Selection of Antidepressants, pt. 2

(Part One is href="">here)
Now we get to the heart of the matter: how does one decide what
antidepressant to prescribe? 

First, let's take a look at the factors that are considered relevant to
the choice.  If you understand what factors are relevant, and why
they are relevant, then you will start to see how the decision is

In the process of deciding which drug to prescribe, there
are two bodies of information.  Well, actually there are more, but
keep things simple, I am going to artificially conflate some of
The two bodies of information are 1) patient-specific information; and
2) general knowledge within the field of psychiatry. 

Patient-specific information includes, in no particular order:

  • Onset  of illness (when in life did it start, was it gradual
    or sudden, what
    was it like when it started?)
  • Precipitants (what was happening in the patients life when it
    started, that may have contributed to the onset?)
  • Premorbid stressors (what happened earlier in the patient's life,
    that may have made it more likely that the illness would occur?)
  • Premorbid functioning (how well was the patient functioning
    before the depression occurred for the first time?)
  • Course over time (has it been persistent, progressive,
    waxing/waning, or intermittent with full remission between episodes?)
  • Symptoms (what does the patient report is happening that is not
    supposed to be happening,
    and what is not happening that is supposed to be happening?)
  • Signs (what additional information is gained in response to
    specific inquiries?)
  • Degree of impairment (divided, somewhat arbitrarily, into
    personal, social and academic/occupational domains)
  • Modifying factors (what makes it better, what makes it worse, on
    an acute level)
  • Sustaining factors (related to modifying factors; what keeps the
    patient from recovering spontaneously, on a chronic level)
  • Family history (are there any family members with the same, or
    similar, conditions)
  • Comorbidities (what other conditions are present in the same
  • Test results

Information that is general in nature, that is, it has nothing to do
with the specific patient, includes the following:

  • Original research, published in journals and/or presented at
    conferences or via personal communication
  • Works synthesized or derived from original research, such as
    meta-analyses, reviews, and textbooks
  • Expert opinions, from editorials, consensus guidelines, or
    personal communication
  • The prescriber's base of training and experience

Not all of this information will be important, at least when it comes
to the first prescription.  But it's always good to start with a
fairly complete data set, if only to feel somewhat confident that the
diagnosis is correct and that the treatment plan is reasonable.

By far, the single most important factor is the patient's prior history
with antidepressant medication.  If the patient has never taken an
antidepressant, this does not count, and the decision is based upon
secondary factors.  But for patients with a prior history, it is
that history that takes precedence over everything else.

Let's start with the simplest scenario.  As a rule of thumb, if
someone took something before (drug A), and it
worked well, and did not cause unacceptable adverse effects, then the
decision tree stops.  They should be prescribed the same thing
worked before. 

Naturally, there are times when it is not that simple.  What are
some exceptions?  It could be that there has been an insurance
change, and now drug A is not covered.  You might want to choose
drug B in that case.  Or, now the patient is taking a drug for
something else, say, hypertension.  Drug A interacts with the
antihypertensive.  That force a reconsideration, although it may
not lead to a change in the recommendation.  Or, now the patient
has a second diagnosis -- either medical or psychiatric -- that was not
present before.  The second diagnosis could influence the
choice.  Another possible factor is that of the patient's
preference.  Yet another is the possibility that new research may
have been published in the interim, which could lead one to reconsider
the original choice.

There is one more possible complicating factor to consider.  What
if drug A worked
well, caused few if any adverse effects, but eventually stopped
working (despite increasing to the maximum dose for a sufficient period
of time)?  I don't think there is a good body of evidence to guide
us in that case.  Probably most psychiatrists would recommend
trying something else (drug B).  In the absence of empirical data,
the fallback position is to use an expert consensus.  Sometimes
there are published expert consensus documents, sometimes there are
not.  If there are not, then it is necessary to employ subjective
judgment.  As an aside, the href="">APA
and the VA have
both produced expert guidelines.  I don't know of any way to
determine if one guideline is better than the other.  In any case,
most experts would not prescribe something again, even if it originally
worked well, given that it eventually stopped working

The absolute simplest scenario is the one described above, in which a
patient previously took a drug, and did well with it.  The next
simplest scenario occurs when a patient has never had any prior
experience with any kind of psychiatric medication.   In that
case, you can pick any antidepressant.  Several references point
out, that the selection of a first antidepressant is based mostly upon
the desire to avoid adverse effects. 

In practice, the factors are these: tolerability, patient acceptance,
and cost.  The rationale for putting tolerability first is this:
the drug will not work is the patient does not take it, so it makes
sense to start with something that has, on average, a low adverse
effect burden.  Although it is possible to get into a long and
complex decision tree by looking at all the possible adverse effects,
the most important thing, at this point, is a statistic known as the dropout
.  That is, in studies, what percentage of people stop
taking the drug because of adverse effects?  The numbers are
comparable for all modern antidepressants, though, so that first cut
does not narrow the field very much.

Then, you look at patient acceptance.  Again, this is based upon
the likelihood of the patient actually taking the drug.  If they
seem predisposed to accept one drug over another, and there is no
reason to do otherwise, then go ahead and prescribe the drug that the
patient wants to try. 

"My roommate took Paxil, and told me to ask for that."

"Fine."  End of discussion.  Again, an oversimplification,
but it illustrates the point.  Obviously it does not always make
sense to give the patient whatever he or she wants, but if it is
reasonably congruent with what would be done otherwise, it is sensible
to do.

The rationale for considering cost is self-evident.  Why prescribe
something expensive, if something less expensive is just as likely to
work?  Fortunately, most of the modern antidepressants are now
available as generics.  Go to (or your favorite
similar site; there are many) and look up the prices for generic Prozac
(fluoxetine), Zoloft (sertraline) and Paxil (paroxetine).  Thirty
20mg capsules of fluoxetine cost $16, about one-third the cost of a cup
of Starbucks every day.  If you get a six-month supply, it costs
25 cents per day.  Not bad.

So far I've covered the two simplest scenarios.  So far, it seems
simple, perhaps surprisingly so.  The thing is, it gets
complicated very quickly, once we have to consider additional
factors.  I'll get to those factors, probably one at a time, over
the next several days.

More like this

I am 55 years old and am taking amatrpiline 150mg per day I would like to change my medication what would you recomend I have been on the tablets this time for 3 mth my side efects are shaking hot sweats

By alan nesbitt (not verified) on 26 Jun 2007 #permalink