Gabapentin: Independent Evidence on Pain, Epilepsy & Misuse Risk

Table of contents

• Evidence summary
• What gabapentin 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 gabapentin
• 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.

Sources cluster in the UK and US. Only one Canadian and one Australian source informs the picture, and none came from a regulator in Asia, Africa or South America. A UK and US lens on prescribing and misuse is therefore built in.

What gabapentin is

Gabapentin is an anticonvulsant and nerve-pain drug. Its chemical name is 1-(aminomethyl)cyclohexaneacetic acid, and despite the name it is not a GABA drug (FDA label). The original brand is Neurontin, sold as capsules, tablets and an oral solution. The FDA's 2019 notice also names two related gabapentin products, Gralise and Horizant, and the sister drug pregabalin (Lyrica) (FDA 2019). Gabapentin was first approved by the FDA in 1993. Many UK generics exist from companies such as Sandoz, Tillomed and Glenmark (emc listing).

Legal status

• UK: a Class C drug under the Misuse of Drugs Act 1971 and Schedule 3 under the 2001 Regulations since 1 April 2019. It is exempt from safe-custody rules (MHRA).
• US: the current label states that Neurontin "is not a controlled substance" (FDA label, section 9.1). We did not check individual US state rules.

How it works

Gabapentin binds with high affinity to the alpha-2-delta subunit of voltage-gated calcium channels. The UK product information says this may reduce release of excitatory chemical messengers in the nervous system (UK SmPC, section 5.1). It does not bind GABA receptors and does not change GABA metabolism. The US label adds that it has no affinity for benzodiazepine, opioid or cannabinoid receptors (FDA label, section 9.2).

Proven in humans versus lab only: the calcium-channel binding and the anti-seizure effect were shown in animal models. The UK document says the relevance of these actions to the anticonvulsant effect in humans "remains to be established" (UK SmPC). What is proven in humans is the outcome in trials: some pain and seizure relief, and sedation.

What it is prescribed for

Licensed uses

The Justice Department states that Neurontin was approved in December 1993 "only" as an add-on seizure treatment (DOJ 2004). Everything else it is used for, from migraine to anxiety, has been off-label.

Where guidelines place it

• Neuropathic pain (NICE CG173): offer "a choice of amitriptyline, duloxetine, gabapentin or pregabalin" as initial treatment, and if that fails offer one of the remaining three (NICE CG173 1.1.8–1.1.9). NICE advises carbamazepine, not gabapentin, for trigeminal neuralgia.
• Sciatica and low back pain (NICE NG59): do not offer gabapentinoids "as there is no overall evidence of benefit and there is evidence of harm" (NICE NG59 1.2.16); do not offer them for low back pain (1.2.30).
• Epilepsy (NICE NG217): gabapentin appears in none of the first-, second- or third-line lists for focal seizures, and the committee noted that "gabapentin and clobazam are no longer recommended" as add-ons. NICE also warns that gabapentin may worsen absence and myoclonic seizures (NICE NG217).

How much it is used in England

In 2025/26 community pharmacies in England dispensed 7,359,436 items of gabapentin at a net cost of about £22.6 million. That ranked 42nd among 2,047 chemical substances by items. Pregabalin ranked 28th, with 10,058,446 items at about £21.1 million (NHSBSA PCA 2025/26 summary tables; release page: NHSBSA). These counts do not say why it was prescribed.

What works and what does not

Benefits by claim

Nerve pain after shingles and in diabetes

The most complete independent analysis is the Cochrane review by Wiffen and colleagues: 37 studies and 5,914 people, mostly at 1,200 mg or more daily for 4 to 12 weeks. About 3 or 4 in 10 people got at least 50% pain relief with gabapentin, compared with 1 or 2 in 10 on placebo (Wiffen 2017). Put plainly, around 6 people must be treated for one extra person to get a large benefit. The earlier review rated its evidence second tier, saying there was "no top tier evidence that was unequivocally unbiased" (Moore 2014). The US label found no extra benefit above 1,800 mg a day (FDA label, 2.1).

