Desvenlafaxine: Independent Evidence on Depression, Side Effects & Withdrawal

Table of contents

• Evidence summary
• What desvenlafaxine is
• How it works
• What it is prescribed for
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions
• Who should avoid desvenlafaxine
• Dosage and how to take it
• Follow the money: who makes it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

The table below grades the main claims about desvenlafaxine. Independent and regulatory sources are listed first where they exist.

Independent evidence and credibility scorecard

Tier 1 means the most independent source: regulators, or government-funded or unfunded academic work with no drug-company money in the analysis. Tier 4 means manufacturer-run analyses. Credibility A–D combines independence with methodological quality.

What desvenlafaxine is

Desvenlafaxine is a serotonin and norepinephrine reuptake inhibitor (SNRI). Chemically it is O-desmethylvenlafaxine, "the major active metabolite of the antidepressant venlafaxine" (FDA label). In other words, when someone takes venlafaxine, much of it is turned into desvenlafaxine in the body. The EMA assessors noted that venlafaxine is "almost entirely (80%) transformed into desvenlafaxine" (EMA withdrawal assessment report).

Brand and approval. The brand is Pristiq, an extended-release tablet in 25 mg, 50 mg and 100 mg strengths. Wyeth's application (NDA 021992) was approved by the FDA as a new molecular entity on 29 February 2008. The FDA record now lists the holder as PF Prism C.V. (Drugs@FDA, NDA 021992). The US label is issued by "Wyeth Pharmaceuticals LLC, a subsidiary of Pfizer Inc." (FDA label). It was registered in Australia on 18 August 2008 (PBAC 2008).

UK and EU status (verified as far as possible). Wyeth applied for EU-wide approval under the name Ellefore in September 2007. At that time the UK was covered by EU centralised approvals. On 13 October 2008 Wyeth withdrew the application, when the EU's CHMP "was of the provisional opinion that Ellefore could not have been approved" (EMA Q&A, 2008; EMA withdrawn application page). Pure City Research found no desvenlafaxine product on the UK electronic medicines compendium (emc), which lists UK-licensed medicines, and no NHS medicine page. We could not query the MHRA's product database directly (it did not load for automated access), so our conclusion is that desvenlafaxine is not licensed in the UK as far as we could verify. Venlafaxine, its parent drug, is widely available in the UK.

Generic status. US generic versions launched in March 2017 under patent litigation settlements (Pfizer 2017 Financial Report). Several generic makers and at least one separately approved desvenlafaxine formulation are listed by the FDA (Drugs@FDA, NDA 204150).

How it works

The FDA label says the exact mechanism "is unknown, but is thought to be related to the potentiation of serotonin and norepinephrine in the central nervous system, through inhibition of their reuptake". Non-clinical studies showed it to be "a potent and selective SNRI" (FDA label). The EMA summary explains that desvenlafaxine "has been slightly modified with the aim of reducing side effects in patients who have problems breaking venlafaxine down" (EMA Q&A). That refers to people whose CYP2D6 enzyme works poorly, which matters for venlafaxine but not for desvenlafaxine.

Pharmacokinetics. Oral bioavailability is about 80%, and only 30% is bound to plasma proteins. It is cleared mainly by conjugation (UGT enzymes), with a minor role for CYP3A4. "The CYP2D6 metabolic pathway is not involved." About 45% is excreted unchanged in urine within 72 hours, which is why kidney function affects dosing. Steady state is reached in about 4–5 days (FDA label). The EMA assessment reported a terminal half-life of 9–11 hours in different studies, rising to as much as 30 hours in people with kidney disease (EMA assessment report). This fairly short half-life helps explain why missed doses and abrupt stopping can trigger discontinuation symptoms.

The empty tablet shell. Pristiq uses an inert matrix. Patients may see something that looks like a tablet in their stool; the label says this is the empty shell after the medicine has been absorbed (FDA label).

What it is prescribed for

What works and what does not

Benefits by claim

Depression: a real but small effect

The FDA approved desvenlafaxine on four positive 8-week fixed-dose trials. The label reports these differences from placebo on the 17-item Hamilton scale (HAM-D17), where patients started at about 23–25 points (FDA label, Table 9):

Two points are worth drawing out. First, these gains of about 2–3 points on the 17-item Hamilton scale sit in the same range as the roughly 1.75-point average across all antidepressants in FDA-held trial data (Stone et al., BMJ 2022). Second, the label table shows only the fixed-dose studies. The EMA's assessors, who reviewed the full European dossier, noted that the flexible-dose studies were all negative: differences from placebo ranged between 0.7 and 1.6 points "and these differences did not reach statistical significance". In one of them, venlafaxine beat placebo by 3 points while desvenlafaxine did not (EMA withdrawal assessment report). The EMA found this telling, because flexible dosing is closer to real prescribing.

