Discontinuation or withdrawal symptoms when antidepressants are stopped are the topic of a polarised debate, particularly on social media and in other media outlets. The sources of controversy have remained the same for many years, including discussions of how discontinuation symptoms should be defined, how often and severely people experience them, and what support should be provided (read Hannah’s Mental Elf blog to learn more about recommendations from those with lived experience).
There have been a few attempts at synthesising the evidence; for example, one previous review found that about 50% of people experience discontinuation symptoms after stopping an antidepressant (Davies & Read, 2019). However, this review had serious limitations because it included data from online surveys that are likely to over-represent people with discontinuation symptoms. Also, it is possible that people will report discontinuation effects after stopping a placebo. We therefore need to compare symptoms reported after stopping placebo with those reported after stopping an antidepressant.
Henssler and colleagues (2024) have published a timely and important systematic review and meta-analysis, which is the first to comprehensively investigate the occurrence and severity of antidepressant discontinuation symptoms. These data can be used to inform clinicians and patients about the probable extent of antidepressant discontinuation symptoms. They can also help the ongoing debate about antidepressant discontinuation symptoms.
Antidepressant discontinuation or withdrawal symptoms are a particularly polarised topic, and have been for many years. Henssler et al. (2024) are the first to undertake a comprehensive systematic review and meta-analysis of occurrence and severity.
Methods
Henssler and colleagues searched three databases for studies which had investigated discontinuation symptoms after antidepressants had been stopped. They included randomised controlled trials (RCTs) and observational studies where participants had discontinued an antidepressant, including those who switched to placebo. Studies may or may not have had a comparison group of another antidepressant, tapering (gradually reducing the dose of antidepressant) an antidepressant at a different rate, or stopping a placebo.
The authors calculated the proportion of patients who experienced any discontinuation symptom after stopping an antidepressant. They also investigated severe discontinuation symptoms. These are difficult to quantify, but the authors used definitions from the original studies. They also used withdrawal from the study as an indicator of severe discontinuation symptoms. The Newcastle Ottawa scale was used to assess risk of bias.
First, the authors analysed data separately for two groups; patients who had stopped an antidepressant and patients who had stopped a placebo. This treated all studies, including the trials, as observational. However, attempting to compare two groups that were never intended to be compared could lead to bias, so the authors also analysed data from the RCTs separately, as this analysis would provide more valid estimates because randomised groups are comparable. The authors subtracted the rate of discontinuation symptoms in the group who had stopped an antidepressant from the rate in the group who had stopped a placebo to calculate an overall percentage. The authors also investigated other variables that could, theoretically, influence the rate of discontinuation.
Results
Seventy-nine studies (44 RCTs and 35 observational studies) were included in the review, including data from 21,002 patients (72% female) with a mean age of 45 years (range: 19.6 to 64.5). Sixty-two study groups provided data for quantitative synthesis and 25 of those were rated as being at low risk of bias.
Among the participants who stopped an antidepressant, 31% (95% CI [0.27 to 0.35]) reported at least one discontinuation symptom. This was compared to 17% (95% CI [0.14 to 0.21]) who stopped a placebo. In these analyses, the antidepressant groups were a mix of study designs. However, the occurrence of symptoms that can be attributed to discontinuation can be roughly estimated as the difference between the antidepressant and placebo groups, so approximately 14%. Further, three percent (95% CI [1.4% to 5.7%]) of people experienced severe symptoms after stopping an antidepressant, compared to 0.6% (95% CI [0.2% to 1.3%]) who stopped a placebo, a difference of around 2%.
The estimates based on RCTs alone were different. Among RCTs, the rate of discontinuation symptoms was lower, with a difference of about 8% (95% CI [4% to 12%]) between active and placebo groups.
Studies that used a structured measurement of discontinuation symptoms (such as the Discontinuation Emergent Signs and Symptoms, DESS) produced a higher estimate of symptoms than those which had not used such an assessment.
