Migraine: Independent Evidence on Treatment, CGRP Drugs & Funding

by
Key takeaways
  • Migraine is diagnosed from the pattern of attacks, not from a scan. The International Classification of Headache Disorders (ICHD-3) asks for at least five attacks lasting 4 to 72 hours, with at least two of one-sided pain, throbbing, moderate or severe intensity and worsening on activity, plus nausea or light and sound sensitivity (ICHD-3, 1.1). NICE advises against scans solely for reassurance (NICE CG150, 1.3.3).
  • For attacks, the cleanest comparative evidence favours triptans and ibuprofen. A BMJ network meta-analysis of 137 trials (89,445 people) found eletriptan, rizatriptan, sumatriptan and zolmitriptan more effective than the newer lasmiditan, rimegepant and ubrogepant (Karlsson et al., BMJ 2024). In a Cochrane review, 26% of people were pain-free at 2 hours on ibuprofen 400 mg versus 12% on placebo (Rabbie et al. 2013).
  • For prevention, guidelines disagree. The 2026 American Academy of Neurology (AAN) and American Headache Society (AHS) guideline gives high confidence to the injectable CGRP antibodies galcanezumab and erenumab in episodic migraine (Pringsheim 2026). The American College of Physicians (ACP) found no clinically important advantage for the newer, costlier drugs and put cost first (ACP 2025). NICE offers propranolol, topiramate or amitriptyline first and CGRP drugs only after three preventives have failed (NICE CG150, 1.3.18 and 1.3.24).
  • Painkillers can make headaches worse. NICE flags medication-overuse headache when triptans, opioids or combination painkillers are used on 10 or more days a month, or simple painkillers and NSAIDs on 15 or more days a month, for 3 months or more (NICE CG150, 1.2.7).
  • Supplements are a mixed picture. NICE says riboflavin 400 mg a day may help some people. A 2019 systematic review with no funding found the evidence for magnesium and CoQ10 insufficient (Okoli et al. 2019), while a 2025 review reported benefit (Talandashti et al. 2025). Butterbur was once rated top-grade, then the AAN retired that guideline in 2015 over liver safety (AAN 2012, retired).
  • Money trail: 73% of the trials in a 2023 network meta-analysis were industry-funded (Lampl et al. 2023). The International Headache Society lists AbbVie, Lundbeck, Pfizer, Ipsen, Organon, Teva, Salvia BioElectronics and Theranica as financial supporters of its corporate roundtable (IHS), and the American Headache Society lists AbbVie, Pfizer, Lundbeck, Teva and others (AHS). Both societies say they keep control of content.
  • Evidence grade: Strong for triptans, NSAIDs and several preventives against placebo. Moderate for acupuncture and exercise. Weak for riboflavin and feverfew. Insufficient for magnesium and CoQ10. Risk for butterbur, topiramate in pregnancy and combined hormonal contraception with aura.

Independent evidence review · Condition

Migraine is a common, disabling brain condition with several treatments that beat placebo. Triptans and ibuprofen are the best-supported way to treat an attack, and propranolol, topiramate, amitriptyline and the CGRP-blocking drugs can cut attack frequency. The newer CGRP drugs work, but head-to-head evidence is thin, the trials were mostly industry-funded, and independent guideline bodies reach different conclusions about when to use them. Supplements are weakly supported, and frequent use of acute painkillers can itself cause headache. Our confidence is high that the main drug classes work against placebo, moderate on how they rank against each other, and low for most supplements.

Best evidence for triptans plus an NSAID for attacks; propranolol, topiramate, amitriptyline and CGRP drugs for prevention; acupuncture and exercise as extras
Main risks medication-overuse headache; topiramate and valproate in pregnancy; combined pill with aura; triptans with heart or blood-vessel disease; butterbur and liver injury
Key rule see a clinician for any red-flag headache, and count the days a month you use acute medicine
Safety first
Call 999 (or your local emergency number) for a sudden, extremely painful headache, trouble speaking, vision loss or double vision, confusion or drowsiness, a seizure, weakness on one side of the body or face, a high fever with meningitis symptoms, or a recent head injury (NHS). NICE lists further reasons to look for another cause: a headache that peaks within 5 minutes, new brain-function problems, a change in personality, headache brought on by coughing, straining or exercise, and headache that changes with posture (NICE CG150, 1.1.1). This article is general information, not advice for your own attacks.

