The Arizona modeling team has been running models to predict hospital demand from both COVID and non-COVID demand drivers. On November 27th, they wrote to the Arizona state government about their most recent model runs. The results are dire:
No matter what actions are taken, Arizona will experience a hospital crisis in the coming weeks. However, if action is not immediately taken, then it risks a catastrophe on the scale of the worst natural disaster the state has ever experienced. It would be akin to facing a major forest fire without evacuation orders….”
This is a dire half paragraph from academics.
And it is based on modeling that has a key assumption that is probably optimistic:
The models that are projecting a baked in crisis and a catastrophic outcome under a no-change scenario are assuming that there is no change in transmission dynamics from the Thanksgiving holiday. That is a shaky assumption. We know that Thanksgiving has lots of people traveling, even if travel was down significantly this year compared to 2019. We know that Thanksgiving tends to have a lot of talking and loud voices in confined, interior spaces. We know that Thanksgiving tends to have a good amount of drinking. I think that making the assumption that Thanksgiving will not change infection trends is a key assumption that requires significant sensitivity analysis.
However, I also think that the modeling exercise with data that probably was fresh on Tuesday or Wednesday of the week for a Friday release is useful. The model is putting out a floor. Thanksgiving could be a nothingburger but business as usual is still a catastrophe by Christmas. Thanksgiving could produce a significant bump in transmission which moves up the no-intervention hospital overload by days or weeks. The model and its results gives a meaningful floor from which action can be taken to divert from the projected outcomes.
But right now, this is what “optimistic” modeling looks like in some states.