4 ms·

Maybe. There are two ways this goes: 1) 3.4% of people who catch it die every year, and life changes 2) It mutates, and gets worse. Unlikely, but not impossib

Maybe. There are two ways this goes: 1) 3.4% of people who catch it die every year, and life changes 2) It mutates, and gets worse. Unlikely, but not impossible. Or it leads to long-term disability which we don't know about yet (not unlikely). Etc. 3) Kids get it early, acquire partial immunity as it passes, and it becomes another variant of the common cold/flu. Most of the time, #3 is the outcome. It's worth taking seriously because #1 and #2 will eventually happen if we ignore these, but....

anoncake 6y ago
> 1) 3.4% of people who catch it die every year, and life changes Why do you think the IFR will increase by a factor of almost 7?
- woofie11 6y ago
  Massachusetts, which has good test infrastructure, had 280k cases, and 11k deaths. That's 3.9%, excluding people who have yet to die. That was also WHO's first estimate (3.4%). I'm now seeing WHO is estimating 0.5-1.0%, which is possible too. I don't think we'll know for some time. One of the problems is people who have yet to die... You can't get a good estimate until long after a pandemic, to catch all the stragglers.
  - anoncake 6y ago
    Unless Massachusetts somehow manages to capture 100% of infections, that's the case fatality rate, which is worthless except as raw data to estimate the infection fatality rate. The 3.4% "estimate" also referred to the IFR as far as I know and if it didn't, it would also be worthless now anyway as it has been superseded.
    - woofie11 6y ago
      Yes. It is raw data. I believe Massachusetts comes as close to capturing 100% of infections as anyone at this point, although that wasn't always the case. There is a lot of test infrastructure. CFR can overestimate or underestimate IFR. In most past pandemics, CFR significantly UNDERestimated IFR during the pandemic, because the raw numbers include a lot of people who are sick, but haven't died YET. There is a tendency to take CFR, apply very liberal corrections, sometimes multiple times, and get artificially low IFR numbers. Those aren't really plausible, even looking at scenarios where we could catalog all cases (e.g. the cruises had an IFR just over 2%, with state-of-the-art medical). My general impression, looking at data, is that the IFR was around 3-ish percent early on, and got better, potentially due to mutations making the virus less fatal, lower initial viral loads due to precautions, and/or better treatment.
      - anoncake 6y ago
        > I believe Massachusetts comes as close to capturing 100% of infections as anyone at this point, although that wasn't always the case. So it does not capture 100%. > CFR can overestimate or underestimate IFR. Incorrect. CFR cannot be higher than IFR. The number of deaths per confirmed infection obviously isn't higher than the number of deaths per infection (confirmed or unconfirmed). I'll continue to believe epidemiologists who actually have a clue what they are doing over your "general impression".
        woofie11 6y ago
        The arrogance. If you want, an exert from an epidemiology text: "A CFR can only be considered final when all the cases have been resolved (either died or recovered). The preliminary CFR, for example, during an outbreak with a high daily increase and long resolution time would be substantially lower than the final CFR." The cited numbers are preliminary CFR, not final CFR. The key question is final IFR. Final IFR can be lower or higher than preliminary CFR. I'm done here.
        jsight 6y ago
        At least in the US, we are no longer in a period that qualifies as an "outbreak with a high daily increase". That implies a high increase relative to the current case count, which really isn't the case any more once you have a population case rate of 5%. I would be cautious about estimating countries that are only now having their first major outbreak (for example Czechia) for exactly the reason that you cite, but not every country is in that state. Put another way, we are no longer living with preliminary data in much of the world.
        woofie11 6y ago
        Let's toss some real numbers here. * Last death of Diamond Princess occurred after more than 2 months. * Over the past two months, the number of known cases in the US roughly doubled from 8 million to 16 million. This means there are a lot of people who haven't died yet, and dividing deaths by cases is an underestimate of CFR by up to a factor of 2. That doesn't mean we don't have the data. We could e.g. take the infections from June and look at how many of those resulted in death as a final CFR estimate. However, the number I gave (deaths over cases) didn't do that. It's a preliminary CFR without that sort of correction, and a correction to final IFR can go up (people who haven't died yet) or down (unidentified cases). anoncake argued this was an overestimate by definition, and anyone who said otherwise didn't have a clue. anoncake was wrong.
- ddingus 6y ago
  Current death rate is computed: Dead people / Recovered people and that is around 3 percent, down from 5 to 7 percent a few months ago. Infections / deaths is not accurate as cases in progress are added to those with definitive outcomes. That can also be seen as a best case, given nobody currently infected does not die. A worst case can be seen assuming everyone infected will die. But if we want to know the percentage of people who get sick and die, we have to compute that percentage with the people who got sick and did not die and their case ran its course to a definitive outcome. Infections lead to outcomes. Outcomes are where the dead vs not dead become known. Everyone infected will eventually arrive at an outcome.
  - anoncake 6y ago
    I have no idea what you are trying to say, sorry.
    - ddingus 6y ago
      It is simple: Divide dead people by recovered people to assess true risk of death from COVID. Dividing dead people by infected people represents a best case, overly optimistic risk due to the fact that cases that ran their full course to a definitive outcome are mixed together with cases in progress. Doing that biases the risk perception down below actual risk and can be dangerous. I know one person who died, who did not have to, and they died because they made choices after having evaluated risk incorrectly.
      - anoncake 6y ago
        I'm sure the epidemiologists who estimate the IFR at ~0.5% are aware of the fundamentals of their field.
        ddingus 6y ago
        As am I, but that does not make the risk assessment I put above wrong. They have their reasons for that metric. That too doesn't make (dead/recovered) wrong. There are dead people in my life, with whom I had this discussion. "Calculated risk" = optimistic. The IFR combines futures with actual and does not represent an outcome based risk assessment. Nature however, will definitely take you, or any of us, to an outcome, should we become infected.

