Well, you perform a safety assessment. In the context of Axis II pathology, you want to explore why they were cutting and what the cutting did for them. Superficial SIB is typically a poor way of coping with psychological distress. If their wounds don't require sutures and they say something to the effect that the physical pain makes it easier to cope with the psychological pain then it may sound like it's low/lower acuity, but you still need to assess other suicidal risk factors- particularly access to lethal means and planning/intent. DBT is great- IF you can find someone who can work your patient in. That's the main trouble I've had with DBT- is finding therapists who can work someone in on short notice. CBT isn't a bad substitute. Medication needs to be used very judiciously in this population, with SSRI/SNRIs deserving primary considerations. Use caution with TCAs for insomnia, but if you do use them, then choose Doxepin 10mg or either dosing of Silenor- but don't Rx for 90 day supplies (the idea being if they OD on #30 Doxepin the total amount ingested is the upper end of the normal dosing range for antidepressant therapy). You could try Elavil 10mg qhs, but Doxepin is much more antihistaminergic at equivalent doses. Being medically defensive and referring them for inpt hospitalization can very possibly cause them to decompensate further. Some might argue benzos aren't unreasonable because of their high therapeutic index but I find comorbid substance abuse runs high in this group which means you can throw the therapeutic index right out of the window. But always remember that ~8% borderline patients actually complete suicide- so don't be outright dismissive. Close follow-ups and adhering to clearly established therapeutic boundaries are vital.
Oh yeah, document thoroughly, but that goes without saying...