A Counterpoint to Matthew Crawford's VAERS analysis
While also acknowledging the overall merit of his efforts
Matthew Crawford, who writes
, is a critical thinker who has been an unquestionable champion of data analysis during COVID. But I have some beef with one of his graphic presentations of the VAERS system found in this Substack.He mentions how VAERS reports could “explode” while background events “barely budge” — and his graphic seems to show that potentiality but also seems to overly show it.
As a talking point, here is how I would present the VAERS system graphically:
As can be seen, according to published estimates, VAERS reports capture somewhere from 2% to 72% of the background events. If they captured 2% one year, it is theoretically possible for them to capture 72% the next.
The 72% Threshold
To find the upper limit on captures rates of passive surveillance requires the combination of two things:
An extremely adverse event (such as lifelong paralysis)
An extremely-obvious tie of that particular event to the vaccine in question
Because paralyzing polio from live oral polio vaccine meets the two conditions required for estimating the theoretic maximum capture rate, the upper bound on the CDC estimation for the capture rate of that particular event is used here.
But notice what that upper limit does to what’s possible regarding the range of increases that you’ll ever see in VAERS reports. It makes for an upper limit on the possible increase or decrease (difference of any two years) in reporting rates of 36-fold — from a plausible minimum 2% up to a theoretic maximum of 72%.
There is only “room” for VAERS reports to rise by about 36-fold, though the probability that they ever will rise that much is miniscule.
VAERS at the Outer Limits
The published estimates reveal that you could not get, say, a 50-fold increase in VAERS reporting rates without ALSO witnessing a rise in background events. There isn’t enough “slack” in the VAERS system to support a 50-fold rise without a corresponding increase in background events.
The probability distribution on rate increases decreases with each new increase in rate, small changes are highly probable, large changes are not. There are good reasons that reporting rates have some stability, rather than full latitude to freely move — and one of them is the human propensity-to-report.
If human behavior were completely malleable — making humans full “trainable” like dogs are — then VAERS reporting rates could move freely, perhaps even asymptotically approaching 100% (though never actually reaching 100%).
But, as economists such as Mises have pointed out, that’s not the case. Human action is principled action, and it only changes for very definite reasons. Besides human behavior being mostly stable, other procedural aspects will also limit by how much a human system like VAERS can ever move from one year to the next.
Reference
[72% as the theoretic maximum amount of capture in passive surveillance of vaccine-associated background events (extremely adverse outcome; extremely related to the vaccine)] — CDC. Reporting sensitivity of VAPP in VAERS. Surveillance for Safety After Immunization: Vaccine Adverse Event Reporting System (VAERS) --- United States, 1991--2001. https://www.cdc.gov/mmwr/preview/mmwrhtml/ss5201a1.htm
[The plausible range of adverse event (AE) capture rates goes from a low of 2% up to a high of 18% (the interquartile range of the 37 drug studies under systematic review)] — Hazell L, Shakir SA. Under-reporting of adverse drug reactions : a systematic review. Drug Saf. 2006;29(5):385-96. doi: 10.2165/00002018-200629050-00003. PMID: 16689555. https://pubmed.ncbi.nlm.nih.gov/16689555/