|
Check back for updates and a companion video to this article by John Cassis, CRC Originally written: April 14, 2020, 7:42 pm ET Originally published: April 20, 2020, 2:16 am ET Last revised: April 21, 2020, 11:00 pm ET Summary: With the release of testing statistics on COVID-19, we can now calculate the True Mortality Rate for the virus. The commonly used calculation of total deaths / total infected, which is 4.2% as of this writing, overstates True Mortality Rate in the US by over 100 times—the result of an overlooked logic flaw in that equation. The current True Mortality Rate in the US is 0.0393% or about 4 deaths per 10,000 people, though this number will continue to increase through the end of the COVID-19 event. In contrast, the 2018-2019 US flu season showed a mortality rate—based on the correct calculation—of 0.0963%, or about 10 deaths per 10,000. COVID-19 is currently 59% less lethal than last year’s flu in the US, and is on track to remain equally or less lethal than the flu by the end of the current pandemic. In a worldwide sample, the mortality rate is approximately 0.0752%, or just under 8 deaths per 10,000, based on 7.4 million tests so far. Though aiming for a final rate somewhat higher than the flu’s, the worldwide True Mortality Rate is nowhere near the 7.3% resulting from the commonly used calculation. The infection rate of COVID-19 is nearly twice that of last year’s flu: about 20% of the US population is infected vs. 10.9% for the 2018-2019 flu. Reasons for this difference are discussed herein. This report presents an analysis of publicly available data. If you like numbers and data, read on and enjoy! Transparency. It’s amazing what you can discover when they give you all the data you need to figure things out for yourself. Why, you might even be able to prove that COVID-19 is far less lethal than the 2018-2019 flu! We’ve all “felt” that the estimated mortality rate of 1-3% was a bit high because not everyone who was sick was included in the case numbers. But all we had was a “hunch”, and it’s risky making decisions affecting billions of people on a hunch. Well “hunch” no more! We can now calculate a True Mortality Rate (TMR). Using information publicly provided by the health reporting agencies and aggregated on worldometer.com—which now includes the number of people tested for COVID-19—we can mathematically calculate the coronavirus death rate in the US, in US states, and in many other nations. Understand that this report is not scientific and does not the create data, it only uses data (believed to be reliable) supplied publicly by health and other authorities. To date, most estimates of COVID-19 mortality involved taking total reported deaths and dividing that by the total number of reported cases (let’s call this the old calculation, or the “Common Method”). Based on the limited data we’ve been provided, it seemed like a most straightforward approach. This has given us mortality rates that have increased over time. In the US we initially saw a rate of around 1% to 2%, but then watched in quiet terror as it increased daily to the current calculation of 4.2% (and still rising). Residents of Spain, France, Holland and Sweden have been terrified watching their reported mortality rates rise to around 10%, while those in Italy and the UK have been faced with a heart-stopping 13% reported mortality rate. Some of our more fancy thinkers went a different direction and measured deaths in proportion to the sum of recovered-cases-plus-deaths (let’s call this the “Fancy Method”), generally providing an even higher mortality rate. But the methods for determining when a patient has “recovered” have not been clearly presented, and are fraught with many irregularities so as to make that measure unusable. [35] Comparing Apples to Oranges But it turns out that these calculations have been completely wrong in a huge way, and the terrifyingly high US mortality rate has been overestimated by a factor of over 100x. There’s an enormous logic flaw in the assumptions to these equations. If you haven’t figured it out for yourself yet, you’ll be furious for not noticing once I point it out…it’s obvious and yet not obvious at the same time. Here it is: the total number of deaths is a POPULATION-wide number, whereas the total number of cases is a SAMPLE-wide number! Yes indeed, the decision makers of this pandemic have been comparing apples to oranges, and making civilization-impacting decisions from a grossly incorrect calculation. They’ve been comparing the number of deaths for 100% of the population to the number of COVID-19 cases for less than 1% of the population. Said another way, unless there are about 100 deceased and uncounted Americans for every one deceased American who has been counted, the Common Method of calculating mortality rate isn’t