Are many more people infected than we think?

Here is the Clive Cookson FT piece (with an irresponsible headline).  Here is the Lourenco new Oxford study, only a few pp.  Miles Kimball offers analysis and numerous references.

Here is the Bendavid and Bhattacharya WSJ piece that perhaps has had the biggest popular influence.  They argue that many more people have had Covid-19 than we think, the number of asymptotic cases is very large, and the fatality of the virus is much lower than we think, perhaps not much worse than the flu.  But their required rate of asymptomatic cases is implausibly high.

The best evidence (FT) for asymptomatic cases ranges from 8 to 59 percent, and that is based on a number of samples from China and Italy, albeit imperfect ones.  Icelandic data — they are trying to sample a significant percentage of their population — suggest an asymptomatic rate of about 50 percent.  To be clear, none of those results are conclusive and they all might be wrong.  (And we should work much harder on producing better data.)  But so far there is no particular reason to think those estimates are wrong, other than general uncertainty.  You would have to argue that the asymptomatic cases usually test as negative, and while that is possible again there is no particular reason to expect that.  It should not be your default view.

Marc Lipitsch put it bluntly:

The idea that covid is less severe than seasonal flu is inconsistent with data and with the fact that an epidemic just gathering steam can overwhelm ICU capacity in a rich country like Italy or China.

Furthermore, the “optimistic” view implies a much faster spread for Covid-19 than would fit our data from previous viral episodes, which tend to come in waves and do not usually infect so many people so quickly.

So I give this scenario of a very low fatality rate some chance of being true, but again you ought not to believe it.  The positive evidence for it isn’t that strong, and you have to believe a very specific and indeed unverified claim about the asymptomatic cases testing negative, and also about current spread being unprecedentedly rapid.

Here is Tim Harford’s take (FT) on all this, he and I more or less agree.

By the way, Neil Ferguson didn’t walk back his predictions.  That was fake news.

So we still need to be acting with the presumption that the relatively pessimistic account of the risks is indeed true.  Subject to revision, as always.


If we are more infected than we thought, the case for strong restrictions on movement is weaker and would suggest a lower overall mortality rate than reported.

An unwise and unwarranted conclusion when a biological agent that is poorly understood is involved. Most public health professionals disagree. How does repeat infection work, for example? A more evidence based approach is sounder.

Not just what Susan Kuhn said, but "asymptotic" cases assumes these people, once mildly infected (and I myself in January came down with a mild flu, as did everybody here in my friends house, posting from a remote Greek village), then develops immunity to Covid-19. Where is the data to support this? The only way you can do this is with a Covid-19 virus antibodies test, which, from what I read online, is a difficult, expensive test (for now). I'd be happy to not advocate quarantine if somebody can show that having Covid-19 antibodies in your blood means you're immune from Covid-19. Wonderful! But we are not there yet.

Bonus trivia: Catinthehat was wrong about the supposed high costs of quarantine, see the new Fed paper released today, below. And I, along with Student, were right. Feels good to be right. I always feel good. - RL

New Fed study finds efforts to slow pandemic don’t depress the economy – Published: March 27, 2020 at 8:52 a.m. ET By Steve Goldstein

The economy isn’t helped by rushing back to work in the face of a pandemic, according to a new study by Federal Reserve researchers.

At a time when the lieutenant governor of Texas has said senior citizens would be willing to sacrifice their lives for a better economy and when President Trump has fretted that the cure may be worse than the disease, a new research paper looked at the impact of the 1918 Spanish flu on the U.S. economy.

^^ Ray Lopez, voice of reason? Surely, we are through the looking-glass.

Interesting paper and perhaps correct, however there’s one significant difference between the Spanish flu and Covid-19. The Spanish flu was atypically fatal to the 20-40 age bracket, so saving their lives was important to economic output. Not as much for Covid-19 as most fatalities are retirees who don’t work.
So if you calculated the average number of years lost per fatality, it would be much higher in the Spanish case. I now have 12.5 years for Covid-19 after using less severe life expectancy assumptions.

@Catinthehat - show your work Cat. I'm sure somebody here, even me, can do an Excel spreadsheet in about 30 minutes showing a cost-benefit analysis, in view of the above Fed paper that implies that GDP will not be impacted much by shelter-in-place (work from home). But we need data for life expectancy by age group and Covid-19 mortality (hard to do unless you have Big Data from hospital). Actually this should have been on the net by now, but it's not. I can't be expected to save the world since from my viewpoint it's unlikely I get Covid-19 (remote, completely self-sufficient Greek village); and there's nothing it it for me since unlikely I have any following in policy spheres even though people in Trump's administration probably read MR (though I do keep secrets of a non-client, non-NDA basis about Trump and his inner circle that could arguably sway the election, nothing illegal, as I've said before, and I'm not the kind to babble private information even if it's not protected by any privilege).

The US demographics can be found here :
The total population is 327 M. it shows each 10 year bracket
the Covid Case mortality is in various places, here is just one example. There isn’t a lot of US data yet
The Italy mortality per age group has more data
life expectancy tables are here. I take the midpoint of each bracket
you create 3 vectors of 10 entries for the 10 age brackets
Relative age demography, 2- relative covid case mortality per age, 3- life expectancy per bracket midpoint
Here is the relative age demography for the US
Here is the relative Covid mortality
relativeCovidMortality=1⁄100{0.01,0.03,0.15,0.3,0.65,2.0,3.8,7.4,9.9,10.}"(*" US"*)"
Here is the life expectancy
We need to scale the life expectancy of the later brackets because ( see the Italian mortality) 48% had 3 comorbidities,etc.. only 1.2% had none. These reduce your life expectancy , for example a 60 year old with diabetes has a 5.4 years reduced life expectancy. The scaling factor is ~ 0.65
You compute:

Where *. is a dot product.
Result -> 13.2 years. A similar calculation for Italy gives 10.3 yrs

not dot product, but element by element (of same index) product yielding a vector of the same length

Well given this week's Fed study that quarantine actually helps economies vs doing nothing, it's clear your analysis is deficient, especially since these days people can work from home using the internet. Case closed.

Their analysis is not about all pandemics, but from a specific one , one that killed disproportionately young healthy people. I am not disputing saving them was important to the economy. Most of the people who die in this one are not in the work force, they’re in their 70s, 80s, and 90s. You can argue it’s important to save them for social/moral reasons not for economic ones.
That case has not been made in my mind.
If working from home is equivalent to the pre Covid regime , why is the purchasing managers index in Europe and here at its lowest level in 2 decades. Why did 3.2 M file for unemployment, a 30 sigma event?Here’s a quote from the paper:
Anecdotal evidence suggests that our results have parallels in the COVID-19 outbreak.
Countries that implemented early NPIs such as Taiwan and Singapore have not only limited
infection growth. They also appear to have mitigated the worst economic disruption caused
Well-calibrated early and forceful NPIs should therefore not be seen as
by the pandemic.
having major economic costs in a pandemic.“
That’s not what the US is doing , we’re not doing NPI, like Taiwan and Singapore. We have brute force , blanket quarantines with widespread economic shutdown with basically no tracking of the infected.

The idea that one does not develop resistance to COVID-19 after recovery from infection is a bold claim. And one that, for some reason, is getting considerable play. This is how viruses generally behave. It is our standard model. Counter claims require extraordinary evidence.

re: 1918. The mitigations in 1918 were quite different than those of today: no large gatherings, wear masks, staggered shifts. These interventions are far less costly than "shelter in place". If I were Sergio Correia, et al I wouldn't be so confident.

The bold claim is quite silly but probably still needs to be refuted through study. Here is a start:

*If* the virus is more widespread than we thought, what would that mean for how we understand R0? If more people have it than we thought, then R0 should be higher, right? Probably much higher as those asymptomatic cases are out there spreading the virus for much longer than those who get laid up by it. If more people have it than we thought, then by definition, the mortality rate and the incidence of severe cases is lower.

But I do take back my conclusion that this makes the case for strong restrictions weaker. It makes the case for early strong restrictions on movement stronger (though, if strong restrictions are delayed, there comes a point at which those restrictions will be too late).

This does make me question the data or the reporting from S Korea. If R0 is much higher than we thought (which is the case if more people have the virus than we thought), should the testing, quarantine, and contact tracing be as effective as it appears to have been in South Korea? Or have measures been more wide ranging and restricting than almost all media outlets are reporting?

Also, while being infected by coronaviruses seems to not lead to permanent immunity, immunity seems to last 1-2 years (

The fearmongers are the real heroes here.

And it's not hard to be one! Remember these three easy rules.

(1) When something bad happens, be sure to report it immediately, with urgency, and repeatedly. Even if it's something as inconsequential as the First Death of a Veteranarian in Wakashaw County.

(2) When things are going OK, just switch over to "Things are Probably Much Worse Than You Think" mode.

(3) And when things are going good, yell at anyone who points this out. Your best angle here is that The Virus Is Just Not Completely Understood.

It's easy and fun! And you really only have until Election Day to do it, so start right away. If Trump wins, it all becomes moot, and he doesn't, posting this crap becomes a hate crime.

But why has this become a religious/ideological issue? I really don't understand why some insist on attacking anyone who reports a position that the virus might be less of a catostrophe than we originally thought. All should agree that all models suffer from a lack of information.
One thing that they will have to keep in mind is that social-distancing will almost certainly become increasingly diluted, no matter what country you are in. Only massive police persecution will keep the young inside as the weather improves, the young become convinced that it is not much of a threat to them, and the tedium of social-isolation becomes unbearable. No matter what action is taken, the increased lack of effectiveness of social-distancing should be taken as a given.

'But why has this become a religious/ideological issue?'

This is extremely bizarre, and pretty much only noticeable in the U.S. And Iran. One could draw some unflattering comparisons, as both countries are demonstrating how not to handle the spread of a novel coronavirus.

Speaking from Canada, the debates and yelling in the US is where the debates are happening. Everyone here tut tuts, but watch and read everything. Any criticisms from either point of view are identical to what is open and vigorous in the US.

This happens with pretty well every policy dispute.

Something to watch for is the Scapegoat. In the jewish law a goat was released into the wilderness to carry the sins of everyone. There are many people to blame here; almost every politician, anyone who travels, etc. Too much blame, not specific enough, but enough to tear asunder society. So someone or something will be the scapegoat. Watch for it.

