Discussion about this post

User's avatar
Robert W. Chandler, MD MBA's avatar

Excellent question. Not at all a gotcha. You raise important issues that should not be ignored. Refusal to give up data sets on request is an important topic. You are not the first to run into this roadblock. This matter should be settled in a collegial manner as progress is made through cooperation and competition not suppression.

These are some additional issues I have with the approach used in Denmark 1 and 2. I missed Max’s presentation at D4CE meeting featured in the video, so I missed an opportunity to ask him directly. These are meant to be discussion points.

1. Methodology: log transformation, hierarchical followed by non-hierarchical cluster analysis then linear regression. Does this lead to similar very discrete clusters across multiple data sets? Independent reproducibility is important. What distortions are introduced with this methodology?

2. Why in the regression for the Blue cluster in Denmark #1 are the top four data points not included in the regression? The answer, I believe, has to do with requiring the regression to originate from the 0 point since there can be no SAEs (Deaths, Severe etc.) if no drug is administered. Including those points would move the x-axis intercept into positive territory, i.e., SAEs with no drug administered. I am just putting a ruler on those points and would like to run the numbers to see but without requiring a 0 origin to see if including those four points displaces the x intercept. Is non-linearity involved?

3. The yellow data set in Denmark #1 is hard to explain since there are SAEs, severe SAEs, and deaths in even in saline placebo groups. So, zero SAEs etc. makes no sense to me. Denmark #2 with the steeper slopes points to a process issue. Remember the kerfuffle over the word "placebo"?

4. The 89 cases need to be accounted for. This might tell us more about the international data corruption process. Hopefully you will get the data and sort out what happened.

5. Why are the slopes so different between Sweden and Denmark if there are truly three clusters with very high R2 but vary across countries? There is more going on than batch differences.

6. Why are the death plots unlike the others? Hot lots should have more deaths.

7. How sound are the time series by proxy plots since the date of release is not the date of administration? US vaxx administration data is highly granular, by daily by county as I remember. Fortunately, I was able to locate CDC data by month of administration for my time series analysis.

My takeaways: Early batches were very dirty and had a profound affect on women. (Geoff Pain just posted a piece on this topic) I believe this is because Process 2 produced a new drug formulation that had very limited testing and something in the formulation changed, other than the shift to a Tris buffer.

BioNTech/PFE knew from animal testing with BNT162b1 and the many versions of BNT162b2 through version 9 that the different formulations produced significant differences in “reactogenicity”. This is documented in the preclinical and phase 1 studies. Reactogenicity in some/many cases foretells significant health problems downstream (my opinion).

The final point is that the Denmark study was intended as preliminary to hypothesis testing research design. Unfortunately, governments around the world have taken away our most powerful tools to study the effects of these drugs leaving us with these flawed passive databases.

Expand full comment
James stutz's avatar

Bob, we are all thankful for your thorough research into this debacle. You, Malone and Wolfe have opened my eyes and heart (pun intended).

Jim S

Expand full comment
17 more comments...

No posts