FROM MY other blog:
The Planetary Ecologist: Examining SARs-CoV-2.
SARs-CoV-2 and the cost of social distancing
Currently there is a vocal minority movement, supported by forces within the Conservative Party and fanned by news cycles that cannot help but be drawn to conflict, whose aim is to open the US economy quickly and with few restrictions. There is even a movement within this movement seriously protesting the wearing of masks in public. These protesters say the economic costs of social distancing measures to the US will kill more people than COVID-19. “It will kill the economy.” They may be right, but I have seen very few predictions of how this might come to pass, based on actual data, or even on reasonable economic models. While I have not seen any good data-based projections for how a four or five-month program of social distancing will result in lives lost due to economic depression (damage to businesses, job loss, reduced family incomes etc), I have some data-based projections for how lives lost from Covid-19 might translate to a monetary cost based on known death rates and economic valuations of human life. I’ll be using Maine to draw example numbers from, and assuming a time frame of however long it would take for the disease to spike and fade without intervention or mitigation, which is basically what the “Open up Now” contingent is asking for.
The Back of Maine’s Envelope. A tale of two fatality rates. (Case Fatality Rate will often be abbreviated CFR)
One of the data points used to make these calculations in the Case Fatality Rate (CFR), which is the number of people dying from COVID19 divided by the number of people infected with COVID19. For the following exercise I’m going to produce two sets of numbers for Maine. I’m first going to use the South Korean case fatality rate of 2.368%. In the second model, I’m going to use Maine’s numbers as they currently stand. While the South Korean numbers may tell us something important about how deadly the virus is in an environment with a robust response program, Maine hasn’t had that response, nor has the rest of the US as a whole. The Maine numbers, while they skew higher, may tell us something important about how Maine will fare with COVID-19.
Using South Korea’s case fatality rate, what could the Pine Tree State expect from SARs-CoV-2? Assume just a third of Mainers get the novel coronavirus. That is probably on the low end. Angela Merkel suggested Germany might face 70% infected, rather than the 33% this model is using. I worry my model will be dismissed as alarmist if it assumes a larger percentage of infected people. Hence the 33%. It is important to remember, my model is not intended to predict actual disease trends or epidemiological outcomes. The purpose of this model is to demonstrate that lives lost to COVID-19 also have real economic costs. As my model will show, these economic costs can be quite daunting.
Maine has a population of 1,344,212 people. Assuming 1/3 of Maine gets infected, that would give the state 448,070 people with COVID-19. With a 2.368% CFR that gives us 10,610 dead. This would be the number expected if the disease only infected 1/3 of Mainers with the South Korean case fatality rate. No time frame, mitigation or further contagion is assumed. At present and with mitigation, we have only lost 65 people and our actual documented cases are massively well below 448,070. I haven’t been able to find the hospitalization rate for South Korea, so we will leave that aside. What would 10,610 deaths cost the Maine economy?
The US government values a US citizen’s life at $10,000,000 USD. With that in mind it is pretty simple to derive an estimate. 10,610 deaths would represent a $106,000,000,000 loss to the Maine economy (10,610x $10,000,000). Even with South Korea’s smaller case fatality rate, this cost would be quite significant. The Maine GDP last year was 59,255,000,000. The economic cost to Maine at the South Korean rate dwarfs our state’s GDP.
Maine’s case fatality rate is 4.9% (calculated on May 8). Maine’s COVID-19 hospitalization rate is 14.9%. Want to update with today’s data? If these numbers are closer to Maine’s case fatality rate, and hospitalization rates, how many deaths, and hospitalizations can Maine expect with 1/3 of the population infected?
448,070 x .049 = 22,153.3 deaths. Economic cost to Maine = $221,530,000,000
448,070 x .1443= 64,656.5 hospitalized.
The economic cost to Maine’s economy of widespread hospitalization due to COVID-19 is harder to calculate, but lost productivity, lost discretionary spending would have to be enormous burdens on Maine’s local economies.
