How they were found, how they actually work (or don't), what the trials really show, and why cross-tapering them isn't like cross-tapering benzos. Every claim sourced from peer-reviewed literature.
This page is for informational purposes only. Never stop, start, or change an antidepressant without medical guidance. Abrupt discontinuation can cause severe withdrawal. If you're considering changes to your medication, work with a prescriber — ideally one who understands tapering. If you're in crisis, contact Lifeline (AU): 13 11 14.
Not a single major class of antidepressant was designed to treat depression. Every one was an accident.
At Sea View Hospital on Staten Island, New York — one of the largest tuberculosis sanatoriums in the US — something unexpected happened in the early 1950s. Patients being treated with the experimental TB drug iproniazid (a derivative of isoniazid) began showing dramatic changes in mood and behaviour. These were patients who were dying of tuberculosis. Many had been bedridden and deeply depressed by their terminal prognosis.
"Patients were dancing in the halls, tho' there were holes in their lungs."
— Associated Press reports, early 1950s (cited in Sandler, 1990; Samouco et al., Bipolar Disorders 2023)The transformation was striking enough that the Associated Press sent photographers. The images captured something the medical establishment couldn't explain: terminal TB patients in celebratory moods, socialising, dancing — a radical change from the depressive stupor typical of sanatorium life.
"Surprising reports of improved mood, vitality, and sociability, despite the fact that the reporting patients were dying."
— Samouco et al., Bipolar Disorders (2023)Psychiatrist Nathan Kline, at Columbia and Rockland State Hospital, took notice. He began testing iproniazid on psychiatric patients and described it as a "psychic energiser" that instilled "a sense of joyousness and optimism." In his 1957 trial, 14 out of 16 depressed patients showed remarkable improvement — some became completely symptom-free. Crucially, Kline observed that it took at least five weeks before the antidepressant effect appeared.
"Deteriorated, regressed patients who had been hospitalised for long periods of time, and who had been unresponsive to other treatment, showed an improvement within a five-week period of observation. They became more alert, responsive and sociable."
— Lasker Foundation, on Nathan Kline's experiments at Rockland State HospitalIproniazid was released as Marsilid in 1958. By the end of the 1950s, over 400,000 psychiatric patients were being treated with isoniazid/iproniazid derivatives. But it was pulled from most markets by 1961 due to fatal liver toxicity (hepatotoxicity) caused by its metabolite isopropylhydrazine, and the discovery of the "cheese reaction" — hypertensive crises triggered by tyramine in fermented foods.
SRC
While Kline was discovering MAOIs from TB drugs, a parallel accident was unfolding in Switzerland. Roland Kuhn, a psychiatrist at the remote 700-bed Münsterlingen psychiatric hospital, was testing compounds from the pharmaceutical company Geigy. The backstory: chlorpromazine (the first antipsychotic) was derived from antihistamines and had revolutionised the treatment of schizophrenia in 1952. Geigy had a similar compound — imipramine — which differed from chlorpromazine by just one side chain.
Kuhn tested imipramine on several hundred patients with schizophrenia. It was a complete failure as an antipsychotic. Some patients actually got worse. But Kuhn noticed something in a subset of patients:
"The patients appear, in general, more animated, their voices, previously weak and depressed, now sound louder; they are more communicative, the lamentations and sobbing have ceased."
— Roland Kuhn, describing the first antidepressant observations (1957), from World J Biol Psychiatry (2022)Specifically, three patients diagnosed with depressive psychosis improved dramatically within weeks. The antidepressant effect was, in Kuhn's own words, "completely unexpected and its discovery entirely accidental." He wrote to Geigy on 4 February 1956, suggesting the compound might help depression. He then treated 37 more depressed patients.
"In 30% of all patients, he reported optimal results, and in around 20%, failure."
— Dr Freyhan, reporting on 46 imipramine patients at the First International Congress of Neuropharmacology, Rome (1958)Imipramine was released as Tofranil in Switzerland in late 1957, then across Europe in 1958 and the US in 1959. It became the prototype for the entire tricyclic antidepressant (TCA) class — a class of drugs all derived from what was originally an antihistamine that failed as an antipsychotic.
Kuhn freely admitted he could not explain how imipramine worked. The mechanism of action remains unknown to this day.
"Up to the present day, the core mechanism of action of antidepressants is still unknown."
— PMC, Roland Kuhn — 100th Birthday tribute (2012)SRC
Two accidental discoveries — iproniazid (from TB drugs) and imipramine (from antihistamines) — both happened to raise levels of brain chemicals called monoamines?Monoamines are a group of brain messenger chemicals that include serotonin (mood), norepinephrine (alertness/energy), and dopamine (pleasure/motivation). They carry signals between nerve cells. The "chemical imbalance" theory is based on these. (serotonin, norepinephrine, dopamine), though through different mechanisms. This coincidence led to the monoamine hypothesis of depression, formally proposed in 1965: the idea that depression is caused by not having enough of these brain chemicals.
Another piece of evidence seemed to support it: a blood pressure drug called reserpine that depletes monoamines was observed to cause depression in some patients. If depleting these chemicals causes depression, and raising them treats it, the theory seemed logical.
Nonetheless, this hypothesis drove all subsequent antidepressant development for the next 60 years — including the SSRIs.
SRC
TCAs and MAOIs worked but were dangerous — TCAs could kill in overdose, MAOIs required strict diets. The pharmaceutical industry wanted something safer and more profitable. The goal was to make a drug that selectively blocked serotonin reuptake without hitting all the other receptors TCAs affected.
Fluvoxamine (1983, UK) was actually the first SSRI marketed, followed by fluoxetine (Prozac) in 1987 (US), then sertraline (Zoloft), paroxetine (Paxil), citalopram (Celexa), and escitalopram (Lexapro). These were the first drugs "rationally designed" to treat depression — but the rational design was based on the monoamine hypothesis, which itself was never proven.
