The benefit and risks of selective serotonin reuptake inhibitors: a review
Samuel Rasche 10770003 Literature Thesis
Supervisors: dr. Huib van Dis and dr. Jos Bosch University of Amsterdam Department of Psychology November 2019
Index
Abstract………....3
Introduction - psychopharmacological treatment of psychological disorders.……….………….4
I. Depression….……….………...4
1.1 - Depression………....………..….4
1.2 - Operationalization of depression……….…..…….5
1.3 - Duration of a depressive episode ……….………..6
1.4 - Response, remission, recovery, relapse and recurrence in depression..………6
1.5 – Neural correlates of depression……….………7
1.6 – Depression as a network of symptoms without underlying cause……….….8
1.7 – Chapter conclusion……….9
II. Working mechanisms of selective serotonin reuptake inhibitors………...…………...………...10
2.1 –-Antidepressants and prevalence of use………..……….……..10
2.2 - Selective serotonin reuptake inhibitors…..……….……….……..10
2.3 - Monoamine hypothesis………..……….…...11
2.4 - Properties of the six SSRI agents……….….…….……12
2.5 - Pharmacokinetic interactions……….………...13
2.6 - Chapter conclusion……….………...14
III. Effectiveness of selective serotonin reuptake inhibitors……….………...…...15
3.1 - Assessing selective serotonin reuptake inhibitor effectiveness……….….15
3.2 - Effectiveness of selective serotonin reuptake inhibitors versus placebo...………16
3.3 - Effectiveness of selective serotonin reuptake inhibitors in children and adolescents…………..….17
3.4 - Placebo efficacy……….………19
3.5 - Statistically significant versus clinically relevant..……….………..21
3.6 - Effectiveness of selective serotonin reuptake inhibitors versus psychotherapy………23
3.7 - Lack of ‘blindness’ in double-blind studies.……….……….24
3.8 - Confounds in causal inference………...26
3.9 - Chapter conclusion…….………...27
IV. Detailed summary of the side effects of selective serotonin reuptake inhibitors.…………...….………..28
4.1 - Side effects……….………28
4.2 - Psychological and nervous system side effects………..…………28
4.3 - Gastrointestinal, hepatic and metabolic side effects……….…...30
4.4 - Genitourinary side effects and post-selective serotonin reuptake inhibitor sexual dysfunction...….31
4.5 - Other side effects………...………32
4.6 - Precautions with other conditions and medicines…….………...……….33
4.7 - Life-threatening side effects……….………….…….34
4.8 - Chapter conclusion….………...…………36
V. Long-term use and the effects of selective serotonin reuptake inhibitor discontinuation……….….38
5.1 - Withdrawal effects after discontinuation……….………..38
5.2 - Incidence of discontinuation symptoms…….………...……….39
5.3 - Long-term use (due to discontinuation) and its side effects...………..…….40
5.4 - Tapering……….………...……….41
5.5 Chapter conclusion…...………...…………42
VI. General discussion...………..………..43
Abstract
Biomedical theories portray psychological disorders as involving neurochemical imbalances that can be resolved with psychoactive drugs. One kind of psychoactive drugs are selective serotonin reuptake inhibitors, the most commonly used antidepressants. Although experienced as effective by many users, the agents may be accompanied by serious adverse effects. This review discusses the benefit-risks profile of commonly used selective serotonin reuptake inhibitors. Various studies have shown SSRIs efficacy, but this has not established beyond doubt. Namely, substantial placebo effects are observed in SSRI-placebo comparison studies, and pharmaceutical company funded and publication biases are suspected of
portraying SSRI efficacy too positively. A final pint of debate is that statistical differences may not be clinically relevant. Based on review of these issues and an additional examination of the side and withdrawal effects associated with SSRI use, the current thesis concluded that the findings point to a negative benefit-risks profile of selective serotonin reuptake inhibitors.
Introduction - pharmacological treatment of psychological disorders
Psychological disorders are commonly conceptualized as brain disorders in which there is some sort of chemical imbalance. Presumably, the homeostatic equilibrium can be restored by taking psychoactive drugs that boost levels of, or responsivity to, certain
neurotransmitters. Neurotransmitters are messenger molecules that affect neural transmission by their release in the synaptic cleft and binding to their cognate receptors, thereby changing psychological and behavioral functioning. By adopting this biomedical model of mental disorders, drugs have become first-line treatments and are regarded as directly treating the underlying cause of the disorder. However, as will be argued throughout this thesis, the
mechanisms of depression and antidepressants are likely to be much more complex. Moreover, psychoactive drugs influence the entire nervous system, hence (serious) side effects are
sometimes observed. It is therefore important to compose a benefit-risks profile for
psychoactive drugs. This narrative literature review shall focus on depression and will aim to compose a benefit-risks profile for the antidepressants ‘selective serotonin reuptake inhibitors (SSRIs)’. First, depression, the working mechanisms of SSRIs and their effectiveness in treating depression will be discussed, followed by chapters that will examine the possible side effects and withdrawal effects with SSRI use.
I. Depression
1.1 – Depression
Depression is a common mental disorder, affecting more than 300 million people worldwide (WHO, 2019). Following the diagnostic and statistical manual of mental disorders (DSM-5, 2013), the diagnosis major depressive disorder is given when an individual meets at least five of the following criteria for a duration of two weeks. One of the first two symptoms needs to be necessarily present:
1. Depressed mood most of the day, nearly every day.
2. Diminished interest in (almost) all activities most of the day, nearly every day.
And an additional four of the following symptoms need to be present:
3. Significant weight gain or weight loss when not dieting, or decrease or increase in appetite nearly every day
4. Insomnia or hypersomnia nearly every day
5. Psychomotor agitation or retardation nearly every day 6. Fatigue or loss of energy nearly every day
7. Feelings of worthlessness or excessive or inappropriate guilt nearly every day 8. Diminished ability to think or concentrate, or indecisiveness, nearly every day
9. Recurrent thoughts of death, recurrent suicidal ideation without specific plan, or a suicide attempt or a specific plan for committing suicide
Moreover, for a diagnosis, these symptoms need to cause clinically significant distress or impairment in social, occupational or other important areas of functioning. Lastly, the depressive episode must also not be the result of substance abuse or a different medical condition (other diagnoses are more appropriate when this is the case).
Most studies include patients suffering from major depressive disorder and therefore the focus of this review will also be on major depressive disorder. However, in addition to major depressive disorder there are many more variants of depression, including dysthymia, premenstrual dysphoric disorder, substance/medication induced depressive disorder,
depressive disorder due to another medical condition and bipolar disorder. Psychiatric comorbidity is also common. Thus, depression is regarded as a category rather than a single condition.
1.2 – Operationalization of depression
How is an indication of depression severity established? Most studies measure the severity of depressive symptoms by means of questionnaires. A frequently used questionnaire is the Hamilton Depression Rating Scale (HDRS or HAMD) (Hamilton, 1960). The original questionnaire consists of 17 items, but there also different versions, such as the HDRS-21 (21
items) and HDRS-24 (24 items), among others. The score on the original scale ranges from 0 to 52. A score of eight is considered to be the cut-off score between asymptomatic and
symptomatic status. a score above 24 is considered a severe depression (Table 3) (Zimmerman et al., 2013). It should be noted that the questionnaire is scored by psychiatrists. Hence, the HDRS score is subject to the interview skills of the psychiatrist and the weight that is given to the frequency and intensity of a symptom by that interviewer, which can be problematic for the reliability of the scale (Potts et al., 1990).
Depression severity Range of score
None 0-7
Mild 8-16
Moderate 17-23
Severe ≥24
Table 3. Score interpretation of the HDRS-17. Obtained from Zimmerman et al. (2013).
