I have thought about it. It might work in this particular case, but we have around 50 different systems with somtimes 20 different checks for one particular system - that was why I put my hope into dependencies.
I don't think it make sense having a service reporting itself as OK when it's dependencies have failed. If there was an option on dependency definitions on how to handle status for the dependent service I think that would have been really elegant in this case.
I also don't want to put more preassure on the system with redundant checks, since we already have 6000 checks and 700 hosts leading to occacional latency issues.
I'm more thinking of writing a event handler for the individual checks, that will report status changes to a database, where I can have the overall checks look for the combined status for the different group of checks.