Duloxetine: Independent Evidence on Depression, Pain, Side Effects & Withdrawal

Table of contents

• Evidence summary
• What duloxetine 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 duloxetine
• 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

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator or government body; 2 = academic or Cochrane team with no declared commercial funding for the work; 3 = independent authors analysing mostly manufacturer-run trials, or minor declared industry ties; 4 = manufacturer-funded or manufacturer-authored. Credibility runs from A (highest) to D.

What duloxetine is

Duloxetine is an antidepressant used to treat depression, anxiety and nerve pain. In the UK it is only available on prescription (NHS). Pharmacologically it is a serotonin and norepinephrine (noradrenaline) reuptake inhibitor, or SNRI (FDA label). It is sold as gastro-resistant capsules. The coating lets the contents pass through the stomach intact so that stomach acid does not break down the active substance (EMA Cymbalta EPAR).

• Brand names: Cymbalta (depression, anxiety, nerve pain) and Yentreve (stress urinary incontinence in the EU/UK). Both are authorised to Eli Lilly Nederland B.V. (EMA Cymbalta, EMA Yentreve).
• First approvals: The US FDA approved Cymbalta (NDA 021427, sponsor Lilly) as a new molecular entity on 3 August 2004 (Drugs@FDA). The EU authorised Yentreve on 11 August 2004 and Cymbalta on 17 December 2004 (EMA, EMA).
• Originator: Eli Lilly and Company, headquartered at Lilly Corporate Center, Indianapolis, Indiana, US (Lilly SEC filings).
• Generic status: Cymbalta lost US patent exclusivity in December 2013, "resulting in the immediate entry of several generic competitors" (Lilly 10-K for 2013). Many generic duloxetine capsules are licensed in the UK (eMC listing).
• Prescription status: Prescription-only in both the UK and the US.

How it works

According to the FDA label, the exact mechanisms of duloxetine's antidepressant, pain-inhibiting and anti-anxiety actions in humans are unknown. They are believed to relate to increased serotonergic and noradrenergic activity in the central nervous system. In preclinical studies duloxetine was a potent inhibitor of serotonin and norepinephrine reuptake and a less potent inhibitor of dopamine reuptake. It has no significant affinity for dopaminergic, adrenergic, cholinergic, histaminergic, opioid, glutamate or GABA receptors, and it does not inhibit monoamine oxidase (FDA label §12).

The EMA explains that blocking reuptake leaves more serotonin and noradrenaline between nerve cells in the brain and spinal cord. Because these chemicals are involved in mood and in dampening pain signals, the effect can improve depression, anxiety and neuropathic pain (EMA Cymbalta EPAR). For stress urinary incontinence, the EMA says the mechanism "is not clear". The working theory is that higher serotonin and noradrenaline levels at the nerves controlling the urethral muscle close the urethra more strongly during urine storage (EMA Yentreve EPAR).

Pharmacokinetics that matter in practice:

• The half-life is about 12 hours (range 8–17), and steady state is reached after about 3 days.
• It is cleared mainly by the liver enzymes CYP1A2 and CYP2D6, which drives most of its interactions.
• More than 90% is bound to plasma proteins.
• Smokers have about one-third lower exposure.
• In moderate cirrhosis, exposure rose 5-fold.

(FDA label §8, §12.3)

What it is prescribed for

Licensed indications differ by country

Sources: FDA label, UK SmPC, EMA Yentreve, EMA refusal report. The NHS page also mentions nerve pain such as fibromyalgia, which in the UK is an off-label or guideline-based use (NHS).

