back

by alexandercrohde·6y ago·view on hn ↗
Unfortunately garbage-in -> garbage-out.

That chart shows US at 233 confirmed cases.

According to NYT, over 10x that number are self-quarantined in NYC alone [1].

Now it's technically true that those cases aren't medically 100% factually confirmed, because NYC has only performed 35 tests total so far. But it gives a very inaccurate picture of reality. Garbage in, garbage out.

1. https://www.nytimes.com/2020/03/05/nyregion/coronavirus-new-...

11 comments
This dashboard is not intended to visualize computed estimates from models. It tracks the official confirmed cases, which is a very different and useful piece of data.

You are unfair by stating it is garbage and misrepresenting what it shows (confirmed cases vs. estimated cases).

The comment above is partially true. The data sources for each country are different, and thus have different error profiles.

It's statistically incorrect to assume they can be merged and treated as a single data set.

What is more useful is to examine well bounded and vetted subsets of the data.

Once the South Korean cases become more mature in a couple weeks, I think we will have very high confidence mortality data.

Including poor data quality sources will just introduce biases.

What's garbage is the US response to it.

Italy is running 2500 tests a day.

Oregon has the capacity to run a total of about 40.

The WHO fact-finding mission to China reported that they ran 350,000 tests in the province of Guangdong alone (though it must be noted that its population is 110 million) ...
Agreed. Seems like people in Oregon choose to mostly ignore the issue
It's not just Oregon; that's just a number I saw this morning in the local paper, but I suspect it's similar everywhere. The US did not use the advance warning we had to prepare.
You are right but unfortunately almost everyone calculates the mortality rate based on confirmed cases alone, and that leads to a gross over-estimation of the dangerosity of this virus. So these visualisations are still kind of misleading in a way.
Isn't that true of all flu viruses, every year? Meaning this measure is still valid to compare with other flu epidemics.

So that leaves this one at 10X or more fatalities, however you calculate it.

I am not sure. Looking at the CDC website [1], it looks like they extrapolate cases for the flu. They seem to estimate that roughly half of the people contaminated with the flu do not see a doctor (I would have guessed the numbers would be even lower, I had the flu a couple of times, it didn't even cross my mind to see a doctor).

[1] https://www.cdc.gov/flu/about/burden/how-cdc-estimates.htm

Yeah but if every virus is modelled this way then the data is representative...
I don't agree. The sample is not representative of the whole. The only useful inference is a lower bound on how many cases there are, but this is more likely to mislead that inform.

If you know that officially confirmed cases don't reflect reality, and you care about reality, then the dashboard is not useful.

According to this estimate[0] based on viral mutations, there were 600 cases in Seattle on March 4th. That puts the multiplier at about 20x cases per confirmed case. With an estimated doubling time of 7 days[1], that's about one month worth of spreading.

(Disclaimer: all numbers approximate, high uncertainty everywhere.)

[0] https://bedford.io/blog/ncov-cryptic-transmission/ [1] https://bedford.io/projects/ncov-phylodynamics/

Why is it inaccurate to state the number of confirmed cases, over reporting the number of self-quarantined? If I go to the hospital w/ flu-like symptoms and they tell me to stay at home just to be safe and it turns out I just got another disease, do I count towards the self-quarantine numbers? Does my family? If I work from home by choice to avoid exposure to public places, does it count? Does the number track compliance rates? Are we really getting accurate meaningful data from that measurement or are we feeding a bias towards alarmism?
It's inaccurate because the US is not testing people with symptoms who certainly would have been tested in other countries.

South Korea has performed over 100,000 tests. NYC has performed ~ 35. [1]

I'm really curious what possible reason there could be for that. Regardless, there is no reason to believe "tested cases" is going to give a remotely accurate picture of "actual cases."

1. https://www1.nyc.gov/site/doh/health/health-topics/coronavir...

> I'm really curious what possible reason there could be for that.

It's right in the page you linked:

> If you are experiencing fever, cough or shortness of breath, and traveled to an area where COVID-19 is spreading, call or go to your health care provider. Your provider will work with the Health Department to determine if you need COVID-19 testing.