The manufacturer's own pooled analysis of nine registration trials found an average advantage of 1.9 HAM-D17 points, with response in 53% vs 41% and remission in 32% vs 23% (Thase et al., 2009). That is roughly one extra responder for every eight or nine people treated. The independent Cipriani 2018 analysis gave desvenlafaxine a response odds ratio of 1.49 (95% CrI 1.24–1.79). That is tied with clomipramine for second-lowest of 21 drugs, above only reboxetine (1.37). Its dropout odds ratio of 1.08 (0.88–1.33) was no different from placebo (Cipriani 2018, figure 3). Cipriani could not include desvenlafaxine in its head-to-head analysis because it "had only placebo-controlled trials" in the dataset (Cipriani et al., 2018).

Is desvenlafaxine better than venlafaxine?

Here the regulators split:

• European Medicines Agency (2008). The CHMP concluded that "the benefit-risk of desvenlafaxine is not positive". It found the efficacy documentation "not convincing" and said that, relative to venlafaxine, "desvenlafaxine appears less effective with a similar safety and tolerability profile". It also objected that maintenance of effect had not been shown at the proposed dose and that too few elderly patients had been studied (EMA withdrawal assessment report).
• Australia's Pharmaceutical Benefits Advisory Committee (2008). Using the sponsor's indirect comparison via placebo, the PBAC accepted that desvenlafaxine was non-inferior to venlafaxine. The median difference was −0.27 HAM-D points (95% CrI −1.17 to 0.65). The PBAC treated desvenlafaxine 50 mg as equivalent to venlafaxine 75 mg, and listed it on a cost-minimisation basis, meaning no price premium. The committee "was concerned with the evidence presented showing no increase in clinical benefit of doses of desvenlafaxine above 100 mg per day" (PBAC public summary, 2008).
• Independent meta-analysis (2015). Across head-to-head trials against other antidepressants, desvenlafaxine had a lower chance of response (RR 0.90, 95% CI 0.82–0.98) and remission (RR 0.82, 0.71–0.95). The authors concluded it "might not be as efficacious as other antidepressants" (Laoutidis & Kioulos, 2015).

Pure City Research's reading: the best available evidence suggests desvenlafaxine is at most equivalent to venlafaxine, and possibly slightly less effective at the doses used. Its genuine advantages are practical ones: a single fixed dose with no titration, no dependence on CYP2D6, and a low interaction burden.

Relapse prevention

In the label's 50 mg maintenance trial, people who had responded and then stayed stable for 12 weeks on 50 mg were randomised to continue or switch to placebo. At 26 weeks the Kaplan-Meier relapse estimate was 14% on desvenlafaxine vs 30% on placebo. In a higher-dose trial (200–400 mg) it was 29% vs 49% (FDA label). These "randomised withdrawal" designs have a known weakness: people switched abruptly to placebo can develop withdrawal symptoms that look like relapse. The label's own discontinuation warnings make this relevant here.

Menopausal hot flushes: an indication that was not approved

In a 12-week trial of postmenopausal women with at least 50 moderate or severe hot flushes a week, 100 mg daily cut hot flushes by 7.3 a day (62%) vs 4.5 (38%) on placebo. However, 10.0% on desvenlafaxine vs 3.7% on placebo dropped out because of adverse events (Pinkerton et al., Menopause 2013). A meta-analysis concluded it "is effective in the treatment of hot flashes but it is strongly associated with several adverse events and treatment discontinuation" (Berhan & Berhan, 2014). The FDA issued an "approvable" letter to Wyeth in July 2007 requesting a further one-year safety study, then a "complete response" letter in September 2011. Pfizer withdrew the application in February 2012 (Pfizer 2011 Financial Report; Pfizer statement).

