Thread by @AlanBixter

COVID Mortalities, Shades of Truth, and Under the Spin (Thread).
We know the raw case mortality rate of COVID in Australia: 1,039 deaths from 61,609 cases = 1.69%.
If we take all COVID deaths to the beginning of June '21, we get 910 deaths from 30,118 cases = 3.03%. /2

It looks like we got better at looking after COVID, right. But we can go further: Vic 2nd wave (804/2,0393=3.94%) vs NSW current wave (126/28,321= 0.44%). Wow - that must mean Labor is shit and Liberals are brilliant. Or vaccines are magic. /3

The problem with all of those raw mortality calculations, except the one embedded entirely in 2020, is that they assume that that a person who dies of COVID immediately they are informed of the diagnosis. Which is obviously a stupid assumption. /4

Despite being a stupid assumption, it is used a lot. The conclusions based on this stupid assumption got so extreme, that NSW Health in their COVID-19 Weekly Surveillance Report Published 30 August 2021, includes a new note - Read the last sentence: /5

So the problem is, that from the time people are diagnosed to the time they die of it, varies, distributed over a period of time, and to get the true mortality rate, we need to wait until they all recover or die. /6

Basically we need to know how far back we need to look at the total case numbers to match up with today's total of mortalities. It won't be perfect, but it will be better than pretending today's cases will never die. /7

NSW Health make their point without applying any correction. But they have a strange choice of central tendency to talk about the problem: Median. If you wrote down the numbers of days to death of each patient from shortest to longest, the Median is the middle number. /8

This is a strange choice, because obviously the numbers can only range down to 0, but may range up to 100's. The distribution is not symmetrical it is skewed by a "left wall effect" You can't have a number of days less than 0, so the shape is probably something like this. /9

In the VIC 2nd wave we did not use the median; we used the mean (what most people intend when they say average): all the lengths of time added together divided by the number of deaths. In the VIC 2nd wave, this was 4 weeks. /10

So if we look at the NSW outbreak, on the 4/9/21 a total of 126 deaths for 28,321 cases gives a raw mortality rate of 0.44%. Corrected using the median time to death by backing up case numbers 111 days, we get 126 deaths for 14,850 cases: 0.85%. /11

But if we use the correction used for the VIC 2nd wave, 28 days, we get 126 deaths for 5,218 cases and 2.41%.
You will notice that most media quote the 0.85% figure but then compare it to the VIC 2nd wave mortality calculated by either actual, total, deaths, or the mean. /12

This is deceptive and creates a false impression that Delta magically both kills quicker, but fewer people; which at face value would be an odd thing. So, until NSW Health publish the actual distribution of times to death, I'll keep using the mean of 28 days as a comparator.

@ThreadReaderApp
unroll