8 ms·

Coronavirus: South Korea’s aggressive testing gives clues to true fatality rate

61 points by ryannevius 7y ago

nraynaud 7y ago
I was watching a conference, they were saying the cruise ship was the best data, because everybody got exposed, tested and had a follow-up.
- arno_app 7y ago
  The cruise ship has a massive age bias. Most passengers were 60+
  - nraynaud 7y ago
    that's the difference between "best" and "perfect"
  - koheripbal 7y ago
    Seems like an easy variable to adjust for.
  - dogma1138 7y ago
    The cruise ship also didn’t had proper medical facilities. This is somewhat the problem with these epidemics the mortality rate is highly dependent on the available level of care. If the healthcare services get overwhelmed the mortality rate can easily spike.
- samvher 7y ago
  They were old on average though and all got relatively good treatment (not too many overcrowded hospitals). Something that data makes very clear is that there are many asymptomatic cases though - around 50% of the people who tested positive did not experience any symptoms. Which suggests that areas where only people exhibiting symptoms get tested are probably missing a large proportion of carriers.
mytailorisrich 7y ago
This corroborates Chinese numbers. China has also been very aggressively testing and mortality outside Wuhan is no more than 1% (according to Dr Aylward, following the WHO fact finding mission to China). This implies very serious efforts, though, which I'm not sure we're seeing in the West.
- trenchgun 7y ago
  Yep. Early detection makes for better outcomes and smaller fatality rate.
  - mirekrusin 7y ago
    How come, I thought there was no cure?
    - nobrains 7y ago
      Good question. Since once someone is infected, how can early detection and being under quarantine help in recovery? The answer is that with early detection, and keeping the virus numbers low, the health care network can be kept free for caring for the cases that do come through. And ICU level care cannot cure the virus but can reduce fatalities.
      - mirekrusin 7y ago
        Early detection means up to 100x more people overloading hospitals, doesn't it? It's much better to just stop as much as possible for 2 weeks - work from home etc.
    - d6e 7y ago
      Early detection does two things: 1. Allows you to isolate the person before they spread it further. 2. Allows you to be ready to treat their pneumonia which increases their survival chances.
      - 7y ago
DataDrivenMD 7y ago
I caution against reading too much into the South Korean fatality rates because many of the patients are younger and generally healthy (http://www.koreaherald.com/view.php?ud=20200303000714 http://www.koreaherald.com/view.php?ud=20200303000714). Whatever the global mortality rate shakes out to be in the long-run, it’s impossible to ignore the fact that this virus killed nearly a dozen nursing home residents in the span of 2 weeks. There’s a lot we don’t understand about COVID-19, certainly more than a single metric can tell us. Based on what we’re seeing in the state of Washington and around the world, it’s safe to say that older patients are at higher risk for developing respiratory failure. Beyond that, it’s simply way too early to draw conclusions.
- xbmcuser 7y ago
  Rumors of a few young people discharged after they have recovered and testing negative have later died with the virus coming back. I agree we have too many stories with even stories of 2 different strains one being deadlier than the other and people being infected by both at the same time. Actual mortality rate we will find out after 1-2 years only even that only if they have a test where can check if a healthy person ever contracted the infection in the past then test a few thousand people randomly. Btw China numbers show that people either survived or died in the third or fourth week after their symptoms got worse in 14-20 days after infection. So Korean mortality rate might even go up as it is still too early to tell .
  - KarlKemp 7y ago
    Don’t spread “rumors”.
- D_Alex 7y ago
  The article's 0.6 mortality rate seems to have been obtained by dividing the number of confirmed infections (6284) by the number of deaths so far (40) - source: https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6 https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594... This is completely misleading, unless the other 6244 people have already recovered. The low death rate so far is because the detected illnesses have not yet run their course. Edit: The source above says that the number of people in South Korea who have recovered from the illness is... 135.
mirekrusin 7y ago
Uh, just looked at map and Pyongyang/North Korea is holding great on stats - zero infections. Does anyone know if it's becase a) almost nobody travels there, or b) there's no outgoing information or both?
- manquer 7y ago
  It is also c) they may not have testing kits or may not be using what WHO sent to various countries.
- pi-rat 7y ago
  I'm guessing a combination of information blackout, and lack of testing equipment. Where are the world's PCR machines located?
- MarkSanghee 7y ago
  I saw a couple of articles written in Korean - those stated, in North Korea, there are only limited people, in a novel class, are able to travel outside the country. Also, unbelievable point was, North Korean government executed one inspected person immediately after finding out.
  - solveit 7y ago
    Two people were executed. One for smuggling(The regime was figuring out how he had contracted the virus after NK barred Chinese people from entering the country and they found out that he was meeting Chinese smugglers), and the other for breaking quarantine and going to a public bath.
- n0rbwah 7y ago
  I've heard that their border with China is pretty porous and that people cross it on a daily basis to get contraband into the country. So I'd say b) is most likely. People are criticizing China's release of information but compared to NK, they're pretty transparent. Even if the situation on the ground was disastrous, I'd bet that the government would do everything to hide it.
- 7y ago
patrec 7y ago
Does anyone know of any good, publicly accessible case data (for any country)? E.g. for each case, date of detection, age of patient, ongoing|recovery date|date of death
- DataDrivenMD 7y ago