Risks and all side effects

Common side effects, by dose

The label's pooled 8-week fixed-dose trials show clearly how side effects climb with dose while benefit does not (FDA label, Table 2):

Overall, 12% of patients on 50–400 mg stopped because of an adverse reaction, vs 3% on placebo. At the recommended 50 mg the rate was 4.1%, similar to placebo (3.8%). At 100 mg it was 8.7% (FDA label). Sponsor pooled data likewise showed adverse-event discontinuation rising from 4% to 18% across the dose range (Thase 2009). CANMAT lists desvenlafaxine among antidepressants associated with lower rates of sexual side effects (CANMAT 2023). The label data suggest this holds mainly at 50 mg.

Serious and important risks

Withdrawal (discontinuation) symptoms

The label lists nausea, sweating, low or irritable mood, agitation, dizziness, "electric shock sensations", tremor, anxiety, confusion, headache, lethargy, insomnia, hypomania, tinnitus and seizures after stopping serotonergic antidepressants. It adds: "There have been postmarketing reports of serious discontinuation symptoms with PRISTIQ, which can be protracted and severe. Completed suicide, suicidal thoughts, and severe aggression (including hostility, rage, and homicidal ideation) have been observed in patients during reduction in PRISTIQ dosage." Other reports describe visual changes and raised blood pressure after stopping. "In some patients, discontinuation may need to occur over a period of several months" (FDA label).

• Henssler et al. 2024 (unfunded, 79 studies, 21,002 patients): after stopping any antidepressant, 31% had at least one discontinuation symptom vs 17% after placebo, a placebo-adjusted incidence of about 15%. Severe symptoms affected 2.8% vs 0.6%. "Desvenlafaxine, venlafaxine, imipramine, and escitalopram were associated with higher frequencies of discontinuation symptoms", and "either desvenlafaxine or venlafaxine" with higher severity (Henssler et al., Lancet Psychiatry 2024).
• Manufacturer data (Montgomery et al. 2009): across Wyeth's trials, the commonest symptoms after stopping were dizziness, nausea, headache, irritability, diarrhoea, anxiety, abnormal dreams, fatigue and sweating. Symptom scores were significantly higher than placebo after stopping 50 and 100 mg (Montgomery et al., 2009).
• Pfizer taper trial (Khan et al. 2014): after 24 weeks on 50 mg, a 1-week taper at 25 mg was statistically equivalent to stopping abruptly on a symptom checklist (Khan et al., 2014). This trial had Pfizer-employed authors. It tested only a very short taper, so it says nothing about the slower, proportional tapers NICE now recommends.
• CANMAT classes desvenlafaxine as "moderate risk" for discontinuation symptoms, and paroxetine and venlafaxine as "high risk" (CANMAT 2023).

The label also warns that "discontinuation symptoms have been reported when switching patients from other antidepressants, including venlafaxine, to PRISTIQ" (FDA label). Switching from venlafaxine to its own metabolite is not automatically seamless.

Dependence and overdose

Desvenlafaxine "is not a controlled substance" (FDA label). Withdrawal symptoms reflect physical adaptation, not addiction. There is limited overdose experience with desvenlafaxine itself, so the label relies on venlafaxine data. That includes tachycardia, seizures, ECG changes and deaths, and an overdose risk "compared to that observed with SSRI antidepressant products, but lower than that for tricyclic antidepressants", though how much of this reflects the patients rather than the drug "is not clear" (FDA label).

All interactions

Desvenlafaxine has fewer metabolic interactions than venlafaxine, fluoxetine, paroxetine or fluvoxamine. The label states it "does not inhibit CYP1A2, 2A6, 2C8, 2C9, 2C19 CYP2D6, or CYP3A4" in vitro and is not a P-glycoprotein substrate or inhibitor (FDA label). The serotonin and bleeding interactions still apply in full.

