NedNotes (not blog): COVIData Round-up 21aug20

 

Angelo Codevilla, "The Attempted COVID Coup of 2020",
Real Clear Politics, 13th August 2020. 

https://www.realclearpolitics.com/articles/2020/08/13/the_attempted_covid_coup_of_2020_143952.html

Purpose: analysis of this article as a proxy for the argument that recorded cases of COVID are understated by as much as ten times. The political side of the article lies beyond the scope of this essay.

B.L.U.F. (bottom-line, up-front): conceptually a plausible argument; data and conclusions presented by Real Clear Politics (R.C.P.) are internally inconsistent and conflict with experience.

Introduction
R.C.P. is a right-wing publication tied to the ‘alt-right’ brand of conservatism associated with conspiracy theories generally supportive of President Trump. Most alt-right publications are too biased for serious debate or analysis. 

Of these periodicals, R.C.P. is considered an intellectually honest, if biassed, publication. Even so, the article reviewed here asserts a conspiracy that is difficult to accept and will not be discussed in this round-up. Political biases are at play; this presentation seeks to avoid these topics.

The basic scenario underlying the conclusions is a reasonably sound one. When the coronavirus contagion emerged six months ago and accelerated into the COVID-19 pandemic, epidemiologists guesstimated global penetration rates of 40-70% over the following year. For the U.S. population, that implies a range of 130-235 million people contracting the virus.

Essence of the argument argued by the publication
Six months on, U.S. confirmed cases stand at 5.7 million, or about a 1.7% penetration rate. To reach the expected penetration rate over the next six months, infections would have to double every thirty days,  an aggressive scenario. The implied period for cases doubling, based on current growth rates in the bellwether states, detailed in Table-2, in the South and Southwest (S.&SW.), is roughly six weeks.

Not only do time horizons not match well, some of the ‘spike’ states in the S.&SW. are slowing the growth of confirmed cases. Reporting deficiencies understate penetration due to technical constraints of a recent switch in reporting protocols. The change exacerbates the other possible source (i.e., the thesis pursued by R.C.P.): that cases are vastly under-counted due to a large majority being asymptomatic and untested.

Evaluation of the content presented in the article
Using the same data from the Centres of Disease Control (C.D.C.) underlying the article, it is possible that undetected cases could elevate the penetration by nearly four times – not the ten times disseminated by the ‘New York Times’. At that point, cases would need to double every eight-to-ten weeks to reach the middle of 130-235 million range, a defensible scenario.

The underlying C.D.C. data, however, fail to substantiate the implicit claim of 10x. Taking the 59% excess of ‘recovered’ cases recorded of Worldometre (i.e., 3.1 million) over Johns Hopkins (2.0million) as asymptomatic cases, one can estimate that 1,150,000 asymptomatic cases detected through the testing to date. That corresponds to about one fifth of the confirmed cases. The worst case C.D.C. percentage of asymptomatic cases is 70%.

Depending upon how one slices and dices the data, applying the 70% asymptomatic assumption yields a level of undetected asymptomatic cases in a wide range of 4-15 million across the United States. That may sound like a lot, but these cases would represent less than 5% of the population. The current mortality run rate (i.e., number of deaths as a percentage of confirmed cases), or what the C.D.C. calls the “infection mortality ratio” equals 3.1% in the United States.

Table-1: C.D.C. Planning Assumptions

Adding in the maximum number of guesstimated undiagnosed cases of 15 million brings the mortality run / infection fatality rate down from its current 3.1% level down to 0.87%, or 3.3x above rate asserted in the article of 0.26%. The premise underlying the argument that the COVID epidemic in the United States does not add up.

The difficulty posed by the article is its 0.26% infection fatality / mortality run rate; in the C.D.C. table, the only perceptible combination of data to yield this lower fatality rate results from multiplying the base case (i.e., best estimate) values for percentage of cases that are asymptomatic (i.e., 40%) by the mortality run rate of 0.65%. Yet such a calculation makes no sense, especially in calculating an adjusted penetration rate.

By displaying its bias in favour of the Trump Administration’s contention that recent increases in the number of cases manifest more testing rather than spikes attendant to pre-mature re-openings by Texas, California, and Florida, the author weakens the internal consistency of the article’s thesis; in short, the presentation breaks down. 

Why? Because the two theses argue potentially incompatible points.

1. The Admin. is capturing a significant number of asymptomatic cases to create an appearance of accelerating penetration. VERSUS
2. The testing barely scratches the surface of asymptomatic cases run rampant.

Criticisms
While Real Clear Politics makes a sincere effort to argue that the coronavirus exerts a mortality burden similar to that of a bad flu, the evidence presented displays several deficiencies with this reasoning:

• inability to trace the data used in the argumentation to the C.D.C. report;
• the lowest possible mortality run-rate calculated from current data 3½x higher than the asserted 0.26%;
• unwillingness to confront high growth rates in the number of deaths shrouded by very high growth rates in confirmed cases; as well as,
• failure of the essay to account for hard evidence of excess deaths exceeding 200,000 last Spring, well above the concurrent COVID death toll of 140,000 at the time.

Though the growth rates in cases have slowed this past week, notably, in two of the five bellwether states in the South and Southwest (i.e., AZ, CA, FL, TX, and OK) as displayed in Table-2, the weekly growth rate in deaths remains in double digits, portending total deaths in the S.&SW. region as a high as 300-500,000 by year’s end, the earliest possible date of a vaccine.

Since a vaccine will likely not be tested and produced by the end of the first year of the U.S. epidemic (i.e., 28th of February 2021) the total mortality for the country could approach 700,000 deaths in a worst case scenario. This virus is not the equivalent of a harsh flu (i.e., 2x the number of deaths over a normal flu) at all.

In fact the COVID death toll is already approaching 3x the highest death toll for the flu for the past decade (in 2018-2019), essentially forecast to repeat this year. Additionally, scaling the smaller first wave of the 1918 Flu epidemic to the current population calculates out to 142,000 deaths, the best-case scenario. That is to say, the current death toll is already 25-65% larger than that benchmark.

Conclusion 
Real Clear Politics may be the most restrained of the alt-right publications. Accordingly, it makes sense to explore its thesis of massive numbers of asymptomatic cases not being diagnosed. Yet the thesis does not hold up too well under the scrutiny of checking the primary source materials it cites and from the current trends realised to date. 

TABLE-2: Bellwether states in the South and Southwest

Table-3: normalised / comparable mortality table for 33 states and six territories