Scienceme Street
The CDC issues a blockbuster "teen Covid vaccine efficacy rate" report. Is it real-world data, or a statistical fairy tale? And why should we even care?
Say you want to know if something you were doing worked. Like, you were putting helmets on local baby birds, in case they fell out of their nests. You helmeted up 1,000 baby birds. So, curious experimenter that you are, naturally you head to the veterinary hospital, and count how many of your helmeted-up baby birds are to be found in the Fell from a Tree Ward. There are 3. And then, naturally, you go count how many of your helmeted-up baby birds are in for something else. There are 50! That’s incredible!
But, hold on.
That’s not science. You need to compare the portions, dum-dum!
Fortunately, you realized your mistake before leaving the hospital, and set to counting. There’s 100 birds overall in the Fell from a Tree ward, and 100 birds in the other part. Ah, now you’ve got science. Your helmets are 94% effective at preventing hospitalization for falling from a tree!
100% x some math = 94%!
As you walk out of the hospital, another one of your birds is brought in for helmet-associated injury.
This evaluation demonstrated that 2 doses of Pfizer-BioNTech vaccine were highly effective in preventing COVID-19 hospitalization among persons aged 12–18 years. Findings reinforce the importance of vaccination to protect U.S. youths against severe COVID-19.
So proclaims the new report released by the CDC’s notoriously propagandistic MMWR platform, manufactured using the “case control” method depicted above.1 19 pediatric hospitals in a smattering of states were inspected. The patients for “Covid 19” and the patients for “something else” were counted up. There was a much higher percentage of vaccinated patients in “something else.” Some math was done. Et voilà.
But to analyze or critique yesterday’s report further perhaps diminishes a more significant contextual backdrop. Since the end of summer, the MMWR has conspicuously gravitated toward the topic of “kids and Covid;” at this point the outfit resembles a deflated tarp draped over the outline of a singular object. Whether that object is intrinsic curiosity, or external command, I leave to the reader to judge. The end result is the same: A string of cherry-picked, misleading “real world” studies that add fuel to the media/state’s endeavor to achieve, by one means or another, the permanent denaturalization of human childhood.2
The Pandemic™, of course, has not been exclusively or predominately, but most significantly a war on childhood itself; and nothing after all is more essential to that “self” than the obviation of adult anxieties over death. Innocence of death, therefor, must be demonized, catastrophized, legally suppressed, and psychologically tortured into nonexistence.
That is the war, and in most of the country (by population), the media/state are winning it. Yesterday’s report merely represents one more pamphlet dropped out of the airplanes to demoralize enemy parents.
Additionally - on the subject of whether critiquing this study is really worthwhile - it’s not like the results are even that outlandish; and neither do they lead to the asserted recommendation anyway. Did the experimental Pfizer/BioNTech Covid vaccine (the only one emergency-authorized for American teens) prevent 93% of would-be hospitalizations “for Covid-19” among recipients between June and September? Sure. Why not. Teen vaccination uptake was very low on June 1, when the emergency-authorization was expanded to include 12-15 year-olds.3 And the Covid vaccines appear to have a limited window of strong “infection efficacy,” something like three or four months.4 If you can’t test positive for “infection,” you probably won’t experience severe outcomes from infection, and you certainly can’t meet the condition of being hospitalized “for Covid-19.” I hope that has muddied the waters between objective and statistical reality enough.
The so what on this likely result has two major points:
Will “severe outcome efficacy” hold up more than four months after the injection, as it has with other groups, or will teens display a normally obscured detrimental effect due to how highly protected they were to begin with? It is entirely possible that once “infection efficacy” dwindles, rates of severe illness accompanied by a positive test for SARS-CoV-2 will increase dramatically among the young, owing to the removal of innate immunity protection.5 If that turns out to be the case, the Covid vaccines will have essentially aged the immune system of young recipients. Yesterday’s report does nothing to rule out that outcome.
