Physiotherapy · Primary headaches

Migraine

A patient comes about their neck. They say they have “had headaches for ever”, that it comes from stress, that their osteopath mentioned C1. This article deals with migraine, a primary headache of neurological origin, where the site's other “head and neck” pages deal with pain arising from the cervical spine or from a cranial nerve: cervicogenic headache, occipital neuralgia, trigeminal neuralgia and cervicogenic dizziness. The distinction is not academic: a migraine treated as neck pain means months of sessions that do nothing to the disease, and a patient left without a preventive treatment when effective ones exist.

Summary written from primary sources verified one by one: every identifier was resolved through the PubMed E-utilities API and every abstract read before being cited. Each figure carries its reference where it is written; the full bibliography, with PMIDs and DOIs, appears at the end of the article. The classification codes are those of the International Classification of Headache Disorders, 3rd edition, taken from the official text.

What is a migraine, and what is not one?

The diagnosis of migraine is clinical, positive, and rests on counting. Knowing it by heart changes a consultation: you stop asking the patient what they think they have, and you get them to describe what they experience.

Migraine belongs to the first part of the International Classification of Headache Disorders, that of primaryheadaches: diseases in themselves, and not symptoms of something else. Its heading carries the ICHD-3 code 1. Migraine, with two main forms, 1.1 Migraine without aura and 1.2 Migraine with aura, and an advanced form, 1.3 Chronic migraine (PMID 29368949).

The criteria for migraine without aura run to five lines, and it is their simplicity that makes them usable in the treatment chair.

Criteria taken word for word from the official text of the International Classification of Headache Disorders, 3rd edition (ichd-3.org); publication reference: Cephalalgia 2018 (PMID 29368949).

Three remarks from the official text are worth knowing, because they close as many false trails. Migraine in children is more often bilateral than in adults, one-sidedness appearing only in late adolescence. An occipital headache in a child is rare and calls for diagnostic caution. Finally, a migraine attack can be accompanied by cranial autonomic features, lacrimation or nasal congestion included: their presence therefore does not by itself point to cluster headache.

The diagnosis of migraine does not rest on what the patient thinks they have. It rests on the duration of an attack, two characteristics of the pain, one accompanying feature, and five episodes.

What migraine is not

The site's corpus already covers four neighbours of migraine, and each is distinguished by a single sign. This table is made to be read from right to left: you start from the sign and work back to the diagnosis.

Codes and criteria: ICHD-3, Cephalalgia 2018 (PMID 29368949). The distribution of causes of chronic dizziness is set out, with its sources, in the site's article on cervicogenic dizziness.

Who is affected, and how many migraine sufferers do not know it?

The figure that matters to a physiotherapist is not the prevalence. It is the proportion of migraine sufferers who do not know they are, because that measures exactly the place you can take.

In the Global Burden of Disease study, migraine affected 1.2 billion people in 2021, against 2.0 billion for tension-type headache. Tension-type headache is more common; migraine is markedly more disabling, and the burden is concentrated disproportionately in women aged 30 to 44 (PMID 40972580). Migraine remains the second leading cause of years lived with disability worldwide (PMID 33974014).

In France, the benchmark measurement is old but solid, because it rests on a representative sample of the general population. Across 10,585 people screened, the standardised prevalence of migraine in the strict sense (categories 1.1 and 1.2) was 7,9 %, with a considerable difference by sex: 11.2 % in women against 4.0 % in men. Including incomplete migraine pictures, the total prevalence reached 17.0 % (PMID 12136063). Two years later, with the revised criteria that include probable migraine, the FRAMIG 3 survey identified 1,179 migraine sufferers out of 10,532 people surveyed, that is 21.3 % (PMID 16643573). The whole French series, eleven studies over more than thirty years, has been the subject of a recent review (PMID 39627053).

The last figure is the one that explains the most consultations. Chronic daily headache affected 2.98 % of the French general population, and two thirds of those people had a migraine-like picture. Their impact and their use of healthcare clearly exceeded those of the episodic migraine sufferers in the same sample: 28.2 % reached grades 3 or 4 on the MIDAS scale against 12 %, and above all, only 6.6 % were receiving a preventive treatment while 88 % were taking non-specific analgesics (PMID 12620605). That is the very definition of the situation that produces medication-overuse headache.

