Data & Numbers: COVID 19

Bits of the following have appeared in the past couple posts on this current coronavirus kerfuffle; for clarity and reference sake, thought I’d put them all in one place. Nothing original here, just compiling the information in one key place.

I plan to stop posting on this for the next 2-3 weeks, at which time it should be clear if I’m wrong or if I’m right. I will then crow/eat crow, depending.

Data or ‘Facts’

Before one can start talking about numbers, what exactly one is counting should be spelled out. This is called ‘being reasonable’ or ‘the scientific method.’ Note that all the dreadful numbers, death rates, how fast the virus is spreading, how infectious and deadly it is, all hinge on these few numbers. Knowing how they are created and measured is therefore critical to understanding what is going on.

Infected: Ideally, we should only count a person as infected if they have been tested for the infection at least twice. The unknown but almost certainly high number of false positives makes testing only once imprudent.

In reality, we do not now and never will know the number of people infected with the virus. It seems many people infected, possibly as high as one half, have no symptoms. They feel fine, don’t go to the doctor, and, except in unusual cases such as cruise ships and NBA teams, don’t get tested. It is from those unusual cases that the information that about half of infections show no symptoms, e.g., “Nearly half of Diamond Princess cruise ship passengers and crew who had coronavirus were asymptomatic when tested, CDC report says”

Of the 712 passengers and crew members of the ship who tested positive for coronavirus, 331 — or 46.5% — were asymptomatic at the time of testing, the CDC said.The agency said that the high rate of asymptomatic infections could partly explain the high rate of infection among cruise ship passengers and crew.

And in Iceland, where a larger percentage of the general population has been tested than just about anywhere:

[Iceland’s chief epidemiologist Thorolfur] Guðnason is quoted here

https://futurism.com/neoscope/half-coronavirus-carriers-no-symptoms as saying “Early results from deCode Genetics indicate that a low proportion of the general population has contracted the virus and that about half of those who tested positive are non-symptomatic.”.

hat tip to Foxfier

Point #1: Keep in mind that nobody knows how many people are infected, since asymptomatic people will never get tested except in somewhat unusual circumstances, and that it’s possible half or more of the people infected never get sick. The numbers of infected you hear or read are the outputs of models – not data – or otherwise pulled out of thin air.

Cases: Again, ideally, a case of COVID 19 infection should mean a situation where a person was tested at least twice and found to have the infection.

In reality, it appears many cases of COVID 19 infection are diagnoses without benefit of testing. In China, many cases of COVID 19 were diagnosed based on symptoms and CT scans:

On Feb. 12, the national health committee announced that it was changing the way cases were counted in Hubei province. People with symptoms of the disease and with evidence of pneumonia on a CT scan but who had not been tested for presence of the virus — in other words, people who had a clinical diagnosis — would be added to the list of cases.

That day, the country reported a massive increase in cases — over 15,000. But since then, the daily numbers have tumbled. On Wednesday, China announced it was reverting to reporting laboratory-confirmed cases only. When the National Health Committee issued its update for Wednesday, it reported a net increase of only 394 cases — the first time in weeks the daily case increase was under 1,000 cases.

That was then; now, in every country around the world, we have some largely unknown number of ‘unconfirmed cases’ – still counted as cases, but not confirmed by tests.

When diagnosis is based on symptoms, plus sometimes on travel and contact history, things get more and more uncertain: someone who has typical cold symptoms but has a friend who recently came back from a trip to China might become an ‘unconfirmed case’ and then, depending on the availability of tests, might not get tested right away. And that’s in First World countries.

Here are the CDC’s top symptoms:

• Fever
• Cough
• Shortness of breath

These symptoms are common to colds, flus, pneumonia, and probably more diseases.

Point #2: What a case *is* is not clearly defined, and is different over time and from country to country. Most important, cases are NOT equal to infections. Many, many more people may be infected than show up in case counts, however defined. DO NOT CONFUSE CASES WITH NUMBER OF PEOPLE INFECTED.

Deaths: Ideally, what we want to know is if COVID 19 killed somebody. Such a determination is clear, or as clear as it is likely to get, when a healthy person tests positive for the virus, and dies of respiratory failure after having progressed through some or all of the less serious symptoms.

Such cases are remarkably rare. Instead, we have COVID 19 deaths almost entirely restricted to vulnerable populations, people who have underlying health issues, including, especially, the overall weakness and infirmities associated with advanced age.

As in cases, it seems what constitutes a COVID 19 death may vary from country to country, and does not correspond to what a layman thinks when he hears ‘died of COVID 19’. Here’s how it is done in England:

Lee explains that in the U.K. if someone dies of a respiratory infection, the specific cause of that infection is not usually noted unless it is identified as a rare “notifiable disease,” not seasonal infections, like the flu.

