The CDC Gave Me Whiplash
They said they weren't monitoring VAERS for safety signals, then they said they did, then they said they didn't but recently did. Confused yet?
Here’s a long overdue update on the ongoing saga with the CDC’s safety signal fiasco. Strap in—it’s a rollercoaster of a ride.
To refresh your memory, on June 16 , in a response to a FOIA request I submitted together with Children’s Health Defense, the CDC admitted that they had not monitored VAERS for safety signals from the COVID-19 vaccines, although they said they would. My posts on that story are here and here, and this is a link to my Defender article on their response and the safety signals I found when I analyzed the data they provided.
Now here’s where things get interesting. On July 23, the ever intrepid Zachary Stieber of The Epoch Times reported that the CDC reversed itself and stated in an e-mail to him that the “CDC has been performing PRRs since Feb 2021, and continues to do so to date.” That comes from none other than Dr. John Su, head of the VAERS team at the CDC’s Immunization Safety Office. If anybody should be able to provide a definitive answer to that question, it’s him. Sieber’s article was aptly titled “CDC Says It Performed Vaccine Safety Data Mining After Saying It Didn’t.”
But wait, there’s more! On August 11, Stieber published another article after the CDC reached out to him, doing an about-face on John Su’s assertion and giving me a case of whiplash. Turns out, unbeknownst to the head of the VAERS team, the CDC had not been performing PRR’s since February 2021, after all. Apparently, they only got around to doing them starting on March 25, 2022 and stopped on July 31. That article was titled “CDC Admits It Gave False Information About COVID-19 Vaccine Surveillance.”
Why the confusion? Get a load of this, from Stieber’s article:
The spokeswoman said it “misinterpreted” both Children’s Health Defense and The Epoch Times.
Children’s Health Defense had asked for the PRRs the CDC had performed from Feb. 1, 2021, through Sept. 30, 2021. The Epoch Times had asked if the response to the request was correct.
The spokeswoman said the CDC thought “data mining” referred only to empirical Bayesian (EB) data mining, a different type of analysis that the FDA has promised to perform on the VAERS data.
“The notion that the CDC did not realize we were asking about PRRs but only data mining in general is simply not credible, since our FOIA request specifically mentioned PRRs and their response also mentioned that they did not do PRRs. They did not say ‘data mining in general,'” Josh Guetzkow, a senior lecturer at The Hebrew University of Jerusalem who has been working with Children’s Health Defense, told The Epoch Times via email.
“There is also no credible reason why they waited until March 25, 2022, to calculate PRRs, unless it was in response to our initial FOIA filed in December 2021, which was rejected on March 25, 2022—shortly after they say they began their calculations.
“It means the CDC was not analyzing VAERS for early warning safety signals for well over a year after the vaccination campaign began—which still counts as a significant failure.”
The CDC also hasn’t released the results of the PRRs.
“PRR results were generally consistent with EB data mining, revealing no additional unexpected safety signals,” the agency spokeswoman said. “Given it is a more robust data mining technique, CDC will continue relying upon EB data mining at this time.”
The CDC is behaving like it has just been caught with the maid in the pantry with its pants around its ankles and is frantically and unconvincingly trying to explain why “this isn’t what it looks like, honey.”
Let’s acknowledge the fact that the head of CDC’s VAERS safety monitoring doesn’t know his right hand from his left, and move on. There are a couple of other important things to note about their response:
The CDC claims they started VAERS safety monitoring only on March 25, 2022. Why then? Well, we had filed our original FOIA back in December. After not hearing from them, I asked the legal team to write to them for an update. The date of that e-mail was…March 22nd. Quite a coincidence, wouldn’t you say? On March 31, they sent a letter to CHD rejecting the initial request for being broad, vague and unduly burdensome.1 Here is a screenshot, full letter embedded below:
I don’t know about you, but I find the timing awfully suspicious. I can’t think of any other reason why they would have started conducting PRRs on March 25th of all days — assuming, of course, we take them at their word that they have been (I don't). They refused to release the PRR calculations to Zack. If they had them and they showed nothing, why would they refuse?
