Dinah posted about How Psychiatrists Select Antidepressants, which was a very thoughtful and concise description of the factors we take into consideration. Supremacy Claus commented on Dinah's pragmatic, plain-speaking distillation (talk about plain-speaking pragmatism, check out this legal eagle's excellent blog). Dr Smak (another great blog... she and Dinah should go shoe-shopping together) was surprised to find no mysterious revelations, and "The" Shrink (a great new psychiatrist blog... welcome!) felt doctor preference was a missing element.
I started a comment, but it got so long and non-plain-speaking (sorry, S. Claus) that I moved it here.
Shrink, not so sure about the physician preference part (or maybe I am atypical... ha). I don't have a "favorite" or fall back antidepressant, as I find that when I apply Dinah's list (which is quite comprehensive and a good list), I am usually left with one or a couple drugs, and can still find a reason to pick one over the other (eg, cost). I feel I am quite familiar with the zillions (ok, maybe it's only 15 or 20) and ready to pick whichever seems best.
I do think Dinah's #4, 5, and 6 should be expanded on, and #4 should be split into 2 separate sections (I'm a splitter)... #4a being Other Medical Issues (eg, Seizure -x-> Wellbutrin; Psoriasis -x-> Lithium; Hypertension -x-> Effexor; etc) and #4b being Drug Interactions.
Drug Interactions is a whole 'nother post, and is a BIG factor for me when prescribing. Many of my pts are on multiple meds, so it becomes really important to think about this. Prozac and Paxil, for example, are famous for 2D6 interactions, so I avoid it when folks are on drugs which are solely metabolized by that enzyme. Luvox is a great hs antidepressant, but will muck with 1A2-metabolized drugs. Serzone and 3A4 drugs (though, haven't seen Serzone in years now... too bad, was a great drug to have around, esp if you knew how to use it... great for blocking SSRI-induced sexual side effects).
#5: Target Sx - When I think thru these, I think in terms of receptors (may be a tomato-tomahto thing here). I hear "no appetite" and think "I want histamine antagonism"; I hear "can't concentrate" and think "dopamine agonism"; I hear "no energy" and think "norepinephrine".
#6: Side Effect Profile - This is the one I spend the most time on with a pt. For any given side effect that is either desireable (sleepy, energizing, stimulates appetite, reduces appetite, etc) or undesireable (weight gain, wt loss, sexual, rash, seizure, nausea, etc), I have a pecking order in my head of drugs and their propensity to cause -- or not cause -- the particular side effect (other term is "adverse reaction", though they are only "adverse" when undesireable). The above 3 sections are where a better understanding of psychopharmacogenetics would come in handy.
The above may be where Dr Smak noted the perceived "secret way" in which shrinks pick 'em. What may be different in the way in which psychiatrists and PCPs select antidepressants is just in the way these thought processes get all merged together, or maybe thought about in an explicit way (my receptor-tomahto approach) or an implicit or nonverbal way (Dinah's best guess-tomato approach).
This "gut feeling" about which drug to use is merely the end result of a massive probability calculation which is automatically performed in the brain, based on all the above input about which side effects or target symptoms should take precedence for that specific pt, which drugs are more or less likely to deliver them based on receptor and metabolic profiles and based on literature and personal experience, in addition to the other factors like likelihood for compliance and affordability, all boiled down to a single "I think you should try Effexor". Much of that calculation is not conscious, and I think Dinah belies the complexity under the surface by simply (but honestly) stating "my best guess".