>This change wasn’t adopted at a national level in a single moment; that would have led to a single step-wise change in mortality rates. Instead, the measuremen

>This change wasn’t adopted at a national level in a single moment; that would have led to a single step-wise change in mortality rates. Instead, the measurement change was adopted state by state, which led to a gradual rise over 14 years. You know what else would also lead to a "gradual rise over 14 years"? Actual worsening poverty conditions and access to healthcare. If it's due to measurement change, a state by state adoption of new measurements indeed would not lead to a single step-wise change. But it would lead to multiple step-wise changes (the step points being after each new set of adopting states publishes their new numbers, especially for the larger states). Those would be hard to tell from a gradual organic change, and it would rather mask any gradual organic change (e.g. a 15% percent "organic" increase that would have been there even without new measurement applied, would be masked by a 80% measurement-change-related increase). So, it's not so clear whether there was a decline in actual numbers or not. Just that more deaths that were missed are now measured. And their plot conveniently stops at 2017 - if it didn't we'd have 6 years where the "measurement change" doesn't apply (as it's the year all states have already adopted the new measurement), and we could see if there's still an increase.