Putting an end to the “holiday surge” meme – there was none

Summary

  • There were no holiday surges
  • The acceleration in new cases flipped in mid November before the holidays
  • On February 18, 2021, the UW’s Trevor Bedford reported Rt began declining in November (same as what I found)
  • California cases peaked before Christmas
  • Oregon Health Authority said there was no holiday surge in Oregon
  • Sacramento public health said there was no holiday surge in the Sacramento metro area
  • Update Mar 4, 2021): UK saw R-t start to decline as of Dec 21, before the holidays.
  • No one has a unified theory as to why cases have dropped rapidly. But it turns out it correlates perfectly with solar activity, therefore, I propose we turn off the sun (see chart below – too funny). /satire

Hospitalization Data

Here is the hospitalization data from The Covid Tracking Project:

Peak day of hospitalizations was Jan 6th. In Oregon, the time between diagnosis and hospitalization is a median of 12 days (OHA). The time between diagnosis and death is 14 days (OHA) – NOT the conventional wisdom 3-4 weeks. I can explain the CW discrepancy, if interested.

Note chart is 7-day moving average which time shifts peaks to the right of actual time.

This next chart is the 1st derivative of the epicurve above which highlights inflection points in the epicurve.

This “acceleration” curve peaked on Nov 14th and then reversed course. This is before Thanksgiving and well before Christmas. (Earlier spike is a day some one updated lagging hospital reports.) This is an important finding – the wave acceleration was dropping off in mid-November. (Trevor Bedford, below, appears to confirm this.)

The red line is a 7-day trend on a 7-day moving average of hospitalizations, hence, time shifted many days to the right. Ignore right most data point (due to how I store the data in the spreadsheet).

Oregon

Oregon Health Authority said in a statement there was no “holiday surge” in Oregon.

California

The situation in California was blamed on “holiday surge”. But this is not possible.

  • Cases began their climb in late October – not post Halloween, not post Thanksgiving, not post Christmas
  • Peak day of new cases was Dec 20th.
  • The 2nd and 3rd highest days were Dec 17 and Dec 18th.
  • The 7-day moving average peaked on Dec 22 (which is time shifted right)
  • Cases began rising in October and rose at a steady pace thereafter.
  • Sacramento metro public health has said there was no holiday surge
  • People seemed to have drawn correlations because it was kinda sorta close to Christmas but this was because California’s peak ran behind most other states’ peaks (none of which showed holiday surges).

(7-day average chart has time shifted the trend to the right)

Scientist Trevor Bedford

Trevor Bedford is a scientist at the University of Washington and posted this item on February 18th- Rt started to decline in Nov – which is the same as the 1st derivative decline I spotted, above.

Sturgis South Dakota Motorcycle Rally

This rally was held in early August and one forecast had this ½ million attendee event creating  over 260,000 additional cases of  Covid-19.

In the weeks that followed, South Dakota traced 124 cases back to Sturgis, and other states traced up to 290 cases back to Sturgis (because the patient had been there but could not rule out they picked it up elsewhere).

The forecast “surge” never happened.

Florida Superbowl

The Florida Superbowl was forecast to be a “superspreader” event that would cause a “surge” of new cases.

11 days later, the FL epicurve looks like this:

National CDC Epicurve

NOTE National curve is likely underreporting due to weather causing test access and reporting delays. But neither FL or National curves show a post superbowl “surge upon surge” at this time. Will be interesting to re-check in a week when more test results are available.

Travel Data

Airline passenger counts, by week. Slight peak before Thanksgiving, then drop during Dec, then a rise the week before Christmas, peaking on Dec 27th. Obviously, people travel by car also. However, the media focused on airline travel. After the end of December, there was a sharp drop in travel. Compared to overall national population (about 328 million people) this delta in air travel is tiny.

Apple Mobility Data

Some people point to car travel, saying mobility data explains the decline in new cases. But the decline – in most ALL countries – started in October just as cases were rising -and begin rising slighty in January, just as cases continued their sharp decline.

Is mobility data a proxy for close, in person, extended contact? We have no idea but many people assume it is and therefore, the peak in cases would have been in late summer. Nationally, the peak was in late December – and in fact, this data appears inversely correlated to cases.

Update Mar 4, 2021: UK Saw R-t decline beginning December 21, 2020

This was before the holidays, even in the U.K.

Conclusion

I cannot find a holiday surge.

I cannot find surges after major events forecast to cause “surge upon surge”.

Some epidemiologists say the current drop in cases is because a “holiday surge” ended. This is not supported by any data (see above). Their claim seems based on their pre-holiday assumption and forecasts of surges – and an unwillingness to acknowledge they were wrong. Some scientists now think the decline is due to herd immunity effects taking hold – but to acknowledge that is to further acknowledge that the field of epidemiology failed – since they said this is not possible until late 2021 when everyone has been vaccinated. So they will stick with their “surge upon surge” holiday travel lie.

I do not think anyone knows what is going on – I’ve probably seen ten different explanations and they are as likely accurate as my finding that an increase in solar sunspot activity in 2020 correlates perfectly to the Covid-19 pandemic!!!!

Note the blue spikes in April, July and then rising in October, November and December, and the fall in January. Clearly, this “correlation” explains everything and all we need to is stop the sun. Problem solved. Easy peasy.

Seriously, I think this analysis is as accurate as the hand waving the experts are doing. But we don’t know if Covid-19 pandemic waves cause solar sunspots or if solar sunspots cause Covid-19 pandemic waves. Heh.

/SATIRE

Afterword

  • OHA agreed with my finding on their death data reporting and now has a correctly displayed chart of Oregon deaths, by actual date of death. Peak day of deaths was December 9, not mid-January as their public health director had said at a January press conference. OHA has since backtracked.
  • I discovered that ICU bed census is an excellent proxy for “actual death date”. This data metric is real time whereas death data lags by up to months. Actual peak in deaths is usually within a day of peak ICU beds in use. This is a much better way to gauge deaths than the very lagging daily body counts.
  • Iowa and Ohio acknowledged the same “lagging” death report problem as was in Oregon.

Reminder – I am a brain injured idiot with no relevant experience. I do not make predictions: I make observations and ask really stupid questions. Covid-19 is a real disease, infecting too many people, and killing too many people, and is not a hoax.