The evidence base

Overall, randomised controlled trials show that depression ratings decrease slightly more with antidepressants than placebo. Studies are inconsistent, however, and differences are small, especially when unpublished trials are included. Reviews of the literature on tricyclic and older antidepressants concluded that ‘in well-designed studies, the differences between antidepressants and placebo are not impressive’.Reference Smith, Traganza and Harrison ³ Meta-analyses of trials of selective serotonin reuptake inhibitors (SSRIs) and other modern antidepressants that include unpublished trials have found mean differences between antidepressants and placebo ranging from 1.80 to 2.55 points on the widely used Hamilton Rating Scale for Depression.Reference Khin, Chen, Yang, Yang and Laughren ^{4 –} Reference Turner, Matthews, Linardatos, Tell and Rosenthal ⁶

The clinical significance of such small differences is doubtful. The Hamilton scale has a total score of 54 points. A recent analysis comparing scores on the Hamilton scale with scores on the observer-rated Clinical Global Impression (CGI) scale suggests that a difference of three points on the Hamilton scale is equivalent to a rating of ‘no difference’ on the CGI scale, while a difference of eight points is required to obtain CGI scale ratings of ‘mild improvement’.Reference Leucht, Fennema, Engel, Kaspers-Janssen, Lepping and Szegedi ^{7 ,} Reference Moncrieff and Kirsch ⁸ Antidepressant/placebo differences therefore appear to fall well below levels required to make a noticeable difference in someone's condition.

Antidepressant effects and severity

It is often suggested that antidepressants are more effective, or perhaps only effective, in severe depression, and that this can explain their poor performance relative to placebo in trials with mixed populations. Some meta-analyses have found a gradient between the size of the antidepressant/placebo difference and the severity of initial depression across trials,Reference Kirsch, Deacon, Huedo-Medina, Scoboria, Moore and Johnson ^{5 ,} Reference Fournier, DeRubeis, Hollon, Dimidjian, Amsterdam and Shelton ⁹ although differences in people with the most severe depression still fall well below those equating to ‘mild improvement’ on the CGI. Other meta-analyses have not identified a severity gradient.Reference Gibbons, Hur, Brown, Davis and Mann ^{10 ,} Reference Walsh, Seidman, Sysko and Gould ¹¹

Older evidence suggests that antidepressants are not particularly helpful for inpatients with depression. A Medical Research Council trial, for example, found little difference between imipramine, phenelzine and placebo. ¹² Trials of antidepressants for the treatment of depression in people with bipolar disorder have also found no differences between antidepressants and placebo.Reference Zhang, Yang, Yang, Liang, Dai and Wang ¹³

Antidepressant effects and the drug-centred model of drug action

The accepted view of drug action in psychiatry is that psychiatric drugs work by targeting a putative underlying brain abnormality. I have called this the ‘disease-centred’ model of drug action. An alternative explanation is the ‘drug-centred’ model, which suggests that psychiatric drugs influence symptoms of mental disorder and distress through their psychoactive effects. ‘Psychoactive’ drugs, sometimes referred to as ‘mind-altering drugs’, include recreational drugs, drugs prescribed for mental health problems and some other medical drugs (e.g. steroids, anticholinergics). They modify normal thoughts, emotions and behaviours in characteristic ways. According to the drug-centred model of drug action, for example, antipsychotics reduce the symptoms of psychosis through their ability to produce a state of mental slowing and emotional restriction, a state they produce in animals and humans, regardless of the presence of psychiatric or behavioural problems. Anxiolytics reduce symptoms of anxiety through their well-known sedative and relaxant effects, which occur independently of any psychiatric disorder.

Elsewhere, I have outlined the lack of evidence for the disease-centred view of drug action for any class of psychiatric medication.Reference Moncrieff and Cohen ¹⁴ The serotonin and noradrenaline theories of depression, which appear to explain the action of antidepressants in a disease-centred manner, are not supported by evidence or expert opinion.Reference Moncrieff and Cohen ^{15 ,} Reference Lacasse and Leo ¹⁶ Moreover, numerous randomised trials have shown that drugs that are not considered as antidepressants, and have actions on other neurotransmitter systems, including benzodiazepines, opiates, stimulants and antipsychotics, are as effective as recognised antidepressants in people with depression.Reference Moncrieff and Cohen ¹⁵

The drug-centred model suggests that the mental and physical alterations produced by antidepressants account for the differences between antidepressants and placebo in randomised trials. The psychoactive effects of individual antidepressants vary in strength and character, depending on chemical class and composition. Tricyclic drugs are strongly sedating and impair psychological test performance.Reference Dumont, de Visser, Cohen and van Gerven ¹⁷ SSRIs have weaker and more subtle effects, but can induce a state of emotional numbing or restriction, lethargy, reduced libido and sexual impairment. They also occasionally produce a state of agitation and tension, especially in young people.Reference Goldsmith and Moncrieff ¹⁸ Antidepressant-induced emotional numbness may directly reduce the intensity of people's feelings, but mental and physical alterations may also produce an amplified placebo effect, by revealing to people participating in randomised trials that they are taking an active drug. The fact that drug/placebo differences are so small, however, suggests that antidepressant-induced alterations are not clinically useful, whether they act through pharmacological or psychological means.

Adverse effects

By emphasising that psychiatric drugs change the normal state of the brain and body, the drug-centred model highlights the likelihood of adverse effects. Although modern antidepressants are usually well-tolerated, there is mounting evidence of less common but serious effects, including increased suicidal thoughts,Reference Sharma, Guski, Freund and Gotzsche ¹⁹ fetal malformations,Reference Reefhuis, Devine, Friedman, Louik and Honein ²⁰ bleeding, a prolonged and severe withdrawal syndrome,Reference Fava, Gatti, Belaise, Guidi and Offidani ²¹ and persistent sexual dysfunction after discontinuation.Reference Farnsworth and Dinsmore ²² The widespread use of antidepressants may also produce nocebo effects by undermining people's sense of self-efficacy, potentially setting them up for a lifetime of chronicity and dependence on services.