1.3 – Duration of a depressive episode
The duration of a depressive episode varies per person and can be influenced by severity and the presence of comorbidity, among other factors. Various studies report a median duration of three months and a decreasing rate of recovery over time (Furukawa et al., 2000; Spijker et al., 2002; Eaton et al., 2008). In the study of Spijker et al. (2002) the participants were further analyzed by level of care. Interestingly, for those individuals who had not received any professional help, the median duration was also three months.
1.4 – Response, remission, recovery, relapse and recurrence in depression
Various terms are used to indicate whether a depressive episode is dissolving or
returning. When the symptoms have been reduced by 50% (50% reduction in baseline score on the HDRS), there is said to be a ‘response’ to the treatment by the patient. Reaching a state of ‘remission’ is commonly the treatment goal (Thase, 2003). Remission is achieved when all symptoms are cleared. This equals a score of seven or below on the HDRS. If the state of
remission is maintained for longer than six months, the person is considered to be ‘recovered’. However, this recovery does not entail that the person is cured, since a depressive episode may still reoccur. Falling back into a state of depression during remission is known as ‘relapse’, whereas a returning depressive episode during recovery is known as ‘recurrence’. Sadly, to reach a state of remission can be quite challenging. Only one third of clinical patients reach this state on their first antidepressant treatment. Moreover, only about two thirds of patients reach remission after receiving four different antidepressant treatments for a period of a year (Stahl, 2013).
Relapse and recurrence of major depressive disorder is commonly observed. In a 23-year follow-up study of first-onset general population patients, only half of the cases had no further episode (Eaton et al., 2008). In clinical settings it is even higher, about 85% of patients experience recurrence in 15 years (Hardeveld et al., 2010). Moreover, Eaton et al. (2008) also reported that about 15% of cases are unremitting. Reaching a state of remission is clinically relevant, because residual symptoms increase the likelihood of falling into relapse or
recurrence. Londen et al. (1998) found that patients with residual symptoms relapsed after an average of four months, whereas remitters relapsed or recurred after an average of 12 months. Unfortunately, relapse or recurrence is still frequently observed in remitters. Moreover, the likelihood of relapse or recurrence increases with the number of pharmacological treatments it takes to reach remission (Stahl, 2013). Of course, patients that receive more treatments are usually more treatment resistant. The possibility exists that these patients are more prone to depression in general, therefore more likely to relapse.
1.5 – Neural correlates of depression
Much research has been dedicated to discover the neural correlates of depression. Major depressive disorder has been related to impaired functioning in a broad variety of brain
areas, mostly in the (medial) prefrontal cortex and subcortical structures, such as the amygdala and basal ganglia (Drevets et al., 2008). All these areas that have been related to depression are involved in the expression, interpretation, evaluation and experience of mood and emotions (Phillips et al., 2003), hence it is not surprising that these areas are implicated in depression. Nevertheless, currently there is no biomarker of depression in the brain.
In addition, neurochemical theories have been proposed to explain depression. The classic monoamine hypothesis suggests that depression is caused by a deficiency or depletion of the monoamines serotonin, norepinephrine and/or dopamine (Hirschfeld, 2000). This theory is problematic however, because both low and high levels of serotonin are observed in
depressive patients (Bowden et al., 1984).
In depression, genetics and the environment also seem to play a major role (Goldstein et al., 2011). It is thus this interaction between biological susceptibility and stressful life events that may cause a major depressive disorder in some individuals.
1.6 – Depression as a network of symptoms without underlying cause
Further complicating labelling depression and finding its neural correlates is that the symptom profile may differ between patients. This conflicts with the idea of a single construct that equally underlies all these symptoms; i.e., depression as a latent variable. As described above, research has been dedicated at discovering this underlying variable that is regarded as the cause; hence a single treatment may be effective for all patients. However, it may be possible that in depression different symptoms influence each other, rather than reflecting an underlying common cause (Cramer et al., 2010; Borsboom, 2017). This network approach to psychopathology contrasts with the latent variable view to psychopathology. Adopting such a theory has certain implications. It would assume that depression may be treated more
effectively by treating the numerous symptoms separately, rather than having one treatment (e.g. a drug) that is used to treat all symptoms (Borsboom, 2017).
1.7 – Chapter conclusion
Many people suffer from depression. It is a complex disorder that varies in type, severity, duration and symptom profile. Theories about the relation between brain structure or functioning and depression have been proposed, but this knowledge has not resulted in a clear biomarker of depression. This may, according to some, indicate that there is no underlying (neural) variable causing depressive symptoms (Borsboom, 2017). Depression is commonly treated with a class of antidepressant called selective serotonin reuptake inhibitor. These medicines will be discussed in the next chapter.
II. Pharmacological treatment of depression
2.1 – Antidepressants and prevalence of use
Antidepressants are commonly used and prescribed. Almost 7% of the Dutch
population used antidepressants in 2018 (Table 1). The amount of users in the United States is even higher, namely 12,7% in the period 2011-2014, and antidepressant use has increased over the years (Pratt et al., 2017). Antidepressants are mainly used to treat major depressive
disorder, but are also commonly prescribed to treat other disorders, including other depressive disorders (e.g. dysthymia and premenstrual dysphoric disorder), anxiety,
obsessive-compulsive disorder and eating disorders. There is a variety of antidepressants, usually divided into four classes: selective serotonin reuptake inhibitors (SSRIs), tricyclic antidepressants (TCAs), monoamine oxidase inhibitors (MAOIs) and a rest-category. Table 1 provides an overview of the number of users in the Netherlands per class.
Antidepressant 2014 2015 2016 2017 2018
Selective serotonin reuptake inhibitors 532.740 543.100 545.210 536.790 550.560
Other antidepressants 280.730 284.800 288.530 286.630 291.750
Tricyclic antidepressants 267.090 271.610 279.460 280.580 286.370 Non-selective monoamine oxidase inhibitors 1.850 1.945 2.035 2.077 2.078 Monoamine oxidase inhibitors type A 1.243 1.147 1.118 1.056 1.019
Total 1.083.653 1.102.602 1.116.353 1.107.133 1.131.777
Table 1. The number of people that used antidepressants in the Netherlands in 2014-2018. Retrieved from
GIPdatabank (2019). https://www.gipdatabank.nl/databank?infotype=g&label=00-totaal&tabel=B_01-basis&geg=gebr&item=N06A
2.2 – Selective serotonin reuptake inhibitors
The focus here will be on SSRIs, since these are the most used antidepressants. About half of antidepressant users in the Netherlands use these agents (Table 1). The preference for SSRIs as a first-line treatment for depression is due to better tolerability compared to other antidepressants, but not necessarily greater effectiveness (Anderson, 1998; Cipriani et al., 2018). There are six main prescribed SSRI brands (summarized in Table 2). Although these six agents have unique pharmacological properties, they all act on the central nervous system
by inhibiting the serotonin transporter. This will be discussed in more detail in the next paragraph. SSRI 2014 2015 2016 2017 2018 Citalopram (Cipramil ®) 160.270 167.090 169.810 169.930 177.710 Paroxetine (Seroxat ®) 179.790 175.490 168.630 158.990 154.470 Sertraline (Zoloft ®) 72.972 78.478 81.983 83.644 89.008 Escitalopram (Lexapro ®) 61.040 63.246 65.049 65.727 70.022 Fluoxetine (Prozac ®) 54.054 54.547 55.407 54.540 56.650 Fluvoxamine (Faverin ®) 20.070 19.423 18.627 17.625 17.156
Table 2. Amount of users of the six major SSRI agents in the Netherlands for the period 2014-2018. The brand
names of the SSRIs are shown between parentheses. Retrieved from GIPdatabank (2019).
https://www.gipdatabank.nl/databank?infotype=g&label=00-totaal&tabel=B_01-basis&geg=gebr&item=N06AB
2.3 – Monoamine hypothesis
In accordance with the monoamine hypothesis (described in 1.5), antidepressants would relieve the patient from depression by increasing the levels of monoamines, especially serotonin, thereby modulating activity in the areas that are affected by depression and
restoring the chemical balance in the brain. SSRIs achieve this increase of serotonin in the synaptic cleft by blocking the serotonin transporter (SERT). This transporter can be regarded as a presynaptic pump that takes serotonin back into the synapse where it is stored for
subsequent use (Stahl, 2013). Thus, by inhibiting this transporter, serotonin cannot be transported back, hence, the amount of serotonin is increased in the synaptic cleft. The
inhibition raises the level of serotonin rapidly. This is however problematic for the monoamine hypothesis, since relieve of depressive symptoms is not experienced until weeks later.