Who even benefits from 93% protection? There were only 173 (unvaccinated) teens hospitalized for “Covid-19” in the 19 examined pediatric hospitals through the entire four month study period, surely a surprise to any American news consumer who has been assured that the summer “surge” in pediatric cases was overwhelming hospitals left and right. If something like 1.5 million American teens - a very crude guess of how many were still unvaccinated in the coverage area and time-frame6 - would need to have been dosed to prevent 350 hospitalizations (assuming some teens in the same region ended up in nearby regular hospitals), that’s 1,499,650 pointless injections.
That in no way supports the assertion, “[The] Findings reinforce the importance of vaccination.” The findings reinforce the near-futility of doing so! Teen Covid vaccine recipients would have to win the lottery of bad luck to even potentially benefit from this experimental intervention into their immune system.
Put differently, the benchmark the Covid vaccines would have to meet in order to make sense for younger recipients is not 90%, it is 9,000%. Meeting this benchmark is intrinsically impossible, and there is really no point studying the results to begin with.
The question of “How can I keep my child safe from Covid-19” should be as interesting to parents as the question, “How can I prevent my child from getting her arm stuck in the gap beneath a grocery shopping cart handle at exactly the moment a frenzied moose appears?” Kids do not need the Covid vaccine.
Onto discussion of “case control,” and the flaws in this particular example. From the study, brackets in original converted to parenthesis:
During June 1–September 30, 2021, among 572 eligible patients, 108 were excluded, including 56 who were partially vaccinated or who completed vaccination 0–13 days before illness onset [!], [30?] who were hospitalized >14 days after illness onset, 14 case-patients who received a positive SARS-CoV-2 test result but were admitted for non–COVID-19 reasons, and 18 who were excluded for other reasons [including 14 cases of “onset of COVID-19–like illness after admission”!]. The 464 patients in the final analysis comprised 179 case-patients and 285 controls (122 (43%) test-negative and 163 (57%) syndrome-negative).
Per the case control design, the “control” group used to derive efficacy rates is other kids in the hospital for other stuff.
Now, you might be saying to yourself, “How could that actually work?” Or, “That sounds profoundly, incredibly stupid.” Or, you might have already donned a tattered shawl and started throwing rotten produce at your flatscreen display while hissing and cursing. I wouldn’t blame you. But, I will explain how the “case control” method works: Vaccine developers and the researchers who carry water for them agree to speak and behave as if it works, no matter if it renders their entire occupation into a ludicrous fiction, akin to a round-the-clock game of fort where the cat is narrativized as an enemy raider whenever it is kind enough to walk by - and therefor it works.
The fiction, however, does occasionally almost overlap into practical use. For example, to a public health researcher, the fact that sick people are not representative of most of the population - and thus, what makes sick people likely to have a given problem is not the same as what makes a healthy person likely to have it - might seem like something that adds, rather than diminishes from the value of a revealed association. Let’s use the trope of red meat - say it turns out, that among the elderly or obese, eating red meat = higher chance of “bad.” Well, it wouldn’t be helpful to those groups to look at overall association of red meat with “bad,” would it? So, it’s taken as totally appropriate to build selection bias for ill health into study designs. After all, it’s not like the results will go on to be used to sell drugs to people who don’t need them. No, that would never happen.
Vaccines, in particular - and flu vaccines most of all - have become a popular subject for this built-in-bias type of “study” design. Here researchers have exerted themselves by practicing a modicum of self-policing. To allegedly diminish the selection bias of “unhealthy people,” the “control” population is whoever shows up to a place to get a flu test, and walks out with a negative.8 The rationale for why this leads to more accurate results is that…
Well.
The rationale is…
No. Hold on. I’ve almost got it.
Oh yes! The rationale is that vaccine researchers agree to speak and behave as if there is a rationale, and this is what they happened to agree to say that rationale is.
Note, however, that despite claims to the contrary,9 yesterday’s study doesn’t meet the standard practice for evaluating vaccine efficacy with “test-negative” studies, fraudulent as that practice is: It does not look at walk-in test results. Test-negative studies are not, in practice, used where there is either asymptomatic screening for the virus under scrutiny, nor where hospitalization is one of the selection criteria. The MMWR paper simultaneously features both. These two competing factors might in fact cancel out, but there’s no way to know. The sticking point is that the design of yesterday’s report flaunts epidemiological custom by using a (semi-nonfunctional) “infection efficacy” methodology to measure severe outcome efficacy.