even close to correct. The root of the logic flaw stems from the fact that the number of deaths is exactly the same for both the whole population as well as for the small sample of the population that has been tested. But the deaths are only representative of the entire population of a country or state, and not of the sample. How do we know this is the case? Join me in a thought experiment. The Logic of Population-wide Mortality As of this writing (7:42 pm ET on April 14, 2020), we are told that in the US about 3 million people have been tested and about 611,000 cases have been diagnosed, with just under 26,000 deaths. The 611,000 cases are the result of a SAMPLE of 3 million people. Now what if the US had tested 30 million people as of now instead of 3 million, so ten times as many? Well, we would expect to find more infected people, maybe roughly ten times more give or take, so around 6.1 million instead of 611 thousand. But we would not expect to find ten times more people who have passed away. Why? Because of a very obvious reality: with few exceptions, everyone who dies gets counted and gets assigned a cause of death. Deaths don’t just go unrecorded in very large numbers (and both deaths and causes of death have received record scrutiny during this pandemic). Let’s continue our thought experiment. We now expect that everyone who passed from COVID-19 has already been counted or nearly so, no matter if we’ve tested 3 million or 30 million or 300 million. If number of deaths were based on the sample size, then we would expect the number of deaths to increase if we increased the number of people we tested. But testing, in and of itself, won’t affect the number of people who have passed away. We won’t find ten times as many already-deceased individuals if we increase our testing tenfold. So deaths are not a sample-based number. An exception to this truth would be if the health reporting agencies were goal-seeking (i.e. fabricating) the death count based on number of infections (relax…I’m not suggesting this is happening, but a complete analysis must consider alternate possibilities, no matter how unlikely). But assuming that’s not the case, number of deaths would be unchanged regardless of the number of people tested. Therefore, number of deaths is related to the POPULATION. The Common Method compared the number of deceased (a population-based figure) to the number of infected (a sample-based figure). But we can’t do that…the resulting ratio just won’t have meaning and certainly won’t be accurate. What we need to do is to compare the total number of deaths for a population with the total number of COVID-19 cases for that same full population (which is exactly the way the CDC has presented its flu statistics for at least the last nine years). The Most Correct Mortality Calculation To Date Without testing everyone in a state or country, the total number of people with COVID-19 in the entire population at any given time can’t be exactly determined down to the person. But we’re in luck! In the USA, we’ve now tested so many people for COVID-19—nearly one percent of the total population in fact—that we can make a very reasonable estimate of the infection rate across the entire population. Many states and countries have also tested a significant percentage of their populations, with that percentage growing every day. And with sample sizes in the hundreds of thousands and even millions, the statistical probability of a sample result close to the true number for the population is pretty good (statisticians, please feel free to quantify this). We can now use some simple math to make a meaningful estimate of total people infected in a population. Let’s start with the infection information for the United States. As of 7:45 pm on 4/14/2020, worldometer.com reported the following (Figures 1 and 2):   Comparing COVID-19 to the Prior US Flu Seasons We now have a fairly concrete mortality rate that we can compare to last year’s flu mortality. Why is this important? Because last year’s flu mortality rate did not cause world and local leaders to order citizens into lockdowns and businesses into closure. Practically nobody altered their routines over last year’s flu…those who were sick stayed home, those who weren’t didn’t. But this year, civilization as we know it was effectively canceled, and people were pressured into distancing from and being afraid of their friends, family, and neighbors. Surely that could only happen if COVID-19 were believed to be much more lethal than last year’s flu, right? So for these reasons, a comparison of mortality rates is imperative (Figure 3).  