Analyzing Roger Seheult's explanation of the VAERS death curve

We can use a new metric - Death-ratio-per-thousand (DRPT) - and infer that the VAERS death curve is very different from the ones for previous vaccines

Summary:

• There is a precipitous drop in VAERS death reports around Day 3, and Roger Seheult MD says it is an artifact of reporting. He also uses another pandemic year 2009 to explain the sudden rise in total number of reports

• I introduce a metric - death-ratio-per-thousand - where I calculate the ratio of death reports to all reports for each value of NUMDAYS (days to symptom onset) to show that the death reports for COVID19 vaccines are still very different from those of 2009

• If we consider the area-under-the-curve for this metric as a proxy for the cumulative seriousness of adverse reports over time, the COVID19 vaccine remains “dangerous” for well over 30 days

• I then modify this metric to normalize it (to dampen some of the noise), and there is a clear pattern of danger for a prolonged period of time compared to other vaccines

• By itself, this metric does not prove the vaccine caused the VAERS deaths, but it does show that the usual explanation for the VAERS death curve needs a lot more analysis

Recently Roger Seheult MD posted a link to a video from mid-2021 where he explained the VAERS deaths curve.

Link to tweet

This is the video debate. It also has chapters, so you can just jump to the relevant part where they discuss VAERS (from minute 35)

Here is a summary of his arguments (for why he thinks the mRNA vaccines are not dangerous):

1. personal anecdote of disbanding the temporary second ICU in his hospital after April-May 2021 because they did not see as many patients with severe COVID19. He asks: “If the spike protein from the vaccine was so dangerous, how were we able to contract down our hospital ICU intensive care services during  that period of time?“

2. there was a 2017 study which demonstrated how the number of VAERS reports suddenly jumped in 2009 for H1N1 and then went back to normal the following year. The study claimed that only 11% were serious, and that there were no death reports.

3. The FDA made the reporting requirements a lot more stringent for the COVID19 vaccine which in turn led to a lot more reports

4. there are a lot of deaths occurring every day in the US, and what you see in VAERS would be a small fraction of it even if you suppose a large underreporting factor like 40

5. people forget to file adverse events as we get farther out from the date of vaccination (to explain the sudden drop-off in number of reports)

6. why isn’t there a second peak corresponding to the second dose, given that the COVID19 vaccines follow a multi-dose schedule?

First of all, I don’t have any opinion about the personal anecdote simply because there are also other people who have their own personal anecdotes, from the same time period, which come to the exact opposite conclusion.

The US nursing home data is devastating for the narrative; now it's time for our own peer review before we submit this to a journal

What can you conclude from personal anecdotes?

And why would you trust one person’s recollection over another’s?

Second peak

Number 6 is the easiest argument to refute.

There is no second peak because the VAERS reports are always based on the most recent dose.

In fact, there are plenty of reports where you see the phrase “had the same reaction” or “had the same symptom” in the writeup.

I compiled them into a list and turned them into a search demo.

Click here to see the list of reports

Note: this search demo has a lot of false positives too.

You can use Machine Learning to improve the classification and reduce the false positive rate, but I have left it as it is to make my main point - which is that if a patient experiences the same symptom after first dose and second dose, the days to symptom onset is still calculated only based on the most recent date.

So there will not be a new second peak for dose 2, because VAERS reports only consider the latest dose (if anyone knows of a report which clearly states otherwise, please do let me know in the comments).

There is another way to interpret Roger’s comments here: if someone gets a reaction to the vaccine after Dose 1, and the D1 reaction is filed in VAERS, they will probably get the same reaction after Dose 2 ALSO and that should cause a bump up in the curve.

But VAERS does NOT update follow up information into the original report.

And even if they did, it will probably not matter for this particular argument because the NUMDAYS is a single valued field and there isn’t really any way to add two different events into it.

This could be considered a problem in the way the VAERS data schema has been designed in the first place, but my main point is that the current schema simply cannot reveal the the bump-in-the-curve that Roger expects.

All Cause Mortality

I agree with the all cause mortality argument (Number 4), but I also don’t think it matters one way or the other.

I wrote this about ACM in a previous article:

It sets an extremely low bar, and even worse, it now makes this an acceptable bar for future vaccine rollouts.

Link

As you can see, Dr Roger Seheult’s calculation explains why this is.

Even with a very high (low?) under-reporting factor, it will not even make the slightest dent in the ACM.

Nor should it!

Why would anyone want a vaccine which is so disastrously bad that it even shows up in All Cause Mortality statistics?

Besides, anyone using the ACM as a measure will need to account for a ton of confounders. Maybe someone will be able to do it, but I don’t expect it to happen without the cooperation of major health agencies.

And in case you haven’t noticed, the health agencies are in no mood to cooperate.

Link to tweet

A new metric for measuring the adverse event “recall” effect

Now I will group arguments 2, 3 and 5 by introducing a new metric for doing this analysis.

As you can see, Roger is making some points about how people recall adverse events over time.

Unfortunately we cannot read people’s minds to know how people do this during pandemic versus non-pandemic years, so we have to create some kind of proxy metric to try and measure this.