Where guidelines place it

• Depression (NICE NG222): Duloxetine is not named individually. NICE says antidepressant treatment "can be" an SSRI, an SNRI or another antidepressant. SSRIs "should be considered as the first choice for most people". Switching to a different class, such as an SNRI, is a further-line option (NICE NG222). In practice this makes duloxetine a second-line option in UK depression care.
• Generalised anxiety disorder (NICE CG113): Offer an SSRI first, with sertraline considered first because it is the most cost-effective. If sertraline is ineffective, offer an alternative SSRI or an SNRI (NICE CG113). This also makes duloxetine second-line.
• Neuropathic pain (NICE CG173): Offer a choice of amitriptyline, duloxetine, gabapentin or pregabalin as initial treatment (except trigeminal neuralgia), so duloxetine is first-line (NICE CG173).
• Chronic primary pain, including fibromyalgia-type pain (NICE NG193): Consider amitriptyline, citalopram, duloxetine, fluoxetine, paroxetine or sertraline for adults after a full discussion of benefits and harms. This was off-label in April 2021 (NICE NG193).
• Low back pain (NICE NG59): "Do not offer selective serotonin reuptake inhibitors, serotonin–norepinephrine reuptake inhibitors or tricyclic antidepressants for managing low back pain" (NICE NG59). The UK advice is therefore not recommended, even though the FDA licenses duloxetine for chronic low back pain.
• Stress urinary incontinence (NICE NG123): Do not use duloxetine first-line, and do not routinely offer it second-line. It may be offered second-line if a woman prefers drug treatment to surgery or is not suitable for surgery (NICE NG123).

What works and what does not

Benefits by claim

Depression

The most complete independent comparison is the Oxford-led network meta-analysis of 522 double-blind trials (116,477 participants). It searched unpublished trials and regulatory websites as well as journals. Every one of the 21 antidepressants beat placebo, with odds ratios for response ranging from 2.13 (amitriptyline) to 1.37 (reboxetine) (Cipriani et al., Lancet 2018). In the study's forest plot of all trials against placebo, duloxetine's response OR was 1.85 (95% CrI 1.66–2.07), behind amitriptyline and mirtazapine. Its acceptability OR, which measures dropout for any reason, was 1.09 (0.96–1.23), meaning no significant difference from placebo (Cipriani 2018, figure 3). In the head-to-head analysis, duloxetine was among the drugs with the highest dropout rates (ORs 1.30–2.32 for that group).

What an OR of 1.85 means clinically: response is common on placebo too. That is why effect sizes shrink when trials are pooled honestly. In Lilly's four pivotal US depression trials, the drug–placebo difference on the 17-item Hamilton scale ranged from −2.2 to −4.9 points (FDA label, Table 8). In one of those trials, for example, placebo patients improved by 8.3 points and duloxetine patients by 10.5. Most of the improvement seen on the drug is therefore also seen on placebo. The drug adds a real but moderate increment. Published literature can also overstate antidepressant benefit. An FDA-records analysis of 12 antidepressants found that 94% of published trials looked positive, compared with 51% in the FDA's own analysis, and that publication inflated effect sizes by 32% overall (Turner et al., NEJM 2008). This is why the independent reviews above deliberately included unpublished data.

Against other antidepressants, a Cochrane review of 16 head-to-head RCTs found no statistically significant efficacy differences. More patients dropped out on duloxetine than on escitalopram (OR 1.62, 95% CI 1.01–2.62) or venlafaxine (OR 1.56, 1.14–2.15) (Cipriani et al., Cochrane 2012). For relapse prevention, patients who responded to duloxetine and continued it had a longer time to relapse than those switched to placebo (FDA label, Study MDD-5).

Generalised anxiety disorder

An unfunded UCL network meta-analysis covered 89 trials and 25,441 patients. Duloxetine (MD −3.13), pregabalin (−2.79), venlafaxine (−2.69) and escitalopram (−2.45) were more effective than placebo "with relatively good acceptability". Quetiapine had a larger effect (−3.60) but was poorly tolerated (Slee et al., Lancet 2019). A 3-point average difference on the Hamilton Anxiety scale is meaningful but modest. NICE still recommends trying an SSRI (sertraline) first for cost-effectiveness reasons (NICE CG113).

Diabetic nerve pain

Cochrane found duloxetine 60 mg daily effective for painful diabetic peripheral neuropathy in the short term. The chance of at least 50% pain reduction at 12 weeks had an RR of 1.73 (95% CI 1.44–2.08), and the number needed to treat for one extra person to benefit was 5 (4–7). Lower daily doses did not work. The reviewers concluded there was "adequate amounts of moderate quality evidence from eight studies performed by the manufacturers" and that further trials were not required (Lunn et al., Cochrane 2014). NICE lists duloxetine as one of four first-choice options (NICE CG173). The UK SmPC advises assessing response after 2 months, as extra response after that is unlikely (UK SmPC).

