We’re Testing the Wrong People

100% there is a shortage of reagents and kits. There are certain countries that are not letting reagents and kits ship out, and we are in no position to call them out on that.

>That doesn't stop the American press making it sound like not achieving those 500,000 tests a day is the result of some uniquely American political failing though.

I don't think we're unique in where we are in comparison with other countries as far as testing, but we are in a unique position in regards to being able to put some strong leverage on getting other countries to work together to respond to this pandemic. We're not doing that and as a result, other countries are having to "roll their own" response to this pandemic just like we are. The lack of coordination within our country is putting us at a unique disadvantage, even if we had better coordination world wide.

This pandemic is a exceptional global crisis in a way we've never seen before, and thus should've led to a addressing it at a global scale in a way we had never seen before. Instead the global response has been crippled in numerous ways at various different points in time.

The American political failing is in not being the true world leaders, and the world is, IMHO, worse off for it.

There are even plans to increase testing to 4,5 million(!) tests per week in Germany, or ~650k per day. Currently there is unused testing capacity, so it looks like the old restrictive conditions for testing (basically: symptoms and (contact to infected person or stay in high risk area)) are still in place and should be updated ASAP.

The political will should be fewer American lives lost. Politicians are framing this about the economy and states rights and other BS as a way to distract the public from the fact that people have died because we didn't act soon enough and decisively enough.

I also don't think we need to get the word out about how testing is important. People are cooped up in their homes precisely because we don't know who has it and who doesn't. Testing could resolve that.

I have little faith in the legislation from Washington for many reasons. The primary one being that before the crisis, during it, and at this very moment there is a lack of acknowledgment of the actual scale of this pandemic. Even worse a large part of the legislature fails to take this seriously, thus making the scale of the issue a secondary problem. What's been passed so far has already missed the mark as far as funding testing and properly ensuring that those who need the money get it. Instead, big business gets "relief" with little oversight, which given the past actions of this administration clearly isn't just by accident.

The reality is that testing is still significantly behind and efforts to increase testing won't make the news cycles. I strongly doubt that we'd be able to scale up quickly enough to make a meaningful dent in this anymore than strictly following stay at home orders. Unfortunately even that is too much to ask us to do as a society it seems.

You can get part of that with excess death rates. From the weeks overall deaths subtract the average death rate for this week in past years. That should give both people who died at home and consequential deaths.

The number of deaths may be off but that is likely to be dwarfed by the much higher number of total cases that haven't been confirmed. Between this study, the LA study, and the Santa Clara study there could be 10-80x as many cases. That is more significant that the death count being off by a couple percent when it comes to determining the fatality rate.

Most nursing home deaths are not getting tested either.

>> Maybe it's reliable in the statistical sense, but not reliable in the sense that a layperson or a politician would be able to rely on it to guide decision making.

This seems to be what we're relying on in the UK. Hospital admissions for Covid show a good/relatively quick estimate of how measures are working and then a few weeks later a reflection of that trend in deaths confirms it.

In the US, if the person tested positive, it is classified as covid19 death.

Also it seems that 98% of covid19 deaths have at least one comorbidity, 75% at least two. So I would expect that in many cases the virus may have just accelerated something that would have happened within a few months anyway. So you kind of need to look at how the full year death rate evolved to have an idea of the net impact of the virus.

This is tricky to understand, but "all cause excess mortality" is pretty well understood by statisticians. They use it all the time for flu.

Here's one document from the UK that mentions the kind of testing they do to arrive at "deaths to flu" figures. They have international programmes to see what respiratory diseases are circulating, they test people who arrive in hospital for those, they know roughly how many people in the population they expect to die (because RTAs, cancer, heart disease, etc all stay sort of stable from one year to the next). They also do survey work to see if people have flu symptoms but haven't seen a doctor, and they can see how many people attend various healthcare settings for flu.

It's all a lot more complicated than "did someone write flu on a death certificate".

https://assets.publishing.service.gov.uk/government/uploads/...

Many coroners in my state have been quoted as saying they are unable to test the dead, as tests are prioritized for the living. Our state health department says that they do not report suspected COVID-19 deaths to the CDC if the person did not test positive.

