One of the weaknesses in our public policy is our tendency to grasp for hope. In the last month, the virus has really exploited this as people who want to believe letting the virus “wash over us” can happen without mass deaths. Republicans have been claiming that we can grow cases with minimal deaths, and that cases are just a matter of “more testing.
This week it became clear neither are true.
The looming tragedy of Texas
Texas provides a great example. Until recently, one could look at the below charts and feel safe that while cases were spiking as we gave up on social distancing, deaths weren’t.
One possibility was that we had deaths under control as we protect senior centers. Another is the virus is becoming less deadly.
It’s always been apparent that there’s a sneakier cause – lags. Let’s look at Texas.
- Cases began to spike on June 13.
- Hospitalizations on June 18
- Deaths on July 5
So for Texas, it took only 5 days for people to start being admitted, but 3 weeks for them to begin to die.
What’s happening to the “death rates” as this is going on? It’s collapsing of course – not because fewer people are dying – or will die – but just because you’ve got a huge flood of people getting sick, many of whom will die in 3 weeks, but they haven’t died yet.
Our death rate will thus be much lower while the epidemic goes exponential because the exponential growth of cases started earlier. It’s very simply a measurement/timing artifact and a misleading measure. Europe’s death rate will be much higher because they’ve got the epidemic under control, cases aren’t growing, we’ll see the “true” death rate which will eventually play out for us.
Another sign of trouble is the percent of tests coming back positive. If this percent is high, it means the epidemic is outstripping testing, not the other way around. Texas is at 20% of tests positive – an alarmingly high number.
Looking at another state, Arizona:
- Cases spike June 1
- Hospitalizations June 5
- Deaths June 30
So they had a longer lag to deaths (4 weeks), but eventually deaths caught up and started to spike.
30% of Arizona’s tests are coming back positive, so the epidemic is outstripping their ability to test.
- Cases spiked on June 15
- Hospitalizations on June 22
- The reaper arrives July 9
So, 3 weeks to deaths.
A bright spot – California’s tests are only 8% positive, implying their growth may be more driven by testing. This may explain the modest increase in deaths – but we’ll see.
- Cases on June 6
- they’ve been hiding hospitalizations
- deaths spiked July 8.
4 weeks to deaths. And Florida’s tests are 20% positive – so it’s not testing.
So, let’s look at Washington.
We have sadly ben spiking cases, starting June 23 in earnest.
Hospitalizations are up, shortly thereafter (June 28)
When will deaths come? assuming a 3-4 week lag, expect them July 21-28. But it’s hard to say for sure, because our cases do mirror testing to some extent. And our % positive tests is still only at 7%.
By Special Request, Indiana…
I don’t know what they’re doing but Indiana is in pretty solid shape. Their cases are heading up again – modestly – and 8% of tests are coming back positive. If I were the governor I’d be looking at this as a leading indicator that things are too open, but maybe only a little. I’d be ordering a mask mandate *on businesses* and hoping that’s enough. And keeping bars and theaters closed, for gods sake.
What do I conclude?
- Death rates are a really deceptive measure to hang your hat on.
- It’s possible under this there’s also a decrease in death rate as we protect old people better – but the deaths will come (and are coming). How bad can it get? Assume that as long as cases are spiking, deaths/day will grow along with them (with a lag).
- Clearly, “testing” is not causing these hospitalizations and deaths in the spiking states.
Politicians and hopeful people citing these numbers are increasingly engaging in deceptive numerology. The only real solution to this is to get the epidemic under control as fast as possible. And remember – today’s numbers represent our behavior a month ago – 30 days of these growth rates are utterly baked in, so if we start behaving today, the growth in deaths will continue for a month.
And in Florida, Texas, and Arizona, they’re not even starting to behave yet.
If you live in a state that is behaving, thank your Governor.
So, I agree (not surprisingly) with everything you are saying. But I actually think things are worse than you make it out to be. In Florida, surplus deaths have been twice the COVID deaths. In Arizona, test results are coming back at least a week late, so someone dying at home without a COVID diagnosis will take a week to show up on the death list, if they even get tested at all. Also in Arizona, if you look at their department of health dashboard, you can see a giant lag in deaths (far fewer deaths than expected in the last week — the deaths for a given day don’t show up on their dashboard until a week later.) Arizona also has a weirdly low ratio of hospitalizations to deaths (not enough testing?). But in Arizona, they’re now bringing in refrigerated trucks because the morgues are overflowing. I think the overflowing morgues tell you more than any statistics. In Texas, the number of people dead on arrival at hospitals (going critical at home, calling an ambulance, and not making it) has spiked.
So, most of what we are seeing is what you just said – lag. But I think there’s also a significant second factor: as infections jump to the point where they overwhelm testing, we start to miss cases, and the test lag also increases.
The visual analysis would be more complete and provide full perspective if the satrting point for the graphs were February.
Your wish is my command. Swapped them all to full range starting with first case.
(Sigh, third try to leave this comment. I hate WordPress.)
Thanks for these! Any chance I could talk you into adding Indiana?
> If you live in a state that is behaving, thank your Governor.
I’m more inclined to thank my city (Bloomington) and my employer (Indiana University), which have been science-based and proactive.
Wow, thanks! You’re giving me hope for my state of residence 😉
And just to end with a cheap joke: Elsewhere on the interweb, you said your birthday was coming up and you wanted a new transmission. What, fomite isn’t enough?