For most of the twentieth century, migraine had a tidy explanation. Blood vessels in the head constricted and then dilated; the dilation stretched pain-sensitive structures; that was the headache. It explained the throbbing quality, it justified the drugs of the era, and it was wrong in the way useful theories usually are — not absurd, just describing a downstream consequence as though it were the cause.

What the Vascular Account Could Not Explain

It could not explain the hours before the pain. The irritability, the food cravings, the yawning, the sensitivity to light that arrives before anything hurts — a vessel-diameter theory has nothing to say about a prodrome that precedes the vascular event.

It struggled with the aura, a slow-moving disturbance of vision or sensation that travels across the cortex at a rate that does not match blood flow.

And it could not explain why the aftermath lasts. The day of flattened, foggy recovery that follows an attack is not what you would expect from a vessel that has returned to its usual diameter.

None of that made the vascular observations false. Vessels do change calibre during an attack. The error was one of ordering — treating a visible downstream event as the origin of the thing, which is a recurring hazard when the accessible measurement is not the important one.

What Replaced It

A 2026 review states the change directly: once viewed mainly as a vascular headache, migraine is now understood as a heterogeneous neurobiological syndrome arising from interactions among genetic susceptibility, hormonal influences, environmental exposures, and cortical, brainstem, trigeminovascular and neuroinflammatory networks (MedComm, 2026).

Each word in that list is doing work. Heterogeneous means it is probably not one condition. Networks means the unit of explanation is a system rather than a structure. Genetic, hormonal, environmental means susceptibility and trigger are different questions.

The mechanistic models the review covers are correspondingly varied: cortical spreading depolarisation, trigeminovascular signalling, CGRP biology, neurogenic inflammation, glial activation and brain network dysfunction. That is not a field converging on one answer. It is a field describing several interacting processes.

The Scale, Which Is Easy to Underestimate

About 1 billion people worldwide, with the burden falling disproportionately on women and on working-age adults.

Both halves of that distribution matter. A condition concentrated in working-age adults is a condition that removes people from work and family life during the years those demands are heaviest. A condition that disproportionately affects women is one with a long history of being under-investigated and under-treated.

It is also why the hormonal thread in the mechanistic account is not incidental — it connects to the broader question of how hormonal transitions act on the brain.

What Has Genuinely Improved

The review is clear that treatment has changed substantially. CGRP-pathway therapies, gepants and neuromodulation have transformed care — that is the review’s own word, and it reflects a shift from borrowing drugs developed for other purposes to targeting a mechanism specific to migraine.

The distinction is worth spelling out. For decades the mainstay treatments were compounds developed for other purposes and found, empirically, to help — drugs that constricted vessels, or that were designed for blood pressure, epilepsy or depression and turned out to reduce attack frequency. They worked for some people, and nobody could say precisely why.

That progress came directly out of the reframing. You cannot target CGRP biology while you believe the problem is vessel diameter.

A mechanism-specific therapy is a different kind of object. It is designed against a target the field believes is doing the work, which means its successes and its failures both inform the underlying model rather than sitting outside it.

Why Treatment Is Still About More Than Pain

A second 2026 review takes up what the pharmacological advance leaves unfinished. It describes a biopsychosocial model proposed to explain migraine’s complexity — the interconnections between emotional, psychological, social and biological factors — in contrast to a strict medical model (Nature Reviews Neurology, 2026).

Its argument is explicit: this modern view recognises that relying solely on pharmacological treatment might not be sufficient, and that psychological approaches should be considered both to help people manage the condition and to enhance the effectiveness of medications.

The interventions it describes are relaxation training, biofeedback, cognitive behavioural therapy, mindfulness, and acceptance and commitment therapy — with patient education named as the first step of care.

It is also a claim with a specific scope. The review describes these interventions as relevant to managing the condition and to supporting the effect of medication; it does not present them as reducing attack frequency on their own, and reading them that way would overstate what is being argued.

Two things are worth noticing about that framing. It is additive: the argument is that these work alongside medication, not instead of it. And putting education first treats understanding the condition as part of the treatment rather than a preliminary to it.

What Remains Unsolved

The 2026 review names the gaps rather than glossing them. Major ones remain in disease stratification — telling apart what is probably several conditions currently sharing a label — in preventing chronification, the transition from episodic to chronic migraine, and in the management of special populations.

Chronification is arguably the most consequential of the three. The shift from occasional attacks to near-continuous ones changes the condition’s entire character, and preventing it would be worth more than most improvements in treating individual attacks.

It is also worth being careful about what a review can establish. Both papers here are reviews — syntheses of a literature, not new experiments — and a mechanistic model that appears in one is a current best account rather than a settled fact. The heterogeneity the field keeps describing is a genuine obstacle to certainty, not a rhetorical hedge.

None of it is diagnostic for an individual, and none of it substitutes for a clinician. What it supports is the narrower point that a condition understood as a network problem is one where the things that act on the whole system — including sleep and sustained stress — are part of the picture rather than a distraction from it. That connects to why sleep disruption has consequences well beyond tiredness.