johnh wrote:
Not sure if I posted that. As I mentioned before I’ve been spending (far too much) time on epidemic simulation. One thing it shows is that things are very flat above R=1. It’s not atr all the case that at R=1.000001 everybody drops dead on the spot, you just have a long, long period with a relatively small number of infected people.
Thanks Johnh.
Put simply at values a little above 1, it rumbles on slowly.
Have you done any modeling as to how long it will take to get to around the magic 65% immunity, but assuming remaining with the capacity of the NHS at say 40% occupancy of COVID specific capacity?
Some slightly more optomistic vaccine news today with Oxford seeing good success with the first animal trials – a long way to of course.
johnh wrote:
One thing it shows is that things are very flat above R=1. It’s not atr all the case that at R=1.000001 everybody drops dead on the spot, you just have a long, long period with a relatively small number of infected people.
Sure, at ‘1.0000001’. But at just 1.1 you get a big increase, and pretty quickly.
I remember Merkel talking a few weeks ago. If the R rises to 1.1, the German health service will be overwhelmed by October. At 1.2, by July, and 1.3 by June.
A little before the UK’s peak, we were thought to have an R of 3.
1.3 and June seems counter intuitive. As you say we were thought to have 3 at its peak, and presumably high figures on the shoulder each side, and yet we peaked at less than 50% capacity of the NHS and possibly less, depending which set of data you use. I am not disputing the rapid rise, perhaps more than in the real world it hedges on the lower side of the predictions as opposed to the higher. Perhaps this is because we have tended to overestimate the R? By simple extrapolation of hospital cases as I said earlier it seems to me the overall level of infection in the population should be higher than it actually is based on the first results that have been published of the UK’s population testing survey. The slight worry is this could also lead to the conclusion that the number of people who end up in hospital or passing is in fact a little higher than the mortality figures suggested would seem to indicate.
At the beginning, the number of cases was tripling every 3-4 days, so the difference between peaking at 50% and being overwhelmed is around 2 days.
Interesting new data in the Daily Trash
This is interesting, and not really surprising knowing that so many people in London live and move more or less on top of each other
I am glad there is some fallout taking place over the way the UK population has been treated as a load of morons. Not everybody is thick, and they are not telling us anything useful. There is still no data on infections, which is properly localised. You just get it by the county, but if say there were 50 cases arising out of some event, nobody will tell you…
I remember Merkel talking a few weeks ago. If the R rises to 1.1, the German health service will be overwhelmed by October. At 1.2, by July, and 1.3 by June.
I don’t buy that. But then everybody means something different, and politicians have to manage the herd which is much more complicated than epediemology. If by R=1.3 you intend to mea that the health system will be overwhelmed by June, then for sure you are right (or she is). The reality of the math is that R=1.3 just gives you a long, long low level of infection. But nobody knows how to behave to make R=1.3, or 1.001, or 2.163421 or any other number, so the important thing is to scare people into staying keeping distancing up.
One thing my model shows clearly is that the effect of distancing is very non-linear. If 0 is business as usual and 1 is Wuhan, then anything under about 0.6 makes little difference.
Isn’t R 1.3 equivalent to around 2700 after 30 days, so assuming some base of the population being infected (and just a few thousand in Germany would be a conservative base), several million would be infected over the period with a relevant proportion needing ICU. Relevant wrt ICU capacity which reduces as health staff burn out get infected.
R0 change with time and Rt goes to zero with time as those infected recovered are taken out of population, I think anything with R less than 1.30 falls under “controllable” in one month, like the 30% APR on my credit card that I check on a monthly basis, “surprising” but “controllable” 
RobertL18C wrote:
Isn’t R 1.3 equivalent to around 2700 after 30 days, so assuming some base of the population being infected (and just a few thousand in Germany would be a conservative base), several million would be infected over the period with a relevant proportion needing ICU.
That’s my understanding.
RobertL18C wrote:
Isn’t R 1.3 equivalent to around 2700 after 30 days, so assuming some base of the population being infected (and just a few thousand in Germany would be a conservative base), several million would be infected over the period with a relevant proportion needing ICU. Relevant wrt ICU capacity which reduces as health staff burn out get infected.
Correct summary. That is why Merkel is right on the matter and any R > 1 spells trouble eventually, the higher the number the earlier.
As for reduced ICU capacity:
The first member of our department has just tested SARS-CoV-2 positive, which result in a further 6 colleagues having to be quarantined as a precaution. That demonstrates how devastating staff infections in health care can be for overall treatment capacity.
Luckily, this happened at a time where we have loads and loads of spare capacity. Infections have stabilized in Germany at a rate of roughly 1000/day over the last week. At this rate, it would take ages to achieve herd immunity, however, so that’s not a solution we can aim for.
Interestingly, testing capacity is underused, standing at about 1 million tests/week available but less than 400.000 used last week. This is rightfully drawing criticism from experts snd journalists alike.
Imho, hospital staff could be tested in intervals even if asymptomatic. You wouldn’t even have to test individuals: you can take samples from say ten people and combine them, and only test each individual if the combined sample was positive. This allows to use capacity for testing both broadly and effectively.
