4. What psychiatric drugs do by class

4.2.1 History

During the 1950s, certain drugs were tried out on people who were depressed, which started to be called antidepressants. One group of these drugs, which are similar in structure to some of the early antipsychotics, is known as the tricyclic antidepressants. Another group is the monoamine oxidase inhibitors or MAOIs. These were the main types of antidepressant used until the late 1980s. Prozac was launched in 1988 and was the first of a series of new antidepressants introduced during the 1990s called the ‘selective serotonin re-uptake inhibitors’ (SSRIs). These were joined by a variety of other sorts of drugs also branded as antidepressants (including venlafaxine, duloxetine and mirtazapine).

From the beginning of the 1990s, industry advertising campaigns and professional publicity increased the prescribing of these drugs substantially. Antidepressants are now by far the most commonly prescribed class of psychiatric drug, and their use continues to rise. In 2016 in England, over 65 million prescriptions were issued for antidepressants, a 6% increase on the previous year and over 500% increase since 1992.¹

Antidepressants are regarded as useful treatments for depression and a range of other conditions and their use is recommended in various situations by official guidance.

4.2.2 Common short-term uses

Antidepressants are recommended for what is judged to be moderate or severe depression, and for less severe depression that is not helped by psychological interventions¹. SSRIs are usually the first choice.

People diagnosed with depression who recover after being prescribed an antidepressant for the first time, are generally advised to continue the drug for at least six months.² Antidepressants are usually prescribed for a longer period if the patient has suffered several episodes, the symptoms have not disappeared entirely, the person has a long-term physical health condition, or has ongoing life stressors. In such cases, antidepressants are recommended to be continued for a minimum of two years², but increasingly people end up taking these drugs for multiple years and beyond.

They are also prescribed for individuals who have received a diagnosis of a variety of other mental health difficulties, including anxiety, obsessive compulsive disorder, panic disorder, phobias, bulimia and post-traumatic stress disorder. Tricyclic antidepressants are sometimes also used to treat chronic pain, particularly pain of a neurological origin, or insomnia, usually at lower doses than those recommended for depression.

4.2.3 Theories of action

The traditional view of antidepressant action, based on a disease-centred model (outlined in section 2), suggests that antidepressants help correct a chemical imbalance presumed to be present in depression. They are said to increase the availability of certain neurotransmitters that are thought to be deficient in depression. Older drugs, like the tricyclic antidepressants and the MAOIs are thought to act by increasing the availability of the neurotransmitter noradrenalin. The SSRIs are still generally believed to improve depression by correcting a deficiency of serotonin.

Although the idea that depression is caused by a chemical imbalance has entered the public consciousness, the ‘monoamine’ theory of depression is not supported by evidence or expert opinion.^3,4 Studies of serotonin receptors, for example, show contradictory findings, with some showing that receptor numbers are reduced in people with depression, compared to people without, some showing no difference, and some showing they are increased. Studies that aim to induce a lowering of serotonin levels through dietary means do not show any association with the onset of depression in people with no history of depression, although some studies show a deterioration of mood in people with a previous history of depression who have been treated with SSRI antidepressants. Evidence on noradrenaline is also contradictory.⁵ In addition, numerous randomised trials have shown that drugs that are not thought of 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.⁵ Leading psychopharmacologists have concluded that direct evidence for the monoamine hypothesis is lacking.^6,7 Indeed, the whole ‘chemical imbalance’ theory of depression is now dismissed as overly simplistic by academic psychiatry.⁸ Some official sources continue to suggest that antidepressants work by increasing ‘levels of chemicals in the brain’ that are linked with depression⁸, but others, such as the Royal College of Psychiatrists public information leaflet, no longer mention reduced serotonin as a potential cause of depression.⁹

The drug-centred model as outlined in section 2, suggests that antidepressants produce mental and physical alterations, which interact with the symptoms of depression. These may potentially account for certain differences between antidepressants and placebo in randomised trials. For example, the sedation produced by older antidepressants may be experienced as helpful by some people with anxiety and insomnia, while the emotional numbness induced by some antidepressants may reduce the intensity of negative feelings for some. The mental and physical alterations may also reveal to people participating in randomised trials that they are taking an active drug, increasing the placebo effect.

4.2.4 The alterations produced by antidepressants

This section describes evidence about the alterations to normal physical and mental functioning produced by different sorts of antidepressants. Their effects on ‘symptoms’ and ‘disorders’ is covered below in the sections on ‘efficacy’.

