Tomorrow, New Caledonians are gonna get their 3rd attempt at independence; their 2nd under the current settlement. If it fails, they've only got one more shot at it in 2022, then probably another decade or two without a vote. https://www.forbes.com/sites/alexledsom/2020/10/02/this-rich-little-island-state-seeks-independence-but-risks-racial-upset/#48dcbd9e18cb

The article has an error though; flights are suspended until March 2021, not March 2022.

Death Day 2!

Let's talk cause of death.

I'm newer working with the cause of death data. So much as last week I corrected an error from the week before, this week I must correct another error.

I double counted all deaths in the by-cause file. I didn't realize it included all states individually AND the US in total. This has no effect on the relative balance between causes of death or time trends of any kind. But I wanted to share it in the interests of disclosure.

However, it does have an impact on the total number of by-cause excess deaths.

Combined with updated data and an improving handle on revision patterns by cause, rather than 475k natural causes, 375k respiratory, and 50k external causes, it's now 271k natural, of which 190k respiratory; 20k external causes.

Here's the graph of weekly excess deaths by cause. You can see that external causes deaths are rising, but excess "natural causes" deaths are still by far the biggest, and respiratory deaths are the biggest share of that.

Here it is simplified.

What you can see is excess deaths of respiratory causes are tightly correlated with excess deaths of other natural causes, which STRONGLY suggests that these are deaths "inevitably arising from the COVID pandemic."

External causes have a different trend.

Let me be clear:

External causes deaths ARE up, which could include rising suicides or drug ODs or homicides.

HOWEVER....

Excess "Likely COVID deaths" being classified as NOT COVID are MORE NUMEROUS than excess external causes deaths.

What I mean is: things that were actually COVID deaths *misclassified* as e.g. dementia deaths or heart disease are much bigger than whatever spike there may be in suicides or drug ODs.

But ya know what? Rising external causes deaths are still a big deal! Their lack of correlation with COVID deaths also suggests they are not an inevitable result of the pandemic, and in that sense represent an avoidable tragedy.

HOWEVER, they can't be avoided by ending lockdowns. Here's excess deaths by state vs. an index of Google mobility data. More negative numbers on the X axis means a more intense lockdown.

There's no correlation between lockdowns and higher external causes deaths.

Nor is this an economic thing. In fact, states with a bigger spike in unemployment over April to August have actually had LESS of an increase in external causes deaths. The big increases are in states that have NOT been economically hard hit!

The state with the biggest rise in unemployment, by the way, is.... Hawaii.

If we take a vector of lockdown intensity AND unemployment spike, again, a bigger "COVID economic impact" is not AT ALL associated with higher deaths from external causes.

It's still possible that external causes deaths are related to COVID somehow. Maybe even to lockdowns and economic losses. But it's not as direct and simple a relationship as you might think.

The evidence linking lockdowns or economic losses to external causes deaths is MUCH weaker than the evidence linking suicides and drug abuse to.... voting for Donald Trump.

Vermont is a weird one there. ANd this link may be totally spurious or non-causal! But really folks, we have to say right now we don't know for sure why exactly external causes deaths are rising.

ANYWAYS.... here's how excess deaths are stacking up in absolute and relative terms, now vs. the 1918 pandemic (which, recall, had its first wave in March/April 1918, not October as people often think).

I don't think we'll have a 1918-style spike. History never repeats. But it's worth understanding how bad things can get. So far <15% of Americans have been infected.

oh lol I reattached a previouis chart too, whoops!