I'm trying to do do a logistic regression using R.
I want to regress using binary dependent variable Y as granted or not grated (1 or 0) using the model in the following way (just showing few covariates here):

glm(Y ~ age + score + wage + employmenttype + co_person + co_score + ..., family = binomial)

Now here, "co_person" is a dummy variable and "co_score" has a value only if "co_person" = 1. Otherwise in my raw-data my "co_schore" has value "NULL".
For example like this:

co_person, co_score
1 __________ 2.3
1 __________ 5.2
0 __________ "NULL"
0 __________ "NULL"
1 __________ 8.0

As you see, the co_score for co_person = 1 is NOT 0 since when there is no co_person, there is no co_score so there is a missing value that is a "hole" as "NULL".
Now I wonder, how could I intreprate this when creating my model. How should my regression formula look like? Should I use interactions or some other method? Or is there some smart algorithm in R that can do it? Or should I simply change my "NULL"'s to 0? But that dosen't feel right since then I am giving a value that does not exist for co_score in question.