May 2020

Moderna, Inc. today announced outstanding early data for its Covid-19 vaccine candidate, in a trial led by the National Institute of Allergy and Infectious Diseases (most Americans will recognize NIAID as Dr. Anthony Fauci’s agency), which is part of the National Institutes of Health. Key points from the press release follow, with some plain English translation.

Dose dependent increases in immunogenicity were seen across the three dose levels, and between prime and boost within the 25 µg and 100 µg dose levels. All participants ages 18-55 (n=15 per cohort) across all three dose levels seroconverted by day 15 after a single dose. At day 43, two weeks following the second dose, at the 25 µg dose level (n=15), levels of binding antibodies were at the levels seen in convalescent sera (blood samples from people who have recovered from COVID-19) tested in the same assay. At day 43, at the 100 µg dose level (n=10), levels of binding antibodies significantly exceeded the levels seen in convalescent sera. Samples are not yet available for remaining participants.

At this time, neutralizing antibody data are available only for the first four participants in each of the 25 µg and 100 µg dose level cohorts. Consistent with the binding antibody data, mRNA-1273 vaccination elicited neutralizing antibodies in all eight of these participants, as measured by plaque reduction neutralization (PRNT) assays against live SARS-CoV-2. The levels of neutralizing antibodies at day 43 were at or above levels generally seen in convalescent sera.

Translation: The trial is looking for the best dose, and tested three dose levels on 15 patients each, for a total of 45 patients, two administrations of vaccine for each. Every patient developed some antibodies within two weeks of the first dose. Eight patients progressed to 43 days, which is two weeks after the second dose, and all 8 developed antibodies at or higher than levels seen in patients who have recovered from Covid-19.

The vaccine candidate was generally safe and well-tolerated, and a Phase 3 trial (the trial designed to prove safety and effectiveness to statistical significance, which is needed for regulatory approval) will begin in July.

For those of you for whom hope is a strategy, which is most of us, this news is very good.

The prospects for a vaccine against Covid-19 are very good, hedging about the durability of immunity notwithstanding. There are more than 100 other programs working on Covid-19 vaccines around the world, and tremendous pressure, including geopolitical competition, to get to the finish line first.

Following, a fair use excerpt from a Wells Fargo analytical report on vaccine programs (sadly, no link):

Summary. This is the inaugural issue of our Covid-19 Vaccine Tracking Report. In this report, we are tracking a select set of vaccine candidates targeting the novel coronavirus designated SARS-CoV-2. These candidates must meet three criteria: (a) currently in human
clinical trials, or with imminent starts, (b) sufficient capitalization to
carry through multiple R&D stages, and (c) partnerships or
manufacturing plans to credibly release substantial quantities of
vaccine within the next year.

Current Situation. Well over a hundred Covid-19 vaccine candidates
are in various stages of research and development. These vaccine
candidates target the SARS-CoV-2 virus, typically one or more aspects
of that virus’s spike (S) protein. Virtually every vaccine type that has
ever succeeded with some other pathogen (e.g., inactivated, subunit,
vectored) or is viewed as promising new technology (e.g., DNA, mRNA) is being explored. Our three criteria lead us to track ten vaccine candidates being developed by BioNTech, CanSino, Inovio, Johnson & Johnson, Moderna, Novavax, Sanofi, Sinopharm, Sinovac, and the University of Oxford (alphabetically by sponsor)….

Timing: We anticipate that positive data at end of phase 2 trials could
be sufficient to trigger EUA [Emergency Use Authorization] status with national regulators. Based on public statements to date, this could be as soon as late 2020 for several of the candidates. At present, JNJ and SNY project later trial starts and thus later EUA decision points. Presumably, those two companies are assessing how to compress their own development timelines. Once products are distributed under EUA status, some of the highest risk populations can be offered some degree of protection. Meanwhile, data collection will continue, even as distribution broadens, to build the evidence base that would warrant full reviews and approvals from FDA, EMA, and their counterparts
around the world.