We are cave dwellers with fancier tools.

A common theme across many religions, and hence human nature, is the notion that we must morally cleanse ourselves through abstinence and atonement, in this case morally cleanse ourselves of excessive consumerism. Not only is there resistance to any suggestion that the virus is not as catastrophic as the worst case assumptions imply, there is also resistance to any suggestion that we might be able to deal with it through any means other than abstinence-and-atonement measures like locking ourselves in our homes until a vaccine becomes available (>= 1 year). Other means might include expanding supply of testing, ventilators, masks, hospital beds, etc.; wearing masks and testing in workplaces; targeting lockdowns to hotspots.

We see a similar phenomenon with climate change alarmists. Indeed, in both cases we see the rhetorical weaponization of alleged deference to "The Science" and "experts": scientists are viewed as prophets whose teachings imply the need for abstinence and atonement. No, the secular view of science is that it's a tool for human innovation to avoid abstinence.

"there is also resistance to any suggestion that we might be able to deal with it through any means other than abstinence-and-atonement measures like locking ourselves in our homes until a vaccine becomes available (>= 1 year)"

That's a strawman of your own creation.

Right on cue, Tyler includes this as "Friday assorted link" No. 2:

One will note the intermittent distancing through 2021 or 2022. The Imperial College Ferguson paper, awarded an Emergent Ventures Prize for Good Policy Thinking on this site for laying out the "basic framework" for policy discussion, also shows intermittent distancing for 1+ years until a vaccine is found.

If by "strawman", you mean that the general public has not been adequately informed about the excessively long duration of (intermittent) abstinence --- years not weeks --- then I would agree with you. The "flatten-the-curve" graphs circulating the internet deceptively leave the time axis unlabeled. Many people understandably may think that extreme social distancing just means abstaining from regular life for a few weeks, 1-2 months tops, to "flatten the curve" before the all-clear signal sounds. They probably are not aware that at least some policy makers consider "everyone doing their part" to mean home confinement on and off for at least 1 year, longer if necessary to find a vaccine.

Don't forget "quarantine shaming," which parallels the shaming rituals of certain monasteries and convents.

We don't need such a degree of pop-sociology in this case. The most obvious comparison is with the trend in virtue signaling in general and, in particular, the shaming that goes on in wartime of those who are seen as insufficiently patriotic or somehow failing to "support the troops." American flag lapel pins strike me as going out of fashion but they were practically mandatory for politicians in the years after 9/11 and Obama was once duly shamed for not wearing one.

Except that politicians are not regarded as ersatz clergy like medical and scientific experts (e.g., Speaker Pelosi: "I say science is an answer to our prayers") who tend to the secular idol of bodily health. And call me a pop-Straussian since I was a student of Straussian professors.

Good point. Even more deeply, the disgust mechanism - known to be a driver of political orientation is deeply linked to disease transmission - it's what it evolved for. So it's not a surprise that when there's a contagion going around that's going to push all sorts of deep psychological buttons.

I don’t think asymptomatic rate is the right metric. The real number we want is fraction requiring hospitalization (which is probably closely related to death rate). That number is influenced by the asymptomatic rate but also by the “I thought I had a cold” rate. It’s this latter number that some people think could be incredibly large.

The Icelandic study is telling here. The headline was that half were asymptomatic. But the other thing was that 0.89% of the sample was positive. That was in a country of 350k people with 6 hospitalizations and 1 death at the time. Combining these three numbers with the proper respect for margins of error and the right corrections for time lags is going to be complicated. But the “crude” rate of hospitalizations and deaths from those numbers is 2 orders of magnitude lower than the commonly accepted values. That should be enough to provide us some hope that the virus is not as deadly as we think.

But what was the rate of false positives in that 0.89%? There is so much we don't know yet.

Not 0.89% is the "dylexic" number. The correct number was 0.86%.

That should be "Not 0.89%. 0.89% is the dylexic number." I guess I'm suffering from dylexia myself today.

Correct. Tyler misunderstood the WSJ article, which didn’t make a claim on the asymptomatic rate

Minnesota governor Tim Walz explains his decisions to limit activity and travel to "essential services":

"The concern that people are showing about the economy, I hope they're feeling the sense that I have of that, too. I understand the impact that that has on people. I also understand that not taking these actions will result in the deaths of Minnesotans," he said.

"The idea that this is a false choice between the economy and protecting people, if there is a better way other than just send everybody back, because my data shows me – and it was my first scenario run by the University of Minnesota and Minnesota Department of Health – shows what will happen if you just send them back."

Data provided by institutions can't be accepted. The data is supplied by individuals that must be held responsible for it. Walz, like politicians in general, is taking a bogus form of responsibility, similar to that of Janet Reno in the Branch Davidian disaster.

Maybe you are the outlier.

Mayo Clinic, Minnesota Department of Health, Osterholm, the UMN group on are smarter than they are.


You're the smartest guy in the room. That's why you say: "Data provided by institutions can't be accepted."

What facts did you offer in your comment. None. In fact, you disclaimed data offered by others. What did you cite as evidence.

Cool, so if one can't prove the data and assumptions are wrong, then they must be right! Skeptics need not apply.

Skeptics are more than welcome to apply. All they need to do is a bit of math - two constants and a variable to get a percentage. The constants (loosely defined) are reported confirmed cases at the time the equation is performed and population, with the variable being the ratio of detected to undetected cases. Using (an imaginary) 1 to 100, as of today, the results seen in Italy come from a total of 13% of Italy's population being infected.

And a skeptic is more than welcome to also wonder how well Italy (or any county, for that matter) could handle that percentage climbing to 39% in the next three months.

Yet, not every country is Italy.

Using that fantasy ratio, and excluding China where a country percentage would be meaningless, here is current list using today's worldometer info -

Spain - 64,059 confirmed, population 46 million - 14%
France - 29,155 confirmed, population 67 million - 4%
Germany - 47,373 confirmed, population 83 million - 5.7%

The French system cannot handle the Alsace outbreak at this point, but the Germans are still hanging in there, to the point of accepting a few French patients.

And for a new number, let's try the Swiss at 11,951 cases, 8.5 million population, 14% rate.

These are fantasy figures, of course. Realistically, it is extremely likely that Switzerland has something around 1.4% infected, France .4%, and Germany at almost .6% using a more realistic ratio of 1 to 10.

What's the point of showing various fantasy rations of 0.1%. 0r 0.01% and saying without argument that the first is more realistic than the second? I can as well answer that 0.01% is less realistic than 0.001%.

If you make a poll among the persons exiting from an electorate meeting of a minority party (say "Lutte Ouvrière" in France), you will probably get a result predicting an elector victory of this party with close to 100% of the votes, and yet you will be off by a factor one hundred.

To demonstrate the utter implausibility of using 1 to 100 or 1 to 1000? Especially when infection rate could provide another way to demonstrate how utterly implausible such numbers would be.

Maybe analogy is a touch better than math. If I say that a car's gas tank hold 5 gallons of gas, and then I say that after driving 100 miles, the gas tank needs to filled with a further 5 gallons, and someone says no, that cannot be true, I have the same model car, and when I put in 5 gallons it goes a 1,000 miles, there is absolutely no reason to believe that claim.

Boundaries are quite possible in the real world, which is why orders of magnitude is such an elegant concept to rapidly filter between the plausible enough and essentially impossible. 1 to 10, 1 to 12.33, 1 to 20, 1 to 7, 1 to 6.5? Sure, why not, within the realm of plausibility, particularly when supported by data. To go from 1 to 10 to 1 to 100 without anyone anywhere on the planet noticing it? Implausible is a polite phrasing in such a case.

"Yet, not every country is Italy."
Or, you might note that:
Not every country is Italy... yet.

You've completely missed the point, which is that when decisions are made, the basis for those decisions must have attribution. Walz should say that he has been advised by "Dr. XYZ", administrator of study "ABC", that following this course is the optimum procedure.

Oddly, the example given of Janet Reno's press conference following the Branch Davidian holocaust isn't posted on youtube. However, in it she claimed to have based her decision to attack the "compound" on the advice of some "experts". These experts were never named. So Reno washed her hands of responsibility and passed it on to invisible, unknowns. This is the ideal course of events for political figures.


Admit it. You didn't do any research. Have no facts.
Here, to fill your info void, is what I found:

You can watch the video of the statement and supporting data.

If I can do simple research, you can too.

from the article: Walz used modeling done by the University of Minnesota to explain his decision, and to show the consequences of different responses by the government. Under a no-mitigation situation, the model projects Minnesota would reach peak epidemic in about nine weeks (late-May), with ICUs reaching capacity in early May.

Minnesotans have already reduced contact by 50 percent with social distancing measures. Under the conditions he announced today, the model projects an 80 percent reduction in person-to-person contacts. This would put off the peak of the epidemic until early July, and ICUs would reach capacity around the second week of June.

Evidently, you have a reading problem. Nowhere in the web page you referenced is any individual other than Walz mentioned. The University of Minnesota is an institution, it can't talk, study diseases or mow the lawn. Those things are done by individuals. The model mentioned must have been constructed by one or more individuals. Who are they? On whose expertise is this model based? When and if the Minnesota economy regresses to a bronze age model Walz will say, "I did what the UofM said to do. It ain't my fault". In fact, the UofM said nothing, some figure employed by it did. Who is it?

I'm one of those unnamed researchers. We've been working nights and weekends at Mayo Clinic to synthesize data from all over the US and all over the world. Our only objective is to make the best predictions we can. You can take it or leave it.

You, too, must be missing the point or avoiding it. When the UofM Gopher basketball team fails against lowly Nebraska the media doesn't get quotes on the disaster from the governor, it talks to the coach or some of the players, who are clearly identified. Of course, that's sports, a subject much more important than the health, physical or economic, of the state. And the coach is, after all, paid more than the governor, earning $2 million annually, plus bonuses compared to the governor's $174,000.

Get over it. You have a commitment to a position that will not be refuted by facts.
When people see that, they know who you are and how you think.
You should be embarrassed and ashamed.
Here is the center that studies pandemics at the U of Mn.
Act like an adult, if you are one.
You still have not offered any facts.
Shame on you. No one listens, or should listen, to someone who offers no facts.