When does a headache stop being a migraine?

This is the most important chapter in this article, and the only one whose omission can cost a life. A primary headache is only diagnosed once secondary headaches have been ruled out, and that step cannot be delegated.

The proportion of secondary headaches depends entirely on where you work. In an international cohort of 5,293 patients attending emergency departments for headache, 6.1 % had a serious cause identified (95 % confidence interval: 5.5 to 6.8). The independent predictors were a new neurological deficit, then a history of cancer, an age over 50 and a recent head injury (PMID 38658053). In private practice the proportion is far lower, but the consequence of missing one is the same.

The SNNOOP10 list

The reference tool is the SNNOOP10 list, published in Neurology in 2019, which sets out fifteen warning signals for secondary headaches (PMID 30587518). Its performance has been measured: in a prospective series of 100 patients triaged in the emergency department, of whom 46 had a high-risk headache, the sensitivity of the complete list was 100 % (95 % confidence interval: 90.2 to 100). The most sensitive items taken individually were neurological deficit or dysfunction (75.5 %), a change in pattern or the recent onset of a new headache (64.4 %) and onset after the age of 50 (64.4 %). The most specific were post-traumatic onset (94.5 %), a history of cancer (89.1 %) and systemic signs (89 %) (PMID 36003002).

List and wording: Do TP et al., Neurology 2019 (PMID 30587518). Sensitivities measured: García-Azorín D et al., Cephalalgia 2022 (PMID 36003002). What to do in the clinic is a transposition by the editorial team, not a recommendation from the authors.

Why should cerebral venous thrombosis concern you?

This disease does not belong to physiotherapy, and nothing that follows is treated in a clinic. It appears here for one reason only: its most frequent presentation is an isolated headache in a young woman, which is exactly the patient who walks through a clinic door talking about their migraines.

Cerebral venous thrombosis is occlusion of a dural venous sinus or of a cerebral vein. Its incidence is estimated at 1.32 per 100,000 a year in high-income countries, and it preferentially affects children, young adults and women, particularly during pregnancy and the post-partum period (PMID 27816347). In the benchmark international study, ISCVT, which followed 624 patients in 89 centres and 21 countries, 465 of the 624 patients, that is 75 %, were women, younger than the men and more often presenting with headache as the main feature (PMID 19478226).

The trap: headache can be the only sign

This is the point that justifies this chapter. In the multicentre VENOST study, of 1,144 patients with cerebral venous thrombosis, 287 presented with an isolated headache, against 857 with other associated signs. Onset there was more often subacute or chronic than in the group with associated signs, where acute onset dominated. A decisive point for a practitioner who will see the patient again: 29 % of the isolated headache group developed other neurological signs during follow-up (PMID 34254656).

A French series had established it earlier, and even more disturbingly. Of 123 consecutive patients, 17 had headache as their only manifestation, with a normalbrain CT and cerebrospinal fluid examination. The lateral sinus was involved in 15 of them. Onset was gradual in 11, acute in 3 and thunderclap in 3. Once established, the headache was continuous in 15 and unilateral in 13, on the side of the occluded sinus. The authors conclude that magnetic resonance imaging and venography must be requested for any recent, progressive or thunderclap headache, even when the CT and the lumbar puncture are normal (PMID 16024884).

The outlook is good once the diagnosis is made: in ISCVT, after a median 16 months of follow-up, 57.1 % of patients had no symptoms and 22 % had minor ones. But 8.3 % had died, and a subgroup of about 13 % remained at high risk of a poor outcome, identifiable on admission: coma, impaired alertness, haemorrhage on the initial CT, deep venous system thrombosis, central nervous system infection, cancer (PMID 14976332). Treatment is anticoagulation, started in hospital, and its conduct is the subject of a GRADE-based European guideline (PMID 28833980).

A young woman, on combined oral contraception or in the weeks following childbirth, describing a new and continuous headache: it is not a migraine until a doctor has said so.

