Fear Of The Unknown During Crisis Times

“The oldest and strongest emotion of mankind is fear, and the oldest and strongest kind of fear is fear of the unknown.” – H.P. Lovecraft

Like so many bloggers, I’ve been reading lots about the coronavirus, puzzling over numbers, looking at charts, double-checking facts. I don’t usually spend this much time studying secondary and tertiary sources. Since my beat is business and wealth building, I prefer to conjure up my ideas and advice from my personal experience. But in this case, I have no choice. If I’m going to write about the virus, I’m going to have to study it. And that means relying on information I am not qualified to evaluate. Are the data accurate? Is the logic sound? Are there missing pieces?  Reading the News Again: How Many Could Die?  For 20 years, I’ve read what news I read in the evening. I never wanted to deplete my morning energy by focusing on problems that were beyond my control. But for the past month, I’ve been starting each day by checking two charts: one that tracks the stock market and another that tracks the coronavirus. I have a detached curiosity about the stock numbers. But my interest in the coronavirus numbers is visceral and strong. By next week, if not before, everyone in the United States will know someone who has been infected. I already know a half-dozen. The pandemic and its economic aftermath is going to have a psychological effect on Americans that will last for the next 50 years. Since I began tracking the data, the number of diagnosed cases has gone up every day. So too, happily, the number of diagnoses. But the data point I find myself stuck on is the number of deaths. It has gone up every single day – and in the past week, at an increasingly alarming rate. So every day, I wonder: How many will die? Will it be millions? Will it be hundreds of thousands? Or will it be less than 100,000, putting the coronavirus pandemic in “bad flu” territory? Thirty days ago, the numbers were small but the projections were big. Based on the “consensus” opinion then, the virus had an infectious rate of 3 (one person infects an average of three others) and a case fatality rate (CFR – the number of deaths compared to the number of cases diagnosed as positive) of 6 (6% of those diagnosed die). Putting these numbers into probability calculations, the mathematical models I was looking at were projecting an infectious rate of nearly 100% of the population and a death rate of 6%. That amounted to a projected US death toll of about 20 million people (6% of 330 million). And that wasn’t counting the many more that would die indirectly from heart attacks, strokes, and car accidents because access to hospital beds and ventilators would be so limited. That was the direst study I found. Others projected the number infected would be 200 million, with a death toll of 12 million. The most optimistic projection as to number infected was 60 million, with a death toll of 3.6 million. On March 16, the Imperial College of London issued a report based on slightly lower infectious and case fatality rates. This report projected that the US death toll would reach 2.2 million by the end of August. Again, that wasn’t counting the indirect deaths. The most optimistic projection I saw at that time was a death count of 1.8 million (based on a CFR of 3 and 60 million cases). All of these projections were being reported in front-page stories and on every TV news show. And hundreds more were being discussed online, along with heart-wrenching human-interest stories and all sorts of conspiracy theories. I was spending four hours a day reading. And every day, I felt like I knew less than I did the day before. And Then I Figured Something Out…  What I didn’t know then was that most of those early death tolls were projections of what would happen if the virus continued to spread and kill at the speed and rate it had been spreading and killing up to that point. What those early calculations didn’t take into account was what epidemiologists call adaptiveness – the ways a population changes its behavior as awareness of a significant danger spreads. This includes all the things people are doing now to lessen the chance of catching the disease – washing hands, disinfecting surfaces, social distancing, and isolating. These behaviors slow the rate of contagion. With social distancing, for example, the infectious (or reproductive) rate declines. Instead of each victim infecting three others, that rate might drop to 2.5, then to 2.0, and so on. Once it falls below 1.0, the number of people that get infected starts going down. So too does the number of deaths. Another problem with those early mortality estimates was how they determined the lethality of the