Danish study on lot variation and associations with SAEs

Thoughts and comments

Update: I have spoken to two of the authors of this work and one of them, Max Schmeling, has replied with the following reply that you can read in the comments.

Thank you for all your work and thank you very much for your thoughts on our study. I just wanted to add, that the lot size data, we obtained from The Danish Serum Institute, is in fact the number of doses pr. batch, that were shipped from the Danish Serum Institute to all the Danish vaccination centers. It is perhaps the data, that comes closest to being the real number of administered doses pr. batch, since the shipped doses exclude any doses, the Serum Institute might have in stock.

I should also mention that the ADR data, is in fact collected much the same way as VAERS, but the quality is in my opinion much better. Of all the known reported batch numbers only, a mere fraction does not validate correctly towards the lot size list.

Something to consider when balancing the equation.

Hi there loyal readers! Well, I couldn’t help myself. The thunder storm of the season woke me up far before the birds today, and so I decided to look into this new Danish study examining lot-to-lot variation in the context of severe adverse events (SAEs). The study is entitled: “Batch-dependent safety of the BNT162b2 mRNA COVID-19 vaccine” and it was published in the European Journal of Clinical Investigation on March 30, 2023. It’s awesome that people are doing this work and examining pharmacovigilance data world-wide. More please.

The conclusion of this study is that, within the Danish population, of the people who got 2 doses of Pfizer mRNA product, their SAE occurrences were linked to the batch number of the product. They claim that further investigations of this lot-to-lot variation need to be done, and I couldn’t agree more. The manner in which this is done is important.

In conclusion, the results suggest the existence of a batch-dependent safety signal for the BNT162b2 vaccine, and more studies are warranted to explore this preliminary observation and its consequences.

The data comes from the Danish Medical Agency (DKMA) which is akin to VAERS in that it is a passive reporting system, and SAEs were classified as per the MedDRA coding system: same as VAERS. Data for 3,748,215 people in the context of 52 different BNT162b2 vaccine batches, ranging from 2340–814,320 doses per batch, was collected for batches ‘administered’ between December 27, 2020 and January 11, 2022 in Denmark. 7,835,280 doses were administered to these people so each is presumed to have received 2 doses. 43,496 SAEs were registered in 13,635 people. So the numbers are good. Plenty of data.

They plot their findings and it looks like this:

This is compelling, no doubt. It certainly looks like there are batches that have fewer doses but resulting in much higher incidence and reporting of SAEs, doesn’t it? There appear to be at least 4 batches associated with greater than 7,000 SAE reports with fewer than 100,000 doses per batch. This really seems like it definitively means that these batches are more toxic, right? Well… not exactly.

Without knowing how many of these doses from each batch were actually injected into arms, we cannot definitively draw this conclusion. Granted, it does warrant further investigation!

They calculate the SAE rate like this:

The total number of SAEs associated with each batch was divided by the number of doses in the batch to obtain the rate of SAEs per 1000 doses.

I know this is going to make me re-heeeeeallly unpopular, but this is incomplete.

A more accurate way to calculate the SAE rate would be like this:

The total number of SAEs associated with each batch was divided by the number of doses actually administered in the batch to obtain the rate of SAEs per 1000 doses.

Let me make something utterly clear here. I firmly believe there is lot-to-lot, and even vial-to-vial variation. Also, I know that people are different immunologically and that not all people were dosed in an appropriate way using aspiration, for example. I also am very convinced that there is dsDNA contamination and perhaps even LPS contamination in certain batches. I also believe that even if the contents of vials were deemed to have been injected into people, they might not have been. There are accounts of administrators simply not injecting the contents. I have even heard of injection saline in place of the vial contents. Perhaps the administrators couldn’t live with themselves knowing the contents were experimental?

Providing evidence of contamination and potential subsequent toxicity is tricky business. Even if there is contamination, how do we prove that associated adverse events were caused by this particular contamination? It is hard to prove. (By the way, the onus is NOT on us to proof these things aren’t safe; it is on them to prove they are.) I commend the authors for providing this evidence and I believe definitive proof is coming very soon.

I commend the authors for providing this evidence and I believe that we must examine the contents of the vials directly as an obvious next step. Grab 100 samples from each batch and sequence them. Do mass spec. Whatever. It’s not like the manufacturers don’t have the resources to do this. And then, once we have a better idea of which batches are contaminated and how they are contaminated, for example, then we can cross-reference these batches back to deaths and other SAEs to more precisely define why people are experiencing a variety of adverse events. Perhaps we can prove that if the %RNA integrity is low, fewer severe adverse events ensue? I am not saying that we can’t do this in reverse, ie: by examining SAEs and cross-referencing back to VAERS or the DKMA data, but it’s much harder to do because, again, we can’t know how many people got injected from each batch so we can’t calculate the actual rate of SAE occurrence per batch.

Here’s an example that I used recently to explain why I think we can’t make definitive claims about death/SAE rates without knowing how many shots went into arms from each batch.

With regard to the batch size vs death counts (or SAEs or any AE), we actually can't know how many actual doses were administered per batch, thus it's not possible to make statements about differences in toxicity between lots. 

ie: It's possible that more shots were doled out from a batch that had fewer doses which would make it look like this batch was more toxic. 

For example, if batch 1 has 50,000 doses and batch 2 has 100,000 doses, and batch 1 killed 50 people and batch 2 killed 5 people, then yes, it looks like batch 1 is much more deadly. But if only 1/20th of batch 2 was distributed, then they have the same death rate: 1/1000. 

Let’s illustrate this.

So even though one batch is associated with 10 times more death and has fewer doses than another, without knowing how many doses went into arms per batch, we cannot say that one batch is more toxic than the other. For example, if a batch associated with only 5 deaths has 100,000 doses then according to the authors, the death rate is 0.00005 but if only 5000 of these doses were administered, then this rate becomes 0.001 which incidentally, is the same death rate for the batch with the lower number of doses that killed 50 people. See where this consideration makes a difference?

I don’t mean to be the annoying person with their hand up in class, but, I am that person.

Their study is very important and more studies like this need to be done. I think to make an even stronger case in support of studies like this, we should really push batch assessment at the bench level, and then combine this with observations from pharmacovigilance data bases.

Peace out.