COVID Winter Predictions; Napkin Calculation #6, Part 2

We continue from COVID Winter Predictions; Delta+/AY.4.2; Prediction Follies, Part 1.

This paints a picture:

A picture of the failure of COVID vaccines to benefit public health in the way they benefit the individual, which is indisputable. While medical knowledge of the disease is significant and growing, it paints a picture of almost total ignorance of the epidemiology of COVID. The distinction is important, since the medical profession is under siege by a large contingent of the public who have descended to idiocy, exacerbated by unscrupulous politicians and social media.

Quoting from COVID Winter Predictions; Delta+/AY.4.2; Prediction Follies, Part 1,

This is one of those occasions where it is constructive to undermine. To allow the predictions to remain unchallenged invites possible public health whiplash, as happened with (CNN, 2021/05/14) The CDC says masks are no longer needed inside or out if you’re fully vaccinated. By now, we should realize that COVID has more fake-out moves than the Harlem Globetrotters.

A model is a means to simulate a real world process, usually on a computer, sometimes on a napkin. The current state of the art, COVID-19 ScenarioModelingHub, Round 9, takes as input the plateau-uptick that began on 10/23/2021. But as of  11/20-present, the uptick barely registers in the predictions. In isolation, you may be forgiven your trust. But you can’t ignore Europe, which is not fundamentally different from the U.S. A sinking suspicion invades your brain: You can do better reading the news.

The predictions of COVID-19 ScenarioModelingHub, Round 9 look no better, and possibly worse, than a math technique that dates to about 1960, the linear Kalman filter.   In the 60’s, this mainstay of systems engineering was a game toy, used to predict things like weather, for which it should not have worked at all. It worked better than nothing.

I’m not going to bother you with the meaning of “linear”, except to note that many problems can be broken into linear pieces. This is why the ScenarioModelingHub models seem to work — until there is a change in the trend. The inability to predict a change in trend is characteristic of a bad model. Only the simplest part of the models work, the crude linear approximation. Garbage in, garbage out. 

Our napkin calculations are guilty of same, but they make no pretense of accuracy.  The wisdom of John von Neumann applies: “There’s no sense in being precise when you don’t even know what you’re talking about.” A napkin calculation is not to be believed, but usefully undermines pretenses of sophistication.

Let’s enumerate recent public health pronouncements for contradictions:

  • Two shots are all you need. This was recently turned on its head.
  • Cell mediated immunity, plasma cell growth, and antibody quality continue to improve long after antibody titers decline, providing durable protection. On its head.
  • The reproduction numbers, R_o and R, are useful in prediction modeling. False beyond trivial napkin calcs.
  • A vaccination sequence that works when people stand 6 feet away works when people stand 3 feet away. Numbers, breakthrough versus distance, versus date of vaccination, unknown.
  • Three shots are all you need. Doubtful.
  • Breakthrough cases are  minimally infectious. Correctness,  with normal waning of immunity is  unknown.
  • When, after this winter, in analogy with  German Health Minister Jens Spahn, (CNBC) “probably by the end of this winter, pretty much everyone in Germany will be vaccinated, recovered or dead…”, we will finally be done with COVID. Doubtful.
  • COVID is like other respiratory diseases. Manifestly false. It seems to re-arm after a short interval.
  • Herd immunity will happen. Not to the extent you might hope.

Taking note of this poor state of knowledge, we paint what went wrong with a broad brush.

Empirical. Diligence in social distancing has declined. At the Willow Grove, PA Home Depot store, 80% of employees faked the wearing of masks, nose uncovered, or around their necks. At the Moreland Road Walmart, 50% compliance was observed.

Theoretical.  Quoting Becker’s Hospital Review) States ranked by percentage of population fully vaccinated: Nov. 23,

“As of 6 a.m. EDT Nov. 22, a total of 196,398,948 Americans had been fully vaccinated, or 59.2 percent of the country’s population, according to the CDC’s data.”

Not anymore, they aren’t. In the crude world described by John von Neumann, where we don’t know what we are talking about, cut that in half, to 30%. It will do for argument. Now dredge up Napkin Calculation #4, which resides at (CNN) White House frustrated with Irresponsible Delta Variant Coverage…Napkin Calculation #4. The relevant equation:

    • Effect of a vaccine: R = R_o * (breakthrough%)/100.

With loss of social distancing, we have to pick a number for R_o. Estimates have been as high as 9. (Journal of Travel Medicine) The reproductive number of the Delta variant of SARS-CoV-2 is far higher compared to the ancestral SARS-CoV-2 virus offers the result of a survey, R_o = 5. Since some people are still wearing masks properly, let’s fold it into R_o, as R_mask = 3.5, so we can see what the vaccine does.

Now pick a breakthrough rate for the poorly vaccinated U.S. population. 70% breakthrough is plausible. Reduced immunity from aged vaccination status may tamp down mortality, but many breakthroughs will experience the worst sickness in their lives. Let’s plug in the numbers:

    • For Pfizer, possibly Moderna, and Delta, R = 3.5*( 0.7) = 2.45. Disaster!

On July 10, I began to agitate for a booster: Pfizer versus FDA & CDC; Booster Shot Needed Sooner than Later. On July 12, Pfizer Booster is Needed? Breakthrough Case Infectivity; Managing Public Health Communications. Quoting from  {9/15) FDA advisers consider Covid-19 booster shots Friday, 

So why the resistance? FDA/CDC are constituted to render decisions of the kind that A nationwide study… presents fact in the state of Israel. Nothing in the FDA/CDC charters permit rubber-stamping the authority of another national establishment, even one as competent as Israel.

Following form, and the requirements elucidated by Drs. Philip Krause,  Marion Gruber, and Paul Offitt, for evidence based medicine,  the FDA did not approve a booster. On 11/19, they approved boosters for everybody, though 6 months after the 2nd shot may not be timely.

When a system of thought produces a decision as wrong in delay as the FDA decision,  it demands critical examination — of the system, not the decision.  Otherwise, the system will continue to deal out similar, deadly errors, correct in form, wrong in hindsight. No decision process should have sanctity. Figure it out.

If I was prescient, napkins deserve the credit. Napkin calculations facilitate mental flexibility, because they are so easy to change. The models of ScenarioModelingHub are enormously complicated, involving automated data collection, normalization, and complex calculations. The modeler becomes a slave to the model.

As for the rest of us, we can free ourselves.