Distinguishing Relapse from Discontinuation Symptoms: A Clinical Algorithm
When a patient deteriorates during or shortly after an antidepressant taper, the prescriber faces a high-stakes diagnostic question: is the original disorder returning, or is the patient experiencing pharmacological withdrawal? The answer determines whether to reinstate, slow the taper, or treat as a new episode — and the cost of misclassification is substantial in both directions. Misreading withdrawal as relapse commits patients to indefinite pharmacotherapy they no longer require; misreading relapse as withdrawal exposes them to preventable morbidity and suicide risk. This post outlines a structured, evidence-based algorithm for differentiating the two.
Why the two are confused
The diagnostic confusion is not a clinical lapse; it is built into the symptom overlap. Selective serotonin reuptake inhibitors (SSRIs) and serotonin–norepinephrine reuptake inhibitors (SNRIs) downregulate postsynaptic 5-HT receptors and modulate noradrenergic tone over weeks. When the drug is removed faster than the receptor system can re-upregulate, the resulting neuroadaptive mismatch produces somatic, affective, and cognitive symptoms that mimic the depressive or anxious phenotype the drug was originally prescribed for.
The Maudsley Deprescribing Guidelines (Horowitz & Taylor, 2024) and a 2019 systematic review by Davies and Read in Addictive Behaviors (n = 14 studies, pooled incidence approximately 56% of patients experiencing some discontinuation symptoms, 46% of those describing them as severe) establish that discontinuation symptoms are common, not rare. Earlier industry-influenced literature framed them as transient and self-limited; current pharmacovigilance data, particularly from paroxetine and venlafaxine, show that protracted withdrawal lasting months is a real, if minority, phenomenon.
The DSM-5 does not contain a discontinuation diagnosis distinct from substance withdrawal, and ICD-11 only recently introduced one (6C4G.4, "antidepressant discontinuation syndrome"). The result is that prescribers often default to the diagnosis they already know — the patient's primary condition — when symptoms re-emerge.
The five-axis differential
A reliable bedside algorithm rests on five axes: timing, symptom profile, trajectory, response to dose reinstatement, and prior taper history. No single axis is diagnostic; the combination is.
Axis 1 — Timing relative to the dose change
Discontinuation symptoms typically begin within 36–96 hours of a dose reduction or cessation, with onset tightly linked to the parent drug's elimination half-life. For paroxetine (t½ ≈ 21 h, no active metabolite), symptoms commonly emerge within 1–2 days. For venlafaxine (t½ of parent compound ≈ 5 h, O-desmethylvenlafaxine ≈ 11 h), onset can be within 12–24 hours. Fluoxetine, with its 4–6 day half-life and norfluoxetine metabolite half-life of 7–15 days, can delay onset by 2–6 weeks, which is why fluoxetine is sometimes used as a bridging agent.
A relapse of major depressive disorder, by contrast, follows a slower trajectory. Recurrence typically takes weeks: the STAR*D follow-up data and Cochrane reviews of maintenance pharmacotherapy show median time to relapse after antidepressant discontinuation in placebo-controlled discontinuation trials is on the order of 8–24 weeks, not days.
A useful clinical heuristic: if the patient deteriorates within one week of a dose decrement, withdrawal is the leading hypothesis. If deterioration emerges 4–12 weeks after the last reduction and is gradual, relapse moves up the differential.
Axis 2 — Symptom profile
Discontinuation syndromes have a characteristic somatic and neurosensory signature that depression does not. The FINISH mnemonic captures this: Flu-like symptoms, Insomnia, Nausea, Imbalance, Sensory disturbances, Hyperarousal.
The most discriminating features are the neurosensory ones. "Brain zaps" (paroxysmal electrical-shock sensations, often triggered by lateral eye movement), vertigo on head turning, paresthesias, electric-feeling jolts in the limbs, visual trails, and akathisia-like inner restlessness are not part of the depressive phenotype. Their presence shifts the diagnosis strongly toward withdrawal.
Conversely, anhedonia, early-morning awakening with diurnal mood variation, psychomotor retardation, hopelessness specifically about the future, guilt and worthlessness ruminations, and suicidal ideation that has cognitive content (a plan, a target) are more characteristic of recurrent depression.