Oh, give it up, Chuck.
We get stats from the IRS, from budget offices, from the UN, from government departments, from universities and on and on and on and on. Data can be accumulated and reported and studies can be done by groups, by teams, by organizations...

I was underwhelmed by the WSJ piece. Their "Orders of Magnitude" suggestion remains a huge stretch. I agree that it's plausible that the infections per capita rate is two or three times what the (clearly inadequate) data allows, but a factor of 100 or more? No way. OTOH, the very low rate in the young suggests that it is possible that they have encountered some related bug - meaning that once we have 'serum' tests for antibodies, it may show a significant false positive rate with the children & younger adults because they're resistant to something quite similar. Just another way of saying that the map is not the territory.

Even assuming the fantasy ratio of 1 to 100, you can see the clear results of 13% of Italy's population being infected with covid-19.

It has been a real surprise seeing just how many innumerate people are unable to use two constants and a variable to get a percentage.

So, even using that 1 in 100 ratio, what is happening in Italy does not even represent 1 in 5 of the population being infected to date.

"No way" is a very strong argument, I have to admit. Let me use it too: "The real real of infection only two of three times the measured rate? No way!" Now, to use another argument, if we recall which people are tested (only those who present serious symptoms or are considered high risk for their recent long frequentation of infected people), you see that the number you obtained is in no way a reliable data for the population in general.

If you say No Way! to only 2 to 3 times, you would have the full backing of the Italian government, where the current estimate, after experiencing a still growing number of confirmed covid-19 cases, is 10 unknown cases to 1 confirmed case.

There is a bit more data available this week than last week, or the month before.

And really, by now, you should be able to do the basic math. Even if the truly imaginary ratio of 1 confirmed case to 1000 unknown cases is used, then less than half of everyone in France has been infected in around 3 months.

France - 29,155 confirmed, population 67 million, a ratio of 1 to 1000 - 40%

To be honest, at this point, one truly has to question your motives at continuing to defend your position.

To be honest, I do not understand your point.

My point is that we do not know the ration of "confirmed cases / real cases". It may be 0.1 or less, 0,.01, or even 0,001. No one in the comments
has produced any serious argument against any of these possibility (or everywhere in between). Tyler tried, by saying that other pandemics didn't propagate so rapidly. I am not sure that's true, but at least that's an argument.


As noted above, you may have a bit better time with an analogy, and the concept of orders of magnitudes. Boundaries can be set between the thoroughly plausible, the less plausible, and the utterly implausible when one has at least a couple of reliable numbers. In this case, number of confirmed cases and country population allows one to use a variable/a ratio to create plausible, less plausible, and utterly implausible scenarios. And even if 1 to 100 seems considerably less plausible (in part because there is zero empirical evidence to support it), using it as a variable translates to France having a current infection rate of 4%, which has caused a collapse of the health care system in the Alsace.

Peak prior_approval level autism.

Knowing the ratio is critical to the medium to long term path forward. That remains true even if in the very short term it’s not relevant to a doctor in Lombardy.

We need to know the ratio, and we need to invest in randomized population testing.


I think at this point, anti-body testing of random population samples would be hugely valuable. We need to determine the parameters of the issue we are dealing with.

I have read that this former GMU employee was fired 30 years ago for either sexual harassment or sexual assault.

In the spirit of the metoo movement, I think that should be noted every single time you identify this person. Personal courage is important - people who have personally experienced sexual harassment or sexual assault, possibly including yourself in such a traumatic situation, need to be supported unconditionally.

You mean that Yet Again = Bill = Prior and that this person was fired for sexual something from GMU some 30 years ago ? Maybe, I don't really care.

But even if he had been fired for one million rapes and murders, that would not change anything to the value of his arguments.

We need way more data. Orders of magnitude more data. One possible future for us is: Once a week or so you go to your local pharmacy for a Covid-19 test costing <$50; If it is negative you get a large, goofy badge prominently displaying the date of test; If you want to go to work, school, or a restaurant then you can only get in if your yellow badge is dated less than X days ago; Leftists complain that the tests aren't free; Rightists complain about the Leftists ; Pharmacy chains publish their results. This way we generate billions of tests. The tests don't need to be perfectly accurate because with that sort of sheer volume we would still have a clear idea of where the problem is and how to cope with it. Normal life resumes.


The Wuhan Flu likely landed in the Land of the Spree in December 2019 or January 2020.

Of course, I will persist. My hands are bleeding from washing. I don't go out except for prescriptions/food. I keep six feet away. I disinfect everything everything I touch. I microwave money when I get it. [...]

Take for example NYC. It is the most densely populated major city in the USA. Until (what?) ten days ago, each weekday two to three million people rode densely-packed busses and subways day; one million children and 50,000+ teachers/employees went to public schools. And, they want us to believe that their number (40,000 out of 7,400,000) reflects all Wuhan Flu infections.

I call it "Winnie the Flu", if only to annoy the CCP apparatchiks.

It is worth repeating these numbers - Italy confirmed cases 80,589, population 60 million.

Using a ratio of 1 confirmed case to 10 unknown ones (a number that the Italian authorities consider reasonable at this point in the spread of the pandemic in Italy), currently 1.3% of Italy's population has been infected with covid-19, with the painfully clear results that arise from having 1.3% of your population infected.

If one uses a ratio for which there is absolutely no empirical evidence, 1 to 100, then Italy demonstrates what happens when 13% of your population is infected in a fairly compressed period of time.

This is really straightforward math, but somehow, people seem to completely unable to perform an equation that is number of confirmed cases times a variable representing the number of unknown cases then dividing by the population of the country to get a percentage.

And for those actually able to do the math, what would a country like Italy look like if the number of infected was to increase to 39% in the next 3 months? Using a ratio of 1 to 100, a truly imaginary figure.

Someone else can deal with the math involving the infection rate of a disease with a ratio of 1 to 100 of confirmed infected patients to completely undetected cases, with a starting point around November in a single geographic location.

If you want to assume a large asymptomatic population you need to build in a high R0 (have to mathematically or else the disease couldn't have had a chance to infect that many people). A high R0 should mean we are almost at the peak of the infection (good thing) but if easily disproven if we continue to see rise in # of cases detected past that point, even accounting for a 5 day incubation time before symptoms.

This is right. If the number of daily deaths in Italy continues to climb for two more weeks, this would be strong evidence that as of today, we were far from having a large proportion of the population infected.

Or it would show that shutting down contact is effective in slowing disease transmission. As if China was not demonstration enough.

It is crazy that we know so little. Caution says that we should assume the worse, take drastic measures to protect, buildup treatments and medical responses as fast as possible. Fight a guerrilla war where we hide from the opponent until we are strong enough to launch a counter-offensive.

Is it burning through some populations at amazing rates? Is it possible some subgroups are truly protected to some degree? Although health workers who have significant consistent exposure seem to do very poorly. Are the asymptomatic simply people with low viral loads but who don't have antibodies. They just become carriers, (in a normal flu season 5-35% of people are asymptomatic carriers).

"The presence of asymptomatic carriers can hamper control efforts and make it difficult to uncover the true details of the natural history of an infectious disease and to estimate the total infection prevalence. Here, we have shown that if carriage is not correctly incorporated into mathematical models used to inform control decisions, there is a risk that these models may produce substantially misleading predictions.

Some of the deaths in Italy seem to be a result of too many sick at the same time. The high death rate for the elderly seems to be a choice to take extreme interventions with younger victims and morphine for older victims.

"For infectious pathogens such as Staphylococcus aureus and Streptococcus pneumoniae, some hosts may carry the pathogen and transmit it to others, yet display no symptoms themselves. These asymptomatic carriers contribute to the spread of disease but go largely undetected and can therefore undermine efforts to control transmission. Understanding the natural history of carriage and its relationship to disease is important for the design of effective interventions to control transmission. Mathematical models of infectious diseases are frequently used to inform decisions about control and should therefore accurately capture the role played by asymptomatic carriers. In practice, incorporating asymptomatic carriers into models is challenging due to the sparsity of direct evidence. This absence of data leads to uncertainty in estimates of model parameters and, more fundamentally, in the selection of an appropriate model structure. "

"Some of the deaths in Italy seem to be a result of too many sick at the same time. The high death rate for the elderly seems to be a choice to take extreme interventions with younger victims and morphine for older victims."

I'll preface this by saying that I'm not intending to diminish the human tragedy of these deaths, including elderly people dying and unable to be with their loved ones for last good-bye's because of this infectious disease.

If we go by the reports that we've seen on the demographics of deaths in Italy - median age about 80, half with 3 or more existing health conditions, another 25% with 2 existing health conditions - is it really likely that many of these patients would have been saved by having more ventilators available? People who do recover from COVID-19 after being on ventilators reportedly need them for 10 to 20 days. I have a difficult time believing that ventilators would have changed the fatality rate of this population by much.

Your description of 80-year-olds with severe comorbidities paints one picture. Why put someone on a ventilator if it has little effect on disease progression or ultimate outcome? When my father was in the hospital near death they performed cardiac surgery on him. It was unnecessary given he was expected to die within weeks. I was angry at the time.

But some stories out of Italy are with cases that are less clear. In Italy, 45% of deaths are between 60 and 79. In Lombardy officially 13 % of those infected have died. But they aren't sure officially how many are carrying the virus. Still, of those who make it to the hospital, 13% die. If you assume that 15-20% of victims need a hospital and can get to a hospital. then 2-2.5% of the infected in the general population die at the hospital. I assume that if the hospitals were not overrun then the death rate could be cut in half. And if the 1% dying are the 80-year-olds in poor health this is sad but not a national tragedy. It is the other 1.5%, not a trivial number, that is the tragedy.

Italian hospitals report being short of ventilators, oxygen, and protective equipment. Patients lie on mattresses thrown on the floor. It is chaos with health care providers making tough choices about allocating scarce resources.

BTW if the assumption that 80% of people get a mild version is wrong then this all gets a lot worse. If 50% are mild, 6.5% die. If 25% are mild then we start to approach 10% die. Based on Italy's deaths in the hospital. Not to mention the deaths from crowding out.

I am not sure what to make of that, but Italy and Spain, the two countries hardest hit by the Corona (in terms of the number of deaths and its rate of growth) are also the two countries with highest life expectancy in the EU, and among the few highest in the world.

