Statistics

Coronavirus Update II, Stats & Predictions

The code to do the plots yourself is here, a post which should be read for all the caveats, explanations, and so on. The original coronavirus post is here, which talks about panics, planning and the effectiveness of quarantines—which are on the increase.

Today, some updates and cautions. The first is that the easy access to social media has spurred things along, people racing off in all directions. I can’t help thinking people want to believe the worst. Any possible good news, or positive interpretation, is received with mild hostility.

Here’s the weirdest one:

Gist: South Korean faux-Christian death cult seeks to bring about the apocalypse via viruses. I didn’t know these guys, but I do know South Koreans love themselves some funky religion. My experience is with the Dahn Yoga cult, which is neither here nor there as far as fevers go.

Loyal reader Mark points us to this Barron’s article: China’s Coronavirus Figures Don’t Add Up. ‘This Never Happens With Real Data.’.

I had it at one time, but the article disappeared behind a pay wall. Anyway, if memory serves, it featured a statistician complaining that the sort of naive model we’re using fit too well, which made him suspicious.

Recall in the previous post I said that if China wanted to cheat, it would have to use a model in the same genre as ours to produce good-looking fake numbers. Maybe they have, and maybe they haven’t.

Problem with looking at the cumulative numbers, pictured next, is that they’re automatically smoother than the daily new cases and deaths, which are much rougher. Meaning the cumulative model can give the false appearance of accuracy.

These numbers are as of about 8 PM EST 24, February. These are the reported numbers; if they’re wrong, I’m wrong.

About that: many assume China is lying because, well, they’re China. This isn’t a bad reason to suspect their reports, but it’s not conclusive. We have this in their favor: the numbers outside of China are behaving as if China was reporting accurate numbers. Here’s the daily data, with the fudge factor applied to the first three weeks (see the code page for more).

About that death peak, see below.

Yes, the virus is peaking and spreading in areas outside China. Which is exactly what we would expect after the infection spreads beyond the hot spot. Be careful looking at percentage increase reports, which tend to exaggerate or are used to frighten (see the original post on this).

Recall what happened with SARS. Peaked in China, then spread out, causing infections to peak well after the original peak. Also, and worse, the death rate was higher outside China. Canada, for instance, had a large number of cases, relative to everywhere else, and a death rate of 17%!

The explanation is obvious: cases spread less easily outside the hotspot, only sicker people get it, and they die easier. Gruesome, but that’s the way it is.

Now this new virus is about 10 times higher in case numbers and deaths, so far, so it is more worrisome than SARS. The good news is that these approximations look to be holding. The peak in China is over.

Again I say, Beware of Percent and Raw Number Increases! Here’s the headline: Iran announces 50 dead in Qom coronavirus outbreak. Sounds scary, but here’s the sub-headline: “Tehran says deaths date back to February 13, though first cases were only announced days later; Kuwait, Bahrain announce first cases in people entering from Iran”.

Holding back numbers for two weeks and then reporting them all at once makes it appear worse than it is. The death numbers pop up. Here’s the simple estimate of mortality rate, which will be likely under the true rate (see second post):

This is starting to increase because new cases are slowing down. The trick is to guess where it peaks, which will be the true mortality rate. 3-6%?

Yes, Chinese goods are likely to become scarcer. How scarce, nobody knows. But do recall the cases are on the decrease in China, so people will be going back to work in a week or so—if we believe the numbers. And, yes, the stock market here took a big hit.

I am not saying those who preach doom—I am, as all know, a connoisseur of doom—but I agree with the headline of this article: The Media Fail Miserably At Coronavirus Coverage.

Take Bloomberg and The Guardian, for instance, both of which just reported that about two-thirds of the planet could become infected by coronavirus. That hypothetical number came from an epidemiologist in Hong Kong.

Here’s the truth: Any projection about what coronavirus is going to do in the future is little better than wide-eyed speculation. Previous coronavirus outbreaks followed very different trajectories.

SARS (severe acute respiratory syndrome) appeared suddenly in 2002 and soon disappeared. Ten years later, MERS (Middle East respiratory syndrome) emerged and stuck around. A handful of cases still occur each year.

Given how far the Wuhan coronavirus has spread already, it is reasonable to conclude that it might continue, especially in poor countries with inadequate healthcare. But it’s also equally valid to conclude that China will get the outbreak under control and the virus will stop dead in its tracks inside wealthy countries. Or, the virus could become seasonal, like influenza.