Epilepsy

For people whose focal seizures continue despite other drugs, add-on gabapentin cut seizures by half or more in more people than placebo (RR 1.89, 95% CI 1.40 to 2.55; 6 studies, 1,206 people). Withdrawal from treatment was no different from placebo (RR 1.05). The reviewers state the trials were short and give "no evidence for the long-term efficacy of gabapentin beyond a three-month period", and the results "cannot be extrapolated to monotherapy" (Panebianco 2021). Placebo-controlled trials tested roughly 600 to 1,800 mg a day, with about 25% responding at 1,800 mg (full text). NICE now lists other add-ons ahead of it (NG217).

Low back pain and sciatica

This is the clearest case of off-label overuse. Nine placebo-controlled trials of topiramate, gabapentin or pregabalin (859 people) found no pain or disability benefit in 14 of 15 comparisons, with more side effects (Enke 2018). NICE agrees (NG59).

Anxiety, insomnia and bipolar disorder

A systematic review of 55 double-blind trials of gabapentin and pregabalin found moderate evidence for anxiety states but minimal evidence for bipolar disorder and insomnia (Hong 2022). Gabapentin cut preoperative anxiety more than placebo (SMD −0.92, 95% CI −1.32 to −0.52), which says little about long-term anxiety disorders. For those, see our generalised anxiety disorder guide.

Gabapentin versus pregabalin

Both bind the same calcium-channel subunit and both became Class C drugs in the UK on the same date (MHRA). NICE treats them as interchangeable first-line options for neuropathic pain. For focal epilepsy NICE lists pregabalin, but not gabapentin, as a second-line add-on (NG217 5.2.5). MHRA counted more UK reports of abuse and dependence for pregabalin (113 and 98) than for gabapentin (11 and 9) to 10 April 2019. Those are report counts, not rates, and prescribing volumes differ. We cover that drug in our pregabalin review.

Risks and all side effects

The US label has no boxed warning in the text we reviewed. Its warnings are listed in section 5, and its most common adverse reactions in trials are dizziness, somnolence and peripheral oedema (nerve pain), and somnolence, dizziness, ataxia, fatigue and nystagmus (epilepsy in people over 12) (FDA label).

Common side effects (licensing trials)

Source: FDA label, section 6.1. In the postherpetic neuralgia trials 16% on gabapentin and 9% on placebo stopped treatment because of an adverse reaction. The Cochrane pain review pooled all conditions: at least one adverse event in 63% versus 49% on placebo (NNH 7.5), withdrawal because of adverse events 11% versus 8.2% (NNH 30), and serious adverse events 3.2% versus 2.8%, which was not a significant difference (Wiffen 2017). In the epilepsy review dizziness had RR 2.43 and somnolence RR 1.93 against placebo (Panebianco 2021).

The UK document lists other effects, including weight gain, increased appetite, blurred or double vision, tremor, dry mouth, constipation, nausea, impotence, low white blood cells and confusion as common. Rare or unknown-frequency effects include respiratory depression, pancreatitis, hepatitis, hyponatraemia, rhabdomyolysis, gynaecomastia and worsening myasthenia gravis (UK SmPC 4.8).

Serious and rare risks

• Suicidal thoughts: pooled analyses of 199 antiepileptic-drug trials found about twice the risk of suicidal thinking or behaviour versus placebo (adjusted RR 1.8, 95% CI 1.2 to 2.7). Suicidal behaviour has also been reported after stopping gabapentin (FDA label 5.5).
• Falls and accidents: dizziness and sleepiness can raise the chance of falls; confusion and loss of consciousness have been reported after marketing (UK SmPC 4.4).
• Seizures: a small increase in seizure frequency or new seizure types can occur, and gabapentin may worsen absence seizures (UK SmPC).
• Children aged 3 to 12: emotional lability, hostility, thought disorder and hyperactivity were seen in epilepsy trials (FDA label 5.9).