Does the neck make the migraine, or does the migraine hurt the neck?

This is the question that decides everything else, and the literature answers it with a nuance: both, but not in the same patient. Knowing which side the person in front of you is on is the physiotherapist's own competence in this field.

That migraine comes with neck pain has long been established. In an old but much-cited series, across 50 migraine sufferers questioned phase by phase, 32 reported neck pain or stiffness: 10 during the premonitory phase, 30 during the headache itself and 10 during the recovery phase (PMID 8163372). Neck stiffness is in fact one of the canonical premonitory symptoms of an attack, alongside yawning, thirst and frequent urination (PMID 30074545), and that premonitory phase has been documented by electronic diary in a multicentre study (PMID 12654956).

A subgroup does have an upper cervical dysfunction

Kerstin Luedtke's team measured what palpation of the upper cervical spine gives in 179 migraine sufferers and 73 matched controls, examined by a physiotherapist blinded to the diagnosis. The technique combined oscillatory movements and sustained pressure. It separates migraine sufferers into three groups: 11 % with no pain, 42 % with local pain alone, and 47 % in whom sustained pressure reproduces pain referred to the head (PMID 28952052). The same authors have since summarised their whole research programme: musculoskeletal dysfunctions are frequent in migraine sufferers, and it is the reproduction of pain referred to the head, and not simple local tenderness, that appears to identify the relevant subgroup (PMID 37331926). An earlier study had found, in 55 women with episodic migraine and 16 with chronic migraine compared with 22 controls, alterations in upper cervical mobility and in cervicocephalic kinaesthesia (PMID 28118694).

But the neck pain of a migraine sufferer is not neck pain in the usual sense

This is the indispensable counterweight, and it comes from an Australian team. Comparing 124 migraine sufferers (106 episodic, 18 chronic) with 32 healthy controls and 21 patients with idiopathic neck pain, using a complete battery (cervical range and movement accuracy, segmental dysfunction, neuromuscular and sensorimotor measures), the authors conclude that neck pain associated with migraine does not necessarily reflect cervical musculoskeletal dysfunction. Patients with a comorbid cervical disorder had been excluded, precisely to isolate the question (PMID 34214181).

The practical consequence is direct. Faced with a migraine sufferer whose neck hurts, the question is not “should the neck be treated?” but “does this neck show, between attacks, the picture of a cervical disorder?”. Range, endurance of the deep flexors, segmental dysfunction reproducing the usual pain, examined away from an attack: if none of that is found, the neck is a symptom, and it will be treated by treating the migraine.

Neck pain is present in nearly two thirds of migraine sufferers. Cervical dysfunction is not. Confusing the two is the error that has necks rehabilitated for months without changing the number of attacks.

What can physiotherapy actually do, and at what level of evidence?

This chapter announces no spectacular good news, and that is precisely its value. An effective but modest modality, correctly presented to a patient, produces a better alliance than a promise the facts will disprove within three months.

Aerobic exercise, the best-supported modality

Two meta-analyses converge, and the gap between them is instructive. The first retained six trials: it concludes that there is moderate-quality evidence that aerobic exercise reduces the number of migraine days, with a mean reduction of 0.6 days a month (standard deviation 0.3). The authors could not conclude on pain intensity or attack duration, for want of homogeneous measures (PMID 30764753). The second, across 10 articles and 508 patients, finds significant short-term effects on intensity (standardised mean difference 1.25; 95 % confidence interval: 0.47 to 2.04), frequency (0.76; 0.32 to 1.2) and duration (0.41; 0.03 to 0.8), on low to moderate quality evidence. The authors themselves report a significant publication bias on the quality of life outcome (PMID 31904889). Recent network meta-analyses have looked for the optimal dose and put protocols combining aerobic work and strengthening at the top (PMID 41142313).