To solve this issue, Stahl (2013) proposed a more elaborate neurotransmitter-receptor hypothesis that fitted the results better. This theory suggests that in addition to a monoamine deficiency, there is an upregulation of postsynaptic monoamine receptors in depression. Antidepressants would downregulate neurotransmitter receptors and alter their sensitivity. More specifically, taking SSRIs would initially increase serotonin mainly in the
somatodendritic area (near the cell body), instead of the axon terminal and synaptic cleft. At the somatodendritic area, the increased serotonin would bind to 5HT autoreceptors (a type of
serotonin receptors), which causes downregulation and desensitization of these receptors over time. Once desensitization at the somatodendritic area has occurred, serotonin can no longer inhibit its own release, causing an impulse down the axon and consequently serotonin release in the synaptic cleft. Here, downregulation and desensitization of post-synaptic 5HT1A
receptors would occur as well, resulting in an overall increase of serotonin. This whole process takes at least two weeks and would coincide with the onset of therapeutic effects. Side effects are experienced immediately, and would presumably be caused by acute serotonin increase at different receptors and brain pathways. Besides indications for 5HT1A receptor desensitization
by fluoxetine in rats (Raap et al., 1999), not much empirical evidence could be found that supports this neurotransmitter-receptor theory.
2.4 – Properties of the six SSRI agents
Even though all SSRIs increase serotonin levels by inhibiting the serotonin transporter (SERT), all drugs have unique secondary properties (Table 4). This may explain why
individuals can experience different efficacy and tolerability with each agent. Depending on the type of depression symptoms and comorbidity certain SSRIs may be preferred over others, since the drugs are associated with different psychological effects (Table 4).
SSRI Additional side of action Neurochemical effect Psychological effect
Citalopram (Cipramil ®)
- R and S enantiomer - Antihistaminergic Calming, sedating Paroxetine
(Seroxat ®)
- M1 receptor binding
- Norepinephrine transporter (NET) inhibitor
- Nitric oxide synthetase (NOS) enzyme inhibitor
- Antihistaminergic - increased norepinephrine
Calming, sedating. NOS inhibition may contribute to sexual dysfunction
Sertraline (Zoloft ®)
- dopamine transporter inhibitor - σ1 receptor binding
- Increased dopamine - σ1 actions not well
understood
Improved energy, attention and concentration. σ1 actions
are related to anxiolytic and antipsychotic effects Escitalopram
(Cipralex ®)
- S enantiomer - Removal of R enantiomer removes antihistaminic properties
Anxiolytic effect. Best tolerated SSRI
Fluoxetine (Prozac ®)
5HT2C antagonist Increased norepinephrine
and dopamine
Improved energy, attention and concentration
(Fevarin ®) understood effects
Table 4. Secondary properties of SSRIs and the presumed neurochemical and psychological effects of these
additional actions. Obtained from Stahl (2013).
2.5 – Pharmacokinetic interactions
Besides the pharmacodynamic actions of psychoactive drugs (discussed above for SSRIs), many also have pharmacokinetic actions, meaning that they affect the cytochrome P450 (CYP) enzymes. This is a family of enzymes that is involved in the metabolic
breakdown of drugs. The various SSRIs differ in their pharmacokinetic properties; most importantly in their half-life (Stahl, 2013). The half-life of a drug is the time it takes for the drug concentration in the blood plasma to be halved, and thus gives an indication of the time needed for the drug to be eliminated from the body. Whereas most SSRIs have a half-life of approximately one day, fluoxetine has a long half-life (3-6 days) and its active metabolite norfluoxetine has an even longer one (7 to 16 days) (Table 5) (FDA, 2016). The long half-life fluoxetine is proposed to be advantageous, because the slow clearance from the body would reduce the withdrawal reactions that are caused by sudden discontinuation of SSRIs (Gury & Cousin, 1999) (Chapter 5). However, with a long half-life it is especially important for the patient to have a long ‘wash out’ period wherein no treatment is given before starting with another agent (Stahl, 2013). This way, it can be ascertained that the drug is eliminated from the body. If this is not done, interaction effects may occur (e.g., the serotonin syndrome (chapter 4) with premature MAOI administration after SSRI discontinuation) (Stahl, 2013).
With regard to antidepressants, an important CYP enzyme is CYP 1A2 (Stahl, 2013). Various antidepressants are substrates of this enzyme, including the SSRI fluvoxamine. Fluvoxamine is a substrate and also an inhibitor of this enzyme, meaning that the absorption, metabolization and excretion of other substrates of this enzyme is slowed down, thereby influencing the half-life of other substrates. That is, if substrates of CYP 1A2 (which are numerous drugs, including many antipsychotics, anticoagulants, analgesics and caffeine) are
simultaneously administered with fluvoxamine, the blood plasma and brain levels of these substrates may increase, potentially causing side effects (chapter 4).
SSRI Half-life (in hours)
Citalopram 35 Paroxetine 21 Sertraline 26* Escitalopram 27-32 Fluoxetine 72-384** Fluvoxamine 15-22
Table 5. Elimination half-life of the six SSRI agents. *Sertraline has the less active metabolite
‘N-desmethylsertraline’ with a half life of 62 to 104 hours (FDA, 2016). **fluoxetine has a long active metabolite ‘norfluoxetine’.
2.6 – Chapter conclusion
Selective serotonin reuptake inhibitors are commonly used to treat depression. These agents would restore the homeostatic neurochemistry of the depressed brain through its serotonergic actions. Its efficacy in treating major depressive disorder will be discussed in the following chapter. In addition to influencing areas presumably related to depression,
psychopharmacological agents also affect other bodily and brain areas. Hence, psychological and physiological side effects are commonly observed. These will be discussed in chapter 4.
III. Effectiveness of selective serotonin reuptake inhibitors
3.1 – Assessing selective serotonin reuptake inhibitor effectiveness
How can the efficacy of a drug treatment be established? Effectiveness of
antidepressants is usually recognized when there is a response, meaning a 50% reduction in score on scales such as the HDRS (Thase, 2003; Cipriani et al., 2018). A 50% reduction in scores does not always entail remission, since remission is achieved when there is a HDRS score of 7 or below, while scores can reach a maximum of 52 (chapter 1). Antidepressants are unfortunately not effective for many patients (Spijker et al., 2002; Eaton, 2008; Stahl, 2013; Cuijpers, 2018). Only about one third to half of patients reach remission with their first treatment and antidepressants (Stahl, 2013). Although there are no strong predictors for remission, severe cases seem to be very treatment resistant (Spijker et al., 2002; Eaton, 2008). Nevertheless, since SSRIs are commonly prescribed they surely must be more effective than placebos. Sadly, even this has been difficult to prove.
