Here’s another guest blog from Dr Carter. (If you didn’t read her post on the origins of COVID-19, be sure to take a look because it’s chilling.) This is food for thought as we try to reopen the economy.
–Caroline
Let’s look at some recent epidemiological data to see what we have learned in the last few weeks.
What’s the real mortality rate (MR) of COVID-19 based on Seroprevalence?
New York City
What’s the real mortality rate (MR) of COVID-19? Meaning, what percentage of people who are infected with COVID-19 go on to die from it? We have some idea of how many have died from COVID-19. The information we are missing is how many total individuals were infected. Since many people have had the disease with minimal symptoms, the only way to determine the number of people infected is to test the population for COVID-19 antibodies (“seroprevalence”). According to a recent survey of 1,300 NYC dwellers, 21% tested positive for COVID-19 antibodies. If we factor in the population (20.3 million), that means about 4.3 million residents have had the virus. Current mortality estimates are running around 12,050, which means the crude mortality rate (MR) is around 0.3%. Given the high prevalence rate of the disease and the accuracy of the test, as well as the problem of possible undercounting of deaths, there will be some uncertainty around these numbers BUT it’s probably safe to say that the MR of COVID-19 is a lot less than 2% and likely less than 1%. The calculations for the state of New York are similar to those of NYC.
California
California conducted two surveys in Los Angeles and Santa Clara counties. The data have been analyzed and posted as preprints, meaning they haven’t been published in peer-reviewed journals yet. The seroprevalence numbers are around 3.5% and 1.5%, respectively. There is currently the equivalent of a war going on involving statisticians arguing the numbers, which has to do with the sensitivity and specificity of the test (think false positives and false negatives), but the overall conclusion is that there are a LOT more cases of COVID-19 in these counties than have been counted. This will also drive down the MR.
Other Countries
Other countries are starting to publish their own seroprevalence data with ranges from 1% of individuals having been infected (Scotland) to 15% of the population infected (a hard-hit town in Germany).
National Virus-coping Strategies
The Kiwi Model
Epidemiological pundits point to New Zealand and Australia as the best models for coping strategies.
New Zealand started shutting down its borders very early with 14-day mandatory self-isolation, along with a ban on large group gatherings and shutting down schools. They then started a very hard 4-week lockdown from which they are just starting to emerge. There are no seroprevalence data as yet.
Currently, they have 1,469 known cases and 19 deaths. If this was scaled up to UK levels, this would equate to 20,566 cases and 266 deaths, which is vastly lower that what the UK has experienced to date.
Verdict on the Kiwi Model? One of the best results, but it comes with a very heavy economic price.
The Swedish Model
Ever since the outbreak hit Europe, Sweden has been the odd duck. Consistently going against the grain and ignoring hordes of clinicians and epidemiologists who predicted doomsday for their reckless behaviour. And yet, Sweden is doing OK. What did they do?
- Moderate social distancing was advised (but not enforced)
- Focus was on protecting the elderly
- Businesses were advised to limit density of crowds
- Schools were NOT closed
What happened? Sweden does not count positive RNA PCR tests as “cases,” but rather calculates caseloads from patients presenting with acute symptoms to the emergency hospitals, known as “IVA.” The Health Ministry recently said: “It’s rather flat. Looking at Sweden and the number of IVA cases, the IVA cases have not moved. We have 15,000 new cases reported 1,158 people who have been treated at IVA so far.”
According to their regional testing, about one third of the Swedish population is positive and not sick while two thirds have yet to be exposed. That’s quite a high seroprevalence. A total of 1,536 sought medical care for COVID-19, with 1,209 people hospitalized. Also, no children got any form of COVID-19. They claim to still be working on a mortality count due to the virus. Data from Johns Hopkins puts the confirmed cases at 18,640 and deaths at 2,194, which, case-wise, is not too dissimilar to the numbers that the Swedish Ministry is reporting. On a population basis, this would equate to 125,000 cases and 14,700, which is less than the United Kingdom (UK) but not by a whole lot. It should also be pointed out that the population characteristics of Sweden are not too dissimilar to that of the UK.
Verdict? They did a bit better than the UK despite no lockdown.
The UK Model
This would be the example of not doing anything and then getting a nasty wake-up call one morning from your epidemiological modelers telling you that the Apocalypse is coming. To say that everything was conceived on the fly would be an understatement, but it’s a good example of what happened to many countries.