Right now, the city is only testing symptomatic people that have been outside the country recently to one of the five nations with widespread cases, and people in direct contact with confirmed cases in the city. It's a (possibly too) conservative rationale, but it is a rationale.

You can't realistically track "actual cases" accurately anyways. For example, people might not go to the hospital. This is a well known challenge with population estimation e.g. domestic violence numbers)

What you seem to be getting at is that a) this chart doesn't display the total number of tested cases and b) external reports of number of tested cases point to very low number of tests in the US

But that's not a reporting problem, it's a risk/cost analysis problem by the part of the responders. Yes, you could in theory test everyone in the country every two weeks, but it would arguably be very wasteful and impractical to do so, so CDC/hospitals/etc need to figure out what is the right balance as the epidemic unfolds. Whether they are doing a good job is indeed debatable, I agree.

This is just a straw man. No one is suggesting that they test everyone in the US, but at the level that epidemiologists and the WHO says is absolutely necessary to understand and respond to the spread of the disease.

That they’re not doing testing at anywhere near the necessary levels, while countries like South Korea are, is not “debatable,” but evidence of their failure.

Also your supposition about tracking actual cases is false. Singapore has significantly curbed the spread of the disease by doing extensive contract traces, testing, and quarantines.

It's not a straw man, the spectrum goes from no testing to testing everyone, and _measuring_ "actual cases" accurately (especially after community transmission stage) requires one to be on the latter end of the spectrum. I did say that it's unrealistic to go that far, and you're right that more aggressive testing will obviously lead to better visibility. My point merely was that the rationale for not ramping up testing to the wazoo could be attributed to risk/cost analysis. Incidentally, that might even explain why Singapore with a population half the size of Wuhan's would be able to respond more elastically than the US with a ~300M population and high levels of bureaucracy and government fragmentation.

Btw, I should clarify that when I said the word "debatable", I was using an euphemism.

This is almost certainly getting cause and effect backwards. That is, South Korea has been testing a large number of people because there were a large number of likely cases to test, mainly due to that one cult inadvertantly spreading the coronavirus to so many people. You can see this in the number of tests that turn out positive: it's hovering somewhere around 4% in South Korea and slowly dropping now they've got through the really likely targets, but was in the high single digits earlier. Back when the CDC was only testing the most likely cases, their number was around 3%. A lot of the media reporting on this is dubious and might lead people to believe that the US had South Korea-level numbers all along and just wasn't spotting them due to not testing, but that's unlikely.
CDC’s lackluster response is absolutely confounding. One could suspect political involvement, but I can’t see who would stand to benefit.
Wasn't the CDC budget cut drastically by the current administration? You reap what you sow.
Nope this was misinformation sowed for political purposes: https://apnews.com/d36d6c4de29f4d04beda3db00cb46104
No, the budget was not cut.
Well Congress is the ones who actually pass the budget so no?
Why would you have any confidence with CDC?

Go browse through the weekly bulletins from 2003 regarding the SARS 'epidemic' in the US. https://www.cdc.gov/mmwr/preview/mmwrhtml/mm5218a2.htm

You'll read of the hundreds of 'confirmed' and 'probsble' cases and dire warnings and the like.

Then, when the dust settles and years pass, the CDC reluctantly admits the following: ------ In the United States, only eight people had laboratory evidence of SARS-CoV infection. All of these people had traveled to other parts of the world where SARS was spreading. SARS did not spread more widely in the community in the United States. https://www.cdc.gov/sars/about/fs-sars.html#outbreak ----

The market and the current administration both benefit from a low count of confirmed cases. The administration has already been criticized for repeatedly attempting to slash the CDC budget[1] and terminating the epidemic response team[2]. Trump also keeps attempting to downplay the severity of the virus so it would not be a good look if the true spread of the disease was widely known. Knowledge of the spread would also probably trigger more panic selling in the markets which the current administration also tries to avoid as much as possible.

[1]: https://abcnews.go.com/Politics/trump-cut-cdcs-budget-democr...

[2]: https://www.washingtonpost.com/news/to-your-health/wp/2018/0...

That just doesn’t make sense to me. Uncontained, this disease will grow exponentially until tens of millions are infected at once, and it eventually burns itself out. But by then, the disaster will be obvious and we will have the same result. Only with more casualties.

