Using data from the NCHS/CDC mortality surveillance system, I get the following
2018*: 2,831,836
2019*: 2,845,796
In 2020, official reporting in the NCHS mortality surveillance system extends through September 19. I have no idea how you would say "through 9/24", when deaths are only officially reported through 9/19, to my knowledge.
The count through 9/19 is 2,267,854. However, there is lag that means the last 3 or so are too low and will continue rising. However, to be conservative, I'll just use the average of 2018/2019 for those weeks and re-calculate. That number becomes 2,314,260 (over 38 CDC weeks ending in 2020)
If we were to extrapolate that to a year, we would actually get 3,166,882.
So we can say she's wrong.
However, extrapolating that to a full year isn't the right thing to do. Excess deaths are down to about 500/day (which is also about where CDC-reported deaths/day are) as of 9/12 (too much error in trying to estimate that 9/19 week so I don't use that data in my calculations). If we do not see a second wave, that excess death number will likely continue dropping and could go negative if pull forward deaths (older people that would have died this fall but died this spring) exceed COVID + lockdown related deaths. Lockdown deaths (particularly in patients with alzheimers) continue to climb so it will be interesting to watch. Cancer deaths not showing a strong trend up yet, and in fact are trending a little below average I think.
Point being, too many unknowns from here to take that last 38 weeks and extrapolate those to the next 14. We've had about 275,000 excess deaths through 9/12 by my calculations. That number will continue going up for a while...but could drop a bit if pull forward deaths become significant.
*I used the 52 CDC weeks ending in the year of interest to calculate that year's deaths. It will contain a few days from the previous year, but also doesn't contain a few days at the end of the year, depending on what day of the week Jan 1 falls on.
