A weird finding is that, whether you look at raw counts or rates, the year of 2018 is almost exactly (less than 1% away from) the midpoint of 2019 and 2020.
With an error of less than 1%, it suggests that perhaps an equivalent proportion of deaths were "saved up" in 2019, only to then be "unloaded into" 2020. If you attempted this, then you'd use 2018 as your best estimate, your "set-point", as to how much to shift each time.
The share of all obituaries using one of the two words went down in 2023, which would be a good sign, except it is still up to 55% higher than each of the previous 7 years -- even including 2020.
2019 is too low
A weird finding is that, whether you look at raw counts or rates, the year of 2018 is almost exactly (less than 1% away from) the midpoint of 2019 and 2020.
With an error of less than 1%, it suggests that perhaps an equivalent proportion of deaths were "saved up" in 2019, only to then be "unloaded into" 2020. If you attempted this, then you'd use 2018 as your best estimate, your "set-point", as to how much to shift each time.
Will they ever go down?
The share of all obituaries using one of the two words went down in 2023, which would be a good sign, except it is still up to 55% higher than each of the previous 7 years -- even including 2020.