Dependence, misuse and withdrawal

The UK document says gabapentin "can cause drug dependence, which may occur at therapeutic doses", with higher risk in people with past substance misuse or mental health conditions, and says tolerance can develop (UK SmPC 4.4). The US label says misuse has been reported, mostly in people with a history of polysubstance abuse, but also that the abuse and dependence potential "has not been evaluated in human studies" (FDA label 9.2–9.3). Withdrawal symptoms reported after stopping include seizures, agitation, confusion, anxiety, insomnia, nausea, pain, sweating, tremor, headache and low mood, usually within 48 hours. The label says to taper over at least a week. We give no tapering advice; that is for the prescriber.

Overdose

Overdose symptoms include double vision, slurred speech, drowsiness, lethargy and diarrhoea, and fatal respiratory depression has been reported. Gabapentin can be removed by dialysis (FDA label, section 10).

All interactions

Gabapentin is "not appreciably metabolized" and does not interfere with the metabolism of common antiepileptic drugs (FDA label 7.2). That makes its main interaction pharmacodynamic: adding sedation on sedation. Please see our guides to clonazepam and opioid-induced constipation for related medicines.

Who should avoid gabapentin

Contraindications

The only UK contraindication is hypersensitivity to gabapentin or an ingredient (UK SmPC 4.3). It must not be restarted after Stevens-Johnson syndrome, toxic epidermal necrolysis or DRESS.

Use with care

• Breathing and sedation risks: people with chronic obstructive pulmonary disease, other lung or neurological disease, kidney impairment, or taking opioids or other depressants (UK SmPC 4.4).
• Past misuse: people with a history of substance or alcohol misuse or mental illness, who face a higher risk of dependence.
• Myasthenia gravis: exacerbations were reported after marketing.
• Absence or myoclonic seizures: NICE says gabapentin may make these worse (NG217).
• Back pain or sciatica: NICE advises against starting it (NG59).

Older adults

Gabapentin is cleared almost entirely by the kidneys. In the postherpetic neuralgia trials, people aged 75 and over had a larger treatment effect at the same dose, which the label links to reduced kidney function (FDA label 8.5). Dose must be adjusted in reduced renal function (label 2.3–2.4).

Pregnancy and breastfeeding

The US label says the totality of cohort data "has not indicated an increased risk of major birth defects or miscarriage", but warns of important methodological limitations. Animal studies showed developmental toxicity at or below clinical doses. Neonatal withdrawal has been reported when gabapentin is combined with opioids close to delivery (FDA label 8.1). The UK document cites a Nordic study of more than 1,700 first-trimester exposures with no higher risk of major malformations or neurodevelopmental disorders, and limited evidence of higher risk of low birth weight and preterm birth. It says gabapentin can be used in the first trimester "if clinically needed" and should not be stopped suddenly in epilepsy (UK SmPC 4.6). Gabapentin passes into breast milk, and its effect on the infant is unknown. Any decision on pregnancy belongs with a specialist. Our sleep and pregnancy guide covers wider pregnancy issues.

Children

The US label sets add-on use in epilepsy at age 3 and over, and the UK at age 6 and over. The effect of long use (over 36 weeks) on learning and development in children "has not been adequately studied" (UK SmPC).

Driving

Both labels warn that gabapentin can cause drowsiness and dizziness and that people should not drive until they know how it affects them (FDA label 5.3–5.4). See drowsy driving.

Dosage and how to take it

These are label and trial doses, not personal advice.

• US label, postherpetic neuralgia: 300 mg on day 1, 600 mg a day on day 2, 900 mg a day on day 3, then titration to 1,800 mg a day (600 mg three times daily). Efficacy was shown at 1,800 to 3,600 mg a day, without proven extra benefit above 1,800 mg (FDA label 2.1).
• US label, epilepsy (age 12 and over): starting dose 300 mg three times a day; maintenance 300 to 600 mg three times a day; the gap between doses should not exceed 12 hours (FDA label 2.2).
• UK label, adults and adolescents: effective range 900 to 3,600 mg a day in trials, maximum 3,600 mg a day, increased in 300 mg steps every 2 to 3 days (UK SmPC 4.2).
• Trial doses: the Cochrane pain review pooled doses of 1,200 mg or more a day; the epilepsy review pooled 600 to 1,800 mg a day.
• Stopping: the US label says to taper over at least a week. The UK document says to discuss a plan for ending treatment before starting, and to use the shortest duration possible outside epilepsy (UK SmPC 4.2).
• Antacids: separate by at least 2 hours.

