Best way to code a variable in a hierarchical (multi-level) regression?

#1
The outcome is binary, and I have individual-level data and geographical-level data. It's the latter I need help with.

Here's the best example: imagine the geographical unit is county, and a variable I want to account for is the sanitation grades of fast food restaurants. I care about both the number of restaurants in each county, as well as the grade each of them got.

I thought of coding it as % of restaurants in the county with each grade with missing values if no restaurants, but I'm not sure that is the most appropriate way? Any thoughts/suggestions would be greatly appreciated!