By March of 2022, rates of mortality appeared to have reverted to normal, but a closer look reveals all age cohorts have paid a heavy price over the course of two years of poor public policy.
Troubling charts. The confounding of drugs with the vaccines makes the analysis difficult. Still among working age people, drug use in not typically normal. Despair might play a part but by middle age coping skills ought to exist.
Will look forward to the state by state analysis. That might bring clues.
Just so I understand what percentage excess mortality means. If the normal number of deaths in a period is say 100 but actually there were 150 deaths then that would be a 50% excess mortality. Is that right?
Excess mortality in absolute terms (50 fatalities in your hypothetical) can be illuminating in its way. But expressing excess mortality as a proportion of benchmark mortality (50% in your hypothetical) can also be illuminating in its own way. The latter helps pick up important changes in rates of mortality in populations that don't generate much mortality. Such is the case with very young people. Very few of them die from anything, and we might not really notice if, say, in a country of 330 million people, 135 more young people each start dropping off. But that constant simmer of 135 per week adds up. The "excess" of young people dying is like a the steady simmer of a silent and nearly invisible Vietnam War. (The American toll in Vietnam exceeded 60,000 over eight years.)
It would be interesting to see how sensitive the analysis is to different measures of excess deaths. I have seen other analysis where the picture changes totally depending on what years are included in the baseline.
Do you also get more of a breakdown of cause of death from this data? In the UK it has been heart disease type issues that have stood out more than anything.
(1) The method I apply is very linear. I basically estimate a linear trend and toss a sine wave on top of it. I use data from the years 2015-2019 to estimate that baseline.
Projecting that baseline post-2019 amounts to projecting a slowly increasing trend in expected fatalities.
(2) the linear trend would be little skewed were I to estimate with data going back further t0 2012. I have noticed that mortality rates had increased around 2012. (It looks like Baby Boomers were just starting to retire en masse. Perhaps they started dying at higher rates, too.) But then the annual increase in rates settled down again.
(3) Some people use moving averages that exploit the lagging five years or so. Including the last few years in a baseline would amount to folding in some of the elevated, post-COVID mortality. I can imagine two effects on estimates of excess mortality: (a) a level effect in that one might end up with a higher "expected" mortality and thus lower estimates of excess mortality; (b) insofar as COVID effects align with the usual cold-season peaks in mortality: folding previous COVID peaks in to the baseline would certainly diminish estimates of seasonal peaks in excess mortality going forward.
(4) One may certainly wonder if the age distribution of the population has changed in a way that is easily discernible in available data (from, say, the Census Bureau estimates of annual population). I might think of using some crude measures of age distributions to adjust estimates of benchmark mortality. I expect such work would make estimates of benchmark mortality more modest ... and thus make excess mortality appear more extreme. Hmm ...
On 4 I would have imagined that US population is aging (,a quick search seems to confirm). So if your benchmark was done for the whole population your might be overestimating younger cohorts hence underestimating the excess there (and the opposite for older cohorts).
I actually estimated benchmarks for each cohort separately. BUT, I can say that the linear "non-linear regression" (a regression equation that features a linear time trend with a sine wave) does not fit the youngest cohort as well as other cohorts. (One can see evidence of that in a graph that I included in my post of May 23.)
Your comments have gotten more interested in digging into the younger cohorts ...
I must admit that I hadn't realised how many excess deaths there were even before the vaccines were rolled out.
When the pharmaceutical companies started bringing out their vaccines but before they'd rolled out the big vaccination programmes those companies were quoting figures like 70-90% effectiveness. At the time I assumed that this was the effectiveness against infection. So if it had 90% effectiveness and 100 people were exposed to the virus 10 people would catch it and 90 would not. I'm sure I'm not the only one who thought that. It's obviously wrong. Vaccination doesn't stop you catching the disease at all. So what did the pharma companies mean by effectiveness? Effectiveness against dying? Also have they changed their definition of effectiveness?
Yes, Steve, what does vaccine "effectiveness" mean? I had to do some digging around on that point, but last January I did make contact with it. See, "The Orpheus Fallacy" at https://dvwilliamson.substack.com/p/the-orpheus-fallacy
Pfizer had to submit a study to the FDA to secure its EUA ("emergency use authorization"). Pfizer jabbed about 22,000 people with the vax and jabbed another 22,000 with a placebo (saline). Over the course of six months, some number of people in control group contracted COVID. Only about 5% as many people who got vaxxed contracted COVD. Hence the claim that the vax was "95% effective."
But, get this: more people in the vaxxed group died over the course of six months than in the control group. So, on net, did the vax really protect people? No. And that was Pfizer's own study.
I have trouble understanding it. Everyone I know has had covid despite being fully vaccinated. My three daughters and their partners, my mother (96), my brother and my two sisters and their partners have all had it. I could go on. The only people I know who haven't had it is my wife and myself. We live out in the sticks on a farm and we don't go out much. So it appears to me that vaccination does not stop anyone from catching the disease at all. Therefore to base an effectiveness calculation on the number of people who did or did not get the disease is meaningless.
