the classroom at risk.

So it was already designed in a way to miss a lot of transmission that may happen asymptomatically within classrooms among those not defined by close contacts by DfE (based on their archaic unevidenced definitions) - despite being at risk from exposure to a case.

Neither did it study onward transmission to members of the household, or risk to them from the trial. In any case the numbers studied, and the low participation meant that any difference, even large ones, would be impossible to detect with this design.

One thing the trial did show was that sensitivity of only 53% for LFDs vs PCR. This means only half of those positive on PCR were detected by LFDs. One could argue that LFD may pick up more infectious cases, but the current study doesn't provide any real data on this.

So, what does the trial show all in all?
Nothing, except that a PHE ethics committee was happy to approve a trial on the basis that a <50% increase in transmission in schools, as a result of the intervention was acceptable, to the extent that the trial wouldn't even detect this!

And a trial team, including scientists were happy to carry this out, with public funding, knowing fully that it might increase risk to children & communities, and despite it not being able to detect even *huge* differences in infection between the intervention and control arms!

Not only were they comfortable with that, they were happy to continue, even as infection rates rose, delta spread through schools and back into communities, arguing the risk hadn't changed!! Even as schools participating closed.

This trial tells us nothing about DCT vs isolation

It does tell us a lot about the state of science, and the (non-)independence of ethics committees.

And about a govt and scientific leadership, and members of ethics committees who are happy to put children at risk in the name of research, that is neither rigorous nor ethical.

Full preprint here:
modmedmicro.nsms.ox.ac.uk/wp-content/upl…

And our earlier piece in the BMJ discussing all the ethical and scientific issues (our concerns now fully vindicated by the trial results) here:
blogs.bmj.com/bmj/2021/06/17…

I also refer you to the 'events' trials which were also designed to not detect an effect (PCR tests *within* 5 days of event), where only 15% tested, but were hailed as showing events were safe. Even as 15% of those from Scotland who attended Wembley stadium were infected...

These trials are sadly a sham. I get a corrupt govt doing this, but researchers undertaking such non-rigorous work to facilitate a government agenda that puts people at risk is frankly shameful.

Just want to add that the ethics committee that approved this trial was the PHE ethics committee that reviews PHE research. I'm specifically critiquing them here. Notably, the Oxford ethics approval doesn't seem to have been sought, despite the chief investigator being at Oxford.