The most common antidepressants, at least in the UK, are sertraline, fluoxetine and citalopram. These were associated with a lower rate of discontinuation symptoms than some of the other antidepressants, such as venlafaxine and paroxetine.
There was no evidence that other factors influenced the rate of discontinuation symptoms. These factors included:
- Whether studies were high or low risk of bias
- Whether studies used a tapering regime
- The amount of time participants were followed for, after the study had begun
- How long participants had been prescribed an antidepressant
- Participants’ diagnoses (which included any mental, behavioural, or neurodevelopmental problem)
- The presence of funding from a pharmaceutical company
However, there was a lot of variation in how these studies were conducted and the way that discontinuation symptoms were measured. We should therefore interpret any negative findings with caution.
The findings of this review suggest that between 8% and 14% of people will experience discontinuation symptoms when they stop an antidepressant, and that for 2% of this group, these symptoms will be severe.
Conclusions
The authors concluded that the study had several key findings:
First, across all studies and antidepressants, we found that approximately every third patient discontinuing antidepressants will have antidepressant discontinuation symptoms of any kind […] Second, even in studies of people receiving a placebo, discontinuation symptoms (which could be called discontinuation-like symptoms) occurred in approximately one in six patients […] Third, severe discontinuation symptoms occurred in around one in 30 patients discontinuing antidepressants.
Strengths and limitations
There are many strengths to this review, including:
- Pre-registering their plans for the study online before they began, which increases transparency and reduces the opportunity to bias results by changing plans after the results have been seen.
- Following international guidelines for systematic reviews to accurately report the important methodological steps, which aids clarity, rigour, and transparency.
- Using two independent reviewers to screen the included studies, which enhances the reliability that appropriate studies were included, and relevant studies were not missed.
- Searching databases with no date, language, or publication restrictions, reducing the likelihood of publication bias and increasing the chances of all relevant data being included.
- Investigating the potential influence of a range of variables on their findings, providing greater certainty regarding the meta-analytic findings.
However, any systematic review is limited by the quality of the studies it includes. The limitations of the studies included were:
- Most were funded by the pharmaceutical industry and many had unclear funding statements. This means that possible conflicts of interest could not be identified, which could bias results.
- Most included participants had who been receiving antidepressants for a relatively short period of time (e.g., possible that discontinuation symptoms are more frequent in people who take them long term; in fact, most people take antidepressants for the long term as maintenance (read more about this in the context of relapse in Asha’s Mental Elf blog).
- Most studies were high risk of bias, which reduces reliability.
- The authors used the definitions of discontinuation symptoms provided by the studies themselves and measurements differed widely; this adds heterogeneity to the meta-analysis and could mean that different constructs were examined between studies.
However, the authors found that funding, risk of bias, and duration of treatment had no influence on the findings.
Other limitations of the studies included:
- Data on ethnicity were not consistently reported, potentially limiting generalisable.
- There were no studies of several widely used antidepressants (e.g., mirtazapine, bupropion, amitriptyline) so the results are less applicable to current practice.
There are also several potential limitations of the review itself:
- The settings of the studies included in the review were not reported. Most studies were probably conducted in secondary care (specialist mental health services), whereas people with depression are usually managed in primary care, and findings from specialist mental health services may not generalise to this setting.
- Many of the most common discontinuation symptoms are also symptoms of depression and anxiety (e.g., fatigue, nervousness, irritability, dizziness). Because of the overlap in symptoms, it is challenging to distinguish antidepressant discontinuation symptoms from depression relapse. Neither the review, nor the studies it included, addressed this problem.
- People with lived experience were not involved in the study. This could have enhanced the interpretation of the data and the directions for future research.
Because of the overlap in symptoms, it is challenging to distinguish antidepressant discontinuation symptoms from depression relapse. Neither the review, nor the studies it included, addressed this problem.