Another "trivia" is that in a typical flu season, Italy has three times (!) more deaths per capita due to flu that the US, and 50% more than France. (Was to check if you which too with wikipedia).

Certainly it is due to age distribution, but perhaps we can be more precise than just saying median age that is very old in Italy (the second oldest in the world, have I read here). After all the median is just one aspect of a distribution.

For having lived in Northern Italy for one year, I know that some clichés about the Italians are true: that they eat well and that they have a lot of physical activity, walking and biking -- at every age. You often see very old persons (of both sexes) walking up a very steep street carrying an heavy bag of vegetables and fruits without apparent difficulties, or the same biking up and down. Certainly old Italians have a lower rate of cancer and heart-related illness and type II diabetes than in other rich countries, especially the US. So we can imagine a lot of people in Italy getting very old, without any chronic illness, but of course as they get old they become more and more fragile, and vulnerable to flu or coronavirus.

Seeing those people dying in huge numbers is disheartening. But at a personal level, I would prefer to die after a 3-day coronavirus infection at 87 that at 72 after a long cancer.

And I would take Covid19 over dementia.

It is sad to think of a baby born during the struggles of WW2 now dying from this flu. But celebrate the life and try to understand the death.

Assume everything about this is the same as the normal flu. Except the population lacks immunity. Than 5-35% of people are asymptomatic carriers.

Stopping social isolation seems to lead to a big increase in cases and by extension deaths. Even more, if the health care system is overwhelmed. Unless we can develop antivirals or other treatments, wishful thinking.

Look at Ohio today 1,137 confirmed cases. 276 in hospital, 19 dead.

So 25% require hospital, 9.4 % in ICU, 1.6% fatal. assume the death total will go up as new patients decline. But by how much?

State Health officials project 10,000 new cases a day at peak. 40-70% of Ohioans infected.

Is there a bright side?

"Is there a bright side?" I don't know, but I just saw the number of deaths today in Italy. More than one thousand. That's bad and sad.

But Italy's percentage increase in deaths has been declining and is now at 9% to 12% a day increase for the past four days, down from a 16% to 20% range that ended five days ago.

Today it's about a 30% increase, but it has been bouncing around. I certainly hope the lockdown is doing something

I'm not sure there is a reliable decrease, but indeed the growth rate in daily deaths crapped out last Saturday and hasn't changed -

Growth in confirmed cases stopped about 11 days ago -

Apologies, growth in confirmed new cases per day.

"If you want to assume a large asymptomatic population you need to build in a high R0 "

That's precisely the conclusion I reached, for the infected numbers to be this high, it implies that a lot of people were already infected several cycles ago. Since the cycle time is 5-7 days, and we are talking a period of 8 week or so.

The only way for the population numbers to be this high is that the R0 is very high and the CFR is an order of magnitude less than we thought. In that case it's a very infectious, bad flu and it should burn itself fairly quickly.

Should not be too hard, just use any country's confirmed cases, their population, and whatever ratio of confirmed to unknown cases you desire.

And it would be illuminating to the innumerati.

And the start date of first infection, of course. At least in some cases, the genetic fingerprint of the virus allows a fairly precise initial date for a specific country.

So, let's assume the Icelandic rate of 0.86% is accurate and not a sign of false positives.

They had 48 positives out of 5,571 cases. So if the test they were using has a false positive rate higher than 0.1%, then the numbers aren't reliable.

A number of 0.86% (assuming it's typical for the entire country) implies that there were (364K * 0.86%) 3,130 cases for the country.

If you assume conservatively that the R0 = 2 and the cycle time = 7 days, and that the infections average out at 7 days, that implies the infection will triple every 7 days. (I'm going to make this a week for convenience.)

Let's assume that conservatively that Iceland had 4 different infected patients show up on the same day, rather than 1 patient 0.

In order to have the 4 cases grow to 3K, it would take, 6 cycles. The testing date was roughly March 13th. So that would imply that their were 4 patients in Iceland on January 31st. (first confirmed infection was February 28th).

That's possible. However, the current CFR seems to be around 1%. If you assume that patients deaths lag cases by 2 weeks , you would expect to see Covid9 deaths in Iceland starting much earlier.

Projections: week-infected (0=Jan 31st)
0=4, 1=12, 2=36, 3=108, 4=324, 5=972, 6=2,916, 7=8,748, 8=26,244

5=1, 6=3, 7=9, 8=29

Reported deaths in Iceland are currently: 2

Conclusion: Either Covid19 had a CFR 2, Iceland results aren't typical, the testing was flawed(high false positives) or some combination.

If you ran the same numbers for NYC, the results would be even more out of line with our current estimates.

Ugh, the post stripped out the greater than less than sign and everything in between!

Either Covid19 had a R0 greater than 2
CFR was less than 1%
Deaths lagged case by more than 2 weeks.
Iceland results aren't typical
The study was flawed(high false positives, etc)
Asymptomatic cases are a high number (90+%)
Or some combination.

Also, note the scenario implies that Iceland currently has 26K cases but only 866 are currently reported.

So, even if the CFR was lower, one would expect the detected case count to be

If feels like the most likely scenarios if for a higher R0, combined with a lower CFR.

;tldr Covid19 may be More infectious, but less deadly than we believe.

The problem in your math is that deaths lag hospitalizations by about 4 weeks. (People are in the hospital about four weeks before they die.) That doesn't count the incubation period. So the number of deaths you predict for week 5 should be the number in week 8 or 9. If your other calculations are correct, then the 2 deaths so far are consistent with a 1% fatality rate.

Iceland is in the early stages of infection and there is a confidence interval on that 0.86% estimate. We also need to balance it out with South Korea, which is not in the early stages and has had over 350,000 of its citizens test negative so far.

Believing that covid-19 is 10 times more common than the official numbers suggest and it is one-tenth as deadly requires you believe that about 8 million Italians are infected but only 100,000 Koreans. And that leads to the question of why, under this theory, there are 80 times the number of cases in Italy that there are in Korea. These countries are not that far apart in terms of population and an 80-fold difference in cases is just not credible. In any case, Korea shows it is possible to keep a lid on on the spread of the virus and that countries can avoid the fate of Spain and Italy through smarter policies.

You can't really just declare that two countries are the same size, therefore the overall infection rate must be the same...? There are lots of countries the size of Italy which likely have fewer cases per head (UK, Germany, etc.).

Why can't the actual number of cases in Italy be something like 80 times higher than South Korea? (That doesn't prove 8 million vs. 100k - just the ratio.)

The number of COVID-19 deaths in Italy is now 65 times the number in South Korea.

South Korea is a country where people routinely wear masks even in normal times and where the population takes these matters very seriously due to prior scares with MERS and SARS. It also, informed by those experiences, did a very good job of containing by testing and tracing.

Italy is a country where people routinely greet each other by kissing on the cheek and where a lot of people's first reaction to any government edict is "what's my angle to get around this rule?" They conceivably might have been seeded with several coronavirus carriers arriving from China at a similar time in the part of the country with extensive business ties to China.

The point about which to be skeptical is that South Korea has 139 reported deaths but is extremely unlikely to have 100,000 cases. They have just over 9,000 confirmed cases and have been finding only about 100 new cases each day. They've almost certainly missed a few people, but it's very unlikely that they've missed 90,000 cases.

"Believing that covid-19 is 10 times more common than the official numbers suggest and it is one-tenth as deadly requires you believe that about 8 million Italians are infected but only 100,000 Koreans. "

Yes, which is why I started with the most likely explanation is that the R0 is higher than 2.

If the R0 is higher than you can reach the current number much more quickly. It doesn't presume that the virus has been latent in the population for 8-12 weeks.

The current tests are PCR tests. Established tests using PCR for variety of illnesses have a false positive rate of 5-10%. Our infection rate in Seattle right now in about 1 in 2500.

Test 10K people that do NOT have it, and the tests will tell you 500 to 1000 have it. So you test again. And again. You will have to test some people 4 times to arrive at the 1 in 2500 threshold.

That’s not the only conclusion. Some people may have immune immunity perhaps due to repeated exposure to other milder Coronaviruses. Children could be such a group.
Some of these people are not infected today, so Ro doesn’t have to be super high. Many would test negative.
When infected, they become the asymptomatic or mildly ill. They may even only briefly ( 3 days) have a detectable virus. For SARS also , a more lethal Cov-1 , children below 12 only developed a cold. Only a sensitive serological test would detect these uninfected but immune people.
For Ebola too in Africa, there’s a significant number of immune people. There’s not an explanation of why this is so but there undoubtedly is.
The indirect evidence for this is the diamond princess, a lot of mingling for a while but only 17% tested positive.
In Wuhan the virus had free rein ( Nov 17 to Jan 23 , 66 days to the lockdown ) but only 80000 managed to test positive. That’s less than 1% of the population. One possibility is certainly native immunity.

Not “ immune immunity” in the 2nd line but “natural immunity “

Tyler : " the “optimistic” view implies a much faster spread for Covid-19 than would fit our data from previous viral episodes, which tend to come in waves and do not usually infect so many people so quickly. "

Is that true ? According to the CDC, the annual number of persons infected each year (during the decade 2010-2019) by the flu during the 2-3 months flu season varies between 9 million and 45 million.
We now know that the coronavirus has begun to circulate in Italy at the beginning of January, and probably not much later in other developed countries. We are at the end of March. Why is it unrealistic to suppose that the propagation of the coronavirus, in a population with no immunity at all, was as quick as the propagation of the flu a normal year ?

And if we have either 8 or 45 millions infected people currently in the US, then the mortality rate is not equal, it is one or two factors of magnitude
lower than the flu.

Has anyone challenged the oft cited # of deaths from the flu?

Well, Joël, we shall see in a few months if you're right or not. The evidence will eventually shake out. If the # of deaths continues to rise, you will have been proven wrong.

Italy's daily deaths have plateaued over the past week. The evidence on the pattern of deaths is clarifying every day.

One would expect the death / hospitalization ratio to improve with experience.

And the hospitalization rate to dictate the end of the wave.

Here is some regional data from Italy posted two days ago, with data through 2/23.

Steep growth in some areas. They are saying it is a lockdown story, rather than any kind of natural plateau.