Antidepressants come from many different chemical classes, and therefore can be expected to vary in the effects they produce. Tricyclic antidepressants, for example, appear to be pharmacologically similar to some of the older type of antipsychotics. They are strongly sedating drugs. They increase sleep and cause drowsiness during the day. Studies with healthy volunteers show that taking tricyclic antidepressants makes people slower in their reactions and impairs intellectual abilities such as attention and memory. Taking them is usually an unpleasant experience for volunteers (it is associated with ‘dysphoria’ in volunteer studies).^14,15

SSRIs have more subtle effects in volunteer studies apart from their effects on the gut (most of the body’s serotonin is present in the gut). They commonly cause nausea and sometimes diarrhoea and vomiting. SSRIs also commonly produce mild drowsiness but can also cause insomnia. They can induce a state of emotional numbing or restriction.¹³ In addition, they can cause lethargy, reduced libido and sexual impairment. They also occasionally produce an unpleasant state of agitation and tension, especially in young people.^14,16 These effects can be difficult to recognise.

4.2.5 Evidence of efficacy

4.2.5.1 Short-term use in depression

Antidepressants are one of the standard recommended treatments for depression and many people regard them as useful. Their use is based on evidence from hundreds of placebo-controlled trials, which show that antidepressants are slightly better than a placebo in terms of scores on a depression rating scale, the principle outcome measure of these trials. Studies are inconsistent, however, and differences are small, especially when unpublished trials are included.

The small difference between antidepressants and placebo raises questions about whether the effects are, indeed, worthwhile. For example, in an analysis, which combined the results of several American trials of SSRIs and other new drugs, the difference between the drugs and the placebo was less than two points on the commonly used Hamilton Rating Scale for Depression (HRSD).¹⁷ Other meta-analyses, including the largest ever conducted, published in 2018, report similar small differences between antidepressants and placebo.¹⁸ The HRSD usually has 17 items and scores up to 54 points. When a difference of around two points is compared to ratings on a commonly used global measure of people’s overall condition, the Clinical Global Impressions Scale¹⁹, it does not register as showing any difference at all. Indeed, a difference of eight points on the HRSD would be required to register as a ‘mild’ level of improvement on the Clinical Global Impressions Scale, a difference that is way above that found in any combined analysis of placebo controlled antidepressant trials.²⁰ An analysis of trials conducted by the (then) National Institute of Clinical Excellence²¹ also found that the difference in depression scores between people randomised to antidepressants and people randomised to placebo was so small that it was, in the words of the Institute’s report, ‘unlikely to be of clinical significance’.²¹

Although depression rating scales scores are the principle outcome measures of placebo-controlled trials, results are often presented in terms of the proportion of people who show a ‘response’ to the antidepressant compared with the proportion that respond to the placebo. The largest antidepressant meta-analysis reported, for example, that people randomised to take antidepressants were one and a half to two times more likely to show a ‘response’ than people allocated to placebo.¹⁸ There is no objective marker of ‘response’, however. It is simply defined, quite arbitrarily, as a certain level of reduction in depression measurement scale scores. When scores are categorised in this way, however, the difference between the groups can be inflated, so that small absolute differences in scores become quite large differences in response rates.²² Therefore the depression scores are the most reliable measure of the outcome of these trials.

The small difference between antidepressants and placebo that is indicated by depression scale scores may not even be a genuine difference in actual levels of depression, however, but may be an artefact of research designs or a consequence of the mental alterations produced by antidepressants. Publication bias, not accounting for withdrawal effects from previous treatment and various statistical issues may have artificially inflated differences between antidepressants and placebos in randomised trials and meta-analyses of these trials.²³

Additionally, antidepressants may produce alterations that reduce depression-related symptoms without actually acting on depression itself. Depression often involves insomnia or sleeping difficulties and sometimes involves anxiety and agitation. Any drug with sedative properties will improve this aspect of the problem. The HRSD, for example, contains three items on sleep alone and these items can score up to six points. So, any difference between drugs and placebo may reflect the sedative qualities of some commonly used antidepressants (tricyclic antidepressants and mirtazapine, for example).

Any drug that alters our consciousness may also obscure or suppress depressive feelings. SSRIs appear to dull or numb emotions, which could reduce the intensity of depressive feelings.^16,24 Tricyclic antidepressants may also promote a state of emotional indifference, given their affinity with antipsychotic drugs that are known to have this property. All these effects may reduce scores on depression rating scales.