Table of contents

Evidence summary

Claim Evidence Source Funding / conflict Strength
Ibuprofen 400 mg treats a migraine attack 9 trials, 4,373 people. Pain-free at 2 hours: 26% vs 12% on placebo (NNT 7.2). Headache relief at 2 hours: 57% vs 25% (NNT 3.2). Side effects similar to placebo. Rabbie et al., Cochrane 2013 Oxford Pain Research Funds; NHS Cochrane programme grant; NIHR (protocol-stage statement). Two authors have consulted for drug companies. Strong
Oral sumatriptan treats an attack 61 trials, 37,250 people. Pain-free at 2 hours NNT 6.1 (50 mg) and 4.7 (100 mg). Treating while pain is still mild worked better. Derry et al., Cochrane 2012 Oxford Pain Research funds; NIHR; NHS Cochrane programme grant. Authors declared consulting for sumatriptan's maker GlaxoSmithKline; the company had no involvement. Strong
Which acute drug is best 137 trials, 89,445 people. Pain-free at 2 hours vs placebo: odds ratio 1.73 (naratriptan) to 5.19 (eletriptan). Eletriptan, rizatriptan, sumatriptan and zolmitriptan beat lasmiditan, rimegepant and ubrogepant. Karlsson et al., BMJ 2024 NIHR Oxford Biomedical Research Centre and a Lundbeck Foundation professor grant; funders had no role. Several authors declare pharma consulting and speaking fees. Strong (triptans)
Add a triptan to an NSAID ACP strongly recommends adding a triptan for adults who do not respond adequately to an NSAID (moderate-certainty evidence). NICE offers a triptan plus an NSAID or paracetamol. ACP 2025; NICE 1.3.10 The accompanying ACP systematic review lists its primary funding source as the American College of Physicians itself; disclosures and conflict management published by ACP. NICE funded by the UK Department of Health and Social Care. Strong
Rimegepant (a gepant) treats an attack 1,186 people, one attack. Pain-free at 2 hours 19.6% vs 12.0% on placebo (difference 7.6 points, 95% CI 3.3 to 11.9). Lipton et al., NEJM 2019 Funded by Biohaven Pharmaceuticals, the drug's developer, later bought by Pfizer. Moderate (maker-funded)
Topiramate prevents attacks 9 trials, 1,737 people: 1.2 fewer attacks per 28 days (95% CI 0.8 to 1.6). Responder rate about doubled (RR 2.02), NNT 4. 100 mg a day worked as well as 200 mg. Linde et al., Cochrane 2013 International Headache Society and Lifting The Burden paid editorial costs. Authors declare pharma travel grants, trial roles and advisory fees. Strong
CGRP antibody erenumab prevents attacks 955 people with episodic migraine. Migraine days fell by 3.7 a month on 140 mg vs 1.8 on placebo; 50.0% vs 26.6% had at least halved their migraine days. Goadsby et al., NEJM 2017 (STRIVE) Funded by Amgen and Novartis, the makers. Moderate (maker-funded)
How preventives compare 74 trials, 32,990 people. High-certainty evidence for CGRP antibodies, gepants and topiramate; moderate for beta-blockers, valproate and amitriptyline; gabapentin may be no better than placebo. Lampl et al. 2023 No funding. 73% of included trials were industry-funded. Several authors, including the European Headache Federation president, declare ties to AbbVie, Lilly, Lundbeck, Novartis, Pfizer and Teva. Strong (vs placebo)
New drugs add little clinically over old ones ACP found no clinically important advantages for the newer, costlier medicines in episodic migraine and ranked treatments using cost and patient values. All three recommendations are conditional and low-certainty. ACP 2025 Funded by the American College of Physicians (stated in the linked systematic review); disclosure forms published; no members recused. Moderate
Acupuncture prevents attacks 22 trials, 4,985 people. Versus sham: small benefit (SMD -0.18, 95% CI -0.28 to -0.08); at least halved frequency in 50% vs 41% (NNT 11). Versus no treatment: larger, but not blinded. Linde et al., Cochrane 2016 US NIH (NCCIH and cancer-centre grants). Several authors use acupuncture clinically or have been paid by acupuncture societies. Moderate (small vs sham)
Riboflavin (vitamin B2) 400 mg prevents attacks NICE says it may help some people. A meta-analysis of 9 studies (673 people) reported fewer migraine days, with very high inconsistency between studies. NICE 1.3.23; Chen et al. 2022 NICE: UK government. Chen: no funding statement found in the record we opened. Weak
Magnesium and CoQ10 prevent attacks Okoli 2019 (18 trials): CoQ10 and magnesium did not significantly cut frequency or severity; evidence insufficient. Talandashti 2025 (22 trials): magnesium cut monthly migraine days by 1.66 and CoQ10 cut frequency by 1.73. Okoli 2019; Talandashti 2025 Okoli: no funding. Talandashti: authors declare no conflicts; funding statement not retrieved. Insufficient / conflicting
Feverfew prevents attacks 6 trials, 561 people. The largest, best trial found 0.6 fewer attacks a month than placebo; the reviewers rate this low-quality evidence. Wider et al., Cochrane 2015 International Headache Society paid editorial costs; authors declared no interests. Weak
Butterbur is a safe preventive The AAN retired its 2012 guideline in 2015 over serious safety concerns about butterbur. The UK regulator warned of liver toxicity. AAN (retired); Medsafe 2012 Regulator and professional-body statements; the largest case review disclosed funding from Petadolex's maker for some safety studies. Risk
Withdrawing overused painkillers, with a preventive, treats medication-overuse headache 120 people. Cured of medication-overuse headache: 96.8% (withdrawal plus preventive), 74.3% (preventive only), 88.9% (withdrawal, preventive optional later). Reduction in headache days did not differ significantly (P = .20). Carlsen et al., JAMA Neurology 2020 Grants from TrygFonden and the Danish Medical Society. Authors declare lectures, travel grants and trials for several drug companies. Open-label. Moderate
Combined pill plus migraine with aura raises stroke risk US claims-data study: odds ratio 6.1 (95% CI 3.1 to 12.1) for ischaemic stroke versus neither factor. Absolute risk in young women is low (11 strokes per 100,000 females). Champaloux et al. 2017 US CDC intramural funding; no conflicts reported. Observational. Risk

