How more testing can — and can’t — help Minnesota reopen its economy

As Minnesota meets the goals set by Gov. Tim Walz when he extended the shutdown on April 8, does his new justification for extending it again make sense?

Yesterday, Gov. Tim Walz extended Minnesota’s stay-at-home order (SHO) again, this time until May 18. The original aim of the shutdown was to ‘flatten the curve,’ to ensure that the peak of new cases was never so great that it overwhelmed the health system’s capacity to cope. The stated aim of the extension on April 8 was to push the forecast peak back to mid-July to buy time for extra capacity to be added to meet it. On Wednesday, Walz announced that this had been accomplished. On Thursday, he extended again anyway.

So why this new extension? Sadly, we seem now to be groping forwards without any very clear guiding policy. Economically, this is a disaster. At his press conference on Thursday, Walz repeatedly mentioned ‘herd immunity,’ of which more later, possibly. But another theme which has emerged is testing, specifically the mantra ‘test, trace and isolate.’ How will this help?

How more testing can help

One of the major problems faced by policymakers worldwide in dealing with COVID-19 is that we know very little about it. This isn’t too surprising, it is brand new after all. But it is, nevertheless, a severe handicap.

For instance, we don’t even know how deadly it is. By that, I mean we don’t know what share of people who get COVID-19 die from it: the Infection Fatality Rate (IFR), or ‘death rate.’ We don’t know this because some percentage of those who get infected with COVID-19 never know (they are ‘asymptomatic’) or their symptoms are very mild and they don’t seek medical treatment. As the Star Tribune reports:

State officials stressed that the actual death rate is much lower, because there are as many as 100 unconfirmed infections for every one lab-confirmed case in Minnesota.

The people we in Minnesota have been testing so far — and this is the same in many places — are those presenting themselves as possibly being infected, mostly because they have the symptoms. This sample is more likely to give us a positive result.

We do know what share of people with positive results die from COVID-19: this is the Case Fatality Rate (CFR). In Minnesota, with 5,136 positive results reported by the Department of Health and 343 deaths (to date), this gives us a CFR of 6.7 percent (343 / 5,136).

But the CFR isn’t the ‘death rate.’ It doesn’t tell you how deadly COVID-19 is because it doesn’t account for all those asymptomatic carriers or those who don’t seek treatment. However much we might want to fill in the blank where the IFR should be, the CFR cannot fill it: it is like trying to find out how safe driving is by calculating what share of people in car crashes survive.

Until now, we have focused testing on identifying whether people who appear to be infected actually are. Given limited resources, this makes sense. But as we increase testing capacity, we can test random samples of the population, like an opinion poll, and this would start to give us some idea of what COVID-19’s IFR is.

There is little doubt that the IFR is somewhat lower than the CFR, but by how much? Assuming that those state officials are correct and there are 100 unconfirmed infections for every one lab-confirmed case in Minnesota, that would mean 513,600 COVID-19 infections in the state (5,136 x 100), 9.1 percent of the population. And, assuming that that is true for a moment, that would mean an IFR for Covid-19 — a ‘death rate’ — of 0.1 percent (343 / 513,600). This would be about the same as seasonal flu and is below most estimates of COVID-19’s IFR, which are currently around 1.4 percent.

Even if state officials are overestimating the number of unconfirmed cases by 100 percent and there are only 50 for each confirmed case, those numbers don’t rise by much. Total infections fall to 256,800 (5,136 x 50) or 4.6 percent of the population, and the IFR is still 0.1 percent (0.13 percent to be more precise — 343 / 256,800 – up from 0.07 percent).

Clearly, as Minnesota’s policymakers weigh the costs and benefits of the lockdown, they would be helped by having a better idea of how many infections there have been. It would, for example, give them a clearer idea of what share of infections end up needing ICU care (using the estimate above, it would be 0.1 percent — 358 ICU admission so far / 513,600 cases, as opposed to 6.3 percent of cases requiring ICU treatment — 358 / 5,136). This would enable them to better judge how close Minnesota’s health care system could come to being overwhelmed. Here, testing, and lots of it, will be a big help.

How more testing won’t help

What is proposed though is not random testing designed to discover the true number of infections, it is testing as a precursor to tracing and isolating. This is what South Korea did with such success, without locking down. But that doesn’t mean that simply copying that strategy will do much good here.

South Korea, with the experience of SARS under its belt, got onto testing and tracing almost immediately. If you get on this when there are few infections it is quite possible that you can identify a share of them, trace their contacts, isolate everybody who tests positive, and halt the spread of the infection. You don’t ‘flatten the curve,’ you chop its tail off.

But state officials suggest that something like 513,600 Minnesotans have been or are infected with COVID-19. Even if it is half that, 256,800, can we identify all of them? If we do, can we trace all their contacts? If we can, can we quarantine them all? To ask the question is to answer it. Quite simply, if the state officials are right about a case-to-infections ratio of one percent, Walz is not going to be able to test, trace, and isolate.

Test, trace and isolate might have been a very sensible strategy when infections were few in number, but if state officials are correct that horse has bolted. Now it is just mindless, hopeful imitation.