Slow Taper Is Not Enough: Why Stopping an Antidepressant Should Come With Psychotherapy
TL;DR
A 2026 Lancet Psychiatry network meta-analysis found that tapering slowly (over more than four weeks) with psychological support cut relapse risk by about as much as staying on the drug (RR 0.52 versus 0.51).
Slow tapering alone did not significantly beat stopping abruptly (RR 0.81, 95% CI 0.56 to 1.18).
My position: taper pace manages withdrawal, and psychotherapy manages relapse. A remitted patient who wants to stop needs both.
The evidence is moderate at best, comes mostly from depression, and cannot say which kind of support works best.
Staying on the antidepressant is a legitimate choice, and nobody should stop without a plan from their prescriber.
Most conversations about stopping antidepressants circle one question: how slowly should the dose come down? Patients ask it, online communities argue over it, and clinicians, myself included, spend years refining tapering schedules. A new analysis in The Lancet Psychiatry suggests we have been answering only half of the question. I think the other half, psychological support, deserves to be the default and not an optional extra.
What did the Lancet Psychiatry analysis actually find?
The analysis pooled 76 randomised trials of 17,379 adults in remission from depression or anxiety and compared ways of stopping antidepressants. Slow tapering with psychological support lowered relapse risk versus abrupt stopping (RR 0.52), about as much as simply continuing the drug (RR 0.51).
Relapse was the primary outcome, with a mean follow-up of 46 weeks. Slow tapering meant reducing over more than four weeks, and fast tapering meant four weeks or less. Against abrupt discontinuation, four strategies lowered relapse: continuing the standard dose with psychological support (RR 0.40, 95% CI 0.26 to 0.61, NNT 4.3), continuing the standard dose alone (RR 0.51, NNT 5.3), slow tapering with support (RR 0.52, 95% CI 0.38 to 0.72, NNT 5.4), and continuing at a reduced dose (RR 0.62, low certainty). Certainty was moderate for the first three, and tolerability was comparable across strategies.
Why is a slow taper alone not enough?
Tapering and relapse prevention solve different problems. A slower dose reduction gives the nervous system time to adjust, but it does nothing about the vulnerabilities that produced the depression. In the analysis, slow tapering alone did not significantly beat abrupt stopping (RR 0.81, 95% CI 0.56 to 1.18).
Two cautions apply. The interval is wide, so this does not prove tapering alone is useless. It means the trials could not show a relapse benefit. And relapse is not withdrawal: withdrawal data were limited or missing in many of the included studies, so the analysis says little about how much a slower taper eases withdrawal symptoms. The rest of the pattern is still telling. Fast tapering with support (RR 0.52, 95% CI 0.27 to 1.01) and abrupt stopping with support (RR 0.73) did not significantly beat abrupt stopping either. Adding psychological support to either fast or slow tapering also improved outcomes, although the certainty of that added benefit was low. Pace and support seem to work together, but the data are not as firm as I would like.
What does psychological support add?
In my view, psychological support gives patients a structure for the months when relapse risk is highest: spotting early warning signs, riding out withdrawal-like sensations, and telling a wobble from a returning episode. In the pooled data, adding it to a slow taper moved the relapse estimate from RR 0.81 to RR 0.52.
This fits an older trial. In PREVENT, adults with three or more previous depressive episodes on maintenance antidepressants were randomised to mindfulness-based cognitive therapy with support to taper or to staying on medication. Over 24 months, time to relapse did not differ between the two groups (HR 0.89, 95% CI 0.67 to 1.18). Therapy plus a supported taper performed comparably to the drug.
The honest limit is that the review found the evidence on psychological support itself is limited. We know it helps in these trials. We do not know which form helps most, or how brief it can be.
Does the pace of the taper still matter?
Yes. Discontinuation symptoms are real, and how often they occur is debated. A 2024 meta-analysis estimated that about 15% of people have symptoms specifically attributable to stopping, and around 3% have severe ones. A reanalysis of the most systematic studies put the figure at 25% to 39%.
Horowitz and Taylor argued that dose cuts should shrink as the dose falls, because the relationship between dose and serotonin transporter occupancy is hyperbolic. An observational study of 608 people, mostly on venlafaxine or paroxetine, found that withdrawal was limited and fell as tapers got longer. That study was not randomised, so it cannot tell us whether those patients stayed well afterwards. This is the gap I keep coming back to: a slow taper protects the patient from withdrawal, and therapy protects them from relapse. I have written before about telling withdrawal from relapse, and the two are easy to confuse when there is no support in place.
What does this mean for patients in India?
It means the recommendation runs into workforce reality. India has about 0.75 psychiatrists per 100,000 people against a WHO minimum of three, and trained therapists are scarcer still outside big cities. Telling every patient to “add psychotherapy” is easy. Delivering it is harder.
The review's co-author Debora Zaccoletti has said that short, scalable, remotely delivered therapies should be tested and prioritised. I agree, and remote care already widens access. Where a therapist is out of reach, I would still rather a patient have scheduled reviews, education about withdrawal, and a written relapse plan than a taper with nothing around it. That is my clinical judgement, and these trials did not test it. It is how I approach deprescribing at ManoMitra.
Stopping Antidepressants Safely: What Is Deprescribing?
Where might reasonable clinicians disagree?
The certainty is moderate at best, and the population is narrow. Participants were mostly White (88%), 79% of trials were in depression and 21% in anxiety, and “psychological support” was not a single intervention. A network meta-analysis also leans on indirect comparisons. The Critical Psychiatry Network published its own promises and pitfalls response, which is worth reading.
There is a fair objection to my “always.” Some patients taper on their own without trouble, and a therapy requirement could become a barrier or a delay for them. Averages also hide individuals. I still favour offering support as the default, since the downside is small and the relapse signal is not. Staying on the medication, with or without therapy, was at least as protective as any way of stopping. Coming off is a choice, not a goal.
The bottom line
Stopping an antidepressant is a change in treatment, not the end of it. The pace of the taper decides how the body copes, and the support around it seems to decide whether the person stays well. Doing one without the other leaves the job half done. This post is educational and is not individual medical advice. Please do not stop or reduce an antidepressant without a plan from your prescriber.
Frequently Asked Questions
Can I stop my antidepressant slowly on my own?
In the trials, slow tapering alone did not significantly reduce relapse compared with stopping abruptly (RR 0.81, 95% CI 0.56 to 1.18). Safer results came when tapering was paired with psychological support. Plan the taper with your prescriber and ask what support can run alongside it.
How long is a slow taper?
In this analysis, “slow” meant more than four weeks. A sub-analysis of tapers longer than 12 weeks showed a similar pattern, but few trials contributed. People at higher risk of withdrawal may need many weeks or months, and your prescriber sets the pace.
What counts as psychological support?
The review found the evidence on psychological support limited, so no single form can be called best. The PREVENT trial used mindfulness-based cognitive therapy alongside support to taper. Ask your clinician which structured option is available to you, in person or remotely.
Is staying on the antidepressant just as good?
For relapse prevention, yes. Continuing the standard dose lowered relapse versus abrupt stopping (RR 0.51), and continuing with psychological support did best (RR 0.40), according to the pooled results. The decision depends on side effects, preferences, and past episodes.
Do these findings apply to anxiety or bipolar disorder?
Only partly. The analysis covered remitted depressive and anxiety disorders, and 21% of trials were in anxiety, so evidence there is thinner. It was not designed to answer questions about bipolar disorder, and those decisions need individual specialist advice.