Manual therapy, a recent result that calls for nuance

The most relevant trial for our practice was conducted in primary care, on exactly the population that consults us: migraine sufferers who also have neck pain. Thirty-six patients received manual therapy (mobilisations, exercises, myofascial techniques) and thirty-one usual general practice care, with follow-up at 12, 26 and 52 weeks. On the primary outcome, the number of migraine days, there is no significant difference, nor on most secondary outcomes. The manual therapy group did, however, have higher pressure pain thresholds over the occipital muscles and better perceived recovery at 12 and 52 weeks, and used less preventive medication. The authors conclude that manual therapy is not superior to usual care on migraine days, but that patient preference and satisfaction may come into it (PMID 41705211).

Earlier, smaller trials had reported favourable effects of articulatory protocols (50 subjects randomised, PMID 33989990) and a three-armed trial compared a multimodal approach (manual therapies, cervical exercises and pain neuroscience education) with its individual components in 75 participants (PMID 40460766). The overall picture remains one of an uncertain effect on the disease and a plausible one on the cervical impact.

Behavioural approaches, to know about and to refer for

The most recent systematic review covers 50 trials and 6,024 adults. It concludes that cognitive behavioural therapy, relaxation training and mindfulness-based therapies may reduce attack frequency, on evidence described as low, and that education alone targeting behaviour may improve impact. For biofeedback, acceptance and commitment therapy and hypnotherapy, the data were insufficient to conclude. Most trials were at high risk of bias (PMID 39968795).

What can medical treatment do, and why does that concern you?

A physiotherapist does not prescribe. But they see their patient twenty times where the doctor sees them twice, and they are therefore best placed to spot the two situations that spoil a treatment: the absence of a preventive treatment, and analgesic overuse.

Migraine prevention has changed its face. The American Headache Society published an update to its consensus statement on integrating new treatments in 2021 (PMID 34160823), then in 2024 a position statement affirming that treatments targeting calcitonin gene-related peptide (CGRP) are a first-line option for prevention, where they had previously been reserved for failures of non-specific treatments (PMID 38466028). The International Headache Society for its part has called for the goals of preventive treatment to be raised, aiming at freedom from attacks rather than the halving that has become the standard in trials (PMID 39980456), and for earlier treatment to prevent disease progression (PMID 41134822).

Medication-overuse headache

This is the point where your observation is worth its weight in gold. Medication-overuse headache affects between 1 and 2 % of the world's population, and is defined as a headache occurring on more than fifteen days a month in a patient with a pre-existing primary headache, arising from the regular use of acute treatments (PMID 38568489). Recall the French figure: among people with chronic daily headache, 88 % were taking non-specific analgesics and only 6.6 % had a preventive treatment (PMID 12620605). The question to ask is simple and takes one sentence: “how many days a month do you take something for your head?”. Beyond ten days for specific treatments or fifteen for simple analgesics, there is something to raise with the family doctor.

The transformation of episodic migraine into chronic migraine has been the subject of a systematic review of its risk factors, which identifies in particular initial attack frequency, medication overuse, obesity, sleep disorders and painful and psychiatric comorbidities (PMID 30589090). Several of those factors are exactly the ones a physiotherapist already works on.

Migraine in women: what do hormones change?

Migraine is three times more common in women, and its relationship with hormones has a direct consequence for vascular risk. It is the only chapter of this article where the information you pass on can change a prescription.

Migraine with aura is associated with an increased risk of ischaemic stroke, and that association is amplified by combined oral contraception and by smoking (PMID 23432442). The subject has been debated since pills came to contain oestrogen doses far below those of the 1960s and 1970s on which the original contraindication rested (PMID 28806162), and a case-control study has specifically examined the effect of oestrogen dose and migraine subtype (PMID 36752588). Recent work on British general practice data has compared stroke risk between combined oral contraceptives and progestogen-only pills in women with migraine with aura (PMID 42096712). The French synthesis on migraine and contraception sets out the prescribing markers (PMID 25727163).

What that implies in the clinic comes down to one sentence: if a patient describes visual auras and also tells you she takes a combined oral contraceptive, the only thing to do, and it matters, is to ask her whether her doctor knows about those auras. Many have never mentioned them, because they do not connect them with their contraception.