Several reasons make the efficacy of antidepressants difficult to establish. First, the efficacy of antidepressants over placebos is established by showing that patients in the active drug group reach a reduction of 50% in HDRS score, whereas the patients in the placebo group do not. Studies have found efficacy of antidepressants over placebos (e.g., Waugh & Goa, 2003; Hieronymus et al., 2016). Yet, spontaneous recovery may occur and there is a considerable response to placebos as well (Spijker et al., 2002; Furukawa et al., 2016; Bijl, 2019). This questions the contribution of the psychoactive substance to the alleviating effects, with some authors arguing that the contribution is negligible (Kirsch, 2003; Moncrieff & Kirsch, 2005; Bijl, 2019). Second, another common critique in antidepressant research is publication bias: ‘positive’ results would be published, whereas studies not finding any result would be left unpublished. This would lead to an unbalanced reflection of the findings in favor of studies reporting positive results (Kirsch, 2003; Ioannidis, 2008; Turner et al., 2008). Third,
studies funded by the pharmaceutical industry have been accused of overestimating the efficacy and tolerability of antidepressants due to a biased interpretation of the results (Lexchin et al., 2003; Als-Nielsen et al., 2003; Whittington et al., 2004; Bijl, 2019). These reasons have led various authors to question antidepressant efficacy, with some concluding that antidepressants cause more harm than benefit (e.g., Moncrieff & Kirsch, 2005; Healy, 2015; Gøtzsche, 2015; Bijl, 2019).
3.2 - Effectiveness of selective serotonin reuptake inhibitors versus placebo
Recently, a large meta-analysis by Cipriani et al. (2018) found effectiveness of antidepressants over placebos. A total of 522 randomized control trials were analyzed and all types of antidepressants were included. The study included 86 (16%) unpublished studies, which have been associated with less favorable outcomes for antidepressants, and included 409 (78%) studies that were funded by the pharmaceutical industry, which have been associated with more favorable results for antidepressants (Lexchin et al., 2003). The study analyzed 21 different antidepressants and all were found to reduce symptoms by 50% relative to placebos. The authors did note that the risk of bias was substantial and that there was a small effect size of d = 0.3. Thus, these results led the authors to conclude that all
antidepressants were more effective than placebos, but there may be risk of bias and non-response may occur.
The findings of Cipriani et al. (2018) elicited many reactions. Some have regarded the study as proof for antidepressant efficacy (Adlington, 2018), whereas others have remained skeptical over the effectiveness of antidepressants (Hengartner & Plöderl, 2018; McCormack & Korownyk, 2018; Moncrieff, 2018; Bijl, 2019). For example, some authors discussed statistical outcomes that undermine the validity of the results of the meta-analysis by Cipriani et al. (2018) (Hengartner & Plöderl 2018; Bijl, 2019). Their critique is based on the following
argument: assuming a normal distribution, when selecting a random person from the antidepressant group with an effect size of d = 0.3 (the effect size found by Cipriani et al. (2018)), there will be a chance of 58% better outcome than in the placebo group (50% being no difference at all). Assuming a placebo response rate of 35-40%, they calculated how this effect size would translate to efficacy and concluded that, relative to placebo, 9 patient would need to receive antidepressant treatment for one to benefit. This means that 8 out of 9 patients would benefit equally from placebo. Others have estimated that the ‘number needed to treat’ would be between 5 to 10 patients (Thase et al., 2011; McCormack & Korownyk, 2018). Thus, based on these findings of Cipriani et al. (2018), some may benefit from antidepressants but many patients seem to equally benefit from placebos.
3.3 – Effectiveness of selective serotonin reuptake inhibitors in children and adolescents Just as in adults, SSRIs are the first-line pharmacological treatment in child and adolescent depression. In the Netherlands, over 30.000 people under the age of 25 use SSRIs (GIPdatabank, 2019). In the past, antidepressants were thought to be beneficial for the pediatric population, but this has also been receiving a lot critique throughout the years. The reason for the criticism are mainly two factors: questionable efficacy and the increased risk of side effects among underage users (chapter 4).
Older studies showed effectiveness of SSRIs in child and adolescent depression. Wagner et al. (2003) found that 69% of the children and adolescents (aged 6-17) responded to sertraline, whereas 59% responded to placebo. This differed significantly, leading the authors to conclude that sertraline is effective in this population group. The incidence of certain adverse effects was at least double in sertraline compared to placebo and occurred in at least 5% of patients, but it was concluded that sertraline was generally well tolerated. Another study, by Keller et al. (2001), found paroxetine efficacy in treating major depressive disorder
in adolescents (aged 12-18). In addition, withdrawal rates did not differ much from placebo. Interestingly, a difference between drug and placebo on parent- and self-rating measures was not found, which may question the clinical relevance. Nevertheless, the authors concluded that paroxetine was effective and well tolerated by adolescents. It was also found that combination therapy (fluoxetine and cognitive behavioral therapy) was the most effective treatment in child and adolescence depression (March et al., 2004), although the results of this study have been challenged due to lack of blinding (Lenzer, 2004). Additional support for the effectiveness of fluoxetine in children and adolescents came from a meta-analysis that also included
unpublished data (Whittington et al., 2004). The authors found efficacy of SSRIs over placebos in the patients (5-18 years), but concluded that the risks may outweigh the benefit, except for fluoxetine. Interestingly, the abovementioned studies by Wagner et al. (2003) and Keller et al. (2001) were discussed in the meta-analysis, but paroxetine and sertraline did not yield to a benefit-risk balance in favor of the SSRI. The favorable outcome of fluoxetine in the study by Whittington et al. (2004) has been challenged due to methodological issues
(Wohlfarth et al., 2004). Thus, although these findings point to SSRI efficacy over placebo, the findings are again not established beyond doubt.
An alarming finding emerged when the study by Keller et al. (2001) was reanalyzed. Le Noury et al. (2015) used the same method (as far as possible) and the available data of the study by Keller et al. (2001) and found that paroxetine did not differ significantly from placebo in efficacy. In addition, there was an increase in adverse effects with paroxetine compared to placebo, including an increase in suicidal behavior. Thus, the authors concluded that paroxetine was not effective in treating adolescent depression and the active drug was associated with increased negative effects. The study by Keller et al. (2001) was funded by the pharmaceutical company GlaxoSmithKline (GSK). Therefore, in addition to other cases (Lexchin et al., 2003), the reanalysis by Le Noury et al. (2015) casted doubt over the reliability
of studies funded by pharmaceutical companies. It is worth mentioning that the other discussed studies that found positive effects of SSRIs in treating child and adolescent
depression (except for the study by Whittington et al. (2004)) were funded by pharmaceutical industries as well.
Although limited, meta-analyses have recently been conducted to investigate the efficacy of SSRIs in pediatric populations. The effectiveness of SSRIs over placebo for children and adolescents was found, although the effect size was small (Hedges’ g = 0.3). In addition, the authors found that the placebo response was large, and serious side effects were more common with SSRIs (Locher et al., 2017). Cipriani et al. (2016) only found effectiveness of fluoxetine over placebo and concluded that given the benefit-risks balance, SSRIs do not seem to be beneficial for children and adolescents. If any SSRI-treatment would be
considered, fluoxetine should be preferred. Nevertheless, many children and adolescents still use other SSRIs than fluoxetine (GIPdatabank, 2019; Ramchandani, 2004). Thus, the current volume and type of SSRI prescriptions in the pediatric population do not seem to be in accordance with the current scientific outcomes.