“So the vast majority of respiratory deaths in the UK are recorded as bronchopneumonia, pneumonia, old age or a similar designation,” writes Lee. “We don’t really test for flu, or other seasonal infections. If the patient has, say, cancer, motor neurone disease or another serious disease, this will be recorded as the cause of death, even if the final illness was a respiratory infection.” The result, he says, is that U.K. certifications “normally under-record deaths due to respiratory infections.”

But COVID-19 is now included in the updated list of “notifiable diseases,” Lee explains. “That means every positive test for COVID-19 must be notified, in a way that it just would not be for flu or most other infections.” If any patient dies after having tested positive for COVID-19, “staff will have to record the COVID-19 designation on the death certificate — contrary to usual practice for most infections of this kind.” That might give COVID-19 “the appearance of it causing an increasing numbers of deaths, whether this is true or not,” writes Lee. (Read Lee’s op-ed here.)

Similar practices, or practices with the same result of over reporting deaths, seem to prevail elsewhere around the world. New guidelines for the US on how to fill out the cause of death on death certificates were released just yesterday. Minnesota Public Radio says “But new guidelines from the National Center for Health Statistics, which Minnesota follows, will err on the side of pinning more deaths on COVID-19, at least provisionally.”

Point #3: What the medical reporting means by a COVID 19 death is not what we commoners mean by it. As reported, a ‘death’ means ‘died while infected with COVID 19’ not ‘died OF a COVID 19 infection.’

Numbers and Percentages

The first thing to note, and the thing to keep in mind, is the vast amount of uncertainty in all this. We don’t know, and are very unlikely to ever know, how many people are infected with this virus. The number of cases can, and repeatedly has, shifted as the definition of what constitutes a ‘case’ has shifted. Deaths are in the same boat: Yesterday, for example, France added 884 deaths from some unknown number of cases in nursing homes, since they had not been counting deaths and, presumably, cases, from all their nursing homes up to that point. So, a major locus of coronavirus deaths – nursing homes, full of precisely the most vulnerable population – is not (they’re still counting it up) accurately included in the reported totals.

It would beggar belief to think France’s counting issues are unique or even rare.

Let’s revisit the key numbers:

How many people are or have been infected with the coronavirus so far? What is the peak likely to be? The honest if uncomfortable answer is: nobody knows. Yet, all sorts of numbers, ratios, and percentages that include some guestimate of the number of infected people get tossed out as if they are facts, or at least, strong enough guesses that policymakers are urged to impose martial law and cripple the economy based on them. One such estimate, as discussed here, was that 200 million Americans would get infected over the next 200 days. Baked into this number is the assumption that 100% of people are susceptible to infection.

Are we? On the Diamond Princess, only 18% of the people on board got infected; in Iceland, only about 6% got infected. These are places where something like a population was tested; all the national numbers you hear, like Italy’s or China’s, are almost entirely based on tests administered to a tiny subset of people with symptoms or with exposure to people with symptoms. We would expect much higher infections rates as determined by tests when all your test subjects are only people who you have good reason to think have the disease.

So, assuming 200 million infected people in the US, 60% of the population, is ridiculous. Then, applying a death rate based on one subset of cases, the author projects 11 million deaths. This scary scenario is created by applying a ratio of deaths to cases to a projection of number of infections pulled out of a hat. For about a week, this number – 11 million dead Americans if we don’t act now! – was treated as Gospel; then it became more like 500,000 dead Americans, despite actions being taken not nearly as drastically nor as quickly as demanded. Now? I’m still hearing numbers in the neighborhood of 250,000. Do not be surprised if the worse case numbers bandied about fall under 100,000, or, as I like to say, bad flu season level, over the next 2-3 weeks as Spring kicks into gear. We will pretend, often willingly, that we didn’t wildly overreact based on 11 million dead! and that we only had, say, 50,000 dead because we took such admirable, if extreme, actions as fast as we did.

We keep hearing about outrageous and terrifying death rates from COVID 19 infections: 4.5% in China, 11% in Italy, even 2.5% here in the US. Rarely do we hear it discussed that these are *CASE* fatality rates (CFR) NOT the percentage of people infected with the virus who die from it. The CFR DOES NOT represent our risk of dying if we catch the virus; rather, it expresses the risk of someone identified as a case having COVID 19 show up on his death certificate. That risk is much higher than your or my risk of dying from COVID 19.

To get some idea of our real risk of death from COVID 19, we’d need a better estimate of the number of people infected versus the number of cases. Then, we’d want to reduce the number of deaths to include only those where COVID 19 was the primary cause of death. Then, we’d have a number – inescapably, an estimate – that says how risky this whole plague is to the average Joe.

Can we do this? At least, get closer to reality than the CFR? All is not hopeless: we can, over time, refine our guess by intelligently comparing them to reality, or at least to much harder numbers.

Let’s start with the asymptomatic cases: about 50% of those who test positive in the general population (as in Iceland) for the virus show no symptoms. Applying this knowledge to the cases numbers means we can double the number of cases: if as many people who are infected show no symptoms as do show symptoms, and asymptomatic people are very unlikely to get tested and therefore show up as ‘cases’, that means there are (at least – more to come) twice as many infections as cases.

