Mortality rate in the ICU

[MarcusUppmax]

Mortality rate in ICU should depend on whether ICU resources are overwhelmed. Just testing the code a little bit with 100k population yields scenarios with up to 43 ICU patients per 100k, which is about 7 times the number that Sweden usually maintains.

I recall seeing an article a couple of weeks ago estimating the change in CFR between a non-overwhelmed healthcare system and an overwhelmed one, but I cannot locate it anymore.

An alternative, if mortality rate cannot be quantified accurately, is to not report deaths in this simulation. Number of hospital and ICU beds required is perhaps more useful information in any case.

[jasminegardner]

Thank you for your comment. We are going to add this functionality but at the moment it is more important to get initial data and understand the base case data without extraneous assumptions. It is obvious that hospitals being overwhelmed is an issue which will highly increase the rate of death (as we see in Italy). In such a situation, it is hard to quantify the actual effect because 1) we do not definitely know how many ICU beds are actually in Sweden (this is not public data), 2) the Swedish government and folkhälsomyndigheten are committed to increasing the number of ICU and hospital beds as the situation rises (it's hard to pin down a moving target) and 3) we have not directly quantified the rates of death without proper ICU care (we are working on getting a reasonable estimate). Ideally, we would like to show a scenario where theoretically there are enough hospital beds and the death rate (which is fairly well quantified at this point) is at normal rates and show a secondary worst case scenario which involves the collapse of the healthcare system.

[MarcusUppmax]

Thanks, that makes sense.

Could you write an issue or a comment somewhere about what the goals are with this project? Is it a goal to run a full 10 million individual simulation or something smaller (right now I think it breaks at 1 million or so)? I've been looking over the code and now I'm not sure what to work on. Would a parallelised version be useful, for instance?

[peterkasson]

We have a lot of effort going on right now, thanks to great involvement from the centers--I think Åke is working on a parallelized version. Asking him might be best. Right now we're planning on ensembles of full-sized simulations. Thanks again, --Peter

…

On Tue, Mar 24, 2020 at 8:05 AM MarcusUppmax ***@***.***> wrote: Thanks, that makes sense. Could you write an issue or a comment somewhere about what the goals are with this project? Is it a goal to run a full 10 million individual simulation or something smaller (right now I think it breaks at 1 million or so)? I've been looking over the code and now I'm not sure what to work on. Would a parallelised version be useful, for instance? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACRWZNMCLACD4PL4JRZSFSLRJDD2HANCNFSM4LSWE57A> .

[akesandgren]

I've added an Issue, #5, for keeping track of things other people can start looking at, please add to it if you come up with other things.

[jasminegardner]

I'm going to close this one and we can keep discussion on Åke's thread. Thank you for your feedback and continuing help, Marcus!