But what we do know is that the year 2009 had a lot of adverse event reports compared to the year 2008 and 2010 according to the 2017 paper cited by Roger.

So the death reports from 2009 can be used as a baseline to see if things look different for the COVID19 vaccines.

As you know, the NUMDAYS field in the VAERS DATA CSV file tells us how many days after the vaccination the symptom first appeared.

The metric I will be using is the ratio of the number of deaths reported for a certain value of NUMDAYS over the number of all adverse events reported for the same value of NUMDAYS. Since the percentage is a very small number, I will instead use per 1000 reports (basically this will be the percent multiplied by 10).

So let us call this metric Death-Ratio-Per-Thousand - DRPT for short.

For example, suppose there are 5000 reports of all types where the symptom onset is the same as the day of vaccination (which is recorded as NUMDAYS = 0 in the VAERS CSV file).

And suppose 75 of them are death reports.

So DRPT = (75 * 1000)/5000 = 15¹

I first calculated the DRPT for each NUMDAYS value from 0 to 30 for all the years 2000 - 2022 for all vaccines².

Click here to see the source dataset

Based on the above dataset I generated a chart which plots DRPT as a function of NUMDAYS for each year. That is, every year is represented by its own curve.

So what does this chart tell us about the difference between 2009 and 2021/2022?

a) unlike the DRPT values for 2009, the DRPT values for 2021 and 2022 rarely drop close to zero - so for every single day until day 30, a fairly large chunk of VAERS reports are actually death reports.

b) the curves for 2021 and 2022 are more similar to each other than they are to 2009

c) the “area under the curve” (AUC) is much larger for 2021 and 2022 compared to 2009. This area under the curve can be considered a proxy for the overall seriousness of the VAERS reports

d) the usual explanation provided for the 2009 curve does not explain the much larger AUC for 2021 and 2022. For example, a sudden rise in the total number of reports because “the vaccine was in the news” also implies almost none of them were serious³ . While this could be true for 2009, the wide gap between the 2009 and 2021/2022 curves shows that this argument doesn’t make much sense for COVID19 vaccines. That is, if that same trend held for the COVID19 vaccine, it should have also resulted in a lot of near-zero DRPT values after Week 1.

Normalized DRPT

While the DRPT itself is a fairly useful metric, notice what happens if you include 2009 and the years 2018-2022.

DRPT chart

Even though the AUC is still clearly larger for 2021/2022 compared to the other years, this chart is a little harder to read and we also see a lot of noise because of the “peaks” introduced by small denominator values.

One way we can “dampen” the peaks is to introduce a small delta value to both the numerator and the denominator and calculate a “normalized” DRPT.

For example, the DRPT is nearly 60 for NUMDAYS = 16 for the year 2009 primarily because only 65 reports were provided for that date, and 4 of them were deaths. This gives a DRPT of 4000/65 = 61.53

So I subtracted 3 from the numerator, and added 3 to the denominator (for all NUMDAYS and all years) to calculate a new metric N_DRPT.

Here is the N_DRPT for the same NUMDAYS = 16 and YEAR = 2009

N_DRPT = (4-3)*1000/(65+3) = 14.7

Now see the difference between DRPT and N_DRPT for NUMDAYS = 16 and the COVID19 pandemic year 2021

As you can see, there isn’t much difference between DRPT (42.32) and N_DRPT (40.92) because both the numerator (94) and denominator (2221) are much larger numbers.

Let us look at a comparison between 2009 and 2021/2022 for the N_DRPT values.

The difference becomes even more clear once you add in the years 2018-2020

You can barely see the peaks now.

We can, in fact, show ALL the years on the chart since it makes the AUC aspect a lot more obvious.

N_DRPT chart

Consider the “red band” I have added into the chart.

This would be N_DRPT values between ~20 and ~50.

While we see many values fall inside this band (or come very close to it) for all the other years until around Day 7, after that the trend lines completely diverge.

Beyond Day 7, there is almost no data point which falls outside the red band for 2021 and 2022, and almost no data point which falls within the red band for the other years.

As a sanity check, here is the N_DRPT chart for the foreign dataset⁴ exhibiting a similar pattern.

Foreign N_DRPT chart

Conclusion

Is this enough to conclude that the nature of the COVID19 death reports is different from the previous vaccines? Yes.

Does this mean we can infer that the vaccines caused all these deaths? No. But I certainly would not dismiss the COVID19 vaccine death reports as “nothing unusual”, which is what the vaccine pushers are doing.

1. Notice that you can also derive this number by multiplying the percent by 10. In this example, 75/5000 = 1.5%

2. All calculations are based on the last VAERS CSV files I downloaded about 2 months back. Since my calculations are only up to year 2022, a newer version of the CSV file will not make much difference.

3. Because if they were, we have much bigger problems to worry about if the Doctors are citing the sudden rise as if it was nothing!

4. This is based on the latest “good” dataset from Nov 2022 before the CDC stopped publishing the writeups for EU VAERS reports.

The foreign data set was gutted this week in VAERS and the cancer signal was halved, the myocarditis dose 3 response signal was lost and 994 spontaneous abortions/still births were dropped