Your migraine trigger might be the migraine already starting

Short answer

Many of the things you have written down as triggers are not causes. They are the first symptoms of an attack that has already started in your brain, hours before the pain. The craving, the light sensitivity, the stiff neck, the yawning — these come from the same early phase. You did not eat the chocolate and get a migraine. The migraine made you want the chocolate.

That reframe has real evidence behind it. In a study of 1,010 people at a headache centre, those who had light sensitivity in the hours before the pain were the same people who named bright or flickering light as their trigger; the same pairing showed up for food cravings and "certain foods," and for smell sensitivity and odours. The authors concluded that most commonly reported triggers are "most likely just misinterpreted" early symptoms. (link.springer.com) A 2025 REFORM analysis of 632 people found the same overlap in numbers: 59.4% of people with pre-headache light sensitivity named bright light as a trigger, 49.0% of those with pre-headache sound sensitivity named loud noise, and 77.0% of those with pre-headache tiredness named sleep problems. (journals.sagepub.com)

And you are not imagining the warning. If you have ever said "I knew it was coming," you were right, and it is measurable. In an electronic diary study, people who reported early symptoms correctly predicted the headache from 72% of those diary entries. (neurology.org) In the much larger PRODROME screening period, 95.1% of early-symptom events were followed by a headache within 24 hours. (pmc.ncbi.nlm.nih.gov) The problem was never your perception. The problem is that the standard trigger diary asks you to write down what you did, when the useful information is what your body was already doing.

If your head has its own logic and you have never had the pattern behind it looked at, this is a decent place to start: take the 2-minute check-in — it maps what is happening in your nervous system behind the headache pattern, not just the headache itself.

What is the difference between a migraine trigger and a prodrome?

A trigger is supposed to be a cause: something external that arrives, and then an attack follows because it arrived. Remove it, and you remove some attacks.

A prodrome — also called the premonitory phase — is the opening act of the attack itself. Brain changes are already under way; the pain has not started yet. Typical early symptoms are tiredness, trouble concentrating, yawning, neck stiffness, mood change, thirst, food cravings, and heightened sensitivity to light, sound, and smell. (nature.com)

Both show up as "the thing that happened before my headache." That is the whole problem. From the inside, a trigger and a prodrome feel identical — something happened, then your head hurt. The difference is direction, and direction is invisible from where you are standing.

Why do the same things appear on both lists?

Because the early phase of a migraine changes what your senses do.

The prodrome lowers your threshold for light, sound, and smell. Light that was fine yesterday is unbearable today. A perfume that you normally walk past now makes you feel sick. Your brain is not responding to a stronger stimulus — it is responding with a lower threshold to the same one.

So the sequence you actually live is: attack starts quietly → light becomes intolerable → you notice the light → pain arrives. And the sequence you write down is: light → pain.

The 1,010-person study put the overlap plainly. Pre-headache symptoms starting two or more hours before the pain were present in 38.9% of patients, the most frequent being a tense neck, sound sensitivity, and difficulty concentrating — and the people who reported light sensitivity in that window were significantly more likely to name bright light as their trigger. (link.springer.com)

A provocation study made the same point from the other end. Fifty-three people were given nitroglycerin, which reliably provokes migraine; 83% developed a migraine-like headache and 98% had early symptoms first. Light as a reported trigger lined up with pre-headache light sensitivity, food as a trigger lined up with food cravings, and skipping meals lined up with cravings too. (link.springer.com) Nobody in that room was reacting to a real-world trigger. The drug started the attack, and the attack produced the "triggers."

Is chocolate really a migraine trigger?

This is the cleanest test case, because someone ran the experiment properly.

Chocolate is the most-blamed food in migraine. The largest double-blind test gave 63 women with chronic headache — about half with migraine — two samples of chocolate and two of a carob placebo matched for taste and texture, in random order, without anyone knowing which was which. Chocolate was not more likely to provoke a headache than carob in any diagnostic group, and the result did not depend on whether the person believed chocolate was their trigger. (doi.org)

That is not the only study, and it would be dishonest to pretend it is. A smaller 1991 trial found a migraine after chocolate in 5 of 12 people against 0 of 8 on placebo — suggestive, but short of statistical significance (p = 0.051) and tiny. (doi.org) A 2020 review that pooled everything found that chocolate is reported as a trigger by anywhere from 1.3% to 33% of people across 23 studies, while all three provocative studies failed to find a significant difference between chocolate and placebo, and concluded there is insufficient evidence to tell migraine patients to avoid it. (doi.org)

The explanation that fits is the prodrome. Food cravings — often specifically for sweet or chocolate — are a documented early migraine symptom. (nature.com) The attack makes you want chocolate. You eat chocolate. Hours later the headache arrives on the schedule it was already on. The chocolate is a witness, not a suspect.