Pregnancy and the post-partum period deserve separate vigilance, and for a reason that is not migraine itself. They are prothrombotic states, and cerebral venous thrombosis is over-represented in them: it is item 11 of the SNNOOP10 list (PMID 30587518). A new headache in a pregnant woman or one who has recently given birth is never unremarkable. The question of the safety of later pregnancies after cerebral venous thrombosis has in fact been the subject of a dedicated follow-up of the ISCVT cohort (PMID 28974635).

What do published cases teach us?

The cases that follow are all genuinely published and referenced reports. None is reconstructed, and none is a flattering example: they were chosen because they illustrate errors our profession makes.

A cervical manipulation on a headache that had changed

The most instructive case for a French physiotherapist was published in the Annales de réadaptation et de médecine physique. A 39-year-old woman, a current smoker, with migraine for ten years, presents with an unusual headache associated with neck pain. She is treated with cervical spine manipulation. Seven hours later, she develops a crossed syndrome with a right sensorimotor deficit and a cerebellar syndrome, a picture of vertebrobasilar ischaemia. The authors, themselves physical medicine physicians, recall that the incidence of these accidents is underestimated and that the risk falls if the contraindications are observed (PMID 18586346).

Two features of this case appear word for word in the list of red flags: the unusual headache in a known migraine sufferer, and the associated neck pain with a new headache, which must raise the possibility of arterial dissection before any technique. The association between cervical manipulation and cervical artery dissection has been the subject of dedicated work, including an analysis of American health databases comparing several study designs (PMID 37422607) and a review of 64 medico-legal files that could not identify either the at-risk patient or the type of manipulation involved (PMID 12195461). Two recent reports of carotid dissection after manipulation have been published (PMID 34971321).

A cerebral venous thrombosis in a woman on contraception

A 22-year-old woman, on oral contraception, presents with a right-sided headache, blurred vision and dizziness of four days' duration. Magnetic resonance venography shows complete thrombosis of the right transverse sinus, the sigmoid sinus and the internal jugular vein, together with partial thrombosis of the superior sagittal sinus and the left transverse sinus (PMID 33942790). Another report describes a cerebral venous thrombosis in a 31-year-old woman after a brief exposure to oral contraception, prescribed for endometriosis (PMID 35786211). The initial reason for consulting, in both cases, could have been referred to a physiotherapy clinic had it been labelled “headache and dizziness of cervical origin”.

What these cases do not prove

A case report does not demonstrate a frequency. These reports establish neither that cervical manipulations are dangerous in general, nor that headaches in women on contraception are thromboses. They establish something more limited and more useful: these sequences exist, and they all begin with a signal that was on the list of red flags.

How do you apply this in the clinic?

This chapter sums up what to do in five acts, three of which are not technical at all.

1. Ask three questions of every patient with a headache

They take less than a minute, and cover the most sensitive items of the SNNOOP10 list: was the onset sudden? Is this headache new, or different from usual? Is there a neurological deficit, a fever, a pregnancy or a post-partum period? A single positive answer is enough to suspend the treatment plan and refer.

2. Have them describe an attack, never ask for a diagnosis

Duration, side, pulsating quality, effect of exertion, nausea, photophobia and phonophobia: six elements, which are enough to recognise the picture. Six migraine sufferers in ten do not know they are (PMID 16643573); asking them whether they get migraines therefore means asking the question they do not have the answer to.

3. Examine the neck between attacks, and look for referred pain

Overall range, upper cervical mobility, endurance of the deep flexors, and above all sustained palpation of the upper cervical spine looking for pain referred to the head, the sign that identifies the relevant subgroup (PMID 28952052). In the absence of a cervical picture in its own right, do not build a cervical treatment.

4. Build the programme around aerobic exercise

It is the best-supported modality, and it is announced for what it is: a mean reduction of the order of half a migraine day a month, obtained through regular, progressive practice, and not through passive sessions (PMID 30764753). Adding strengthening work is consistent with the most recent network meta-analyses (PMID 41142313).

5. Count the days of acute treatment

One question per assessment, recorded in the notes. Beyond ten or fifteen days a month depending on the drug class, the information goes back to the family doctor. It is the most cost-effective act on the list, because it targets the only mechanism of worsening that is entirely reversible.

Frequently asked questions

On the same axis, in our resources