Breaking my goal of radio silence on the current viral kerfuffle to note a couple developments:
- That last post, in which I tried to simply summarize the current state of the data being used to fan the panic, and make a few calm, reasonable conclusions, blew up this humble blog, thanks to a link on Instapundit by Sarah Hoyt (who reads this blog!?! Wow.), and subsequent links by some other people. Got a year’s worth of traffic over about 36 hours. A lovely discussion was engendered in the combox. Thanks to all who participated.
- Got told I was virtue signalling. By analyzing the numbers? I guess? And to whom would I be signaling, exactly?
- While I did spend a little longer than usual on that post, getting some links and trying for a less impressionistic logical structure, I’m still not entirely happy with it. It could be better. That’s what the following post is. I didn’t put in all the links this time.
- Italy seems to have clearly turned the corner. The sawtoothiness (I insist that’s a real word) of the numbers in the US still leave things in doubt, but, if I’m right, we should see a serious downturn in deaths within 3 weeks, probably sooner. (Note: when I say ‘if I’m right’ I’m just one guy looking at numbers and history, helped essentially by the likes of William Briggs and Malcolm the Cynic, and others reporting from the field. I hasten to add that any errors are all mine, but these and others pointed me in the direction of interesting stuff.
After this brief post, I’ll lay off COVID 19 for the next 2-3 weeks, and go back to blogging on education history, SF&F, and the Eternal Endless Front Yard Brick Project of Doom, subjects, historically speaking, 20K+ people won’t come here to check out. Be true to your school. So:
Reframing the question:
How likely am I or someone I love to get infected with the coronavirus, to get sick, or to die? How bad is this thing?
Spoiler: looking at the numbers, at history, and using common sense, the answers are: Not very, very unlikely, vanishingly unlikely. You probably won’t catch the virus; if you did, you almost certainly will either not get sick at all (about 50% of the time) or get sick very little (about 48.75% of the time). If you were healthy before getting the virus, chances are near 100% that you will not get very sick and you will not die from it.
Near 100% means there will be exceptions, but that these exceptions, added together, are a tiny percentage of total deaths, in the ballpark with ‘get eaten by a shark’. Certainly, every death is a tragedy. We live with potential and actual tragedy every day, and take precautions based on how bad we think the risk is. Most of us look both ways before crossing the street, but few of us wear a helmet in case we get hit by a meteorite. There’s risk here, as everywhere.
This is what the data says, intelligently understood. Judging by almost everything you read or hear, people certainly don’t understand any of this.
Let’s start with deaths. The key steps:
- Your likelihood of dying of COVID 19 or any other disease if you were to catch it is: the number of people who die from the disease divided by the number of people who catch the disease. You must grasp this simple fact before you go on.
- the Case Fatality Rate (CFR) is NOT your likelihood of dying if you catch the disease. That is simply not how it is defined. The CFR is: the number of death certificates listing COVID 19 divided by the number of diagnosed cases of COVID 19.
- The two components of the CFR do not correspond to the two components of the formula for your likelihood of dying:
- the number of death certificates listing COVID 19 would include, for example, a man in England diagnosed with COVID 19 who gets hit a lorry crossing the street; an 85 year old woman in Italy with lung cancer if she is suspected of having COVID 19, a seriously ill American given 2 weeks to live who catches COVID 19 and dies. Each nation and sometimes areas within nations have rules about what must be included on death certificates; it’s not consistent, but strongly tends to overstate COVID 19 as a cause of death. One Italian doctor said that the virus was a significant cause of death in at most 11-12% of such cases.
- the number of diagnosed cases of COVID 19 includes only people who were infected AND went to the doctor. 95% of cases have minor symptoms that look a lot like a cold or a flu; some large percentage of people don’t go to the doctor for that level of sickness. Such people don’t show up in case count. Then, 50% of infections are asymptomatic – no symptoms at all.
- Thus, the CFR includes too many deaths and too few infections to be an accurate gauge of your or my or anyone’s risk of dying from COVID 19.
Let’s plug in a few numbers, just to put a stake in the ground to argue about: say only half of the people with minor symptoms go to the doctor. Half going to the doctor seems high to me, because I’m not going to see the doctor for sniffles, achiness and a low fever – I’m popping a couple Tylenol and going back to bed to sleep it off. But maybe that’s just me. So let’s use half. This means we can effectively double the number of cases, since these slightly sick ones make up 95% of the total cases. In other words, if 100 people with minor symptoms go to the doctor, we’re assuming for the sake of argument 100 more are sleeping it off or taking Dayquil and muscling through it. So, in this example, there are 200 infected people but only 100 cases.
Applying this to the world, at the moment there are about 1.3 million cases – so, to start, just to allow for people who don’t run to the doctor with minor cold or flu-like symptoms, we will double that number to 2.6 million people.
