Understanding the COVID-19 data
March 21, 2020 1:16 PM   Subscribe

Why do countries some countries appear to have a much higher death per total case rate from COVID-19 than others? The data confuses me.

I'm looking at the novel coronavirus data compiled on Worldometers, which, as far as I can tell, is accurate. I'm confused as to why many (most?) countries that report total cases in the tens of thousands seem to have a death rate in the thousands, whereas a few have a much lower apparent death rate.

If I sort the data on the Worldometers page, there are six countries with a reported total case number above 20,000: China, Italy, Spain, USA, Germany, and Iran. Of those, China, Italy, Spain, and Iran are reporting deaths in the thousands. By contrast, as of this writing, the US is reporting 288 deaths and Germany is reporting 83.

I'm nothing close to an epidemiologist, and the wide variance in this data confuses me. Is it due to a difference it population demographics? Response? Overwhelmed hospitals? Availability of tests? Is it simply that countries like the US and Germany are earlier in the timeline, and we can expect the death rate to rise to the levels we see in other countries with tens of thousands of cases?

I do not want to be gruesome or grim with this question--I'm just trying to understand what's going on, and how to read the data. I don't understand how, for example, the US, Germany, and Spain can have a nearly identical number of reported cases, but such a wide difference in number of deaths. Is this a hopeful sign for the countries with fewer deaths, or do the countries suffering more right now indicate where the former group is headed?
posted by scarylarry to Science & Nature (16 answers total) 2 users marked this as a favorite
You’ve already named many of the factors but I think a very large one is that key term “reported cases.” The numbers will change wildly until you actually broadly test the public.
posted by raccoon409 at 1:21 PM on March 21 [7 favorites]

Death rate is being calculated as the percentage of people who test positive and then go on to die (or are determined to have died as a result of COVID-19 after death).

Rates of testing make a huge difference. If you're only testing people who arrive at hospitals in respiratory distress, your percentage will be much higher than if you're testing more widely.
posted by pipeski at 1:22 PM on March 21 [9 favorites]

In the UK the testing is abysmally low so the mortality rate looks really high.cant speak for other countries.
posted by BAKERSFIELD! at 1:22 PM on March 21 [4 favorites]

It’s not only the testing rates (which is probably the biggest factor), but also the fact that we don’t know the outcome of all the cases yet... in many countries, there are far more people who are currently ill than people who have died or recovered. Some of those currently ill people will die, some will recover, but we won’t have a true percentage until those cases resolve.

Also demographic differences in populations are not to be discounted. For example, Italy’s population skews older than many other countries’, while most of the people who got sick in South Korea were younger.
posted by eirin at 1:35 PM on March 21 [9 favorites]

It's a testing issue. There's no way that the number of confirmed cases resembles the number of actual people that have the virus.

The virus spreads quickly and in many cases with minor symptoms, and most places are still testing only people who have severe symptoms or have had contact with those who have had severe symptoms. The actual virus is spreading at least one or two steps beyond that.

Only something like 1/5 of the cases end up with severe symptoms. The average case spreads it to between 2 and 3 other people. It's not unreasonable to guess that the number of confirmed cases is undercounting by at least a factor of 10 unless there is population wide testing and surveillance in place. And that's not even counting the fact that there's a few days lead time between testing and reporting.

Without the capability to test a statistically significant random sample of the populace, it's really hard to accurately estimate how many actual infections there are. The best even the professionals have is slightly better educated guesses.

But in some sense, the exact numbers don't matter. The actions that need to be taken are the same - as much social isolation as possible. Everyone needs to act like they're carrying the virus because by the time you know that you have it, you've already spread it to more people.
posted by Zalzidrax at 1:45 PM on March 21 [5 favorites]

Some countries are not performing post-mortem testing of people who die without having been tested first.
posted by EndsOfInvention at 1:47 PM on March 21 [6 favorites]

It's almost all in the rate of testing. The reported number of cases could be orders of magnitude off from the actual number of infected, if only the seriously ill are tested. That said, there will be other smaller variables; the demographics, behavioural factors such as rate of smoking, and also how well the hospitals are able to cope, such as ventilator shortages.

I've actually been watching the percentage number of deaths vs. recovered creep up the past few weeks, from about 6% to 12%, and I think it's because of spread in countries who are only testing the seriously ill like the UK and the US.
posted by stillnocturnal at 1:49 PM on March 21 [3 favorites]

I don't want to derail/hijack, but a related item.
I've checked several places, only in the USA is the number of deaths higher than the number of recovered. Like, lots higher. Why is this?
posted by rudd135 at 2:25 PM on March 21

I found this article (titled "Deciphering the pandemic: A guide to understanding the coronavirus numbers") to be a good explainer of the basic considerations involved in coming up with these numbers.
posted by forza at 2:41 PM on March 21 [2 favorites]

only in the USA is the number of deaths higher than the number of recovered. Like, lots higher. Why is this?

Educated guess: Americans tend not to actually seek out medical care in cases of mild sickness (because of our lack of public health care, which discourages seeking care below a certain threshold of discomfort/seriousness) and as a result low-intensity cases are recovering at home and not included in recovery statistics (which are presumably compiled from among those actually receiving medical care).
posted by jackbishop at 2:49 PM on March 21 [9 favorites]

Anecdotally it's lack of testing, as someone coughing to the point of throwing up, with a fever that comes & goes & trouble breathing, I've basically been told to go home take steroids for five days & a cough suppressant for five days & treat it like I have bronchitis, because even if the test came back positive, that's how they'd want me to treat it. If my lips or finger tips go blue, I'm to go back in & then I'll get a test, otherwise I'm classed as low risk so not worth testing. I understand why, the doctors/nurses where great and obviously frustrated at not being able to just test people. I feel like we're never going to have an accurate picture of how many people actually get this.
posted by wwax at 2:54 PM on March 21 [11 favorites]

This article is now a bit out of date (as in it’s a week old), but it was looking at why Italy was so badly hit compared to South Korea. There’s also a partial explanation at the bottom as to why the German death rate was lower at that point - the proportion of people who had the virus over 60 was only about 15%.
posted by scorbet at 2:55 PM on March 21 [3 favorites]

rudd135: I've checked several places, only in the USA is the number of deaths higher than the number of recovered. Like, lots higher. Why is this?

It's still early days here. The people who haven't recovered yet are still sick.
posted by capricorn at 3:36 PM on March 21 [7 favorites]

I found this article to be very helpful in understanding the kind of statistical anomalies you are describing. FWIW.
posted by forthright at 5:54 PM on March 21 [3 favorites]

It's also interesting to look at the Diamond Princess to find the true death rate. Out of 2666 passengers, 700 contracted the disease and 8 died (so far). The sample is far from random, but probably better than the "we test people if they have bad symptoms" approach. Some researchers have looked closer at the data, but there seems to be a lot of uncertainty still.

Then there's the town Vò in Italy, which tested all of its 3300 people. A quick search hasn't turned up definite numbers of infections/deaths.
posted by milan-g at 6:44 AM on March 22 [4 favorites]

« Older Google forms quiz tracking   |   How to help my community affected by the... Newer »

You are not logged in, either login or create an account to post comments