This review of the history of the platform is hopefully sufficient to convey the significance of the division between the “122 test-negative controls” and the “163 syndrome-negative controls” in the report: It is the arbitrary application of influenza vaccine research customs to a context in which such customs are even more meaningless than usual. The fact that 122 children who were hospitalized had symptoms similar to “Covid-19” and tested negative is not a relevant determination of anything. The fact that 163 children were hospitalized without such symptoms (almost all were still screened for SARS-CoV-2 anyway10) is not a relevant determination of anything. The authors are just off in la-la land. There is no reason at all that vaccination rates among either the “tested” or un-“tested” hospitalized teens should be used to estimate vaccine efficacy! The whole exercise is pure, farcical nonsense.
In fact, what if you were simply to guestimate the average rate of vaccination in the entire 12-18 population during the study period, and use that as a “control” instead? In my “crude guess” of the unvaccinated rate earlier on, I merely stuck my thumb at the bulge in the CDC’s Covid-vaccination tracker curve that I felt looked like the “average” for both orange groups.11 It came out at 30%.
Among the “control” groups in the MMWR report, what was the “fully vaccinated” rate? 32.6%.
It took 29 authors and a surrogate investigative team to screw in this lightbulb, folks.12
Here, we could get into the weeds of why the authors screened out so many patients - “108 were excluded, including 56 who were partially vaccinated or who completed vaccination 0–13 days before illness onset” - or why not all hospitals that are part of the “Overcoming COVID-19 Investigators” network were included in this study.13 The potential for cherry-picking is quite high; and the authors’ exclusion of results for the partially vaccinated seems incredibly reckless and deceptive.
But, again, the central result - “severe outcome efficacy in the first four months” - lands in reasonable territory anyway. And what is particularly galling about the method the authors used to concoct their “real-world” proof of this result is not so much that it might have involved cherry-picking, but that it potentially rewarded the Covid vaccines for harming children.
If, as in the baby-bird-helmet analogy, the Covid vaccines were driving kids to show up in the hospital, this would mathematically have boosted the vaccine efficacy derived by this study. And is such an outcome even so implausible? What to make, for example, of the result that the Covid-vaccinated were potentially more likely to have neurological or cardiac issues (though we can only guess based on the higher overall rates):
“Underlying condition” ceases to have the same meaning when evaluating the 163 “syndrome-negative control” children who were not hospitalized for a respiratory infection specifically. Many, if not most of the children in that “control” group would have been hospitalized for one of the given conditions (as well as some in the “case” group, despite the authors’ attempt to only use actual hospitalization for Covid-19 in selecting that group) - you don’t end up in a hospital because nothing is wrong with you.14 It could turn out that few of the 101 cardiovascular and neurologic “underlying issue” patients were fully Covid-vaccinated. Or, it could turn out that almost all of them were - and that their given issue only began after Covid-vaccination. We have no way of knowing - the authors do not even bother to sort their data between “test-negative control” (patients with symptoms similar to Covid-19) and “syndrome-negative” patients.
But isn’t it incredibly odd that not only does hospitalization for “Covid-19” rise in August and September while summer wave cases are at their peak, but hospitalization for other things rises while Covid-vaccination is rising?15
Is this the result of the authors’ attempt to match cases and “controls,” or does it correspond to an actual surge in general childhood illness? In their rush to support their pre-ordained conclusion, the authors don’t appear to say.
Which favors the argument that the MMWR’s interest is intrinsic, I suppose - what could be more near and dear to a professional “public health expert” than robbing human life of all vitality, especially where it is at its most vital. Nonetheless the timing is conspicuous. There were several reports concerning schools in the spring, but without the distinct weaponizability that has characterized the reports concerning schools and children after the summer. Additionally, the four relevant studies before yesterday’s are paired by posting date and topic, and three of these appear delayed, which may be an artifact of fishing through older available data as if by prompt:
For a tidy rebuke of the mask studies, see: Roche, Kevin. “CDC Brings the Garbage.” (2021, September 28.) The Healthy Skeptic.
For my review of the “Outbreak in Marin” study, see “Travels to Maskladesh.”
