Something Worth Looking At

I have been, frankly, too in despair to do a lot of data mining the past few days.  But today I sucked it up and downloaded my usuals and did the basics.

And, lo and behold, there is something new in the data.

Let's look at WA state first.  This is total deaths, daily deaths, and a 3 day mean on daily deaths:
Look at the right end of that graph (the bar graph with the grey 3 day mean line): daily deaths are monotonically down for five days from a local peak of 22 on the 28th of April to 4 yesterday (Sunday 3 May).  I have noted that there is an approximately weekly dip in the past, but this is the first time we've seen a monotonic decline over this period of time, and the first time we have seen a sub-5 death day since the 22nd of March.

We need to be cautious because the data remains noisy and there are many possible reasons why we might see a revision of recent death counts.  That said, we have reason here in WA state to be cautiously optimistic about deaths.

If we look at the graph of WA state cases, there is a similar (but not so lengthy) monotonic decline in cases from 10 April to 13 April, with low but not monotonically decreasing days extending until 16 April.  10 April to 28 April is 18 days or 2.5 weeks, and we can suggest a 2.5 week period from case identification to death.


Moving to the US daily deaths, take a look at the right end of the graph:
We've got a four day monotonic decline in daily national deaths starting on 29 April.  This decline is driven partly by the decline in daily deaths in the NY, NJ, CT cluster, where 5 day mean deaths have dropped from about 1300 on 17 April to about 700 on 3 May.  We have not seen a four day decline in deaths before this, and the last two day decline in deaths was 17 April to 19 April.

Does this mean we're ready to reopen?

No.  In fact, No.  Actually, NO!

But it does mean that we've demonstrated that mitigation via lockdown and social distancing works.  And it can work well enough to not just "flatten the curve" but to reduce both cases and deaths on a longer-term basis.  We have, temporarily, beaten the exponential.  These reductions mean that R(t) for COVID-19 has been temporarily reduced sufficiently to choke off the spread.

However, in WA state that cases minimum back in the first half of April did not last, and that probably means the deaths minimum will also not last.  Nor is the US deaths reduction adequate (the 3 day mean remains above 1500 deaths daily) to call COVID-19 remotely contained.

We've proved the tools at hand can solve the problem.  We just need to deploy them correctly, maintain them stringently, and stay the course.