But what if the 2018-2019 flu season were a fluke? Well, it turns out that it was: it was among the least deadly of the last nine years! Behold more calculations (Figures 4 & 5) from data on the CDC’s website [5][6]:   It’s unclear if the data we have on COVID-19 is of the same quality as the data estimates used by the CDC in tabulating prior flu seasons, but COVID-19 certainly has received more than its fair share of attention and resources. I make no representations on the methodologies for any of these tests. Calculations in this analysis are made using only the data provided by the reported sources. Comparing the True Mortality Rate with Calculations Used To Date Now let’s compare the True Mortality Rate, which uses only population-based measures, with the Common Method and the Fancy Method, which use mixed sample-based and population-based measures [3][4]:  QED True Mortality Rates for Select States and Nations We can take this same analysis and use it to determine mortality rates for select states (Figures 7 & 8) and countries (Figures 9 & 10), as well as collectively for all tested countries, again using data from worldometer.com between 7:42 pm and 9:30 pm on 4/14/2020 [1]-[4],[7]-[16]:    Make no mistake that, while countries like Italy are experiencing COVID-19 mortality well above that of the 2018-2019 US flu season, it is still nowhere near the Common Method calculation (Figure 10). For example, while the Common Method mortality rate for Italy is 13%, the True Mortality Rate is 0.2302%. Those same figures for Spain are 10.49% and 0.1346%, respectively.  Data from Cruise Ships and Navy Vessels While we empathize with the misfortune of those stuck on certain cruise ships and navy ships experiencing COVID-19 breakouts, they may be comforted in knowing that their suffering gave us data confirmation we could not have otherwise. During the term of their coronavirus infections, these ships served as a closed system where, in general, no one got in or out until they were tested. We have testing statistics for five such cruise ships (Figure 11):  Since passenger ship demographics were likely very different from naval ships’, let’s consider them separately, starting with the former. The infection rate on the cruise ships averaged 20.9%, which is very close to the 20% we’re seeing for the US and the 18.7% we’re seeing worldwide. The Diamond Princess, the only large ship for which 100% of the population was tested, produced a similar 19.2% result. But the 1.3% true mortality is far higher than our 0.04% estimate for the US (33x higher, in fact). We don’t have the data needed to determine the reasons for this difference. But we can expect that the typical age of cruise passengers might be greater than the typical age of the US population in general. Further research does in fact prove this out (Figure 12):  So how can we analyze this? Let’s begin by finding mortality rates by age range, which turns out is quite limited. Two studies of the initial outbreak in China produced a mortality table by age range that has been published for some time on worldometer.com. These figures are represented in Figure 13, along with extrapolations of expected Common-method mortality rates for cruise ship passengers, sailors, and the US population. One important note: the snapshot from the two Chinese studies produced a mortality rate well before the end of the COVID-19 outbreak in China, so we could expect our extrapolated mortality results to be lower than actual season-end Common-method mortality results.  Worldometer.com also provided the methods for calculating case fatality rate [24], which turns out to be based on the Common Method and Fancy Method previously mentioned. The age statistics shown above are based on the Common Method, so we will employ that method in our analysis of the ships. The Common Method mortality rate for the cruise ships was 1.5% (Figure 11). However, when applying the age-based mortality from Figure 12 to the age distribution of cruise ship passengers from Figure 13, we get an expected Common-method mortality of 2.5% for those cruise ship passengers & crew (Figure 13). Thus, the actual Common-method mortality rate on the cruise ships was below what we would have expected for the estimated age distribution on those cruise ships. The mortality rate on the cruise ships remains a bit puzzling, though. The Common-Method mortality of 1.5% for the closed-population cruise ships (Figure 11) is multiples below the Common-Method mortalities of 4.2% for the US (Figure 2) and 7.3% worldwide (Figure 10). Yet if we use the true mortality rate of 1.3% for cruise ship passengers & crew, it is multiples above the 0.0393% for the US and 0.075% worldwide. It is convenient yet accurate to say that the number of deaths observed on the cruise ships was a statistically small number compared to the data we have since compiled on state-wide and nation-wide levels. So perhaps the death rates produced by the cruise ships are not especially usable. But with over 