Who should avoid desvenlafaxine

• Contraindicated: people allergic to desvenlafaxine, venlafaxine or any ingredient (angioedema has been reported), and anyone taking an MAOI or within the washout periods (FDA label).
• Uncontrolled high blood pressure, or heart or cerebrovascular disease that a rise in blood pressure could worsen: use with caution, and control hypertension first (FDA label).
• Bipolar disorder or a family history of mania, seizure disorders, narrow-angle glaucoma, bleeding disorders: caution or specialist review (FDA label).
• Children and adolescents: not approved. Two paediatric trials failed, and weight loss of 3.5% or more occurred in 22% (low dose) and 14% (high dose) vs 7% on placebo (FDA label).
• Pregnancy: there are no published studies of desvenlafaxine itself in pregnancy. The label draws on venlafaxine data, which have not shown a clear link with major birth defects. It warns of a possible increased risk of pre-eclampsia with mid-to-late pregnancy exposure, a less than 2-fold rise in postpartum haemorrhage with exposure in the month before delivery, and newborn complications after late exposure. It also cites evidence that stopping antidepressants in pregnancy raises relapse risk (FDA label). Do not stop suddenly on finding you are pregnant. Speak to your prescriber.
• Breastfeeding: in a study of 10 breastfeeding women, the mean relative infant dose was 6.8% (range 5.5–8.1%), and no adverse reactions were seen in the infants (FDA label). Discuss this with your prescriber.
• Older adults: higher rates of orthostatic hypotension (8% vs 2.5%) and hyponatraemia. Possible reduced kidney clearance should be considered when setting the dose (FDA label). The EMA assessors objected that too few elderly patients had been studied to establish a safe and effective dose. Exposure was about 50% higher in people over 75 (EMA assessment report). Note: we could not access the full 2023 AGS Beers Criteria tables during this review, so no Beers-specific wording is given here.
• Kidney impairment: maximum 50 mg/day in moderate impairment (creatinine clearance 30–50 mL/min). In severe impairment or end-stage disease, 25 mg daily or 50 mg every other day, with no extra doses after dialysis (FDA label).
• Liver impairment: in moderate to severe impairment the recommended dose is 50 mg/day, and escalation above 100 mg is not recommended (FDA label).

Dosage and how to take it

Source: FDA Pristiq label. Doses in other countries may differ. Australia's PBAC, for example, considered 50 mg and 100 mg tablets (PBAC).

How to take it

• Take it at about the same time each day. Swallow the tablet whole with fluid; do not divide, crush, chew or dissolve it (FDA label).
• An empty tablet shell in the stool is expected and harmless.
• Have your blood pressure checked before starting and regularly during treatment.

How long before it works

NICE advises that, if an antidepressant is going to work, benefits are usually felt within 4 weeks. It recommends a first review within 2 weeks, or at 1 week for people aged 18–25 or at risk of suicide. Treatment usually continues for at least 6 months after remission (NICE NG222).

How to stop safely

Do not stop suddenly. The label advises gradual reduction whenever possible, returning to the previous dose if symptoms are intolerable, then reducing more slowly. It notes that "discontinuation may need to occur over a period of several months" for some people (FDA label). NICE's approach for all antidepressants is to reduce step by step, each time to a proportion of the previous dose (for example 50%), with smaller steps (for example 25%) at lower doses. It advises allowing 1 to 2 weeks to judge each step, and notes that drugs with a short half-life "will need to be tapered more slowly" (NICE NG222). Because Pristiq tablets must not be split or crushed, very gradual tapers need a plan made with your prescriber. Do not improvise.

Follow the money: who makes it and who funded the evidence

Originator and current owner

• Originator: Wyeth (US). Wyeth held the original FDA application approved in February 2008. It also filed in Europe (Wyeth Europa Ltd) and Australia (Wyeth Australia Pty Ltd) (Drugs@FDA; EMA; PBAC). Wyeth also developed venlafaxine (Effexor), then its largest product (see our venlafaxine review).
• Pfizer (New York, US). Pfizer completed its acquisition of Wyeth in October 2009 (Pfizer press release). The current US label is issued by Wyeth Pharmaceuticals LLC, "a subsidiary of Pfizer Inc.", and the FDA lists the application holder as PF Prism C.V. (FDA label; Drugs@FDA).
• Timing. Pristiq was approved in 2008. Pfizer's filings show Effexor revenue falling 61% in 2011 as generics arrived, while Pristiq revenue rose 24% that year, "primarily driven by promotional activities in the U.S., and targeted international markets" (Pfizer 2011 Financial Report). Pure City Research notes the pattern of a patented metabolite launched as the parent drug faces generics. We do not infer intent beyond what the filings say.

Revenue (company-reported)

Documented legal and regulatory events

• EU application withdrawn (2008). Wyeth withdrew its EU application at day 166 of the 210-day review. The CHMP's provisional view was negative (EMA Q&A).
• Hot-flush indication not approved (2007–2012). An FDA "approvable" letter in July 2007 requested more safety data, a "complete response" letter followed in September 2011, and Pfizer withdrew the application in February 2012 (Pfizer 2011 Financial Report).
• Securities litigation. Pfizer's filing reports that in late 2007 and early 2008, purported class actions alleged that Wyeth and former officers "violated federal securities laws by misrepresenting the safety of Pristiq" before the July 2007 approvable letter for the hot-flush indication (Pfizer 2011 Financial Report). These were allegations. We did not find their final outcome in the filings we checked.
• Generic entry by settlement. US generics launched in March 2017 "as a result of a patent litigation settlement with several generic manufacturers" (Pfizer 2017 Financial Report).

