Diamond Princess for COVID-19 Analysis

[Bo Deadly]

Basically all of the numbers of COVID-19 cases are largely incorrect because we're not testing people equally. People generally only get tested if they already have a fever and cough. We know that the majority of people who get Coronavirus have mild symptoms. So when calculating the usual percentages, the denominator is far too small.

One of the first significant COVID-19 outbreaks was on the Diamond Princess cruise ship which was quarantined in Japan for over 3 weeks. All disembarked on March 1. As of today March 24, 712 of the 3,711 passengers and crew were infected and 10 died.

The Diamond Princess is very interesting because all of the passengers were almost completely isolated and the quarantine was weak. It would be difficult to stage a study that could simulate an outbreak in isolation as well as it occurred naturally on that ship.

So if we assume that the data is a better representation of effects of Coronavirus, we can now compute the usual percentages with greater accuracy:

Infection %: 712 / 3,711 = 19%
Morality %: 10 / 3,711 = 0.27%

And it is worth recognizing that these numbers are almost certainly somewhat conservative because 1) the demographic is people who tend to be a little older and 2) they were literally trapped in a relatively small space.

 

[Script]

Yes, very interesting case indeed, I fully agree. The wake-up call, at least for Japan.

Also worth noting is that these were among the very first people outside China that we know of who got group-infected -- and at a time when health infrastructure (meaning hospitals) could cope with cases, unlike in Wuhan and Italy... ...

But also worth noting is that people were finally taken of that ship in March, BECAUSE quarantine conditions turned out to be weak and ineffective for some.

 

[dogears]

We stopped the infection experiment arbitrarily at 20%. It didn’t stop on it’s own.

And the fatality rate is deaths per case in most of these discussions, not typically mortality. Mortality is generally given as a rate of deaths per capita.

So you need to divide deaths per case, and you’ll get something like 1%.

 

[Script]

Possibly -- and with health infrastructure fully functional.

 

[Bo Deadly]

We stopped the infection experiment arbitrarily at 20%. It didn’t stop on it’s own.

True. But because it was such a small space, the infection % would have eventually become unnaturally high. In the wild people would scatter and isolate. So the number is not completely meaningless if you consider it as a function of time.

And the fatality rate is deaths per case in most of these discussions, not typically mortality. Mortality is generally given as a rate of deaths per capita.

So you need to divide deaths per case, and you’ll get something like 1%.

I don't understand. Please explain.

I would think that because it was a closed system with the same individuals present at the start and end that it would be a pretty good measure of the number of deaths in a population which is what I think of as mortality.

 

[JohnRoberts]

Speaking of the princess they found traces of covid virus on cruise ship surfaces over two weeks after passengers left....

Traces do not equal infection vectors, and this was before the ship was scrubbed..

https://thehill.com/policy/healthcare/489213-cdc-says-traces-of-coronavirus-found-on-cruise-ship-surfaces-after-two-weeks

We still have things to learn about this virus.

JR

PS: During today's press conference I saw Dr Fauci touch his face multiple times.

 

[dogears]

True. But because it was such a small space, the infection % would have eventually become unnaturally high. In the wild people would scatter and isolate. So the number is not completely meaningless if you consider it as a function of time.

Epidemics stop / slow based on herd immunity. Unless everyone "freezes" indefinitely, infections keep going until they hit a percent of the susceptible population to slow them down. Without intervention it's likely that ~50-70% of the population will get this in 3-6 months. WITH intervention its likely 50-70% get it in the next couple of years. Then it dies out or becomes endemic like the flu. BUT, this virus has a RNA "checksum" or proofing that the flu doesn't, so hopefully it won't become endemic.

All above changes bigtime with vaccine obviously.

I don't understand. Please explain.

I would think that because it was a closed system with the same individuals present at the start and end that it would be a pretty good measure of the number of deaths in a population which is what I think of as mortality.

Yeah, deaths per population is fine, but we stopped the infection in the population. So better to look as deaths per infection, because we don't know when infections would have stopped (20% is probably low).

 

[Bo Deadly]

Epidemics stop / slow based on herd immunity. Unless everyone "freezes" indefinitely, infections keep going until they hit a percent of the susceptible population to slow them down. Without intervention it's likely that ~50-70% of the population will get this in 3-6 months. WITH intervention its likely 50-70% get it in the next couple of years.

Where are you getting those numbers? Provide a citation.

Those numbers sound very high to me. Even for a densely populated city.