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator or government body; 2 = academic, Cochrane or professional-body work with no declared commercial funding for the work (author ties may still be declared); 3 = work paid for or hosted by a society that takes industry sponsorship, or by authors with extensive declared industry ties, or an industry-sponsored body itself; 4 = manufacturer-funded. Credibility runs from A (highest) to D. Our sources cluster in the UK, US and Europe: UK, US, Danish-led, Austrian-led, German-led and Norwegian-led reviews, so the verdict reflects Western guideline culture and Western-funded trials. Many authors are headache specialists who also advise or lecture for the companies whose drugs they assess, which is hard to avoid in a field where most trials are industry-run.

Source Who funds it Country Independence tier (1–4) Credibility (A–D) Why it would tell the truth / residual bias
NICE CG150 and linked technology appraisals UK Department of Health and Social Care UK 1 A Public accountability and a cost remit make it cautious about expensive drugs. It relies on company submissions for newer drugs, and net NHS prices are confidential under discount schemes.
NHS migraine pages UK government (NHS England) UK 1 A Written for patients with no product to sell and reviewed on a fixed cycle. Brief, so it omits effect sizes.
MHRA topiramate pregnancy-prevention announcement UK government (regulator) UK 1 A Legal duty to protect patients and an incentive to avoid a public safety failure. Regulators are sometimes criticised for acting slowly.
Cochrane acute-treatment reviews (ibuprofen 2013, sumatriptan 2012) Oxford Pain Research Funds; NIHR; NHS Cochrane programme grant UK 2 B+ Pre-specified methods and reputation for neutrality. Authors declare pharma consulting; the data are old (searches to 2011 and 2013).
Karlsson et al., BMJ 2024 network meta-analysis NIHR Oxford Biomedical Research Centre; Lundbeck Foundation professor grant Denmark-led, with UK (Oxford) and other European authors 2 A− Registered protocol, patients on the design panel, and the BMJ's disclosure rules. The Lundbeck Foundation owns about 69% of migraine-drug maker Lundbeck; the funders had no role, and several authors declare pharma fees.
American College of Physicians guidelines (acute and prevention), 2025 American College of Physicians (professional body; it funds its own reviews) US 2 B+ Includes cost evidence and publishes conflict management. Conditional, low-certainty recommendations; physicians' body that may favour primary-care drugs.
AAN and AHS prevention guideline and systematic review, 2026 American Academy of Neurology; co-developed with the American Headache Society, which lists drug-company sponsors US (Canadian lead author) 3 B Large evidence base (217 studies) and a formal process. We could not open the panel's conflict disclosures, and both societies take pharma money.
Linde et al., Cochrane topiramate review 2013 International Headache Society and Lifting The Burden (editorial costs) Norway-led (Trondheim), multinational authors 3 B Standard Cochrane method; the funder is a society with pharma sponsors, and authors declare pharma travel grants, advisory fees and trial roles. Searches ended in 2013.
Lampl et al., J Headache Pain 2023 No funding Austria-led, multinational European authors with a Canadian methods team 3 B GRADE ratings and 74 trials. The European Headache Federation president is a co-author with declared AbbVie, Lilly, Lundbeck, Novartis, Pfizer and Teva ties, and 73% of the underlying trials were industry-funded.
Okoli et al., Can J Neurol Sci 2019 No funding Canada 2 B Unfunded, so no sponsor to please. Only two supplements could be pooled, with few small trials; one author declared past Allergan research funding.
Linde et al., Cochrane acupuncture review 2016 US NIH (NCCIH and cancer-centre grants) Germany-led, with US, UK and Italian authors 2 B Public funding and a sham-controlled analysis. Reviewers who practise or teach acupuncture have a professional interest; blinding against no treatment was impossible.
Manufacturer-funded pivotal trials: STRIVE (erenumab, 2017) and rimegepant (2019) Amgen and Novartis; Biohaven Pharmaceuticals US, Switzerland 4 C Registered, randomised and regulator-reviewed, and a failed trial hurts the sponsor's reputation. Sponsors choose the design, comparator and outcomes, and the makers profit from a positive result.
IHS and AHS sponsorship pages Member and industry support, as the pages themselves state UK (IHS); US (AHS) 3 B The societies publish their own sponsors, which is against easy self-interest, and say they keep control of content. We cannot check that claim, nor the money amounts.

What migraine is and how it is diagnosed

Migraine is a long-term neurological condition with recurring attacks. The NHS says migraine affects about 10 million people in the UK (NHS). The BMJ review cited above puts the global figure at more than one billion and describes migraine as the leading cause of disability in girls and women aged 15 to 49 (Karlsson 2024).