5,000 COVID-19 tests administered on the cruise ships, the sample size and the resulting infection rate statistics can be argued to be much more significant. Thus, we can reasonably rely on at least the infection rates from the ships (and that is more convenient also, as infection does not appear to discriminate by age or health the way that mortality does). Now let’s look at the Navy ships (Figure 14). We see that the infection rate of sailors averaged 23.5%, which is in line with, if not slightly above, our US and world calculations. And with over 6,000 tests administered, we can say that the infection rate appears statistically meaningful. The mortality rate of Navy sailors averaged 0.061%, nearly identical to the worldwide average of 0.075% (despite the median age of US sailors being about eight years younger than the median age of the average American). However, with a mortality count of one for the naval ships, a meaningful conclusion can’t be drawn from the conveniently similar mortality rates.  Future Trend of True Mortality Rate Another thing we can do with our True Mortality Rate is analyze it over time. We can capture the same data from day to day and see the trends for any population. As long as the infection rate is mostly unchanged, the True Mortality Rate can be expected to increase, as the cumulative number of deaths in the population rises while the total population remains nearly unchanged. But as we pass the peak of mortality, the True Mortality Rate will stabilize as we near the end of the COVID-19 season. Before analyzing trends in the True Mortality Rate, we need to consider some reports that have surfaced slightly after I started writing this report. We learned of possible changes in the way the US mortality rate could be tallied on or around April 14, 2020. [33][34] The CDC had previously issued an alert regarding methods used in determining deaths caused by COVID-19. [30] Also, there are reports that two heavily populated states [31][32] may be increasing reported mortality counts. We have in fact seen some unusual behavior in the reported US moralities, beginning with an outlying spike on April 14, 2020 (Figure 15), which could weaken the integrity of our data hereafter.  It appears that, in an attempt to maintain some continuity and integrity of data, we should consider making some adjustments to figures reported on or after April 14, 2020, as we update the True Mortality Rate hereafter. Figures 15 and 16 show two possible methods: one that permanently and prospectively removes the strange spike in deaths reported on April 14th, and one that further adjusts the suspiciously overall higher death rates reported from April 15-17, compared with the clear declining trend in deaths reported from April 11-13.   Why Is the True Mortality Rate of COVID-19 So Much Less than Last Year’s Flu’s? So how can COVID-19, in state after state and country after country, be so much less lethal than recent US flu infections? We will explore several possibilities, but we will leave the most likely answer—an incomplete COVID-19 season—for last. One politically popular suggestion for the lower mortality rate might be that the shelter-in-place, “social distancing”, and other mitigation orders imposed by many (but not all) states and countries must be the cause. However, these measures would only reduce the infection rate of the population, not the death rate. Last year’s flu season exhibited an infection rate of 11%, whereas COVID-19 exhibits an infection rate of 20%. Why is this so much higher? Could it be an error? Some countries have exhibited far lower infection rates (10.0% for Germany, 6.0% for Canada, and 1.6% for Russia, per Figure 10). Nonetheless, the 20% infection rate has been supported across several other countries and states, as well as on cruise ships and naval vessels. Furthermore, a recent small-sample (200-person) antibody test found a 32% infection rate among residents in Chelsea, MA (if this rate were proved to be the ultimate final infection rate, the True Mortality Rate would be cut nearly in half). [48] By reducing exposure to other potentially infected people, logic says that quarantining almost all of the American population should have resulted in drastically lower infection rates for COVID-19 compared to last year’s flu. But this did not happen. Also, states like Iowa [26] and nations like Sweden [27] that did not impose lockdowns have not experienced greater infection or death rates than those with such mandates (Figures 8 & 10). Further weakening the case that quarantine-style lockdowns worked is a recent government test that found sunlight destroys the virus quickly (especially in hot and humid environments). [25] But this can’t happen if infected people are locked indoors. Could this be why the COVID-19 infection rate is