Italy, new deaths on
March 24 - 743
March 25 - 717
March 26 - 712

That plateau means that 2172 people have died in 3 days. The Italians are desperately hoping that represents a plateau, because they are really starting to have a problem dealing with that many infected dead bodies, day after day.

Three days is a pause, not a plateau. Here's the last six days:

March 21 793
March 22 651
March 23 601
March 24 743
March 25 683
March 26 712

That's not how exponents work.

I understand Italy's system is being pushed to the limit, they had 20,600 active cases on March 14 and 62,013 yesterday. But the pace of new cases is slowing and Italy is seeing 1,000 recoveries a day at this point, so the trend here is encouraging too.

919 dead Italians today. Oof. Bad news for everyone.

I agree that's bad news.

Has anyone seen a location for Italian COVID-19 statistics by day *by region* (i.e., the sort of information at Worldometer, but down to the granularity of region)? Idea would be to see how much trends are spread to other regions of Italy versus Lombardy tragically being hit even harder than it has been.

Have at it:

Thanks. Looking at the graph and also finding some daily numbers from Reuters, here are the numbers just for Lombardy:

16-Mar 220
17-Mar 319
18-Mar 209
19-Mar 381
20-Mar 546
21-Mar 361
22-Mar 320
23-Mar 402
24-Mar 296
25-Mar 387
26-Mar 541

Italy usually has 1700 or so deaths a day. I'd check the provenance of any media reports about them not being able to handle it.

"not being able to handle it?"

What is wrong with you? And are you wearing your "do not ventilate" bracelet?

You know that the reported covid19 deaths are deaths related/caused by covid19. The earlier average of 1700 per day remains more or less unchanged - though car accidents are undoubtedly significantly lower, strokes and heart attacks are likely unchanged.

Or do you think the Italian Army is called in to haul away the dead on a regular basis?

I’m confused when people say that the numbers have plateaued so we don’t need to worry. They only plateaued after Italy shutdown the whole country. This doesn’t tell us anything about when things would plateau if there was no policy response and no shutdowns. But many who see the plateaus act as if it’s evidence against the very policies that led to the plateau!

Who said we don't need to worry? The surest sign that this isn't the apocalypse is how quickly we return to insipid political finger-pointing.

Sorry, should have clarified that I wasn't directing my comments at you.

It is definitely not just finger pointing. Lives are literally on the line.

If "an Italian plateau" mutes US response, literally as NYers share ventilators, more people die.

You and aMichael should just yell at each other.

I can't help the stupidity of New York to date. If "Italian deaths have plateaued over the past week" leads DeBlasio to issue an all clear, well, I can't help that either.

How is that a rational response?

Do you think NY is unique in ICU/pop ratio or transmission rate?

I'd bet NYC is at the top of the US in transmission rate, because it is cosmopolitan and has by far the densest population in the US, plus delayed response.

New York should have led the response. Instead, city officials were wonderfully sensitive to Chinese New Year racism, AOC chided racists for shunning Chinese take-out, and DeBlasio was giving out show recommendations.

Well, if by "delayed response" you mean "slow to shut down," do you at the same time understand the meta-argument regarding total infections and hospitalization rate?

The argument is being made, as Tyler says with bad data, that higher infection rate means we can reduce our concern, and that same response.

The implicit argument is that we can treat NY like a one-off, the same way Seattle was a on-off, the same way Wuhan was a one-off.

But in a global pandemic there are not really one-offs. It just takes a while to reach you.

"Are many more people infected than we think?"

The consensus on this board, which I think is correct, is "No".

There is no implicit argument. You are inferring in error.

I'm not Brian, but I think New York's response seems uniquely bad within the US so far. It stands in contrast to, for example, California's response. I'm pleasantly surprised and I think Gov. Newsom deserves props.

But it's early days, and as far as I can tell luck and the distribution of the cases you originally imported from China and then Europe has a lot to do with how regions are doing at the moment.

Our California cases are still growing. We won't know that we actually avoided the Italy/NY scenario until cases peak, we hope before ICU saturation.

As of yesterday: "The number of cases is doubling in Orange County every three days or so."

That seems dangerously consistent if indeed it is through lockdown.

A lot of that has to be increased testing, or getting the results from testing that was already done but backlogged in labs. It doesn't mean infections are growing at that rate, at least not in a straightforward manner.

I don't think anybody can be confident about anything at this point, to be clear, but a lot of people expected California to fair quite badly and at least so far it's doing better than that. I don't know where this will end up, but even arriving at a high level of infection a few weeks later is a major victory, because that gives time for the build-up of resources to handle the surge. Delay, delay, delay.

If anything,

"A report from the Society of Critical Care Medicine found that more than nine out of 10 intensive care beds nationally are in metropolitan areas with a population of more than 50,000. Only 1 percent of America's ICU beds are in rural areas."

An uncharitable interpretation is that you'd like to see Red America more on the sharp end of this.

Dude. You are more political than me in this moment. "AOC." "Red America."

What I want is everyone to take this seriously and to prepare.

I'm sorry if you feel like I'm attacking Democrats, but they are the people that run New York government.

It is unconscionably stupid to "center" identity politics the way New York did in the face of a serious problem.

Just like it is unconscionably stupid to say "I don't know anyone who has the virus, we need to get back to normal now", which you get from the right.

But we were talking about New York, and all of a sudden you start talking about rural ICU beds. What's up with that?

If in fact NY has a high ratio of ICU/pop, that means it is not a one-off to be ignored when those resources are swamped, it is an extreme caution to anyone with a lower ICU/pop ratio.

Who is ignoring New York or treating it as a one-off? It is possible to be extremely concerned about the situation in New York while realizing that most places can and should do better.

I don't think flu season has a point source, or is a single disease. It is a total of all the strains. From what I understand there are reservoirs of the virus all over the world waiting for the next season.

This is different because it was a point source, and might enter your town by just one or two carriers.

And so the doubling takes a long time to be noticed.

Regarding flu comparisons...yes. "Flu" is a number of different viruses, so it is isn't helpful to compare the "flu" to single virus.

Serological testing from San Miguel County CO (Telluride) disproves the widespread infection hypothesis. 645 first responders and family members, tested, 0 had immunity.

Is San Miguel a hotspot? 1 case confirmed in a county via a PCR test. Also I'd agree first responders should be a hot spot, but testing 645 out of 7900 isn't quite there yet.

"Is San Miguel a hotspot? "

Even if it's not a hotspot it tends to disprove the widespread infection hypothesis.

"Even if it's not a hotspot it tends to disprove the widespread infection hypothesis."

Maybe, but not sure that it's that strong of evidence. It's conceivable that not many people are infected in a remote Colorado county. Granted, the presence of Telluride means a decent amount of outside travel there, but perhaps they were lucky in not having COVID-19 carriers visit (or only having a few with unusually good hygiene).

I think the bigger claim is that what we saw in Lombardy and Wuhan, and are unfortunately starting to see in New York, is from vastly higher rates of infection than implied by confirmed COVID-19 counts. In other words, perhaps Lombardy's health system hit the breaking point as COVID-19 spread rapidly and infected about 10% to 20% of the population. (Lombardy's population is about 10 million.)

The idea is a combination of factors. Several unidentified cases showing up around the same time. (Chinese traveling from Wuhan in northern Italy, international travelers in New York.) High R0 due to population density and mass transit.

Also, one other option: perhaps this serological test simply doesn't work. Has it been used successfully in other places?

"Maybe, but not sure that it's that strong of evidence."

+1, yes one case is not strong evidence. It might be an aberration.

I don't believe the virus has passed through a large percentage of the population. Iceland and Luxembourg are doing extensive testing on their small population. So far, these two countries have identified 0.23% of the population as being infected.

Tiny San Marino (population 33,000) has suffered the highest per capita death rate in the world, yet even here only 0.6% of the population has been identified as infected. Given how small and hard hit San Marino was, I'd have thought they'd have tested most people by now.

I could be hopelessly confused but I think what could be happening to some extent are exposure events to the virus which lead to short-duration asymptomatic cases. Short duration could mean several days, during which it is likely that no test would have been performed (I know one case locally that was asymptomatic positive Tuesday and negative by Friday; testing was done due to direct exposure to a symptomatic positive case). What is needed is an antibody test to find out how many cases may have been missed by the PCR test even those countries that are testing aggressively.

A one time PCR pos in someone is not indicative of either immunity or asymptomatic infection. PCR is so sensitive it can pick up a piece of crud that got into your nasal mucus and didn't actually infect any cells. Especially if you test negative the following day, that doesn't tell you have cleared the disease or have immunity. A serological antibody test is the only way to know. Luckily Germany is funding a large scale study:

Well this individual self-isolated with her husband and now her husband has a dry cough. In any case that's why I said an antibody test is the only reasonable way to address the ambiguity in these cases. This is a new disease that may have an unusually large distribution in case severity: the fat tail on the left are asymptomatic spreaders and the fat tail on the right are queueing at the hospitals.

... More than 80,000 Americans died of the flu in the winter of 2017-2018,

Covid-19 deaths are about 1,000 so far

what do the numbers tell you?

That the Trump Administration was utterly unable to handle a flu season with fewer deaths than the swine flu season in 2009?

And that you need to wait at least a few days before making comparisons, because that 1000 figure is currently at 1306 according to worldometer.

Fixed it for you

>... More than 80,000 Americans died of the flu in the winter of 2017-2018,

Covid-19 deaths are about 1,000 *in the last 7 days*

what do the numbers tell you?

Well just looking at the numbers, the flu season is roughly 19 weeks.

80,000/19 = 4,210 deaths per week during the flu season.

So, I'm not sure we can conclude anything at all.

Dude look at the graph of daily deaths

I have. but without knowing where there going to peak, we don't know the long term effects. It's not a pure exponential curve, it's a sigmoid function (logistics curve). So, how far away is the inflection point is the question? 7 days away and this won't be as bad as a bad flu season, 30 days away and it will be terrible.

But the fact that the Italian new infection count & new death count both hit a high mark 6 days ago is a promising sign.

Oops, Italian new death just shot up. That's bad news.

The problem is that if you are successful in stopping or controlling a pandemic, others will say: it wasn't necessary, see, you overreacted.

So, your success gets rewarded with criticism.

That's why its good to test, test, test, and have the data along the way.