These and other alterations also mean that people involved in antidepressant trials are sometimes able to detect whether they are taking the active drug or the placebo. This may produce an unequal, amplified placebo response in people who are taking antidepressants in randomised trials. If people can improve by taking an inert placebo, what is known as the ordinary placebo effect, then people who take a drug that has noticeable effects may have an amplified placebo response. Conversely, people who take the placebo may realise this because they do not experience any of the ‘side’ effects they have been told to expect. Such people may do worse than they might do if they had not been enrolled in a trial in the first place. As such, the difference between antidepressants and placebo detected in clinical trials may be a result of ‘amplified’ placebo effects.²⁵

The idea that antidepressants may be working through inducing ‘amplified’ placebo effects is supported by the finding that other drugs with noticeable effects, including stimulants, benzodiazepines, opiates, and antipsychotics have been found to have equal effects to standard antidepressants in randomised studies in people with depression.⁸

In summary, antidepressants are only marginally better than placebo in randomised trials in people diagnosed with depression. Some evidence suggests the differences are unlikely to translate into meaningful clinical benefit. Moreover, there is no current evidence that strongly supports the idea that antidepressants produce their effects by acting on the underlying biological mechanism of depression.²⁶ Although much research has been conducted to look for the underlying mechanisms of depression, no such mechanism has been confirmed, and there remains little evidence that serotonin or other neurochemical abnormalities are associated with depression, or account for antidepressant action. Moreover, there are other convincing explanations of how antidepressants affect people with depression.

4.2.5.2 Antidepressants in severe depression

It is commonly stated that antidepressants are most effective in severe cases of depression. Overall, the evidence around this is contradictory. A NICE review claimed antidepressants have their most marked benefits in people with more severe depression, but the data actually found the greatest effects compared with placebo in people whose depression was in the middle range of severity, rather than in those with the most severe depression.²¹ A recent meta-analysis that specifically examined this issue found that the severity of depression was not correlated with drug-placebo differences.²⁷

4.2.5.3 Long-term use for relapse prevention in depression

There are several studies that show that if you take people whose depression has improved while they are taking antidepressants, and you randomise some of them to have their antidepressant stopped and substituted with a placebo, then the people transferred to placebo will have more ‘relapses’ of depressive symptoms.²⁸ Based on these studies, people who have had a single episode of depression are recommended to continue taking antidepressants for at least six months. People who have had recurrent episodes are recommended to take antidepressants on a longer-term basis.

However, the interpretation of these studies has been challenged particularly because the people transferred onto placebo are liable to experience withdrawal effects provoked by stopping antidepressants (see section 4.1.5).^29–31 These effects include anxiety and mood changes and may be mistaken for a relapse of the original problem.³²

In addition, people who experience withdrawal effects may realise that they have been swapped onto the placebo and this may make them anxious and vulnerable. The next time they experience problems they may lapse into a state of depression because they have come to believe that they need the drug to remain well and because they realise that they have been taken off it. This situation is likely, because the participants of these maintenance treatment trials are a selected group who have made a good initial response to treatment.³³ They may already be persuaded of the benefits of drug treatment, or, at least, they are likely to be nervous about having it withdrawn.

However, non-randomised observational studies provide no evidence that antidepressants improve long-term outcomes of depression. In fact, some studies indicate that long-term antidepressant use is associated with increased relapse rates,³⁴ and worse long-term outcomes,^35,36 compared to people who do not use antidepressants.

One such recent non-randomised study analysed the association of antidepressant use from the age of 20 and depressive symptoms over the course of the succeeding 30 years. Involving 159 people, it found that those who used antidepressants were more likely to have more severe symptoms during follow-up. However, it is likely that all these studies reflect the fact that people who take antidepressants generally have more severe problems initially than those who decide not to take them, which may account for their worse outcomes. Some studies have taken indicators of initial severity into account in the statistical analysis to some degree, but it is difficult to exclude this problem altogether.³⁷

4.2.5.4 Use in anxiety disorders

A recent meta-analysis of studies of the treatment of anxiety showed that SSRI and SNRI antidepressants were superior to placebo in reducing scores on anxiety rating scales, but again the effect was modest. The difference in improvement between people taking the drug and those taking placebo was between two and three points on the Hamilton Anxiety Rating Scale, which has a maximum score of 56 points.³⁸ Another meta-analysis of 12 trials of the SSRI drug paroxetine found that, on average, people randomised to take paroxetine improved by 2.3 points more than people randomised to placebo.³⁹ Studies comparing SSRI antidepressants with benzodiazepines for anxiety symptoms find that benzodiazepines have larger effects.⁴⁰

SSRIs and other antidepressants, particularly clomipramine, one of the old tricyclic antidepressants, are commonly prescribed to people who are diagnosed with obsessive compulsive disorder (OCD). They improve symptoms more than a placebo by around 3.2 points on a 40-point OCD measurement scale.⁴¹ Behaviour therapy has larger effects than medication, but most studies of therapy include people who are also on prescribed drugs.⁴²

