COVID-19: Is there a Second Wave?
There was a bit of a disturbance in the force this week as discussion of a second wave of COVID hit the news and the stock market. A resurgence of COVID-19 could lead to states re-closing after re-opening and thus six more weeks of quarantine. Perhaps we a ceremony involving a groundhog who sees or does not see his shadow to predict what’s going to happen.
But what does the data say? On a national level we appear to be experiencing an agonizingly slow decline in new daily cases and deaths. It’s hard to interpret the data about cases because there’s more testing going on than there was in the early days of the pandemic. Even the death data can be suspect, with certain early opening states motivated to keep the official statistics low.
The daily numbers show weekly peaks due to cases being underreported on weekends and then officials catching up during the work week. It’s better to look at the seven day moving average (orange) line. We can see that we actually had a double peak in April and another small uptick in late May.
In terms of fatalities the peak occurred in mid-April, followed by a plateau of a couple weeks but trends have been fairly consistently down since then.
The hospitalization data is more encouraging. After the mid-April peak in hospitalizations in mid-April all the curves are coming down fairly steadily. There looks to be some aberrations in the ICU data as well as the ventilator data around the beginning of May. Unfortunately many states aren’t reporting all three of these categories of hospitalization data so please consider these as incomplete.
In terms of tests we’ve finally reached the half a million per day rate. You may think that’s great (especially relative to March numbers) but that means 660 days to test the entire country.
So why are people talking about a second wave then? I believe that this is because we’re seeing a rotation in the set of states that are reporting cases. The chart below illustrates this fairly well:
The states that are seeing an increase in new cases are on the left whereas the ones showing a decrease are on the right. At the far right we can see the states that suffered the most in the early days of the pandemic: New York, New Jersey. At the left states that were relatively unaffected early on are showing large increases: Arizona, Florida, Texas. California is probably the most troublesome state. It was heavily impacted early on and is now seeing a resurgence. It really should be at the far right with the other early sufferers. Note that in the chart above, roughly 40% of the states are showing increases while roughly 40% are showing decreases.
In terms of deaths the situation is a bit better with more states showing larger decreases than the number of states which are seeing increasing fatalities.
Zooming in on California, we can see that while our early efforts at containment were good (much lower per capita infection and fatality rates than NY) we have yet to reach a peak in new cases and daily fatalities have been fairly constant for almost two months.
We can look at the data at the county level as well. We can see that the New York/New Jersey/Connecticut area is thankfully mostly green (cases going down) whereas the south-east is a sea of red. This particular map uses percentages instead of absolute numbers so some of those counties may be showing large percentage increases on very small numbers but the Appalachians look pretty bad, as does Florida, Texas, and Arizona. California is a mixed bag but the northern Sierras up into Oregon and eastern Washington are all showing increases.
As usual all this work is available on GitHub. Most of the analysis for this post comes from this notebook whereas the map was generated in this one.