Birx was almost tipping some of her cards in the news conference yesterday with respect to the presumed infection rates based on country level data. I think she might be leaning towards an epidemiological model where viral prevalence in many countries is much higher than official numbers. Doubt it's 100x but would believe even 5-20x.

Those White House press conferences are really good and informative. They do not raise my confidence in our news coverage, though.

+1 good for americans to see the profoundly unserious "media" in vivo

At least those idiot reporters stopped asking variations of the question "you shut down travel from China; what kind of racist does that make you?"

I counted seven variations of that in one of the press conferences (and zero questions about asymptomatic cases)

we see why the "media" want the conferences censored!
Americans understand context& causation better than "media"

Yeah, if she has any cards.
Maybe its a bluff to please Trump.
It will all come out with oversight.

I appreciate this platform and Tyler's reasoning here. I just wish someone in the White House would call him.

The administration hears the arguments from the WSJ piece:
1. We need more data.
2. The data might show that a universal quarantine is not worth the costs.

and uses argument #2 to justify not doing #1. When the leader is in denial and is a narcissist, do you really think he's going to change his behavior based on data?

Anyone who uses decimal points right now in their calculations is pulling your leg. The early tests from china had a serious bias towards positive results; they knew it wasn't accurate and wanted false positives, not false negatives. What test was Iceland using, and what is it's reliability? How much of this discussion is driven by incoming data challenging models, as opposed to incoming data generating insights?

There are a lot of tantalizing hints in Europe that size matters.

Small countries (Luxembourg, Switzerland, Portugal) have better metrics across the board (in particular, per capita death rates) compared to their larger neighbors (Spain, Italy, France, Netherlands, Belgium).

Either North America is well behind the European timeline or the virus seems less virulent here (more spread out?). Either way, it seems like good news for North America. Canada (big, but small by North American standards) is crushing it.

Geography matters. I'm 8 hours drive away from two of the hotspots in Canada. We have a sprinkling of cases none in hospital. I'm not a specialist, but I'm getting the impression you have to work a little to get this thing to spread; being in a metal tube for hours with a few infected will probably do it, or having a few million people putting their hands on the same stainless steel handhold.

Yes. I am 55 and alert for the first time in my life to door handles, revolving doors, elevator buttons (I work in a 100-story building), and touching my face. Small behavioral changes can have a huge impact.

I'm increasingly feeling like we will get over the first hump in the next couple weeks, but not without people setting their hair on fire when the daily US death toll hits 500.

The longer-term is more uncertain. Japan is well ahead of our timeline and may lead the way here: a famously-disciplined and hygenic people but packed like sardines. If they can strike a balance between being careful and a functioning economy, we can learn from them.

Indeed. My work takes me into lots of places but I can avoid people. My daughter lives in a large city, and she is staying home; in crowds it can't be avoided.

There are two data points that matter, the rest is otherwise unoccupied people finding something to do. The collapse of the medical system first in China and then in Italy. The second is the unemployment numbers. They both signify a collapse, with extreme second and third order effects that are as unpredictable as everything else in this mess. Or maybe predictable, a better word is unthinkable.

So far outside of media and anyone who pays attention to those idiots, it seems everyone is doing the best they can. Lots of complaints around the margins, but this is a mess, and everyone in authority is desperately trying to not make it worse. Sometimes in spite of themselves.

Intriguing anecdote for Derek's point. The NBA has 14 positive cases across 7 teams. ( )

Given the amount of time that players are around each other (and coaches, trainers, etc.), I'd think that something highly contagious for which people have no immunity would spread more rapidly within each team.

Neil Ferguson didn't walk back his predictions. He changed a parameter in his model. What is the fake news? That the model is of academic interest and utterly useless as a predictive tool?

Isn't it common knowledge that if your inputs are imprecise, that your model may be simply a random number generator?

As for the plea for better data, it will arrive, but by definition these situations cannot produce data in a timely way. If you don't want to fly blind, don't fly. If you don't want 3.2 million unemployment applications in a week, don't set up an economic system that depends on the public health systems in a place far away.

This is an ugly situation not simply because of the deaths. Imagine the humiliation of so many when the realize that Nassim Taleb was right.

There were several right of center outlets that were reporting yesterday that Furguson had "reversed" himself. Implying that decisions based in his model were gross overreactions.

So he didn't reverse himself, he changed a parameter that changed the output by orders of magnitude.

In other words a random number generator.

I think more accurately real life changed a parameter by going from business as normal to lockdown- right?

and the "media " didn't get that fact

Nah, we just having fun in Florida.

God is punishing us all for electing an adulterer for president. Fail to keep His commandments and we must pay the price.

Yes! The US should have elected the woman who hugged a child after she had collapsed due to pneumonia.

Why did God wait a half-century, though? He can't STILL be mad at JFK, can He?

OR .... the original strain of the virus did not actually cross the species line in November of 2019, but much earlier, maybe a year or two ago. It just mutated into a more infectious form in Nov 2019 in China. So it's been spreading in the world population a lot longer than we think, and a lot of people have already been infected and recovered.

(Resuscitating personal theory about my own viral bronchitis infection last June. Everyone thinks I'm crazy, but the symptoms were pretty much identical.)

Great post. I agree with Tim and Tyler. I have a suspicion that I got this in early February... but I have no way of knowing currently. Sure would be nice to know for certain... because if I knew It’s likely I could rejoin society full fledged (though there is a small chance of reinfection... low but it’s a possibility).

The case for having got it:

I got really sick in early February. Sicker than I can remember having been in well over a decade. Coughing up blood and fever of 104 sick. I live in an area with much interaction with Southeast Asia. I got sick about a week after Chinese New Year. Everyone in my family got sick which I wouldn’t expect with regular seasonal flu because I would expect someone in my family to have immunity. The spread between us was amazingly fast as well. That never seemed to happen when someone got the normal flu before. Usually only one of us would get it. Got CT scans (cuz I coughed up a decent amount of blood and that never happened to me before). The scans revealed gound-glass opacification, which is consistent with covid19.

The case against:

It was very early in the outbreak only a week after Chinese New Year. Had the travelers even gotten back yet with time to catch it from someone that caught it from someone else?

My kids go to a large school system with maybe thousands of teachers. There were no deaths reported and if the CFR was 1:50 or 1:100, one would expect to have seen some teachers die. That didn’t happen.

I doubt that a new virus could circulate in the community without the medical community becoming aware of it quickly.

Would sure like to take an antibody test. Wonder when that will be available large scale. Imagine if we got that out there and we found out who was likely immune... markets might rebound really quick. If it is the case that herd immunity has been achieved or close to it or even that some fairly sizable share could act normally, things would be a lot better.

This could be huge. IMO, antibody testing (if it is the case that many have really had it) could be one of the quickest ways things get back to normal quickly.

There have been a number of stories this winter similar to yours where they were tested and didn't have that specific virus. As you say it would be nice to know.

I doubt that a new virus could circulate in the community without the medical community becoming aware of it quickly.

I don't. Urgent care when I had my illness last year was just like "there's some shit going around. We don't know what it is, it's viral. There's no treatment, here's an inhaler." And given what we know about the restrictions of testing and test development , how would they have had the tools to identify the virus? It's quite possible that doctors in your area suspected COVID-19, but either weren't permitted to test or didn't have a test available.

One more thing in favor of having had it. One of my kids got tested for influenza A and B, both were negative. Could have been false negatives... but I am almost certain it was some kind of flu, but testing suggested not type A or B.

I found this paper via the LW database. He looks at Influenza like infection reporting data over the past 20 years and sees some suggestive historical spikes this past season. The most notable thing for me was the 20 year highs in washington in dec/jan. But also some surprising other areas have historic highs. And nearly all zones hit 20 years high in past couple weeks which would correspond to underreported Covid. He does mention that it could simply be some other unknown pathogen or pathogens trggering these reports. Obviously, just one set of correlations, but interesting nonetheless.

The only data that tested a total (or nearly total) population is from the Diamond Princess cruise ship. Roughly 50% of cases were asymptomatic, similar to the Icelandic data. 50% asymptomatic is he most reasonable prior. What is still unknown is the transmission rate of the asymptomatic. Case reports show that transmission is possible, but we don’t know if the degree of contagion is substantially less than that of the symptomatic.

A disease that is often invisible and and highly contagious by those people is nearly impossible to contain without severe measures. Anybody not recognizing this possibility has their head in the sand.

Yet even with the 50% non-contagious proviso, China, Japan, Singapore and South Korea have managed to slow or contain the virus.

Some combination of extensive testing, contact tracing, and social distancing might help slow or contain the spread. Maybe people can go back to work but we keep restrictions on travel, crowded bars and restaurants, concerts, big gatherings such as conventions and religious services, and a few other things for several more months. For contact tracing, I know it's a tough sell in the U.S. but get as many people as possible to install an app to allow health authorities to monitor people's movements and alert them if they come into contact with someone infected. I think Singapore and South Korea are trying version of this and almost certainly saving lives.

But first, we need to slam the breaks on and give people who are already infected time to self-isolate and recover.

Could the tracking app be required due to OSHA requirements?

note about the The diamond princess study -- If i am reading it correctly, they say that they assume some number of asymptomatic cases were actually presymptomatic, and revise their final asymptomatic number to 17%.

The tests were for the virus. Not for the antibodies. The entire ship may have been infected. 3,711 on board . . . skewed elderly . . . and 8 died.

Yeah, I've seen that theory. Would be super awesome if we had robust anitbody testing. And random sampling. Would help bring a lot of things into focus.

"(And we should work much harder on producing better data.)"

And/Or: we should work on producing data. (Modest contribution thereto: Colorado--363 cases 3/19, 475 cases 3/21, 912 cases 3/23, 1430 cases 3/25 . . .)

in this viral pandemic
isn't it likely that many of the "asymptomatic" infections
are false positives

Or presymptomatic

would suspect/ bet that if tested positive but presymptomatic
epidemiologic followup would kick them into the symptomatic /infected column?

I don't really mind TC's argument here, though time will tell if it's wrong (as I would suspect to be likely).

Accusations of sunk cost fallacy (and I-cant-reverse-myself syndrome!) meet accusations of wishful thinking.