Product quality was not a theme in our sources; gabapentin is sold as a licensed medicine. Medicine bought online is a separate risk, and the UK document asks prescribers to ask about medicines obtained online.

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

Who made it, and who sells it now

• Originator: Parke-Davis, a division of Warner-Lambert, which Pfizer acquired in June 2000 (DOJ 2004). Pfizer's registered business address is in New York (SEC EDGAR).
• Brand today: DailyMed lists a current Neurontin label from Viatris Specialty LLC (version published 6 August 2026) and an older one from "Parke-Davis Div of Pfizer Inc" (April 2026) (DailyMed listing). The UK Neurontin product is held by Viatris, "formerly Mylan or Upjohn" (UK SmPC). Viatris Inc. has its SEC business address in Canonsburg, Pennsylvania (SEC EDGAR).
• Generics: many UK manufacturers (emc listing). Gabapentin is now a cheap generic: about £3 per NHS item in England in 2025/26, from the NHSBSA figures above. No single company has a strong incentive to fund new trials.
• We could not open a document describing how the brand moved from Pfizer to Viatris, or any citable sales figure for Neurontin at its peak.

Documented integrity event: the Neurontin off-label marketing case

On 13 May 2004 the US Department of Justice announced that Warner-Lambert had agreed to plead guilty to two counts under the Food, Drug and Cosmetic Act and pay more than $430 million. Pfizer, the parent, agreed to a corporate compliance programme. The charged conduct predated Pfizer's June 2000 purchase (DOJ 2004). The money was split:

• $240 million criminal fine, which the release calls the second largest criminal fine ever imposed in a health care fraud prosecution.
• $190 million civil: $83.6 million to the federal government for Medicaid, $68.4 million to states for Medicaid, and $38 million to states for consumer protection.

Prosecutors alleged that Parke-Davis promoted Neurontin for bipolar disorder, pain, migraine, restless legs syndrome, attention deficit disorder, ALS, alcohol-withdrawal seizures and as first-line single treatment for epilepsy, none of which was approved. Alleged tactics included "medical liaisons" posing as scientific experts, lavish paid "consultant" meetings, and planted questions at sponsored education events. The department also alleged the company chose not to seek approval for new uses partly to protect a planned follow-on drug. These are the department's allegations as set out in the settlement announcement.

Who funded the evidence

• Trial publication bias: Vedula and colleagues obtained internal Pfizer and Parke-Davis documents for 20 off-label gabapentin trials, of which 12 were published. Eight of the 12 had a main outcome in the paper that differed from the protocol. Trials with non-significant main outcomes "either were not reported in full or were reported with a changed primary outcome" (Vedula 2009). A follow-up of 21 trials found the term "intent to treat" was defined six different ways, so the numbers analysed often disagreed between document and paper (Vedula 2013). The 2013 paper discloses that one author was paid by plaintiffs' lawyers for help with an expert report in litigation against Pfizer. We found no funding statement for the 2009 paper in the PubMed record and could not open the full text.
• Epilepsy evidence: the Cochrane review records that Parke-Davis sponsored nine of the included trials and Pfizer funded one more. The review itself was funded by the UK NIHR (Panebianco 2021).
• Pain evidence: the 2014 review states that no author received funds from a company with an interest in gabapentin for gabapentin research, though several disclose consulting and lecture fees from Pfizer and others (Moore 2014). The 2017 update discloses Pfizer-funded research grants for one author's laboratory and a 2019 audit finding that it did not meet the current Cochrane sponsorship definition (Wiffen 2017).
• Harm evidence: the opioid-death study was funded by the Ontario Ministry of Health (Gomes 2017); the psychiatric-use review by UK Department of Health research grants (Hong 2022).

What this means: The benefit of gabapentin for shingles-related and diabetic nerve pain, and as a short-term epilepsy add-on, rests largely on trials run by its maker. Independent reviewers pooled those trials and found a modest, real effect. Where the drug spread beyond its licence, the picture is worse: the maker promoted it illegally, its off-label trials were selectively reported, and independent re-testing for back pain found nothing. The safety concerns come largely from regulators and publicly funded observational studies, which have no commercial reason to overstate them. Both sides of the question point the same way: the licensed uses hold up modestly, and the wider uses do not.

