@buzzkill@diaspora.schoenf.de So, I got my peak dates off a bit, I was running off memory and thinking of some northern/midwest states, e.g., Wisconsin, peaking 21 November. Though the underlying story still holds.

It's really helpful to distinguish reporting artefacts from ground truth. And healthcare data are full of these artefacts. I used to work with the stuff and picked up some sense for this at the time.

Using US daily new cases data, what you'll see around the holidays is a characteristic dip, almost always followed by a bit of a rise. For 2020, the federal public holidays occurring during the COVID-19 pandemic were Memorial day (25 May), Independence day (4 July), Labour Day (7 September), Thanksgiving Day (27 November), and Christmas Day (25 December). Coincident with each of those, using the 7-day moving average smoothing at Worldometers (linked above), you'll see a characteristic dip and bump for each of these, though for 4th of July, there seems to be a rise before the date. The notches for Thanksgiving and Christmas are most notable.

Smoothing out those bumps is a bit dicey, but it's clear that already as of Thanksgiving or shortly after the increase in new cases was slowing. If there had been a sudden surge of cases during the holidays, you'd have seen a persistent acceleration as those cases were reported and as subsequent infections developed because of them.

(These trends are even more evident in the deaths data. Coroners apparently don't like working holidays.)

Both the dips and the rises following were very likely largely due to testing availability or inclination of people to be tested, and of lags in reporting those tests. So a slump in people being tested during the holiday followed by a surge afterwards gives this characteristic rotated 'S' shape: first a dip, then a rise.

I can remember the first time I was working with healthcare data, plotting what I was seeing (patient visits), and looking for a smoothing interval that would make the pattern clear. As I increased days from 2 to 7, suddenly the signal shone through, and the first thing I noticed was that holidays were dramatically evident in the pattern. So I had a pretty good idea of what to look for here.

**If you want to see what an event-driven surge looks like, pull up the plots for North Dakota and South Dakota. The 2020 Sturgis Motorcycle Rally was held in Sturgis, SD, from August 7--16, with some 460,000 vehicles attending (some may have had multiple riders).

New cases started tracking upwards about August 24, about 17 days after the start of the rally, with a first peak on September 2, then a second and much larger rise running September 14 -- November 16. In North Dakota, the rise began August 17, the day the rally ended, and continued through November 17. My suspicion is that preventative measures noncompliance during the event, combined with the onset of cooler weather and more indoor activities, as well as schools, all compounded. But the result was a three month massive tenfold rise in new daily cases.

And that's precisely what we did not see following Thanksgiving and Christmas.

Note in particular that ND and SD new daily cases were falling nearly continuously through the holiday season.

I'm not saying that no infections occurred over the holidays. But as compared to the Sturgis event, it was exceedingly minimal, and didn't interrupt what was already a peaking and declining trend.

"Smoothing out those bumps is a bit dicey, but it’s clear that already as of Thanksgiving or shortly after the increase in new cases was slowing. If there had been a sudden surge of cases during the holidays, you’d have seen a persistent acceleration as those cases were reported and as subsequent infections developed because of them."

No.

First of all, Thanksgiving was Thursday, November 26th. Not the 27th.

Second of all, US daily cases reported Oct 30th was the highest ever reported to date : 99,615

Nov 4 2020 110855 we hit 100,000 for the first time. We did not go below this again until FEBRUARY
Nov 10 2020 141357 we went over 130,000 for the first time and did not go below it again until JANUARY 31st
Nov 16 2020 154602
Nov 18 2020 168584
Nov 24 2020 173775

There was a steady, humunguous increase ALL THROUGH NOVEMBER. THERE WAS NO SLOWING BEFORE THANKSGIVING AT ALL

November 10th was a record, then Nov 11, then Nov 12, then Nov 13, then Nov 19, then
Nov 20 2020 199608

from November 16th record of 150,000 cases we did not go BELOW 150,000 UNTIL November Nov 28
THERE WAS NO DECREASE

"(These trends are even more evident in the deaths data. Coroners apparently don’t like working holidays.)"

No.

There are usually peaks on MONDAYS because less official reporting is processed on WEEKENDS and HOLIDAYS. Also, there is RADICALLY LESS TESTING AVAILABLE on HOLIDAYS.

"Both the dips and the rises following were very likely largely due to testing availability or inclination of people to be tested, and of lags in reporting those tests. So a slump in people being tested during the holiday followed by a surge afterwards gives this characteristic rotated ‘S’ shape: first a dip, then a rise."

No. There WAS NO DIP BEFORE THANKSGIVING.

then Dec 2 2020 206882 we broke 200,000 new record
Dec 11 2020 252017 exactly two weeks after Thanksgiving, as predicted, we broke 250,000. A new record, and a MASSIVE spike

Numbers started to come down a LITTLE BIT - WHY? Not because it was BEFORE CHRISTMAS - because the THANKSGIVING SURGE ENDED

We still did NOT GO DOWN BELOW 183,405 before CHRISTMAS even once and for the most part STAYED ABOVE 200,000
Dec 22 2020 201923
Dec 23 2020 233118
Dec 24 2020 198820
Christmas was Dec 25 2020 137786

then TWO WEEKS AFTER, EXACTLY, AS PREDICTED, we hit a MASSIVE SPIKE and ALL TIME RECORD of

Jan 8 2021 315006 249593

It turns out to be quite difficult to find plots of patient care (hospital admissions, ER admissions, doctor visits, clinic visits) via online image search, though I tried on both DuckDuckGo and Google Image Search. (Google Search here, adjust for terms as listed above if you wish.)