"SSRIs are thus the first class of rationally designed therapeutic drugs in psychiatry."
— Frazer (1997), as cited in ScienceDirect, Pharmacokinetics of SSRIs (1999)The irony is that while SSRIs are described as "selective," this selectivity is relative, not absolute. As we'll see in the Drug by Drug tab, each SSRI has its own unique off-target effects — some hit dopamine (sertraline), some are anticholinergic (paroxetine), some antagonise 5-HT2C receptors (fluoxetine). They are pharmacologically diverse drugs lumped under a misleading class name.
Venlafaxine (Effexor) arrived in 1993 as the first SNRI — the idea being that adding norepinephrine reuptake inhibition to serotonin reuptake inhibition would provide superior efficacy. The evidence for this superiority over SSRIs is weak. What IS clear is that venlafaxine has one of the worst withdrawal profiles of any antidepressant.
Duloxetine (Cymbalta) followed in 2004, also an SNRI, also with significant withdrawal risk. Desvenlafaxine (Pristiq) is literally the active metabolite of venlafaxine, marketed separately — a common pharma strategy when patents expire.
Bupropion (Wellbutrin), approved for depression in 1989, works primarily on dopamine and norepinephrine and barely touches serotonin — making it a fundamentally different drug from SSRIs/SNRIs.
Mirtazapine (Remeron), approved 1996, works by blocking receptors rather than inhibiting reuptake — another entirely different mechanism.
The newest entries include vortioxetine (Trintellix, 2013), marketed as "multimodal," esketamine (Spravato, 2019), a nasal spray ketamine derivative for treatment-resistant depression, and dextromethorphan-bupropion (Auvelity, 2022), which acts on NMDA glutamate receptors. These newer drugs suggest the field is finally moving beyond the monoamine hypothesis — though esketamine's approval was controversial and its long-term data limited.
TB drug → accidental euphoria → MAOIs. Antihistamine → failed antipsychotic → accidental antidepressant → TCAs. Unproven hypothesis from accidents → SSRIs/SNRIs. 70 years later, the mechanism of action is still unknown, the foundational hypothesis is unconfirmed, and the most important trial's results may have been inflated by half.
The "chemical imbalance" story told to hundreds of millions of patients. What does the evidence actually say?
You've probably heard it from a doctor, seen it on a website, or been told by a friend: "Depression is caused by a chemical imbalance in your brain. This pill corrects it." It's simple. It's reassuring. It sounds like science. And it has been told to hundreds of millions of people worldwide since the late 1980s — exactly when SSRI antidepressants launched and needed a selling point.
The only problem is that it was never proven.
"The popularity of the chemical imbalance idea of depression has coincided with a huge increase in the use of antidepressants. Prescriptions for antidepressants have sky-rocketed since the 1990s, going from being rare to a situation now where one in six adults in England and 2 percent of teenagers are prescribed an antidepressant in a given year."
— Moncrieff & Horowitz, press release for the umbrella review (UCL News, 2022)The NHS website stated: "it's thought that SSRIs work by increasing serotonin levels in the brain." Versions of this explanation were given to patients in consulting rooms worldwide. The Royal College of Psychiatrists eventually removed all references to "chemical imbalances" from their website — but only after the Moncrieff review made international headlines.
"I had been taught that depression was caused by low serotonin in my psychiatry training and had even taught this to students in my own lectures. Being involved in this research was eye-opening and feels like everything I thought I knew has been flipped upside down."
— Dr Mark Horowitz, psychiatry trainee and co-author of the umbrella reviewPublished in Molecular Psychiatry (July 2022), downloaded 1.8+ million times, cited 839 times. This wasn't a new experiment — it was an umbrella review?An umbrella review is a "review of reviews" — the highest level of evidence in science. Instead of looking at individual studies, it gathers ALL the previous reviews and analyses on a topic and asks: taken together, what do they actually show?, which gathered up ALL the previous research on serotonin and depression and asked: after decades of research, does the evidence actually support the theory?
They looked at six different ways researchers have tried to link serotonin to depression:
"The main areas of serotonin research provide no consistent evidence of there being an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity or concentrations."
— Moncrieff et al., Mol Psychiatry 28, 3243–3256 (2023)"It is always difficult to prove a negative, but I think we can safely say that after a vast amount of research conducted over several decades, there is no convincing evidence that depression is caused by serotonin abnormalities, particularly by lower levels or reduced activity of serotonin."
— Professor Joanna Moncrieff, lead author (UCL News)SRC
In June 2023, 36 researchers led by Jauhar et al. at King's College London published a formal critique in the same journal. Their core arguments:
"Perhaps the worst flaw of the 2022 review is that it casts doubts on the effectiveness of antidepressant medication, without presenting any evidence in support of this conclusion. There is incontrovertible evidence that antidepressant medications are effective in treating people with clinically-significant depression."
— Professor Carmine Pariante, co-author of the rebuttal (KCL News, 2023)It's important to note: even the rebuttal authors don't defend the simple "chemical imbalance" version of the hypothesis. They argue serotonin is "implicated" in depression, but in a much more complex way than "low serotonin = sad."
SRC
A separate commentary (PMC, 2023) noted important nuances. The serotonin hypothesis isn't one hypothesis — it's a bundle of related sub-hypotheses. Not confirming the bundle doesn't mean no individual sub-hypothesis has merit. The complexity of the topic is real.
However, the commentary also noted something important about Moncrieff's overreach:
"This treatment-related conclusion is highly problematic and cannot be directly inferred from the result of the umbrella review. Furthermore, the efficacy of antidepressant treatment, including serotonergic antidepressants, is well supported by the evidence. In principle, we can view the efficacy of antidepressants independently from the validity of the serotonin theory of depression."
