despite reams of data dating back nearly a year that schools and children are not significant or even relevant spread vectors for covid, historical data that school closures do not stop respiratory epidemics, and the data from the half of the US that is in class full time showing no risk from in person classes, school opening remains a hot issue and a political football.
worse, the demands to open schools are perfectly calculated to ensure that schools will immediately close once more. it all seems innocent and reasonable enough, but it is not. it is, in fact, an inescapable cycle of pseudo-science and data-crime that runs like this:
demand mass, random testing of all school children as a requirement to return to school “because, safety.”
use PCR tests that are so oversensitive that they are literally contra-indicated for solo covid diagnosis in non-symptomatic individuals.
find lots of false and non-clinical positives among kids who are not, in fact, sick
report this rise in positives as raw data without adjusting for the huge spike in sample rate caused by the rise in child focused testing.
use this misleading raw data to claim that “cases are up in kids!” and “new variants affect kids disproportionately!”
use that false fear narrative to close schools.
never, ever admit that it was just a rise in testing and a change in who was tested that drove this.
repeat until no child in america can read.
it’s simple, insidious, and preys upon the credulous and the statistically illiterate. unless you have dug deep into the data and science in 3 or 4 different places, this crooked game of 3 card monte looks compelling. but it’s not. it’s just statistical legerdemain and they are about to use it to try to close puerto rico schools. again.
this one hits me close to home. the children of puerto rico have been savaged by over-zealous covid rules. they have had perhaps 3 weeks of in-person instruction over the last 12 months. the effects of this on scholastic learning have been shown over and over all over the world and they are DIRE even in the BEST CASE scenarios. puerto rico is not a best case scenario. computer ownership and home internet access are low. this is putting a terrible burden upon our most vulnerable and is going to have effects on scholastic achievement that may endure a lifetime.
to pay such a price would require dire danger and if one seeks to pander to powerful teacher’s unions without admitting one is doing so, then a dire crisis must be ginned up. and oh, how they are ginning up a false crisis. all public proclamation on PR is “cases are spiking, we need to close schools!” but cases are not spiking. raw, reported positive tests are spiking. looked at alone, this looks frightening, but as i have been saying for nearly a year now:
“reporting case data without reference to testing level is tantamount to lying.”
it’s data-crime. ignoring sample rate is a statistical error so basic that no credible health official could possibly make it or fail to correct it if they did. you’d fail your first quiz in “statistics for poets” making an error like this, and for good reason. failing to correct for sample rate gives the whole pandemic in the US the wrong slope. the red line is “reported” the blue line is what really happened in terms of prevalence.
i work though this in some detail (and address the baseless claims about “variants” being more dangerous and school transmission) HERE for those inclined to dig deeper.
so what does this mean for puerto rico?
pretty much everything. we can see it readily on this graphic taken from the NYT covid reporting DATASET.
cases are up 482% from the february 24th low and are now at approximately the levels from december (908 now vs 928 in dec using 7 day moving avg). this data looks frightening until one contextualizes it. the rise appears meteoric. but, when one looks at the rise in testing this narrative unravels.
since the february low testing has risen by 3,473%. that is not a typo. testing is up 35X from feb vs “cases” being up only 5X.
% positive has dropped from an utterly implausible 69.6% (certainly an artifact of extremely low testing levels of 224 a day) to 8.5% (still likely well below the minimum viable sample rate for strong accuracy)
the rise in “cases” perfectly aligns with the rise in testing. this data is not consistent with a rise in covid prevalence. it appears to be being driven entirely (and then some) by a rise in sample rate.
looked at longer term, we can see that the peak of the 7 day moving avg for cases (useful because cases get reported in boluses and weekends drop a lot then monday-tuesday spike) was on dec 22. it was 928. on that day, only 1,101 tests were performed. (again, using 7DMA for both) that implies an utterly implausible positivity number and i suspect either bad data or lag because PR is slow to report tests. so let’s use the test peak instead of 2565 on dec 11. if we adjust current data for that sample rate, the 908 raw figure drops to 290 and is now 69% lower than the december 22 figure.
this is always a bit approximate (though far better than not adjusting it). perhaps this is off by 10 or 20% (though i doubt it as i already slanted this calculation heavily to favor the current case count) but even if it were, it makes it clear that once more, using raw data is making a decline look like a rise because it cannot tell a rise in testing from a rise in disease.
this is is a fundamental and serious error and it provides absolutely no basis for the curtailment of schools. but this error gets much worse and much more insidious because of the targets of all the new testing: children.
there is an old joke about a drunk looking for his keys under a streetlight. when asked if that’s where he lost them, he replies, “no, but this is where the light is.” we all see the fallacy and that’s what makes it funny. this ceases to be funny when it is used as a basis for public health and a pretext for policies harmful to children.
