Slow Antidepressant Tapering: What Evidence Shows

What if the apparent benefit of slow antidepressant tapering depends less on tapering alone and more on what happens around it?

A 2026 systematic review and network meta-analysis examined 76 randomized trials involving 17,379 adults with fully or partially remitted depression or anxiety disorders. Its headline finding was encouraging: slow tapering combined with psychological support appeared comparable to antidepressant continuation for preventing relapse.

That conclusion deserves attention. It also requires more restraint than some summaries have shown.

What the researchers actually asked

The Lancet Psychiatry analysis compared several strategies: abrupt discontinuation, fast tapering lasting four weeks or less, slow tapering lasting more than four weeks, continuation at a reduced dose, and continuation at a standard dose. Each strategy could occur with or without psychological support.

The primary outcome was relapse by the end of the trial. The average follow-up was about 46 weeks, although duration varied substantially. Sixty of the 76 trials involved depression, while only 16 involved anxiety disorders.

The population was 67.5% female, with a mean age of 45. Participants were predominantly White. These details matter because a large sample does not automatically produce broad generalizability.

The central result was a combination, not tapering alone

Compared with abrupt discontinuation, slow tapering plus psychological support reduced the relative risk of relapse by about 48%. The reported number needed to treat was 5.4. In other words, compared with abrupt stopping, roughly five people would need the combined strategy to prevent one additional relapse during the study period.

Standard-dose continuation also outperformed abrupt discontinuation, with a number needed to treat of 5.3. The investigators concluded that slow tapering plus psychological support was similar to medication continuation for relapse prevention.

But slow tapering alone did not produce a statistically clear advantage over abrupt discontinuation. Its relative risk was 0.81, with a 95% confidence interval from 0.56 to 1.18. That interval includes both a potentially meaningful benefit and no benefit.

This is the first major interpretive boundary: the strongest finding supports a package of slow tapering and structured psychological support, not slow tapering by itself.

“Similar” does not mean proven equivalent

A network meta-analysis combines direct comparisons within trials with indirect comparisons across a connected evidence network. This is useful when few trials compare every strategy head to head.

It also adds assumptions. The populations, relapse definitions, psychological interventions, medications, and taper schedules must be similar enough for the indirect comparisons to remain credible. The authors found moderate heterogeneity and no statistical evidence of network inconsistency, which strengthens the analysis.

Still, failure to detect a difference is not the same as proving two strategies equivalent. The finding supports clinical similarity at the group level. It does not establish that tapering with therapy and continuing medication will be interchangeable for a particular person.

The withdrawal problem remains unresolved

Relapse was the primary outcome, but relapse and withdrawal can overlap. Anxiety, sleep disruption, low mood, agitation, impaired concentration, and physical symptoms may appear during dose reduction. Timing, symptom pattern, and response over time help interpretation, which is why the symptom timeline matters.

A major discontinuation trial, ANTLER, found higher relapse and withdrawal scores after discontinuation. It also illustrated the methodological difficulty: symptoms emerging during or soon after a taper can influence depression ratings and decisions to restart treatment.

The 2026 meta-analysis does not fully solve this problem. Withdrawal outcomes were less consistently measured than relapse, and the analysis relied on summary-level trial data rather than individual participant timelines. There is no single meaningful dropout rate for the review because attrition and missing-data procedures differed across the included trials.

This distinction should shape how success is defined. Preventing a symptom-scale increase, completing discontinuation, maintaining daily functioning, and preserving quality of life are related outcomes, but they are not the same outcome. Before medication changes are considered, it helps to define what success is supposed to mean.

“Slow” was still a blunt category

Any taper lasting more than four weeks counted as slow. That category may combine a five-week linear taper with a months-long, progressively smaller reduction schedule.

It did not specifically test hyperbolic tapering. A systematic review of serotonin transporter imaging found that antidepressant dose and transporter occupancy have a nonlinear relationship. Low doses can retain substantial biological activity, making the final reductions pharmacologically larger than their milligram size suggests.

That imaging evidence provides a plausible rationale for smaller reductions at lower doses. It does not, by itself, prove that a specific taper schedule prevents withdrawal, relapse, or functional impairment. Biological plausibility and clinical outcome evidence are different levels of evidence.

What the study establishes

For adults with remitted depression, abrupt or rapid discontinuation performed worse than several more supported strategies.
Slow tapering combined with structured psychological support produced clinically meaningful relapse-prevention results.
Continuation remained effective, and reduced-dose continuation might help some groups, although that evidence was less certain.
The review itself reported no funding and no competing interests.

Funding of the review, however, is separate from funding and reporting practices within the original trials.

What it cannot establish

That slow tapering alone prevents withdrawal or relapse.
That more than four weeks is the correct taper duration for everyone.
That linear and hyperbolic tapering are equivalent.
That early symptom worsening necessarily represents relapse rather than withdrawal, expectancy effects, or both.
That results apply equally to active illness, complex medication combinations, severe prior withdrawal, or populations poorly represented in the trials.

This is a consequential study because it moves the conversation beyond the false choice between indefinite continuation and unsupported stopping. But its most defensible message is not simply “taper slowly.” It is that medication reduction occurs within a psychological, biological, and clinical context.

If tapering outcomes depend on both the shape of the dose reduction and the support surrounding it, why are those two components still studied as though one generic protocol could settle the question?

Sources and Further Reading

Educational content from DrTeralyn.com
Brain Health · Psychology · Psychiatric Medication Literacy

Is Slow Tapering Enough? What a 2026 Antidepressant Meta-Analysis Really Found