— PMC (2023), "Is the serotonin hypothesis still relevant? Methodological reflections"This is a key point: a drug can "work" (produce a measurable effect) even if the theory about WHY it works is wrong. Aspirin reduced fevers long before anyone understood prostaglandin synthesis. But the critical difference is that patients are being TOLD the mechanism is known and proven, when it isn't.
"We do not understand what antidepressants are doing to the brain exactly, and giving people this sort of misinformation prevents them from making an informed decision about whether to take antidepressants or not."
— Professor Joanna Moncrieff (UCL News, 2022)"Thousands of people suffer from side effects of antidepressants, including the severe withdrawal effects that can occur when people try to stop them, yet prescription rates continue to rise. We believe this situation has been driven partly by the false belief that depression is due to a chemical imbalance. It is high time to inform the public that this belief is not grounded in science."
— Professor Moncrieff, UCL press release (2022)Current alternative hypotheses include: HPA axis dysregulation, neuroinflammation, gut-brain axis effects, neuroplasticity changes, BDNF/neurotrophin signalling, glutamate system dysfunction, epigenetic changes, and psychosocial factors. The field has no consensus.
Approved on weeks. Prescribed for decades. The foundational trial may have been inflated by half.
The 2025 analysis in The American Journal of Medicine (Ward, Haslam & Prasad) examined 52 studies of the 10 most commonly prescribed antidepressants. What they found is staggering:
25% of US antidepressant users have taken them for over 10 years — approximately 8.8 million adults. Yet not a single trial has evaluated outcomes beyond one year. The FDA considers 6-8 week trials adequate for regulatory approval.
"A substantial discordance exists between the typical 8-week duration of clinical trials and the median 5-year real-world use of antidepressants. This gap, compounded by inadequate monitoring for withdrawal effects and post-treatment outcomes, raises important questions about the evidence supporting current long-term prescribing practices."
— Ward, Haslam & Prasad, Am J Med (2025)SRC
What is STAR*D? The biggest, most influential antidepressant study ever done. The US government paid for it. 4,041 depressed outpatients at 41 clinics. Published in 2006. It tested what happens when the first antidepressant doesn't work — do you switch? Add another drug? Try therapy?
What did they claim? That after trying up to four different treatments, 67% of patients got better (remission?Remission means symptoms are almost completely gone — not just "a bit better" but actually feeling well. It's a higher bar than "response" which just means symptoms reduced by half.). This number became gospel. Textbooks cited it. Guidelines quoted it. Doctors told patients: "Don't worry, with persistence, two-thirds of people recover."
What actually happened? In 2023, independent researchers (Pigott et al.) got hold of the original raw data and re-analysed it. What they found was explosive:
"In contrast to the STAR*D-reported 67% cumulated remission rate after up to 4 antidepressant treatment trials, the rate was 35.0% when using the protocol-stipulated HRSD and inclusion in data analysis criteria."
— Pigott et al., BMJ Open (2023)"For us in psychiatry, if the BMJ authors are correct, this is a huge setback, as all of the publications and policy decisions based on the STAR*D findings that became clinical dogma since 2006 will need to be reviewed, revisited, and possibly retracted."
— Psychiatric Times (2023), "STAR*D Dethroned?"The original researchers defended themselves in the American Journal of Psychiatry (Dec 2023). The controversy is ongoing. But the implications are enormous: the single study that told psychiatry "keep trying, most people get better" may have overstated the success rate by nearly double.
SRC
The Cipriani 2018 Lancet meta-analysis — the largest ever (522 studies, 116,477 people) — found all 21 antidepressants were "statistically better than placebo." But what does that actually mean for a real person deciding whether to take one?
"Is exposure to the adverse effects of drug treatment worth the 12% chance of a better outcome? Or to put it another way: 12% of patients will benefit from the treatment, while the remaining 88% will suffer the adverse effects of treatment without any additional therapeutic benefit beyond placebo."
— Robert Whitaker's analysis of Cipriani data (2018)Cochrane review (primary care): NNT for SSRIs was 7 (48% drug vs 42% placebo response). NNT for TCAs was 9. But the number needed to harm (NNH) — withdrawal due to side effects — was 4-30 for TCAs and 20-90 for SSRIs.
For children and teenagers: The 5 most recent studies (2010-2018) showed drugs worked in 65% vs sugar pill in 60% — a 5% gap that researchers called "clearly not clinically meaningful."
The "they can tell" problem: Antidepressants cause obvious physical effects — dry mouth, nausea, sexual changes, jitteriness. In theory, studies are "double-blind" (neither patient nor doctor knows who got the real drug). But when the drug causes noticeable side effects, both sides can often guess. Research by Irving Kirsch found 78% of patients and 88% of psychiatrists correctly guessed who was on the drug vs the sugar pill. If everyone knows who's getting what, the "blind" test isn't blind anymore — and the placebo effect gets smaller because placebo patients lose hope, inflating the apparent drug benefit.
SRC
How "maintenance therapy" is justified: Take a group of people who've been on antidepressants for months. Keep half on the drug. Abruptly switch the other half to a sugar pill (without telling them). Wait and see who gets worse. When more people in the sugar-pill group deteriorate, conclude: "See? They need to stay on the drug."
The obvious flaw: When you suddenly yank a brain-altering drug from someone after months of use, the resulting withdrawal symptoms — anxiety, insomnia, crying spells, mood swings — look almost identical to the original depression coming back. These studies can't tell the difference between "the depression returned because you stopped the drug" and "the drug created a physical dependence and stopping it made you feel awful."