Chronic pain across conditions

The 2023 Cochrane network meta-analysis compared 25 antidepressants across 176 trials. It concluded that "the only antidepressant we are certain about for the treatment of chronic pain is duloxetine". Duloxetine 60 mg gave substantial pain relief with OR 1.91 (1.69–2.17) and reduced pain intensity with SMD −0.31 (−0.39 to −0.24), both moderate-certainty evidence. The standard dose worked as well as a high dose (Birkinshaw et al., Cochrane 2023). The same authors stress several limits:

• Trials excluded people with low mood.
• There is no reliable long-term evidence.
• Safety evidence was very low certainty.
• The findings "should not be read as an encouragement to prescribe antidepressants where other non-pharmacological intervention could be equally effective" (Birkinshaw et al., HTA 2024).

Fibromyalgia

Cochrane reported RR 1.57 for ≥50% pain relief at 12 weeks (NNTB 8) and RR 1.58 at 28 weeks. It noted that an NNTB of 8 "is not an indication of substantial efficacy" and that the benefit may come more from improved mental symptoms than from reduced physical pain (Lunn et al., Cochrane 2014). When Lilly applied to add fibromyalgia to the European licence, the EMA's scientific committee concluded in October 2008 that the benefit/risk balance "remains negative". It found that the effect size "is at best rather smaller than the one observed for other therapies", that the data "show a clear link between drug effect and mood", and that "the only study including EU patients was negative" (EMA refusal assessment report). The FDA label lists one adult fibromyalgia trial (FM-3, 16 weeks) that did not demonstrate efficacy (FDA label §14).

Back pain and osteoarthritis

An independent BMJ meta-analysis of 33 trials found moderate-certainty evidence that SNRIs (mainly duloxetine) reduce back pain by 5.30 points on a 0–100 scale at 3–13 weeks. The authors concluded this effect is "small and not clinically important". For osteoarthritis the reduction was 9.72 points (low certainty), and a clinically important effect "cannot be excluded" (Ferreira et al., BMJ 2021). This is the clearest example of a gap between regulatory approval (the FDA's chronic musculoskeletal pain licence) and independent judgement of clinical value.

Stress urinary incontinence

Cochrane pooled 9 RCTs (3,327 adults). Duloxetine improved incontinence quality of life (WMD 5.26) and roughly halved leak frequency. Objective pad-test measures showed no benefit, and subjective cure favoured duloxetine by only 3%. About one in three reported treatment-related side effects, most often nausea, and about one in eight stopped because of them (Mariappan et al., Cochrane 2005). In the EMA summary, leaks fell by 52% on Yentreve vs 33% on placebo. The drug beat placebo only in women with more than 14 episodes a week (EMA Yentreve).

Risks and all side effects

The NHS lists these common side effects (NHS):

• headaches
• nausea and vomiting
• dizziness or drowsiness
• dry mouth
• diarrhoea or constipation
• insomnia
• sexual problems
• weight loss

Most ease within a couple of weeks. In pooled US placebo-controlled trials (8,100 on duloxetine vs 5,655 on placebo), the most frequent effects were those below. Across trials, 8.4% (depression) to 17.5% (fibromyalgia) of patients stopped because of side effects, versus 4.6% to 10.1% on placebo (FDA label §6.1).

Withdrawal (discontinuation) symptoms

The UK SmPC states that withdrawal symptoms are common, particularly after abrupt stopping. In clinical trials, adverse events after abrupt discontinuation occurred in about 45% of patients on duloxetine and 23% on placebo. Symptoms usually start within the first few days and are generally mild to moderate, but can be severe. The SmPC advises reducing the dose over at least one to two weeks (UK SmPC). The FDA label lists these symptoms after stopping: dizziness, headache, nausea, diarrhoea, paraesthesia, irritability, vomiting, insomnia, anxiety, excessive sweating and fatigue. For SSRIs and SNRIs generally it also notes "electric shock sensations", confusion, tinnitus and seizures (FDA label §5.7).

