Covid and My Inner Skeptic

I have taken a pretty strong position on covid-19: it will kill a ludicrous number of Minnesotans/Americans unless we shut things down pretty hard until we have the ability to suppress it (either through treatments or through testing and contact tracing). I don't think it's possible to remain shut down until we have a vaccine, because we simply don't have the resources (financial, mental, logistical) for an 18-month lockdown. But I think we will reap the rewards, in both lives and dollars, if we continue lockdown until we have some means of suppressing or defanging this virus.

For once, I seem to be on the side of The Experts about something! Disease experts, from Dr. Gottlieb and Dr. Christakis on the (relative) right to Dr. Lipsitch and Dr. Rasmussen on the (relative) left, have formed something resembling a consensus in favor of this approach.

There's a Bloomberg article going around lately about what we should learn from coronavirus "skeptics," who oppose this consensus. A couple of them are experts themselves. I'm not going to link the Bloomberg article, because, even though it's an earnest, well-intentioned article, it makes some dumb rookie mistakes. Nevertheless, it makes a good core point -- there ARE things to be skeptical about in the current experts' consensus. Some of what we're doing is based on best guesses that we know could be off.

Here are some of the arguments my inner skeptic has with me:

1. Outdoor transmission doesn't seem to be a big thing based on the evidence so far. Playgrounds/beaches/mayyyyyyyybe swimming pools could potentially reopen at, say, half capacity.

Why my inner skeptic still loses this argument: I'm waiting for a little more confirmatory evidence before embracing it. My state is starting to see a few playgrounds reopen, and it will be interesting to see how that goes.

***

2. We still don't know how many undetected cases there are for each detected case. If we're catching 2-5% of all cases (as I currently believe), then covid is killing 1 in 200 (ish) of those who catch it, and we're doing about what we should be. If we're only catching 1% of all cases (as Minnesota currently believes), then what we're doing still seems like the best option, but it's a much closer call.

On the other hand, If we're only catching 0.1% or 0.01% of all cases, then covid isn't all that deadly and we should chill the heck out.

Why my inner skeptic still loses this argument: there's very little evidence supporting those extremely low detection rates, and lots of evidence pointing to the more modest estimates. Here's a rough-and-ready summary of evidence (of varying quality) collected up to this week: PCR and Serological Studies, Google Sheets

***

3. While we know quite a bit about the characteristics of the virus itself, we DON'T really know very much about the social dynamics of mitigation, lockdown, and so forth. Much of what we do know is derived from not-entirely-analogous situations from the flu pandemic of a hundred years ago -- not the best data set.

Consider the famous Imperial College model: the parameters Imperial College used to describe covid (r0 = 2.4, case fatality rate = 0.9%) were based on the best science available at the time. Although that science has shifted (happily, in a positive direction!), it hasn't changed more than everyone reasonably expected with a novel virus. Their analysis of the virus's behavior remains sound. But the Imperial College model also made assumptions how people would behave. (See page 6 of their report.) For example, they assumed that, if infected households were asked to quarantine, those households would reduce community contacts by 75% and that 50% of households would comply. They assumed that, under a broad stay-at-home order, everyone would reduce external contacts by 75% (while increasing internal contacts by 25%).

Are these numbers reasonable? They seem reasonable to me. Are they correct? ...how could we possibly know? We've never done anything remotely like this in modern times. The paper certainly cites no sources. These are educated guesses about human behavior. And if you tweak these parameters just a twinge, it can make the difference between r=1.1 (the virus grows slowly but exponentially) and r=0.9 (the virus slowly dies out like in New Zealand).

So it's quite possible that government mitigation efforts combined with people's natural desire to stay home in the midst of an epidemic does a better job of reducing disease spread than the experts think. It's also possible that the added benefits of legally-binding stay-at-home are smaller than we think they are. Either way, moving to a "Sweden model," where the government issues strong recommendations and tightly enforces social distancing in public spaces, but does not actually make you stay home, might lead to fewer deaths than one would expect based on the virus's lethality.

Why my inner skeptic still loses this argument: actually, I'm really starting to suspect that some version of this is true, to some degree. Covid still seems to be the kind of virus that could kill tens of millions of people worldwide, but the way it's spreading and killing in various countries that don't have legally binding stay-at-home orders makes me think that individual people are, by and large, making good choices to slow its spread no matter what their governments say.

But before I'm willing to say, "Yes! We can reopen businesses and it won't make that big a difference as long as we still mostly stay home voluntarily!" I really want to make sure that isn't going to kill a million Americans. And that's a tricky business. The difference between a virus with an effective reproductive number just slightly below 1 and a virus with an effective reproductive number just slightly above 1 is the difference between a few hundred Minnesotans dying and tens of thousands of Minnesotans dying; even a lockdown that is only slightly effective is worth it if it's the only thing keeping us below 1. (But is it? We don't know!)

