@TheABC
Does anyone have links to reliable data on hospitalisation? The NHS in our area is bring fucked over anyway by the social care crisis and I don't think there's much more give in the system.
Oliver Johnson
@BristOliver
Now that some previous omicron hope has expired (it does seem to be growing faster, it's not just in South Africa) the main game in town is the m-word. Is it milder? But I think that's tricky to work out, particularly at the moment.
1. You can't tell anything from whether hospitalizations grow exponentially or not. Basically it would be a huge surprise if they didn't, when cases are growing exponentially. The default is that they'll grow at more or less the same exponential rate.
2. In order to determine whether it's milder, you need some confidence in what fraction of infections or cases lead to hospitalization. But this gets into questions of lag: how long from case to admission, when you try to line up the two time series against each other.
3. For example, the @Spectator Data Hub says 2% in the UK at the moment, based on a 10 day lag. But even if you use a different lag you wouldn't get a very different answer, because everything is broadly flat (up or down only 10% or so per week).
4. Whereas in South Africa, where cases are growing something like 500% per week (30% per day!) if you misestimate the lag by a couple of days in either direction, your estimate of "fraction hospitalized" might be out by 70% or more in either direction.
5. You can try to estimate the lag by lining up "the start of the wave" in terms of cases and hospitalizations, but this is tricky because there isn't even a fixed lag, it's a probability distribution (some people are admitted on day 0, some a lot later).
6. Honestly, it's a complicated mess, and I think (boring answer) we need to wait for more data based on actually tracking particular patient outcomes before coming to any firm conclusions.
I was surprised by the % he thinks you could be out by with the data we have atm.