a couple of quick things so far ...

(1) the primary literature is not published in "The BMJ" ... it's in BMJ Public Health ... these are both rigorous, but The BMJ along with JAMA are possibly the most-sought-after places to publish. If anyone states it's in "the BMJ" I would take anything after with a grain of salt. We had a Dean in an ancient university here (UK) that used to say "The BMJ" and it was BMJ Open and he didn't make it so long before being made redundant. Those small differences better when assessing competency.

(2) The raw data/variables employed can be accessed here: https://sls.lscs.ac.uk/variables/

(3) The 1991 Census ... which I believe is the starting data for the study has two interesting codes shown below: