A reporter pointed me to this article, Life Expectancy and Mortality Rates in the United States, 1959-2017, by Steven Woolf and Heidi Schoomaker, and asked:

Are the findings new? Can you subdivide data, like looking at small populations like middle aged people in Wyoming and have validity? Can you make valid inferences about causes and effects? And why aren’t children and older people suffering an increase in mortality?

My reply:

This link from a couple of years ago might help.

The short answers to your questions are:

1. Mortality trends vary a lot by age as well as geography, so it makes sense to look at different age groups separately.

2. Causes of deaths are much different for children, middle-aged people, and old people—so it makes sense to see different trends in different age categories.

3. Sample sizes are large enough that you can look at individual states (as you can see from the above link).