What do the stats look like? There are no current seroprevalence data but to date (Johns Hopkins data) some 20,800 persons have died. Since the UK is quite a dense population, especially London, we could make a low and high estimate of the MR. If we use seroprevalences of 3% and 10%, then the overall MRs would be 1% and 0.3%, respectively. I would suspect the seroprevalence is closer to 10% than 3%, so again it is likely the MR is a lot less than 1%. Just to compare with the USA on a population basis, if the same proportion of deaths experienced in the UK had happened in the USA, the USA would have experienced 102,000 deaths. That’s a lot more than the 54,900 the USA currently has, even allowing for some undercounting.
Verdict: For a country that had to suddenly develop a strategy overnight, they paid a heavy price in terms of deaths and economics.
Soul Searching
I suspect when the first wave is over in the next few months, there’s going to be a lot of soul-searching:
- Did some countries under-react at the beginning and then over-react later?
- What will the true economic cost be of the shutdown?
- Why is it that over the past 20 years, we have done nothing about hundreds of thousands of flu deaths (i.e., they weren’t special as COVID-19 cases are)?
- Why didn’t we appreciate that the R0 (infectivity) could be so high (like 6 or more) in densely populated areas?
Marissa J Carter MA PhD MAPWCA
President
Strategic Solutions, Inc.
37 Voyager Lane, Bozeman, MT 59718, USA
Website: http://www.strategic-solutions-inc.com
Dr. Fife is a world renowned wound care physician dedicated to improving patient outcomes through quality driven care. Please visit my blog at CarolineFifeMD.com and my Youtube channel at https://www.youtube.com/c/carolinefifemd/videos
The opinions, comments, and content expressed or implied in my statements are solely my own and do not necessarily reflect the position or views of Intellicure or any of the boards on which I serve.
Look at the Czech Republic model. The were very successful.
Dr. Carter:
As a person with some understanding of exponential sequential growth modelling, it seems to me that the ultimate goal of management of any infectious disease outbreak is to bring the R nought below 1 (or optimally as close to zero as possible) locally and globally. The R nought is, when it can be approximated over long time frames, the best measure of success in controlling the spread of any infectious disease.
Dr. Goldberg
Dr. Goldberg, yes it is. But what is the best way to get there?
Maintain social distancing, have a co-ordinated national approach, ramp up contact tracing measures and test. These are all things that both Australia and New Zealand are doing. COVID19 is a test of both the leadership and governments of all countries right now. Finally, there is a real need to stop the confusing messaging within the United States and focus on what matters, but unfortunately this is not occurring.
Dr.Carter:
I hope that the following does not sound like rambling.
It’s rare when a biomathematical concept can compactly (but approximately) describe a given complex situation. Such a concept is R nought which compactly (but approximately) describes the general progression of an outbreak such as Covid 19 in any given population (as finding it depends on a host of known, approximately known, unknown or unknowable factors.) It seems to me the best way to measure it is to contact trace a small number of RANDOMLY chosen PCR tested individuals with mitigation in a series of given small areas and then periodically PCR test these same index individuals and all contacts and note who in these cohorts becomes symptomatic over time. This would give a series of various R nought cohorts. This data could then be combined using PCA analysis incorporating as much public health information as possible to give a rough estimate of the R nought in any larger population (and corresponding larger geographic area) so that specific further geographically suppressive strategies could be scientifically employed to decrease viral spread. This collective R nought could then be successfully measured and with specific mitigation strategies, decrease it over time. I am not being facitious, but I seem to recall a NUMBERS episode which used this concept in a similar vein.
Respectfully,
Dr. Goldberg
Yes, Dr. Goldberg, excellent idea. Alas, that kind of testing isn’t happening for a variety of reasons. Years ago when I did a lot of epi fieldwork we used to periodically sample small countries or regions of countries to estimate the prevalence of chronic disease and then resample after an intervention to start judging effectiveness. No reason why it shouldn’t be done at a county level in key states for viral antibodies. I suspect it’s a lack of testing, resources, foresight, and politics. Also for low prevalence you need large samples if you want decent 95% CIs. That’s what the statisticians are arguing about in CA.
Dr. Carter:
Thank you for your response. I see your point. One more well known (but simple) point that I would like to make is the following: that even the simplest exponential geometric growth model (eg finding a rudimentary R nought) quickly shows very large time dependent disparities in growth with very small changes in the constant geometric parameter. Maybe the lay public should be taught this fact (in addition to the flattening of the curve concept) so that they truly understand the reason that physical separation is so important.
Stay safe and stay well.
Respectfully,
Dr. Goldberg