For whom: We expect initial supplies of Covid-19 vaccines to be focused on people at highest risk of infection or essential to community function. Some of these will be occupational groups (e.g., healthcare workers, first responders, grocery-store workers, schoolteachers). Others will be social groups (e.g., nursing-home residents, seniors, communities with high attack rates). Starting points for prioritizing vaccine distribution will be similar to those used in previous vaccine shortages (e.g., influenza vaccine in 2004 and 2009)….

Assessment of Production Capacity. Ideally, to assess leading manufacturers’ plans for Covid-19 vaccine production capacity, each company would describe a common set of production goals and corresponding dates. For example, doses to be produced through December 2020, as well as goals for doses per month to be released in June 2021 and in December 2021. In reality, the companies’ public statements provide varying levels of detail, tempered by uncertainties inherent to technology transfer, industrial scale-up, facility repurposing or construction, dosing requirements not yet settled, regulatory approvals, and related factors. A major source of uncertainty will be resolved for each candidate once the human dose is set for the phase-3 trials of vaccine efficacy. The quantity of protein per dose bears directly on how many doses can be produced from any given manufacturing asset (e.g., fermenter).

Bold emphasis added.

The very last point provides some important context for the Moderna trial discussed at the top of the post. Why not just go for the highest safe dose to get the most robust antibody levels? Because that will slow down production and delay vaccination for millions of people. The dose determination needs to be optimized, not for profits — although we certainly expect some ignorant sniping in that direction once the popular press starts focusing on dosing decisions — but to maximize the number of people who can get the vaccine, even if some of those people will not develop antibody levels that confer perfect immunity. Whether there is a magical dose level for any safe and effective vaccine that reliably confers immunity and allows for grillions of doses very quickly remains to be seen.

Stay tuned.

In dramatic lore (and great sportswriting), the Four Horsemen are Famine, Pestilence, Destruction, and Death. In St. John’s original construct, “War” stands in for Destruction. We prefer Destruction, because it captures the many types of war not imagined in Biblical times.

This morning we reread the first paragraph of Barbara Tuchman’s classic work on the worst century in Western history, A Distant Mirror: The Calamitous 14th Century, reproduced below. Tuchman proposed seven horsemen:

Plague, war, taxes, brigandage, bad government, insurrection, and schism. Broadly defined, we’ve got six of them running around the United States right now. “Brigandage,” which involves unemployed soldiers gallivanting around the countryside looting the undefended and disarmed locals, is common in the world but has not been a feature of American life since the years following our Civil War, no matter how you might characterize gun rights demonstrations of recent moment or the violent crime of the ’60s to the ’90s.

Plague? Check. Not literally “plague,” of course, which is a specific disease with a precise cause and an effective remedy, but plague in the sense that people who do not consort with medievalists or infectious disease experts use the word.

War? Nineteen years and running. We even confess to having supported those wars once upon a time, which is more than most people will admit. My guess is that nobody will care so much about terrorism now, so maybe we should generally withdraw and let all those people resume killing each other. But what about “Destruction”? We have made a policy decision (which we admit we supported, for a while) to destroy our material well-being to save lives from plague, and there are those who argue that we need a good deal more destruction still. Maybe that policy choice is yet the most cost effective — we won’t know for several years which choices were best — but all Americans, including especially the WFH overclass, ought to have the courage to call it by its name: Destruction.

Taxes? They are coming hard. Beautiful taxes like you’ve never seen before, in every American jurisdiction.

Bad government? No matter who you are, you have your favorite examples. As we have pointed out, everybody agrees that there have been massive failures of government in the United States. One’s opinion as to the cause of those failures is a Rorschach test for one’s pre-pandemic predilections.

Insurrection? We are closer than we have been for some time. Google “defies.” We have hair salon owners defying judges, mayors defying governors, and governors defying the president, all of which seems weirdly reasonable under the circumstances. Nobody is shooting yet, but we are one out-of-proportion bad judgment enforcement action away from another Ruby Ridge or Waco. Brace yourself for the “national conversation” about that.