The ICHD-3 criteria

The ICHD-3 is the classification used worldwide and is published for the International Headache Society. It sets these main criteria:

  • Migraine without aura: at least five attacks lasting 4 to 72 hours untreated; at least two of one-sided pain, pulsating quality, moderate or severe intensity and worsening with routine activity; and nausea or vomiting, or both light and sound sensitivity. In under-18s, attacks may last 2 to 72 hours (ICHD-3, 1.1).
  • Migraine with aura: at least two attacks with fully reversible aura symptoms (visual, sensory, speech or language, motor, brainstem or retinal) and at least three of six features, such as gradual spread over 5 minutes or more and each symptom lasting 5 to 60 minutes (ICHD-3, 1.2).
  • Chronic migraine: headache on 15 or more days a month for more than 3 months, with migraine features on at least 8 of those days (ICHD-3, 1.3).

NICE uses a similar table and also defines episodic migraine as fewer than 15 headache days a month. It suggests a headache diary for at least 8 weeks, including links to menstruation (NICE CG150, 1.1.3 to 1.2.1).

Symptoms and red flags

A typical attack brings a throbbing headache, often on one side, with nausea, vomiting, and sensitivity to light and sound. Attacks usually last 4 hours to 3 days, and early symptoms such as tiredness can start up to 2 days before the head pain. Aura symptoms such as zigzag lines, flashing lights, pins and needles, or difficulty speaking should not last longer than 1 hour (NHS).

Red flags that need prompt assessment
NICE says to consider investigation or referral for headache with fever, a thunderclap onset (maximum intensity within 5 minutes), new neurological deficit or confusion, a personality change, head injury in the past 3 months, triggering by cough, exercise or posture, signs of giant cell arteritis or acute glaucoma, or a substantial change in the usual pattern. Weakness, double vision, one-eyed visual symptoms, poor balance or reduced consciousness during an apparent aura are also reasons for further assessment (NICE CG150, 1.1.1, 1.1.2 and 1.2.4). The NHS advises an urgent GP appointment or NHS 111 if an attack lasts over 72 hours, aura lasts over an hour, or you are pregnant or have recently given birth (NHS).

Causes and risk factors

The NHS says it is not known what causes migraine, but attacks involve changes to nerves and blood vessels in the brain. Close family history and being female make it more likely, and common triggers include periods, anxiety, stress, tiredness, skipped meals, too much caffeine and too little exercise (NHS causes). The newest drugs target calcitonin gene-related peptide (CGRP), a signalling molecule that the antibody makers describe as playing an important role in migraine (Novartis and Amgen 2018). That the drugs work shows CGRP matters in humans; it does not by itself explain the whole condition. Trigger lists are mostly self-reported, and we found no high-quality independent trial that tests individual triggers.

Stress, low mood and anxiety often travel with migraine. For that overlap see our guides on stress and physical health and mindfulness-based therapy.

Treatments that work, and how well

Option Verdict Evidence Key caveat
Triptan plus NSAID or paracetamol for attacks WORKS Strong ACP recommendation after an NSAID alone fails (ACP 2025); first offer in NICE (1.3.10). Triptans are unsuitable with some heart and blood-vessel conditions (NHS).
Ibuprofen, aspirin 900 mg, paracetamol alone WORKS (MODEST) Ibuprofen NNT 7.2 for pain-free at 2 hours (Cochrane). NICE lists each as a single-drug option (1.3.11). Under-16s should not be given aspirin because of Reye's syndrome (NICE 1.3.11).
Gepants for attacks (rimegepant, atogepant) MIXED Beat placebo, but less effective than the best triptans in the BMJ analysis. NICE allows them only after at least 2 triptans and NSAIDs or paracetamol have failed (TA919). Useful when triptans are unsuitable. Trials are maker-funded.
Propranolol, topiramate, amitriptyline for prevention WORKS NICE first-line options (1.3.18). Topiramate: NNT 4 (Cochrane). Amitriptyline has moderate-certainty benefit but more drop-outs (Lampl 2023). Propranolol overdose risk; topiramate pregnancy rules; amitriptyline withdrawal. See our amitriptyline review.
CGRP antibodies and gepants for prevention WORKS High-certainty benefit versus placebo (AAN/AHS 2026 review; Lampl 2023). NICE: only after 3 preventives fail, with a 12-week stop rule. ACP: use after older options. Few head-to-head trials.
Botulinum toxin A (chronic migraine only) WORKS (CHRONIC) High-confidence benefit in the AAN/AHS review for chronic migraine; NICE offers it after 3 preventives fail (1.3.26). Not for episodic migraine. Injected repeatedly in clinic.
Acupuncture MIXED Small benefit over sham, larger over no treatment (Cochrane 2016). NICE: up to 10 sessions if the three drugs fail. The gap between sham and real is small.
Regular exercise MIXED 21 trials, 1,195 people: moderate-intensity aerobic exercise cut monthly migraine frequency by 2.2 (95% CI 1.1 to 3.3) versus control (Woldeamanuel 2022). Small trials, no blinding; funded by a US NIH neurology grant, no conflicts declared.
Gabapentin, ergots, opioids AVOID NICE advises against gabapentin for prevention (1.3.20) and against ergots and opioids for attacks (1.3.16). Opioids raise the risk of medication-overuse headache.