In summary, based on these findings there is some evidence for SSRIs effectiveness over placebos in the pediatric population, similarly to adults. However, this small advantage of SSRIs over placebos may be negligible, as will be discussed in the following paragraphs. In addition, side effects with SSRI use are commonly observed. Notably, the increased risk of suicidality in pediatric population is concerning. The side effects will be discussed in chapter 4.
3.4 – Placebo efficacy
As mentioned above, studies investigating antidepressant efficacy observe large placebo responses in patients suffering from depression. The response rate to placebos has
been shown to be around 30-40% (Brown, 1994; Sonawalla & Rosenbaum, 2002; Furukawa et al., 2016). Although depressed patients tend to improve over time, the spontaneous recovery cannot explain the entire magnitude of the placebo effect (Rutherford et al., 2012). Posternak and Miller (2001) found a mean decrease of 15% in symptoms for patients assigned to a 4-8 week waitlist (and thus receiving no treatment). This shows the importance of positive
expectations and contact with clinicians. In fact, is has been demonstrated that the variance in HDRS score due to medication was lower than the variance due to psychiatrist, leading the authors to conclude that the psychiatrist effects were larger than the medication effects
(McKay et al., 2006). In addition, Khan et al. (2008) demonstrated that the placebo response is not short-lived. In their study 79% of placebo responders did not relapse, compared to 93% antidepressant users. Although these groups differed significantly, there was nevertheless a large portion of placebo responders that remained well. Others have even reported lower relapse rates for placebos (33%) than for antidepressants (50%) (Shea et al., 1992).
The severity of the depression also seems to matter for the effectiveness of
antidepressants compared to placebos. Namely, drug-placebo differences seem to increase as a function of baseline severity (Kirsch et al., 2008). The efficacy of antidepressants over
placebos is negligible in mild cases (Hegerl et al., 2012; Cameron et al., 2014), but the difference is larger in severe cases, although still small (Kirsch et al., 2008).
Interestingly, effect sizes of antidepressants have declined over the years (Leucht et al., 2012; Furukawa et al., 2016). An argument to explain for this effect has been that the
responsiveness to placebos has increased over the years. Advocates of antidepressants argue that this would explain the increasing ‘failures’ in showing antidepressant effectiveness, because the detectability of an effect would be decreased (Rutherford & Roose, 2013). However, Furukawa et al. (2016) managed to replicate the effect of increased placebo responsiveness over the years, but demonstrated that when controlling for methodological
confounds (e.g., trial duration, sample size and baseline severity) the effect was not significant. This suggests that placebo responsiveness has been stable for over the years. Therefore, another possibility is that methodological advances in more recent research may explain the effect of decreased SSRI effectiveness. Furthermore, from a patient perspective, the question whether there is decreased SSRI effect or increased placebo effect is perhaps less relevant, since the goal is to alleviate the depressive symptoms, regardless of type of
treatment.
In conclusion, both antidepressants and placebos seem to lessen depressive symptoms. Although challenged, there are some indications of positive effects of antidepressants over placebos since the response rate is about 50% and 35%, respectively. However, a critical question remains: is this statistical difference between drug and placebo clinically relevant?
3.5 – Statistically significant versus clinically relevant
As mentioned by the American Statistics Association (ASA), a p-value does not measure the size of an effect and can therefore not give indication of importance. With a large enough sample size, it is possible to make any small difference between a drug versus placebo statistically significant (Wasserstein & Lazar, 2016; Bijl, 2019). For illustration, let us take the following example (for another example, see Kirsch, 2003):
This example demonstrates that a small difference in HDRS score in the active drug group compared to the placebo group can reach statistical significance. Let us assume a baseline HDRS score of 20 for both groups and a sample size of N = 50 in each group. Randomly generated in MATLAB, the group receiving the active drug had a mean HDRS score of 9.52 and a standard deviation of 3.58 after treatment, whereas the placebo group had a mean HDRS score of 11.36 and a standard deviation of 3.91 after treatment. This means a
reduction of 43,2% in HDRS score in the placebo group and a reduction of 52,4% in the active drug group. Significance between the two groups was calculated using the Wilcoxon rank-sum test and resulted in a significant effect (p = 0.012) and an effect size of d = 0.49. Thus, there was a response (50% reduction in HDRS score) in the active drug group and not in the placebo group, and the score after treatment differed significantly between the two groups. Therefore, it is concluded that the active drug is more effective than the placebo in alleviating depression.
This example demonstrates that, what seems to be a small difference in score on the HDRS, can reach statistical significance. In this example, the difference on the HDRS between the two groups is only 1.84 points, which may be considered negligible since the questionnaire may be scored differently by different psychiatrists. Yet, this difference was significant and yielded an effect size of d = 0.49, thereby favoring the active drug over the placebo. Of course, this data was simulated and should therefore not be taken too strictly.
What is the impact of this difference in practice? It could be possible that a statistically significant but small drug-placebo difference on the HDRS may in clinical settings not be observed by the patient nor the psychiatrist. To investigate the influence of small differences in HDRS score on the experienced distress of depressive patients, Leucht et al. (2013) compared HDRS scores to the Clinical Global Impressions Scale (CGI) (Guy, 1976). This scale assesses the ‘global impression’ of a patient’s mental state, thereby giving an indication of clinical relevance during treatment. The authors concluded that a reduction of three points on the HDRS corresponded to a CGI score of four, which is a score that reflects ‘no change’ in mental state. Therefore, it has been argued that at least a difference of seven points on the HDRS between antidepressants and placebo is needed to show effectiveness of antidepressant treatment over placebos. This seven point difference is considered ‘minimal improvement’ and a reduction of 14 points would be ‘much improvement’ (Leucht et al., 2013; Moncrieff &
Kirsch, 2015). Since a p-value does not give an indication of degree of effect, effect sizes are used. In the study of Cipriani et al. (2018) the reported effect size was d = 0.3. Moncrieff (2018) argued that an effect size of d = 0.3 equals an HDRS score difference of about two points, which would be interpreted as ‘no change’ on the CGI scale.
A recent meta-analysis by Jakobsen et al. (2017) investigated the efficacy of SSRIs. The authors focused on clinical relevance, defined as a drug-placebo HDRS difference of three points. Although statistical significant effects were found for the efficacy of SSRI over
placebos, there was no significant effect with the stricter three point difference in HDRS score between drug-placebo, also not in severe cases. The three point difference on the HDRS is not detected by clinicians and thus minimal (Leucht et al., 2013). Nevertheless, even with this arguably small difference in HDRS score there was no advantage of SSRIs over placebos, leading the authors to conclude that the harm of SSRIs outweigh the benefit.
3.6 - Effectiveness of selective serotonin reuptake inhibitors versus psychotherapy The studies mentioned above do not point to strong SSRI efficacy compared to placebo. Possibly, psychotherapy may be more effective than SSRIs. Although limited, various meta-analyses have been conducted to compare the effectiveness of pharmacotherapy and psychotherapy in treating depression. These studies suggests that both drug- and
psychotherapy are similarly effective in treating major depressive disorder (Casacalenda et al., 2002; Cuijpers et al., 2013). In addition, a meta-analysis by Cristea et al. (2017) found that pharmaceutical industry-funded studies indicated pharmacotherapy effectiveness over
psychotherapy, whereas non-industry-funded studies did not find a difference between the two treatments. This led the authors to conclude that the industry-funded studies were biased in favoring pharmacotherapy over psychotherapy compared to non-industry-funded studies.