Schism? The Papal Schism of the 14th century was so scary because each pope excommunicated the followers of the other. When one believes that this life is the misery one must endure for immortal paradise, excommunication is the equivalent of killing one of Tolkien’s Elves. The loss of immortality is a tragedy greater than mere mortal death, because the sacrifice is so great. Our schism today, which involves profound contempt verging on unqualified hatred for people who have a different vision of the meaning of the United States, destroys the purpose of our country, unique in the world, that moved our extraordinary ancestors to overcome challenges vastly more difficult than Covid-19. That is, or would have been to many Americans of old, a tragedy greater than mere mortal death.

The most profound sentence in Tuchman’s first paragraph may, unfortunately, be the last: “All but plague itself arose from conditions that existed prior to the Black Death and continued after the period of plague was over.”

Let us hope that history does not repeat itself.

We’ve been wondering about the effects of Vitamin D, which one mostly gets more of by exposure to sunshine, on Covid-19 mortality. Yesterday saw new news from Trinity College in Dublin on the point:

Whereas there are currently no results from randomized controlled trials to conclusively prove that vitamin D beneficially affects COVID-19 outcomes, there is strong circumstantial evidence of associations between vitamin D and the severity of COVID-19 responses, including death.

Here’s a recent study from Northwestern that points in the same direction:

Backman and his team were inspired to examine vitamin D levels after noticing unexplained differences in COVID-19 mortality rates from country to country. Some people hypothesized that differences in healthcare quality, age distributions in population, testing rates or different strains of the coronavirus might be responsible. But Backman remained skeptical.

“None of these factors appears to play a significant role,” Backman said. “The healthcare system in northern Italy is one of the best in the world. Differences in mortality exist even if one looks across the same age group. And, while the restrictions on testing do indeed vary, the disparities in mortality still exist even when we looked at countries or populations for which similar testing rates apply.

“Instead, we saw a significant correlation with vitamin D deficiency,” he said.

Google yields more such early data.

Now, in the United States there is a huge range in mortality rates from one state to the next. Using the Worldometers data from yesterday, in the 20 states with the most confirmed cases the mortality rate ranges from 1.67% in Tennessee to an astonishing 8.97% in Michigan. Even a few weeks ago, these disparities were easily dismissed as an artifact of testing rates — some states had high mortality rates because they had insufficiently tested and therefore the denominator was artificially small, the argument went. As testing has risen in to the hundreds of thousands even in the smaller of these Top 20 states, that explanation is weakening.

So why the huge differences in mortality rates?

We compared the mortality rates in the Top 20 states and their average hours of sunshine per year. The results were interesting, albeit subject to the qualifications below the scatterplot:

We did adjust the data in one respect: New York’s “hours of sunshine” calculation was based on readings in Syracuse, which is a very gloomy place. Since most of its deaths are in the tri-state region, we used New Jersey’s number for New York, which is probably a decent quick and dirty correction.

Obviously, there are many limits to this approach, the first being that hours of sunshine in a state is only a very loose proxy for Vitamin D levels. The actual Vitamin D readings for hospitalized patients who do and do not die would tell us a lot, and even better would be the levels from a sampling of mildly symptomatic or even the famous asymptomatic cohort.

There is also the problem that we used the annual hours of daylight, rather than during the two months during which the Covid-19 deaths actually occurred. We reason that the correlation could actually be stronger if we had the urge to unpack the sunshine data for those two months, insofar as many of the high mortality states in the north probably accrue most of their sunny hours in the summer and fall rather than in March and April, but who knows?

No doubt actual scientists could think of many other wrinkles. Regardless, understanding the massive differences in mid-pandemic mortality rates between American states will be fodder for countless graduate students in the years to come.

Covid-19 is spreading through the White House, ever closer to the President. Judging from the comments to the linked post, many people who hate or have contempt for Donald Trump would be delighted if he caught the disease.

Our guess is that if Trump or Pence did get Covid-19 the left would do a terrible job at concealing its glee — schadenfreude being the most difficult to resist of the ugly emotions — and that will ensure the GOP ticket’s victory in November. Covid-19 in the White House might in the end be Troll Level: Universal Overlord.