Next, the vast majority of cases show only mild symptoms:

from today’s Worldometer

Mild symptoms are largely indistinguishable from common cold or flu symptoms, as we saw above. So, how many people go to the doctor when they have the sniffles or a low fever? Compared to how many people just pop a couple of Advil, call in sick and climb back into bed? Or even take some Dayquil and go to work? Now, imagine you are a member of a less affluent society than ours: do you go to the doctor for what looks like the common flu?

So I would suggest that at least as many infected people with mild symptoms don’t end up as cases as do. This would push our infected estimate up to 4 times the number of cases, under the assumption that there are as many asymptomatic infections as symptomatic ones. This cut our real death rate to 1/4 of the CFR, or from 2.5% to 0.6% in the US.

(This also means only 1.25% – that’s 1/4 of 5% – of people infected with the virus end up with serious symptoms. 98.75% have mild or no symptoms.)

But that’s still not enough, again according to Dr. Lee cited above (He’s using British numbers for a couple weeks ago):

Also, we’re only dealing with those COVID-19 cases that have made people sick enough or worried enough to get tested. There will be many more unaware that they have the virus, with either no symptoms, or mild ones.

That’s why, when Britain had 590 diagnosed cases, Sir Patrick Vallance, the government’s chief scientific adviser, suggested that the real figure was probably between 5,000 and 10,000 cases, 10 to 20 times higher. If he’s right, the headline death rate due to this virus is likely to be 10 to 20 times lower, say 0.25 percent to 0.5 percent. That puts the COVID-19 mortality rate in the range associated with infections like flu.

Spectator

If Sir Patrick is even a little correct, and we’re only off around one order of magnitude, real death rate will be 10% of the CDR – 0.25% in the US, well within common flu range.

And this counts everyone with COVID 19 on their death certificate! Which means we’d need to push that number down some more. How much? I’ve heard estimates of as little as 11-12% of of official COVID 19 deaths where the viral infection caused or seriously contributed to death, but alas, I can’t find the reference now. The point remains: given all the other uncertainty and out and out misrepresentations of the numbers, even if the reporting inconsistency is small, COVID 19 remains, at worst, a little more deadly than a routine flu.

But what about…

“1 out of every 6 people infected with COVID 19 get permanent lung damage.“

You mean, one out 6 of the 1.25% of people with a coronavirus infection who show serious symptoms? Surely this can’t mean that 1 out of 6 asymptomatic people – that’s 1/2, remember – suffer such damage? When? Without any symptoms? Or even that 1 out of every 6 people with mild symptoms? Their mild symptoms end up causing serious lung damage?

If this claim is, as I suspect it is, based on people with serious symptoms, 1 out of 6 our of 1.25% would be 1 out of every 500 people infected with the virus. We’re getting into ‘killed by a tornado’ levels of unlikelihood. I’d like to see the details of this claim – 5 minutes of web searching pulled up nothing.

“Hospitals overrun! Healthcare collapsing!“

Did you know that Italy has a history of being overwhelmed by respiratory cases? They have had higher than typical death rates from flu outbreaks over much of the last decade.

Did you know Alabama had to declare a state of emergency in 2018 when a flu outbreak threatened to overwhelm their health care system?

Did you know NYC was home to the #WuhanStrong movement right before the outbreak got bad, where people were encouraged to hug and otherwise fraternize with those who had recently been to or come from China, in order to prove they were not evil raaacists like the Orange Man? This might figure into their problems.

In general, all healthcare everywhere makes capacity trade-offs. No one can afford to build and staff a hospital such that no emergency will ever catch them underprepared. They take an educated guess at what they’ll need, adjust it by what they can afford, try to coordinate with surrounding systmes, and hope for the best. And then, hey screw it up, because they are human. A local example: we in California know it is almost certainly just a matter of time before the next big earthquake or ARkStorm. You think our healthcare system could handle that? I’ll spare you the suspense: no, it could not. BUT – there’s no way our healthcare system could be maintained for years on end with the sort of capacity such disasters would demand. So they do what they can, and hope for the best.

“exponential growth!”

Not really.

from Worldometers

The red line is 1, meaning a linear type increase. The current trend, if you can even claim to make it out, seems pretty flat, with occasional dips below the line. What is clear is a lack of any clear exponential explosion.

Check back in 2-3 weeks. If this virus acts like a typical air- and surface-borne virus, sunshine and nice weather will be the end of it (with a possible reprise in the fall). But, while it is tragic when people die, you’d be a lot better off getting some exercise and cutting out some calories than in worrying yourself over this virus.

Of course, I could be wrong. I’m just a numbers guy. This thing could explode tomorrow, mutate into something really deadly, and kill us all. It just doesn’t look like it will, from the available evidence and history.

So, WHY is everybody panicking over this? WHY is dominating the news and causing us to commit economic seppuku? Those are very good questions.