Does the problem directly cause death, injury, or disability? If so, real problem. Everything else is just a matter of optimizing your level of comfort in the absence of real problems.

We can then talk about grading problem importance by its distance from death, injury, or disability (e.g. it's a problem in the solution to a problem in the solution to a real problem), and the probabilities and amounts of death, injury, or disability it does, can, or might in the future cause.