This though "Furthermore, the “optimistic” view implies a much faster spread for Covid-19 than would fit our data from previous viral episodes, which tend to come in waves and do not usually infect so many people so quickly." is ultra dumb. It's a novel virus; it's not gonna spread near as slow of get to a plateau near as fast as "previous viral episodes"(which refers to ?).

Data is actually showing that corona infection cases today are massively down in the US. Good news!

It's not the end of the day yet.

Maybe they tested far fewer people today (so far).

Only 21 states have reported any data today so far.

That third paragraph makes my head hurt in a coronavirus-like way.

> You would have to argue that the asymptomatic cases usually test as negative, and while that is possible again there is no particular reason to expect that. It should not be your default view.

There are several, particular, reasons to expect that. 1) The person was infected at low dose, sufficient to generate an immune response, but not sufficient for the virus to be detectable. Our tools are very sensitive, but they have limits. See LabCorp's FDA document and search for LOD (limit of detection): 2) The person was infected at high dose but has preexisting immunity (which might explain why kids, especially, who are often suffering various corona virus infections, might be less susceptible) which rendered the infection outright undetectable or much shorter than the typical timeline. 3) The person was infected but either the strain of the virus, or the genetic makeup of the person infected (or both?), made the person less susceptible, rendering the infection undetectable.

Separately, do not overly rely on serological tests for immunity. For the types of infection mentioned above it might take weeks or months for the immune response to be detectable by our available tools, *if at all*. And that response may not be sustained at a detectable level. The idea that we will be able to universally test for immunity and allow these people back into the economy is flawed. As with any test, there will be false negatives.

Yes, all of this is speculative. But, I can just as easily say that there is no reason to expect that social distancing is effective. We do not yet have proof that it is. Implemented *correctly*, it absolutely should be. But I see little evidence of this. Kids are still getting together for "birthday" parties. High school seniors are going on "pilgrimages" to their schools, just to sit outside and reminisce, in shock from the rites of passage they are missing. People in the neighborhood are out and about in abnormal numbers enjoying the weather, people from different households walking together and talking. Office buildings I see in the suburbs have mostly full parking lots. We have likely altered the steepness of the curve, but it is by no means clear by how much.

I echo your call for much, much more testing, and fast. It is essential. But even testing has limits! The faith of the complacent class in exponential curves, effective testing (especially given our track record so far) and an inevitable vaccine is disconcerting, bordering on terrifying, given the stakes. If any lesson was learned from Theranos, it should have been to pay much more attention to people like John Ioannidis, who warned us then (, well before the WSJ, and is again warning us now. And, dare I say, pay less attention to economists on this particular matter?

single data point:

Exclusive: Captain of aircraft carrier with growing coronavirus outbreak pleads for help from Navy

"between 150 and 200 sailors had tested positive"

Not clear out of how many. Presumably only those showing some symptoms. They state whole ship of 4000 will be tested at some point.

"none of the infected sailors has shown serious symptoms"

Likely that few, in any, of these would have been tested if they were in the general population.

If You Have Coronavirus Symptoms, Assume You Have the Illness, Even if You Test Negative

"Across the world, people with signs and symptoms of Covid-19 are testing negative and wondering what it means. They are not showing up in the statistics, and they are left in limbo about what to do next.
The problem may be with the test. Current coronavirus tests may have a particularly high rate of missing infections. The good news is that the tests appear to be highly specific: If your test comes back positive, it is almost certain you have the infection."

China Concealed Extent of Virus Outbreak, U.S. Intelligence Says

"The Chinese government has repeatedly revised its methodology for counting cases, for weeks excluding people without symptoms entirely..."

Antibody Tests for the Coronavirus

"On the other side, a negative result really doesn’t mean much, because there’s always the chance that a person generated antibodies that don’t recognize the antigens that the test kit has built into it for detection. You can’t rule it out. It is also quite possible that a person has been infected but hasn’t had time to generate enough antibodies for the test to detect yet."

> do not overly rely on serological tests for immunity. For the types of infection mentioned above it might take weeks or months for the immune response to be detectable by our available tools, *if at all*.

Britain has millions of coronavirus antibody tests, but they don’t work

"John Newton said that tests ordered from China were able to identify immunity accurately only in people who had been severely ill and that Britain was no longer hoping to buy millions of kits off the shelf."

"A lot of people are (understandably) talking about having some sort of “immunological passport” system to clear people for work, etc. before we are able to vaccinate the population, and these results may be telling us that that will be a complicated process, one that might not clear as many people as one would hope."

“Once you’re a couple of days into infection, chest CT scans don’t miss,” said an emergency medicine physician in Louisiana who asked not to named. With the swab test missing 30% to 50% of cases, physicians in China called for the diagnostic use of CT early in the outbreak there, and “fever clinics” set up in Wuhan and elsewhere began routinely using them.


Of every 100 symptomatic people who test negative for Covid-19, 30 are actually infected. The test missed them.


“False negatives are going to be a problem and could definitely undermine” re-opening hopes.


Begun, the testing war has. There is going to be so much room to question and critique all of the testing variants. Any chance of consensus will slip further and further away. This particular article focuses on symptomatic cases. Cases which are so symptomatic they resort to an expensive, radioactive CT scan. Seeing this, plus knowing the population skew for susceptibility (30% of US pop > 55 years old), what are the chances that the number of people "unknowingly" infected is not at least 10-fold what we are measuring, with the consequent impact on CFR? How much evidence do we need? I despair that we will ever see the degree of *effective* testing necessary. At least, not in a time frame that's going to make a difference to the economic impact. It seems like we are rapidly approaching an economic tipping point, if we haven't yet unknowingly passed it.

Patient-derived mutations impact pathogenicity of SARS-CoV-2

"Importantly, these viral isolates show significant variation in cytopathic effects and viral load, up to 270-fold differences, when infecting Vero-E6 cells. We observed intrapersonal variation and 6 different mutations in the spike glycoprotein (S protein), including 2 different SNVs that led to the same missense mutation. Therefore, we provide direct evidence that the SARS-CoV-2 has acquired mutations capable of substantially changing its pathogenicity."

This also relates to your other post, but comments are closed there:

and comment:

If a second wave results from a more pathogenic variant, like may have happened with the 1918 flu, Hanson's idea of variolation, at low dose, with a relatively benign strain, looks increasingly attractive.

If the virus spreads as quickly as they claim, you would expect people to show up at hospitals at nearly the same time over much larger geographical areas. The staggering with which people at different places go to hospitals provides an upper limit to the rate of spread.

Unless there are somehow two strains of the virus, one that spreads much faster, but is much less deadly and one that spreads much slower and can only take hold in some bodies that somehow didn’t manage to build immunity due to the first infection. Then you could have many asymptotic cases fast and a slower spread of deadly cases.

Asymptotic -> asymptomatic

A doctor in the Faroe Islands (population 52,000, with 92 confirmed cases) has perhaps found different chains of contact that appear to have different rates of transmission. Not proven, but intriguing:

"We saw that the first cases we had were not contagious. We could see that although they had had kissing contact, they hadn't passed it on.

Then a different type came from Denmark and from Iceland which spread really quickly, so these two chains of contagion became completely central."

There's not yet information if the the fatality rate differs.

Link that I originally forgot to include -

It’s not just the infected but asymptomatic, but the uninfected but naturally immune. That’s why a serological test may help.
If there’s a large population of infected but asymptomatic, it follows there’s an even larger population of uninfected and naturally immune.
This infected but asymptomatic becomes silent as they no longer test positive.

Can you clarify this for me? Do the "naturally immune" produce antibodies to the virus or not?

It's not the only way the body fights infection but they should have some levels of antibodies, which destroy the virus before it penetrates cells ( where the virus is less vulnerable), either directly or through agglutination of viruses into a bigger target or activation of phagocytes.

It is not neccesarily through antibodies.
The naturally immune persons could have a genetic difference in their cells that prevent the virus (our other pathogens) to penetrate them. Like, if you have one of the allele for the sickle-cell anemia, you don't have the illness (you need the two alleles, because it is recessive), but you have many anormal red blood cells that the plasmodia of Malaria can't get in, and you get some natural protection from Malaria.

For the statisticians here, what are the odds that the 2 most famous people in the U.K., Boris Johnson and Prince Charles both have the virus, if we assume different proportions of the population have the virus? If 50% have the virus what are the odds that they both have it? Assuming very famous people represent x percentage of the population?

Lots of fairly famous-ish people have it, which is odd if you have well below 0.01% prevalence as suggested by confirmed case counts, but clearly it's difficult to estimate from the because folk like BoJo and Charlie Boy might be more connected into networks that deliver it them than the average person.

I imagine both see many more people per day than the average person does, and the people they see every day see many more people per day, etc.

The probability that two given peopler are infected (say the two most famous) in a country where a proportion x of the population is infected
is x^2. So if 50% of the population is infected, it is a 1-in-4 odds; if 10% is infected, this a 1-in-100 odds; if 1% is infected, it is a 1-in-10000 odds. And if we talk the official figures of 11,000 infected, then it is 1-in-36 million odds.

But these obvious computation assume two crucial things. That the people in question have the same odds to have contracted the virus that anyone in the population, and that they are independent, concretely not evolving in the same circles. Both assumptions are false here. Boris and Charles see many people, so they have a much more chance of being infected than a standard British subject. And they see each other, or at least they both regularly see the Queen, so they are not independent. (By the way, has the Queen been tested?)

The bottom line is that I don't think there is anything reliable we can draw from this trivia that both the PM and the heir-prince are contaminated in UK.

I agree. They both meet a lot of people. They both would ordinarily shake a lot of hands (not sure at what point they stopped shaking hands). There are lots of people in common who are 1 to 2 degrees of separation from both Johnson and Prince Charles.

Many of those people close to both Johnson and Price Charles themselves have high risk factors for contracting COVID-19. Those common contacts also meet lots of people. There's also far more international travel among this group than across the general population.

They do have a few lower risk factors, but probably not enough to offset. They don't travel by mass transit. I assume that they rarely open a door themselves in a public place or touch elevator buttons because of security / staff.

