it occurs to me that continued or increased testing may be counting more cases in healthier people who are not going to die.
BTW, I think this is likely the dominant factor, with the populations infected being a secondary factor.
If we say that new cases have been roughly flat for the last four weeks (approximately true)....
Test positivity has been dropping, while the number of tests has been increasing from around 300-350k in mid-May to 400k+ per day now.
This means we are indeed identifying more of the "healthier" cases - ones we did not find before. If you were to plot the actual infections vs. time, it's been nowhere near flat the last four weeks - the flatness results from the flattening effect of the amount of testing we've been doing. If we had scaled testing even more rapidly, we might have even seen tests increase, while simultaneously infections were dropping.
So even if we assume a constant IFR, the above ensures that the deaths, which are lagging, will also continue to drop. Because deaths are proportional to infections, not cases.
This has been discussed here previously.
Of course and that seems like the most likely explanation for why daily deaths have fallen while daily case numbers have remained constant over the last couple months.
I think it's actually more that the scaling of testing has increased, but certainly as discussed earlier, there are likely more young people getting infected overall, too, which helps.
A very good summary report from the CDC today. Gives some data on death rates in children, and in other demographics. Just remember when reading this report that the actual number of
infections is likely closer to 13 million (at the end of the report data window May 31st - might be closer to 15 million now). This is a guess, but most experts and serology results put it in that ballpark (+/-20%).
https://www.cdc.gov/mmwr/volumes/69/wr/pdfs/mm6924e2-H.pdf
Someone I follow (Dr. Eric Topol) summarized the mortality in children, which might be of interest to most people. N(%) is the number of cases. Hospitalizations are not tabulated here. Again, this is out of ~13 million total infections, and if we assume we can uniformly apply the % (may not be true as younger infections may be less likely to be detected (more frequently minor))...that would mean 0.037*13e6 = 480k infections in 9-19 years, and 195k infections from 0-9 years.
Out of that, 46 children died, 6 of whom had no underlying conditions. So about 0.007% mortality rate (including underlying conditions). About 0.05% ICU admission rate. And probably about 0.25% hospitalization rate.
We knew this of course - deaths in this age bracket are thankfully very rare, though the approximate hospitalization rate is still troublingly high.