They talk about correlation, not causation. And the data is from surveys, not actual measurements:
Researchers analyzed data from all 3,143 U.S. counties over six years, from 2019 to 2025, comparing how much sleep people reported getting with average life expectancy in each county. [Emphasis mine]
So saying that you sleep poorly is a predictive measure, but things that cause shorter life spans probably just screw up sleep (or at least ones perception of it) as well. Translating a correlation into individual advice is disingenuous at best.