I looked through their scientific paper and couldn't find any data about the demographics of the declined patients. The only information I can find is the classification of their groups, which seems to exclude the two classes I mentioned above.

If we assume that the majority of "non HCQ" patients were of group (e), young people, then their rate of mortality in this paper would be at least an order of magnitude higher than that of any previous paper, therefor I would expect group (d), old people with complications, to be more likely to have higher representation in their "non HCQ" group.

I didn't conduct this study so I don't have access to the raw data not provided in the paper.