Implications for practice
The findings of this review suggest that between 8% and 14% of people will experience discontinuation symptoms when they stop an antidepressant, and that for 2% of this group, these symptoms will be severe. The finding of discontinuation symptoms in the placebo group is also important, as it suggests that many symptoms which are classified as discontinuation are also general, non-specific and experienced in the wider population. In research studies it is essential to compare rates of discontinuation symptoms after stopping an antidepressant with the rates after stopping in a placebo group. In clinical practice, we must be careful when we infer that any symptom is causally related to stopping antidepressants. Patients might report symptoms after discontinuation, but it is difficult to be sure the symptom is caused by the drug being stopped. It is also important to reassure patients  that some of their symptoms might have occurred by chance and not as a result of discontinuation.
For patients and clinicians, the main issue is how to manage adverse symptoms that may occur after antidepressants are stopped. Psychotherapy and support via the telephone and internet can reduce the incidence of discontinuation symptoms (Kendrick et al., 2024). More severe and long-term symptoms might best be managed by tapering more slowly, or they might be symptoms of relapse, so remaining on the antidepressant might be the best option.
For individual clinicians, severe discontinuation symptoms will seem uncommon, and most patients will not experience them. However, antidepressants are prescribed to millions of people worldwide, so in aggregate, this will still affect a large number of people. This explains the many people who report discontinuation symptoms in online surveys and in the media. All patients who attempt to discontinue antidepressants should be supported, particularly those who develop severe symptoms.
Recommendations for future research
Prescriptions for an antidepressant have risen substantially in recent decades and this increase is largely due to people staying on antidepressants for longer to reduce relapse. Future research could investigate how many people experience discontinuation symptoms, given that antidepressant use has expanded since many of the studies included in this review were done and people take them for much longer than in these studies. We should also investigate the specific symptoms caused by discontinuation and whether these can be distinguished from depression relapse. There should also be more research into the kinds of tapering regimes that will minimise discontinuation symptoms.
Severe discontinuation symptoms may seem rare at 2%, but antidepressants are prescribed to millions of people worldwide, meaning that this is still a substantial issue.
Statement of interests
Gemma Lewis receives funding from NIHR and Wellcome Trust and is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (grant number 223248/Z/21/Z).
Glyn Lewis’s university receives grant funding from NIHR, UKRI and Wellcome Trust. Glyn Lewis was given travel and accommodation expenses to attend ECNP 2023.
Links
Primary paper
Henssler, J., Schmidt, Y., Schmidt, U., Schwarzer, G., Bschor, T., & Baethge, C. (2024). Incidence of antidepressant discontinuation symptoms: a systematic review and meta-analysis. The Lancet Psychiatry, 11(7), 526–535.
Other references
Bowers, H. (2024). Antidepressant withdrawal: recommendations for support from people with lived experience. The Mental Elf.
Davies, J., & Read, J. (2019). A systematic review into the incidence, severity and duration of antidepressant withdrawal effects: Are guidelines evidence-based? Addictive Behaviors, 97, 111–121. 7
Horowitz, M. A., & Taylor, D. (2019). Tapering of SSRI treatment to mitigate withdrawal symptoms. The Lancet Psychiatry, 6(6), 538–546.
Kendrick, T., Stuart, B., Bowers, H., Haji Sadeghi, M., Page, H., Dowrick, C., Moore, M., Gabbay, M., Leydon, G. M., Yao, G. L., Little, P., Griffiths, G., Lewis, G., May, C., Moncrieff, J., Johnson, C. F., Macleod, U., Gilbody, S., Dewar-Haggart, R., … Geraghty, A. W. A. (2024). Internet and Telephone Support for Discontinuing Long-Term Antidepressants: The REDUCE Cluster Randomized Trial. JAMA Network Open, 7(6), e2418383–e2418383.
Ladwa, A. (2024). Risk factors for depression relapse while on long-term maintenance antidepressant treatment. The Mental Elf.