I myself almost reviewed the mask studies, but never put anything on the page. Nonetheless the empty draft was saved with the title, “Hear Me MMWR.”
From https://covid.cdc.gov/covid-data-tracker/#vaccination-demographics-trends, accessed October 20, 2021:
For the most recent view into the “infection efficacy window,” see the review of the UK real-time data for mid-October in “yUcK!”
For discussion of and proposed explanation for the apparent degradation of innate immunity caused by the Covid vaccines, see “Forever Spike.”
Here, I used 30 million for the amount of 12-18 year olds in the US. I simplistically used the figure 250 for the amount of pediatric hospitals in the US to derive a sampling modifier for the report’s 19 hospitals. I used .7 for the share of unvaccinated teens in the imaginary “cumulative exposure value” which corresponds with the 173 hospitalizations, though there is a selection bias (becoming infected selects for unvaccinated status). That leads to an approximation of “unvaccinated+exposed teens in the coverage area” of ~1.6 million:
(30,000,000x.7)x(19/250) = 1,596,000
See: Fukushima, W. Hirota, Y. (2017.) “Basic principles of test-negative design in evaluating influenza vaccine effectiveness.” Vaccine. Volume 35, Issue 36, 24 August 2017, Pages 4796-4800:
Rationale for applying the test-negative design in evaluating influenza VE
[…]Typically, [“test-negative case control”] study subjects are patients who visit medical institutions due to ILI [influenza-like illness] during the influenza season. Subjects with positive test results for influenza are classified into cases, while subjects with negative results are classified as controls, and then vaccination status during the season can be compared between cases and controls. As the subjects are likely to visit a medical institution soon after ILI onset, both cases and controls are considered to be similar in their health care-seeking behavior. [How? - remember, the whole goal is to discover that “cases” are less vaccinated, which means they “sought” less health care at some point!] Therefore, the test-negative design can minimize confounding by health care-seeking behavior in evaluating influenza VE [How?] even though the outcome measure is laboratory-confirmed influenza, which is expected to resolve the dilemma [that is encountered] in cohort studies. [How?!]
From the study:
“This study used a test-negative design, similar to other postauthorization VE [vaccine efficacy] evaluations”
No, it was not.
Presumably, if testing positive, to be lumped in with the 14 excluded patients who “received a positive SARS-CoV-2 test result but were admitted for non–COVID-19 reasons” and 17 excluded patients who either tested positive [3] or developed symptoms [14] “after” onset of illness or admission.
From the report:
An earlier study using the Overcoming network involved 58 hospitals. See: Son, M. et al. “Multisystem Inflammatory Syndrome in Children — Initial Therapy and Outcomes.” New England Journal of Medicine. 2021; 385:23-34. (Hospital affiliations are listed in the appendix pdf).
I once woke up in a hospital after a bike accident. Many broken bones. I suppose that counts as having a significant thing “wrong,” though nothing productive was actually done in the hospital.
Part of this must necessarily be due to a rise in cases matching symptoms for Covid-19 (such as from the rumored surge in RSV infections among youth), and part for cases which do not, since both the “test-negative” and “syndrome-negative” control groups outnumber total cases for June and July. Additionally, the authors’ exclusion of “partially vaccinated” patients from both the “test” and “control” groups likely selects out many June and July patients. This might skew non-excluded cases for both groups into the latter months (edit, October 21: it should skew subjects in the less Covid-vaccinated “case” category into earlier months and in the “control” category into later months).
/ Substack tracking scrambler link (?): Matlock: Season 3 DVD Set (amazon.com) /
After over a decade of reading nutrition studies and statin studies, I've learned to be very wary of statistical manipulation. So much of it is pure obfuscation. The words "relative risk" should set a flashing red lights in your mind. A useful number is NNT, number needed to treat. I really appreciate the baby bird analogy.
I checked the CDC covid database about a week ago and the deaths for kids by covid in the US across both 2020 and 2021 was something like 518. Over almost two years. One death for all October 2021 as of the time I read it. The data is hidden in plain sight and yet people are panicking because most people skim news headlines or watch TV and don't both to go read the data for themselves.