An unfunded 2024 meta-analysis across all antidepressants put the incidence of at least one discontinuation symptom at 31% after stopping an antidepressant and 17% after stopping placebo. Severe symptoms occurred in 2.8% vs 0.6%. It estimated the drug-attributable incidence at about 15%, or one in six to seven patients. Venlafaxine and desvenlafaxine, the SNRIs most similar to duloxetine, were among those with more frequent symptoms. The review did not single out duloxetine (Henssler et al., Lancet Psychiatry 2024). The NHS says your doctor will reduce the dose gradually "over several weeks or months" (NHS).

Dependence: duloxetine is not a drug of addiction in the classic sense. There was no drug-seeking behaviour in trials, and it showed no dependence-producing potential in rats (FDA label §9). The body does adapt to it, however, which is why withdrawal symptoms occur. NICE addresses antidepressants in its guidance on medicines associated with dependence or withdrawal symptoms (NICE NG193).

All interactions

Sources: FDA label §4, §5.12, §7; UK SmPC 4.3–4.5; NHS.

Who should avoid duloxetine

• Must not use (contraindications):
  • allergy to duloxetine
  • current or recent MAOI use
  • liver disease resulting in hepatic impairment
  • severe kidney impairment (creatinine clearance below 30 mL/min)
  • use with fluvoxamine, ciprofloxacin or enoxacin
  • starting treatment with uncontrolled high blood pressure
  (UK SmPC 4.3). The FDA advises avoiding it in chronic liver disease, cirrhosis, substantial alcohol use and severe renal impairment (FDA §2.7, §5.2).
• Needs specialist caution: epilepsy, heart disease, glaucoma, or a history of mania (NHS). Also bipolar disorder, a recent heart attack or unstable coronary disease (these patients were excluded from trials), conditions that slow stomach emptying, and bleeding disorders (FDA §5).
• Children and young people: In the US it is licensed for GAD from age 7 and fibromyalgia from age 13. It failed in two paediatric depression trials (FDA §8.4). In the UK it should not be used under 18 for depression (UK SmPC). The boxed suicidality warning applies up to age 24.
• Pregnancy: The NHS says duloxetine can be used in pregnancy if needed, at the lowest effective dose, and hospital birth may be advised (NHS). The UK SmPC reports several findings:
  • Two large observational studies (about 2,500 and 1,500 first-trimester exposures) do not suggest an overall increased risk of major malformations.
  • In the EU study, late-pregnancy use was linked to preterm birth: less than 2-fold, or about 6 extra premature births per 100 women.
  • US data show a less than 2-fold increased risk of postpartum haemorrhage with use in the month before birth.
  • A possible persistent pulmonary hypertension of the newborn signal is described for SSRIs.
  (UK SmPC 4.6). Stopping antidepressants in pregnancy also carries relapse risk (FDA §8.1). Decide with a clinician; do not stop abruptly.
• Breastfeeding: The NHS says it is sometimes used and it is rare for babies to have side effects, but check with your doctor (NHS). The UK SmPC does not recommend it while breastfeeding. It estimates the infant dose at about 0.14% of the maternal dose (mg/kg) (UK SmPC). The FDA notes reports of sedation, poor feeding and poor weight gain in exposed infants (FDA §8.2).
• Older adults: The American Geriatrics Society 2023 Beers Criteria say to avoid SNRIs in people with a history of falls or fractures unless safer alternatives are not available, noting "newer evidence suggests that SNRIs may increase falls risk". They also advise using SNRIs with caution because of hyponatraemia/SIADH, with close sodium monitoring (AGS Beers Criteria 2023). The FDA label reports more falls than on placebo, and hyponatraemia more often in older people (FDA §8.5).
• Liver and kidneys: In moderate cirrhosis, clearance fell to about 15% of normal, with a 5-fold rise in exposure. In end-stage kidney disease on dialysis, duloxetine levels doubled and major metabolite levels rose 7- to 9-fold (FDA §8.9–8.10).

Dosage and how to take it

Sources: FDA label §2; UK SmPC 4.2; EMA Yentreve.