Who funded the evidence

• Registration and maintenance trials: Wyeth, later Pfizer. The label data and the pooled analyses by Thase (2009) and Montgomery (2009) draw on the sponsor's trial programme (Thase 2009; Montgomery 2009). The Khan (2014) taper trial lists Pfizer-employed authors (Khan 2014).
• Independent and regulatory assessments: Cipriani 2018 (NIHR and JSPS), Henssler 2024 (unfunded), the EMA assessment (EU regulator) and Australia's PBAC (government payer). All of them relied on the same underlying company trials. Where they judged efficacy, they found it modest, not better than venlafaxine, and no better at doses above 50 mg.
• Guidelines: CANMAT's guideline had no industry funding, but many authors declared industry ties, including to Pfizer (CANMAT 2023).

Related research

• Venlafaxine: the parent drug, licensed in the UK
• Stress, anxiety and depression: prevention and management guide
• Supplements for stress, anxiety and depression: the evidence
• St John's wort: do not combine with SNRIs
• Saffron for depression
• Ashwagandha: evidence and safety
• Magnesium
• Sleep prevention guide (insomnia is a common side effect)
• Sibling medicine reviews: duloxetine, sertraline, escitalopram, fluvoxamine, vortioxetine, mirtazapine, bupropion

Frequently asked questions

Sources and funding notes

• Pristiq prescribing information (DailyMed / FDA): manufacturer label (Wyeth Pharmaceuticals LLC, a Pfizer subsidiary), approved by the US FDA. Efficacy data come from manufacturer trials.
• Drugs@FDA NDA 021992 and NDA 204150: US government approval records.
• EMA, Questions and answers on the withdrawal of the marketing application for Ellefore (2008), withdrawal assessment report and withdrawn application page: EU regulator (mostly fee-funded).
• PBAC Public Summary Document, desvenlafaxine (November 2008): Australian government medicines funder. Sponsor submission by Wyeth Australia.
• NICE NG222, Depression in adults (2022): UK government-funded guideline body.
• CANMAT 2023 depression guideline (Lam et al., 2024): Canada. Internal funds, with no industry funding for the guideline. Many authors declared industry ties.
• Cipriani et al., Lancet 2018 (full text): NIHR Oxford Health BRC (UK) and JSPS (Japan), with no funder role. Some authors declared pharma fees.
• Stone et al., BMJ 2022: FDA staff analysis of trial data submitted to the FDA. No external funding.
• Laoutidis & Kioulos, Pharmacopsychiatry 2015: Heinrich Heine University Düsseldorf (Germany). Funding and conflicts not shown in the PubMed record.
• Thase et al., CNS Spectr 2009: pooled analysis of the manufacturer's registration trials. Lead author at the University of Pennsylvania (US). Funding not shown in the PubMed record.
• Montgomery et al., Int Clin Psychopharmacol 2009: analysis of manufacturer trial data. Lead author at Imperial College (UK). Funding not shown in the PubMed record.
• Khan et al., J Clin Psychopharmacol 2014: taper trial with Pfizer-employed co-authors (US).
• Henssler et al., Lancet Psychiatry 2024: University of Cologne and Charité Berlin (Germany). No funding and no competing interests.
• Pinkerton et al., Menopause 2013: hot-flush trial. Sponsor not shown in the PubMed record.
• Berhan & Berhan, Ethiop J Health Sci 2014: academic meta-analysis (Ethiopia). Funding not shown in the PubMed record.
• EMA PRAC, May 2019 (persistent sexual dysfunction wording): EU regulator.
• Pfizer 2011 Financial Report, 2016 Financial Report, 2017 Financial Report and 2021 Form 10-K: company filings with the US SEC, used for revenue, regulatory and litigation facts.
• Pfizer statement on the Complete Response Letter for the VMS indication: company statement.
• Pfizer, "Pfizer Completes Acquisition of Wyeth" (2009): company press release.

Last reviewed: 26 September 2026. Educational information from Pure City Research, not medical advice. Medicine decisions belong with a qualified prescriber.