When lack of safety equals safety

New FDA study claims to show covid vaccines are safe for children even though the study shows significant increase in serious heart conditions

I was contacted by both the Epoch Times and the Children’s Health Defence (CHD) to comment on this FDA article that appeared this week.

Some of my comments and those of other colleagues are included in both the Epoch Times article and the CHD article, but not all the problems we raised are included.

For the record, here are the full set of comments we made.

I said:

I’ve had a quick look through it.

The conclusion that “these results provide additional evidence that COVID-19 vaccines are safe in children” is pretty ludicrous given the fact that a) they show the safety signal for myocarditis and pericarditis is very strong (and is likely underestimated given some obvious weaknesses of the study); and b) the lack of any benefits of the vaccine to children of this age since they are at no risk from covid and (as others studies show) are actually more likely to get covid with the vaccine.

Issues about the study:

• There is a clear ‘healthy user’ bias (or other systemic bias like misclassification or under-reporting of outcomes) in the sample and analysis methods as can be seen for the conditions excluding myocarditis, pericarditis and anaphylaxis. Unless the vaccine has magical health benefits why would incidence of things like bells palsey and appendicitis be so much lower in the vaccinated? In fact, the authors admit that they were unable to follow up extensively on the outcomes.

• I don’t like the fact that, where number of outcomes is  “<11”, the exact number is not provided.

• Given that the types of outcome are inevitably going to be rare in children I’m a bit concerned that the separation into 3 age groups is diluting the results. For example, I suspect that if the groups were combined there could be a significant safety signal for Anaphylaxis based on the reported risk ratios for the 12-15 and 16-17 groups.

• There may be conflicts of interest among the authors.

Dr Scott Mclachlan said:

The “risk window” for some events seems arbitrarily too short.

It feels to me that all they are really giving us is “myocarditis” - likely because we already know (and they already admit to) that one. For everything else it just feels like they tried to hide the noise in the numbers.

Another bias is that they included patients on an insurance plan (who are therefore more likely to have a higher socioeconomic status because their parents are employed and therefore provided with or able to afford expensive health insurance). I think the outcome numbers would be vastly different in a group of children whose parents are not as well off.

It is also retrospective (health records) based. I don’t believe they declare what happens when a parent loses their job (and therefore their health plan) and the child ceases to be an active health plan member. I have seen studies out of Optum and Kaiser Permanente in California and Boston in the past where they assume the health plan member had no reaction simply because nothing was written in their health record, but it was assumed they were still a member - sometimes in error. In America this is an incorrect assumption because if you can’t afford to attend the Health Plan doctor, you have to go to a community clinic or the doctor funded by your new health plan provider. That whole “the absence of something is not proof it didn’t happen” thing springs to mind.

Prof Josh Guetzkow said:

I agree with Norman that the main weakness of this design is the comparison they are doing. They are comparing a highly select group of child and adolescent insured vaccine recipients to a historical baseline population consisting of everyone in the relevant age group who were insured. But the people who get vaccinated tend to be healthy, sometimes much healthier, than others. The Nature paper on increases in POTS after vaccination showed this very clearly because the baseline risk values for the unvaccinated were many times higher than for the vaccinated group. 

Amazingly, this limitation isn't even mentioned in the article!

I also agree with Scott that the risk windows are rather short and arbitrary. Narcolepsy is a perfect example: the narcolepsy cases following the Swine Flu vaccine in 2009 took months to manifest and be diagnosed. If this was the same then you wouldn't expect to see the incidence increase 42 days after jabbing. 

Another thing that is arbitrary are the ratios they establish that would trigger a signal. This is in the protocol of the study, Table 5 page 15: https://bestinitiative.org/wp-content/uploads/2022/12/C19-Active-Monitoring-Protocol-Addendum-2022.pdf

Why, for example, is the ratio for appendicitis 1.09, for DVT it's 1.25 and for GBS it's 2.38 (for the .5-4 year olds)? That means that in order for a 'safety signal' to trigger, you'd need a ratio of 2.38 the incidence rate of GBS among vaccinated compared to expected pre-COVID 'background' rate and the ratio would need to be statistically significant. Oh but wait a second! Why don't the authors include sequential test results for GBS in their study? It's in the protocol. They have the background rates. So why no comparison to background rates? The same can be said of some of the other AEs for which there is only descriptive results. You can triangulate between the study protocol (linked above), the published tables (especially in the supplement) and the protocol for establishing background rates to see exactly which ones should have a sequential comparison but instead are only descriptive.(MIS-C is an exception since that is a new diagnosis developed after COVID started.)

Also, according to the protocols, they are supposed to have some 'negative case' controls -- AEs that would be unrelated to vaccination but could detect other differences, like care-seeking. Why aren't those results reported? 

Finally, I think the idea that they look at only 20 very specific AEs then declare them safe upon not finding anything is very myopic. There are many different ICD 10 codes for blood clots or strokes, for example, but here they look at very specific ones like DVT. What might they find if they looked at, for example, a broader range of blood clot diagnoses grouped together? 

Also (and related), they say in the protocol that they would add AEs to look at based on signals identified from other sources. What about all the concerning safety signals identified by the CDC for these age groups?