the big spike in testing on puerto rico was caused by our children returning to school. schools to which i have spoken stated that they were testing, either at random or systematically, 5% of their students daily. this was not targeted by risk or by symptoms. it was done on healthy children as a matter of policy and mandated as a condition to open.
the spike in testing on PR was the mass testing of kids going back to school and if they were being tested at a 5% rate vs the general population (well under 0.01% on puerto rico) you have populations that cannot possibly be compared. the sample rate for kids is (conservatively) 500X that of adults. (it’s likely closer to 5,000-50,000X) can it be any wonder that suddenly kids are a big part of the “new cases”? we’re not seeing where the disease is overall, we’re just seeing what’s under the light we chose to shine.
if you had a test with a 5% false/clinical false positive rate then even if no kid on this island had covid and 100% of adult tests were positive, kids would STILL look like they had 25 times the cases of adults.
even at 1% false positivity they would be 5X adults and that’s if there was literally not a single case of covid in any boriqua child. there is no test with a false clinical positive that low.
my friends, this is not a scary new variant affecting kids. it’s a rise in testing prevalence being mistaken for a rise in disease prevalence.
this is a massively “salted” dataset either being misunderstood or being misrepresented. i see no other interpretation consistent with the data. neither case makes the local health officials look trustworthy.
a word on PCR
PR is mostly using the saliva based roche PCR test on schoolkids.
large scale community testing of asymptomatic individuals is a gross misuse of PCR tests (as even their inventor kary mullis, who won the nobel prize for so doing would tell you were he still alive.)
they are being systematically run at much higher levels of amplification than could possibly be clinically relevant. tests are at 40 cycles of amplification. (40 doublings of source material) this is called a “cycle threshold” and abbreviated Ct.
that’s 100-1000X the amplification needed vs ~30-3 for the ragged edge of being able to culture live virus from a sample. if you cannot culture live virus and get it to replicate, it is neither infective nor contagious. you may find RNA, but the positive is non-clinical. this is why germany uses a 35 Ct and much of asia uses a 24.
it’s also being used in non-symptomatic patients (which it should not be). PCR is not suited to task there. even the NYT cottoned on to this back in august and pushed for a shift to rapid tests that look for antibodies (IgM/IgG), not trace genetic material. health departments did not listen.
using PCR indiscriminately upon an asymptomatic population never made sense. PCR positivity is not proof of clinical infection or infectivity. at a 40 Ct, you’re using 1 trillion X amplification of source RNA material and will pick up even trace and fragmentary RNA from virus long dead. it could be from months ago and emerge during routine cell necrosis.
the WHO has also (finally) agreed on this issue that so many of us have been harping on since this all began: MEMO HERE
this was, of course, known all along by those familiar with these sorts of diagnostic tools. it is not an accident that when they designed their vaccine trials the pharma companies and the FDA did not accept a positive PCR test as a truth standard nor did they, in most cases, test for covid in subjects that were not displaying symptoms. you can see the moderna study design here:
the FDA and drug companies knew full well that PCR testing of full populations of asymptomatic people would make a mess of the data. the question remains: “why do so many public health officials STILL not know this?” (or why, if they do know this, do they persist in doing it?)
closing schools does not work anyway
we’ve known this for some time and this study from last summer was quite telling. SWEDEN STUDY . it compared students and staff in swedish and finnish schools. sweden was open, unmasked, and un-distanced. finland was closed. there was no difference in outcomes. this is because asymptomatic people do not spread covid STUDY DATA HERE and neither do children to any great degree.
neither do lockdowns or masking do anything to stop covid spread or hospitalization. this was in all the global pandemic guidelines as recently as last year. SOURCE
this longstanding data driven guidance was ignored to great cost and no benefit.
data tracking social mobility using google mobility data and comparing it to covid outcomes (similar to what i did last april and onward) validated these results. it was recently published in NATURE.
we see this again if we normalize hospital census for population size (shown per million here). locked down and masked NY, NJ, MI etc look exactly like open and in person schooling states like FL and TX. since covid went endemic, you’d be hard pressed to tell them apart without labels.
even german courts are now realizing this (having weighted the evidence) and are striking down closures and masking for kids in school.
come on guys
the idea that in person schooling is a risk to children and communities simply has no basis in fact and the data, when looked at with even rudimentary rigor, has steadfastly refused to bear out or validate such claims.
this is not “following the science,” it is abrogating it entirely in favor of superstition and partisan politics.
this is not “playing it safe” it’s tossing the basic concept of cost/benefit aside to embrace panicked, emotional decision making.
is this what we wish to teach our children? is this what we wish to allow to be inflicted upon them for no good reason? is this what you wish to be remembered for?
it’s time for the adults in this room to stand up and act like adults. the data here will only become clearer and clearer and those who are on the wrong side of history and epidemiology and refused to admit it will be remembered and remembered with disgust. and they will have earned it.