Read on Twitter
Hello peeps!Got more data from Scotland for you. I have been mulling this over for a while.
So here is positive tests to death. I think the delay is a little on the long side.... not sure. But matching peaks gives 18 day lag and good correlation on the decline.
1/
Deaths reported fluctuate wildly on a day-to-day basis and Sunday and Monday always very low.2/
Applying this 18 day delay and taking the ratio of deaths per positive test, we see that covid appears to be now about 3 times more deadly compared to September. I speak to clinicians. They do not accept this is true at all. I believe them. But the numbers say different.3/
Now ITU *occupancy* to positive tests. (Admissions more tricky because data are appallingly inconsistent & change back 2 weeks in to past some days.)Again, good correlation matching peaks. But the 16 day delay seems very long. Correlation lost somewhat in last 3 weeks.
4/
25% fewer patients per case ending up in ITU compared to Sept. So while covid getting more deadly, also putting fewer people in ITU? Those two conclusions don't sit together.Are patients kept out of ITU bc it won't do them any good? But more dying outside ITU elsewhere?
5/
Here, we see relative size of ITU peaks in 2 waves (2nd wave much smaller&broader). Deaths comparable in 2 waves (but smaller in 2nd). Have used deaths/ITU occupant, not because that's where people die of covid (I know it's not), but because it's measure of severe illness.6/11
... so again, we see similar deadliness against ITU occ'y to first wave in second and yet, ITU occupancy is much lower this time around. I have used a 2 day delay, based on data from tweets 1&4 above and it's possible the 2 waves should have different lags.Thoughts!
7/11
This is a longer scale than above, going back to start of September, showing that covid has become 4 times more deadly in this time. New *cases* (sorry) have not been increasing very significantly in this time.8/11
Lots going on in this graph from Mar. Number of tests is extremely high now compared to spring, but it also fluctuates wildly day to day. In the spring although we weren't doing as much testing, I do think we caught those who died of covid so I think in many ways....9/11
... the deaths per case were more reliable then. Now we test SO MUCH we are finding so many cases and many fewer deaths relatively, and yet deaths/case rising now.I genuinely want your thoughts. I'm not really putting out any explanations.
10/11
I will say, in my very limited data analysis experience, a lot of this doesn't make any sense. The statements don't sit well together. They don't work.We are doing too much testing. I do worry that the real clinical picture is being obscured by #pcrtest
11/11.
