At the population level, there will be FOUR TYPES of #COVID19 #mortality:

1) Direct-direct
2) Direct-indirect
3) Indirect
4) Competing risks

In this short thread, I will discuss the four.
1) Direct-direct

Basically, these are the deaths that are being counted in real time (on your dashboards, etc.) and attributed to #COVID.

These are deaths, typically in hospital, typically with a positive test. Might be coded as underlying cause of death ICD-10 J22...
Direct-direct (cont'd)

These are, basically: "This person died of COVID"

This category (D-d) is what everyone (dashboards, the press, etc.) is counting, and it's essentially a tally.

Many of these deaths will also be coded as U07.1:…
2) Direct-indirect

These are early COVID deaths that were mistaken for influenza, and later COVD deaths that are somehow not recorded as such.

These will be coded things like: ICD-10 J11, J18, etc.

These *do* *not* show up in your dashboards or in the press.
Direct-indirect, cont'd

These are NOT simply a tally and will have to be inferred from seeing excess deaths in J11, J18, etc., relative to expectations.

There are relatively non-controversial models to calculate expected, and, hence, excess, mortality, but it's not a tally.
3) Indirect.

This is — for example — someone who dies of a broken femur (ICD 10 S72 in this example) because this injury was not treated in the emergency department altho under normal circumstances it would not cause death in most cases.
Indirect, cont'd.

This mortality is *not* *counted* in the dashboards, etc.

It is not a tally and will have to be inferred afterwards from models, as with Direct-indirect.

This is one of the huge "X-factors", especially as regards mortality at ages below 60.
4) Competing risks.

This isn't reallly a category of deaths, per se, but is highly relevant.

This refers to the fact that — for example — someone who dies of #COVID now may have been otherwise destined to die of a heart attack in a July heat wave...
Competing risks, cont'd.

Competing risks means that (e.g.) heart disease deaths could — in principle, anyways — *decline* due to #COVID2019

But it's a decline for the wrong reasons. We can only die once, +some COVID deaths will have been borrowed, so to say, from other causes.
Competing risks, cont'd.

This is another "X-factor". We don't really know how much this will matter.

Demographers will pore over the data (ex-post) and use multiple-decrement life tables to understand this phenomenon....
Stay tuned... but it will be a while before we understand all the dimentions of #COVID19 mortality. /end.
Addendum 1: There are health effects of recessions. Per past research, these often — counter-intuitively — reduce mortality, on net.

This will be difficult to disentangle, except for the period in which the recession out-lasts the epidemic.
Addendum 2: Scarring.

It's too early to say if COVID will see this kind of effect, in which survivors of infection have durable health problems. We will have to monitor for years...
Scarring, cont'd

There is an hypothesis that 1918 flu caused cardiac scarring, which in turn generated the 20th century heart disease mortality epidemic in the decades thereafter.

Full disclosure: I have argued that for 1918, this isn't correct. Link:…

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Andrew Noymer

Andrew Noymer Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @AndrewNoymer

Aug 24, 2021
Let's talk Junes. Short thread.

June 2020. Pandemic was still new and chaotic. Here in California, our first big wave was brewing (would peak in July). We didn't have the devastaion in March/April that NY etc. had had.

We were hoping for seasonality, but no... summer wave.

So, in June 2020, we didn't know that much. There were diverging opinions about whether the fall would bring more and when the wave(s) would peak.

June 2021. California has low community transmission; very low. Statewide color-coded tier system is scrapped.

This past June, we ditched masks, "opened up the economy". Etc., etc.

Well, good vibes didn't last long. Here we are.

June 2022. I think by then we'll have sense of where we stand.Boosters, winter waves, variants, natural immunity, we will have better persepective.
Read 6 tweets
Apr 29, 2021
Thread on why mRNA vaccines may be giving better immunity vs. SARS-CoV-2 than surviving natural infection.

disclaimer: we still don't understand the long-term correctness of the claim that vaccines are better protection than natural infection.

TL;DR: it's different kind of vax
In this thread I will offer some speculation as to why it's scientifically plausible that the vaccines offer better protection than surviving natural infection.

This is what it is, speculation.

As I said in the parent tweet, it remains to be seen how true this phenomenon is.

But if it is true, I don't think it defies logic. And here I will explain why.
Read 15 tweets
Mar 28, 2021
Regional covid epidemiology in the US of A.

A short 🧵.

Living in California, I have been increasingly optimisitc of late. Pic related.

... continued

But, New York and New Jersey, OTOH, are giving me the heeby jeebies...

... my thinking *before* this pandemic was that the next pandemic would see rapid spread, leading to regions being in phase with one another. Ex., there has not been a ground-stop of aviation.

And it's not just the northeast. Here's Michigan:

Clearly, the US of A remains a country with epidemics, plural, playing out at least with different timing in different regions.

Read 6 tweets
Mar 28, 2021


Covid deaths per million residents; minimum 100 covid deaths:

Imperial 3,916
Los Angeles 2,284
San Bernardino 1,815
Stanislaus 1,804
Tulare 1,745
Riverside 1,733
San Joaquin 1,661
Fresno 1,610
Kings 1,592
Merced 1,580

California counties, covid deaths per million population


Orange 1,479
Madera 1,465
Kern 1,350
Ventura 1,139
Shasta 1,122
Sutter 1,074
San Diego 1,059
Sacramento 1,038
Santa Clara 998
Santa Barbara 983

California counties, covid deaths per million population


San Luis Obispo 898
Yolo 889
Marin 851
Alameda 837
Butte 826
Monterey 781
Santa Cruz 733
San Mateo 711
Placer 669
Contra Costa 662
Sonoma 620
El Dorado 561
San Francisco 534
Solano 428.
Read 4 tweets
Feb 10, 2021

Counties. Covid-19 deaths per M population (minimum 100 deaths):

Imperial 3,151
Los Angeles 1,817
Stanislaus 1,590
Tulare 1,419
Riverside 1,393
Merced 1,336
Fresno 1,265
San Joaquin 1,234
Madera 1,199
Kings 1,169
Orange 1,072

California. Counties, cont'd

San Bernardino 1,006
Shasta 911
Sacramento 877
San Diego 853
Santa Clara 813
Ventura 803
Santa Barbara 779
Yolo 771
Kern 740
Marin 697
Monterey 668
Butte 640
Alameda 631
San Luis Obispo 627
Santa Cruz 583
San Mateo 581

California. Counties, cont'd

Sonoma 554
Placer 549
Contra Costa 499
San Francisco 392
Solano 313.

These 32 counties, with at least 100 deaths per county, account for 98% of recorded Covid-19 mortality in the pandemic to date.

Read 4 tweets
Dec 13, 2020
Remember this 👇🏻 chart?

I received request for a breakdown by age groups.

I forget who it was; sorry.

Well, ask and ye shall (sometimes) receive.

*THREAD*: All-cause mortality, weeks 1 thru 35 (early Sept). 2015–20, BY AGE, w/trend-line and 95% prediction interval.
Here is ages 0–24 (L) and 25–44 (R).

Deaths and 95% prediction interval. Input data from @NCHStats.
Here is ages 45–64 (L) and 65–74 (R).

Deaths and 95% prediction interval. Input data from @NCHStats.
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!


0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy


3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!