DEI at the PEI

Ezra 6:2

Recently and quite unexpectedly, the Paul Ehrlich Institut (PEI, the German government agency responsible for drug monitoring) published raw data they had collected on suspected adverse events after Covid vaccination. Some people have commented (in German: here, here) on the issue – maybe many more; I wouldn’t know because I am not on FIXT (Facebook, Instagram, X, Telegram). I will provide my own perspective on what can, and also what cannot, be inferred from the data.

There are two files, a large one containing everything reported before 2024, and a small one containing everything reported during the first half of 2024. Altogether, there are 1,001,539 adverse events (I will simply speak of events) for 348,987 cases. A case is the combination of a person and (events after) a vaccine dose. Someone having experienced different events after different doses may result in more than one case, with no chance of matching the person.

The whole thing gives the impression of raw data. Hospitals, general practitioners, nurses, affected individuals themselves or someone close to them can all access the reporting system. Errors, duplicates, missing data and inconsistencies are to be expected in such a system but the PEI does not seem to have done much about it.

Let me list a few interesting, strange or ridiculous aspects:

• Of the events, 71% had been experienced by women, only 28% by men (and 1% by people of unknown sex). The proportion of women among cases is 68%, which means that not only did women allegedly suffer more often, but suffered more events on average. While it is plausible that some conditions pertain to one sex more than to the other, such divergence is more likely due to massive underreporting (by men in particular).

• Speaking of sex-specific events, there are 20,250 related to menstruation (search: “menstr”). Of these, sex is “unknown” for 135, and there is even one unlucky man (case 1,259,702).

• But this is balanced by a “female” case (1,339,842) with erectile dysfunction (“Erektionsstörung”). Which doesn’t explain anything but the stupid title of this post.

• For 6,781 of the 836,199 instances where both date of vaccination and date of event are given, the event predates vaccination (there might be various explanations for this, e.g., earlier vaccine doses, or conditions that were present before vaccination but then became worse).

• Unfortunately, the dose number is not in the data although the PEI has criticized others for omitting the dose number (this was in the context of the Danish study on batch dependence, see also my take here).

• For 601,723 events, the years 2020 or 2021 are given as the year of reporting, but the PEI’s safety report as of 07.07.2022 (which covers all of 2020 and 2021) only mentions 244,576. But of the 2,255 deaths mentioned in that safety report, I could track only 1,033 (searching for “tod”, and then excluding certain conditions like near-death experience (“Nahtoderlebnis”)). So on one hand, the data are more extensive than claimed previously, and on the other hand they are more restricted.

• The data are very unbalanced regarding detail. For example, consider two cases related to the Johnson&Johnson vaccine: case 1,004,632 lists 15 events (I won’t translate: Synkope, Kopfbeschwerden, Herzfrequenz erniedrigt, Schwindelgefuehl, Tinnitus, Gefuehl anomal, Sehen verschwommen, Schmerzen im Oropharynx, Panikattacke, Tremor, Angststoerung, Schwindelgefuehl, Leistungsfaehigkeit erniedrigt, Arbeitsunfaehigkeit, Transitorische ischaemische Attacke); but case 1,004,937 lists only one event: death. Just death? From what?

• There are 6,292 distinct events (top 3: headache: 71,424, fatigue: 63,747, influenza-like illness: 49,070). Note that total numbers of events are difficult to compare (cf. above: issues with menstruation can only occur with women, and only in a limited age band).

• There are 4,327 events of myocarditis or pericarditis, at least 300 of these for children. Altogether there are at least 17,448 events for children (0-17).

Most events occurred at the date of vaccination or on the very next day. The following diagram shows the distance in days between date of vaccination and date of event:

For 569,854 events (from 187,196 cases) the vaccine batch is in the data. The following diagram shows numbers of events (blue) and of cases (red) by batch, ordered by number of events:

The numbers vary widely by batch but since we have no information about batch size it is impossible to infer anything from these data.

Since most events occurred close to the vaccination date, it makes sense to plot the number of vaccine doses per calendar week (left axis, black line; from week 53 of 2020 until week 52 of 2022) with the number of events recorded per that week (right axis, columns):

Events indeed scaled with vaccinations, except in early 2021 when AstraZeneca (AZ, blue part of the columns) wreaked havoc.

Vaccination with AZ was stopped in Germany on 19.03.2021 (during calendar week 12). According to the FAQ, the main reason for hitting the brakes were seven (!) events of sinus venous thrombosis. Now, the total number of events of sinus venous thrombosis in the data is 708 (319 of these connected to AZ). While I get it that the total number of AZ vaccinations was small compared to the other vaccines, and the seven cases occurred during a short period of time, I need a convincing explanation why a hundredfold increase in events did not produce a signal.

While we are at it: Dr. Erich Freisleben (see also Stephan Sander-Faes’ reports here and here) meticulously documented more than 250 cases from his own practice, and reported them to the PEI. For example, his case no. 6 matches case 1,344,624 from the PEI database (but without mentioning death of the patient as event!). However, I tried a few more cases and could only match a tiny number of them. It might have to do with the fact that I have no medical training and am therefore bad at synonymous symptoms, but who knows? From the database it is hard to infer how the PEI actually dealt with all the reports they were bombarded with.

And bombarded with reports they were. The PEI also has analogous data from the pre-Covid era (as pdf only, a little hard to evaluate). The total numbers of cases in the years since 2004 look like this:

Even after (approximate) scaling by the number of vaccine doses dealt out in Germany, the image does not change much:

The sheer total number of events makes it obvious that Covid vaccination was a bad idea (considering that it didn’t prevent infection, didn’t prevent transmission, didn’t prevent serious illness, and didn’t prevent death). Underreporting may be gigantic. Pre-Covid, everybody knew that underreporting rates of 90% were not uncommon, and I do not see a reason why this should have changed with the Covid vaccines. Note, for example, that I took one dose of J&J for the team (I won’t go into the details, but it was to no avail anyway; I was not allowed to enter shops early in 2022, or take my children to the pool), and the side effects I experienced are not in the data, nor are those of anybody I know.

Apart from that basic insight, the data are not worth much. But the PEI is sitting on a second, even larger, data set, the data from the SafeVac app (around 3 million entries). Many people are waiting for their release (FOIA requests are running), and new insights might be gained from comparison of the two datasets.

I suspect that the true total number of cases is of the order of magnitude of 2-3 million (for around 200 million vaccine doses). This suspicion is compatible with the following back-of-the-envelope calculations:

• Early in 2022, Andreas Schöfbeck estimated 2.5-3 million events for 2021 only (and the courageous man lost his job as CEO of a large health insurance company).

• As mentioned above, underreporting rates of 90% have been common knowledge in the past, which gives 3.5 million cases.

• Dr. Freisleben recorded 52 cases among his own patients. Take that number, multiply by 55,000 GPs in Germany, and also get 2.86 million cases.

What are we to make of the PEI’s data dump? Is it an attempt to shed responsibility? Is it a first step towards open discussion? Is it a cry for help?