[ I posted this on Radford’s blog but not sure if it posted. Will post here as well:]
Might I suggest another hypothesis: that in the metro areas of major cities in most countries that have been hit hard (i.e. those with linear growth rates in deaths), the infection rate among the frail WHO WERE GOING TO DIE OF OLD AGE ANYWAYS reaches saturation point. Then, all the old age deaths (that would have happened anyways), start being classified as covid deaths. In reality, they are probably best considered covid deaths+other causes. It would be fair to say covid expedited the death, but then so did every other infection and disease that they had at the time.
Then, what you get is the death rates are just representing the baseline natural causes death rate, where all those deaths are covid deaths now.
It would also be sufficient to explain the linear deaths if there was some set property of countries that made the “percent of the frail who get covid” being a constant. This to me also seems feasible. Each country will have a set “percentage of the elderly who live in nursing homes.” Not all do, but let’s suppose close to 100 percent of those in nursing homes contract covid. Their body will be fighting covid from now until the day they die (their immune system is weak, after all). Now, they won’t die from it right away; but they will die eventually soon anyways. The key point is that if we assume baseline old age death rates stay roughly constant, then the linearity can be explained by “a saturated percentage of old folks who die all having covid.”
Note then that this would have nothing to do with the mitigation measures done outside of nursing homes in each country. The death rates per day will simply be proportional to the percent of old folks near life’s end who have covid. This is likely just related to the percent of old folks whose living situation isolates them from the virus or not. This would be a constant per country.
That seems like a plausible explanation. I wonder how one could confirm it? If there were two sets of covid death counts based on strict and on loose criteria, one could see if there is a divergence in their rates of increase.
Your comment didn’t show up at my blog. Not sure why...
Fuirst, I don’t think looking at “number of cases identified” is useful at all for this analysis. Mainly because this is too much a function of how much testing you do that it’s not very meaningful.
Let’s write down the hypothesis: “the linearity observed is primarily due to the arrival of “deaths door” for those approaching end of life. A fixed fraction of these deaths have covid in any country because a fixed fraction are in living situations that give them some minimal exposure.”
So what are some consequences of this hypothesis?
We could look at the death rates of young folks, say those under 50. We’d assume that for that set, their death would not be caused by frailty with covid. If they die with covid and covid-related symptoms, this would not be a death due to frailty. We know that among those under 50 only a small percentage have complications, so this absolute number should be very small. However we know it is not 0.
What we should see is that THAT number should go up exponentially (though remain small) as reopenings begin. However, since that number will STILL be small even at full infection rates, this shouldn’t effect the totals too much because that will be dominated by the linear arrival of “imperfectly isolated frail folks dying with covid.”
As well, the test cases too look at would be cities and regions that have not seen major infection rates or death rates. We might expect that in a place like NYC (where serological tests are suggesting up to 20 percent infection rates), that the young folks who were going to die of this may have already died.
But perhaps the interior US will allow us to test this hypothesis.
My bet would be: a small absolute number of young folks dying of the disease, but that that absolute number will go up exponentially in regions with very small numbers of covid right now.
[ I posted this on Radford’s blog but not sure if it posted. Will post here as well:]
Might I suggest another hypothesis: that in the metro areas of major cities in most countries that have been hit hard (i.e. those with linear growth rates in deaths), the infection rate among the frail WHO WERE GOING TO DIE OF OLD AGE ANYWAYS reaches saturation point. Then, all the old age deaths (that would have happened anyways), start being classified as covid deaths. In reality, they are probably best considered covid deaths+other causes. It would be fair to say covid expedited the death, but then so did every other infection and disease that they had at the time.
Then, what you get is the death rates are just representing the baseline natural causes death rate, where all those deaths are covid deaths now.
It would also be sufficient to explain the linear deaths if there was some set property of countries that made the “percent of the frail who get covid” being a constant. This to me also seems feasible. Each country will have a set “percentage of the elderly who live in nursing homes.” Not all do, but let’s suppose close to 100 percent of those in nursing homes contract covid. Their body will be fighting covid from now until the day they die (their immune system is weak, after all). Now, they won’t die from it right away; but they will die eventually soon anyways. The key point is that if we assume baseline old age death rates stay roughly constant, then the linearity can be explained by “a saturated percentage of old folks who die all having covid.”
Note then that this would have nothing to do with the mitigation measures done outside of nursing homes in each country. The death rates per day will simply be proportional to the percent of old folks near life’s end who have covid. This is likely just related to the percent of old folks whose living situation isolates them from the virus or not. This would be a constant per country.
That seems like a plausible explanation. I wonder how one could confirm it? If there were two sets of covid death counts based on strict and on loose criteria, one could see if there is a divergence in their rates of increase.
Your comment didn’t show up at my blog. Not sure why...
Fuirst, I don’t think looking at “number of cases identified” is useful at all for this analysis. Mainly because this is too much a function of how much testing you do that it’s not very meaningful.
Let’s write down the hypothesis: “the linearity observed is primarily due to the arrival of “deaths door” for those approaching end of life. A fixed fraction of these deaths have covid in any country because a fixed fraction are in living situations that give them some minimal exposure.”
So what are some consequences of this hypothesis?
We could look at the death rates of young folks, say those under 50. We’d assume that for that set, their death would not be caused by frailty with covid. If they die with covid and covid-related symptoms, this would not be a death due to frailty. We know that among those under 50 only a small percentage have complications, so this absolute number should be very small. However we know it is not 0.
What we should see is that THAT number should go up exponentially (though remain small) as reopenings begin. However, since that number will STILL be small even at full infection rates, this shouldn’t effect the totals too much because that will be dominated by the linear arrival of “imperfectly isolated frail folks dying with covid.”
As well, the test cases too look at would be cities and regions that have not seen major infection rates or death rates. We might expect that in a place like NYC (where serological tests are suggesting up to 20 percent infection rates), that the young folks who were going to die of this may have already died.
But perhaps the interior US will allow us to test this hypothesis.
My bet would be: a small absolute number of young folks dying of the disease, but that that absolute number will go up exponentially in regions with very small numbers of covid right now.