In regards to interpretation of the numbers, could it be that we are (all) confused by what the numbers actually indicate? A friend told me about the Simpson's Paradox and how it may apply to C19 numbers, and sent me a post by a doctor which basically said (I'm paraphrasing the posts below):
In essence the USA is experiencing:
- increasing total positive cases
- increasing hospitalizations
- increasing deaths
For a time it appeared that hospitalizations were decreasing and total positive cases were flat. It would be reasonable to assume that means "young people are getting it now, not old people?", or "we've become better at treating it, so the death rate has fallen", or "we're testing more people, so we're seeing more cases". There is some truth to these conclusions since more young people getting it (for now), we have become a bit better at treating it, and we are testing more and finding more cases. But none of these conclusions explain the effect.
The C19 outbreak is not one outbreak spread evenly across the US; it is many outbreaks spread unevenly. We need to look at state and county data to understand the spread. As an example, look at AZ and TX where cases and deaths have been increasing for weeks. FL is similar (but their data is incomplete...on purpose, thank you FL). Look just at Miami and Houston, and it's worse.
This is where Simpson's paradox comes into play. "If you pool data without regard to the underlying causality, you'll get erroneous results". This doctor continues to say that in the coming weeks it will be horrible, as dozens of areas go through what happened in NYC earlier this year. ICUs will exceed capacity, ventilators rationed, and too many deaths.
A short (4.5 min) video on Simpson's paradox can be found at
For just one example, let's look at Phoenix.
Phoenix, AZ (Maricopa County)
Dundas BI made by Dundas Data Visualization
phdata.maricopa.gov
Positive Cases
(there is an eight day reporting delay)
Hospitalizations (21 day reporting delay, so the numbers are "likely to increase")
Deaths