disease. The early numbers – first from Wuhan and then from Seattle – were very high: 6% and higher. The mainstream interpretation was that 6 or more of every 100 people that caught the virus would die from it. But that was wrong for several reasons. First, the deaths in the USA in the first two weeks were concentrated in nursing homes and cruise ships, where the average age of those infected was considerably higher than the norm. Since, as is the case with most viruses, this one is much more likely to kill older people and people with compromised immune system, one would expect those early fatality rates to be disproportionately high. In the state of Washington, for example, the first cases were in nursing home residents. That produced a highly distorted CFR. (At one nursing home, 34 of 81 infected residents died. That is a CFR of 42%!)  This anomaly, along with the data coming from Wuhan, is the reason the early projections for the US were between 6% and 12%. So that was the first problem: an overstated estimate of how infectious the virus is. The second problem was the way the early media coverage misunderstood the data the CDC (and other groups) were publishing about the lethality of the disease that coronavirus causes: COVID-19. The lethality of COVID-19 was expressed in terms of the CFR, which, as I said, is a ratio that compares the number of deaths to the number of diagnosed cases. It doesn’t take a degree in statistics to figure out what’s wrong with that: * Since the symptoms, for most people, are similar to the flu, many people that get it wouldn’t go for testing and, thus, wouldn’t be diagnosed. * Of those that would go for testing, any that didn’t have advanced symptoms and a connection to a carrier would be turned away because of the scarcity of testing kits. I asked a doctor friend of mine about this. My hypothesis was that if you could know how many people were actually affected, and compared that to the number of deaths, you would have a real fatality rate that was lower than the 3% figure being talked about then. He agreed. He said, “They call it the denominator problem.” It works like this: When you underestimate the denominator, you overestimate the numerator. Thus, for the reasons cited above, the denominator (cases diagnosed) is likely to be a gross understatement of the meaningful statistic (the number of people that actually have the virus). So why were they using this faulty ratio? “Because,” my friend said, “you cannot measure what you don’t know.” To make a scientific measurement, you must stick to the facts. In the case of measuring lethality, there are only two relevant facts: the number of cases diagnosed as positive and the number of deaths. In the beginning of the outbreak, the CFR will give you a rate that is higher, even considerably higher, than the real death rate for the reasons pointed out above. But as the days and weeks go by and you get a larger percentage of the population tested, this distortion will diminish. And that’s what has happened since I’ve been looking at it. The CFR in the US has dropped from 6% to 3% to about 1.7% today. Will that continue to drop? Definitely. Up to now, we’ve had just a fraction of 1% of the US population tested. As tests ramp up quickly, so will the diagnosed cases. And as the ratio of diagnosed cases to deaths increases (as it will), the CFR will continue to go down. To get to a realistic lethality rate, we have to take another guess: We have to guess how many Americans have the virus but have not yet been diagnosed with it. This is the denominator problem I mentioned above. Considering that 80% of those that get COVID-19 have mild symptoms, and that we’ve been able to test so few, my guess has been that for every person diagnosed, there were 10 others that had it but had not been diagnosed. A recent report I read that summarized estimates from top epidemiologists concluded that the percentage of diagnosed cases versus actual cases is 9%. Close enough. So let’s use my 10% guess to keep the arithmetic simple. What that means is that the number of Americans that have the disease right now (as I write this) is about 10 times larger than the diagnosed cases. Ten times the current diagnosed cases (139,061) is about 1.4 million. So now we have an “adjusted” infected rate of 1.4 million and a death count of 2428. And to get a realistic fatality rate, all we have to do is divide 2428 by 1.4 million. Right? But wait… there’s more Alas, no. There is another problem with the CFR: It doesn’t make sense to compare the number of deaths to date to the number of cases to date. That’s because people that die from COVID-19 don’t die overnight. Based on the numbers so far, it seems to take 10 days to two weeks. Therefore, the