When you sign up at CuriosityStream.com/Wendover, you'll also get access to Nebula—the streaming

site that Wendover is a part of.

Covid-19 is a devious disease.

It's not deadly enough to kill a large proportion of its hosts, so they go on to spread it to

an incredible number of individuals, meaning its fairly low mortality rate has already

killed hundreds of thousands.

It disguises itself, presenting like a common cold, yet can escalate to something far more

worrisome.

It also hides, with the possibility of being spread by someone who does not exhibit any

symptoms whatsoever.

All together, this leads to one simple truth: until there is an effective treatment, a vaccine,

or herd immunity, perhaps the most important thing a healthcare system itself can do to

minimize the number of cases and deaths is test.

Worldwide, diagnostic testing for Covid-19 has ramped up incredibly fast.

The first genetic sequence of the virus was published on January 5th, and within days,

the first testing protocols were published by researchers.

In the second half of January, the world was testing hundreds of individuals per day.

By February, they were testing thousands per day, then in March, hundreds of thousands,

and now, in July, 2020, millions of Coronavirus tests are performed and processed each and

every day.

Believe it or not, though, this is not enough.

The process has been mired in issues and, to get to the point where testing can actually

minimize the damage, rather than just quantify it, we need to rethink the system itself.

The core of any Covid test is a laboratory method known as a polymerase chain reaction.

This is a commonly-used procedure that copies DNA or, through reverse transcription, RNA,

enough so that it can be studied.

The various steps in the process—denaturation, annealing, and elongation—require various

specific temperatures, and so performing this process requires a machine called a thermal

cycler.

These thermal cyclers can reach precise temperatures incredibly quickly and accurately, but it

comes at a cost.

The cheapest thermal cyclers are priced in the thousands of dollars, but the high-throughput

ones used to process COVID tests often cost in the many tens of thousands of dollars.

Therefore, at this time, it is not practical for every doctor's office to have one, especially

as many less-automated variants require trained technicians to operate.

There are some faster point-of-care tests hitting the market, but they tend to be less

accurate, more expensive, and can process fewer tests at a time.

Therefore, while samples can be collected almost anywhere, processing them is a greater

challenge.

One of the largest testing providers in the US is a company called Quest Diagnostics.

At the time of writing, they have processed about 17% of all tests performed in the US,

and have the capacity for 130,000 each day.

They carry out this work in one of their twenty-eight labs spread out across the United States.

Of course, it's easy to see that there are wide swaths of the country without a lab nearby,

so, to get samples from where they're collected to where they're processed, they rely on

their own small logistics network.

This system involves plenty of ground transportation, but also, in order to quickly transport samples

from far-away areas to the lab, they use airplanes.

Each night, after the day's samples have been collected around the country, Quest Diagnostic's

fleet of airplanes get to sky starting at around 7 pm.

These twenty or so aircraft are primarily based in Reading, Pennsylvania and Lawrenceville,

Georgia.

As an example, on July 16th, 2020, one aircraft, a Pilatus PC-12 registered as N338QD, took

off at 8:45 pm from Reading, Pennsylvania and flew to Elmira, Syracuse, Rochester, Niagara

Falls, Allegheny County, Manassas, and Baltimore where samples were trucked to the lab a few

miles from BWI airport.

The aircraft then returned to Reading, Pennsylvania by 3:30 am.

On that night, the majority of their aircraft focused their flights on the American south,

though.

This is a region without many Quest Diagnostics labs and a surge of cases, at that time.

Another PC-12, for example, flew from Lawrenceville, Georgia to Jacksonville, Gainesville, Tampa,

Orlando, Valdosta, Lawrenceville, Roanoke, Manassas, and Greensboro, before returning

to Lawrenceville, all in one night.

Filling in the rest of the routes, you can see just how much ground this small airline

covered in one night—stopping at a large number of cities and towns, going to just

a few labs.

Now, in theory, a system like this, bringing a whole countries worth of samples to a few

central sites, should provide economies and efficiencies of scale, and, to an extent,

it does, however, other economic concepts have simultaneously decreased its effectiveness.

To start with, there's a heavy supply and demand imbalance for crucial equipment.

To perform a test, about twenty different consumable supplies are needed including nasal

swabs, transport tubes, reagents, and other equipment.

Without any of these twenty items, a test cannot be performed, so a bottleneck in any

one of those twenty supply chains stops testing.

In addition, economic forces are actually decreasing overall efficiency.

Particularly in a country like the US, where testing is overwhelmingly conducted by private

organizations rather than the government, thousands of providers are each bidding against

each other for equipment.

In conventional economics, one typically wants to avoid monopolies and oligopolies because

a few number of players have too much power.

This is why, in countries like the UK, for example, where 75% of the grocery market is

cornered by four firms, products like milk are sold for just about what it costs to produce—farmers

barely turn a profit.

That's because, if a farmer can't sell their milk to Tesco, Sainsbury's, Asda,

or Morrisons, they don't really have another option, so they'll take any price they can

get and meanwhile, there are plenty of farmers, so the grocery stores have plenty of options

if one says no to their low price.

Meanwhile, in the Covid test supply market, the producers have thousands upon thousands

of potential customers, the testing providers, so if one customer won't accept a high price,

there are plenty of others and one that most desperately needs the supplies inevitably

will.

If, for example, the federal government was the only customer in the US, or just the largest

one by far, the supply producers would just have to accept what the government paid because

there would be no alternatives.

Now, resources might not be allocated as efficiently using a centrally-planned system, but it would

at least come at a lower cost.