"Logically, whether withdrawing a drug is associated with harm is different than whether the initial long-term use was associated with sustained clinical benefit."
— Ward, Haslam & Prasad, Am J Med (2025)They are NOT interchangeable. Each drug hits different targets despite sharing a class label. Click any drug to expand.
S-enantiomer of citalopram. Classified by some as an "ASRI" (allosteric serotonin reuptake inhibitor). Several meta-analyses show marginally better efficacy and tolerability vs other SSRIs — though the clinical significance of this difference is debated. SRC
Short half-life + no active metabolite + anticholinergic + non-linear kinetics?Most drugs have "linear kinetics" — double the dose, double the blood level. Paroxetine doesn't work like that. At higher doses, small dose increases cause big jumps in blood level. Going the other way, small dose cuts can cause unexpectedly large drops. This makes tapering unpredictable. = worst withdrawal of any SSRI. Pharmacovigilance data consistently ranks it highest-risk. Non-linear kinetics mean small dose reductions can produce disproportionately large drops in blood levels.
SRC
The only SSRI with clinically relevant dopamine transporter binding. This makes it pharmacologically distinct from every other SSRI. It cannot be assumed to be interchangeable. FDA-approved for PTSD (along with paroxetine). SRC
The cultural icon. Only SSRI approved for children 8+ (FDA). Long half-life = lowest withdrawal risk but enzyme interactions persist 5-6 weeks after stopping. The least selective SSRI — it's a 5-HT2C antagonist, which may explain its activating profile (and may also explain why the "bridge to Prozac" strategy is problematic — it's not just hitting SERT like the drug you're replacing). SRC
Most selective SSRI for serotonin over norepinephrine. But carries the highest QT prolongation?QT prolongation is a heart rhythm problem. The "QT interval" is a measurement on a heart trace (ECG) that shows how long the heart takes to reset between beats. If it gets too long, the heart can develop dangerous irregular rhythms. This is why citalopram has a dose cap — higher doses increase this risk. risk of all SSRIs — dose capped at 40mg (20mg in elderly) due to cardiac arrhythmia risk. R-enantiomer actually inhibits the S-enantiomer's binding, which is why escitalopram (pure S) was developed. SRC
Primarily used for OCD?Obsessive-Compulsive Disorder — a condition involving intrusive, unwanted thoughts (obsessions) and repetitive behaviours (compulsions). Several SSRIs are used for OCD, but fluvoxamine and clomipramine are considered particularly effective for it. rather than depression. Its strong sigma-1 receptor activity may contribute to cognitive effects. Potent CYP1A2 inhibition means dangerous interactions with caffeine, theophylline, and clozapine. Short half-life means higher withdrawal risk than fluoxetine or sertraline. SRC
5-hour half-life + dual serotonin/norepinephrine effects = severe, rapid-onset withdrawal even from missed doses. Pharmacovigilance data consistently ranks it alongside paroxetine as highest-risk.
SRC
More balanced SERT/NET ratio than venlafaxine but still significant withdrawal risk. Enteric-coated capsules that cannot be split or opened make gradual tapering extremely difficult without compounding pharmacy access — a practical disaster for tapering. SRC
Desvenlafaxine IS the active metabolite of venlafaxine (O-desmethylvenlafaxine), marketed separately — a common pharma strategy when patents expire. There is no evidence it is clinically superior to venlafaxine. Same withdrawal risk.
The most effective antidepressant in the Cipriani 2018 meta-analysis — but also among the least tolerable and most dangerous in overdose. A "pharmacological shotgun" hitting serotonin, norepinephrine, histamine (sedation/weight gain), muscarinic (dry mouth/constipation/urinary retention), and alpha-1 (orthostatic hypotension). Now more commonly used at low doses for pain, migraine, and insomnia than for depression. SRC
Kuhn's accidental discovery. The prototype TCA. Effective even in severe, treatment-resistant depression. Can cause mania/hypomania in up to 25% of bipolar patients. Narrow therapeutic index makes overdose lethal. Still used for enuresis in children and in veterinary medicine. SRC
Better tolerated than amitriptyline due to less antihistamine activity. One of the few TCAs with a well-defined therapeutic window. Often the TCA of choice when an SSRI/SNRI has failed. STAR*D Step 3 tested it alongside mirtazapine.
The most potent serotonin reuptake inhibitor among the TCAs — essentially a "dirty SSRI" with all the TCA side effects. Gold standard for OCD treatment. Cannot be cross-tapered with SSRIs due to extreme serotonin syndrome risk. SRC
The most potent H1 antihistamine among TCAs. At antidepressant doses (75-300mg) it's sedating and causes significant weight gain. Remarketed at ultra-low doses (3-6mg as Silenor) specifically for insomnia, where only the antihistamine effect is active.
A hydrazine derivative. Considered highly effective for atypical depression?Atypical depression is a subtype where instead of the "classic" symptoms (can't eat, can't sleep), people oversleep, overeat, feel heavy in their limbs, and are extremely sensitive to rejection. It's actually quite common — some estimates suggest it's the most common subtype of depression. (oversleeping, overeating, rejection sensitivity). Requires strict dietary tyramine avoidance — cheese, aged meats, fermented foods, wine, beer. Hypertensive crisis?A sudden, dangerous spike in blood pressure. With MAOIs, this happens because the drug blocks the enzyme that normally breaks down tyramine (found in aged cheese, cured meats, wine, etc). Without that enzyme, tyramine floods your system and causes your blood pressure to skyrocket. This can cause stroke or death. It's why MAOIs come with strict food rules. from tyramine can be fatal. Cross-tapering with any serotonergic drug requires minimum 2-3 weeks washout (5+ weeks from fluoxetine). SRC
Structurally related to amphetamine. More activating/stimulating than phenelzine. Same dietary restrictions. Used in STAR*D Step 4 as a last-resort option vs venlafaxine+mirtazapine combination. Same strict washout requirements apply.
