3 ms·

This happened again today, with Washington Post publishing an article based on a study that was submitted to Nature on March 22, 2020. They used this old data (

This happened again today, with Washington Post publishing an article based on a study that was submitted to Nature on March 22, 2020. They used this old data (2 months ago is old in this case) to claim that the lockdown saved millions upon millions of lives. To be that disingenuous can't be accidental, I am afraid. And I am normally a fan of WaPo

mthoms 6y ago
The data is indeed old but China began its lockdown on January 24th and South Korea a month later. Those countries seem to be the central focus of the study with regards to evaluating the effectiveness of the shutdown(s). I'm only a layperson but it doesn't seem too far-fetched that they could extrapolate from how those countries "flattened the curve" to come up with a ballpark estimate of global fatalities averted. After all, other countries mostly followed a similar "curve flattening" pattern (as far as I'm aware). At the very least, I don't see anything inherently disingenuous. Side note: FWIW, the study itself says the authors last "updated" it on May 13th (not the data itself though it appears).
- salmon30salmon 6y ago
  IMO, it is pretty easy to see why an extrapolation is faulty. Let us look at two different reasons: 1. Density. China and South Korea are far more dense than most of the western world. Seoul has a density of 42,000 per square mile. New York City, the densest city in the US, has a density of 23,000 per square mile. The next top 10 most populous cities in the USA have an average density of 5,000 per square mile. Over half of South Koreans live in an urban environment, while less than 25% of Americans do. Density, and the number of interactions someone has in a day, is a major determining factor in the spread of any airborne virus. Density also has implications for transit use, which in NYC was shown to be a large source of transmission. 2. Demographics. Italy, the United States and Sweden all have very different demographics and cultures. In Italy, many people live in multi-generational housing, and nursing home usage is far lower than the United States. The average age of Italy is also higher. Sweden is more like the US in terms of nursing homes, but still lower. As the older folks are far more likely to die from this, or even have detectable infections, the way each country manages their aging population will have a huge impact on mortality. For example, people going about their day like normal, and then returning to their shared home with their Grandma would put Grandma at a much higher risk of infection. That is the Italy scenario. In the United States, where the old either live independently or in a group home, their vectors are much more limited. Test and trace everyone going to and working at a nursing home. It should also be noted that the areas if NYC most heavily hit are also more multi-generational than the rest of the city and country. So see, you can't take one nation and extrapolate it out to the world. That is a faulty model and those who championed it should know better.
  - mthoms 6y ago
    What you're saying makes sense, thank you. I'm wondering though, are you saying it because the study did indeed fail to account for all those factors or are you basing it my (probably poor) interpretation/description of it? :-) If you're going off my comments, I may have done a disservice as my laypersons understanding of the topic is pretty rudimentary. FWIW, the study does indeed model other countries like the US and Italy so I'd be surprised if much of what you speak of was not taken into account. I certainly don't think they modeled one nation and simply extrapolated that out into the world without other considerations. I apologise if my words gave that impression... Even I can see how stupid that would be. Here's the link if anyone is interested https://www.nature.com/articles/s41586-020-2404-8_reference.pdf https://www.nature.com/articles/s41586-020-2404-8_reference....
    - salmon30salmon 6y ago
      Hey, Not sure if you are comfortable with R or Python, but you can look at the code here: https://codeocean.com/capsule/1887579/tree/v1 https://codeocean.com/capsule/1887579/tree/v1 They do not seem to be taking cultural or demographic information into account. A lot of the data is sourced from the Johns Hopkins data set and fed into an SRI type model, with some data about when and how strict lockdowns where. It is actually pretty sloppy to be given so much weight.
      - mthoms 6y ago
        Cheers, I will indeed take a look. Thanks.

mthoms 6y ago

The data is indeed old but China began its lockdown on January 24th and South Korea a month later. Those countries seem to be the central focus of the study with regards to evaluating the effectiveness of the shutdown(s). I'm only a layperson but it doesn't seem too far-fetched that they could extrapolate from how those countries "flattened the curve" to come up with a ballpark estimate of global fatalities averted. After all, other countries mostly followed a similar "curve flattening" pattern (as far as I'm aware). At the very least, I don't see anything inherently disingenuous. Side note: FWIW, the study itself says the authors last "updated" it on May 13th (not the data itself though it appears).

salmon30salmon 6y ago
IMO, it is pretty easy to see why an extrapolation is faulty. Let us look at two different reasons: 1. Density. China and South Korea are far more dense than most of the western world. Seoul has a density of 42,000 per square mile. New York City, the densest city in the US, has a density of 23,000 per square mile. The next top 10 most populous cities in the USA have an average density of 5,000 per square mile. Over half of South Koreans live in an urban environment, while less than 25% of Americans do. Density, and the number of interactions someone has in a day, is a major determining factor in the spread of any airborne virus. Density also has implications for transit use, which in NYC was shown to be a large source of transmission. 2. Demographics. Italy, the United States and Sweden all have very different demographics and cultures. In Italy, many people live in multi-generational housing, and nursing home usage is far lower than the United States. The average age of Italy is also higher. Sweden is more like the US in terms of nursing homes, but still lower. As the older folks are far more likely to die from this, or even have detectable infections, the way each country manages their aging population will have a huge impact on mortality. For example, people going about their day like normal, and then returning to their shared home with their Grandma would put Grandma at a much higher risk of infection. That is the Italy scenario. In the United States, where the old either live independently or in a group home, their vectors are much more limited. Test and trace everyone going to and working at a nursing home. It should also be noted that the areas if NYC most heavily hit are also more multi-generational than the rest of the city and country. So see, you can't take one nation and extrapolate it out to the world. That is a faulty model and those who championed it should know better.
- mthoms 6y ago
  What you're saying makes sense, thank you. I'm wondering though, are you saying it because the study did indeed fail to account for all those factors or are you basing it my (probably poor) interpretation/description of it? :-) If you're going off my comments, I may have done a disservice as my laypersons understanding of the topic is pretty rudimentary. FWIW, the study does indeed model other countries like the US and Italy so I'd be surprised if much of what you speak of was not taken into account. I certainly don't think they modeled one nation and simply extrapolated that out into the world without other considerations. I apologise if my words gave that impression... Even I can see how stupid that would be. Here's the link if anyone is interested https://www.nature.com/articles/s41586-020-2404-8_reference.pdf https://www.nature.com/articles/s41586-020-2404-8_reference....
  - salmon30salmon 6y ago
    Hey, Not sure if you are comfortable with R or Python, but you can look at the code here: https://codeocean.com/capsule/1887579/tree/v1 https://codeocean.com/capsule/1887579/tree/v1 They do not seem to be taking cultural or demographic information into account. A lot of the data is sourced from the Johns Hopkins data set and fed into an SRI type model, with some data about when and how strict lockdowns where. It is actually pretty sloppy to be given so much weight.
    - mthoms 6y ago
      Cheers, I will indeed take a look. Thanks.