  The best and closest to what you’re looking for comes from Johns Hopkins: - https://data.humdata.org/dataset/novel-coronavirus-2019-ncov-cases https://data.humdata.org/dataset/novel-coronavirus-2019-ncov... The GitHub repo with datasets: - https://github.com/CSSEGISandData/COVID-19 https://github.com/CSSEGISandData/COVID-19 Start there, and reach out to the staff if you’re looking for something more specific. They may be able to point you in the right direction.
  - patrec 7y ago
    Yeah, I'm aware of the JHU dataset. Unfortunately working out useful CFRs from that is not easy. You need to be able to figure out infected_at_t_and_dead/(infected_at_t_and_recovered+infected_at_t_and_dead) for some time t sufficiently in the past so that basically everyone falls into these two categories. And if you want to draw transferable conclusions you also need an interval where treatment conditions were relatively uniform (e.g. already completely overwhelmed health system, or health system that could cope, throughout). As well as an age breakdown of recovered and dead. The data from Italy does not look promising, they have an old population but seem to have reacted well and still have capacity to treat people. And yet CFR seems 3.8%, without excluding people who are ongoing and will die in the next days and weeks. It would be very interesting to have a more detailed comparison with the South Korean outbreak, where again reaction has been very good as well and we could look at a subset of people for whom we have no undetected cases, so very good estimates ought to be possible.
- Kliment 7y ago
  There is some good data on 1100 early cases in China. See https://www.nejm.org/doi/full/10.1056/NEJMoa2002032 https://www.nejm.org/doi/full/10.1056/NEJMoa2002032
scarmig 7y ago
My theory on the coronavirus. I've managed to convince myself to have moderate levels of confidence in it; please tear it apart. 1) The Diamond Princess is our best source of data. Every person who got infected on it was identified (~700), and those who were infected got the best medical care possible. This led to a ~1% mortality rate (6/700) and a ~5% ventilator support rate (35/700). 6/700 is subject to significant error in both directions, but let's go with it. 2) Their average age was in the upper 40s, so assuming younger people don't suffer the most serious ill effects from it, a guess for the general population might be a 0.5% base mortality rate and 2.5% ventilator support. This is a rough order of magnitude estimate: obviously expected mortality should be adjusted for actual age distributions. 3) South Korea's mortality rate is 0.6%. Cases outside of Wuhan but within China have a reported mortality rate of 0.4%. Singapore (at ~100 cases) is 0%. These are all consistent with the estimated mortality rate I extrapolated from the DP. 4) All of the areas mentioned in 3) have very aggressive testing protocols. No evidence has come up suggesting widespread infection among the broader community. This includes China randomly sampling some areas and not finding meaningful unknown pockets of infection. 5) But what about Wuhan? It has a much higher mortality rate (>3%). The reports on the ground suggested that it was the medical system being overwhelmed that drove the mortality through the roof. If we assume everyone who needs a ventilator but doesn't get one dies, that explains in one fell swoop Wuhan's elevated mortality rate. 6) What about Italy? Its mortality rate is ~3.5%. But, compare the number of hospital beds in Italy and SK. It's ~3/1k vs ~10/1k. SK is already running into issues with hospital bed availability, so it stands to reason that the situation is much worse in Italy. Italy also has a significantly older population than SK. So it's starting to run into the same nightmare scenario as Wuhan. Of note, the US has a younger population than Italy, but fewer hospital beds. 7) What about Iran? It has ridiculous mortality rates. My best explanation is that testing constraints probably lead to more undercounting of infections in the hardest hit areas, so Hubei, Italy, and Iran probably all have artifically elevated mortalities, while the reality for all of them is much closer to 2.5%. 8) There's no need for weird hypotheses about Asians having different proteins in their lungs, there being multiple strains that act radically differently from each other, China running secret death camps for the infected, etc. 9) Policy-wise, we need to keep the rate of infection low enough such that the medical system is not overloaded. This is possible, as it's been done in multiple countries, and that's the difference between a severe-but-normal flu season and over a million dead in the USA. We also have to start taking decisive action yesterday, with today being a second best option.
Grue3 7y ago
South Korea already has huge experience with SARS and MERS. Most other countries don't, so "true fatality rate" would be higher. The demographics(age distribution) is also different in other countries.
- archseer 7y ago
  The response to MERS was a total shitshow (no laws against breaking quarantine, quarantined patients playing golf, flying to China, etc.) and we were lucky the virus didn't spread too easily. But they seem to have learned from that and are handling things really well this time: https://edition.cnn.com/2020/03/02/asia/coronavirus-drive-through-south-korea-hnk-intl/index.html https://edition.cnn.com/2020/03/02/asia/coronavirus-drive-th...
krzyk 7y ago
Isn't it that there are two main strains, one (30% of cases) is less lethal (and this one is predominant in South Korea) and the other (70% cases) is e.g. in Italy? Two strains source: https://www.newscientist.com/article/2236544-coronavirus-are-there-two-strains-and-is-one-more-deadly/ https://www.newscientist.com/article/2236544-coronavirus-are...
- mlang23 7y ago
  What speaks against this theory is that in austria, which is north of italy, the cases so far have been pretty mild.
  - koheripbal 7y ago
    Those infections are still recent. The virus takes weeks to kill you.
- imustbeevil 7y ago
  Based on the top response here: https://news.ycombinator.com/item?id=22499569 https://news.ycombinator.com/item?id=22499569 There are more than 100 strains, which you can see here: https://nextstrain.org/ncov https://nextstrain.org/ncov
- ashildr 7y ago
  The two strains paper seems to be hot air: http://virological.org/t/response-to-on-the-origin-and-continuing-evolution-of-sars-cov-2/418 http://virological.org/t/response-to-on-the-origin-and-conti...