That single reframe returns a food to a lot of people who have been avoiding it for years for nothing.

Does this mean stress and skipped meals do not matter?

No — and this is where honesty matters more than a neat story.

Some things do look like real precipitants rather than early symptoms. Sleep disruption, alcohol, hormonal shifts around menstruation, and fasting have mechanisms that make sense as causes. In REFORM, the most frequently named triggers were sleep disturbances (70.1%), stress (67.7%), and alcohol (59.0%). (journals.sagepub.com) Nothing here says you should start skipping sleep.

Stress deserves its own note, because it behaves strangely. The best-documented stress pattern in migraine is not stress itself but the let-down — the attack that lands after the pressure lifts, on the first day of holiday, on Saturday morning. A study testing that hypothesis found the risk rose in the hours after a decline in perceived stress, not during the stress. (europepmc.org) If you have ever felt punished for finally relaxing, that is a described phenomenon with a name.

And there is a mechanistic argument that some triggers are partly a readout of how excitable your brain currently is. In a study of people taking a preventive migraine drug, the number of triggers that were actually followed by a headache fell by 38% in treatment responders against 13% in non-responders. (doi.org) The triggers did not change. The brain's willingness to turn them into an attack did.

How long before the headache does the prodrome start?

Longer than most people assume, and the answer depends on which symptom.

The PRODROME screening period is the largest dataset: 920 participants logging 4,802 early-symptom events. The most common symptoms were light sensitivity (57.2%), fatigue (50.1%), neck pain (41.9%), sound sensitivity (33.9%), and dizziness (27.8%). Of those events, 81.5% were followed by a headache within 1–6 hours, 90.5% within 6 hours, and 95.1% within 24 hours. (pmc.ncbi.nlm.nih.gov)

A smartphone study that sampled people four times a day for weeks found the sharpest changes in the final 12 hours — sensory sensitivity, pain and stiffness, and fatigue all rose — while impaired concentration and flattened mood were detectable earlier, in the 25–36 hour window. (doi.org)

Put together: a day or two out, the changes are cognitive and emotional and easy to explain away as a bad week. In the last half-day, the sensory and physical symptoms arrive — and that is exactly when you are most likely to bump into something and call it a trigger.

Can you tell a real trigger from an early symptom at home?

Not reliably in one attack. But over several attacks there are three questions that separate them better than a trigger list does.

Did the exposure precede the change in you, or follow it? If the light became unbearable before you walked into the bright room, the sensitivity came first. Write down the time you first noticed the sensitivity, not the time you met the light.

Does the exposure cause an attack every time? A real cause should be reasonably consistent. If bright light gives you a migraine on one day in eight and nothing on the other seven, something on those days is different — and what's different is likely your internal state, not the lamp.

What else was already happening? Yawning, thirst, a stiff neck, unusual tiredness, craving, irritability, trouble finding words. If two or three of those were present before the "trigger," you were already in the attack.

A practical version: for two weeks, log your body twice a day — before noon and before bed — rather than logging events. Rate energy, concentration, neck, and light tolerance from 1 to 5. Then mark attacks. What you are looking for is not what you ate. It is whether the numbers had already moved before the day you blamed.

What does this change about treatment?

Quite a lot, and this is the practical payoff.

If the attack has already started during the prodrome, then the prodrome is a treatment window rather than a warning. That is no longer theoretical. In a phase 3 randomised trial, a migraine drug taken during the early phase — before any pain — improved early symptoms: at three hours, neck pain was gone in 28.9% of treated events against 15.9% on placebo, and fatigue in 27.3% against 16.8%. (doi.org)

Read that number again from the other direction. A drug developed for migraine relieves the neck pain you get before a migraine. That is about as close as clinical evidence comes to saying out loud that the neck pain is the migraine.

The second change is subtraction. Every food you avoid for nothing costs you something. If chocolate, cheese, and red wine are on your list because of a pattern that was actually the prodrome, testing them one at a time — on a day when you have no early symptoms — may hand a few of them back.

The third is accuracy with your clinician. "Bright light triggers my migraines" points a neurologist toward avoidance. "I get light-sensitive and can't concentrate about six hours before the pain, most times" points them toward early treatment, which is a different and usually better conversation.

What does the evidence not show?

Three honest limits, because the field is less settled than the confident version of this story suggests.