I believe, at least in New York, people are being counted as covid-19 deaths if they were in contact with someone who was diagnosed. They can count as a covid-19 death without ever being tested for it.

I disagree. In order for a hospitalization to count, they have to have a positive test. First, we have to factor in test inaccuracy and then the fact that doctors are still triaging tests even if a patient has symptoms. These numbers can't be considered accurate until we have enough tests to test everyone in the waiting room indiscriminately.

Gah, I've passed the editing window. If it's confusing to read, it is supposed to say:

> the specific construction lets you pinpoint the people who come up positive

So you might together the samples before testing?

germany has been doing this for a while now

This is exciting. I've been following the idea for a month or so. Did you know that a group in Israel has tested the idea and found that it worked with the swab tests?

I have a link on my notes below. I'll add your site to the reference list at the top tonight.

https://github.com/EFavDB/pooling

The confidence intervals are wide enough it wouldn't be very useful. If you sample 1000 people and get 0 positives, the 95% confidence interval is 0% - 0.36%. If you're the governor of Illinois, with a population of 12.67 million people, that upper bound equates to 45,612 people: far, far too many to handle with contact tracing and isolation. So you need to look to other metrics to make the decision about whether to issue or revoke a stay at home order.

So what are they looking at? Basically, if they have any evidence of community transmission, they need to issue the order or keep it in place. That means that if there are too many symptomatic people to even attempt contact tracing, you're not even close. Once you're low enough you can do contact tracing, you wait until you have a handful at most of symptomatic people you can't trace back to a known case. Until then, all you can do is keep the stay at home order in place, buy/beg/borrow/steal ventilators and PPE, and build testing and contact tracing capacity.

Seattle has been running a test like this. I believe they're allocating 300 tests per day.

https://scanpublichealth.org

It's a problem of limited resource allocation. The _point_ is that we have limited tests, so given limited supply why are we wasting the tests on those we think are sick. The limited supply could still be directed towards asymptomatic carriers. Wouldn't be testing everyone we want, but it's still more valuable than the allocation of tests right now.

But what's the point of testing people that are already sick?

We don't treat Covid patients any differently than any other patients that develop pneumonia

> At the beginning there were some authorities (China, WHO) claiming there was no human-to-human transmission.

I don't understand how this misconception persists. That is not what the WHO statement said.

The statement was that the outbreak was not being sustained by community transmission, and that they believed (this was very early on) that the cases from the Wuhan market were all due to exposure to a single case or pathogen.

This was, obviously, wrong. But at the time it wasn't unreasonable to believe. That's the way most outbreaks work when you catch them early.

At no point did anyone say that a respiratory coronavirus was not contagious, which is how people insist on treating that statement. Of course these things are contagious, that's how they spread.

Oh, wow, that was my mistake. Thanks. I was getting the data from here [0], which for me had a lot of countries selected but not all of them. I didn't realize that it did not include by default a number of countries. When I include them all I see that the U.S. is in the middle of the pack.

[0] https://ourworldindata.org/grapher/full-list-cumulative-tota...

Sometimes, I feel like concepts such as these need to be translated into terms the average hackernews reader will understand.

Let's say your key metric you're trying to maintain or improve is P50 request latency. Its relatively easy to exhibit an acceptable P50 to 10 users, from 1 VM. Its actually harder to deliver the same P50 to 100 users, from 10 VMs. Even harder for 1000 users and 100 VMs, snd so on. You run into previously fine bottlenecks you didn't know existed. Issues which were invisible at ten users become cascading failures at thousands.

It just keeps getting harder the more users you add, even though you're adding proportionally more underlying VMs. There is no end to how difficult it gets, until eventually, it becomes impossible.

Testing is a problem with difficulty that scales with population size. Of the top ten countries by population size, there are only two countries that have higher testing per capita than the US: Russia and China. Those three countries are pretty close; no one else, anywhere in the top ten, is remotely close to the US, Russia, and China.

(Kind of funny how the top two testers in the top 10 countries by population are dictatorships, who have every reason to maintain a strong, powerful image. Funny how it works out like that, isn't it? Do you believe their numbers?)