Flu? As of 15 February the CDC says: “CDC estimates that so far this season there have been at least 29 million flu illnesses, 280,000 hospitalizations and 16,000 deaths from flu.”

In the USA alone.

To support this site and its wholly independent host using credit card or PayPal (in any amount) click here

Categories: Statistics

28 replies »

  1. Thank you for including the US flu statistics. People tend to ignore how DEADLY the flu can really be, yet panic with a new strain. I do wonder about the economic effect (the stock market drops at any sign it can–it’s a scare tactic and sorts out those who do not understand “buy low, sell high”–separating the wheat from the chaff) but beyond that I think the whole thing ranks up there with the invasion of the Reptilians and other conspiracy nonsense. Sensible precautions should be taken, of course, but this is no where near pandemic or apocalyptic. That doesn’t stop the conspiracy nuts, of course……

  2. Greetings from Shanghai.
    I have lived here for many years (I was even here for SARS). Things look pretty bleak these days, but we humans have a limited attention span and, sooner or later, even disasters get boring.
    I seldom leave my apartment. There are few places to go and plenty of temperature check points. So, I just eat a bunch of ramen and make up Gwyneth Paltrow jokes.
    The weather is softening up; often a herald of flu season’s departure. Fingers crossed….
    …and you should all get ready for a bumpy ride.

  3. “About that: many assume China is lying because, well, they’re China. This isn’t a bad reason to suspect their reports, but it’s not conclusive. We have this in their favor: the numbers outside of China are behaving as if China was reporting accurate numbers. Here’s the daily data, with the fudge factor applied to the first three weeks (see the code page for more).”

    That said, it appears East Asians are the most vulnerable to wuflu, with whites being the least vulnerable. So China may still be fudging the numbers quite a bit, but the numbers elsewhere could be lower due to genetic differences.

    Anyone for popcorn?

  4. Flu and pneumonia kill over 50,000 people a year in the USA and nobody gets excited. About 3 million people die each year in the USA. These numbers are from the CDC.
    Number of deaths for leading causes of death:
    Heart disease: 647,457
    Cancer: 599,108
    Accidents (unintentional injuries): 169,936
    Chronic lower respiratory diseases: 160,201
    Stroke (cerebrovascular diseases): 146,383
    Alzheimer’s disease: 121,404
    Diabetes: 83,564
    Influenza and Pneumonia: 55,672
    Nephritis, nephrotic syndrome and nephrosis: 50,633
    Intentional self-harm (suicide): 47,173

  5. The problem with death by well-known causes is that they are well-known. The Corona virus is very much an unknown, so it might have the potential to be as bad as the Spanish Flu of a hunderd years ago. That is a valid reason for keeping very close track. People won’t suddenly start falling of kitchen stairs in droves, so that cause of death is not a reason to worry about. People worry about finding more effective ways of preventing people from falling of kitchen stairs.

  6. Ray,

    The CDC reports:

    “In 2016, 623,471 legal induced abortions were reported to CDC from 48 reporting areas.”

    However, the CDC numbers are derived from counts reported to state health departments, but California, Maryland, and New Hampshire do not publicly report abortion totals.

    A more accurate source is the Guttmacher Institute:

    https://www.guttmacher.org/fact-sheet/induced-abortion-united-states

    They report every three years, and in their most recent reporting year, 2017, they estimate:

    “Approximately 862,320 [US] abortions were performed in 2017, down 7% from 926,190 in 2014.”

    Thus you were mistaken about heart disease being the leading cause of death in the US.

    But your point is well taken: a lot of things kill a lot more people than corona virus with minimal panic.

  7. And just to repeat a point I made on an earlier post, the population of Hubei Province is 58,500,000 according to Wiki.

    The death rate in China was 7.29 deaths per thousand population in 2019 according to Knoema. (Note — that death rate does not include abortions).

    Thus on an average day (before COVID-19) in Hubei Province approximately 58,500,000 x .00729 / 365 = 1,168 people died of all causes (not counting abortions).

    According to Brigg’s model/graph, deaths per day from COVID-19 peaked on Feb 15th at 125. The death rate thus increased ~10% at its peak, assuming that all the deaths were in Hubei Province and that the COVID-19 victims were not going to die anyway. That percentage is now declining.