Mixed results have been found for the combination of psychotherapy and
pharmacotherapy. A small advantage of the two therapies combined (compared to either one alone) has been observed in severe cases of depression (Thase et al., 1998), but another meta-analysis did not detect any advantages of combined therapy (von Wolff et al., 2012). More recently, an advantage of combined therapy as a long-term treatment was found compared to pharmacotherapy alone. Psychotherapy was similarly effective as combined therapy in alleviating symptoms in the long-term (Karyotaki et al., 2016). A possible explanation for the latter finding could be that, during psychotherapy, patients may learn coping strategies which can be deployed in future episodes of depression or help with sustained well-being. Thus, based on these studies, psychotherapy or combination therapy would be preferred over drug treatment alone.
Regarding psychotherapy as treatment alone, a meta-analysis by Cuijpers et al. (2014) found that 62% of the patients receiving psychotherapy remitted. However, of the patients who did not receive treatment, 48% remitted, suggesting a 14% additional value of psychotherapy. Thus, similar to drug-placebo studies, psychotherapy is effective, but the response in the control groups is also high. One crucial advantage of psychotherapy is that it is generally associated with less side effects, in contrast to drug treatment.
3.7 – Lack of ‘blindness’ in double-blind studies
Studies assessing therapeutic effectiveness typically have a randomized ‘double-blind’ design, meaning that subjects are assigned to either a group receiving the therapeutic
intervention or a placebo-group, and neither the subject nor the researcher knows what kind of treatment (that is, therapy or placebo) the subject is receiving. One difficulty in studies
assessing psychotherapy efficacy is that ‘unblinding’ effects may occur, due to lack of ‘placebo-therapy’ (Thase, 2003). Unblinding occurs when either the patient or researcher
discovers what kind of treatment the patient is receiving. Typically, the control group in therapeutic effectiveness studies is a waitlist-group, which makes it possible for both subject and researcher to know who has been assigned to the psychotherapeutic group. This
unblinding effect may lead to an overestimation of psychotherapy effectiveness compared to no treatment, because positive expectations may be increased in patients receiving
psychotherapy, whereas the dissatisfaction of the subjects assigned to the waitlist condition may have adverse effects on their depressive state (Hegerl & Mergl, 2010). Hence, there may be a more favorable outcome for psychotherapy in studies comparing psychotherapy to pharmacotherapy when a placebo-pill is included, thereby questioning the results of equal effectiveness of both therapies in the abovementioned meta-analyses (Hegerl & Mergl, 2010). Without a placebo-group, both therapies are presumably unblinded, hence no unblinding advantage of psychotherapy should be expected. Cuijpers et al. (2015) investigated this effect of placebo-group inclusion in studies that directly compared pharmacotherapy to
psychotherapy and found no difference between the therapies when a placebo-group was included. However, when no placebo was included, hence neither therapies were properly blinded, there was an advantage of pharmacotherapy over psychotherapy, albeit small.
By the same token, drug-placebo studies may also not be completely double-blind. Namely, the active drug produces physiological changes or side effects, allowing the subject to guess that he or she has been allocated to the pharmacotherapeutic group, since placebos produce no changes (Moncrieff, 2018). Thus, in similar fashion to psychotherapy efficacy overestimation, this may lead to an overestimation of psychoactive drug efficacy. Using ‘active’ placebos, placebos that mimic the side effects of antidepressants, Moncrieff et al. (2004) found that the differences between active drugs and active placebos were small in treating depression, suggesting that the unblinding effect of inactive placebos may overestimate the effectiveness of antidepressants.
3.8 – Confounds in causal inference
As mentioned above, it can be argued that the effectiveness of SSRIs is not established beyond doubt. However, many individuals have the impression that antidepressants have worked for them. Although there is evidence that SSRIs may be effective (Cipriani et al., 2018), the possibility also exists that the alleviating effects are due to third factors, and therefore mistakenly attributed to the drugs when consuming them. For example, two confounds in causal inference may be at play: maturation and regression toward the mean (Shadish, 2007).
Maturation refers to natural development over time that could be confused with treatment effects (Shadish, 2007). As mentioned above, the median duration of a depressive episode is about three months and the symptoms tend to spontaneously lessen over time. If however patients take antidepressants during this period and feel alleviation, chances are likely that the positive effects are attributed to the drug. Critics of antidepressants have argued that the observed delayed therapeutic effect of SSRIs could be explained by maturation rather than the working mechanisms of the drug (Bijl, 2019).
The other potential confound of causal relationship is regression toward the mean. This is the statistical phenomenon that when at the first measurement the score is extreme, the second measurement will usually be less extreme (Shadish, 2007). People suffering from depression are likely to visit a general practitioner at their depressive ‘peak’, thus when most distress is experienced (Bijl, 2019). Subsequently, the depressive symptoms tend to decrease for some individuals. Again, if antidepressants are prescribed during at the first measurement and then the symptoms are analyzed at the second measurement, chances are likely to find a reduction in distress, which is then attributed to the agents.
3.9 – Chapter conclusion
In sum, the effectiveness of SSRIs is questionable. There is a response, but this is similar to the response to placebo and psychotherapy. With a large enough sample size, significant differences can easily be achieved. This statistical difference may not be clinically relevant. The small advantage of antidepressants over other treatments could be considered in some (severe) cases and in patients who do not respond to placebos.
In consumption of psychoactive agents it is important to investigate the trade-off between benefit and risks. If the side effects are mild, one may consider using the agents. However, as discussed above, in the pediatric population the potential risks seem to outweigh the benefit. Unfortunately, also in the adult population are SSRIs (presumably the best
tolerated antidepressants) associated with serious side and withdrawal effects. The side effects will be discussed in the next chapter.
IV. Detailed summary of the side effects of selective serotonin reuptake inhibitors
4.1 – Side effects
SSRIs are presumably the best tolerated antidepressants (Cipriani et al., 2018).
Nevertheless, both physiological and psychological side effects are commonly observed. In an incidence study of SSRI side effects, 86% of patients reported at least one side effect (Hu et al., 2004). Although the severity of side effects is difficult to establish and differs per individual, Hu et al. (2004) reported that 55% of patients state that they experienced
bothersome side effects. Moreover, 17% experienced at least three bothersome side effects. An overview of the side effects of SSRIs according to the package inserts can be seen in Table 6-9 (Lundbeck Ltd, 2016; Milpharm Ltd, 2019; Ranbaxy Ltd, 2018; Lundbeck Ltd, 2019; Eli Lilly and Company Ltd, 2018; Wockhardt UK Ltd, 2019). The described side effects in the package inserts are mainly based on observations of short-term studies (Moret et al., 2009; Bijl, 2019). Although the therapeutic effect of SSRIs takes time to occur, the side effects are usually immediately noticed if they occur. Many side effects observed during SSRI treatment initiation disappear after 2-3 weeks. The long-term side effects are not well investigated.
4.2 – Psychological and nervous system side effects
Psychological and neurological side effects are among the most reported side effects of SSRIs (Table 6). This is not surprising, given that central nervous system is directly affected by the psychoactive drugs. Mood, sleep, appetite and cognitive functioning is regulated by serotonin and other monoamines in the central nervous system. Therefore, increasing serotonin levels may influence emotion, cognition and behavior. Interestingly, the most frequently reported bothersome side effect (reported by 38% of patients) in the earlier mentioned study by Hu et al. (2004) was drowsiness. This side effect is however not reported in the various package inserts (except by the package insert of fluvoxamine). The scarcity of post-marketing
research in side effects (Bijl, 2019) and the use of different terminology to describe similar side effects may contribute to this finding. Moreover, the use of various terms may downplay the incidence of side effects. For example, if nervousness, restlessness and agitation would be clustered as one phenomenon, the incidence may be higher.