These kinds of numbers are interesting but what do they mean when set against the Maine economy itself? Maine GDP for 2019 was $59,275,000,000 dollars. Now that we have a couple of estimates of COVID-19 cost based on different case fatality rates, we can make some simple comparisons.
South Korean CFR economic cost $106,000,000,000 > $59,275,000,000
Maine CFR economic cost $221,530,000,000>>$59,275,000,000
Even the current small number of Maine’s COVID-19 fatalities, (65 as of 13 May 2020) amount to a hefty $650,000,000 cost to Maine’s economy.
These are two examples, of course, and with this template one can examine a wide range of consequences by adjusting the CFR. I encourage everyone to play with these ideas. Personally, I would like to believe that the actual fatality rate for SARs-CoV-2 is much lower than any of the CFRs we have seen. We won’t know more until testing in the US becomes much more reliable and accessible.
What is the actual case fatality rate of COVID-19?
There is an argument about what the real CFR is. So far, scientists don’t really know. Many people believe that SARs-CoV-2 is already wide-spread, not reflected in the confirmed cases’ because testing large swaths of the population hasn’t been possible. That would mean the actual pool of infected people (the denominator) is much larger, and thus the actual case fatality rate much lower than what is deduced from confirmed cases. This hypothesis almost certainly has some truth to it. We know there are asymptomatic cases. Given that we don’t know the true depth of infected cases, how can we proceed? How should we proceed?
We should proceed cautiously and empirically. There are open questions to be sure, but the data we do have doesn’t give us a lot of reason to just assume COVID-19 (more specifically SARs-CoV-2) has a fatality rate below 1%, as many people posit. It may be. So far no one has been able to demonstrate that it is below 1%. I think, for the time being, we are stuck using the case fatality rates on offer, as wide ranging as they are. Above, I chose to use the South Korean case fatality rate of 2.368%. I think their rate is going to be closer to the actual fatality rate of COVID-19. This is certainly an assumption, but my reasoning is as follows. The South Koreans tested aggressively, isolated infected people and did robust contact tracing. Their health care system was never overwhelmed, and never experienced the reductions in quality of care present in overwhelmed systems. From the outside looking in, South Korea was able to keep infections low, find and isolate enough cases to keep transmission down, and manage treatment of patients with the full force of an unburdened modern health care system. I also chose the South Korean numbers because, on the low end, they seem conservative. Using the South Korean numbers is not wildly optimistic (by which I mean it doesn’t assume incredibly low fatality rates that have yet to be demonstrated). The estimate also isn’t high like the numbers in the UK or Sweden, which seem to be outliers on the high end (14.327% and 11.148% respectively- as of 13 May 2020), given that so many other places have case fatality rates in the 3-5% range.
I did not try to factor in effects of mitigation or the effects of overwhelmed healthcare systems on the numbers, in part because I don’t want to make more assumptions and in part because I have no idea how to credibly add such effects to my model. By demonstrating that lives lost also have significant costs, all I really want to offer is an estimate of the economic cost of opening the economy too soon, causing more deaths than would otherwise happen, using a hopefully reasonable estimate of case fatality rate, and the empirically derived rate for Maine. At the end I will offer a sketch of what I think the implications for the numbers would be with and without mitigation efforts. I’m not factoring in population demographics (and fatality rates really shift a lot depending on age, economic status, and health related risk factors, so bear that in mind). This model is also uncomplicated. I hope, though, that its pared down simplicity helps people think about the problem clearly.
A note about fatality rates.
The case fatality rate and the actual fatality rate are not the same thing. The case fatality rate is a number derived from known diagnosed cases vs deaths from diagnosed cases. It may overestimate or underestimate the threat of a disease-causing pathogen. How close this estimate (CFR) is to a pathogen disease process’ actual fatality rate will depend on how reliably that disease process leads to fatalities. For instance, a CFR will massively overestimate deaths if a pathogen is actually widespread but not very prone to actually causing symptoms or killing people. For instance, if it turned out that most of the world was already infected with SARs-CoV-2, the current CFR would be massively overestimating the lethality of SARs-CoV-2.
