Key dataframe needs to be available across all work, makes sense to have it created early on rather than part way through manuscript.
Adds some better summary tables of missing data (there is actually quite a lot).
Tried to add a visualisaiton of missing data using visdat::vis_miss() but it prevents the document from rendering correctly. Image can be seen in this issue though.