The "safer MAOI" — reversible binding means tyramine can still compete for the enzyme, greatly reducing hypertensive crisis risk. Available in Australia but not the US. A meta-analysis found it as effective as SSRIs and non-selective MAOIs for depression. Switchout to other antidepressants requires less washout than irreversible MAOIs but caution is still needed. SRC
Fundamentally different from every SSRI/SNRI — it barely touches serotonin. Minimal serotonergic withdrawal. Also marketed as Zyban for smoking cessation. The go-to when SSRI sexual side effects are intolerable. Combined with dextromethorphan in the newer Auvelity. Dose-dependent seizure risk limits its ceiling dose. SRC
Works by blocking receptors rather than inhibiting reuptake — a fundamentally different mechanism from SSRIs. Paradoxically more sedating at 15mg than 30mg (because at higher doses, noradrenergic activation counteracts H1 sedation). Used in STAR*D Step 3 (remission 12%) and Step 4 (combined with venlafaxine). Cannot be treated as interchangeable with SSRIs. Often combined with SSRIs ("California Rocket Fuel") — a practice with limited controlled evidence. SRC
Mostly used as a low-dose sleep aid now. At antidepressant doses (300mg+) it was one of the least effective in the Cipriani meta-analysis. Can cause priapism — prolonged, painful erection requiring emergency medical treatment.
The newest marketed "traditional" antidepressant (2013). Hits multiple serotonin receptor subtypes beyond just SERT inhibition. Some evidence for cognitive benefits. Branded as having lower sexual side effects than SSRIs. Less long-term data than older drugs. Not an allosteric mechanism — its multi-receptor profile is unique.
The S-enantiomer of ketamine. First truly non-monoamine antidepressant approved. Rapid onset (hours, not weeks). Must be administered in a certified clinic with 2-hour monitoring. Significant dissociative side effects. Expensive. Approval was controversial due to modest efficacy in some trials and lack of long-term data. Represents a genuine paradigm shift in mechanism if not in practice.
Another non-monoamine approach. Dextromethorphan (yes, the cough suppressant) acts on NMDA receptors and sigma-1 receptors. Bupropion is included primarily to inhibit CYP2D6 metabolism of DXM, keeping DXM levels therapeutic. Approved 2022. Very limited long-term data. May represent the beginning of the post-monoamine era — or may be another iteration of limited efficacy with novel marketing.
Every drug above does different things to your brain. Sertraline hits dopamine, paroxetine blocks acetylcholine, fluoxetine antagonises a specific serotonin receptor, bupropion barely touches serotonin at all, mirtazapine blocks receptors instead of blocking transporters. Calling them all "antidepressants" and treating them as interchangeable is like calling a sedan, a motorcycle, and a boat all "vehicles" and expecting one to substitute for another. There are no validated equivalence tables for antidepressants — because they're not equivalent. When someone tells you "just switch to this one," remember that your brain has adapted to a specific drug with a specific profile, and the replacement is doing something different.
Clinical trials focus on physical symptoms. Patients report something far more disturbing.
New Zealand. The largest survey of antidepressant users' self-reported experiences:
36% were told NOTHING about adverse effects. Fewer than 1% were warned about emotional numbing. NONE were told about feeling less like themselves or caring less about others. When people report these effects, doctors often attribute them to "the depression."
SRC
Emotional blunting — feeling "flat," "zombie-like," unable to cry, unable to feel joy, unable to feel anger, unable to feel much of anything — is arguably the most common and most distressing subjective side effect of SSRIs and SNRIs. It goes far beyond the physical side effects listed in package inserts.
"Some experienced little benefit from antidepressants while the side effects, particularly more subtle psychological effects such as feeling numb or 'not like themselves,' seemed a significant issue. The research also raised concerns that, in some cases, managing the side effects of antidepressants might take priority over managing the depression."
— PMC qualitative analysis of 1,747 AD users (2016)A study of 3,243 online user reviews found emotional/behavioral effects were the most commonly reported category (41%) — more than sleep, GI, or sexual effects. And crucially, emotional/behavioral effects had the strongest relationship with user dissatisfaction, far more than physical side effects.
"44% of the overall sample reported some degree of negative experience. Participants were also concerned about the medication undermining the legitimacy of their suffering and undermining their sense of control."
— PMC 4863327 (2016)SRC
Post-SSRI Sexual Dysfunction (PSSD)?A condition where sexual side effects that started while taking an antidepressant DON'T go away after you stop the drug. It can include numbness of genitals, inability to feel pleasure from sex or orgasm, loss of sex drive, and erectile dysfunction. For some people it lasts months, for others it may be permanent. is a condition where sexual problems that started during antidepressant use don't go away after stopping the medication. For some people, these effects persist for months or years. For some, they may be permanent.
This isn't just "low sex drive." Symptoms include: physical numbness of genitals (as in, reduced actual sensation), inability to orgasm or orgasms with no pleasure, erectile dysfunction, and complete loss of libido. Increasingly, researchers are also recognising non-sexual symptoms like emotional flatness and inability to feel pleasure in general.
"In an undetermined number of patients, sexual function does not return to pre-drug baseline after stopping SSRIs. The condition is known as post-SSRI sexual dysfunction (PSSD) and is characterised most commonly by genital numbness, pleasureless or weak orgasm, loss of libido and erectile dysfunction."
— PMC, Epidemiology & Psychiatric Sciences (2024)How common is it? Nobody knows for sure — partly because patients are embarrassed to mention it, doctors often dismiss it, and many people don't realise their sexual problems are connected to the drug they took years ago. But the studies that do exist paint a concerning picture:
"PSSD gained official recognition after the European Medicines Agency concluded that PSSD is a medical condition that persists after discontinuation of SSRIs and SNRIs."