• How to take it: Swallow the capsule whole with water, with or without food, once or twice a day as prescribed (NHS). Do not crush, chew or open the capsule, because this can damage the enteric coating (FDA §2.1).
• Missed dose: Take it when you remember, unless it is nearly time for the next one. Never double up (NHS).
• Too much: Taking an extra dose can be dangerous. In the UK, call NHS 111. If told to go to A&E, do not drive yourself (NHS).
• How long before it works: For depression, a response is usually seen after 2–4 weeks (UK SmPC). In the diabetic nerve pain trials, pain reduction was seen from the first week (EMA Cymbalta). After a response, treatment usually continues for several months to prevent relapse.
• How to stop: Only under supervision, by tapering. The SmPC minimum is one to two weeks. The NHS describes reductions over several weeks or months. If symptoms are intolerable, the prescriber may return to the previous dose and then reduce more slowly (UK SmPC, NHS).

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

Who makes and sells it

Duloxetine was developed and first licensed by Eli Lilly and Company, based in Indianapolis, Indiana, US (Drugs@FDA, Lilly 10-K). In Europe, both Cymbalta and Yentreve are authorised to Eli Lilly Nederland B.V. in Utrecht, the Netherlands (EMA). Since patents expired, many generic companies make it. UK product listings include generics from Dr. Reddy's, Milpharm, Neuraxpharm, Zentiva and Amarox, among others (eMC).

How much money was at stake

Lilly's own annual report for 2013 shows Cymbalta worldwide revenue of $5,084.4 million in 2013 ($3,960.8 million in the US), up from $4,994.1 million in 2012. That made it the company's largest product, at 17% of worldwide revenue (Eli Lilly 10-K for fiscal 2013). US patent exclusivity ended in December 2013, and generic competitors entered immediately.

Who funded the pivotal trials

• The trials behind the licences were Lilly's. The FDA label's efficacy sections describe the company's studies (MDD-1 to MDD-5, GAD-1 to GAD-6, DPNP-1/2, FM-1 to FM-4, CLBP-1 to CLBP-3, OA-1/2). To its credit, the label also lists trials that failed: FM-3, CLBP-2 and OA-2 (FDA label §14).
• Cochrane's pain review states that "almost every study [was] performed or sponsored by the drug manufacturer". It called for "preferably independent investigator led studies" (Lunn et al., Cochrane 2014).
• Cochrane's depression comparison states that "most of included studies were sponsored by the drug industry manufacturing duloxetine" and warns of possible "overestimation of treatment effect due to sponsorship bias" (Cipriani et al., Cochrane 2012).

Who funded the independent reviews

• The Oxford-led depression network meta-analysis was funded by the UK's NIHR and the Japan Society for the Promotion of Science (Cipriani 2018).
• The UCL anxiety network meta-analysis received no funding (Slee 2019).
• The German discontinuation meta-analysis received no funding, and its authors declared no competing interests (Henssler 2024).
• The only large publicly funded duloxetine trial we found in this review ran through US National Cancer Institute cooperative networks (Smith 2013).

The independent reviews confirm that duloxetine works for its core indications. They also consistently temper the size and practical value of the benefit, especially for fibromyalgia and back pain.

Documented regulatory and legal events

• EMA refusal (2008): Europe's regulator rejected Lilly's application to add fibromyalgia. It judged the benefit/risk balance negative and noted the effect was linked to mood (EMA refusal report). The US FDA licenses the same indication.
• Discontinuation litigation: Lilly disclosed that it was named in a purported US class action (Saavedra et al. v. Eli Lilly and Company, filed October 2012) and in about 45 individual lawsuits. Both concerned alleged injuries from discontinuing Cymbalta. The district court denied class certification in December 2014, and a request for multi-district litigation was denied the same month. Lilly stated it believed the claims were "without merit" (Eli Lilly 10-K for fiscal 2014). We did not verify the final outcomes of these cases for this article.
• Publication bias context: An analysis of FDA files found that, across antidepressants, negative trials were often unpublished or published as positive (Turner et al., NEJM 2008). This is a class-wide finding, not a finding about duloxetine specifically.

