by Calculated Risk on 5/25/2018 03:16:00 PM
Friday, May 25, 2018
From housing economist Tom Lawler: US Deaths Jumped in 2017
Provisional estimates from the National Center for Health Statistics (NCHS) indicate that the number of US deaths increased sharply last year, both in absolute terms and adjusted for age. According to the NCHS’ “mortality dashboard”, the “crude” US death rate (deaths per 100,000 of population) averaged 866.2 in 2017, up from 839.3 in 2016. The NCHS’ “age-adjusted” death rate (which adjusts for the changing age distribution of the population) for 2017 was 733.6, up from 728.8 in 2017 and the highest age-adjusted death rate since 2011. These data suggest that the total number of US deaths last year was around 2.821 million, compared to 2.744 million in 2016.
While data on deaths by age (or full-year deaths by cause) are not yet available, these data suggest that the recent alarming trend of significantly higher death rates among teenagers and non-elderly adults (shown in the table below) continued last year.
|US Death Rates (deaths per 100,000 population),|
Total and Selected Age Groups (NCHS)
What is especially striking about this table is the substantial increase in death rates for the 15-44 year old age groups from 2014 to 2016.
While full-year provisional estimates of deaths by cause are not yet available, here are some death rates by selected causes for the four-quarters ending in the second quarter.
|Death Rate (per 100,000 population) by Selected Cause|
|Four Quarter Period Ended:||Drug Overdose||Firearm Injury||Suicide|
With respect to drug overdose deaths, the “trending” quarterly data suggest that drug overdose deaths in 2017 were probably around 70,000. Here is a table showing recent history
|Drug Overdose Deaths by Selected Age Groups, NCHS|
For folks who rely on Census population projections to forecast other variables such as labor force growth, household growth, etc., these recent death statistics suggest that they should not do so. The latest Census long-term population projections not only did not reflect the increased death rates for certain age groups from 2014 to 2016, but also assumed that death rates for most age groups would decline from the assumed rates for 2017. E.g., here is a table showing the projected deaths in the Census population projections (released earlier this year) for the year ended 6/30/2017 compared to the NCHS data for deaths for calendar year 2016.
|Age Group||Census 2017 Projections|
2016 (Calendar Year)
Note that the latest Census population projections were dramatically too high for infant deaths (they made a mistake) and too high for the very elderly, but way too low for the 15-84 year old groups.
Here is another table comparing the death assumptions in the latest Census population projection to NCHS data.
|Census 2017 Projections|
12 month period ended 6/30
|NCHS Calendar Year|
If death rates by age were to move back down to 2016 levels from 2018 to 2020, then deaths from 2018 to 2020 would cumulatively be about 260,000 higher than those in the Census’ latest projection. However, deaths of 15-84 year olds would be whopping 496,000 higher over this three year period than what is shown in the Census’ latest projections, while deaths of 85 year olds and older would be about 180,000 fewer. These are actually fairly large number.
Net, the latest provisional data on US deaths is bad news from a societal perspective. They also indicate that the latest Census population projections are of limited usefulness, and analysts relying on population projection to forecast other key variables must actually produce their own population projections based on reasonable assessments of population growth’s key drivers – births, deaths, and net international migration – the latter of which is an even thornier issue, which I won’t touch on today.
Sometime in the near future I will produce updated population projections by age incorporating the recently release “Vintage 2017” estimates, more realistic death assumptions, and different scenarios for net international migration.