 

[dogears]

The formula for the number of susceptible people s at the end of a pandemic in a well mixed / random population is
log (s) = R0 *(s- 1)

If you rearrange it to

R0(s - 1) - log (s) = 0 = y

and then solve for y you get different roots that correspond to the fraction of the total population of people left who didn't get sick at all.

For R0 of 2 the fraction left is 20%.

For R0 of 1.2 it’s 70%.

Another way to estimate it is (R0-1)/R0.

This disease has a widely reported R0 of around 2.2. So, that’s around 55% at the end.

 

[Bo Deadly]

A simple log model is totally inadequate. That assumes we're a bunch of dumb monkeys. We have intelligence of our situation. You have to consider population density, cultural factors, etc. And even with a single-order model, what I've heard is that the effective R0 is more like 1.5.

 

[dogears]

There have been a huge number of studies showing it at R0 of 2.2 or even higher in the early outbreak stages.

The range given is 30-70%.

This is from an article about modeling this in Nature.

For a completely novel pathogen, especially one with a high (say, >2) basic reproductive number (the expected number of secondary cases generated by a primary case in a completely susceptible population) relative to other recently emergent and seasonal directly transmissible respiratory pathogens, assuming homogeneous mixing and mass action dynamics, the majority of the population will be infected eventually unless drastic public health interventions are applied over prolonged periods and/or vaccines become available sufficiently quickly. Even under more realistic assumptions about mixing informed by observed clustering of infections within households and the increasingly apparent role of superspreading events (for example, the Diamond Princess cruise ship, Chinese prisons and the church in Daegu, South Korea) at least one-quarter to one- half of the population will very likely become infected, absent drastic control measures or a vaccine. Therefore, the number of severe outcomes or deaths in the population is most strongly dependent on how ill an infected person is likely to become, and this question should be the focus of attention.

https://www.nature.com/articles/s41591-020-0822-7.pdf

20% is very low and very unlikely end point.

As reference, 2009 H1N1 had an R0 of 1.3 and ~24% of the world got that. Simple log model suggest 45%.

The simple log model for R0 of 2.2 is 85% infected. I was “de-rating” with a ~50% estimate.

To be clear on a cruise ship you’re more likely to approach the theoretical limit, since it’s a small population and there’s no good way to prevent the spread.

 

[Bo Deadly]

I think you're playing fast-and-loose with the numbers dogears. I'm not reading nature.com. That's too much for me. And I have a BS in Biochemistry.

 

[dogears]

I’m not playing fast and loose with anything. The log model with R0 of 2.2 suggests 85%. The simpler fraction suggests 56%. There are a ton of papers that arrive at R0 of >2.

R0 changes with intervention obviously, so the final outcome is anyone’s guess.

Dr Fauci estimated 30-70%, CDC has said the same, Governor of California said 56%. A paper published by a team from Oxford says 80%, but with a lower severity.

The portion from the Nature paper is quoted. Basically if it’s this infectious, a crap ton of people are simply going to get it. So... how severe is it is the key question.

I mean.. fast and loose would maybe be extrapolating a cruise ship with 3700 people to the whole world right?  ;D

 

[Bo Deadly]

I’m not playing fast and loose with anything. The log model with R0 of 2.2 suggests 85%. The simpler fraction suggests 56%. There are a ton of papers that arrive at R0 of >2.

That's the basic reproductive number (the dumb monkeys number). The effective reproductive number (the number "with control") is almost certainly not > 2. This is not a computer simulation. And just like on the Diamond Princess, the total infection % is a function of time. Eventually most people will get it at some point. But there is no scenario where 70% of people have Coronavirus at the same moment.

 

[dogears]

I never said any of that. What I said was:

Without intervention it's likely that ~50-70% of the population will get this in 3-6 months. WITH intervention its likely 50-70% get it in the next couple of years.

Re in Wuhan didn’t drop below 1.5 til they did the big time shutdowns. 

So, yeah, once quarantines happens all bets are off. But the cruise ship... which is what I thought we were talking about.. would almost certainly not have ended until a substantial portion of the population was infected. Like 50%+.

 

[Script]

SDF Hospital has released a report on treating the passengers from the Princess D.
Newspaper article about it  is here:
https://www.japantimes.co.jp/news/2020/03/25/national/science-health/80pc-diamond-princess-coronavirus-mild-symptoms/

The gist of the report itself is that even asymptomatic patients (no fever, no coughing,) can develop abnormalities in lungs only revealed on CT. And even when tested negative in PCR check.

The report is accessible via the hospital page. They are looking at 104 people who agreed to the study / report.