so much higher than the flu’s? On a side note, one mitigation effort in particular—face coverings—has given people a false sense of security. I’ve heard quite a few individuals lecture others for not wearing masks, and orders have been issued in places mandating their use. Yet at 0.125 microns [38] (and particles as small as 0.06 microns [40]) the average size of the COVID-19 virus is much smaller than the 0.3-micron openings in N-95 masks [39]. One study [41] that measured filtration effectiveness of particles 1.0 to 2.5 microns in size (so 3 to 10 times larger than COVID-19) found the following percentages of particles removed:  So even the N95 mask allowed through 10% of all particles that ranged in size from 3x to 10x larger than the COVID-19 virus. Also, the absence of an airtight seal around the edges of any face covering renders its use practically comical. Please forgive this slight digression in our analysis, but I felt it was important to present these facts. So back to our question: how come the COVID-19 death rate is lower than the flu’s? If our health agencies would provide us with more transparency, we could evaluate one particular and plausible reason. Love it or hate it, the drug combination of hydroxychloroquine+azithromycin+zinc has gotten a lot of attention and use during this pandemic. Could it be that it saved so many lives, its effectiveness resulted in a lower mortality rate than the common flu? For example, if this drug cocktail was administered widely and saved the lives of, say, 37,000 Americans, this alone would account for the difference in mortality rate between COVID-19 and last year’s flu (Figure 19). It’s also possible that varying degrees of administering the drugs between states and countries can explain the respective differences in mortality rates. Maybe this drug cocktail has made a difference in observed mortality and maybe it hasn’t, but we can’t accept or dismiss the possibility without the data. It would be very nice for the health administrations to provide it, if they haven’t already.  But we’ve saved the most likely explanation for lower death rate for last. The key reason that mortality is currently so far below the 2018-2019 flu season is because deaths for the COVID-19 season have not yet ended. Cases began in January and we are currently in mid-April. It is difficult to know if COVID-19 will be as seasonal as the flu and the common cold. The CDC noted that the 2018-2019 flu season was one of the longest in ten years, lasting 21 weeks from November to mid-April. [44] For comparison, The first US case was reported on January 21 [4], and twelve weeks have passed since that date. One study estimates a peak of “active” COVID-19 cases in the US around April 20th [45][46], indicating we’re about half way through. If the COVID-19 “season” were to last as long as the 2018-2019 flu season, and if mortality continues in a bell-shaped curve (i.e. doesn’t end abruptly, but rather tapers off as gradually as it increased), then we would expect the final True Mortality Rate for this season to be roughly double our current estimated rate. In other words, 3.9 deaths per 10,000, times two, equals about 8 deaths per 10,000: still around, if not a little below, the 10 deaths per 10,000 from last year’s flu. Incidentally, this same study correctly forecasted a peak of daily new cases in the US occurring around April 5-7. The case for a bell-shaped curve has been supported by research that also projects a peak US death toll of 60,308. [47] If we were to plug this 60,308 death toll into our model, we can estimate the final US mortality rate for the 2020 COVID-19 “season” (Figure 20): 0.0914%, or 9 deaths per 10,000…reasonably close to the 10 deaths per 10,000 for the 2018-2019 flu season, well below the 13 deaths per 10,000 average over the last nine flu seasons, and far below the 17-18 deaths per 10,000 from the 2010-2011 and 2014-2015 flu seasons.  One final but important consideration for the 0.039% US death rate is that it’s nearly half the current worldwide death rate of 0.075%, which is based on a much larger sample inclusive of the US data. Why is that? One important realization is that Italy and Spain, which are already well above the flu rate and contribute significantly to mortality in our worldwide sample of nations, encountered large mortality numbers well before the US. Let’s compare when each country crossed a major mortality milestone using the charts at worldometer.com. [3] The US surpassed 100 deaths on March 22. Italy’s population is 18% that of the US and Spain’s is 14%, so their equivalent milestones would be at 18 mortalities and 14 mortalities, respectively. Italy achieved that benchmark on March 3 (or three weeks before the US) and Spain on March 11 (or just under two weeks earlier). Both countries are much farther along the event curve than the US, so it’s not surprising they’re exhibiting higher mortality rates. But try as we might, we can’t intuit with certainty if the US or other nations will or will not approach the True Mortality Rates of Italy and Spain, even though previously referenced studies suggest not. Limitations of the True Mortality Calculation As with any analysis of a sample, there will be some sampling anomalies that spawn imperfect results. So now it’s time for some important disclaimers. Assumptions in this analysis that may or may not hold true in the samples evaluated include:
