Weekend COVID-19 Briefing


Top news, reports and insights for today:

  1. Curated headline summaries for Saturday/Sunday:
  • The U.S. Centers for Disease Control and Prevention (CDC) on Saturday reported more than 8 million confirmed COVID-19 cases and 217,918 deaths (Reuters)
  • A small randomized trial shows that the Chinese coronavirus vaccine is safe and does elicit an antibody response. This is a key preliminary stepping stone to showing that the vaccine prevents COVID-19 in the population (Lancet)
  • Ukrainian fitness guru, Instagram influencer, who thought COVID-19 was a hoax, dies of the virus (Daily Mail)
  • In an open letter, more than 1,000 current and former CDC officers criticize U.S. COVID-19 response, decried political interference and called for the CDC to have a larger role (Wall Street Journal)
  • CNN anchor Jake Tapper says the White House refused his invitation to allow coronavirus experts including Anthony Fauci and Robert Redfield to appear on his show (Axios)
  • Rural Midwest hospitals struggling to handle COVID-19 surge (ABC news)
  • Pharmaceutical giant Pfizer delivers “final blow” to President Trump’s hope for a pre-election vaccine. The company, considered a frontrunner in the vaccine race, says it won’t seek an emergency use authorization (EUA) for it’s candidate vaccine before late November (Politico)
  1. U.S. daily COVID-19 cases return to near explosive growth. Deaths stay steady
     The end of this week saw a continuation of a trend toward sharp escalation in daily U.S. COVID-19 cases. Thursday and Friday numbers spiked over 66,000, a number not seen since the end of July (Figure A). The rate of increase is not as steep as the explosive growth seen from June 22 to July 15, but it is steeper than it has been since then. Record high daily case totals were broken on Thursday in Montana (723), Michigan (2,030) and Ohio (2,069) as well as on Friday in Idaho (1,094), New Mexico (812), Wyoming (248), Indiana (2,283), Minnesota (2,280), North Dakota (864), Nebraska (1,286), Wisconsin (3,861) and West Virginia (358). In just 2 days, 12 states recorded more cases than any other day in the outbreak making this one of the most disconcerting days thus far. That brings the total number of states with 100,000 lab-confirmed cases to 26, half of all states. Of those, six states now have at least a quarter of a million cases and three, California (864,000+), Florida (751,000+) and Texas (822,000+) now exceed a three quarters of a million. To put this in perspective, California and Texas would now rank in the top ten most impacted countries with more cases than the entire United Kingdom and about as many cases as Peru, Mexico and France.
     Thankfully, there is little indication yet that the trend in cases over the last month is matched by a trend toward rising deaths (Figure B). The 7-day moving average has been steady between 600-800 a week since September. Based on what we know about the time lag between cases and deaths, most experts expect deaths to begin to rise in the coming days or weeks. In my opinion, that expectation may be tempered by continual improvement in medical management strategies that have gradually lowered the lethality of the disease. However, a major caveat is that there will be an inflection point as the cases rise where the capacity of the hospital system to maintain the highest standard of care will be jeopardized. We are already seeing signs of that in states like South Dakota and Wisconsin (see headline 6). Death rates will soar if thresholds are exceeded in the coming months as most experts now predict.
     The bottom line: Daily cases are now spiking at a rate not seen since the summer crush. The U.S. has now passed 8 million cases, the most of any nation and more than a quarter of the 40 million cases world-wide. The incidence of COVID-19 is now 5 times higher in this country (25,171 cases per 1 million population) than for the entire planet. In my view, this is not a second wave and not a third wave, but another wavelet within the tidal undulation of the first epidemic wave of COVID-19 in the United States. The prognosis is not good heading into colder temperatures, lowered humidity, more time spent indoors and continuing social and economic pressure to return to “normal”. The march of suffering and death continues apace and the U.S. still refuses to do what must be done to quell this viral firestorm.
Figure A
Figure B
  1. Why do U.S. COVID-19 death totals differ wildly across sites tracking the pandemic? The disease detective goes to work.
      Above I reported the daily epidemiologic curves for U.S. cases and deaths as I always do using data compiled by Wikipedia. No data source is perfect. There is, I have argued, a big advantage to sticking with the same data source over time despite its limitations if the primary purpose of surveillance is to track patterns of change over time. That having been said, I am increasingly uncomfortable with my choice to stick with Wikipedia in light of growing discrepancies between this source and most of the high-visibility alternatives. Figure C shows you the current total U.S. COVID-19 deaths from my source (Wikipedia, medical cases) and 7 others. I won’t call these independent data sources because they are all tapping into the same underlying data streams coming from official state reporting agencies. I’ll call them data compilers instead. In theory, given the common data sources, you would think the grand totals should very similar. The first thing to say about Figure C is that no two of these compilers arrive at the same total as of October 16. The highest is from WORLDOMETER and the lowest is from Wikipedia. The average (median in this case) is just over 218,200 as of today (red line). Worldometer is reporting 17,952 more U.S. deaths than Wikipedia. That’s a big difference. As disease (and data) detectives, let’s dig into these numbers and see if we can figure out why.
