The Weekly Sift

So far, Covid-19 has been characterized more by what we don’t know than what we do. That has allowed reporters to write either scary or reassuring articles, depending on what they assume about the unknown. This week I went looking for articles that give the unknown its due respect, and explain a lot of the artifacts in the data that might look like trends, but aren’t.

A good place to start is Ed Yong’s “Why the Coronavirus is So Confusing” in The Atlantic.

Terminology: “SARS-CoV-2 is the virus. COVID-19 is the disease that it causes.” It’s like HIV and AIDS. The epidemic — how the disease develops in a community — is yet a third thing.

The fatality rate isn’t a property of the virus. The reason the death rate (or case-fatality rate or CFR) is so hard to pin down is that it only becomes an objective quantity — number of deaths divided by number of cases — in retrospect. The CFR describes how the epidemic unfolded in a particular place; it’s not some inherent property of the virus.

The CFR’s denominator—total cases—depends on how thoroughly a country tests its population. Its numerator—total deaths—depends on the spread of ages within that population, the prevalence of preexisting illnesses, how far people live from hospitals, and how well staffed or well equipped those hospitals are. These factors vary among countries, states, and cities, and the CFR will, too.

We’re not really sure how the virus causes the symptoms of the disease. We know what a human body looks like after Covid-19 has attacked it. We’re not sure how to separate that into (1) damage the virus does, (2) collateral damage the immune system’s response causes, and (3) side effects of treatment.

The disease seems to wreak havoc not only on lungs and airways, but also on hearts, blood vessels, kidneys, guts, and nervous systems. It’s not clear if the virus is directly attacking these organs, if the damage stems from a bodywide overreaction of the immune system, if other organs are suffering from the side effects of treatments, or if they are failing due to prolonged stays on ventilators.

Others viruses might also have more wide-ranging effects than we know, but we just haven’t seen enough cases to notice them.

“Is COVID-19 fundamentally different to other diseases, or is it just that you have a lot of cases at once?” asks Vinay Prasad, a hematologist and an oncologist at Oregon Health and Science University.

Science doesn’t go straight to the right answer. The back-and-forth nature of the early scientific debate (asymptomatic people can’t spread the disease; yes they can) throws a lot of people, but it’s not that unusual.

This is how science actually works. It’s less the parade of decisive blockbuster discoveries that the press often portrays, and more a slow, erratic stumble toward ever less uncertainty. “Our understanding oscillates at first, but converges on an answer,” says Natalie Dean, a statistician at the University of Florida. “That’s the normal scientific process, but it looks jarring to people who aren’t used to it.”

The upshot is that if your whole view of the virus depends on one study by one lab, you should maybe take a wider look.

Uncertainty is a strength, not a weakness. In politics, the guy who is loudest and most sure of himself tends to win the argument. But expertise doesn’t work that way; real experts understand just how far their expertise goes, and recognize past some point, other kinds of expertise become more important.

The idea that there are no experts is overly glib. The issue is that modern expertise tends to be deep, but narrow. Even within epidemiology, someone who studies infectious diseases knows more about epidemics than, say, someone who studies nutrition. But pandemics demand both depth and breadth of expertise. To work out if widespread testing is crucial for controlling the pandemic, listen to public-health experts; to work out if widespread testing is possible, listen to supply-chain experts. To determine if antibody tests can tell people if they’re immune to the coronavirus, listen to immunologists; to determine if such testing is actually a good idea, listen to ethicists, anthropologists, and historians of science. No one knows it all, and those who claim to should not be trusted.

In a pandemic, the strongest attractor of trust shouldn’t be confidence, but the recognition of one’s limits, the tendency to point at expertise beyond one’s own, and the willingness to work as part of a whole.

The flu comparison is even less appropriate than the numbers make it sound. In Scientific American, Dr. Jeremy Samuel Faust concludes that the comparisons we hear about flu deaths vs. Covid-19 deaths are misguided.

When reports about the novel coronavirus SARS-CoV-2 began circulating earlier this year and questions were being raised about how the illness it causes, COVID-19, compared to the flu, it occurred to me that, in four years of emergency medicine residency and over three and a half years as an attending physician, I had almost never seen anyone die of the flu. I could only remember one tragic pediatric case.

He began asking other emergency-medicine doctors, and found that their memories match his. They remember lots of opioid deaths, gun deaths, and traffic-accident deaths — which are supposed to happen in similar numbers — but not flu deaths. Flu death totals, he came to understand, are not counted deaths — deaths of particular people whose doctors write “influenza” on their death certificates — they’re extrapolations based on models. The models assume that many people die of the flu outside of hospitals, and aren’t officially counted.

In the last six flu seasons, the CDC’s reported number of actual confirmed flu deaths—that is, counting flu deaths the way we are currently counting deaths from the coronavirus—has ranged from 3,448 to 15,620, which [is] far lower than the numbers commonly repeated by public officials and even public health experts.

In other words, (and I’m commenting here, not quoting or summarizing Dr. Faust) the flu death totals we usually hear are more comparable to the Covid-19 death totals we’re starting to get from excess-death demographic models — which show much higher numbers than the 65,000+ you commonly see reported. But if we compare counted Covid-19 deaths during the second week of April to counted flu deaths during the worst week of an outbreak “we find that the novel coronavirus killed between 9.5 and 44 times more people than seasonal flu”.

