Open RongzongWu opened 2 years ago
Method A (pproximation):
To make this more accurate, we need to look at countries with high vaccination rates. I picked South Korea (85% fully vaccinated) and Israel (66% fully vaccinated) as of 1/21/2022, according to google new:
https://news.google.com/covid19/map?hl=en-US&gl=US&ceid=US%3Aen&mid=%2Fm%2F06qd3 for South Korea https://news.google.com/covid19/map?hl=en-US&gl=US&ceid=US%3Aen&mid=%2Fm%2F03spz for Israel
Let's look at South Korea first:
The vaccination starts about April 1, 2021, and plateaued around November 1, 2021. There were two waves of death:
Around Jan 7, 2021: 7-day average death is 23, this is before the vaccination began, Around Dec 25, 2021: 7-day average death is 75, this is after the vaccination has plateaued.
A case wave should occur about one to two weeks before a death wave. The case waves correspond to the above two death waves should be the ones around Dec 25, 2020, 7-day average case is 1047, and around Dec 18, 2021, with a 7-day average case of 6832.
So the death rate before vaccination is 23/1047 = 2.2%. the death rate after vaccination is 75/6832 = 1.1%.
So the death rate reduced from 2.2% to 1.1%, significant but still within the order of magnitude of each other.
Do the same analysis for Israel and USA, and here is the summary:
In summary, based on Method A, it seems that vaccination helps significantly, but the death rate for a person that has contracted COVID is on the same orders of magnitude regardless of the vaccinated status.
Note that since the vaccination rate is not 100%, the effectiveness of vaccination calculated using Method A is probably less than the truth. So these should be viewed as lower bounds. In other words, we should expect that vaccination to reduce death rate by at least 50%.
In order to do Method B (Best Estimate). Let's see if we can find some number from CDC of the U.S. government.
Here is the page that shows the relevant data:
https://covid.cdc.gov/covid-data-tracker/#rates-by-vaccine-status
There is a link to download the data, see the screenshot below:
A copy of the data I downloaded today (1/23/2022) can be found here in this repo:
Note that GitHub won't show all the columns. Also note that the data seems to be slightly different between my last download and this download, even for historical data.
After comparing the data in the CSV file against the chart showing on the CDC site, it is clear that the chart is showing the "Age adjusted vax IR", and "Age adjusted unvax IR" columns. "IR" means incident rate, which is incident (either death or COVID case) per 100,000 population.
Here is all the data showing on the chart before splitting into age groups and vaccine type:
Week | Age adjusted vax case IR | Age adjusted unvax case IR | Age adjusted vax death IR | Age adjusted unvax death IR |
---|---|---|---|---|
202114 | 21.02451063 | 203.9323319 | 0.14802466 | 2.842742135 |
202115 | 18.52872443 | 203.3859786 | 0.159363074 | 2.993971489 |
202116 | 16.66126192 | 175.4118045 | 0.132634061 | 2.943645475 |
202117 | 15.98967457 | 159.6048127 | 0.135112389 | 2.588312523 |
202118 | 11.46537228 | 136.1643299 | 0.121080928 | 2.467597783 |
202119 | 8.574761325 | 116.5863226 | 0.089706092 | 2.003318512 |
202120 | 7.292606748 | 90.22434023 | 0.087131345 | 1.673362986 |
202121 | 5.736668205 | 68.46282246 | 0.070741957 | 1.43467003 |
202122 | 5.141028089 | 55.5179887 | 0.070511 | 1.227541079 |
202123 | 5.19057167 | 50.72798976 | 0.06595192 | 1.175770354 |
202124 | 5.438432335 | 49.18168573 | 0.075719921 | 1.16947474 |
202125 | 7.220768436 | 54.56028842 | 0.09677206 | 1.362497124 |
202126 | 11.04556018 | 72.5216213 | 0.101019844 | 1.738044504 |
202127 | 18.48788738 | 109.0548874 | 0.170868479 | 2.738782493 |
202128 | 33.63993835 | 172.8768935 | 0.295105522 | 4.075500579 |
202129 | 58.39782885 | 283.7455799 | 0.473397441 | 7.225968495 |
202130 | 88.29428495 | 424.1256506 | 0.755289571 | 11.02238333 |
202131 | 112.1479814 | 530.772165 | 0.934321269 | 14.59533223 |
202132 | 122.7498664 | 607.8844795 | 1.124302436 | 17.04765669 |
202133 | 128.3088508 | 657.1131485 | 1.119929796 | 18.16471288 |
202134 | 132.0615953 | 682.8966736 | 1.055002165 | 17.35325648 |
202135 | 125.4784082 | 658.8965665 | 0.987623263 | 16.64901073 |
202136 | 113.1012979 | 595.1731908 | 0.939754274 | 14.43491032 |
202137 | 102.4493681 | 533.0887034 | 0.809496229 | 12.39061186 |
202138 | 87.84098312 | 454.9216632 | 0.710131159 | 10.2544866 |
202139 | 82.97084161 | 408.3782701 | 0.591209942 | 8.676893667 |
202140 | 77.66425169 | 373.4160667 | 0.533698312 | 7.709436743 |
202141 | 69.32257568 | 329.4917907 | 0.50222873 | 7.116481256 |
202142 | 65.30913674 | 305.4796871 | 0.467174931 | 6.431453612 |
202143 | 68.7723518 | 318.0736112 | 0.455656623 | 6.301405512 |
202144 | 78.34089861 | 337.1566391 | 0.461489254 | 6.882448207 |
202145 | 85.36868555 | 377.6721347 | 0.484333139 | 7.471984836 |
202146 | 92.34663458 | 418.2837514 | 0.509388337 | 8.39450078 |
202147 | 82.89861928 | 375.3168955 | 0.510981089 | 7.770561558 |
202148 | 123.852306 | 559.0325355 | 0.465690889 | 9.197037038 |
202149 | 113.359997 | 528.5697208 | ||
202150 | 207.1574417 | 639.8792063 | ||
202151 | 443.6895645 | 1006.387219 |
To calculate the death rate, just divide the death incident rate (IR) by the case incident rate. For example, for the week of 202114 (week 14 of year 2021), the death rate for vaccinated people is 0.14802466/21.02451063 = 0.7%, and for unvaccinated people this is: 2.842742135/203.9323319 = 1.4%. This means a reduction in death rate of 50%.
