Summary and Discussion

I have deliberately restricted this review to a few key papers, mostly ones that are available full-text online, free. Thus the data from which I have drawn the following tentative conclusions are available to any reader with internet access.

Treatment algorithm

1) A good first line antidepressant is escitalopram. It is the S isomer of citalopram, which was used as the Level 1 drug in STAR*D 1. It came a close second in the Lancet study for efficacy 7, and a clear first for acceptability. It was one of the best for sexual side-effects, equaling or bettering placebo on two scores 9. From STAR*D, the trial should last for at least two months.

2) If unsuccessful, Plan B could legitimately be one of several options, in no particular order.

a) Augment with thyroxine (T4). Especially if the measured T4 level is below average, eg <14 mmol/L. (Level 3, STAR*D 1, 4)
b) If failure of Plan A is related to 5HT-2 related side-effects (eg insomnia, agitation, anxiety, panic, sexual dysfunction) there is a logical case for augmenting with a drug with 5HT-2 blocking properties, such as mianserin. 11
c) Change to mirtazapine (Beware of weight gain). (Most efficacious, Lancet meta-analysis 7)
d) Change to sertraline. (Beware of diarrhoea and sexual problems). (Level 2, STAR*D 14, and Lancet 7)
e) Change to venlafaxine (Beware of nausea, sexual problems, discontinuation symptoms, and risk of suicide in adolescents) (Level 2, STAR*D 14, and Lancet 7).

3) Plan C.

a) Combined mirtazapine and venlafaxine. (Level 4, STAR*D 15)

Mysteries

1) If it is true that some depressions reflect disturbance in serotonin in the synapses of key circuits, some reflect disturbance of noradrenaline, and some reflect both (as we have generally believed), then why did STAR*D level 2 not show any difference between switches within class (eg citalopram to sertraline), and out of class (eg citalopram to venlafaxine or bupropion)?
2) Since citalopram and escitalopram are so similar (citalopram contains escitalopram, as well as R-citalopram), then how come citalopram is the worst SSRI for the sexual side-effects of desire and arousal, yet escitalopram is the best (even narrowly beating placebo for arousal)?
3) What is the mechanism of effect of augmentation with T3? Is there any reason not to use T4 instead? 16

Treatments to abandon

1) Fluvoxamine. 13
2) Reboxetine. 7
3) Augmentation with lithium for unipolar depression. 4

Treatments to downplay

1) Paroxetine. 7
2) Antidepressants in adolescents, especially venlafaxine and paroxetine. 8

Treatments in danger of being abandoned prematurely.

1) Tranylcypromine. 'Approximately 30% of participants in the tranylcypromine group had less than 2 weeks of treatment, and nearly half had less than 6 weeks of treatment'. 3, 15.