correct ratio should be the number of deaths to date over the number of cases diagnosed 10 days to two weeks earlier. This sounds like a problem that could be easily solved: Simply compare today’s death count against the number of cases diagnosed 10 to 14 days ago. But if you try that for several days in a row, you will see that the number you get keeps moving because you are working with two sets of numbers – death rates and diagnosed cases moving at the same time. So, no, we can’t arrive at a precise number. But we can arrive at a range. The comparisons I did since the beginning of the month increased the CFR by a factor of 2.5 to 4. That would make the lethality rate somewhere between 0.85% (0.34% x 2.5) and 1.36% (0.34% x 4). Okay, so that gives us a real fatality rate of as a range of 0.85% to 1.02%. How many will be infected?  Let’s move on to the other metric we need to estimate the death toll: the Ro or reproductive rate – i.e., the rate at which the virus will spread from one person to others in close contact. Like the case fatality rate, this one has been going up in the past month. Since I’ve been tracking it, it’s gone down from 3.0 to 2.3. A reproductive rate of 2.3 means that each person that gets the virus will infect 2.3 more. 2.3 might not sound scary, but take a look at how fast it turns into 2.4 million:

1. 3 x 2.3 = 5.29
2. 59 (5.29 + 2.3) x 2.3 = 17.4
3. 0 (17.4 + 7.6) x 2.3 = 57.6
4. 6 x 2.3 = 189.9
5. 6 (189.9 + 82.6) x2.3 = 626.9
6. 5 (626.9 + 272.6) x 2.3 = 2068.9
7. 2,968.4 (2068.9 + 899.5) x 2.3 = 6827.4
8. 9,795.8 (6827.4 + 2968.4) x 2.3 = 22,530.3
9. 32,326,1 (22,530.3 + 9795.8) x 2.3 = 74,350.0
10. 106,676 (74,350.0 + 32,326.1) x 2.3 = 245,355.1
11. 352,031.1 (245,355.1 +106,676) x 2.3 = 714,623.4
12. 1,066,645 (714,623.4 + 352,031.1) x 2.3 = 2,453,304
13. 3,519,949 (2,453,304 + 1,066,645) x 2.3 = 8,095,882

And that gets us to 11.6 million in just 14 exponential steps! That’s just 14 degrees of exponential growth to get from one infected person to more than 10 million. In a crowded city the size of New York, that could happen in a few weeks. In a city as dense and populated as Wuhan, it could happen in just a few days. That is the frightening part. With a Ro of 2.3, the coronavirus is a scarily fast moving bug. But that rate is not fixed. It’s dependent on its ability to move freely from one host to another. Without any barriers, it can grow at these rates. And that’s why some of the earlier articles on social media, the ones that were predicting that half to 100% of Americans would get COVID-19 were wrong. The way Homo sapiens adjust to threat is through adaptive behavior. This has been true since paleolithic times. In the case of the coronavirus, those adaptive behaviors include everything we’re being told to do: hand washing, social distancing, and isolation. Only a week or 10 days ago, the projections for how many Americans will contract COVID-19 ranged between 20 million and 200 million. Today, after accounting for the adaptive behaviors that are taking place, the upper end has come down to 60 million. The final calculations: And this brings us to our final bit of arithmetic. We simply multiply our estimate of the range of true fatality rates (0.85% to 1.02%) against this estimate of the number of Americans that will be infected. At 60 million and a 1.02% true fatality rate, we will have 612,000 deaths. At a fatality rate of 0.85%, the death rate would be 510,000. At 30 million, the death toll would be half of that – as many as 306,000 to as little as 255,000. At 20 million, the death toll would be as much as 204,000 or as little as 170,000. At 10 million, the death toll would be as much as 102,000 or as little as 85,000. That’s quite a range – high of 612,000 to a low of 85,000. Putting it in perspective: * The Spanish flu of 1918 killed an estimated 50 million worldwide. * The Asian flu of 1956 to 1958 killed an estimated 2 million. * The Hong Kong flu of 1968 killed about a million. Conclusion  Although assuming this estimated real fatality rate of 0.85% to 1.02% is correct and remains constant, how many Americans will die in 2020 depends on how many will be infected. And that means that all these drastic measures to reduce the number of people that get infected make sense. This is not the only possible strategy. Another idea, considered and abandoned in England, was to isolate only the most vulnerable and let the rest of the population get the virus since their chances of surviving it are very high – more than 99%. (Remember, the high true fatality rate of 1.02% included a good chunk of the population that is older and health compromised.) The reason that was rejected was because it would overwhelm the health care system, since