— ScienceDirect, Sexual Medicine Reviews (2021)The first report to a medicines regulator (MHRA) was in 1991 — over 30 years ago, involving fluoxetine. The DSM-5 (2013) acknowledged the persistence. The EU updated product labelling. Yet many clinicians still dismiss patient reports of PSSD or attribute symptoms to depression.
"The issue of antidepressant-induced sexual dysfunction illustrates the limitations of the paradigm of short-term clinical trials with regard to drug safety. PSSD is sufficiently rare to make most clinical trials under-powered to discover it."
— PMC, Annals of General Psychiatry (2023)SRC
The iSPOT-D study (Nature, Translational Psychiatry 2021) tracked side effects throughout antidepressant treatment. Key findings:
Which antidepressant was prescribed (sertraline, escitalopram, or venlafaxine) did NOT significantly impact side effect severity — suggesting the problem is class-wide, not drug-specific.
SRC
Real, common, can be severe, and can last months to years. The field fought for decades over even calling it "withdrawal."
The 2024 Lancet Psychiatry review (Henssler et al.) is the most rigorous analysis of antidepressant withdrawal to date. Earlier reviews (Davies & Read, 2019) reported that roughly 56% of people experience withdrawal symptoms, and about 46% of those rate them as severe.
Which drugs are worst for withdrawal? (Based on multiple large-scale analyses of patient reports):
Symptoms typically begin within 2-4 days of stopping or reducing and can include: brain zaps?Electric shock-like sensations in the head, often triggered by moving your eyes. One of the most distinctive and commonly reported withdrawal symptoms. Feels like a brief "zzt" or jolt inside your skull. Not dangerous but extremely unpleasant and disorienting. (electric shock feelings in your head), dizziness, nausea, flu-like symptoms, insomnia, vivid nightmares, anxiety, irritability, depersonalisation?Feeling detached from yourself — like you're watching your life from outside your body, or like you're not "real." A deeply unsettling sensation that can be mistaken for a psychiatric disorder rather than a withdrawal effect. (feeling detached from yourself), akathisia?An intense, unbearable inner restlessness — a feeling that you MUST move, pace, or do something but nothing helps. Often described as the worst symptom of withdrawal. Has been linked to suicidality. Frequently misdiagnosed as anxiety or agitation. (unbearable inner restlessness), and electric shock sensations. Over 50 distinct symptoms have been reported.
"Serotonin reuptake inhibitors (SRIs) such as SSRIs and SNRIs are more strongly linked to withdrawal reactions than non-serotonergic antidepressants such as mirtazapine and — especially — bupropion."
— Psychiatric News, APA (2025)SRC
While most acute withdrawal settles within 1-6 weeks, a subset of people experience protracted withdrawal — symptoms that persist for months or even years after stopping the drug.
"In some cases, the withdrawal syndrome is characterised by the return of the original illness at a greater intensity than before treatment, with additional clinical features and/or the occurrence of psychiatric disorders that never occurred before. This latter symptomatology may persist for months or even years, leading to what has been termed persistent post-withdrawal disorder."
— Fava & Cosci, J Clin Psychiatry (2019)This is critical: the symptoms of protracted withdrawal can include new psychiatric symptoms the person never had before starting the drug — anxiety disorders, panic, akathisia, cognitive problems — raising the question of whether the drug itself created new pathology rather than treating an existing one.
"Observational data reliably indicate greater overall withdrawal risk with prolonged (e.g., >6 months) use and even greater risk — particularly of severe or protracted forms — with longer-term (e.g., >2 years) use."
— Psychiatric News, APA (2025)SRC
For decades, the pharmaceutical industry and mainstream psychiatry insisted on the term "discontinuation syndrome" rather than "withdrawal" — arguing that antidepressants are "not addictive" and therefore can't cause "withdrawal." This was a semantic distinction with real clinical consequences: if it's just "discontinuation," it sounds temporary and manageable.
"The term 'discontinuation syndrome' minimizes the vulnerabilities induced by SSRI and should be replaced by 'withdrawal syndrome.' Gradual tapering does not eliminate the risk of withdrawal reactions. Indeed, a significant advantage of gradual tapering compared to abrupt discontinuation did not emerge."
— Fava et al., as cited in Psychiatric Times (2025)The APA claimed withdrawal symptoms "do not require treatment and resolve within two weeks." Multiple systematic reviews have found this claim is not evidence-based — symptoms can be debilitating and extend far beyond two weeks.
"An average of 56.4% of people said they had antidepressant withdrawal symptoms; and of these people, about 46% rated their symptoms as severe. The duration of these symptoms ranged from a few days to several months."
— Psychiatric Times (2025), citing Davies & ReadPET imaging studies?Brain scans that show exactly how much of a drug is sitting on your brain's receptors at any given dose. (Horowitz & Taylor, Lancet Psychiatry 2019) showed that the relationship between dose and brain effect follows a curve, not a straight line:
"Drug effects rise steeply at doses near the lower end of the approved dose range. But at higher doses and concentrations, many drug effects plateau. Hyperbolic dose tapering attempts to mimic this relationship inversely."
— Therapeutics Letter 157 (July 2025)This means linear dose reductions (40→30→20→10→0) are pharmacologically irrational. The last cuts should be the smallest. A 10% reduction per month of the most recent dose is one commonly suggested approach, but even this is a rough approximation. Practical implementation often requires liquid formulations, compounding pharmacies, or bead-counting from capsules.
