If it truly is panic disorder (and actual panic attacks) then for most patients their first panic attack tends to be seared into their memory. Failing that, an approximate date/epoch of onset should definitely be doable for the interviewee. At least nailing down 'did you start having panic attacks in college (20 years ago) or last week' seems reasonable. However, in VA settings, my experience with trying to nail down timelines regarding onset 99.9% of the time devolve into the (unhelpful) dichotomy of either (a) during/at the end of my military service or (b) I have no clue when in my lifetime any of this started.
I have started leaning heavily on unspecified diagnoses toward the beginning of knowing patients these days (first 1-4 clinical encounters) when this is called for. I'm not averse to using unspecified anxiety disorder and/or unspecified depressive disorder. At the end of the day (at the clinical level) these are merely working clinical hypotheses (under the increasingly beleaguered 'latent disease model' of mental illness). Unfortunately, the reality in VA settings is that, once given, a PTSD diagnosis is viewed as permanent since nobody will be willing to 'touch it' or alter it given the high likelihood of everyone losing their minds and creating Hell for the person daring to reconsider that particular diagnostic determination.