Antidepressants have been in the news recently. The general feeling seems to be that although they are being overused and may have some unpleasant side effects, they certainly ‘work,’ at least in some people (1).
So what is the evidence that antidepressants ‘work’? If you compare them with a dummy tablet or placebo in a randomised trial, scores on rating scales that are meant to measure depression sometimes go down a few points more in people taking antidepressants compared to people on placebo. But what does this mean? Well, firstly, the differences are small. The commonly used Hamilton Rating Scale for Depression has a maximum score of 54 points and across studies differences are less than two points (2). A two point difference is unlikely to have any real (clinical) significance. Whether these scales actually measure a complex emotional state like depression is another question. They consist of lists of symptoms that are sometimes, but not always, associated with depressed mood. A two point change can occur because someone is sleeping better and may have no relation to the individual’s underlying mood.
But the real problem is that placebo controlled trials are not a level playing field. As I have highlighted in my blog on ’models of drug action’ (https://joannamoncrieff.com/2013/11/21/models-of-drug-action/) antidepressants are psychoactive substances. They make people feel different, both physically and mentally. The older ‘tricyclic’ antidepressants, such as amitriptyline, were profoundly sedating. There was no mistaking that you were taking them. The psychoactive effects of the newer antidepressants like fluoxetine (Prozac), paroxetine (Seroxat or Paxil) and venlafaxine (Effexor) are more subtle, but nevertheless present. They seem to make people a little drowsy sometimes, and lethargic. They reduce sexual drive, and in some people they produce a state of emotional detachment or indifference. Some people experience unpleasant feelings of tension or agitation (3).
The psychoactive effects of antidepressant drugs can affect the results of placebo controlled trials in two ways. Firstly, they may directly affect scores on depression rating scales. The emotional detachment produced by selective serotonin reuptake inhibitors (SSRIs) and similar drugs may reduce or blunt negative emotions, so people will rate themselves as less depressed. The sedative effects of the tricyclic antidepressants can improve sleep and reduce anxiety. Since these factors feature prominently in depression-measuring scales, these effects will produce an apparent improvement in depression, despite the fact that there may be no change in the individual’s actual mood (although of course feeling less anxious and sleeping better might improve one’s mood too).
Secondly, the mental alterations produced by psychoactive drugs, alongside their physical effects, may also affect depression ratings in randomised trials by signalling to people that they are taking the active substance rather than the placebo. This is what has been called the ‘amplified placebo effect’ (4). We use placebos in randomised trials because we know that the expectation that the drug will make you better increases people’s chances of actually getting better. Using a placebo is meant to guard against the role of expectations, but if people can guess whether they have had the active drug or the placebo, then this safeguard no longer operates. We know that people can usually guess better than chance whether they are on the active drug or placebo in randomised controlled trials of antidepressants and other drugs used in psychiatry (5).
If this is the case, people taking the active drug will have greater expectations of success than those on the placebo. So people in the placebo group get the ordinary placebo effect of thinking they are taking a drug, but people in the antidepressant group get an ‘amplified placebo effect’ because they don’t just think they are taking a drug, they have evidence (in the form of subjectively detectable drug-induced alterations) that they really are. An ‘amplified placebo effect’ is especially likely to occur if people enrolled in the study have a bias towards drug treatment in the first place. Since people who don’t want to take antidepressants would usually not take part in a drug trial, this is likely to be the case.
The direct impact of the psychoactive effects of antidepressants, together with the amplified placebo effect, mean that we cannot interpret the differences between antidepressants and placebo that occur in some randomised controlled trials as evidence that antidepressant drugs have ‘antidepressant’ effects. In other words, these differences do not demonstrate that the drugs reverse part of the underlying mechanism that leads to depressive symptoms. They only show that the experience of taking a drug with psychoactive effects is different from that of taking a sugar pill.
Consistent with this view, almost any type of drug with psychoactive properties has been shown to have ‘antidepressant-like’ effects in one study or another, including stimulants, benzodiazepines and antipsychotics (6). Substances without noticeable psychoactive or physical effects have not (7). The fact that antidepressants come from a wide range of chemical classes, and produce an enormous variety of physical and mental alterations, also supports the idea that it is the presence of these alterations and not any specific chemical mechanism that produces the effects seen in placebo- controlled trials.
Drugs might be useful in depression, however, even if they are acting through their psychoactive effects and not reversing an underlying pathology. The sedative effects of the older tricyclic antidepressants and some of the newer ones might be useful in facilitating sleep and reducing agitation. The emotional detachment or indifference produced by the SSRIs may come as a relief to some people who are deeply distressed. The wide-spread promotion of the idea that depression is caused by a chemical imbalance and that antidepressants help put it right means that most people do not expect the drugs to work in this way, however. Indeed, there is so little coverage of the psychoactive effects of antidepressants that it is likely that most doctors are only dimly aware of them.
Moreover, the psychoactive effects of the drugs we call ‘antidepressants’ do not come cost free, of course. SSRIs cause high rates of sexual dysfunction, including reduced libido which is probably an aspect of the emotional indifference they produce (3). Occasionally they seem to precipitate suicidal thoughts and inclinations and there are also withdrawal effects to consider. A minority of people have severe and prolonged withdrawal reactions (8).
Using psychoactive substances to cope with negative emotions is a longstanding human response, but also one that is fraught with difficulty. Although drug-induced effects may bring temporary relief, they may also hamper people from finding more lasting solutions to their problems. If people do want to go down this route, however, there seems no reason to restrict the repertoire to drugs currently called ‘antidepressants’. This raises all sorts of thorny questions, of course, about why some psychoactive drugs are legal and others illegal, about what sort of drug use society approves of and what it doesn’t, and why the legal dispensation of many drugs is restricted to doctors: subjects for many future blogs!
2) Kirsch I, Moore TJ, Scoboria A, Nicholls SS. The emperor’s new drugs: an analysis of antidepressant medication data submitted to the US Food and Drug Administration.. Prevention and Treatment 2002;5.
3) Goldsmith L, Moncrieff J. The psychoactive effects of antidepressants and their association with suicidality. Curr Drug Saf 2011 Apr;6(2):115-21.
4) Thomson R. Side effects and placebo amplification. Br J Psychiatry 1982 Jan;140:64-8.
5) Fisher S, Greenberg RP. How sound is the double-blind design for evaluating psychotropic drugs? J Nerv Ment Dis 1993 Jun;181(6):345-50.
6) Moncrieff J. Are antidepressants overrated? A review of methodological problems in antidepressant trials. J Nerv Ment Dis 2001 May;189(5):288-95.
7) Keller M, Montgomery S, Ball W, Morrison M, Snavely D, Liu G, et al. Lack of efficacy of the substance p (neurokinin1 receptor) antagonist aprepitant in the treatment of major depressive disorder. Biol Psychiatry 2006 Feb 1;59(3):216-23.
8) Fava M, Mulroy R, Alpert J, Nierenberg AA, Rosenbaum JF. Emergence of adverse events following discontinuation of treatment with extended-release venlafaxine. Am J Psychiatry 1997 Dec;154(12):1760-2.