anoncake 6y ago

> 1) 3.4% of people who catch it die every year, and life changes Why do you think the IFR will increase by a factor of almost 7?

woofie11 6y ago
Massachusetts, which has good test infrastructure, had 280k cases, and 11k deaths. That's 3.9%, excluding people who have yet to die. That was also WHO's first estimate (3.4%). I'm now seeing WHO is estimating 0.5-1.0%, which is possible too. I don't think we'll know for some time. One of the problems is people who have yet to die... You can't get a good estimate until long after a pandemic, to catch all the stragglers.
- anoncake 6y ago
  Unless Massachusetts somehow manages to capture 100% of infections, that's the case fatality rate, which is worthless except as raw data to estimate the infection fatality rate. The 3.4% "estimate" also referred to the IFR as far as I know and if it didn't, it would also be worthless now anyway as it has been superseded.
  - woofie11 6y ago
    Yes. It is raw data. I believe Massachusetts comes as close to capturing 100% of infections as anyone at this point, although that wasn't always the case. There is a lot of test infrastructure. CFR can overestimate or underestimate IFR. In most past pandemics, CFR significantly UNDERestimated IFR during the pandemic, because the raw numbers include a lot of people who are sick, but haven't died YET. There is a tendency to take CFR, apply very liberal corrections, sometimes multiple times, and get artificially low IFR numbers. Those aren't really plausible, even looking at scenarios where we could catalog all cases (e.g. the cruises had an IFR just over 2%, with state-of-the-art medical). My general impression, looking at data, is that the IFR was around 3-ish percent early on, and got better, potentially due to mutations making the virus less fatal, lower initial viral loads due to precautions, and/or better treatment.
    - anoncake 6y ago
      > I believe Massachusetts comes as close to capturing 100% of infections as anyone at this point, although that wasn't always the case. So it does not capture 100%. > CFR can overestimate or underestimate IFR. Incorrect. CFR cannot be higher than IFR. The number of deaths per confirmed infection obviously isn't higher than the number of deaths per infection (confirmed or unconfirmed). I'll continue to believe epidemiologists who actually have a clue what they are doing over your "general impression".
      - woofie11 6y ago
        The arrogance. If you want, an exert from an epidemiology text: "A CFR can only be considered final when all the cases have been resolved (either died or recovered). The preliminary CFR, for example, during an outbreak with a high daily increase and long resolution time would be substantially lower than the final CFR." The cited numbers are preliminary CFR, not final CFR. The key question is final IFR. Final IFR can be lower or higher than preliminary CFR. I'm done here.
        jsight 6y ago
        At least in the US, we are no longer in a period that qualifies as an "outbreak with a high daily increase". That implies a high increase relative to the current case count, which really isn't the case any more once you have a population case rate of 5%. I would be cautious about estimating countries that are only now having their first major outbreak (for example Czechia) for exactly the reason that you cite, but not every country is in that state. Put another way, we are no longer living with preliminary data in much of the world.
        woofie11 6y ago
        Let's toss some real numbers here. * Last death of Diamond Princess occurred after more than 2 months. * Over the past two months, the number of known cases in the US roughly doubled from 8 million to 16 million. This means there are a lot of people who haven't died yet, and dividing deaths by cases is an underestimate of CFR by up to a factor of 2. That doesn't mean we don't have the data. We could e.g. take the infections from June and look at how many of those resulted in death as a final CFR estimate. However, the number I gave (deaths over cases) didn't do that. It's a preliminary CFR without that sort of correction, and a correction to final IFR can go up (people who haven't died yet) or down (unidentified cases). anoncake argued this was an overestimate by definition, and anyone who said otherwise didn't have a clue. anoncake was wrong.
ddingus 6y ago
Current death rate is computed: Dead people / Recovered people and that is around 3 percent, down from 5 to 7 percent a few months ago. Infections / deaths is not accurate as cases in progress are added to those with definitive outcomes. That can also be seen as a best case, given nobody currently infected does not die. A worst case can be seen assuming everyone infected will die. But if we want to know the percentage of people who get sick and die, we have to compute that percentage with the people who got sick and did not die and their case ran its course to a definitive outcome. Infections lead to outcomes. Outcomes are where the dead vs not dead become known. Everyone infected will eventually arrive at an outcome.
- anoncake 6y ago
  I have no idea what you are trying to say, sorry.
  - ddingus 6y ago
    It is simple: Divide dead people by recovered people to assess true risk of death from COVID. Dividing dead people by infected people represents a best case, overly optimistic risk due to the fact that cases that ran their full course to a definitive outcome are mixed together with cases in progress. Doing that biases the risk perception down below actual risk and can be dangerous. I know one person who died, who did not have to, and they died because they made choices after having evaluated risk incorrectly.
    - anoncake 6y ago
      I'm sure the epidemiologists who estimate the IFR at ~0.5% are aware of the fundamentals of their field.
      - ddingus 6y ago
        As am I, but that does not make the risk assessment I put above wrong. They have their reasons for that metric. That too doesn't make (dead/recovered) wrong. There are dead people in my life, with whom I had this discussion. "Calculated risk" = optimistic. The IFR combines futures with actual and does not represent an outcome based risk assessment. Nature however, will definitely take you, or any of us, to an outcome, should we become infected.