61 points by ryannevius 7y ago

I was watching a conference, they were saying the cruise ship was the best data, because everybody got exposed, tested and had a follow-up.

arno_app 7y ago
The cruise ship has a massive age bias. Most passengers were 60+
- nraynaud 7y ago
  that's the difference between "best" and "perfect"
- koheripbal 7y ago
  Seems like an easy variable to adjust for.
- dogma1138 7y ago
  The cruise ship also didn’t had proper medical facilities. This is somewhat the problem with these epidemics the mortality rate is highly dependent on the available level of care. If the healthcare services get overwhelmed the mortality rate can easily spike.
samvher 7y ago
They were old on average though and all got relatively good treatment (not too many overcrowded hospitals). Something that data makes very clear is that there are many asymptomatic cases though - around 50% of the people who tested positive did not experience any symptoms. Which suggests that areas where only people exhibiting symptoms get tested are probably missing a large proportion of carriers.

mytailorisrich 7y ago

This corroborates Chinese numbers. China has also been very aggressively testing and mortality outside Wuhan is no more than 1% (according to Dr Aylward, following the WHO fact finding mission to China). This implies very serious efforts, though, which I'm not sure we're seeing in the West.

trenchgun 7y ago
Yep. Early detection makes for better outcomes and smaller fatality rate.
- mirekrusin 7y ago
  How come, I thought there was no cure?
  - nobrains 7y ago
    Good question. Since once someone is infected, how can early detection and being under quarantine help in recovery? The answer is that with early detection, and keeping the virus numbers low, the health care network can be kept free for caring for the cases that do come through. And ICU level care cannot cure the virus but can reduce fatalities.
    - mirekrusin 7y ago
      Early detection means up to 100x more people overloading hospitals, doesn't it? It's much better to just stop as much as possible for 2 weeks - work from home etc.
  - d6e 7y ago
    Early detection does two things: 1. Allows you to isolate the person before they spread it further. 2. Allows you to be ready to treat their pneumonia which increases their survival chances.
    - 7y ago