4.2.6 Common adverse effects

Tricyclic antidepressants can slow down the conduction of electrical impulses in the heart and in high doses may cause dangerous irregularities of the heartbeat known as arrhythmias. Overdosing on these drugs is dangerous and often fatal. They also cause postural hypotension (a drop in blood pressure on standing up), which can lead to falls, and they increase the risk of seizures. They tend to have ‘anticholinergic effects’ including dry mouth, constipation, difficulty passing urine and blurred vision. At higher doses they may cause confusion. They also cause weight gain and sexual dysfunction including impotence, loss of libido and delayed orgasm.

The effects of SSRIs and SNRIs are similar, although SNRIs may produce more noticeable effects. Both types of drug commonly affect gut activity and cause nausea, vomiting, diarrhoea, constipation, and abdominal pain. They are also associated with sexual dysfunction, especially delayed orgasm. There are mounting anecdotal reports that the sexual dysfunction associated with SSRIs can occasionally persist after the drugs are discontinued, sometimes for months or years.⁴³

Both SSRIs and SNRIs can cause lethargy and SNRIs may cause drowsiness. The state of emotional numbing or detachment they produce can be experienced as unpleasant and debilitating⁴⁴ and is associated with sexual dysfunction.⁴⁵ They can also produce a state of anxiety and agitation, especially in younger people,^46,47 which can also be extremely unpleasant and may be predictive of increased suicidal impulses (see below).

4.2.7 Other adverse effects

Some SSRIs, particularly paroxetine, have been linked with birth defects,⁴⁸ and as a class these drugs can thin the blood and produce bleeding disorders.⁴⁹

4.2.8 SSRIs and suicide

Several meta-analyses of antidepressant studies in children and adolescents show increased rates of suicidal behaviour associated with use of SSRIs.^50–53 Some meta-analyses of trials in adults indicate small increases in suicide attempts or self-harm in people on SSRIs compared with placebo,^54,55 but others do not.^56–58 A recent re-analysis of one of these negative studies revealed a significant increase in suicidal ideation and behaviour using a different statistical approach.⁵⁹ However, where they have been compared with other types of antidepressants, SSRIs have not been found to be any worse in terms of increasing suicidal ideation and behaviour.^60,54 A recent meta-analysis based on data from original trial reports (which can provide more transparent data than official publications) found increased rates of suicidal thoughts and behaviour in children and young people taking antidepressants compared to those taking placebo, but there was no difference in adults. This analysis also found an increase in reports of aggressive behaviour among young people taking antidepressants compared to those on placebo.⁶¹ This confirms evidence from case reports of violent incidents, including legal reports and data from drug-monitoring agencies.⁶² It appears that these behaviours may be related to the state of agitation that SSRIs and related antidepressants can occasionally produce, which, for reasons that are not understood, seem to be more common among young people.⁴⁷

It is difficult to evaluate the conflicting evidence and claims about the relationship between antidepressants and suicide and violence because these situations are rare. On balance, the majority of evidence suggests that antidepressants can increase suicidal impulses and possibly also violent behaviour in children and young people. The evidence in adults is less conclusive.

4.2.9 Conclusion

Although antidepressants have been claimed to work by reversing underlying neurochemical abnormalities, no consistent abnormalities have been demonstrated in depression, and there is little evidence that antidepressants work in this way. Antidepressants show a minimal degree of superiority over placebo in short-term clinical trials (usually eight weeks) of depression. The small difference could be explained by drug-induced effects of antidepressants, such as sedation and emotional blunting, boosting improvements on depression measurement scales, as well as methodological factors in trial design, analysis and publication, which can artificially inflate drug-placebo differences. Finally, the findings of the many short-term trials do not enlighten us about the effects of long-term treatment. Despite the fact that many people end up taking antidepressants for months and years, there is little robust research on the benefits and harms of long-term treatment.

It is possible that some of the mental alterations produced by antidepressants could be useful for some people diagnosed with depression, at least on a temporary basis. Such effects vary depending on chemical class and composition of the particular antidepressant. The sedating effects of , tricyclic antidepressants, may be useful for insomnia, or to reduce anxiety and agitation. SSRIs, whilst exerting weaker and more subtle effects, can induce a state of emotional numbing or restriction, which may reduce the intensity of people’s feelings, thereby possibly impacting engagement in psychotherapy. However, the fact that evidence from randomised trials shows that drug-placebo differences are so small, and easily accounted for by non-pharmacological factors, suggests that antidepressant-induced alterations may not be practically useful.