Yeah, this gets to the whole concept of superspreaders. People with very large social networks are both most likely to get it, and most likely to spread it to others. Politicians and goverment leaders almost by definition HAVE to have large social networks. The President has to speak to his cabinet ministers, who have to speak to the heads of all the government agencies, who have to speak to their deputies and so on down the line. The branching factor is much larger so the number of people they are two or three degrees of connection away from is far larger. Trump speaks to the head of the CDC, who speaks to a deputy, who speaks to the head of a hospital in new york, who speaks to a doctor that is treating covid patients. That's 4 steps. High probability Trump along with many other heads of governments get infected.

meanwhile in this viral pandemic
-the histrionic sociologists at the who assert it is wrong to blame china for the spread of the virus have decided to blame the president and get this "american evangelicals"!
- Dr. Birx fortunately once again has an objective summary

"(And we should work much harder on producing better data.) "

Then we should throw every piece of data produced by China into the trash.

I'm still confused about Neil Ferguson's predictions. Kevin drum summarized the Imperial College estimates of US deaths as follows:

Do Nothing=2.2 million deaths
Do A Lot=1.1 million deaths
Do A Massive Amount For A Long Time=200 thousand deaths

So the difference in deaths between Do Nothing and Do A Lot is a factor of 2. And even the difference between Do Nothing and Do A Massive Amount For A Long Time is only a factor of 10. How is that in Britain, the difference between Do Nothing and what they're doing now a factor of 25? Did Drum just misinterpret the paper?

"...the number of asymptotic cases is very large..."

Yes, one would think so.

Especially if we're approaching the Singularity.

What if the Singularity is just 1 ?

"The best evidence (FT) for asymptomatic cases ranges from 8 to 59 percent, and that is based on a number of samples from China and Italy, albeit imperfect ones."

China's Data? Italy's data?

Italy hasn't blood tested anyone for antibodies. They code everyone who dies with coronavirus as having died FROM coronavirus (that's why 99% have other conditions) and don't test even a tenth of people reporting symptoms.

China lies about everything. Princess Cruise is nice data. Now can we somehow test the people on that ship who supposedly weren't infected for the antibodies please? Maybe literally everyone from that boat has the antibodies. It's certainly plausible.

And the Department of Public Health is awaiting results of other tests.

In total, Fresno County has 27 confirmed cases of COVID-19.

Department of Public Health stated 14 people contracted the virus through travel, three were infected person-to-person, another three got it through community-spread and 11 cases remain under investigation how they acquired it.

Fresno County has not reported any deaths regarding COVID-19.

The central San Joaquin Valley did have its first coronavirus-related death Thursday, though, with Madera County reporting a man in his 60s passing away.

Of Fresno County’s 27 positive tests, the Fresno County Department of Public Health stated that only one patient has made a recovery. That patient contracted the virus while traveling on a cruise.

Others who’ve tested positive for the coronavirus in Fresno County have been asked to stay at home for 14 days or longer. None have needed hospitalization.
My home town is still in phase one. We have only 200 test kits in the county available for phase two which seems to be on its way. This town has about a half million, the whole county abut a million.

That estimate of 10% to 15% is based, in part, on the testing rates of other countries compared to the U.S., Lover said.

“There’s a lot of data in Italy that shows people with mild symptoms can still spread coronavirus easily, so it’s important for those people to be tested,” he said. “You need to see if they have it and minimize their social contacts to slow the spread.”

About 10% of all cases are being reported, said Thomas McAndrew, also at UMass Amherst’s School of Public Health and Health Sciences. He has been aggregating data from experts’ models.


you say there's no reason to believe it, but it would neatly explain the similarity of the growth curves for every country to date despite very different intervention regimes

as for the heterogeneity of the fatality rates, apart from sampling bias and counting protocols, i have my guess: fecal-oral transmission rates vary by region, locality, and setting (nursing homes, poorly run hospitals, unsanitary homes, etc.) and that particular mode of transmission results in a more severe illness and greater probability of death

because there is also a correlation between age (and comorbidities) and fecal-oral transmission rates, probably much stronger in some regions due to hygiene practices, we see pockets with extremely elevated fatality rates.

in any case, we'll soon find out who is right
Testing results from Covid tracking. This is fairly inaccurate results reported by states:

Positive 82,286

Negative 458,432
Hospitalized 10,465
Death 1,199
Total test results 540,718
The stats tell us that 80k out of 450k have the disease alread. That is about 15%. But other stats are reporting 1%, using different aggregations I suppose.

We still have invalid testing numbers.

What assumption should I make about phase two when it hits my hometown? The worst case assumption seems to be a 17% infection rate.

Then look at this number:

N.Y.C. Death Toll Hits 365 as Case Count Tops 23,000 - The ...

Basically a 1.5% death rate. Except the other stats say some 17% of New York should have the virus, New York at 20 million, then en percent is 2 million. The stats are way off. No one seems to agree on what to count.

"The stats tell us that 80k out of 450k have the disease alread. That is about 15%"

The population of people tested would presumably be much more likely to have the disease than a random sample of the population.

"The worst case assumption seems to be a 17% infection rate."

No, that would be good news. If 17% of the population is already infected, the the hospitalization and death rate is much lower than what we believe it to be.

Btw, on the subject of reactions to the new Oxford Study, for a qualitatively better take than Tyler's here's a bit of real peer review in action -

I'm happy to see that at least one of the professors here suggests "Alternatively, the idea could be tested by large-scale surveys of virus genome diversity – this would confirm the much earlier introduction date that is central to Laurenco et al.’s theory (because this allows a longer period of, mostly hidden, exponential growth). Both serological testing and virus genome analysis are planned in the UK.".

That's an option I raised a few days ago here on a similar thread which no one responded on. So I am gratified to see it was fairly reasonable after all.

And of course the same deCode genetics Icelandic mob are ahead of the curve on that one as well -

This is just one of the startling new discoveries deCODE has uncovered from its analysis of the genetic sequences of 40 COVID-19 strains found in Iceland. According to Kári, the diversity of genetic sequences found in COVID-19 samples taken in Iceland indicate that the virus was brought to Iceland from a wider range of areas than was previously thought. The main origins of Icelandic infections are currently thought to be Italy, Austria, and Britain. A football match in the U.K. is thought to be the source for seven infections in Iceland.

Again, I'll return to the drumbeat that China has been very successful in implementing measures, but this kind of analysis that cuts to the real epidemiology and ecology of the virus seems like stuff that, if they were really as scientifically impressive as they seem, they should have underway.

That first link is gold. Thank you!

Strain diversity assessment and branching times based on US, UK, Iceland, China data -

To begin, Washington State has had 354 viruses sequenced (out of 453 in the US). Here we see that the large majority (83%) of sequenced viruses from Washington State appear to descend from a single introduction event in late Jan or early Feb.... This transmission chain has gotten big enough to throw off its own sparks, with sequenced viruses showing up elsewhere in the US, Australia and Iceland... Sequenced cases from elsewhere in the US and Canada are sometimes related to this transmission chain, but most (77%) derive from separate more numerous and smaller introductions. However, it's possible that further sequencing of other regions will change this picture. ... 201 genomes from the United Kingdom. These show a different pattern in which there are multiple clusters of related viruses, suggesting repeated introductions into the country followed by local spread.

Implies that contact and tracing worked plus luck with un-traced infections may have worked fairly well for the UK to the extent they could - with no local strains going particularly big - but ultimately in the face of a lack of a travel ban, too many incoming infections from all angles are too tough to swat?

Seems like estimates of time are difficult here, because sequence lengths are short and surviving mutations infrequent(?).

I'm hopeful that further regional sequencing will help estimate absolute prevalence and help quantify the degree to which social distancing efforts are slowing transmission.

One interesting factor I haven’t seen discussed is the disconnect between the range of spread (in virtually every country in the world now) but the lack of depth of spread (san Marino highest reported infection rate of population at 0.6%).

Attempts to limit spread have varied greatly between countries, and the disease obviously spreads quickly. As % of population infected rises, endemic infection should become harder to stop not easier. Why is nowhere in the world showing > 1% infection?

That's consistent with communities taking increasingly harsh containment measures as cases grow.

Keep in mind that exponential growth in a large population gets off to a slow start. Going from 100 to 100,000 cases can take as long as a month and just about nowhere in the world is willing to tolerate unhindered growth in cases oncd the consequences are obvious.

I get the nature of the spread and the intent of governments to prevent it. What I’m saying is a bit like the Fermi paradox. Everyone wants to prevent indigenous clusters. 0% of nations have succeeded in this, but 100% have cut endemic transmission before passing the 1% threshold, despite b being harder than a, and b being attained regardless of strategy attempted or when it was commenced.

Evidence from serological testing a CO county suggests very high fraction of US pop remains susceptible,

Here is a way to estimate indirectly:
"Recent data clearly show the spread of Covid-19. On March 19, the share of Americans with temperatures indicating they had flu-like symptoms was about 4.9% when it typically would be expected to be about 4.0%. This was likely a result of the spread of Covid-19, according to Kinsa’s researchers."

.9 percent of the population comes to about 3 million.

Great link; can we really be sure it's Covid19 though, and not something else causing that? That said, if so the actual shadow number of infections would seem to have to be double or something that extra 0.9% to account for asymptomatic share, without elevated temperature or anything. Possibly more so if it also misses mild cases.

Nice to see them talk about the fact that other infections will also get hit by social distancing. It seems that if it works, it will be hard to totally distinguish reduced prevalence of Covid19 from correlated reductions in comorbud flus and other respiratory infections. Also expect that once distancing ends, there will be a lull then spike in flu deaths etc.

Part of what will emerge from this crisis will probably have to end up being depersonalised mass monitoring and data aggregation of bio indicators of health; temperature, shift in skin tone using cameras, breath capacity. Mass analysis may not be good enough for individual diagnosis, but could probably identify if *something* is shifting. Tie that to basic travel data indicators and demographics and get something meaningfully predictive.

For US.WA no. Additional COVID.19 tests do not change the CFR much. For NY yes. The more test done, the more deaths can be attributed to COVID.19 rather than sweeping them under the influenza death carpet. The current CFR for NY is very much smaller than that for WA. The problem with NY is the limited hospital beds.

The distribution of the US regional CFR is bimodal, similar to that for China, though the more virulent clusters like those in Wuhan or US.WA are smaller but an order of magnitude more fatal.

If you could check everyone. Then you can say about 100% of patient statistics. And, of course, far more people are infected than we think ...

Comments for this post are closed