some portion (maybe 10% to 20%) of that younger and healthier population would still need medical attention. Since we are implementing behavior adaptations and since the health, scientific, and business communities are working nearly 24/7 to provide the materials we now lack and come up with treatments and vaccines we need, I’m guessing that the death toll will be at the lower end of this range: somewhere between 85,000 and 205,000. So what can you conclude from this? That you’ve wasted 20 minutes paying attention to the arithmetic of a self-admitted know-nothing? That’s on you. I did this because I wanted to answer my own questions, as best as I could, rather than relying on some reputable institution or someone with a title and a degree (that may also have an agenda). My conclusion is that the response we are making – as drastic as it is – is the right decision if our goal is to avoid crashing our health care system and allowing hundreds of thousands of Americans to die that would otherwise not die. I feel sure we will get through this, but there will be – and has already been – a price to pay. The social shutdown we are living through will almost certainly change the hearts and minds of every person old enough to be aware of what is going on. As I said, we will all know someone that has been infected by – or has died from – COVID-19. But the impact of isolation and submitting to what amounts to police-state governance may be worse, leaving us with fears and trust issues that will not disappear soon. Then there is the financial impact. Our economy has virtually collapsed. And it may well slip into a general depression that will last for many years. Millions have already lost their jobs. And hundreds of thousands of businesses that are shuttered now will never again open for business. Our political system will change. I don’t know how. But you can see changes taking place already. Much of it will be bad as politicians from both sides try to take advantage of the situation to further their own political ideas and personal goals. But it’s not all going to be bad. There will be many that begin to understand some simple truths about the world we live in. That we are all, in the end, responsible for taking care of ourselves and our families. And that responsibility is not just about loving and caring but also about providing the financial resources that we need. That it is foolish to trust anyone or anything to take care of these responsibilities for us. And, as I’ve said before, that our government is not, and cannot be, our savior. For despite its efforts to do so, it cannot guarantee us anything that nature will not guarantee. And nature guarantees us nothing. One more thing…  In case this piece is syndicated to a large audience, which is possible given what’s happened in the past, I have to say this: My knowledge of epidemiology is a month old and an inch deep. And my math skills are rudimentary. I’ve shown you my calculations not to persuade you that they are right but to prompt you to do your own thinking. That’s why I’m spelling out the numbers. So you can review them, make your own calculations, and plan accordingly. To assist you in that, following are links to a few of the many studies,  articles, and models I’ve found helpful in researching this piece. * “The Doctor Who Helped Defeat Smallpox Explains What's Coming” – Epidemiologist Larry Brilliant, who warned of pandemic in 2006, says we can beat the novel coronavirus… but first, we need lots more testing. Click here to read the article. * “Will Coronavirus Ever Go Away? What a Top WHO Expert Thinks” – Click here to read the article.  * Bill Gates has spent much of his recent life working on global health issues. He’s now focusing some of his attention on the coronavirus.  Here he answers some of the most common questions. * Alex Tabarrok is an economist at George Mason University and blogger at Marginal Revolution. Click here to watch him speak on the official responses to past pandemics, which countries are doing things right, and how the government can get a better handle on stopping the spread of this disease. * “The COVID Tracking Project – US Historical Data” – Here is the tracking service I’m using. [LINK 3/26] * WorldoMeter is another data tracking service that’s following the pandemic. Click here. * Here are some early estimates based on China’s early results. * “Why coronavirus antibody testing in one town could provide a way forward” – click here to read the article. * While most of the countries in the Western world are mandating shutdowns and isolation, Sweden is taking another approach. It will be interesting to see how they fare. Click here to read a NYT article about the way they’re handling it.