Just had another look at back calculating R (Covid infection rate) from hospitalisations as well as from deaths and recorded infections. I think I've got it now. 
Hospitalisation requirements change, treatments change, testing regimens change with capacity and surge testing, I'm quite surprised the three calculations align as well as this
What does this tell us? Well, to line up R from hospital admissions we need to go eleven days back. That fits in well with the ONS data saying 6 days, especially for older folks, after showing symptoms (on average 5 days after infection). So peak hospitalisation is 11 days
Whereas peak time to get a test result reported is 8 days after infection - thats what you need to use to line R from infections up with R from deaths (deaths don't lie, its the most tragically unambiguous data point)
...or in other words peak time for symptoms is 5 days, it takes a day or two to get tested and a result logged, so peak test result is day 8, and peak hospitalisation is day 11. Peak day to die is, as we already know, 18 days after symptoms, 23 days post infection.
Pondering what else I can get out of the data adding hospital admissions into the mix. Ideas?
Was thinking about this over lunch (see thread). If peak time to get a test result is 8 days, and peak time for hospitalisation is 11 days, then from that you can work out hospitalisations as a percentage of recorded cases. And thats a wild looking graph.
Settling down now to 0.1% of cases, more or less, ending up in hospital. Thats a result both of mass testind and vaccines.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