Unless, of course, the president dies.

A few minutes ago the US Department of Labor announced the unsurprising news that payrolls had dropped by more than 20 million people in April, and the headline unemployment rate had risen to a catastrophic 14.7%, which is probably a few points lower than the reality because of quirks in the reporting. “We meant to do that,” insofar as these declines were the result of some combination of voluntary decisions to social distance and government policies, in each case to limit the casualties from Covid-19. In other words, we have chosen to take on massively more economic risk in order to reduce the mortality and morbidity risk from Covid-19.

News such as this and what we know from speaking with friends and neighbors naturally invites many questions. Among them: Do the government policies and individual practices to flatten the curve work? If so, in what respect do they work? And, finally, are they worth it?

As we wrote almost a month ago, it will be a long time before we are actually able to know which policies worked, and at what cost. Here are the criteria we proposed:

Ultimately, we will know three things in each applicable jurisdiction (itself a hot topic, of which a bit more below): (1) excess deaths (whether from the virus or the stress of the economy and confinement or from local collapse of the healthcare system) over the baseline experience, (2) excess morbidity and mental health problems, and (3) the decline in the jurisdiction’s GDP per capita against the baseline. Those three things in a given country or state will be the indicators of success or failure.

No matter the claims you read from competing media, politicians, and “experts,” we do not believe that we will be able to determine which policies most efficiently worked against these three values for quite some time.

There is, however, another way to think about the risk Covid-19 has added to American life: How does it compare to health risks within recent memory?

Through today, about 77,000 people have died of Covid-19 in the United States. Yes, the number might be somewhat higher or lower depending on how you count, but at that number we have lost roughly 23 people per 100,000 of population. Even if the total deaths this year quadruple in an awful second wave, we will have lost around 100 people per 100,000. (Neither of these figures are age-adjusted, which puts more weight on deaths of younger people and less on deaths of older people deaths, as most comparable data are.)

That still seems like a lot of dead people, and by recent standards it is. There is much tragedy in 77,000 incremental dead people, and vastly more in four times that number.

But how does the risk of Covid-19 compare to other risks with which we are familiar, and which we readily accept, or accepted?

See, for instance, this graph of age-adjusted mortality from cardiovascular disease in the United States. In 1960 we lost 559 people per 100,000 to cardiovascular disease, and as recently as 1990 we lost 322 people per 100,000. In 2017, that number was 165 per 100,000. In other words, if there had been no improvement in the treatment and prevention of cardiovascular disease since only 1990, in 2017 we would have lost an additional 157 people per 100,000, or more than 500,000 people.

Were you anxious about your risk of dying in 1990? If not, perhaps you shouldn’t be too afraid of Covid-19 this year.

Suppose we say that the deaths from Covid-19 this year will in fact quadruple, to about 100 per 100,000. Without adjusting for age, that probably overstates the Covid-19 death rate compared to cardiovascular disease. Even that many deaths from Covid-19 would be the equivalent of giving up the improvement in death rates from cardiovascular disease since only 2004. For one year!

Were you anxious about your risk of dying in 2004? Not particularly? Not enough to lose weight and strap on running shoes? With that framing, is the crushing poverty, which is increasingly looking long-term, from continued mandatory distancing too high a price for such a small relative benefit? If the numbers were presented this way, we believe that most Americans would say “no,” perhaps except for one thing.

One American does not kill another American by transmitting cardiovascular disease. But Americans do kill each other by transmitting Covid-19. We are obviously willing to risk killing strangers in the abstract, which we prove every time we drive too fast because we left the house too late. But nobody wants to kill their own mother or grandmother. The pain in that possibility is too great.

So, are we paying such an enormous price to avoid Covid-19 deaths, however few they are now likely to be compared to other health risks in recent memory, because we are personally afraid of the disease, or because we do not want to live with the guilt of killing a loved one?

Or, maybe, we humans just do a very poor job of thinking about risk. Never has that failing been more expensive.