@March 20, 2020

A couple of days ago, a good friend of the team, Harean Hettiarachchi, asked for confirmation on how fast COVID-19 was popping up in Sri Lanka (i.e.: what does the curve look like?).

We knew we had been religiously downloading and backing up daily situation reports from the Ministry of Health Epidemiology Unit. We just hadn't had the time to properly get that data into a graph format. With the careful application of coffee and curry, we have now solved that problem. So, as of 10 AM today (20/03/2020):

This what our curve looks like. Let's dig into the story behind this chart.

In medical science, the index case or the first documented patient in a disease epidemic within a population.

Patient Zero was a Chinese national who had come to Sri Lanka as a tourist. According to the Epid unit report on the 28th of January, she arrived "2 weeks ago" - so that's possibly the [28-14] = 14th of January.

While the rumor mill went ham, the Ministry of Health's Epid Unit swung into action, putting people under observation. They also contributed to slowing fake news around COVID-19 and Patient Zero - some of our earliest factchecks on the subject, for instance, were directly from Dr. Sudath Samaraweera, the Chief Epidemiologist at the MoH.

At the start they had five other people under observation, 2 of them Sri Lankan. They started adding more while Patient Zero underwent treatment. Their reports show the first to be put under observation were people from the hospital that P-Zero had been at.

The chart above is slightly misleading. If you keep an eye on the data (ie: if you actually dig into the reports) you can see that those people in observation weren't always the same: you can see them picking up new numbers from various hospitals and patients possibly being shifted around for testing. Here's what our tracker looked like at the time:

They actually tested more people than their own daily reports show, because their daily reports only show the numbers of people being tested and not their identities. This is for ethical reasons, and this is why Grumpy Cat is unhappy: we can't argue with it.

Back to the story. We all know what happened. Patient Zero was discharged on 19/02/2020. Various political drums were banged. She got a nice selfie out of it.

To the public eye, very little happens for the next 22 days until bang! the second patient is discovered, and they're a Sri Lankan.

However, if you were keeping an eye on the Epid Unit's daily reports, you see them briefly de-escalating around P-Zero's release and then starting to test more people despite P-Zero being gone.

So despite, rather braindead posts on Facebook, and WhatsApp memes about how Sri Lankans cured the coronavirus, the experts suspected more damage, and kept at it. Grumpy Cat is larger and more worried.

We have to explain here that the third column shows '1' because we're tracking total cases here; that's the shadow of P-Zero in the data.

We suppose we don't have to explain why this portion of our chart is red. The first column is date. The second, number of people under observation in hospitals. The third column is number of confirmed infections as of 10.00 AM.

Grumpy Cat is gone because Grumpy Cat is social distancing.

And today's stats haven't shown up in the official reports yet (that'll show up at 10 AM tomorrow). So, if we add today, that's 71 total cases detected, 70 of them active (we have to subtract P-Zero, remember). So here's what the curve looks like right now:

This spike in confirmed cases has a lot to do with how symptoms manifest and how testing is done. Symptoms manifest 2-14 days after exposure; some people never show symptoms at all. Of those who do have symptoms, not all go get tested - it's easy to write this off as, say, a flu.

Here's what a least square regression model (rounded up to the nearest whole human) thinks will happen:

Note: image has been updated due to a coloring error in the previous iteration

In reality, while 'the curve' looks hella linear now, it has no reason to stick to linearity. We're hoping it does. We're hoping it curves downwards. But a few more idiots like this:

And our curve becomes a real curve. Upwards.

There's a common argument thrown around about the rate of doubling: some people point out that we're doubling faster than Italy, et cetera, et cetera. We're going to share this chart to show why this argument is a bit problematic:

This chart, from Oxford University's Our World in Data, shows the category of countries we're in right now. It's a bit out of date compared to what we've presented in this blog because:a) it's pulling from the ECDC, the European Center for Disease Controlb) we're faster than the ECDC for Sri Lanka, because we're herebut assuming that most countries are going to be similarly out of date (especially those like Afghanistan), you'll see that the doubling rate tells us nothing. We're in the same bracket as some insanely rich countries, some insanely poor countries, some countries that are mostly warzones, and even a country battling 17,000+ cases atm.

And China and South Korea, both countries that were hit hard, look favourable if you compare the doubling rate:

A better way of comparing would be to study the curve (#flattenthecurve, right?). Here's where the world is going, starting from the 100th case:

https://ourworldindata.org/grapher/covid-confirmed-cases-since-100th-case

Singapore and Kuwait seem to have things quite under control here. Bahrain and Japan seem roughly on the same track, too. Since we're pegged as doubling every 4 days in this data, here's the closest we could find to Sri Lanka: Australia. It's right between the 'doubling every 5 days' and 'doubling every 3 days' lines, and it looks like it's been that way for a while.

https://ourworldindata.org/grapher/covid-confirmed-cases-since-100th-case

But a word of caution here: the number of confirmed cases is lower than the number of total cases. Political crap aside, there's not enough tests to go around. In Hong Kong, they have enough testing material to test dogs. In the US of A, only the rich and the famous humans seem to have priority access right now. Here's a breakdown of tests per million people:

https://ourworldindata.org/grapher/covid19-tests-per-million-people

And here's a chart showing another facet of the data: total tests vs total cases.

https://ourworldindata.org/grapher/tests-vs-confirmed-cases-covid-19-per-million

You'll notice that the US, France, South Korea and Italy are scoring high to the top-right: that means it could be any of these things:a) more people are infected, so that chance of a test turning up a confirmed case are higherb) testing is delayed / happening for a smaller number of people - as in the US, where there's not enough tests to go aroundc) testing is targeted (effective); it's happening largely in impacted areas

Meanwhile, the UAE is doing a huge amount of testing - both as a percentage of its population and per confirmed case. It's effectively throwing tests around like dollar bills in the club.

We're not complaining. In Italy, a little town named Vo Euganeo underwent blanket testing. 3000 people were tested, 3% tested positive, were quickly isolated, and no new cases have shown up since. So, ironically, right now the only 100% trustworthy data is from a COVID-19 hotspot in Italy. A close second would be the UAE, possibly Canada and Belarus.

We frequently check out hpb.health.gov.lk. The Health Promotion Bureau runs a fantastic website for at-a-glance daily stats, although it doesn't show you historical data.

The second is, of course, this epid.gov.lk link that leads to their COVID-19 reports. They put these out as PDFs. We read these at 10 (ish) every day so you don't have to.

Of course, lastly, get the WatchDog app (and leave us a review, because some of us are in self-quarantine and have no contact with the outside world). The app gives you notifications whenever we confirm a piece of news or debunk a rumor. Drop by watchdog.team if you don't like apps.

Thanks for reading, wash your hands, and stay six feet away. Thank you!