Following a conversation about this article on the Guardian, a friend made this observation: "I find reading that graph interesting, because of course in 2011 everyone was in the previous age group as in 2024. So to get a true view of what is happening you probably need to read across and down one. So the 60-64 2019 cohort have a big drop from 131 to 65 while the 75-79 2019 cohort jump from 56 to 93."
It's an interesting point. Probably there is something to say both by evolution of individual age group, and by looking at changes of habits within age group. Looking into this problem further, I believe that there is a general solution only where the age group are exactly the same number of years that separates the two data points.
For example, the data below is based on the UK Census 2011 and 2021, which are 10 years apart. It shows the population in the London Borough of Haringey for 10-year age groups (which is an elaboration of the official data, that is year by year).
2011 Data Table QS103EW
2021 Data Table TS007