A practical comparison:
| Feature | Discontinuation syndrome | Relapse / recurrence | ||---| | Onset relative to last dose change | 36 h – 2 weeks | 4 – 24 weeks | | Brain zaps, paresthesias, dizziness | Common | Absent | | Flu-like aches, nausea, GI upset | Common | Absent or mild | | Akathisia, inner agitation | Common (especially venlafaxine, paroxetine) | Less common | | Anhedonia, psychomotor retardation | Uncommon | Common | | Diurnal mood variation | Uncommon | Common | | Suicidal ideation with cognitive plan | Less typical (more impulsive if present) | More typical | | Response to dose reinstatement | Resolves in 24–72 h | Requires weeks |
Note the agitated, hyperaroused quality of withdrawal versus the slowed, vegetative quality of classical melancholic depression. Atypical or anxious depression complicates this distinction, but the neurosensory features remain the most discriminating.
Axis 3 — Trajectory
Discontinuation symptoms peak within 1–2 weeks of the precipitating reduction and then resolve, even without intervention, in the majority of cases — typically over 2–6 weeks. A waxing-and-waning pattern within that window is common; gradual improvement is the rule.
Relapse trajectories are different. Once a depressive episode begins to re-emerge, untreated, it tends to deepen rather than self-resolve. Symptom severity at week 4 is generally worse than at week 1, and the cognitive content (hopelessness, worthlessness) intensifies.
If a patient is reassessed at 3–4 weeks after a taper step and is clearly improving from a peak at week 1, withdrawal is supported. If they are worse than they were at week 1, with deepening anhedonia and cognitive symptoms, relapse becomes more likely.
Axis 4 — Response to dose reinstatement
Reinstatement is both a therapeutic maneuver and a diagnostic test. The Maudsley Deprescribing Guidelines describe partial or full reinstatement as the standard of care for moderate-to-severe discontinuation symptoms, with the expectation that withdrawal symptoms will resolve within 24–72 hours of an adequate dose.
A patient whose brain zaps, dizziness, and agitation disappear within 1–3 days of restoring the prior dose has, by definition, a pharmacological withdrawal phenomenon. A patient whose anhedonia, insomnia, and hopelessness are unchanged 2 weeks after reinstatement is more likely experiencing recurrence — antidepressants take weeks, not days, to treat depression.
The reinstatement test should generally use a dose at or modestly above the last well-tolerated step, not necessarily the original full dose, particularly for paroxetine and venlafaxine where receptor occupancy curves are steep at low doses.
Axis 5 — Prior taper history and individual vulnerability
Patients with a prior history of difficult discontinuation, long duration of treatment (>2 years), or treatment with high-risk agents (paroxetine, venlafaxine, desvenlafaxine) have a substantially elevated probability of withdrawal being the cause of new symptoms. A 2019 meta-analysis in Psychotherapy and Psychosomatics (Fava et al.) reported discontinuation symptom rates exceeding 60% for paroxetine and venlafaxine, compared with around 17% for fluoxetine.
A patient on fluoxetine for 6 months who deteriorates 10 weeks after stopping is clinically more consistent with relapse than withdrawal. A patient on venlafaxine 225 mg daily for 8 years who deteriorates 4 days after a 37.5 mg reduction is the opposite.
A bedside algorithm
The five axes can be operationalized as a stepwise decision pathway. The prescriber should run through these questions in order:
Step 1 — Establish the temporal anchor. When was the most recent dose change? What was the magnitude (absolute mg and percentage of the prior dose)? What is the half-life of the parent drug and any active metabolites?
Step 2 — Map symptoms to the FINISH framework. Are neurosensory features present? Brain zaps, vertigo on head movement, paresthesias, akathisia, flu-like myalgia? If yes, the differential favors discontinuation.
Step 3 — Probe for cognitive depressive content. Is the patient experiencing anhedonia, hopelessness with future-oriented content, guilt, worthlessness, psychomotor retardation, diurnal variation? If these dominate over somatic features, relapse moves up.
Step 4 — Assess trajectory. Is the patient better, worse, or unchanged compared with 1–2 weeks ago? Improvement supports discontinuation. Worsening with cognitive depressive content supports relapse.
Step 5 — Consider a diagnostic reinstatement. If symptoms are moderate to severe and uncertainty remains, restore the last well-tolerated dose. Reassess at 72 hours. Resolution confirms withdrawal. Persistence at 2 weeks suggests relapse — though incomplete resolution can also indicate protracted withdrawal in patients with prior severe discontinuation histories, which complicates the inference.