Week | Vaxed Death Rate (%) | Unvaxed Death Rate (%) | Reduction in Death Rate (%) |
---|---|---|---|
4/10/2021 | 0.7 | 1.4 | 49.5 |
4/17/2021 | 0.9 | 1.5 | 41.6 |
4/24/2021 | 0.8 | 1.7 | 52.6 |
5/1/2021 | 0.8 | 1.6 | 47.9 |
5/8/2021 | 1.1 | 1.8 | 41.7 |
5/15/2021 | 1.0 | 1.7 | 39.1 |
5/22/2021 | 1.2 | 1.9 | 35.6 |
5/29/2021 | 1.2 | 2.1 | 41.2 |
6/5/2021 | 1.4 | 2.2 | 38.0 |
6/12/2021 | 1.3 | 2.3 | 45.2 |
6/19/2021 | 1.4 | 2.4 | 41.4 |
6/26/2021 | 1.3 | 2.5 | 46.3 |
7/3/2021 | 0.9 | 2.4 | 61.8 |
7/10/2021 | 0.9 | 2.5 | 63.2 |
7/17/2021 | 0.9 | 2.4 | 62.8 |
7/24/2021 | 0.8 | 2.5 | 68.2 |
7/31/2021 | 0.9 | 2.6 | 67.1 |
8/7/2021 | 0.8 | 2.7 | 69.7 |
8/14/2021 | 0.9 | 2.8 | 67.3 |
8/21/2021 | 0.9 | 2.8 | 68.4 |
8/28/2021 | 0.8 | 2.5 | 68.6 |
9/4/2021 | 0.8 | 2.5 | 68.9 |
9/11/2021 | 0.8 | 2.4 | 65.7 |
9/18/2021 | 0.8 | 2.3 | 66.0 |
9/25/2021 | 0.8 | 2.3 | 64.1 |
10/2/2021 | 0.7 | 2.1 | 66.5 |
10/9/2021 | 0.7 | 2.1 | 66.7 |
10/16/2021 | 0.7 | 2.2 | 66.5 |
10/23/2021 | 0.7 | 2.1 | 66.0 |
10/30/2021 | 0.7 | 2.0 | 66.6 |
11/6/2021 | 0.6 | 2.0 | 71.1 |
11/13/2021 | 0.6 | 2.0 | 71.3 |
11/20/2021 | 0.6 | 2.0 | 72.5 |
11/27/2021 | 0.6 | 2.1 | 70.2 |
12/4/2021 | 0.4 | 1.6 | 77.1 |
The death rate per COVID cases is shown in the following chart for vaccinated and un-vaccinated populations respectively. It seems that the death rate changes a lot, not sure why. Overall the numbers are in the same order of magnitude as the ones calculated through Method A.
The reduction in death rate is plotted below:
It seems that vaccination initially only reduce death rate by about 40% then increased to about 60-70%, with a sudden jump occurring around end of June and beginning of July 2021. This seems to coincide with the time when the vaccination rate plateaued.
The 60-70% reduction rate is consistent with the numbers we got from Method A (which provides a lower bound estimate of 50%).
With a 70% reduction, we are looking at 2 out of 3 people that can be saved by vaccination from death, if contracted COVID.
I believe this is significant. However, vaccination is far from being a silver bullet, it does NOT eliminate death, only 70% of it. To say all COVID death is avoidable is wrong.
To answer this question we need to look at the death rate (defined as death per case) among vaccinated people vs. unvaccinated people.
There are several ways to come up with the answer.
One rough (order of magnitude) estimate is to look at the death rate from a country before vaccination begins, and the death rate after majority of the population has been vaccinated. Lets call this Method A. A stands for "Approximation".
The other is to look at more specific data: the death amongst vaccinated cases and death amongst unvaccinated cases. Let's call this Method BG, where B stands for "Best Guest".