I run the fastest website on the planet. No one but my mom visits it, but its still the fastest. Lithuania (the world's top daily tester per capita yesterday) is that website. The US is, uh, lets say Google. Should you bemoan Google for taking longer to load?

> In tests/capita US is 42nd

That requires that you compare the US to Iceland, Faeroe Islands, Falkland Islands, UAE, Malta, Gibraltar, Luxembourg, Bahrain, Estonia, San Marino, Cyprus, Isle of Man, Lithuania, Brunei, Qatar, Liechtenstein, Ireland, Latvia, Andorra, Channel Islands, Bermuda, Aruba.

Nice try.

Back on planet earth, here's the list of actual sane comparables that are in front of the US, in order:

Italy, Germany, Spain, Australia, Russia, Canada

And Russia is fake. They haven't tested much at all.

The US has tested at a higher rate than South Korea (which is behind all of those tiny nations also).

The US testing rate is just behind Belgium, a nation 1/29th the size of the US. And it's not very far behind Singapore at this point, either.

A) You'll know which communities have infections. Many places are saying: "it's not bad here, why should we isolate?"

B) If you know that someone with bad symptoms has the virus, then you can go test everyone they contacted. You'll find more people who currently have mild or no symptoms.

> Non-covid diseases still require the use of PPE.

The PPE required for some COVID-19 symptoms in the absence of clear indication of a droplet/airborne transmissible condition are different than for COVID-19. You don't usually use droplet, much less airborne, protections for every patient presenting with a fever, or GI symptoms, or several of the other ways COVID-19 can present, but you must with the prevalence of COVID-19 until you can rule it out.

And it's even more critical to identify COVID-19 or it's absence in sick patients when part of your containment strategy involves segregating ftreatment facilities (part of the capacity surge has been in establishing either new COVID-19 specific facilities or new facilities to handle non-COVID-19 cases while existing capacity is dedicated to COVID-19.)

> During a pandemic, it's more likely than not that any negative test on a symptomatic person is false.

I don't think we've reached the point where COVID-19 is the leading cause of fever, much less GI symptoms, so unless you define “symptomatic” extremely narrowly (like, “in need of a ventilator”) that's probably not true in this pandemic at this time in the US.

> During a pandemic, it's more likely than not that any negative test on a symptomatic person is false.

There are places with single-digit positive test rates, that would require a 90%+ false negative rate, which, as far as we know, is not the case.

yup, 'stats legend', not history

there's an american math society article about this -- it evaluates the fun version of the story in the context of wald's actual paper and a journal entry by another scientist named wallis

Awesome! Thanks for finding that. So they do give up the extra specificity from the reverse primer by having it be a constant region in the barcode, I guess having a sanger step makes up for that a bit though since you can read what it ligated to past the barcode and see if it's a SARS-COV-2 fragment, but that does slow down getting the results. Like a truck of hard drives barreling down a street versus sending data over fiber. Get a lot of reads in giant batches or get less reads but with less latency.

So I guess I'm still not seeing this. Can you give me a concrete idea of what this looks like? Let's assume that it is the middle of August and the food situation is still somehow stable, I install my contact tracing app, I get a clean test. Then what? If we miss one person, that one person can still infect an entire office building by coughing in an elevator. Because of the two week incubation period, those people will not know they're infected until at least a couple of days in, even assuming testing daily. We don't yet know how long a person can infect before testing positive, but this amount of time is most likely not 0. If this amount of time is less than 24 hours the person can still make it anywhere in the US by air. If any longer they can make it anywhere in the world. How do we not have constant low-level infection everywhere?

You and I are often in disagreement on here, but not on this!

I'm pretty sure GP means actual quarantine. We are not quarantining hundreds of millions. We are advising hundreds of millions to practice social distancing.

These are good points (though broad strokes). Mostly I'm just pointing out subtleties or challenges with this approach.

One difference is if a business re-opens and expects people to come in to work, the US probably needs legal protections so quarantined workers don't get fired or forced to come in illegally.

It's all about putting people into risk categories. Then each group's members can act in the optimal way for their group. Then retest/rebalance periodically.