Medicines in detail

Treating an attack

The first-line recipe in UK guidance is a triptan plus an NSAID or paracetamol, taken early in the attack, with an anti-sickness medicine considered even if there is no nausea (NICE CG150, 1.3.10 to 1.3.13). The Cochrane sumatriptan review found that taking it while pain was still mild gave better results than waiting (Derry 2012). Doses studied: ibuprofen 400 mg and sumatriptan 50 or 100 mg in the Cochrane reviews; NICE names aspirin 900 mg and, for menstrual migraine, frovatriptan 2.5 mg twice a day or zolmitriptan 2.5 mg two or three times a day on the expected days. These are trial and guideline doses, not instructions.

The BMJ network meta-analysis found eletriptan, rizatriptan, sumatriptan and zolmitriptan the most effective oral drugs for 2-hour pain freedom, with eletriptan and ibuprofen best for sustained freedom to 24 hours. Its authors said the best triptans should be the preferred acute treatment and argued for adding them to the WHO Essential Medicines List, while saying that patients with high cardiovascular risk need care (Karlsson 2024).

Preventing attacks

The AAN/AHS update says preventives should be offered to people with 4 or more migraine days a month, 4 or more moderate-to-severe headache days a month, or migraine that disrupts work or daily tasks (AAN press release). Its systematic review of 217 studies found high confidence for galcanezumab and erenumab in episodic migraine; moderate confidence for atogepant, eptinezumab, fremanezumab, propranolol, topiramate and valproate; and low confidence for amitriptyline and several others. In chronic migraine it found high confidence for fremanezumab, galcanezumab and onabotulinumtoxinA (Pringsheim 2026).

The three bodies diverge on order, not on whether the drugs work:

  • NICE offers propranolol, topiramate or amitriptyline first, then acupuncture, and CGRP-targeting drugs for adults with at least 4 migraine days a month once at least 3 preventives have failed, with a stop rule if migraine frequency has not fallen by 50% (episodic) or 30% (chronic) after 12 weeks. Review any preventive 3 to 6 months after starting (NICE 1.3.18 to 1.3.28).
  • ACP suggests metoprolol or propranolol, valproate, venlafaxine or amitriptyline first, a CGRP drug second, and topiramate third. It said the newer drugs showed no clinically important advantage and used cost as a primary factor (ACP 2025).
  • AAN/AHS grades by confidence in the evidence rather than ranking drugs, and lists CGRP drugs among the high- and moderate-confidence options (Guideline Central summary).

One trial of 955 people shows the scale of benefit. On erenumab 140 mg, migraine days fell by 3.7 a month from a baseline of 8.3, against 1.8 on placebo. In plain terms, the drug removed about two extra migraine days a month beyond placebo, and half of users halved their days, compared with about a quarter on placebo (STRIVE, maker-funded). Propranolol's non-migraine use and overdose risk are covered in our propranolol review.

Supplements, lifestyle and other claims graded

Claim Verdict Evidence Key caveat
Riboflavin 400 mg a day MIXED Named in NICE as possibly effective for some people (1.3.23). The NHS also lists vitamin B2 among prevention options (NHS). The 2022 pooled analysis showed very high inconsistency (inconsistency statistic of 89% for migraine days).
Magnesium INSUFFICIENT Not significant in the unfunded 2019 review; positive in 2025. A 2026 network meta-analysis of 14 trials found no significant effect on frequency (mean difference -0.82, 95% CI -2.03 to 0.13) but a drop in severity (Peng 2026). Conflicting reviews. See our magnesium guide.
CoQ10 INSUFFICIENT Two pooled trials (97 people) showed no significant effect in 2019; a 2025 review reported fewer attacks (-1.73). Few small trials. See our CoQ10 guide.
Feverfew WEAK 0.6 fewer attacks a month in the best trial; Cochrane calls it low-quality evidence (Wider 2015). Trials reported only mild effects, such as stomach upset and mouth ulcers.
Butterbur AVOID Medsafe relayed the MHRA's alert on liver toxicity, including cases with extracts that had the toxic alkaloids removed (Medsafe). The AAN retired its guideline in 2015 over this concern.
Sleep, regular meals, hydration, caffeine and alcohol limits, stress management COMMON ADVICE The NHS recommends them as self-help (NHS). We found no trial that tested the bundle. Keep a diary to find your own triggers.

For sleep and relaxation tools often used alongside migraine care, see our sleep guide and CBT review.

Medication-overuse headache

Using acute medicine too often can cause its own daily or near-daily headache. NICE says to suspect it when headache has developed or worsened during 3 months or more of triptans, opioids, ergots or combination painkillers on 10 or more days a month, or paracetamol, aspirin or an NSAID on 15 or more days a month (NICE CG150, 1.2.7). The NHS advises not taking painkillers on more than 2 days a week (NHS).

NICE advises stopping all overused acute medicines for at least a month, stopping abruptly rather than gradually, warning people that headaches will probably worsen before improving, adding a preventive for the underlying disorder, and reviewing after 4 to 8 weeks. It does not recommend routine inpatient withdrawal (1.3.42 to 1.3.48). The Danish randomised trial supports starting a preventive with withdrawal. At 6 months, 96.8% were cured with withdrawal plus a preventive, versus 74.3% with a preventive alone, although the trial was open-label and 18 of 120 people dropped out (Carlsen 2020). We did not find an independent source for how common medication-overuse headache is, so we give no prevalence figure.