    Not good, but not necessarily something that should cause worldwide panic either.

  8. Great piece Briggs. Question: Why do you think these viruses lose virulence away from the epicenter of the outbreak? That seems to be the key to the whole story.

    Also do you have a data source for new daily cases? I would like to track them. Intuitively i agree with your thesis.

  9. Ah found the source via following your links. So never mind about that. Very interesting reading. Still chugging along at 500-1000/day. Lots of individual clusters of 1-2 people which are making the headlines. These either metastasize or they don’t.

  10. The reason people are worried about new viruses is because no one is vaccinated for them. Everyone knows how to protect themselves from the flu: get a flu shot. It’s not perfect, but it usually prevents it from spreading like wildfire across entire countries.

    The new virus appear to be highly contagious, can have a long incubation period, and can survive outside of hosts for up to nine days, and poses a substantial mortality threat to infants and the elderly. Pointing out that other things, like the flu, have killed more people than this new outbreak does not mean that heightened concerns about the new outbreak are unwarranted.

    The best case scenario is that this turns out to be seasonal, but there’s already some evidence that it is not. For instance, there have been outbreaks in southeast Asia where it is quite warm. I’m not suggesting that we panic, but I think there are real reasons why this is a bigger threat than other outbreaks this century.

  11. “poses a substantial mortality threat to infants and the elderly”

    Elderly yes, infants no – or at least, not yet. I’m honestly more worried about needless panic buying & supply side issues than this being the new Spanish Flu.

  12. I believe the Chinese numbers are credible. I used to live there and made many friends who were medical students at the time and are now doctors. Though overburdened and sometimes unprofessional, they’re world class. Other things about the current situation and social climate in China makes me trust the numbers are valid.

    It should be noted that the virus has been smothered outside of hubei as a consequence of a big mobilization (paradoxically) which was equated with a war effort. Hubei itself is kind of an extraordinary situation. Epi-surveillance didn’t detect a new viral infection among the existing sick population until the end of December, and then it was incorrectly attributed to an spill from an animal to a human at a wet market. This was scientifically sound, but we now know the virus most likely did not leap to humans at the wet market and unfortunately the initial belief that it did confounded efforts to understand the nature of the virus and gave it room to breath. Because the 50% of the initial cluster were workers and mangers at the market (not customers) epidemiologists thought the virus may not be strongly transmittable from human to human, and contact tracing and testing focused on the market which masked the severity of the virus. Quarantine, though possibly vindicated, also may have contributed to a moral panic and drove hospital cross-infections very high.

    Basically, the situation “outside wuhan” is very attainable, it was attained in every single Chinese province. Another Wuhan will be unlikely to replicate at this point.

  13. J … Very interesting comment. I have a couple of questions. The big one, which I never see addressed, is why should the virus BE less virulent outside the home-area of outbreak (Wuhan in this case) in hr first place? Just based on the math to date it would seem that it is but that is not entirely clear because there were interdictive measures taken in other provinces.

    That brings us to the second question: It seems Iran may be a second possible extreme outbreak precisely because they are not doing quarantining. So what is it that is really keeping the numbers down outside Hubei: reduced intensity of the virus itself or changes in human behavior? And if the latter … does that mean the virus has to basically do a sweep of the world before fizzling (for this year at least)?

  14. Kgaard … Any mutation of a virus that is less lethal should spread more than a more lethal mutation because someone infected with the less lethal variant has more time to spread the virus. However, there has been very little time for mutations since the outbreak, so this general explantion based on evolution is not convincing here.

    If the virus is less viruelent out side Wuhan (I’m not sure it is), I’d suspect warmer weather. The disease is like a severe cold and cold has its season. The warmer it is, the fewer people get the cold

  15. Briggs I have been following the data flow at BNONews as well as the CDC situation reports. The numbers are still a bit elevated: 500 cases a day in China and 500-1000 elsewhere. We still are not seeing decline in Korea, Italy or Iran, which are the hotspots. Seems to me one or all of these need to get over the hump — while no other countries break out — to declare victory on this thing.

    I will say it is impressive that China’s numbers are going down. Though that almost doesn’t square with the stubbornness of the virus in Iran, Korea and Italy. It is as if the virus follows its own story arc in each country.

Leave a Reply

Your email address will not be published. Required fields are marked *