The prevalence numbers are all over the place. A systematic review and meta-analysis found a pooled prevalence of at least one early symptom of 29% in population-based studies and 66% in clinic-based ones, with very high heterogeneity, and concluded that the evidence was insufficient to characterise the phase reliably. (link.springer.com) The phenomenon is real; the percentages are soft.

Not every early symptom predicts anything. A prospective diary study of 487 people found increased depressive symptoms during the headache phase but not in the prodromal phase, and found no evidence for mood change as an early warning sign. (doi.org) Some of what you notice beforehand is genuinely noise.

The mechanism is debated. The popular explanation — that the prodrome comes from hypothalamic dysfunction — has been challenged as premature, with critics arguing the evidence is limited by methodological problems and inconsistent definitions. (link.springer.com) The clinical observation stands; the neurobiology behind it is still being argued.

And the selection problem: the strongest prediction studies deliberately recruited people who already said they could predict their attacks. 76.9% of PRODROME participants identified early symptoms followed by headache at least 75% of the time — but they were chosen for being good at it. (pmc.ncbi.nlm.nih.gov) If you cannot feel your prodrome, you are not doing it wrong.

When should you see a doctor?

Book an appointment if headaches are getting more frequent, if you are using acute painkillers on 10 or more days a month, if attacks are disabling, if the pattern has changed, or if you have never had a formal diagnosis. About half of migraine is never formally diagnosed, and without the diagnosis, prevention and early treatment are not really on the table.

Seek urgent care for a headache that comes on like a thunderclap and peaks within a minute, a headache after a head injury, headache with fever and a stiff neck, a first severe headache after age 50, or any headache with weakness, confusion, seizure, vision loss, or a change in consciousness.

Before that appointment, it is worth knowing what your own pattern looks like: take the 2-minute check-in — it shows the buildup in your nervous system that sits behind the headache pattern, which is exactly the part that never fits in a trigger diary.

How to bring this up with your doctor — and what to ask for

Bring the timeline, not the trigger list. One page: for your last five attacks, what you noticed in the 24 hours beforehand and how many hours before the pain it started. "Six hours of light sensitivity and word-finding trouble, then pain" is clinical information. "I think it was the wine" is not.

Say the sentence that opens the right door. Try: "I think I can tell when an attack is starting, several hours before the pain. Is there anything I can do in that window?" This asks about early acute treatment instead of avoidance.

Ask these specifically. Is what I'm describing a premonitory phase? Should I take my acute medication when the early symptoms start, or wait for pain? How many days a month am I allowed to use it before I risk medication-overuse headache? Given my frequency, am I a candidate for preventive treatment?

Bring the count. Headache days per month, acute medication days per month, and days when you couldn't work or function. Those three numbers drive most treatment decisions.

If you are dismissed. "I understand. Can you note in my record that I asked about treating the premonitory phase and we decided against it for now?" A request on the record gets reconsidered more often than a request that isn't. And ask for a referral to a headache specialist — migraine care differs enormously between a general clinic and a headache service.

How Welltory helps

Welltory does not detect migraines and does not predict attacks. Nobody should sell you that, and this article's own data section says plainly what we can and cannot see.

What the app does give you is the part a trigger diary throws away: a continuous record of your own physiology. Resting heart rate, heart rate variability, sleep, and daily load, measured the same way every day, against your baseline rather than a population average. When you go back and look at the days around an attack, you are looking at measurements you did not have to remember to write down — which matters, because the prodrome is exactly the period you are least able to observe accurately while it is happening.

And the brain fog finding in our data is worth taking personally. If you live with migraine and you have been treating your foggy days as a discipline problem, half the people in your situation report the same thing. It is a symptom, it has a place in the attack cycle, and it is a reasonable thing to raise with a clinician.

How we made it

The clinical content rests on headache-medicine literature: the trigger–prodrome overlap studies (Schulte & May 2015; Karsan et al. 2021; the REFORM analysis 2025), prediction studies (Giffin et al. 2003; the PRODROME screening period), the chocolate provocation trial (Marcus et al. 1997), and the phase 3 prodrome treatment trial (Goadsby et al. 2025). We deliberately included the counter-evidence: the meta-analysis questioning prevalence estimates, the diary study finding no mood-based early warning, and the critique of the hypothalamic explanation.

The data section uses Welltory's own curated cohort: 4,145 users with wearable-quality data, 396 of whom report migraine in an onboarding survey. Every candidate metric was tested against a verification gate — a difference only survives if it holds within strata of how many other conditions a person reports. Resting heart rate failed that gate and is reported here as failing it. Brain fog passed, and is the only metric in the panel that did. We have no attack log, so no claim about prediction is made anywhere in this article.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data