Quoting Jonathan Del Arroz (jdelar...@gmail.com):

> I’ll just host a party for the unvaccinated instead, since studies are
> showing we spread the new variants less, anyway.  Mine should be safer!

Nope.  Your knowledge of epidemiology needs repair, Jon, as you've
been misinformed.

There _aren't_ "studies" showing that unvaxed spread Delta and
successors less.  That is wrong twice over:  Unvaxed are massively more
likely to get infected at all, and also massively more likely to remain
infectious for longer periods, lengthening potential exposure.

What actually came out wasn't a study, but rather a brief item in the
Aug. 6 edition of CDC's _CDC Morbidity and Mortality Weekly Report_
(MMWR) covering a Delta outbreak at Provincetown, tip of Cape Cod, after
Independence Day.  Social media glommed onto _one_ figure: among 469
PCR-confirmed COVID-19 infections found among seaside party-goers, 346
(74%) occurred in vaccinated persons.  Also, breakthrough (vaxed) infectees
during their period of infectiousness carried about the same viral load
as did the unvaxed infectees (the same "cycle threshold" aka Ct, during
operational cycles of the machine that does PCR testing to detect and
duplicate viral genetic material).  So, whoa!  Being vaxed means you
spread Delta _more_, right?

Nope.  This falls haplessly into Base Rate Fallacy, where you flub
estimating probability by ignoring context.

To understand that, imagine a slightly inattentive researcher studying
alcoholism in Dad's hometown of Kristiansund, population ~25,000.

Say (not real numbers here, but credible ones) results note that out of
550 identified alcoholics, 542, were blond.  Whoa, says he!  If you're
blond and resident of Kristiansund, you've 97% chance of alcoholism.  
We must search for genetic links.

Does this researcher get a Nobel?  Nope, he gets indulgent chuckles,
as he forgot Kristiansund (north of Bergen) has wall-to-wall blonds,
about as many blonds as herrings.  Call that, in round figures, 22,000. 
So, more like a 542 / 22000 * 100 = 2% chance, not 97%.

Getting back to Provincetown:  That 74% is just wildly devoid of base
rate context, in that the study of the outbreak made no estimate of how
many persons got exposed, nor even what percentage of those present
in total were vaxxed.  The 469 infectees were essentially self-reported
people, concerned enough to get PCR-tested.  Nobody got told "We need to
test you if you were in any Provincetown bar between July 10th and 18th":
Thus, there's an automatic "detection bias".  Second, as vaxed
population has increased, especially among people like Cape Cod
holidayers, you get the Kristiansund-blonds problem:  Vaxed persons
necessarily will then constitute a rising portion of cases.

For all of the above reasons plus others I don't even get into, _MMWR_ 
warned the info was "insufficient to draw conclusions about the
effectiveness of COVID-19 vaccines against SARS-CoV-2".  But social
media ignored that, and stumbled straight into the Base Rate Fallacy
trap.  https://www.snopes.com/fact-check/cdc-provincetown-covid-event/
https://www.snopes.com/ap/2021/07/30/study-vaccinated-people-can-carry-as-much-virus-as-others/

As a statistics guy and a Bayesian, that kind of thing makes me sad.

Also, with Delta as with prior variants, vaxed infectees clear virus faster.
https://www.cdc.gov/coronavirus/2019-ncov/variants/delta-variant.html
https://www.technologyreview.com/2021/08/09/1031292/questions-delta-variant-contagious-tranmission-vaccination/

-- 
Cheers,            "Orthodoxy is my doxy.  Heterodoxy is someone else's doxy."
Rick Moen               -- William Warburton, Bishop of Gloucester (1698-1779)
r...@linuxmafia.com
McQ!  (4x80)
_______________________________________________
Basfa mailing list
Basfa@lists.basfa.org
http://lists.basfa.org/listinfo.cgi/basfa-basfa.org

Reply via email to