OK class, pencils down, eyes up front. Today we're going to explain excess deaths again since the @nytimes didn't do their jobs in a recent story that failed to explain an estimate of excess deaths they're using.

1/2
A few days ago, the NYT quoted the CDC "estimate" of over 200,000 excess deaths through July 25, which is the number that accounts for reporting lags. This is 54,000 higher than the official count, they say, than the 146,000+ as of when the research began.

2/3
Excess deaths is a measure put forth by the CDC to illustrate how many total deaths in US are occurring on a weekly basis above normal levels. They give a lower threshold which is merely a 5 year average of that calendar week and an upper level, 2 standard deviations higher.

3/4
There is no right or wrong answer as to whether to use average or upper bound threshold as excess, but because of lack of adjustment for population growth and how misleading using average can be for excess, the CDC generally uses the upper threshold as most should (IMHO)

4/5
What the NYT did not explain is the CDC is not predicting 200,000 excess deaths through July 25; it gives a range of possibilities that currently spans from 156-212,000. This is a model that includes the difference as to using "average" as your threshold or upper bound

5/6
In other words, the model is saying the estimate could be as low as 156,000 or as high as 212,000 depending on which "excess" threshold you use and also taking into account a variance in modeling how many more "excess deaths" will be added to the count with lag.

6/7
So finally, when the NYT says CDC estimates 200,000 excess deaths, that's not true. The CDC is saying the worst case scenario is over 200,000 excess deaths and that's if you use the much lower "average" count of weekly deaths as the threshold.

7/8
Last point to remember, this takes into account all deaths occurring including those resulting from our stay at home orders. Excess deaths will be meaningful to tell us how many died *from* Covid-19, but not until we have extensive lockdown calculations done. This article failed.
You can follow @kylamb8.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled:

By continuing to use the site, you are consenting to the use of cookies as explained in our Cookie Policy to improve your experience.