We appealed their response to our FOIA where they claimed that they had not done data mining, and we are still waiting to hear back on that. Ditto for our FOIA to FDA on the data mining they were supposed to do.
The CDC also claimed that Empirical Bayesian Data Mining is a “more robust” technique than PRRs. In their full response to Zack, they wrote:
“As previously mentioned, PRRs can provide a rough means of examining disproportionate reporting but are at best a crude measure prone to many false signals. Therefore, to detect disproportionate reporting, CDC and FDA chose to rely on Empirical Bayesian (EB) data mining. EB data mining is a more robust data mining technique used to analyze disproportionate reporting and is conducted by FDA using VAERS data.”
I beg to differ. This is what I wrote to Zack when he reached out to me for comment:
Here is a 2005 paper from researchers at FDA on analyzing VAERS. The authors conclude that the different methods “may be superior in certain situations but inferior in others."
Here is a more recent FDA 'white paper' discussing data mining methods. They compare the PRR method with the Bayesian Data Mining method they use. Basically the advantage of the Bayesian method is that you have less false positive signals when the number of reported events is very small. But this is not really an issue for most of the COVID-19 vaccine AE’s because there have been so many reported.
Finally, here is an excellent paper from 2020 on adverse event signal detection, showing that each method has its strengths and weaknesses. Although EBDM does a better job of avoiding false positives when the number of events is small, it is also less sensitive to picking up true relationships. So there is a tradeoff. But if you had to pick one method and wanted to err on the side of caution, you'd use PRRs, not the Bayesian method.
The authors of that paper concluded: "No method was superior to the others in terms of the statistical power and sensitivity of detecting true signals. It is recommended that those conducting drug‒adverse event surveillance use not just one method, but make a decision based on several methods."
So as usual, the CDC doesn’t know what it is talking about and cannot be trusted to keep simple facts straight. Stay tuned for the next chapter in our ongoing battle to wring the truth from the CDC and FDA.
Update, Sept. 7: A few months ago, Albert “welcometheeagle88” Benavides discovered that there were reports being systematically deleted from VAERS, sometime over a thousand every week. He recently tweeted this picture from his amazing VAERS dashboard, showing the distribution of deleted reports from Jan-Sep 2021:
Notice anything? There is a HUGE jump in deleted reports beginning on April 1st! 1,447 cases to be exact. And the mass deletion continues apace until July. So they started this frenzy of deletions at the same time they started (allegedly) doing their safety signal analysis. Coincidence?
Yes, I know, technically the number of deletions jumps a week later than when they said they started their monitoring, and it also switched into lower gear at the beginning of July rather than the end of July. But, we don’t really know what timespan or reports their monitoring included. Still, there is a surprisingly close overlap between the dates. At least we can be sure it didn’t include these deleted reports.
Note the article states that the CDC’s response was on March 25th, but in fact their response was on March 31st. I got the dates mixed up when I discussed this with Zack — so the error is my fault, not his.
Josh, about 15 years ago I questioned the veracity of an FDA paper On anthrax vaccine by Manette Niu. She remains at FDA and heads a group of epidemiologists today. She claimed to use this Bayesian analysis method, but underestimated certain symptom combinations by a factor of 10 back then. She was forced to issue a correction once I threatened her with the charge of scientific misconduct.
What may be of greater interest is the fact that I asked for the algorithm she was using to perform this Bayesian analysis, and I was told it was proprietary and belonged to a company in Massachusetts. The implication was that FDA did not know the method it was using to analyze adverse events.
Might I suggest you look into the algorithm being used today?
Much like the UK HSA, MHRA and JVCI, it's really, really difficult to give a straight, honest answer when everything you do is a fricking lie?!