COVID-19: More Deaths in Vaxxed Persons? Simpson's Paradox

 December 8, 2021

I will fix this formatting mess as soon as the AAPS posts this article--sorry.

 

 COVID-19: More Deaths in Vaxxed Persons?

 

         Are you confused about claims about death rates in vaxxed vs. unvaxxed?

        The chart that I posted previously, which showed that all-cause mortality is twice as high in persons receiving the COVID-19 vaccine, got more than a million hits. And a storm of criticism. As former New York Times reporter and science writer Alex Berenson, who first posted the chart, acknowledged, he (and I) fell into a common statistical trap called Simpson’s Paradox: Conclusions derived from the whole group may be the opposite of those reached from a more detailed analysis. That’s because of “confounders”—overlooked variables that can influence both supposed cause and supposed effect.

        Are there differences between 10-year-olds and 59-year-olds who die, other than COVID vaccination status? Huge differences. So why the age range of 10–59 in the table that we used? The Office of National Statistics (ONS), an official agency of government in the UK, used it. Ask them why.

        The chart attracted so much interest and fury because it contradicts the narrative that COVID vaccination protects you from death. A recently published NPR article “Pro-Trump counties now have far higher COVID death rates. Misinformation is to blame,” which is even quoted by Breitbart, proclaims that residents of counties that voted heavily for Trump were 2.7 times as likely to die of COVID-19 (NOT 2.7 times more likely to die of something) than those that went for Biden. Trump voters are more likely to be unvaccinated, they say. These authors assume that “correlation reflects causation”—a blatant contradiction of the most basic tenet of statistics. They make no attempt to correct for confounders or bias, errors beyond the mere Simpson Paradox. And yet there is no avalanche of criticism this time.

(I have had to omit a section because the computer won't let me publish the graph.)

        

        So, what can we conclude from the statistics? First, proceed with caution. We need teams of expert, independent statisticians with access to ALL the raw data. As one expert put it: The validity of the statistics depends on the validity of the data.

        Is death “from” (which brings greater pay to hospital) or “with” COVID (say a positive test in a car crash victim)? Is the patient “vaccinated” (had last shot more than 2 weeks before diagnosis) or “unvaccinated” (hasn’t received all the shots more than 2 weeks prior to diagnosis and before the expiration date)? Without knowing these things, we cannot reach valid conclusions.

        Moreover, it is impossible to know long-term effects now—without having a time machine.

        Important questions all should ask:

·         How many patients, especially children, are experiencing heart damage? How many have been checked, as with troponin levels or the PULS test, for asymptomatic myocarditis? 

·         Have fertility clinics been surveyed? Is the statement that vaccinated sperm donors are being rejected because “their swimmers aren’t swimming” a mere unsupported rumor or anecdote? The CDC says there is “no evidence “ for a problem, citing a study of only 45 men. 

·         Are pathologists seeing more suddenly appearing new or recurrent cancers in vaccinated persons, or is this just an “outlier” observation by a few patients or pathologists? 

·         How frequent is vaccine-induced immune thrombotic thrombocytopenia (VITT), which can cause bleeding and clotting, with strokes and other serious problems, and how should we deal with it?

 

        Each person has a fundamental human right to decline an injection and to receive an honest account of known and unknown risks.

 

Jane Orient, M.D., Executive Director, Association of American Physicians and Surgeons, jane@aapsonline.org