Real risks of stopping antidepressants: over one quarter of long-term users report severe withdrawal
Official guidelines and large clinical trials say that only 3% of people experience severe symptoms when stopping antidepressants, but those trials only tested people who took the medication for eight to twelve weeks; a survey of long-term users found that among those who had been taking the drug for two years or more, one quarter reported severe withdrawal symptoms.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Nobody warned him that stopping would be this agonizing
Mark Horowitz started SSRIs in his early twenties for low mood and remained on them for a decade without ever considering stopping. When he finally decided to stop, official guidelines told him to expect at most a week or two of mild discomfort; instead, he experienced panic attacks lasting hours, severe vertigo, and eventually suicidal thoughts, requiring him to restart medication. His experience became the genesis of this debate: patients describe withdrawal as torture, while some doctors claim it's trivial.
— Noam HassenfeldScientists don't actually understand how antidepressants work in the brain
Antidepressants are commonly understood as correcting a ‘chemical imbalance’ in serotonin, but scientists have not consistently found this imbalance in the blood or cerebrospinal fluid of people with depression. Psychiatrist Philip Mitchell has stated bluntly that despite years of research, what we know about how antidepressants actually work is ‘very little.’This does not mean the medications don't work—data shows about half of users benefit—but it suggests that the intuition ‘if I've taken this for ten years, I should take it forever’lacks mechanistic support.
Large-scale reviews found that only 3% experience severe withdrawal symptoms
Two large systematic reviews—one in JAMA Psychiatry and one German analysis—synthesizing multiple clinical trials concluded that only about 3% of people experience severe symptoms after discontinuation, with most experiencing mild symptoms or none at all. This led some in the medical community to believe that claims of severe withdrawal had been refuted by data, and they argued for ‘taking the loss’and moving on.
Those trials only tested people who took the medication for 8-12 weeks
Mark Horowitz identified a crucial gap: the widely-cited clinical trials typically included only patients taking the medication for eight to twelve weeks, while most long-term antidepressant users in America have been taking the drug for at least five years. Using benzodiazepines and opioids as an analogy—where longer use produces deeper physiological adaptation and typically stronger withdrawal reactions—he argued that findings from short-term trials could not reasonably be extrapolated to long-term users.
Among long-term users, one quarter report severe withdrawal symptoms
Mark and colleagues surveyed more than 300 people from British NHS free psychological services who had attempted to discontinue antidepressants. Among those on medication for six months or less, most experienced no or mild symptoms. But among those on the drug for two years or more, roughly 40% described their symptoms as moderate, another 25% as severe; one third experienced symptoms lasting more than three months; and four out of five people who tried to stop were unable to do so successfully.
The survey itself has serious methodological limitations
Critics noted that Mark's survey suffered from survivorship bias—the survey respondents were self-selected people particularly invested in the topic, very likely those who had experienced severe withdrawal—and lacked a placebo control group. The German review found that even people who had never taken antidepressants but only discontinued placebo reported similar withdrawal-type symptoms, raising the question: whose fault are these symptoms anyway?
Stopping antidepressants doubles relapse risk, though some never relapse regardless
Professor Glyn Lewis conducted a double-blind trial with 400 patients who had been on antidepressants for two years or more and were doing well. Half were gradually tapered onto placebo, half continued the medication. After one year, 56% of the tapered group were deemed to have relapsed, but 39% of the group that continued medication also showed relapse. The conclusion: discontinuation roughly doubles relapse risk, yet many people do not relapse even after stopping, and whether the risk is worth taking is entirely a personal calculation.
Could what looks like ‘relapse’ actually be misidentified withdrawal symptoms instead
The thorniest problem in this debate is whether some of those deemed ‘relapsed’ in Glyn's trial were actually experiencing withdrawal rather than the return of depression. Clues to distinguish the two include: symptoms emerging within hours to days of stopping suggest withdrawal; physical symptoms like brain zaps, dizziness, and nausea are more specific to withdrawal; and if symptoms differ entirely from your original depression—say, you never had hourly panic attacks before but now do—that hints at withdrawal rather than relapse. Mark himself used this framework to confirm that what he experienced was withdrawal.
Tapering slowly helps, but the final steps pose the greatest risk
Research by Catherine Wallace and Mark found that slow, gradual dose reduction lowers the risk of withdrawal symptoms. The riskiest phase is the very end, when the remaining dose looks negligible, because even tiny amounts of SSRI can have significant effects on the brain. Additionally, people taking SNRIs—which regulate both serotonin and norepinephrine—typically experience more severe withdrawal side effects.
In their own words · checked verbatim
I would wake up in the mornings in full-blown panic terror, like I was being chased by a wild animal.
Mark Horowitz4:43
Despite a lot of good research over many years, we really... understand surprisingly little.
Philip Mitchell9:33
around 3% of people coming off their antidepressants would have what they called severe symptoms
For benzodiazepines, for opioids, the longer you use it, the more your body adapts to us, the more withdrawal you get when you stop it.
Mark Horowitz20:04
People who'd been on the drugs for more than two years generally had a lot of problems coming off their drugs.
Mark Horowitz21:50
Within the year, 56% were considered to have relapsed, is what the word is.
Glyn Lewis30:58
be curious before being certain
Mark Horowitz39:21
Figures
| Proportion of Americans taking antidepressants | One in nine (approximately 30 million people) | 8:12 |
| Efficacy rate of antidepressants among users | Approximately 50% | 10:40 |
| JAMA Psychiatry review: severe withdrawal proportion | Most people experience mild symptoms; proportion with severe is small | 13:27 |
| German review: severe withdrawal proportion | Approximately 3% | 18:39 |
| Mark's survey: moderate symptoms among 2+ year users | Approximately 40% | 21:45 |
| Mark's survey: severe symptoms among 2+ year users | 25% (one quarter) | 21:45 |
| Mark's survey: long-term users with symptoms lasting 3+ months | One third | 21:45 |
| Mark's survey: long-term users unable to successfully discontinue | Four out of five | 22:46 |
| Glyn Lewis trial: relapse rate in discontinuation group after one year | 56% | 30:58 |
| Glyn Lewis trial: relapse rate in continuation group after one year | 39% | 31:30 |
Glossary
- SSRI
- Selective Serotonin Reuptake Inhibitor; the most common class of antidepressants, which work by preventing the reabsorption of serotonin in the brain, thereby increasing its concentration.
- SNRI
- Serotonin-Norepinephrine Reuptake Inhibitor; a class of antidepressant that works by preventing the reabsorption of both serotonin and norepinephrine.
- brain zaps
- A common withdrawal symptom from antidepressants, described as brief electrical shock-like sensations in the head; the cause remains unclear.
- tapering
- The gradual reduction of medication dosage over time to minimize withdrawal reactions.
How to listen
Anyone currently taking or considering discontinuing antidepressants, people accompanying family members through this process, and doctors and researchers concerned with the quality of evidence for psychiatric medication.
The opening segment where Noam introduces himself (0:00-2:00) can be skipped without affecting understanding of the subsequent argument.