Certainly anyone who had that sort of ulterior motive would understand this.

And China may have tested every one in Wuhan, which has almost the same population as New York.
In Seattle, there are not enough tests to test everyone diagnosed with the disease based on symptoms any more. So some patients are getting diagnosed with COVID-19, but if they are not in severe condition, they just get sent home without testing. These patients also don’t count toward the “confirmed” numbers.
No, people are instructed to self-quarantine merely for being exposed to an infected person. The large majority of those people are not expected to eventually test positive.
Burn two weeks of PTO when you may not even be sick. I wonder what the compliance rate is?
Yeah, self-quarantining went great in New Hampshire:

https://www.nytimes.com/2020/03/04/us/coronavirus-new-hampsh...

That's really bad - the guy even knew he was sick. I wouldn't have gone to a party with a fever no matter what virus I was hosting.
If you're lucky to even get PTO! :|
From my perspective, if (whoever is in charge) was really in the business of providing accurate numbers, they would test a statistical sample of these 2,300 people to estimate what percentage of them actually have the virus.
Here in Germany a little known tidbit is that the routine sampling system that is already in place to observe influenza has started to check for SARS-CoV-2 as well. It's only a sampling of people sick enough to visit a doctor (not whole population random sampling), but at least this sampling is not limited to people with a known infection path which would completely blindside you to community spread. I don't think that this has been running for long enough to already have meaningful results, but it will eventually put an end to flying blind. Given the value of discovering undiscovered clusters and getting better statistics I think that this is a much much better allocation of testing resources than checking the most likely candidates whom you'd want to quarantine anyways due to the likelihood of false negatives during incubation.

A very similar test allocation might also be running in America without anyone but a few experts taking notice. Here in Germany it has technically been on the news but it was completely drowned out by the pointless shouting match between "it cannot be worse than a cold" (because it would interfere with my precious vacation plans) and "our healthcare wouldbe just as overwhelmed as in Wuhan" (because, I don't know, maybe I'm just not racist enough to assume that it will somehow only be bad in China)

I think the CDC has been rolling SARS-CoV-2 testing out to the US influenza surveillance system. They actually tried to do this in mid-February but ran into problems with the testing kits. Obviously this would have left the US in a much better position to monitor it, though I doubt that'd stop wild theories about how it was everywhere spreading virally.
Fascinating. I'd love to see a system like that in America.
There are R0 and transmission probability estimates out there. If you sniff around I think you might find the study you're asking for. As usual, media outlets are not scientific, so if you want proactive understanding (as opposed to reactive reporting) you will need to follow the literature yourself. Too bad there is no news service for that.
i smell a market opportunity
And the CDC itself says there are only 99 confirmed cases.

What a shitshow.

As of 3/6/2020 8:22 am PT: https://imgur.com/a/D0ICvoP

Source: https://www.cdc.gov/coronavirus/2019-ncov/cases-in-us.html

164 now, though note that "numbers close out at 4 p.m. the day before reporting" so that page is basically the count as of yesterday.
I agree that it's probably not a 100% accurate picture, but if you included in the "known cases" every single person who stayed home out of paranoia because they had a stuffy nose, then you'd have the opposite problem of numbers being bloated and overblown. The media is already chomping at the bit to sow fear and paranoia in exchange for views, I'm ok with looking at data with the understanding that it's probably underrepresented versus data that is probably just a complete guess. In an ideal world, yes, every single hospital would have thousands of test kits on hand to test everyone but we aren't there yet.
I think if you keep in mind that it reports a specific thing, then it is useful.
The US is already exporting cases to other countries:

https://www.thestar.com/amp/news/gta/2020/03/05/person-who-a...

Garbage out, indeed.

New York Times? Are they like world-class science researchers like John Hopkin Institute or are they just a bunch of shady journalists with a fantasy narrative that they desperately need to maintain?
At first I thought US numbers were ridiculously wrong out of poor accounting but it is becoming increasingly obvious there is a massive coverup from the top down of the administration.

They are likely waiting for tests to be privatized so then they can claim numbers can't be released because they are private.

Except deaths will be recorded, so this is going to backfire badly, ie. there will only be 1000 tests but 1000+ people will have died. The numbers are going to explode at some point, especially in Washington State