DataDrivenMD 7y ago

I caution against reading too much into the South Korean fatality rates because many of the patients are younger and generally healthy (http://www.koreaherald.com/view.php?ud=20200303000714 http://www.koreaherald.com/view.php?ud=20200303000714). Whatever the global mortality rate shakes out to be in the long-run, it’s impossible to ignore the fact that this virus killed nearly a dozen nursing home residents in the span of 2 weeks. There’s a lot we don’t understand about COVID-19, certainly more than a single metric can tell us. Based on what we’re seeing in the state of Washington and around the world, it’s safe to say that older patients are at higher risk for developing respiratory failure. Beyond that, it’s simply way too early to draw conclusions.

xbmcuser 7y ago
Rumors of a few young people discharged after they have recovered and testing negative have later died with the virus coming back. I agree we have too many stories with even stories of 2 different strains one being deadlier than the other and people being infected by both at the same time. Actual mortality rate we will find out after 1-2 years only even that only if they have a test where can check if a healthy person ever contracted the infection in the past then test a few thousand people randomly. Btw China numbers show that people either survived or died in the third or fourth week after their symptoms got worse in 14-20 days after infection. So Korean mortality rate might even go up as it is still too early to tell .
- KarlKemp 7y ago
  Don’t spread “rumors”.
D_Alex 7y ago
The article's 0.6 mortality rate seems to have been obtained by dividing the number of confirmed infections (6284) by the number of deaths so far (40) - source: https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6 https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594... This is completely misleading, unless the other 6244 people have already recovered. The low death rate so far is because the detected illnesses have not yet run their course. Edit: The source above says that the number of people in South Korea who have recovered from the illness is... 135.

mirekrusin 7y ago

Uh, just looked at map and Pyongyang/North Korea is holding great on stats - zero infections. Does anyone know if it's becase a) almost nobody travels there, or b) there's no outgoing information or both?

manquer 7y ago
It is also c) they may not have testing kits or may not be using what WHO sent to various countries.
pi-rat 7y ago
I'm guessing a combination of information blackout, and lack of testing equipment. Where are the world's PCR machines located?
MarkSanghee 7y ago
I saw a couple of articles written in Korean - those stated, in North Korea, there are only limited people, in a novel class, are able to travel outside the country. Also, unbelievable point was, North Korean government executed one inspected person immediately after finding out.
- solveit 7y ago
  Two people were executed. One for smuggling(The regime was figuring out how he had contracted the virus after NK barred Chinese people from entering the country and they found out that he was meeting Chinese smugglers), and the other for breaking quarantine and going to a public bath.
n0rbwah 7y ago
I've heard that their border with China is pretty porous and that people cross it on a daily basis to get contraband into the country. So I'd say b) is most likely. People are criticizing China's release of information but compared to NK, they're pretty transparent. Even if the situation on the ground was disastrous, I'd bet that the government would do everything to hide it.
7y ago

patrec 7y ago

Does anyone know of any good, publicly accessible case data (for any country)? E.g. for each case, date of detection, age of patient, ongoing|recovery date|date of death

DataDrivenMD 7y ago
The best and closest to what you’re looking for comes from Johns Hopkins: - https://data.humdata.org/dataset/novel-coronavirus-2019-ncov-cases https://data.humdata.org/dataset/novel-coronavirus-2019-ncov... The GitHub repo with datasets: - https://github.com/CSSEGISandData/COVID-19 https://github.com/CSSEGISandData/COVID-19 Start there, and reach out to the staff if you’re looking for something more specific. They may be able to point you in the right direction.
- patrec 7y ago
  Yeah, I'm aware of the JHU dataset. Unfortunately working out useful CFRs from that is not easy. You need to be able to figure out infected_at_t_and_dead/(infected_at_t_and_recovered+infected_at_t_and_dead) for some time t sufficiently in the past so that basically everyone falls into these two categories. And if you want to draw transferable conclusions you also need an interval where treatment conditions were relatively uniform (e.g. already completely overwhelmed health system, or health system that could cope, throughout). As well as an age breakdown of recovered and dead. The data from Italy does not look promising, they have an old population but seem to have reacted well and still have capacity to treat people. And yet CFR seems 3.8%, without excluding people who are ongoing and will die in the next days and weeks. It would be very interesting to have a more detailed comparison with the South Korean outbreak, where again reaction has been very good as well and we could look at a subset of people for whom we have no undetected cases, so very good estimates ought to be possible.
Kliment 7y ago
There is some good data on 1100 early cases in China. See https://www.nejm.org/doi/full/10.1056/NEJMoa2002032 https://www.nejm.org/doi/full/10.1056/NEJMoa2002032