Migraine and women's health

  • Menstrual migraine: NICE suspects it when attacks occur mainly from 2 days before to 3 days after the start of a period in at least 2 of 3 cycles, confirmed with a diary over 2 cycles. For predictable attacks that do not respond to standard treatment it names short-course frovatriptan or zolmitriptan (1.2.5, 1.2.6, 1.3.31). The NHS says menstrual attacks are often longer and more severe (NHS). Related: PMDD.
  • Contraception: NICE says not to routinely offer combined hormonal contraceptives to women and girls who have migraine with aura (1.3.30). In a US claims study, migraine with aura plus a combined hormonal contraceptive carried an odds ratio of 6.1 for ischaemic stroke, versus 2.7 for aura alone. The overall rate was 11 strokes per 100,000 females aged 15 to 49 (Champaloux 2017). See also women and cardiovascular disease and stroke.
  • Topiramate: in the UK it must not be used for migraine in women and girls who could become pregnant unless they meet a Pregnancy Prevention Programme (effective contraception, a negative pregnancy test and a signed risk form). The MHRA noted that some birth-control methods are less effective with it (MHRA 2024). The AAN/AHS guideline says clinicians must warn that topiramate can reduce hormonal contraceptive effectiveness at doses above 200 mg a day (summary).
  • Pregnancy: NICE offers paracetamol for attacks, with a triptan or NSAID after discussion; seek specialist advice if a preventive is needed (1.3.32 and 1.3.33). The AAN/AHS summary advises avoiding teratogens such as divalproex and topiramate and lists nifedipine first, then metoprolol or propranolol. The NHS says migraine sometimes improves in pregnancy and often eases after the menopause (NHS). See sleep and pregnancy and perimenopause.

Risks and side effects

Serious or notable risks
Triptans: common effects include flushing, tiredness, dizziness and nausea; rare serious ones include heart attack and allergic reaction. The NHS lists heart, liver or kidney conditions, past stroke or TIA, high blood pressure and peripheral vascular disease among reasons a triptan may not be suitable (NHS). Propranolol: NICE flags a higher risk of self-harm with propranolol in people who have both depression and migraine and urges caution because of overdose toxicity (1.3.18). Topiramate: harm to the unborn child (see above). Erenumab: constipation is common, and serious constipation, sometimes needing surgery, has been reported after marketing; the UK label advises avoiding it in pregnancy where possible (UK label). Butterbur: liver toxicity. If you have thoughts of suicide or self-harm, call your local emergency number; in the US call or text 988; in the UK call Samaritans on 116 123.

Topiramate side effects in trials were usually mild: taste disturbance and weight loss at 50 mg, with more at 100 mg and 200 mg (NNH 3 to 25 at 100 mg) (Linde 2013). In the 2023 network analysis, amitriptyline and valproate led to substantial drop-outs, whereas CGRP antibodies and gepants did not increase adverse events (Lampl 2023). Long-term safety of CGRP drugs rests mostly on maker-funded trials and follow-up studies, and those trials excluded some people with major cardiovascular disease (UK label). Interactions with other medicines are in the product labels; ask a pharmacist before combining acute and preventive drugs.

Who to see

  • GP or pharmacist: first stop for diagnosis, a diary, acute-treatment advice and first-line preventives. The NHS says to see a GP if attacks are severe, worsening, more than once a week or hard to control (NHS).
  • Urgent care (NHS 111 or 999): for red flags above.
  • Neurologist or headache clinic: NICE says to consider specialist referral for withdrawal in people using strong opioids, those with relevant comorbidities, or after repeated failed attempts, and CGRP antibodies and botulinum toxin follow NICE criteria (1.3.24 to 1.3.26, 1.3.47).
  • Specialist advice before pregnancy: for any preventive (NICE 1.3.33).

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

Who makes and sells the main migraine drugs

  • Erenumab (Aimovig): developed under a collaboration between Amgen and Novartis. Amgen kept rights in Japan, and Novartis holds rights in Europe, Canada and the rest of the world; in April 2017 the pair expanded the deal to co-commercialise in the US (Novartis 2018). The UK licence holder is Novartis Pharmaceuticals UK (UK label).
  • Fremanezumab (Ajovy): Teva. At its US launch in 2018 the list price was $575 per monthly dose (Teva 2018).
  • Galcanezumab (Emgality): Eli Lilly (US) (Lilly 2018).
  • Eptinezumab (Vyepti): Lundbeck (Denmark), which bought its developer, Alder BioPharmaceuticals, for up to about $1.95 billion in October 2019 (Lundbeck 2019). The Lundbeck Foundation owns about 69% of the company's share capital and about 77% of the votes (Lundbeck).
  • Rimegepant (Nurtec ODT, sold as Vydura in the EU): Biohaven, which Pfizer agreed to buy in May 2022 for about $11.6 billion (Pfizer announcement) and completed on 3 October 2022 (PharmTech).
  • Atogepant (Qulipta), ubrogepant (Ubrelvy) and Botox: AbbVie, which describes them as its three prescription migraine treatments (AbbVie 2023) and completed its acquisition of Allergan on 8 May 2020 (AbbVie 2020).
  • Triptans and NSAIDs: off-patent, generic and cheap. The Cochrane sumatriptan review names GlaxoSmithKline as a manufacturer of sumatriptan and says the company was not involved in the review (Cochrane record).