SRC
The psychiatric establishment's research is deeply flawed. But the counter-movement's research — Horowitz, Sørensen, the Maudsley Deprescribing Guidelines — also has serious problems that the harmed patient community is expected to accept with gratitude.
For decades, people suffering from psychiatric drug withdrawal were told it wasn't real. That it was "relapse." That it would pass in two weeks. That they were imagining it. The psychiatric establishment dismissed, minimised, and gaslit millions.
Then came Mark Horowitz, David Taylor, Anders Sørensen, and others who took withdrawal seriously. They published in top journals. They developed guidelines. They validated what patients had been screaming about for years. This was an enormous step forward, and it would be dishonest not to acknowledge that.
But here's the problem: the patient community is in the position of someone who has been begging for clean water and is handed a glass that's half-full with some debris floating in it. You're expected to be grateful — and you are, genuinely — but that doesn't mean the water is clean. And pointing out the debris gets you labelled as ungrateful, "anti-science," or unreasonable.
The truth is that the best available tapering guidance — from both the Maudsley Deprescribing Guidelines and Crossing Zero — has significant scientific gaps that its authors acknowledge in the fine print but don't emphasise in the headlines. These gaps matter because real people are making medical decisions based on this guidance.
The entire idea behind "slow tapering" — both in the Maudsley Guidelines and Crossing Zero — comes from brain scan studies called PET scans?PET = Positron Emission Tomography. A type of brain scan that can show how much of a drug is sitting on your brain's receptors at any given dose. It's like a camera that shows which parking spots in your brain are "occupied" by the drug. that measured how much of the brain's serotonin transporters?Serotonin transporters (SERT) are like little vacuum cleaners on your nerve cells that suck up serotonin from the gaps between nerves. SSRIs work by blocking these vacuum cleaners, leaving more serotonin floating around. The more transporters that are blocked ("occupied"), the stronger the drug's effect. are "occupied" (blocked) at different doses. The relationship isn't straight-line — it follows a curve. The first tablet you take blocks MOST of your transporters. Doubling the dose barely blocks any more. This is why the last milligrams are the hardest to taper — each small reduction at the bottom of the dose range frees up a big proportion of your receptors.
The principle is solid. The problem is WHO they studied to get these numbers.
Sørensen's research — the most comprehensive collection of this data — looked at 17 brain scan studies involving 294 people across 10 different antidepressants. Sounds impressive. But dig into who those 294 people actually were:
"As many patients take antidepressants for years, studies should also study patients after long treatment duration."
— Sørensen's own systematic review, acknowledging the limitationCritically, the data they DID have showed that receptor occupancy increases with longer use: duloxetine 20mg showed 65% occupancy after a single dose but 78% after just four days. Citalopram 20mg went from 73% (single dose) to 80% (after 25 days). What happens after 5, 10, or 20 years? Nobody knows. The occupancy curves that inform every tapering table and every "percentage reduction" recommendation are extrapolations from short-term data on healthy people.
The leap from "this is how receptors are occupied in healthy volunteers taking single doses" to "therefore this tapering schedule prevents suffering in people who've been on drugs for 10 years" is unproven.
SRC
The Maudsley Deprescribing Guidelines (Horowitz & Taylor, 2023) is the most influential tapering guide in the world right now. Adopted by the RACGP (Australia's GP college), Queensland and NSW Health, and the NHS. It provides tables telling doctors exactly what dose to prescribe at each step of a taper. The basic principle — slow down as you get lower — is presented clearly. Patient experience is acknowledged.
But when you actually check the maths in those tables — the numbers that GPs across Australia and the UK are following — the problems show up:
SRC
Anders Sørensen's Crossing Zero (2023, 413 pages) takes a different approach from the Maudsley: instead of fixed tables, it teaches the principle of 3-10% reductions from the current dose, adjusted based on how you feel. It covers all psychiatric drug classes. It validates withdrawal as real and physiological. It discusses windows and waves, interdose withdrawal, and reinstatement. The psychology chapters are genuinely interesting.
The problems:
SRC
As of early 2026, there are zero large-scale clinical trials proving that "hyperbolic tapering" actually produces better results than other slow methods.
The 2022 review of clinical guidelines on tapering found that the evidence for ANY specific method is "very low certainty."
Perhaps most striking: the Lancet Psychiatry (2026) analysis found that slow tapering on its own — without psychological support — showed no significant difference from just stopping abruptly for preventing depression returning. What DID make a difference was having psychological support alongside the taper. The human element — having someone who understands what you're going through — mattered more than the specific dose-reduction method.
SRC
The mainstream psychiatric establishment spent decades denying withdrawal was real, approving drugs based on 8-week trials, and telling patients their suffering was "relapse." The evidence base for prescribing antidepressants long-term is built on sand.
The reformers — Horowitz, Taylor, Sørensen, Moncrieff — have done genuinely important work. They validated withdrawal. They established the principle of gradual dose reduction. They challenged the serotonin hypothesis. They published in top journals. They moved the needle.
But the specific tapering guidance they've produced is built on PET studies of healthy volunteers taking single doses, extrapolated to chronic users through unvalidated mathematical models, presented in tables constrained by pill fractions, and tested in zero RCTs. The benzo data comes from monkey studies. The Maudsley tables don't follow their own stated mathematics. Both Horowitz and Sørensen acknowledge in their actual papers that their approaches are untested hypotheses — but the books and clinical guidelines built on those papers don't carry these caveats with the same prominence.
The patient community is left in the worst of all positions: dismissed by mainstream medicine, and expected to be grateful for alternative guidance that its own authors admit is unproven. Neither side has the evidence to fully support what they're telling patients to do.
What patients actually need is what they've always needed: honest communication about uncertainty. "This is the best guidance we have, it's based on these principles, but the specific protocols are unvalidated and you need to be closely monitored by someone who understands that" is a more honest starting point than either "withdrawal isn't real" OR "follow this table and you'll be fine."