Related research

• Pregabalin
• Duloxetine
• Amitriptyline
• Nortriptyline
• Clonazepam
• Buspirone
• Generalized anxiety disorder
• Restless legs syndrome
• Antidepressant withdrawal and dependence
• Opioid-induced constipation
• Drowsy driving

Frequently asked questions

Sources and funding notes

What we could not verify: whether the FDA 2019 notice says anything about risk in healthy people taking a gabapentinoid alone (the notice text we opened is silent, so we make no claim); the full text and funding statement of Vedula 2009 (NEJM blocked automated access); a funding page for the FDA, MHRA, NICE and NHSBSA (we describe them as UK or US government bodies without citing a funding document); any citable peak sales figure for Neurontin; how the brand passed from Pfizer to Viatris; the funding of the Enke 2018 review; and current US state-level legal status. We did not review trials of gabapentin for alcohol use disorder, migraine, hot flushes, post-operative pain or cough, so we make no claim about them. Wording differs across country labels, and we cite the versions named below. The FDA 2019 notice states no case counts, so we give none.

1. FDA-approved Neurontin prescribing information (DailyMed, Viatris Specialty LLC, version published August 2026) — role: US regulator-approved label; funder: US government regulator, text prepared with the marketer.
2. FDA Drug Safety Communication, 19 December 2019 — role: breathing warning; funder: US government.
3. MHRA Drug Safety Update, April 2019 — role: UK control of gabapentin; funder: UK government.
4. Neurontin 300 mg capsules, UK SmPC (Viatris; text revised February 2026) — role: UK licensed uses, warnings, pregnancy; funder: marketing-authorisation holder, MHRA-approved.
5. emc search listing for gabapentin — role: identifies UK generic manufacturers; funder: industry-funded database.
6. US Department of Justice press release, 13 May 2004 — role: Warner-Lambert plea and settlement; funder: US government.
7. Vedula et al., NEJM 2009 — role: outcome reporting in off-label gabapentin trials; funder: not stated in the record we opened.
8. Vedula, Li and Dickersin, PLoS Med 2013 — role: analysis definitions in the same trials; funder: not stated in the text we read; one author paid by plaintiffs' lawyers (declared).
9. Panebianco et al., Cochrane 2021 (full text on PMC) — role: add-on epilepsy review; funder: NIHR (UK); included trials sponsored by Parke-Davis and Pfizer.
10. Wiffen et al., Cochrane 2017 — role: chronic neuropathic pain review; funder: not company-funded; authors' Pfizer ties declared; flagged by a 2019 audit.
11. Moore et al., Cochrane 2014 — role: neuropathic pain and fibromyalgia review; funder: none from any source, per authors.
12. Enke et al., CMAJ 2018 — role: anticonvulsants for back pain; funder: not stated in the record.
13. Hong et al., Mol Psychiatry 2022 — role: gabapentinoids in anxiety, insomnia, bipolar disorder; funder: UK Department of Health grants.
14. Gomes et al., PLoS Med 2017 (full text) — role: gabapentin, opioids and death; funder: Ontario Ministry of Health and Long-Term Care.
15. NICE CG173, neuropathic pain in adults — role: first-line options; funder: UK government.
16. NICE NG59, low back pain and sciatica — role: gabapentinoids not offered; funder: UK government.
17. NICE NG217, epilepsies — role: seizure treatment choices; funder: UK government.
18. NHSBSA Prescription Cost Analysis 2025/26 and its summary tables — role: England dispensing; funder: UK government.
19. DailyMed Neurontin listing — role: current labeller; funder: US National Library of Medicine.
20. SEC EDGAR, Pfizer Inc. and Viatris Inc. — role: company business addresses; funder: US government.

Educational research reviewed 6 October 2026. This article is not medical advice; talk to a qualified clinician or pharmacist before starting, stopping or combining treatments.