Step 6 — Risk-stratify. At every step, suicidality, functional collapse, or psychotic features take priority over diagnostic precision. If the patient is acutely unsafe, reinstate, stabilize, and defer the differential question.
Special and ambiguous presentations
Mixed picture
Many patients present with both — withdrawal-driven autonomic and neurosensory features layered on a re-emerging depressive cognitive picture. In these cases, the clinical move is sequential: reinstate to control the withdrawal, hold for 4–8 weeks at the reinstated dose, then re-evaluate the cognitive symptoms with the autonomic noise removed. If anhedonia and hopelessness persist after withdrawal-related symptoms have resolved, relapse is the operative diagnosis.
Protracted withdrawal
A subset of patients, particularly those with long exposure to short-half-life serotonergic agents, develop protracted post-withdrawal syndromes that can last 6–18 months. These can include emotional blunting, anhedonia-like states, and persistent neurosensory symptoms — features that resemble depression. Distinguishing this phenotype from true recurrence is genuinely difficult; the strongest evidence is a clear temporal link to the taper, absence of the patient's prior depressive cognitive content, and partial response to dose reintroduction. The Royal College of Psychiatrists' 2019 position statement explicitly acknowledged protracted withdrawal as a clinical entity.
Anxiety disorders
Discontinuation from SSRIs and SNRIs in patients treated for anxiety often produces a hyperarousal phenotype that mimics returning anxiety almost exactly. Here the timing axis is decisive. Symptom onset within 72 hours of a reduction, with neurosensory features and autonomic activation, is withdrawal. Onset over weeks, with the patient's prior cognitive worry content, is recurrence.
Reinstatement non-response
Reinstatement does not always work, particularly when the taper has gone too fast for too long, or when reinstatement uses too small a dose. A failed reinstatement does not, on its own, prove relapse. Before concluding recurrence, ensure reinstatement was at an adequate dose and held for at least 2–4 weeks. Hyperbolic micro-dose reinstatement (using liquid formulations or compounded preparations to test small increments) can clarify ambiguous cases.
Concurrent medical contributors
Before settling on either withdrawal or relapse, the prescriber should rule out medical contributors that produce overlapping presentations. Hypothyroidism, vitamin B12 or folate deficiency, occult infection, anemia, and obstructive sleep apnea all generate fatigue, low mood, and cognitive slowing that can be mistaken for depressive recurrence. Vestibular pathology can mimic the dizziness and imbalance of withdrawal. Stimulant or alcohol use (newly increased or newly stopped) interacts with the autonomic and affective picture in ways that distort the differential. A focused review of systems, basic laboratory workup (TSH, CBC, B12, ferritin), and a substance-use history at the assessment visit close out these confounders before the prescriber commits to a withdrawal-versus-relapse interpretation.
Patient self-report and recall bias
Patients often compress timelines when reporting deterioration after a dose change. A symptom that began 6 weeks after the last reduction may be recalled as "right after I dropped the dose" because the dose change is the event the patient is anchoring to. The prescriber should anchor the timeline to objective events — the date the prescription was last filled, the date a pillbox was changed, a dated symptom diary — rather than to the patient's narrative ordering. This is a common source of false-positive withdrawal attribution and false-negative relapse identification, and it is correctable with structured questioning.
Clinical pearls
- Timing is the most powerful discriminator. Onset within 1 week of a dose change favors withdrawal; gradual onset over 4–24 weeks favors relapse.
- Neurosensory symptoms are pathognomonic for withdrawal. Brain zaps, vertigo on head turning, and paresthesias are not features of major depression — their presence essentially rules in discontinuation syndrome.
- Use reinstatement as a diagnostic test. Resolution within 24–72 hours of restoring the prior dose confirms withdrawal; lack of response at 2 weeks supports relapse.
- Anchor the differential to the drug's pharmacokinetics. Paroxetine and venlafaxine produce early, intense withdrawal; fluoxetine delays onset by weeks. Match expected timing to the agent.
- Mixed presentations are common — sequence the response. Reinstate to clear withdrawal first, then reassess for residual depressive cognition at 4–8 weeks.
- Document the symptom profile and trajectory at each visit. Patients and clinicians misremember; serial structured documentation prevents anchoring errors when the differential is reconsidered weeks later.
- When in doubt, slow the taper rather than abandon it. A return to gradual hyperbolic reductions held for longer intervals resolves most ambiguous deteriorations without committing the patient to indefinite treatment.
For more clinician resources on safe deprescribing and tapering, visit tapermeds.com.