I'd specifically excluded terms "covid" and "coronavirus" as the results are otherwise dominated by these terms, for some reason, and I'm trying to show the effect in other cases. Looking for noninfectious conditions (e.g., heart attack, stroke, trauma) conditions is also helpful as these avoid any possibility of increased incidence due to holiday-based infection. (Though other holiday-induced injuries are in fact fairly well documented.)

There is some documentation of weekend and holiday effects in patient outcomes, as in these Lancet (weekend) and AHA (holiday). I'm pointing at these in support of my claims above.

https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(17)30782-1/fulltext

https://www.ahajournals.org/doi/full/10.1161/01.cir.0000151424.02045.f7

From the latter, this image is about the closest I can come to a holiday-effect plot, though it's something of the inverse: cardiac morality spikes at Christmas and New Years, as shown:



As the AHA paper's conclusion notes, _the spikes are likely driven by reluctance to seek immediate care:

Our findings suggest that the Christmas/New Year’s holidays are a risk factor for cardiac and noncardiac mortality. There are multiple explanations for this association, including the possibility that holiday-induced delays in seeking treatment play a role in producing the twin holiday spikes.

(Emphasis added.)

Note too that following the spike, mortality rates fall briefly. Again, when looking at visits / admissions / care data, the obverse pattern (a fall in visits, followed by a rise) is clearly visible.

Again, from clinic-visit data I've seen in the past, though cannot share here (I haven't had access to it in decades), weekly, event-driven, and seasonal patterns in medical care are exceedingly obviously present when looking at 7-day smoothed admissions, treatment, or visit data.

In the case of COVID-19 data, not only is there the delay caused by patients not seeking out care, evident in diagnosis reports, but mortality data also typically dip on weekends and holidays, which I attribute as exceedingly likely due to practices in official confirmation, tabulation, recording, and reporting of deaths.

(Russia, curiously, has had very little weekly signal in its mortality data. That morality data has also been recently called very strongly into question. Whether the lack of weekly pattern and lack of trust in the data are related I'm not sure, but it is, as is much else, a rather interesting little element of this whole affair. Sweden's mortality data aren't as suspect, but have proved to take 2--3 weeks before all death reports seem to be finalised, making current determination of the state of COVID-19 in Sweden difficult to determine. I've remarked on this in numerous posts.)

"Please read what I wrote and compare against evidence. Not what you want to believe I wrte."

I quote you EXACTLY, and responded and disproved EXACTLY what you said. Right down to your misrepresenting the date of thanksgiving.

Perhaps YOU should read what I wrote and actually LOOK AT THE EVIDENCE and "not what you want to believe"

"There is some documentation of weekend and holiday effects in patient outcomes, as in these Lancet (weekend) and AHA (holiday). I’m pointing at these in support of my claims above."

The articles you linked to are for cardiac arrest. that has NOTHING to do with COVID or ANY EPIDEMIC OR CONTAGIOUS DISEASE.

THE TOPIC IS THE INCREASED RATE OF SPREAD OF INFECTION DURING HOLIDAY OF EPIDEMICS. YOU ARGUED THERE WAS NO IMPACT

CARDIAC ARREST AT HOLIDAYS HAS NO REFLECTING ON THE SPREAD OF INFECTIONS!!!!!!!!!!!!!!!!

"From the latter, this image is about the closest I can come to a holiday-effect plot, though it’s something of the inverse: cardiac morality spikes at Christmas and New Years, as shown:"

CARDIAC MORTALITY HAS NOTHING TO DO WITH CONTAGION RATES!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

"Again, from clinic-visit data I’ve seen in the past, though cannot share here (I haven’t had access to it in decades), weekly, event-driven, and seasonal patterns in medical care are exceedingly obviously present when looking at 7-day smoothed admissions, treatment, or visit data."

CONTAGION RATES HAVE NOTHING TO DO WITH ADMISSIONS!!!!!!!!!!!!!!!!!!!!! OR TREATMENT!!!!!!!!!

CONTAGION RATES ARE FROM POSITIVE TESTING!!!!!!!!!!!!!!!!!!!!!!!!!

"In the case of COVID-19 data, not only is there the delay caused by patients not seeking out care, evident in diagnosis reports, but mortality data also typically dip on weekends and holidays, which I attribute as exceedingly likely due to practices in official confirmation, tabulation, recording, and reporting of deaths."

I have already TOLD YOU that mortality rates with covid always SPIKE ON MONDAYS from a lack of PAPERWORK PROCESSING in OFFICIAL REPORTING ON WEEKENDS.

AGAIN, YOU CLAIM THERE WAS A DIP BEFORE ALL HOLIDAYS. THANKSGIVING WAS THE LARGEST TRAVEL AND HOLIDAY OF ITS TIME IN THE WHOLE PANDEMIC AND THERE WAS NO DIP BEFORE IT. THERE ALSO WAS NO 'DIP" BEFORE CHRISTMAS. THERE WERE HOWEVER, AS ORGINIALLY STATED, THE LARGEST SPIKES EVER TWO WEEKS AFTER THEM, as PREDICTED AND EXPECTED. YOU ARE COMPLETELY FULL OF SHIT, and SPREADING MORE SHIT. YOU CLAIM THERE WAS A DIP AND NO SPIKES. I COPIED AND PASTED THE EXACT NUMBERS. THERE WAS NO DIP. THERE WAS A HUGE SPIKE.

"USA #1…

…in being fucking self-absorbed barking ignorant."

I couldn't have said it better myself. Quod erat demonstrandum.