Psychological side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Agitation Common Common Common Uncommon Rare Very common
Anxiety Common Rare Common Common Common Very common
Nervousness Common Common Common Uncommon Common Very common
Confusion Common Uncommon Uncommon Uncommon Rare Uncommon
Abnormal dreams Common Common Common Common Common Not reported
Apathy Common Not reported Uncommon Not reported Not reported Not reported Problems
concentrating
Common Common Common Not reported Common Not reported
Memory impairment
Common Not reported Uncommon Not reported Uncommon Not reported
Aggression Uncommon Unknown Uncommon Rare Rare Uncommon
Hallucinations Uncommon Uncommon Uncommon Rare Rare Uncommon
Mania Uncommon Rare Uncommon Unknown Rare Rare
Panic attacks Unknown Rare Not reported Uncommon Rare Not reported
Restlessness Unknown Rare Not reported Common Common Unknown
Feeling detached from oneself (depersonalization)
Not reported Rare Not reported Rare Uncommon Not reported
Suicidal thoughts Unknown Unknown Rare Unknown Uncommon Unknown
Abnormal thinking Not reported Not reported Uncommon Not reported Uncommon Not reported Drowsiness Not reported Not reported Not reported Not reported Not reported Very common
Neurological side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Headache Very common Common Very common Very common Very common Very common Insomnia Very common Common Very common Common Very common Very common
Fatigue Common Common Very common Common Very common Not reported
Sleepiness Very common Common Very common Common Common Very common
Dizziness Common Common Very common Common Common Very common
Tremors Common Common Common Common Common Very common
Tingling or numbness
Common Not reported Common Not reported Not reported Unknown Fainting Uncommon Not reported Uncommon Uncommon Unknown Not reported
Convulsions Rare Rare Uncommon Unknown Rare Rare
Grinding teeth Unknown Unknown Common Uncommon Uncommon Not reported Unusual or lack of
movements or stiffness
Unknown Uncommon Uncommon Unknown Uncommon Not reported
Involuntary movements
Rare Not reported Unknown Unknown Uncommon Not reported Muscle twitching Not reported Not reported Uncommon Not reported Uncommon Not reported Irresistible urge to
move legs (RLS)
Not reported Rare Not reported Not reported Not reported Not reported Low thyroid
hormones
Not reported Not reported Uncommon Not reported Not reported Not reported inappropriate ADH
hormone secretion
Abnormal coordination
Not reported Not reported Uncommon Not reported Uncommon Uncommon Chills Not reported Not reported Uncommon Not reported Common Not reported Fever Not reported Not reported Uncommon Common Not reported Not reported Flushing / hot flash Not reported Not reported Common Not reported Common Not reported
Weight gain Uncommon Common Uncommon Common Not reported Unknown
Increased appetite Uncommon Not reported Common Common Not reported Not reported
Weight loss Common Not reported Uncommon Uncommon Common Unknown
Loss of appetite Common Common Not reported Common Common Very common
Table 6. Psychological and neurological side effects of SSRIs as indicated by the package inserts. Very common
means in 10% or more of the users, common means 1-10% of users, uncommon means in 0,1%-1% of users, rare means in 0,01%-0,1% of users and very rare means in less than 0,01% of users. Unknown indicates that the side effect has been reported, but the frequency cannot be estimated based on the available data. Not reported means that the side effect is not reported for the given drug.
4.3 - Gastrointestinal, hepatic and metabolic side effects
Most of the bodies’ serotonin is located in the gut. It is therefore not surprising that a sudden increase in serotonin levels may commonly be accompanied by gastrointestinal and hepatic side effects (Table 7). In fact, various gastrointestinal conditions have been related to serotonin abnormalities (Camilleri, 2009).
Gastrointestinal or hepatic side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Nausea Very common Very common Very common Very common Very common Very common
Diarrhea Common Common Very common Common Very common Very common
Dry mouth Very common Common Very common Common Common Very common
Constipation Common Common Common Common Not reported Very common
Indigestion Not reported Not reported Not reported Not reported Common Very common Stomach pain Common Common Common Not reported Not reported Very common
Vomiting Common Common Common Common Common Very common
Flatulence Common Not reported Common Not reported Not reported Not reported Increased
saliva/drooling
Common Not reported Uncommon Not reported Not reported Not reported
Hepatitis Rare Not reported Not reported Unknown Rare Not reported
Abnormal liver function tests
Unknown Rare Rare Unknown Uncommon Rare
Increased cholesterol
Not reported Common Rare Not reported Not reported Not reported Yellowing of skin
and/or eyes due to liver problems
Not reported Very rare Unknown Unknown Not reported Not reported
Low blood sugar Not reported Uncommon Rare Not reported Not reported Not reported
Table 7. Gastrointestinal and hepatic side effects of SSRIs as indicated by the package inserts.
4.4 – Genitourinary side effects and post-selective serotonin reuptake inhibitor sexual
Sexual dysfunction during SSRI use is common (Table 8). In the incidence study by Hu et al. (2004) it was reported as one of the most prevalent side effects (present in 34% of patients) and in addition as most bothersome. For the majority (90-95%) this side effect will remain as long as they take the medication (Clayton & Montejo, 2006). Nevertheless, it is assumed that this side effect disappears after discontinuing the drug. Yet, there have been cases of sexual dysfunction after SSRI discontinuation (Csoka & Shipko, 2006). This
phenomenon is known as post-SSRI sexual dysfunction (PSSD). In contrast to the other side effects and withdrawal symptoms of SSRIs, the sexual dysfunction may persist and thus be chronic. Common symptoms include erectile dysfunction, premature ejaculation, genital numbness, weak orgasms and decreased libido (Bala et al., 2018). PSSD is estimated to occur in 2-16% of SSRI users. It has however been argued that this is likely to be an
underestimation, since these rates are based on spontaneous reports by subjects in pre-market research (Bahrick, 2008).
Genitourinary side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Change in sexual function*
Common Very common Very common Common Common Uncommon Irregular menstrual
period
Unknown Not reported Uncommon Uncommon Not reported Unknown Problems urinating Uncommon Uncommon Uncommon Unknown Uncommon Unknown Flow of breast milk
in men/women who are not breast-feeding
Unknown Rare Rare Unknown Rare Rare
Painful erections Unknown Very rare Rare Unknown Unknown Not reported
Table 8. Genitourinary side effects of SSRIs as indicated by the package inserts. *change in sexual function
includes delayed ejaculation, abnormal erection and decreased sex drive, among others.
4.5 – Other side effects
In addition to the abovementioned side effects, other side effects are also observed (Table 9). These include dermatologic and allergic, sensory, cardiovascular, respiratory,
musculoskeletal and hematologic side effects and can mainly be attributed to effects of SSRIs on the peripheral nervous system.