Related research

For wider, evidence-graded context, see these Pure City Research guides:

• Stress, anxiety and depression: prevention and management guide
• Supplements for stress, anxiety and depression: the evidence
• St John's wort (do not combine with duloxetine)
• Saffron for depression
• Sleep: prevention and management guide
• Magnesium

Sibling medicine reviews:

• Sertraline
• Escitalopram
• Venlafaxine
• Amitriptyline
• Mirtazapine
• Fluoxetine
• Paroxetine
• Vortioxetine

Frequently asked questions

Sources and funding notes

• NHS, Duloxetine medicine page (reviewed 18 June 2026): UK National Health Service, publicly funded. No commercial sponsor.
• FDA-approved prescribing information, Cymbalta (Eli Lilly; label effective July 2025) via DailyMed: US regulator-approved text written by the manufacturer. Efficacy data come from Lilly trials.
• Drugs@FDA, NDA 021427 (Cymbalta): US FDA approval record, government.
• EMA, Cymbalta EPAR and EMA, Yentreve EPAR: EU regulator summaries. Marketing authorisation holder is Eli Lilly Nederland B.V.
• EMA/CHMP, Refusal assessment report, Cymbalta fibromyalgia variation (23 October 2008): EU regulator. Independent of the applicant, but assessing the applicant's data.
• UK Summary of Product Characteristics, duloxetine 60 mg gastro-resistant capsules (Amarox Limited, revised 29/05/2025): MHRA-authorised generic label. Its wording mirrors the originator's.
• electronic Medicines Compendium, duloxetine product listing: UK database of licensed product information. Used only to identify generic licence holders.
• NICE NG222, Depression in adults (2022): UK public guideline body.
• NICE CG113, Generalised anxiety disorder and panic disorder in adults: UK public guideline body.
• NICE NG193, Chronic pain (primary and secondary) in over 16s: UK public guideline body.
• NICE CG173, Neuropathic pain in adults: pharmacological management: UK public guideline body.
• NICE NG123, Urinary incontinence and pelvic organ prolapse in women: UK public guideline body.
• NICE NG59, Low back pain and sciatica in over 16s: UK public guideline body.
• Cipriani et al., Lancet 2018 (open-access full text): funded by NIHR Oxford Health Biomedical Research Centre and the Japan Society for the Promotion of Science. UK-led international academic team.
• Slee et al., Lancet 2019: University College London (UK). "No funding was received for this research."
• Lunn, Hughes & Wiffen, Cochrane 2014: Cochrane (UK). The authors note almost every included trial was performed or sponsored by the manufacturer. One author declared honoraria from immunoglobulin manufacturers (unrelated to duloxetine).
• Birkinshaw et al., Cochrane 2023 and HTA 2024 version: academic team (University of Southampton, UK). Most authors declared no relevant interests.
• Cipriani et al., Cochrane 2012 (duloxetine vs other antidepressants): Cochrane (University of Verona, Italy). The authors note most trials were manufacturer-sponsored. One co-author declared speaking honoraria from several companies, including Eli Lilly.
• Mariappan et al., Cochrane 2005 (SNRIs for stress urinary incontinence): Cochrane (Edinburgh, UK). Conflict of interest "None known". Funding of the underlying trials not stated in the abstract.
• Ferreira et al., BMJ 2021: University of Sydney (Australia). No support for the submitted work. Some authors had unrelated industry ties.
• Smith et al., JAMA 2013: US National Cancer Institute-funded cooperative group trial (Alliance for Clinical Trials in Oncology). Publicly funded.
• Henssler et al., Lancet Psychiatry 2024: German academic team. "Funding: None." Authors declared no competing interests.
• Turner et al., NEJM 2008: academic and US Veterans Affairs authors (Oregon, US). Based on FDA review documents.
• American Geriatrics Society 2023 Updated Beers Criteria: US professional society. Panel conflicts disclosed and conflicted panellists recused.
• Eli Lilly and Company, Form 10-K for fiscal 2013 and Form 10-K for fiscal 2014: manufacturer's statutory SEC filings (US). The company's own figures and legal characterisations.

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