Troubling charts. The confounding of drugs with the vaccines makes the analysis difficult. Still among working age people, drug use in not typically normal. Despair might play a part but by middle age coping skills ought to exist.
Will look forward to the state by state analysis. That might bring clues.
There are more clues. And more puzzles!
Just so I understand what percentage excess mortality means. If the normal number of deaths in a period is say 100 but actually there were 150 deaths then that would be a 50% excess mortality. Is that right?
Right on, brother!
Excess mortality in absolute terms (50 fatalities in your hypothetical) can be illuminating in its way. But expressing excess mortality as a proportion of benchmark mortality (50% in your hypothetical) can also be illuminating in its own way. The latter helps pick up important changes in rates of mortality in populations that don't generate much mortality. Such is the case with very young people. Very few of them die from anything, and we might not really notice if, say, in a country of 330 million people, 135 more young people each start dropping off. But that constant simmer of 135 per week adds up. The "excess" of young people dying is like a the steady simmer of a silent and nearly invisible Vietnam War. (The American toll in Vietnam exceeded 60,000 over eight years.)
It would be interesting to see how sensitive the analysis is to different measures of excess deaths. I have seen other analysis where the picture changes totally depending on what years are included in the baseline.
Do you also get more of a breakdown of cause of death from this data? In the UK it has been heart disease type issues that have stood out more than anything.
I can say a few things:
(1) The method I apply is very linear. I basically estimate a linear trend and toss a sine wave on top of it. I use data from the years 2015-2019 to estimate that baseline.
Projecting that baseline post-2019 amounts to projecting a slowly increasing trend in expected fatalities.
(2) the linear trend would be little skewed were I to estimate with data going back further t0 2012. I have noticed that mortality rates had increased around 2012. (It looks like Baby Boomers were just starting to retire en masse. Perhaps they started dying at higher rates, too.) But then the annual increase in rates settled down again.
(3) Some people use moving averages that exploit the lagging five years or so. Including the last few years in a baseline would amount to folding in some of the elevated, post-COVID mortality. I can imagine two effects on estimates of excess mortality: (a) a level effect in that one might end up with a higher "expected" mortality and thus lower estimates of excess mortality; (b) insofar as COVID effects align with the usual cold-season peaks in mortality: folding previous COVID peaks in to the baseline would certainly diminish estimates of seasonal peaks in excess mortality going forward.
(4) One may certainly wonder if the age distribution of the population has changed in a way that is easily discernible in available data (from, say, the Census Bureau estimates of annual population). I might think of using some crude measures of age distributions to adjust estimates of benchmark mortality. I expect such work would make estimates of benchmark mortality more modest ... and thus make excess mortality appear more extreme. Hmm ...
On 4 I would have imagined that US population is aging (,a quick search seems to confirm). So if your benchmark was done for the whole population your might be overestimating younger cohorts hence underestimating the excess there (and the opposite for older cohorts).
I actually estimated benchmarks for each cohort separately. BUT, I can say that the linear "non-linear regression" (a regression equation that features a linear time trend with a sine wave) does not fit the youngest cohort as well as other cohorts. (One can see evidence of that in a graph that I included in my post of May 23.)
Your comments have gotten more interested in digging into the younger cohorts ...
I must admit that I hadn't realised how many excess deaths there were even before the vaccines were rolled out.
When the pharmaceutical companies started bringing out their vaccines but before they'd rolled out the big vaccination programmes those companies were quoting figures like 70-90% effectiveness. At the time I assumed that this was the effectiveness against infection. So if it had 90% effectiveness and 100 people were exposed to the virus 10 people would catch it and 90 would not. I'm sure I'm not the only one who thought that. It's obviously wrong. Vaccination doesn't stop you catching the disease at all. So what did the pharma companies mean by effectiveness? Effectiveness against dying? Also have they changed their definition of effectiveness?
Yes, Steve, what does vaccine "effectiveness" mean? I had to do some digging around on that point, but last January I did make contact with it. See, "The Orpheus Fallacy" at https://dvwilliamson.substack.com/p/the-orpheus-fallacy
Pfizer had to submit a study to the FDA to secure its EUA ("emergency use authorization"). Pfizer jabbed about 22,000 people with the vax and jabbed another 22,000 with a placebo (saline). Over the course of six months, some number of people in control group contracted COVID. Only about 5% as many people who got vaxxed contracted COVD. Hence the claim that the vax was "95% effective."
But, get this: more people in the vaxxed group died over the course of six months than in the control group. So, on net, did the vax really protect people? No. And that was Pfizer's own study.
I have trouble understanding it. Everyone I know has had covid despite being fully vaccinated. My three daughters and their partners, my mother (96), my brother and my two sisters and their partners have all had it. I could go on. The only people I know who haven't had it is my wife and myself. We live out in the sticks on a farm and we don't go out much. So it appears to me that vaccination does not stop anyone from catching the disease at all. Therefore to base an effectiveness calculation on the number of people who did or did not get the disease is meaningless.
A farm out in the sticks ... I'm envious. I've been keeping my eye out for a space out in the sticks for some time.