Conclusion Thanks to the transparency and data provided by the health agencies, we now have a great and statistically meaningful basis for evaluating the True Mortality Rate of COVID-19. And it’s nowhere near the reported mortality rates of 1% or greater. Many, many people have feared for their safety and the safety of their loved ones during this pandemic. This fear was largely rooted in a very high estimated mortality rate. Now that publicly available data confirms a mortality rate much lower than historical US flu mortality, hopefully readers everywhere can relax their fears at least a little. And hopefully reporting agencies and news organizations will reconsider reporting the inaccurate, fear-inducing Common Method with the much more accurate True Mortality Rate. We have trusted that our economies, businesses, social lives, and physical & emotional well-being have not been sacrificed for no good reason. But if the ongoing response to COVID-19 remains harsh, blind to the fact that the true COVID-19 death rate appears no worse than the flu’s death rate, that trust will be called into question. Finally, please challenge my analysis. If you see something that should be revisited or improved upon, please reach out. You may contact me by e-mail at [email protected]. I may not be able to respond, but just know that I appreciate constructive or supportive comments. I may follow up with updates and/or a Q&A. If you find merit in this analysis, feel free to share it with anyone you think should hear it (especially those living in fear…I completed this study in large part to bring them some peace). This matter is one of the greatest we’ve ever faced in our lifetimes, so let’s make sure that the people in charge are making decisions using the best information available. © 2020, Guardian Financial, LLC, All Rights Reserved Update 1 (April 21, 2020, 11:00 pm ET): Testing now covers just over 1% of the US population, with the infection rate holding just below 20%. The True Mortality Rate has risen to 0.067%, and is approaching the expected rate of 0.091% based on the projected peak morality of 60,308. New York and New Jersey are reporting exceptionally high infection rates—39% and 49%, respectively—as well as very high mortality rates. It is not readily apparent what is causing their experience to be so high, but the rest of the country does not appear to be reflecting the same. If the overall US rate were to be adjusted to exclude New York and New Jersey, we find a currently true mortality rate of 0.0469%. (Figure 22)  Appendix A – Key Calculations COVID-19 Data: % Tested = Tot Tested / Tot Population % Infected = Tot Infected / Tot Population Pop Infected = Tot Pop x % Infected % Mortality = Tot Deaths / Pop Infected % Mort-Old Calc = Tot Deaths / Reported Sample Infected Mortality per 10K = % Mort x 10,000 % Over/Under – COVID vs flu = (Mortality per 10K / 18-19 Flu) – 1 Old Calc/True Calc = % Mort-Old Calc / % Mortality Ships Data: Infection Rate = Cases / Tests Est’d Pop Infected = Infection Rate x # of Crew & Passengers Old Calc Mortality Rate = Deaths / Cases True Mortality Rate = Deaths / Est’d Pop Infected CDC Data: % Infected = Illnesses / Population Hosp Rate = Hospitalizations / Illnesses Deaths/Hosp = Deaths / Hospitalizations Deaths/10K = (Deaths / Illnesses) x 10,000 Adjusted COVID-19 Mortality Using Flu Infection Rate: Pop Infected = Total Population x 2018-2019 Flu Infection Rate Additional Deaths to Increase Mortality to Same Rate as Flu: Total Required Deaths = plug-in figure to cause Mortality per 10K to equal 2018-2019 Flu Mortality Additional Deaths = Total Required Deaths – Reported Deaths Appendix B – Sources       
0 Comments
|
AuthorWelcome to my blog! My name is John Cassis. I'm a business strategist, independent financial analyst (since 1986), and former commercial banker of over two decades. I founded Guardian Financial Consulting with the risk philosophy of “Protect First, then Profit”, to help entrepreneurs, real estate investors, traders, market investors, and individuals think through the “what if’s” before making the best decisions possible. ArchivesCategories |
RSS Feed
All rights reserved. Copyright © 2020-2021 Guardian Financial, LLC