Thread by @mimifontaine

If AstraZeneca increased the risk of common blood clots, as well as the rare and dangerous ones, would we even know about it?

[thread]

The focus of the blood clot risk of the AstraZeneca vaccine has been on the clots in unusual (and dangerous) sites such as the brain and abdomen that have been found alongside low platelets (thrombocytopenia)

Thanks to Norway and Denmark, this new syndrome was rapidly identified and reported as a risk, even if certain parties were initially in denial (looking at you MHRA)

The curious observer naturally wonders, could AZ cause more of the garden variety DVT-type clots as well as the serious, “go to A&E NOW” type?

Luckily we have a world-beating medicines agency to figure this out for us!

J/k, we have the MHRA

Here’s what they said on 6 May:

Well that’s a relief. How does the MHRA know this?

They do smarty pants statistical analysis of course! But they don’t make it public because you, the public, can barely be trusted with slogans of longer than three words, nevermind aKsHuaL nUmBerS and stuff

But worry not -

...luckily we have the bureaucratic, inefficient but helpfully paperwork-loving EMA to pull back the curtain on the statistical wizardry that goes on behind the scenes

In their review of 16 March data the EMA carried out an “observed-to-expected analysis”

This means they looked at what % of blood clots “expected” to happen (independently of the vaccine, based on historical incidence rates) actually *did* happen

The closer that is to 100% (or in numbers, 1.0) the less we would suspect the vaccine of affecting blood clot incidence

Throwing some confidence intervals around the result tells us if the statistical conclusion is “move along, nothing to see here” or if the difference is significant - that is, likely not due to chance

As you can see, the EMA even helpfully interpret this for us:

Does this mean…. That the AstraZeneca vaccine actually *reduces* the risk of blood clots? By like, NINETY EIGHT PERCENT??

Not so fast. Before we alert the medical world to this miraculous blood clot prevention drug, we should check the fine print of exactly *how* EMA knows how many blood clots had happened in the ~20 million people who had received AZ by that date:

Enter EudraVigilance, the EU’s pharmacovigilance system for managing and analysing information on suspected adverse reactions to vaccines or other medicines, similar to the UK Yellow Card scheme or the US VAERS database

Presumably these are state of the art systems, linking each person’s vaccination status to their country’s electronic health records so EMA can track what happened to people after their vaccine using accurate, real-time data, right?

Dear sweet, innocent, naive reader - NO

Reports are made when someone (doctor / patient) *suspects* that the vaccine had something to do with a medical condition and then bothers to report it

These reporting systems are *not* comprehensive databases of everything that happens to someone after vaccination

So, EMA... you’re implying that to detect changes in blood clot incidence, we should compare the entire population baseline to... the tiny subset of the blood clots post-vaccination that were reported into Eudravigilance?

And yet this is how people who really should know better seem to think these reporting systems work

People like Phil Bryan, MHRA Vaccines Safety Lead

(Come on, Phil - you know that the number "reported" isn't anywhere near the number that happened)

People like this random guy someone decided to ask:

"possibly, in fact lower" - not exactly shocking when your monitoring system is literally only for the small number of clots where someone suspects a link to the vaccine...

People like this very independent person called Ann Taylor from a company called AstraZeneca

As a separate point: it was well-known at the time of these statements that the blood clots in question were *not* garden variety DVTs, they were in unusual sites occurring with thrombocytopenia in younger people (see below from the Norwegian Institute of Public Health

Yet Ann and co deliberately referred to blood clots *in general* - not only was that irrelevant but their assertions were based on absurd comparisons of baseline rates to Eudravigilance reports

As if they wanted people to think Norway, Denmark etc paused AZ over a few DVTs...

All that to say - while I am not suggesting that AZ causes blood clots without low platelets, I am not convinced EMA or MHRA would be capable of detecting it anyway

I will finish with a simple example to illustrate how both sides misinterpret vaccine side effect reporting

Reactions to the latest Yellow Card Summary showing 722 UK deaths in total after vaccination with AZ

Antivaxxers: OMG 722 people died *BECAUSE OF* the vaccine!

Vaxx fanatics: *ONLY* 722 people died after the vaccine, statistically there should have been tens of thousands!