Thread by @ScienceVet2

So let's say you have pneumonia. It's not looking so hot.

Then you get terminal cancer. You've got a week to live.

Suddenly someone comes in and stabs you in the chest with a knife.

What do you want marked as the cause of death?

That's right: Bleeding from knife wound

Lucky for you, this is how it works! (mostly)

If you're dyING of something, but something else comes in, takes advantage of that, and kills you (you can't fight back against knife guy because of cancer and pneumonia) it's still the knife that killed you.

My dad has heart disease and diabetes. If someone breaks into his house and kneels on his chest until he dies...

The cause of death is lack of air caused by chest compression, even if his conditions make that easier.

COVID-19 is the same way. If proliferation of SARS-CoV-2 in your tissues causes your lungs to fail and you die, that's the primary cause of death, even if it was easier for the virus to kill you because you also had diabetes, or asthma, or what have you.

Now, actual death certificates are more complicated than just "cause of death." So if someone already has terminal cancer that's trying to kill them, that will be noted as a contributing factor (it made it easier for the murderer to stab you).

The biggest challenge is something like COVID-19 where there's a lot of things it can cause (stroke, organ failure, pneumonia, etc). Early on in an outbreak you may get UNDER reporting; marking "cause of death: stroke" not "cause of death: Covid-19 induced stroke"

Luckily, we can go back and get rough estimates of how close we are because things like stroke follow pretty predictable yearly cycles. So we can use statistics to look at total deaths from a cause and calculate "excess deaths"

These are the deaths in addition to what we'd predict. So, for example, you get pneumonia deaths in winter, but it's usually about X number. If we see 1000*X deaths from pneumonia, and then we discover COVID-19 causes pneumonia... we can roughly calculate how many we missed.

We're pretty good at this, actually. Especially on population sized groups. And the U.S. specifically has a large enough population to have sub populations we can use to back check if there's a number that's notable "off" from others. We can look at why.

I would be remiss, however, if after lauding this system I didn't note a failing. The notable place this cause of death thing fails is police involved deaths. Where suddenly having a heart condition is marked the reason you died, not having your air cut off.