scarmig 7y ago

My theory on the coronavirus. I've managed to convince myself to have moderate levels of confidence in it; please tear it apart. 1) The Diamond Princess is our best source of data. Every person who got infected on it was identified (~700), and those who were infected got the best medical care possible. This led to a ~1% mortality rate (6/700) and a ~5% ventilator support rate (35/700). 6/700 is subject to significant error in both directions, but let's go with it. 2) Their average age was in the upper 40s, so assuming younger people don't suffer the most serious ill effects from it, a guess for the general population might be a 0.5% base mortality rate and 2.5% ventilator support. This is a rough order of magnitude estimate: obviously expected mortality should be adjusted for actual age distributions. 3) South Korea's mortality rate is 0.6%. Cases outside of Wuhan but within China have a reported mortality rate of 0.4%. Singapore (at ~100 cases) is 0%. These are all consistent with the estimated mortality rate I extrapolated from the DP. 4) All of the areas mentioned in 3) have very aggressive testing protocols. No evidence has come up suggesting widespread infection among the broader community. This includes China randomly sampling some areas and not finding meaningful unknown pockets of infection. 5) But what about Wuhan? It has a much higher mortality rate (>3%). The reports on the ground suggested that it was the medical system being overwhelmed that drove the mortality through the roof. If we assume everyone who needs a ventilator but doesn't get one dies, that explains in one fell swoop Wuhan's elevated mortality rate. 6) What about Italy? Its mortality rate is ~3.5%. But, compare the number of hospital beds in Italy and SK. It's ~3/1k vs ~10/1k. SK is already running into issues with hospital bed availability, so it stands to reason that the situation is much worse in Italy. Italy also has a significantly older population than SK. So it's starting to run into the same nightmare scenario as Wuhan. Of note, the US has a younger population than Italy, but fewer hospital beds. 7) What about Iran? It has ridiculous mortality rates. My best explanation is that testing constraints probably lead to more undercounting of infections in the hardest hit areas, so Hubei, Italy, and Iran probably all have artifically elevated mortalities, while the reality for all of them is much closer to 2.5%. 8) There's no need for weird hypotheses about Asians having different proteins in their lungs, there being multiple strains that act radically differently from each other, China running secret death camps for the infected, etc. 9) Policy-wise, we need to keep the rate of infection low enough such that the medical system is not overloaded. This is possible, as it's been done in multiple countries, and that's the difference between a severe-but-normal flu season and over a million dead in the USA. We also have to start taking decisive action yesterday, with today being a second best option.

Grue3 7y ago

South Korea already has huge experience with SARS and MERS. Most other countries don't, so "true fatality rate" would be higher. The demographics(age distribution) is also different in other countries.

archseer 7y ago
The response to MERS was a total shitshow (no laws against breaking quarantine, quarantined patients playing golf, flying to China, etc.) and we were lucky the virus didn't spread too easily. But they seem to have learned from that and are handling things really well this time: https://edition.cnn.com/2020/03/02/asia/coronavirus-drive-through-south-korea-hnk-intl/index.html https://edition.cnn.com/2020/03/02/asia/coronavirus-drive-th...

krzyk 7y ago

Isn't it that there are two main strains, one (30% of cases) is less lethal (and this one is predominant in South Korea) and the other (70% cases) is e.g. in Italy? Two strains source: https://www.newscientist.com/article/2236544-coronavirus-are-there-two-strains-and-is-one-more-deadly/ https://www.newscientist.com/article/2236544-coronavirus-are...

mlang23 7y ago
What speaks against this theory is that in austria, which is north of italy, the cases so far have been pretty mild.
- koheripbal 7y ago
  Those infections are still recent. The virus takes weeks to kill you.
imustbeevil 7y ago
Based on the top response here: https://news.ycombinator.com/item?id=22499569 https://news.ycombinator.com/item?id=22499569 There are more than 100 strains, which you can see here: https://nextstrain.org/ncov https://nextstrain.org/ncov
ashildr 7y ago
The two strains paper seems to be hot air: http://virological.org/t/response-to-on-the-origin-and-continuing-evolution-of-sars-cov-2/418 http://virological.org/t/response-to-on-the-origin-and-conti...