The actual fatality rate of a pathogen is much harder to derive. To discover this requires more intense and often more difficult investigation. In addition to knowing the CFR, epidemiologists must find out who has the disease, who has symptoms, who dies from the disease, who gets it and who doesn’t. This kind of estimate requires massive amounts of work. In relation to SARs-CoV-2, that detailed work is only just beginning. The bigger picture won’t be known for some time. For an examination of the case fatality rate, click this text. Crude as the case fatality rate may be, it is the most reliable data we have.
A few notes about the back of Maine’s COVID-19 envelope
This is a simple model. I haven’t accounted for time frames, or contagion rates, mitigation efforts, or no mitigations efforts. My total deaths simply assume the disease runs its course and infection doesn’t extend beyond 1/3 of Maine’s population. These assumptions were meant to simplify the model. In reality the spread of the disease could be much greater than 1/3, or much lower. The fatality rate for SARs-CoV-2 will undoubtedly be refined as more data is gathered, as will COVID-19’s CFR. I think both will skew downward. The former because I expect we will find significant percentage of infected persons who present with no or mild symptoms, the latter because doctors, nurses and an army of medical scientists are going to figure out better ways to treat COVID-19.
We can, even with this paucity of data, infer a few things, and posit a reasonable course (that can and should be corrected as new data comes in).
Given that the projected losses of life, as well as the costs associated with that lost life can be quite high, keeping as many people as possible from getting SARs-CoV-2 must be a policy goal. The other thing that seems obvious even if the CFR of COVID-19 is only as high as South Korea’s, is that it would be much better for the economy if Maine, the US, and world as a whole, spread those losses over a longer period of time so no one financial quarter absorbs the bulk of the losses. Additionally, a slower disease spread over a longer period of time would mean that health care workers would have greater resources, time, and space to treat sick people as they come in. This would almost certainly mean a lower death toll than any projected in my simple model. If that humanitarian argument doesn’t move you, it is also true that lower overall deaths would mean lower overall economic costs.
Plenty of Quibbling Room
There is plenty to debate in my model. Which estimate is likely to be closer to COVID-19’s actual fatality rate? Is a US life really worth ten million dollars? Does life devalue with age? Some might argue that earnings go down with age, but on the other hand, older people also spend a lot of money in the economy. How do population demographics affect fatality rates? Why are CFRs different across countries? The questions are endless and should absolutely be explored. The purpose of these back of the envelope numbers is more to frame issues in a more quantitative way. I want to present an argument in opposition to “Keeping Maine under social distancing will ruin Maine’s economy forever,” and “More people will die from social distancing than from COVID-19!”
The “Open Up Now” crowd rarely offers numbers to support their hypotheses. The discussion offered is consistently without links to economic analysis of reduced economic activity. We should definitely pay attention to the concerns of those who want to open up the economy immediately. However, those concerns, subjective as they are, can be hard to set into a robust economic analysis. One of the goals of this piece is to encourage anyone one considering the disease management issue to think more quantitatively and less qualitatively. Hyperbole gets us nowhere. Vague worries don’t help either. Quantitative thinking bolstered by facts and trends in the data can lead to much more fruitful discussions. They can help us see each other points and concerns much more clearly.
It appears that at least in the 1918 flu pandemic, physical distancing did not equal financial Armageddon.
Important considerations.
