a therapy conundrum

Alright, imagine you are a therapist and a guy comes in and tells you that he has depression. Yay depression, there are about a million treatment models that are proven to work for that.

Now imagine that you find out the guy is depressed partly because his girlfriend is physically abusing him. Suddenly you have a much bigger psychosocial issue; hell, the guy probably should have some sort of extreme emotional reaction to that because his situation is extreme. Is it just depression now? Is it self-esteem? Would he be better served by a case manager, or a cop, or a lawyer?

Now imagine that it turns out that the guy ends up going from one abusive relationship to another, always finding women to play out the same pattern. Kind of puts the ball back in a regular private therapist's court. Do you support him so that he feels stronger and more confident, or do you confront him about his pattern in an effort to get him to change? Both address relevant causal links in the chain of him getting hurt, but at slightly different levels. The second (deeper) one takes longer but without addressing it the guy will probably keep replaying the same issues in one form or another. The first one probably will go more quickly and solve a lot of the immediate problem (say, he leaves the current relationship and starts taking better care of himself) but there are still subterranean issues that could just as easily re-emerge. And what if there are even deeper layers nested within that?

These are all questions that I have no better idea how to address since I started my program, which is a little frustrating. Oh, I think I know what my supervisors would say, that it is more important to have a clear sense of the questions that it is to have a good answer for them, and that by making these things conscious it will allow you to address the individual client's concerns better. I remain unsatisfied. I'm at the part of the learning process where I know a crap-load about what I don't know without enough framework to really be useful yet. It's a crappy spot in the learning curve.