In the real world, the larger providers like LabCorp and Quest Diagnostics do have higher

buying power, processing more tests per day than most countries, but they still have to

compete against the rest of the world in procuring supplies, which increases cost.

In recent weeks, Quest Diagnostics, and other major testing providers, have been able to

collect far more samples than they have been able to process, and tests now take up to

two weeks, or sometimes more, to come back.

Tests that take so long to return are of little use since many asymptomatic or mildly symptomatic

patients might not quarantine for that full period while waiting for results.

Therefore, major focus has been placed on how to increase the efficiency of testing.

For this, Quest Diagnostics received the first emergency use authorization in the US from

the FDA to implement a technique known as, “pooled testing.”

With this, up to four samples will be combined and tested together.

If they generate a negative result, then all four tests are considered negative.

If the group tests positive, then they are separated and tested individually.

Now, if 100 people are tested traditionally, 100 tests will have to be processed.

But let's say that 100 people are tested using pooled testing.

In the US, on average, about 8% of tests currently come back positive, so with 100 tests, grouped

into 25 pools of four, you would expect about 32% of those to come back positive.

For those positive pools, you'd then process each test individually, so you will have processed

five times for four results.

For the 17 pools that came back negative, you only processed once, meaning that altogether,

this pooled testing technique used 57 processing slots to generate 100 results—effectively

almost doubling capacity.

This pooled testing technique, though, only works if testing is widespread enough that

only a small portion of results come back positive.

If an average of 20% of results came back positive, as was the case in the US in early-April,

the efficiency would diminish quickly.

In that case, you'd expect to have to re-test 20 of the 25 pools meaning that, overall,

you'd use 105 processing slots to generate 100 results—more than if you just tested

them all individually.

Therefore, this technique does not work as well for testing in hotspots, where positivity

rates are far higher.

Conversely, though, if positivity rates are quite low, the efficiency of pooled testing

can increase as one could efficiently group 8 or 12 or 16 or more tests together.

Lower positivity rates can be achieved through collecting more samples, so pooled sampling

can increase testing capacity, therefore increasing its own efficiency.

But no matter how fast PCR testing itself is, it is always a lagging indicator.

That is because it takes a number of days for people to develop symptoms after exposure,

then more days for them to decide to get tested, then more days for the test to come back.

That means that, realistically, it will take weeks for significant upticks in transmission

to show in the data.

Therefore, some focus has been placed on developing a more real-time indicator for COVID transmission.

The solution for that is sewage.

Sewage is naturally aggregated into a few processing plants, so by testing it for the

presence of COVID's genetic signature, one can gain a more accurate picture of how a

whole area is doing, without the potential sampling biases of clinical testing.

This technique can show if an area has coronavirus transmission, but it can also give a big-picture

view of how much.

Not only that, but in research tests, sewage testing indicates spikes about a week before

clinical testing, meaning it can help tell governments if they should increase restrictions

before it's potentially too late to make an impact.

But let's go back to the pooled testing.

Using sewage testing, you can know the relative chance of a single person from a given area

of having Coronavirus, so imagine if you could combine the sewage testing and clinical pooled

testing into one cohesive system.

This technique was proposed by a group of researchers from Berkley.

If you know that, say, there was a 25% chance of a test from Florida coming back positive,

you would know that it would be more efficient to test that on its own, however, if you knew

that there was a 1% chance that a test from North Dakota came back positive, you could

pool that test that with a large group of other low-risk tests and save processing slots.

You could also, for example, know that a sample from a healthcare or retail worker would have

a much higher chance of coming back positive than a test of an businessperson working from

home.

The Berkley research proposes a system of machine learning to scale pooled testing sizes

based on risk factors.

Low-risk samples would be grouped into large pools, and high-risk samples would be grouped

into smaller pools or tested alone so that pooled testing never leads to an efficiency

disadvantage, no longer the positivity rate.

The true genius of pooled testing is that efficiency increases as test numbers increase,

so it only takes twice the resources to test a population once daily as it does to test

once monthly.

Hypothetically, a pooled testing system designed by machine learning could lower the cost of

a test to just $3 to $5, meaning the frequency of testing could increase dramatically.

Now, we already know when the crunch time will be for testing—Fall, 2020.

Sometime between October and December, each year, there begins a massive increase in the

number of people catching the flu, which is highly seasonal.

Covid-19's symptoms closely mirror those of the flu, therefore there will be a massive

increase in people exhibiting Covid symptoms.

Those people will want to get tested, to rule out the possibility of Coronavirus, which

means demand will be higher than ever for testing, even if real infection numbers are

lower.

All the supplies for testing produced now are being used now, there is no stash being

built up, which means testing providers and healthcare systems worldwide have a hefty

task at hand if there is to be any hope of testing capacity matching demand in the coming

winter.

Like so many things going on right now, Covid testing logistics and its issues are strongly

rooted in economic theory, and there's plenty more I could talk about, but this video was

getting a little too long to still appease the YouTube algorithm, so I've posted an

extended cut of this video on Nebula including discussion of exactly that.

Nebula is, of course, the streaming site that myself and a bunch of other of the internet's

top educational creators teamed up to create.

Instead of this ad right now, you could be watching the extended cut of this video, or

any of our other videos early and ad-free, or some of our special Nebula exclusive projects.

In fact, there's a brand-new Nebula original going out tomorrow from our other channel,

Half as Interesting, or you could watch Tom Scott's gameshow, Money, which I'm in.

We're also super excited that Nebula is teamed up with CuriosityStream, which is another

of the internet's best streaming sites—home to thousands of fantastic documentaries and

non-fiction shows, like their one on Gander International Airport.

Specifically, on how it and the town reinvented itself as it shifted from being one of the