The Pharmacology of Coming Off an Antidepressant Slowly

The Pharmacology of Coming Off an Antidepressant Slowly#

A patient who has halved their antidepressant twice with barely a ripple is often surprised when the final small cut brings on dizziness, brain zaps, and low mood. The explanation is not that they have become fragile. It is that a selective serotonin reuptake inhibitor (SSRI) does not act in proportion to the milligrams on the tablet. It acts in proportion to how many serotonin transporters it blocks, and that relationship is a curve, not a straight line. Because the curve is steepest near zero, the smallest doses carry the largest biological weight, so subtracting the same number of milligrams at every step lands progressively harder as the dose drops.

Key points#

• An SSRI's effect tracks receptor occupancy, not dose, and occupancy follows a saturating (hyperbolic) curve.
• Near a normal dose the transporter is almost fully blocked, so large milligram cuts change occupancy little.
• Near zero the curve is nearly vertical, so tiny milligram cuts change occupancy a lot.
• To make each step feel similar, decrements have to shrink as the dose falls, which is what hyperbolic tapering describes.
• Cohort and trial data agree that slower tapering is gentler, but the exact optimal schedule is not yet proven.

Why the last milligrams weigh the most#

Picture a dimmer switch that is already turned nearly all the way up. Rotating it down a quarter turn from the top barely dims the room, because the bulb was already close to full brightness. Rotating that same quarter turn near the bottom takes the room from softly lit to dark. Receptor occupancy behaves the same way. At a standard dose the serotonin transporter is close to saturated, so pulling out a chunk of drug moves occupancy only slightly. At the low end the system is nowhere near saturated, so a small change in dose swings occupancy across a wide range.

Occupancy is not a theoretical construct. Positron emission tomography studies measure directly what fraction of transporters a given dose blocks. When Mark Horowitz and David Taylor gathered these imaging data in a 2019 analysis in The Lancet Psychiatry, the dose-occupancy points traced a Michaelis-Menten hyperbola, the same saturating shape that describes an enzyme approaching its maximum speed. That single observation reframes the whole taper. The quantity that matters to the brain is occupancy, and occupancy does not fall in step with dose.

Putting numbers on the curve#

Concrete figures make the counterintuitive part obvious. In that synthesis, roughly 20 mg of citalopram occupies about 80 percent of serotonin transporters. Halving the dose to 10 mg sounds dramatic but removes only a handful of percentage points of occupancy. Drop to 5 mg and occupancy still sits near 60 percent. Even 1 mg holds something above 20 percent. The stretch of the curve doing the real work sits at the very bottom of the dose range, exactly where a patient might assume the drug has effectively stopped mattering.

That is why the step from 20 mg to 10 mg feels like almost nothing while the step from 5 mg to zero can feel like a cliff. To keep each reduction roughly equal in occupancy terms, the milligram decrements have to get smaller and smaller as the dose approaches zero, producing a schedule that descends in shrinking steps toward very small final doses. That shrinking, curve-shaped pattern is what hyperbolic tapering names. None of this is a prescription; it is an account of the pharmacology, and any change to a real regimen is a clinical decision made with the prescriber.

What the outcome evidence actually shows#

Mechanism is only a hypothesis until outcomes test it, and useful outcome data have begun to accumulate. A 2023 prospective cohort in Therapeutic Advances in Psychopharmacology followed 608 people reducing their antidepressants in small graduated daily steps and collected close to 33,000 daily withdrawal ratings across schedules lasting 28 to 140 days. Overall withdrawal was modest, and it moved inversely with the speed of the taper: the slower the reduction, the milder the symptoms. Younger age and female sex were linked to more withdrawal, which the authors took as a case for individualized, monitored tapering rather than one fixed formula for everyone.

A much larger controlled picture arrived in 2025. A network meta-analysis in The Lancet Psychiatry pooled 76 randomized trials covering 17,379 adults in remission from depression or anxiety. Gradual tapering paired with psychological support prevented relapse about as effectively as staying on the medication, while fast tapering of four weeks or less did no better than stopping abruptly. The cohort and the randomized evidence point the same way: slower is gentler.

Where the evidence still runs out#

Agreeing on direction is not the same as knowing the precise answer, and honesty requires naming the gaps. The 2025 network meta-analysis tracked relapse rather than withdrawal symptoms, and its authors flagged that withdrawal and quality-of-life data were largely absent and that the certainty behind the psychotherapy comparisons was low. The 2023 cohort was naturalistic, not randomized, so it cannot fully disentangle the shape of the taper from the expectations of the people who chose to taper that way.

The withdrawal literature underneath all of this is thinner still. A 2025 appraisal and reanalysis in Psychological Medicine found that most efficacy trials captured withdrawal only through spontaneously reported side effects rather than structured checklists, that participants had usually taken the drug for under six months before stopping, and that only a handful of studies measured withdrawal rigorously at all. The very group most likely to struggle, long-term users coming off after years, is the group least represented in the trials.

So the strong claim, that a specifically hyperbolic schedule beats every other slow taper for those long-term patients, has real mechanistic plausibility and supportive but incomplete outcome data behind it, not settled proof. The receptor-occupancy math is sound and explains why equal dose cuts do not produce equal effects. The clinical takeaway that follows, that tapers should generally be slow, flexible, and monitored, is well supported in direction. The exact curve, the ideal duration, and the size of the benefit for long-term users remain open questions that current trials were simply not designed to answer.