NICE's appraisals for erenumab, fremanezumab, galcanezumab and eptinezumab each record a confidential "simple discount patient access scheme", so the real NHS price is not public (TA682; TA764; TA659; TA871).

Who funded the trials and the guidelines

  • Pivotal trials: STRIVE was funded by Amgen and Novartis; the first rimegepant attack trial in the NEJM by Biohaven. Across 74 preventive trials, 73% were industry-funded, 5% government-funded, 3% institution-funded and 17% did not report funding (Lampl 2023).
  • Older-drug reviews: The Cochrane acute-drug reviews were supported by UK public funds. The Cochrane topiramate and feverfew reviews had editorial costs paid by the International Headache Society, and for topiramate also Lifting The Burden (Cochrane record; feverfew record).
  • Guidelines: the AAN funded its own 2026 guideline (AAN). The ACP guidelines state disclosures are published and no members recused.

Societies and sponsors

  • The International Headache Society, which publishes the ICHD-3, lists AbbVie, Lundbeck, Pfizer, Ipsen, Organon, Teva, Salvia BioElectronics and Theranica as members of its Corporate Advisory Roundtable. It says their financial support enables participation and that it keeps full independence over content (IHS).
  • The American Headache Society, co-author of the 2026 guideline, names AbbVie as its Diamond corporate roundtable sponsor, alongside companies including Axsome, Ipsen, Lundbeck, Pfizer, Theranica, Kallyope, Amneal and Teva, and lists programme support from AbbVie, Amneal, Axsome, Lilly, Pfizer and Lundbeck (AHS sponsorship). Earlier AHS symposia thanked AbbVie, Lilly, Lundbeck, Pfizer and Amgen for educational grants (2023 symposium; 2024 Academy).
  • Author-level ties are heavy. The senior author of the BMJ acute-treatment review is a past president of the International Headache Society and declares consulting and speaking for AbbVie, Amgen, Lilly, Lundbeck, Novartis, Pfizer and Teva, and his work is supported by a Lundbeck Foundation grant (Karlsson 2024 disclosures). The European Headache Federation's president co-authored the 2023 network analysis and declares consulting and lecture fees from AbbVie, Lilly, Lundbeck, Novartis, Pfizer and Teva (Lampl 2023), and a past president of the Federation co-authored the Danish medication-overuse trial and declares trials and lectures for several makers (Carlsen 2020). These are disclosed, normal in the field, and not evidence of wrongdoing.
Follow the money

Research funding at a glance

Funding & backersSource & studyClaim & limits

13 sources graded. The counts below are the independence tiers assigned in the scorecard above. They count sources, not funding amounts or evidence quality.

Tier 13Independent
Tier 25Indirect ties
Tier 34Interested party
Tier 41Self-interested

Public funding alone does not prove independence, and commercial funding alone does not prove a claim false. Named funders, countries and relationships are in the scorecard and source notes.

Documented integrity events

  • The AAN retired its 2012 NSAID and complementary-treatments guideline on 16 September 2015 because of serious safety concerns about butterbur, which the guideline had recommended at the top grade (AAN retirement notice). Safety studies of the Petadolex extract were funded by its maker, Weber & Weber, and one author of that 2019 case review disclosed paid advisory work for the maker (Anderson and Borlak 2019).
  • The MHRA tightened topiramate rules in June 2024 after a review of harm in pregnancy, and said the drug was thought to be prescribed to just over 30,000 female patients under 55 in England in one month (MHRA).

What we could not trace

We could not open the full conflict-of-interest disclosures of the 2026 AAN/AHS panel, the ACP funding statement, or the funding statements of the 2022 riboflavin and 2025 supplement meta-analyses. We did not find independent head-to-head trials of CGRP antibodies against propranolol or amitriptyline, and we found no study of how much money flows from makers to UK headache clinicians. Society sponsorship amounts are not published on the pages we read.

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

Frequently asked questions

Is a brain scan needed for migraine?

Usually not. NICE says not to refer people with a diagnosed migraine for neuroimaging solely for reassurance, and to consider further tests when red-flag features appear (NICE CG150, 1.3.2 and 1.3.3).

Can a triptan and ibuprofen be taken together?

UK guidance offers a triptan with an NSAID or with paracetamol as the first approach, taking account of preference, other illnesses and side-effect risk (NICE 1.3.10). The American College of Physicians strongly recommends adding a triptan when an NSAID alone has not worked (ACP 2025). A pharmacist can check suitability for you.

How many days a month can painkillers be used before they cause headache?

NICE's alert thresholds are 10 or more days a month for triptans, opioids, ergots or combination painkillers, and 15 or more days a month for paracetamol, aspirin or NSAIDs, over 3 months or more (1.2.7). The NHS gives a simpler rule of no more than 2 days a week (NHS).

Are the new CGRP injections better than older tablets?

They work and are well tolerated in trials, but the evidence that they beat older drugs head to head is limited. Only one trial had compared a CGRP antibody with topiramate when the 2023 network analysis was published (Lampl 2023), and ACP saw no clinically important advantage for the newer drugs (ACP). The 2026 AAN/AHS review gives them high or moderate confidence of benefit against placebo (Pringsheim 2026). In England NICE reserves most of them for people who have tried at least three preventives.