Basically, I check Sweden's death counts a lot. Yes, their death rate is quite high compared to their neighbors. But it's also falling, and has been for weeks. (While Sweden's reporting of death data lags, that doesn't account for the full decline. I checked.) This suggests that, at least in Sweden, a lockdown is not necessary to get that effective reproductive number below 1.

Would the same be true for us? It's a question well worth asking, given the extraordinary costs of lockdown. Now that some states are starting to "reopen," it will be put to the test.

***

4. We still have no clear conclusion on whether children actually transmit this thing the same way / to the same extent adults do. We have to assume that they do, but there's reason to believe that they don't. And, if they don't, then opening the schools would have enormous economic benefits with very low costs.

Why my inner skeptic still loses this argument: Evidence is really mixed right now. Some evidence says children are important transmission vectors. Other evidence suggests not so much. (Some of that is the same evidence interpreted differently.) I'm very interested in the outcome of this NIH study, announced Monday.

Fortunately, the school year is effectively over anyway (insert my parental griping about "distance learning" here), which means we have until roughly August to figure this out and decide whether and how to reopen schools. By then, I'm confident things will have changed a great deal, one way or another.

***

5. I've advocated for a suppression strategy: crush the curve, then test/trace/isolate. This is usually contrasted with a "herd immunity" strategy. The difference is enormous. Under a "herd immunity" approach, something like 1 in 250 Americans would die over the next 6-18 months (that's ~1.3 million). Under a "suppression" approach (which is what most models assume we do and what the White House has endorsed), something like 1 in 2,500 Americans would bite it instead (~130,000; we're halfway there already). It's a huge, huge win and it's well worth absorbing extremely large short-term economic costs in order to make it happen.

But what if it's impossible? It could be the case that the virus is too infectious for us to actually get its reproductive number below 1, even with many months of stay-at-home orders. Even if we can, there might already be too many infections for us to get the number of cases down to a manageable number in anything like the immediate future. It may also be the case that our testing capacity is simply not going to be able to keep up with prolonged demand over the course of the entire year.

If any of those very real possibilities proves out, it would mean we'll never be able to get covid under control enough to test/trace/isolate. Test/trace/isolate would be impossible. That would make herd immunity -- and, likely, mass graves -- inevitable, and our job would shift to managing that instead of trying to actually save all those lives.

Why my inner skeptic still loses this argument: It's going to take a lot of evidence for me to consign a million or so Americans to the morgue, and, right now, the best evidence still suggests that test/trace/isolate can work, and within a reasonable timeframe.

But I am going nuts watching the case counts rise in Minnesota while the testing rate keeps struggling to hit a daily benchmark that we really needed to start consistently hitting a week ago. Meanwhile, Minnesota is only just now getting around to approving funding for contract tracers, who need to be on the ground doing contact tracing yesterday. Time matters enormously here, every day that passes puts our goal farther out of reach, and it is slipping away from us. Today, I'm much less optimistic about America's ability to rise to this challenge than I was in March. (On the other hand, some other states with similar strategies seem to be faring better than poor Minnesota.)

***

6. On the other hand, it may be the case that "herd immunity" is impossible. We still don't know how long immunity lasts after an infection, or if getting infected confers immunity at all. If catching covid doesn't provide you with at least several months of safety from catching covid again, then there will never be and can never be herd immunity against it. (See also: influenza, arguably HIV). In that case, lockdown followed by test/trace/isolate is really the only choice.

(God help us if both herd immunity and test/trace/isolate prove impossible.)

Why my inner skeptic still loses this argument: Even since I started writing this article, there's been good news on this front. No proof yet, but there are some good indicators that covid infection does confer some reasonable degree of immunity for at least some period of time.

Plus, I believe it was the Doctor of Doctor Who who once said that he knew there was an override switch somewhere on a console because "I started with the assumption that there was a way to survive and worked backwards from there."

***

7. Not everyone needs to be locked down at the same time, and it's possible that some who locked down when they did did so too early. The whole country is not New York City. Montana currently has low case counts and high testing; there's not a lot of good reason to think that they should be locked down right now because of an epidemic they currently have under control.

Why my inner skeptic still loses this argument: as a matter of fact, it doesn't. I think this is clearly true. As I've said before, we don't have a covid epidemic in the United States. We have 50 covid epidemics in the United States, one per state, and probably many more. Each of these epidemics is at a different stage, with different characteristics. Different responses are appropriate in many cases.

Now, don't oversell this. I'm not saying that Rural America Is Safe And Should Open While You City Slickers Cower. Rural America looks like it's finally starting to catch covid in large numbers as it leaks out of the big cities (and the rural meatpacking plants). Rural America is in for a very sucky few weeks and months ahead, and, just like the rest of us, the moment when they should lock down will likely come about three weeks before most residents realize the epidemic has "busted out" in their community, when case counts are still quite low.

But does Don's Cafe in Morris, MN (confirmed county-wide case count: 1) really need to be closed today just because the meatpacking plants have caused hundreds of cases in other rural counties? It seems dubious. Within large states, a region-by-region or even county-by-county approach seems like it should be developed and deployed quickly.