Read both books. Use the principles (gradual reduction, smaller cuts as you go lower). But supplement heavily with community knowledge, your own body's signals, and a clinician who's actually walked patients through complex withdrawal. No table, no curve, no percentage can replace attentive clinical judgment and the lived expertise of people who've been through it. The full detailed reviews of both books — including verified math and reference checks — are available at guythetaperman.com (Maudsley) and guythetaperman.com (Crossing Zero).
Every claim in this document is traceable. 80+ peer-reviewed and institutional sources.
Discovery & History
Kuhn R. Schweiz Med Wochenschr 1957;87:1135-40 • Am J Psychiatry 2015;172(5):426-9 • López-Muñoz & Alamo, Cell Mol Life Sci 2009 • World J Biol Psychiatry 2022 (serendipity) • PMC 5044476 (Kuhn 100th Birthday) • PMC 8203803 (TB-Depression syndemic) • Samouco et al., Bipolar Disorders 2023 • Discover Magazine 2016 • Lasker Foundation (Kline) • Taylor & Francis (iproniazid) • PMC 4428540 (history of antidepressants) • ScienceInsights (iproniazid)
Serotonin Hypothesis
Moncrieff et al., Mol Psychiatry 28:3243-56 (2023) — Umbrella review • Jauhar et al., Mol Psychiatry 28:3149-52 (2023) — Rebuttal • PMC 9957868 (methodological reflections) • UCL News Jul 2022 • KCL News Jun 2023 • Psychology Today Jul 2022 • Pharmaceutical Journal Aug 2022 • Psychopharmacology Institute
Trial Duration, Design & STAR*D
Ward et al., Am J Med 2025 • medRxiv Feb 2025 preprint • PMC (Jureidini & Amsterdam 2019) — youth MDD • Pigott et al., BMJ Open 2023 (STAR*D reanalysis) • Psychiatric Times 2023/2025/2026 (STAR*D) • PMC 4314062 (STAR*D reexamination) • Rush et al., Am J Psychiatry 2006 (original STAR*D) • Sakurai et al., World Psychiatry 2024 (reappraisal) • medRxiv Oct 2025 (STAR*D RIAT augmentation)
Efficacy & NNT
Cipriani et al., Lancet 2018 (522 trials, 116,477 participants) • Kirsch et al., PLoS Med 2008 • Arroll et al., Cochrane 2009 • PMC 4853640 (NNT reconsidered) • Cambridge Core 2019 (antidepressant standoff)
Pharmacology & Receptor Binding
Owens et al. 1997 • Hyttel 1994/1997 • PMC 4047306 (escitalopram comparison) • Sghendo & Mifsud 2012 • PMC 5962350 (SERT crystal structures) • Springer/Hyttel 1997 (reuptake comparison) • Leonard 1995 (SSRI differentiation) • ScienceDirect 1999 (PK of SSRIs) • StatPearls NCBI (SSRIs 2023) • Wikipedia: SSRI, Imipramine
Withdrawal & Discontinuation
Henssler et al., Lancet Psychiatry 2024 (meta-analysis) • Davies & Read, Addict Behav 2019 • Fava & Cosci, J Clin Psychiatry 2019 • Horowitz & Taylor, Lancet Psychiatry 2019 • Psychiatric News APA 2025 • AAFP 2006 • Psychiatric Times 2025 • PMC 4919171 (switching/stopping) • Harvard Health 2022 • Cleveland Clinic 2026 • PMC 3024727 (paediatric) • Wikipedia: AD discontinuation • Gastaldon et al. 2022 • Zhang et al. 2024
Research Critique (Horowitz/Maudsley & Sørensen/Crossing Zero)
Horowitz & Taylor, Lancet Psychiatry (2019) — SSRI tapering proposal • Sørensen et al. systematic review (2021) — 17 PET studies, 294 participants • Sørensen, Crossing Zero (2023) • Horowitz & Taylor, The Maudsley Deprescribing Guidelines (2023) • Hodson review, BJPsych Bulletin (2024) • Gordon review, BJGP (2024) • Meyer et al. (2004) — PET SERT occupancy • Brouillet et al. (1991) — primate benzo occupancy • Full independent reviews: guythetaperman.com/news/maudsley-review and guythetaperman.com/news/crossing-zero-anders-sorensen-review
Cross-Tapering & Switching
Keks et al., Aust Prescr 2016 • NHS SPS 2025 • Shapiro & Cohrs, J Psychiatry Neurosci 2025 • Therapeutics Letter 157 (Jul 2025) • Psychiatric Times 2025 (switching strategies) • BC Guidelines Appendix D • NPS/Australian Prescriber switching table • GoodRx 2025 • Healthline 2022 • Accountable Health Partners • Carlat Publishing 2025 • Lancet Psychiatry 2026 (deprescribing NMA) • PMC 8841913 (CPG tapering review) • PMC 8092632 (Cochrane discontinuation)
Patient Experience & Side Effects
Read et al. 2014 (1,829 NZ users) • PMC 4863327 (qualitative 2016) • PMC 2719451 (SSRI side effects 2009) • PMC 2071966 (Dutch reporting 2007) • ScienceDirect 2017 (3,243 online reviews) • ScienceDirect 2017 (1,008 UK users) • Nature Transl Psychiatry 2021 (iSPOT-D) • PMC 12927973 (Ethiopian study)
PSSD
PMC 11450419 (barriers to quantifying 2024) • PMC 10122283 (Israeli risk estimate 2023) • ScienceDirect Sex Med Rev 2021 • Wiley Pharmacoepidemiol Drug Saf 2023 • RxISK (PSSD overview) • PSSD Network