Hi Laura,
I agree that the x-axis should be labeled. I quickly looked it up in the original publication now, the time period is from 1975 to 2005. I think this period of 30 years is certainly sufficient but I see your point. I do not think there is a hard time cutoff. But it might be most feasible to go with the census data every 10 years. I think it depends a lot on the natural course of the disease you are studying.
I am not sure about the connection you make between increase in population and decrease in mortality . Can you explain your point?
Thank you, Kerstin