How long should a preventive be tried?

NICE says to review the need for continuing a preventive 3 to 6 months after starting, and to stop a CGRP drug at 12 weeks if migraine frequency has not fallen by at least 50% in episodic migraine or 30% in chronic migraine (1.3.27 and 1.3.28).

Does magnesium or riboflavin prevent migraine?

Evidence is mixed. NICE says riboflavin 400 mg a day may help some people (1.3.23), but pooled studies are inconsistent (Chen 2022). For magnesium, an unfunded 2019 review found evidence insufficient (Okoli 2019) and a 2025 review reported benefit (Talandashti 2025).

Is butterbur safe?

Regulators warned about liver toxicity, including with extracts from which the toxic alkaloids had been removed (Medsafe), and the AAN retired the guideline that had recommended it (AAN). We did not find a regulator source that clears it.

Can women with migraine with aura use the combined pill?

NICE says not to routinely offer it (1.3.30). A US study found about a six-fold stroke odds with the combination, though absolute risk in young women is low (Champaloux 2017). Speak to a clinician about non-oestrogen options.

Is topiramate safe if pregnancy is possible?

In the UK it must not be used for migraine in women and girls who could become pregnant unless a Pregnancy Prevention Programme is followed (MHRA). Do not stop a medicine without advice; see a prescriber about alternatives.

Sources and funding notes

What we could not verify: the full conflict-of-interest disclosures of the 2026 AAN/AHS guideline panel and of the ACP guidelines (the publisher pages blocked automated access); the funding statements of ACP, Talandashti 2025 and Chen 2022; the exact tiers for AHS corporate sponsors beyond the Diamond level; the current UK Medicines and Healthcare products Regulatory Agency (MHRA) text on butterbur, for which we relied on Medsafe's relay of the 2012 UK alert; and any sponsorship amounts. The Cochrane ibuprofen and sumatriptan funding lines come from the protocol-stage record on the Cochrane Library. Disclosure: we accept no advertising or affiliate income for this article. We omitted our usual "no commercial relationship" sentence because this article discusses CoQ10 and magnesium.

  1. NICE CG150. Headaches in over 12s: diagnosis and management (updated June 2025; reviewed 23 September 2026) — UK guideline; Department of Health and Social Care. Linked appraisals: TA919, TA1172.
  2. NHS. Migraine and causes (reviewed 10 March 2026); sumatriptan — UK government health service.
  3. ICHD-3 migraine criteria — International Headache Society classification; the IHS is sponsored in part by drug companies (see IHS sponsors).
  4. Rabbie et al. Ibuprofen for acute migraine. Cochrane 2013 — NHS Cochrane programme, NIHR, Oxford Pain Research Funds.
  5. Derry et al. Oral sumatriptan. Cochrane 2012 — same funders.
  6. Karlsson et al. BMJ 2024 — NIHR Oxford BRC; Lundbeck Foundation.
  7. ACP acute migraine guideline 2025 and prevention guideline 2025, with the ACP press release and acute release — American College of Physicians.
  8. Potrebic et al. AAN/AHS prevention guideline 2026, Pringsheim et al. systematic review 2026, AAN press release (states AAN funded the guideline), Guideline Central summary — AAN and AHS.
  9. Linde et al. Topiramate. Cochrane 2013 — International Headache Society and Lifting The Burden paid editorial costs.
  10. Lampl et al. J Headache Pain 2023 (full text) — no funding.
  11. Goadsby et al. STRIVE, NEJM 2017 — Amgen and Novartis; Lipton et al. rimegepant, NEJM 2019 — Biohaven.
  12. Linde et al. Acupuncture. Cochrane 2016 — US NIH.
  13. Woldeamanuel and Oliveira. Exercise network meta-analysis 2022 — US NIH (NINDS) career grant.
  14. Okoli et al. 2019 (journal page) — no funding; Talandashti et al. 2025 — funding not retrieved; Peng et al. 2026 — no financial support; Chen et al. 2022 — funding not retrieved.
  15. Wider et al. Feverfew. Cochrane 2015 — International Headache Society (editorial costs).
  16. Holland et al. AAN/AHS 2012 guideline (retired); Medsafe 2012 on butterbur — NZ regulator; Anderson and Borlak 2019 — Volkswagen Foundation (Germany) and others; some safety studies funded by Petadolex's maker.
  17. Carlsen et al. JAMA Neurology 2020 — TrygFonden and Danish Medical Society grants.
  18. Champaloux et al. 2017 — US CDC.
  19. MHRA. Topiramate pregnancy prevention 2024; Aimovig UK label — UK regulator and licence holder.
  20. Company and society sources: Novartis; Teva; Lilly; Lundbeck and shareholders; Pfizer–Biohaven; AbbVie; American Headache Society — sellers and societies; their own claims.

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.

Have a question — or want us to cover something?

Ask about anything on this page, or request the next deep dive: an ingredient, a supplement, or a health concern. We use published research, evidence syntheses, and regulatory guidance, with clear source links.

We store your topic, message, optional email, and this page so we can manage and reply to the request. Do not include diagnoses, medications, or other sensitive medical information. See our Privacy Policy.