Next, we notice that 50% of infections are asymptomatic. We know this because in a couple places – Iceland, Israel – authorities have tried to do broad testing, not just testing people who are sick, and found that half the positive test results are from people who aren’t sick. Therefore, for our 200 people who are sick, there are likely another 200 people who are infected but are not sick. So, we can double the number of infected worldwide again, from 2.6 million to 5.2 million. That becomes our baseline number for infected people.
Next, about 73, 000 people have died and have COVID 19 listed on their death certificate as I’m typing this. But we know in some cases, COVID 19 is not the cause of death (England, where the law requires COVID 19 to be listed on the death certificate regardless of whether it had anything to do with the death or not), and suspect it’s not the main cause in most of the others – that’s what saying it overwhelmingly affects vulnerable populations boils down to in plain English: it won’t kill y0u unless you’re already sick. Let’s use the Italian doctor’s high end: in 12% of these deaths, COVID 19 played a significant part in killing the person. Note that this is what normal people really mean: when grandma is fighting cancer and catches a cold and dies, we don’t blame the cold. We all know that, if she didn’t have cancer, the cold probably wouldn’t have killed her. All we’re doing here is applying that common sense thinking.
So, 12% of 73,000 is 8,760. That’s how many people have been killed by the COVID 19 virus. Note that if we limited this number to vigorous, healthy people who came down with virus and died, we’d be talking double-digits. That’s what the numbers say. But we’re including people here who, in the opinion of that Italian doctor, might not have died, at least not as soon, if they hadn’t caught the bug.
Grand finale: the number of people who have been killed by the virus according to a common-sense understanding of what that phrase means: 8,760. A reasonable estimate of the number of people infected: 5.2 million. To determine death rate, divide the first number by the second number: I get here an estimated death rate from COVID 19 of 0.17%. Not the 4.5% CFR that is commonly tossed around, which is about 25 times too high; not 2.5% in the US which would be a 1-in 40 chance of dying if you got the bug, but 0.17% – or less than a 1-in-500 chance.
You should be able to see that the same logic drives the possibility you will get seriously ill (and run the risk of permanent lung damage) if you catch the virus way, way down. Not very likely, roughly the same as your chance of dying of it.
Note: This is just one educated pass at making more sense out of the official numbers based on the definitions of those numbers. If you don’t like the way I adjusted the numbers, say why it’s wrong, and offer an alternative. Just don’t foist off the CFR as if it’s somehow a purer or better number. It most certainly is not. Numbers of infected will ALWAYS be an estimate, unless we test everybody in the world – not practical. Cause of death will often be a judgement call on the part of the doctor or coroner, so these numbers, both the ‘official’ counts and my adjusted totals, will contain a lot of slop. The point here is that using the CFR as if it represented the real risk of dying is ignorant and irresponsible. It is sloppiness by most people, and criminal fearmongering by those who should know better.
How about the risk of coming down with COVID 19? We’re told it’s spreading out of control! People who not only could not factor a simply binomial equation, but don’t even know what the terms mean, are sure the virus is spreading ‘exponentially’, and that that is very bad thing, that it means we’re all going to die, more or less.
But that doesn’t seem to be what’s happening:
Italy, specifically, Lombardy, has been the poster child for The End of the World as We Know It. Here’s what cases and deaths there look like today:
New cases are tricky: they are functions of how much testing is going on and what qualifies as a case (a moving target, as mentioned in the last post) at least as much as how much the disease is spreading. Deaths, while morbid by definition, are somewhat more binary:
In Italy, a nation with a notorious track record of not dealing with even the common flu very well, the explosive ‘exponential’ curve up and died. The politicians are already patting themselves of the back, setting us up to be saved from the next tragedy solely by their expert intervention. But Spring has also sprung – all that sunlight (and UV) and warmth are hell on viruses, so maybe it’s not the political action, or not entirely. Also, once the virus runs through the nursing homes and hospitals – notorious disease vectors, and site of most of the deaths – crassly culling the weak, the survivors are more likely to be resistant. So now the virus has to get out in the sun and infect healthy people – it doesn’t seem to be very good at that. Most viruses aren’t.
The same is happening elsewhere now. The US is a week, maybe two, out from the same factors causing a collapse of cases and deaths – and from politicians patting themselves on the back for making ‘hard choices’ for all of us, like deciding the less affluent don’t need jobs, and retired people need a lot less money in their mutual funds.
So, no, you’re almost certainly not going to get COVID 19. So far, about 0.07% of the world’s population, using my adjusted numbers for many more infections than cases, has gotten it. That’s less than 1 out of every thousand people. For comparison, you are about 10 times more likely to catch malaria, which kills at least 750,000 people a year, than coronavirus.
But it could come back! So could the Black Death. Yes, viruses mutate, sometimes – rarely – into more virulent forms. Is that likely? Any more for the coronavirus than any other virus? In its current forms, the evidence is that it’s done for; it might make a resurgence in the fall – that sometimes happens – but worrying about that isn’t going to make anything better, and, as shown above, it likely won’t be that big a deal if it does.
Unless there is a reason to think it different, history – what has actually happened to viral infections in the past – would say: the coronavirus will die out once spring is in full gear.