     What we know based on detailed analysis of excess all-cause mortality is that about 68% of the actual COVID-19 deaths are being correctly attributed to the coronavirus. Based on that, one could assume that the compiler with the largest number is likely to be closest to correct since it’s much less likely that deaths are being incorrectly attributed to COVID-19 than they are being missed. So for this exercise, we will focus on the high vs. low estimate.
     After digging into the details, I learned that the Worldometer (WO after) is highest in part because they don’t rely just on numbers compiled from state agencies, but include also deaths among “U.S. military” abroad (100), and a category labeled “Veterans affairs” (3,711), Federal prisons (128) and Navajo Nation (571) along with three deaths from the Grand Princess ship. That accounts for about 4,000 of the deaths not reported in Wikipedia.
     Next, what states are responsible for the discrepancy? Figure D is my picture after a careful comparison between these two compiler. Aha! We now know that the two sites are in perfect agreement in about one third of states (N=15). About half of states are plus or minus five (N=27). The bigger discrepancies come mostly from a few select states. Wikipedia has more deaths in ten states (only 2 are more than 10 more), and most of these are because the Wikipedia numbers include the deaths from Saturday. Of the 23 states where the WO count is higher, 13 of them are more than 100 apart. The real story is the two numbers that really pop out: New Jersey (+1,917) and New York (+7,814). Three quarters of the “excess” COVID-19 deaths in the WO data are in these two states with 62 percent just from New York. Why are there almost 8 thousand deaths in WO that aren’t in Wikipedia? The difference arises from where the compiler gets the information. Wikipedia uses numbers from the New York State Department of Health COVID19 tracker while WO uses Johns Hopkins numbers. The Hopkins dashboard says they get their New York numbers from the Governor’s coronavirus newsroom. There is no hint there as to why the two data sources differ so markedly. This disease detective found an important clue on the WO’s page describing the data where there is a very helpful and instructive footnote about U.S. data. This footnote announces a change in the death definition on April 14 to include “probable deaths” following updated CDC guidelines. Have a look at Figure E which is a screen capture of the part of that memo that relates to reporting in New York. The plot thickens!!! It tells us that the New York Department of Health has, for unknown reasons, not yet folded in “probable deaths”, but WO has combined them. A good chunk of the data gap arises from “probable” deaths, mostly in New York City that occurred during the high-mortality period in March and April. It looks like the New York state DOH numbers never included those that could never be directly linked to a positive test. The Hopkins dashboard, along with WO, have included those since mid-April.
     What have we learned: There are big differences in mortality numbers across data compiler sites. Wikipedia and the COVID Tracking Project are below average due, because they are more “conservative” and do not count all the “probable” deaths, relying more tightly on official state departments of health. WORLDOMETER reports the highest death totals for 2 reasons: 1) they channel the higher Johns Hopkins numbers that include more “probable” deaths, and b) they include deaths counted outside official state agencies (e.g., military, and Navajo nation). It is very hard to get to the bottom of these differences. Each data compiler should provide a clear and organized statement about their reporting policies, where the source data comes from. The lion’s share of the discrepancy comes from accounting differences in New York and New Jersey. Even the highest estimated deaths from the WO site represent about 70% of the true death toll from this disease.
Figure C
Figure D
Figure E: Screen grab from: https://www.worldometers.info/coronavirus/us-data/
  1. Quirky Qorner: The ‘coronavirus crush’ is a thing. Why a pandemic is the perfect moment for our infatuation with the unattainable
     What is being cooped up doing to us? Lots we all suspect. Some effects are obvious and others are more subtle and nuanced. I have been reading stories lately about the ‘coronavirus crush’. Apparently, being cooped up all day with just our cats and our thoughts has led many of us to become infatuated with someone not in our bubble. Once again, Vox comes to the rescue with this juicy piece by Terry Nguyen from October 16. Here are some highlights that struck my fancy:
    1. 2020 has been a “big year for yearning” that has translated into a deep, unattainable longing for all sorts of people, places and experiences. I hear that!
    2. Pandemic conditions make crushes lower risk for those already in committed relationships.
    3. Romantic yearning is a convenient and pleasant form of escapism and distraction.
    4. Most of us are craving the pleasant surge of oxytocin and dopamine that a crush can bring forth.
    5. Being on zoom all the time can make us feel like high school kids.
    6. With real relationships now off-line and verboten, our erotic energies are free to play in the realm of fantasy.
Print Friendly, PDF & Email

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.