In Maine we currently don’t see widespread infection, or overwhelmed health care systems. There are only 65 deaths and, it seems, limited disease spread. Currently the Maine infectious rate for SARs-CoV-2 is 1.08 and not the 2.4 found in COVID-19 hotspots. This is incredibly good news and seems to argue against some of the more dire financial and human costs in the model offered. An infection rate of 1.08 does mean that the number of infected is still growing in Maine, but at a substantially slower rate (down from a March 8, 2020 peak of 1.58). It is entirely reasonable to wonder if Maine would ever see a third of its population infected SARs-CoV-2. Maine seems very nearly to have plateaued in infection. We most certainly want to avoid the exponential growth of which this virus is clearly capable.
I hypothesize that the part of that excellent trend is a product of Maine being an early adopter of physical distancing at both the government level and at the level of individuals. I also think Maine has benefitted from the natural physical distancing imposed by the state’s lower population density. There may be other local cultural factors too that have led to a lower overall rate of infection and fatalities. Maybe Mainers don’t actually travel that much within the state. Mainers may stay outdoors a lot more and don’t, especially in more rural areas, congregate in cramped poorly ventilated areas. If trends in grocery and supply hoarding is any indication, People, without direction from the federal, state, or local governments, began to self-isolate and limit their exposure to others. It is possible, even likely, that a combination of factors has likely been responsible for Maine’s good numbers.
Will lifting restrictions lead to a “second wave” of infections? Will the number of infected approach the large numbers assumed in the model? What effect will an influx of tourists have on Maine’s rate of infection? Given that each state in the US is, in many ways, unique, should every state necessarily adopt the exact same strategy in dealing with SARs-CoV-2? Maine’s strategy probably shouldn’t mimic Massachussetts’ COVID-19 strategy for instance.
Summary
Given that the COVID-19 CFR for Maine, and for the US as a whole, is actually quite high, and that CFR represents the only really reliable data we have, we should certainly want our pandemic strategy to move forward cautiously. The economic and human costs associated with even moderately high CFRs make basing COVID-19 management policy on optimistic estimates of the actual SARs-CoV-2 fatality rate potentially quite dangerous. What we know means that letting SARs-CoV-2 run rampant through populations is irresponsible and unwise. The disease it causes is incredibly dangerous. There is no shortage of evidence that COVID-19 outbreaks can be very bad indeed. In the US we have several examples from places like Washington state., Gallup, New Mexico, and New York City. The first known US citizen infected with SARs-CoV-2 arrived from Wuhan China on January 15, 2020. In only about four months, COVID-19 has killed, in the US, 82,461 people (as of 13 May 2020). We have no reliable treatments beyond symptom management. There is no vaccine to prevent it and no medicine to knock it down, or to reduce the severity of COVID-19. Remdesivir looks promising, but more work needs to be done to establish how effective it is in the treatment of COVID-19. There is also no herd immunity to insulate the vulnerable from the virus. A SARs-CoV-2 infection widely spread through the Maine population, even assuming a low COVID-19 CFR, would crush Maine’s economy. A prudent strategy, especially given the documented dangers as well as the many unknowns, would be to seek to limit the spread of SARs-CoV-2., while continuing to refine our understanding of SARs-CoV-2/COVID-19.
Maine is poised to be one of leaders in testing capacity. Within the week (13 May as of this writing), anyone who’s physician wants them to be tested for COVID-19 will be able to do so without needing to present the telltale symptoms. We are still a long way off from reliable antibody testing, and not just in Maine, but everywhere in the US. Maine, along with the rest of the country, must increase the ability and reliability of tests for COVID-19 antibodies. Current anti-body tests have a great deal of cross-reactivity (Abbie Smith personal communication). This means that the current tests have a great deal of trouble distinguishing between SARs-CoV-2 antibodies and antibodies from other coronaviruses. Accurate antibody testing would provide a much more accurate picture of how deadly this novel coronavirus actually is. It would reveal where it has been and give a better idea of who has had it, crucial since we know a significant percentage of those infected with SARs-CoV-2 have either no symptoms, or only very mild symptoms. Robust data collection could answer many questions that remain to be answered while driving a much more precise response to COVID-19, one reminiscent of the very successful South Korean model.