Dermatologic or allergic side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Sweating Very common Common Common Common Common Very common
Itching Common Not reported Uncommon Not reported Common Not reported Prickling of the skin Common Not reported Not reported Common Not reported Not reported Loss of hair Uncommon Not reported Uncommon Uncommon Uncommon Not reported
Rash Uncommon Uncommon Common Uncommon Common Not reported
Sensitivity to sunlight
Uncommon Very rare Unknown Not reported Rare Rare
Swelling of arms or legs
Uncommon Very rare Uncommon Uncommon Not reported Not reported
Swelling of skin Unknown Rare Uncommon Unknown Rare Not reported
Dry skin or blisters and peeling of the skin
Not reported Very rare Uncommon / Rare
Not reported Very rare Not reported
Hypersensitivity Unknown Not reported Uncommon Not reported Not reported Not reported Allergic reaction Not reported Very rare Rare Rare Uncommon Uncommon
Sensory side effect Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Ringing in the ears Common Unknown Common Uncommon Uncommon Not reported Dilated pupils Uncommon Uncommon Uncommon Uncommon Uncommon Unknown
Abnormal taste Rare Not reported Common Uncommon Common Unknown
Visual disturbance Not reported Common Common Uncommon Common Not reported Glaucoma Not reported Very rare Rare Not reported Not reported Unknown
Cardiovascular side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Palpitations Common Uncommon Common Not reported Not reported Very common Faster or irregular
heart beat
Uncommon Uncommon Uncommon Uncommon Common Very common
Slower heart beat Uncommon Rare Rare Rare Not reported Not reported
Respiratory side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Yawning Common Common Common Common Common Not reported
Blocked or runny nose
Common Not reported Uncommon Common Not reported Not reported Sore throat Not reported Not reported Common Not reported Not reported Not reported Coughing Rare Not reported Not reported Not reported Not reported Not reported
Musculoskeletal side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Muscle weakness Very common Common Uncommon Unknown Not reported Very common Pain in muscles and
joints
Common Rare Common Common Common Uncommon
Unusual or lack of movements or stiffness
Unknown Uncommon Uncommon Unknown Uncommon Not reported
Increased risk of bone fractures
Unknown Unknown Rare Unknown Unknown Unknown
Muscle tense Not reported Not reported Common Not reported Common Not reported Chest pain Not reported Not reported Common Not reported Not reported Not reported Back pain Not reported Not reported Uncommon Not reported Not reported Not reported
Hematologic side effect
Citalopram Paroxetine Sertraline Escitalopram Fluoxetine Fluvoxamine
Bleeding disorders Unknown Very rare Not reported Unknown Not reported Not reported Excessive or
unexpected menstrual bleeding
Uncommon Not reported Rare Uncommon Common Unknown
(Brief) change in blood pressure causing dizziness or fainting when standing up quickly
Unknown Uncommon Uncommon Unknown Unknown Uncommon
Low blood pressure Unknown Not reported Not reported Not reported Uncommon Not reported Decreased levels of
sodium in blood (hyponatremia)
Unknown Rare Not reported Unknown Uncommon Not reported
Nosebleed Unknown Not reported Uncommon Uncommon Uncommon Not reported
Table 9. Other side effects as indicated by the package inserts of SSRIs.
4.6 – Precautions with other conditions and medicines
The package inserts also warn for additional serious side effects which may occur in patients with certain conditions: extra caution should be warranted when patients have liver or kidney diseases, diabetes, epilepsy, bleeding disorders, low levels of sodium in the blood, heart problems, eye pressure problems or akathisia (Lundbeck Ltd, 2016; Milpharm Ltd, 2019; Ranbaxy Ltd, 2018; Lundbeck Ltd, 2019; Eli Lilly and Company Ltd, 2018; Wockhardt UK Ltd, 2019). SSRIs may trigger or worsen these conditions. For example, fits may be increased with epilepsy, glaucoma may occur with eye pressure problems, gastrointestinal bleeds may occur with bleeding disorders and torsade de pointes may occur with heart problems.
Moreover, a manic episode may be triggered in bipolar patients (Stahl, 2013). These patients are particularly prone to side effects such as agitation, mania or suicidality and it is therefore not recommended to have antidepressants as monotherapy in these cases (Stahl, 2013). Thus, it is important to distinguish bipolar depression from unipolar depression, especially since SSRI-treatment is the first-line treatment of depression. This may result to be quite
challenging, since bipolar patients seeking treatment for their depressive state may never have had a manic episode before, therefore possibly erroneously be diagnosed as having unipolar depression.
The use of SSRIs in combination with other drugs also warrants caution. Taking SSRIs in combination with serotonin increasing medicine (e.g., MAOIs, TCAs, sumatriptan (used for migraine), tramadol (painkiller), lithium, tryptophan and St. John’s Wort) should be avoided, as it may cause a condition known as ‘serotonin syndrome’ (Stahl, 2013). This condition is caused by excessive stimulation of post-synaptic serotonin receptors and is characterized by fever, agitation, confusion and trembling. In severe cases, serotonin syndrome may result in seizures, cardiovascular collapse, brain damage, coma and death (Stahl, 2013). Other drug combinations that should be avoided include SSRIs in combination with drugs that thin the blood (such as anti-coagulants, NSAIDs or aspirin, substrates of CYP 1A2), since this may increase the tendency to bleed (Lundbeck Ltd, 2016). Combining medicine that decreases potassium or magnesium in the blood with SSRIs increases the risk of heart rhythm disorders (Lundbeck Ltd, 2016). At last, medication that lowers the threshold for convulsions such as antipsychotics (substrates of CYP 1A2), tramadol and mefloquine or chloroquine (used for malaria) in combination with SSRIs may increase the risk of seizures (Lundbeck Ltd, 2016).
4.7 – Life-threatening side effects
In worst case scenario, depression may result in suicide (WHO, 2019). Therefore, a treatment goal is to prevent this. However, paradoxically, suicide is reported as a side effect of SSRIs. Compared to adults, teenagers and young adults would have an increased risk of suicidal ideation when using SSRIs. Various studies report suicidal ideation in pediatric populations (e.g., Lenzer, 2004; Wohlfarth et al., 2005; Sharma et al., 2016). The increase in suicidal behaviors in the pediatric population led the Food and Drug Administration (FDA) to add ‘black-box warnings’ to the SSRI packages in 2004. These labels warn about the
potentially harmful side effects that can occur in the pediatric population group (Leslie et al., 2004). Subsequently, the FDA expanded the investigations and in a meta-analysis it was
concluded that there was a modest increase in suicidality in children and adolescents using SSRIs (Hammad et al., 2006). However, some authors have criticized the study for
exaggerating the prevalence of suicidal tendencies, claiming that the benefit of SSRIs nevertheless outweighs the potential risks (Bridge et al., 2007; Fornaro et al., 2019).
In the so-called ‘black box warning’ of the package inserts it is stated that suicidal thoughts may increase in the first few weeks of SSRI use, before the drug starts taking effect (Lundbeck Ltd, 2016; Milpharm Ltd, 2019; Ranbaxy Ltd, 2018; Lundbeck Ltd, 2019; Eli Lilly and Company Ltd, 2018; Wockhardt UK Ltd, 2019). It is also stated that there is an increased risk if suicidal ideation already took place before using the drug and if the patient is younger than 25 years old. Thus, suicidality may also occur in adults, although the incidence is
unknown (Table 5). Concerns about increased suicidal tendencies due to SSRI treatment were already raised soon after the drugs appeared on the market (Teicher et al., 1990). It has been argued that the incidence of suicidal tendencies is doubled in adult SSRI users (Bielefeldt et al., 2016), but others only found increased risk in adolescents (Barbui et al., 2009). It should be noted that suicidal tendencies are relatively rare (even in depression), making it difficult for studies to obtain sufficient power to detect any effect of suicide risk (Gunnell et al., 2005). In addition, it is difficult to disentangle whether the tendencies are caused by medication or whether it is due to depressive symptoms. Moreover, whereas SSRI prescriptions have increased steadily since their introduction, there does not seem to be an increase in suicide in the general population since the introduction of SSRIs. This is hard to interpret however, since many factors can influence general population suicide rates (Gunnell & Ashby, 2004).
Other side effects of SSRIs that could be life-threatening are the serotonin syndrome (discussed above), hyponatremia (decreased concentration of sodium in the blood) and torsades de pointes. A particular danger of hyponatremia is that patients might not detect this deficiency, since its symptoms are commonly caused by other phenomena and can therefore