What happened in Belgium? If you study those country-by-country death totals, the one that always stands out is Belgium, which has 677 deaths per million compared the US’s 204 or Italy’s 475. What horrible lesson, you might wonder, should we learn from Belgium’s disastrous handling of the epidemic?

Maybe none.

Belgium’s high numbers have less to do with the spread of the disease and more to do with the way it counts fatalities. Its figures include all the deaths in the country’s more than 1,500 nursing homes, even those untested for the virus. These numbers add up to more than half of the overall figure.

The curve has flattened, but hasn’t turned downward much yet. Check out the Washington Post’s graphs of deaths and new cases per day. The peak in deaths was 2,874 on April 21 (assuming we ignore April 14, when New York City created a blip by reclassifying 3,700  previous deaths). But deaths have been in the neighborhood of 2,000 a day for the last two weeks. Trends are harder to detect due to a Sunday/Monday effect, when deaths are lower for some reason I don’t understand. (Sunday April 26 had “only” 1,087 deaths, but yesterday had 1,558.)

The numbers also depend to a certain extent on how they’re being collected. The WaPo numbers come from Johns Hopkins, and list 2,461 on Wednesday, 2,097 on Thursday, and 1,723 on Friday. But the WHO has a different way of collecting deaths and assigning them to days. They announced that 2,909 people died in the US on Thursday, a new high.

If you look at things Monday-to-Monday, as I do, there is a downward trend. 68K today, 55K last week, 40K and 22K the weeks before. So new deaths per week have gone from 18K to 15K to 13K.

As for where the numbers might be going next, 538 collects the projections of a variety of models about how many deaths we’ll see in the next three weeks. From the 65K deaths already recorded by May 1, some models predict as few as 72K deaths by May 23, others as many as 103K deaths. But if social distancing is abandoned too quickly and a second wave starts, all those projections go out the window.

When (and even whether) a vaccine shows up is anybody’s guess. A good summary here is Stuart Thompson’s article in Thursday’s NYT. If the normal vaccine-development timetable holds, a vaccine is years away, or maybe even decades. (There’s still no HIV vaccine, for example, after more than 30 years of looking for one.) But lots of things are being done to speed up the normal timetable, and maybe they’ll work.

There is a process to finding and producing a vaccine, but not one that can be easily predicted.

Clinical trials almost never succeed. We’ve never released a coronavirus vaccine for humans before. Our record for developing an entirely new vaccine is at least four years — more time than the public or the economy can tolerate social-distancing orders.

But if there was any time to fast-track a vaccine, it is now.

The main way you speed things up is that you do everything at once. Rather than take the most likely vaccine candidate, test it, and then test the next one after the first one fails, 95 different vaccines are being worked on at the same time. Some of them are probably very bad ideas.

What if a promising vaccine actually makes it easier to catch the virus, or makes the disease worse after someone’s infected? That’s been the case for a few H.I.V. drugs and vaccines for dengue fever, because of a process called vaccine-induced enhancement, in which the body reacts unexpectedly and makes the disease more dangerous.

That’s why you don’t just dream up a formula and start injecting it into the general public. Normally, there are three phases of testing, with time in between for analysis. But for Covid-19, you might start one phase before the previous one finishes. You also might start prepping a factory for production before you’re sure a vaccine works.

If you do all that and you get lucky, you might have a vaccine in mass production by August 2021.

The most aggressive timetable has been put forward by a group at Oxford, which is talking about availability in September, 2020. That would be a million doses, not the billions ultimately needed. But a lot has to go right before that happens. Human trials started this week. Right now all we know is that it works in rhesus macaques.

Whatever we’re returning to, it’s not “normal”. Another Atlantic article, Uri Friedman’s “I Have Seen the Future—And It’s Not the Life We Knew“, looks at the early signs of post-lockdown life in countries that are ahead of the US: China, South Korea, Denmark, and a few others. The very resemblance to normal enhances the strangeness of it.

In China, Friedman reports, reactions bifurcate as everyone anticipates the possibility of a second wave or a new plague. Some remain constantly on their guard, while others take a live-now approach: You’d better do whatever you can while you can, because it might all be taken away tomorrow. In general, the Chinese are saving more and spending less, frustrating planners who hope for a quick economic recovery.

Denmark seems to have the opposite problem: When the government reopened daycare centers and schools, many Danes took it as an all-clear signal. “[Just] because the schools open, doesn’t mean you should stop washing your hands,” says a University of Copenhagen professor.

Temperature screening at City Hall in Seoul

The NYT has a similar article:

In Hong Kong, tables at restaurants must be spaced at least five feet apart and customers are given bags to store their face masks during dining.

In China, students face temperature checks before they can enter schools, while cafeteria tables are outfitted with plastic dividers.

In South Korea, baseball games are devoid of fans and players can’t spit on the field.

… Libraries in Hong Kong are reopening, but visitors are allowed to be inside for only an hour at a time.

Hair salons in Sydney, some of which had closed because of the virus or financial pressures, are back in business with abundant supplies of masks and hand sanitizer. At some, magazines are no longer handed out to customers.

… In Sydney, schools are reopening in phases, holding classes one day a week for a quarter of the students from each grade and gradually expanding until the end of June.

We can hope that our opportunities increase over the coming months. But normal? Well, not exactly.