Migraine Warning Signs Before It Starts: Find Your Own

An attack does not begin when the pain does. For a lot of people it begins hours earlier, in a phase clinicians call premonitory, and the signs are ordinary enough to be dismissed: yawning, a stiff neck, wanting something sweet, finding the office lights harder than usual, reading the same paragraph three times.

Most pages list those and stop. The list on its own is close to useless, because the same symptoms turn up on plenty of days when nothing follows. What makes them worth anything is knowing which of them predict an attack for you, and how often they are wrong.

The hit rate is measurable, and it is not 100%

This has actually been studied rather than surveyed. In a three-month electronic diary study, people who reported warning symptoms correctly predicted a full migraine from 72% of the diary entries in which those symptoms appeared.

The most common features were feeling tired and weary, present in 72% of attacks that had warning features, difficulty concentrating at 51%, and a stiff neck at 50%.

Sit with the 72% for a moment, because it sets the right expectation in both directions. That is high enough to act on — most of the time, the warning was real. It is also low enough that roughly one prediction in four was wrong, so anything you decide to do on a warning has to be something you would not mind doing unnecessarily.

The half everybody skips

Here is where personal warning-sign hunting goes wrong, and it is the same error that ruins trigger diaries.

You remember the times the sign appeared and an attack followed. You do not remember the times it appeared and nothing happened, because nothing happening is not memorable. So every candidate sign feels like a reliable predictor, and the ones you rely on are simply the ones you noticed most often.

The fix is two columns and a fortnight of honesty.

Column one: did I notice the sign today? Pick two or three candidates, no more. Yawning far more than usual. Neck stiff on waking. Craving something specific. Words not arriving.

Column two: did an attack start within 24 hours? Yes or no.

At the end of six weeks you have four numbers per sign: the times it appeared and an attack followed, the times it appeared and none did, the attacks that arrived with no warning at all, and the quiet days. The second of those is the one that decides whether a sign is worth anything, and it is the one nobody collects.

A sign that appears before nine of your ten attacks looks excellent until you notice it also appeared on twenty-two days when nothing happened. That is not a warning system, it is a description of your Tuesdays.

Two or three candidates, not fifteen

The instinct is to watch everything. It fails for a dull reason: watching fifteen things means one of them is always doing something, so there is a warning every day and the system has no information in it.

Start with the three from the study — tiredness, difficulty concentrating, neck stiffness — or with whichever one you already half-suspect. Run it for six weeks. Add another only when you have retired one.

There is a subtler cost to watching everything, worth naming. Constant self-monitoring for signs of an attack is its own kind of unpleasant, and for some people it makes the good days worse without making the bad ones better. If a fortnight of this leaves you more anxious rather than better prepared, stopping is a legitimate finding rather than a failure.

What a warning is actually worth

Two to twelve hours is enough to do a small number of concrete things, and the value is in deciding them in advance rather than improvising while the fog descends.

Moving the difficult meeting. Getting home before the light becomes a problem. Eating, because the evening you skip dinner is a bad evening to be already in a premonitory phase. Telling one person, so the cancellation later is not a surprise.

Aura, for the people who get it, is the sharpest warning of the lot — and it sometimes turns up with no headache behind it at all, which is still an attack and still worth a mark in the record.

Anything involving medication timing belongs with whoever prescribes for you, and it is a genuinely live question rather than an obvious one — treating early works better, and treating often has its own cost, which is the tension medication overuse headache describes. Do not resolve that from a blog post.

The confusion that corrupts trigger diaries

The premonitory phase is also why so much trigger hunting produces confident wrong answers.

Craving chocolate. Feeling thirsty. A stiff neck. Finding fluorescent lights unbearable. Those are the classic reported triggers, and they are also a decent list of premonitory symptoms — which means that in many cases the attack was already running and it produced the thing that later got blamed. How to track migraine triggers without fooling yourself covers the confusion in full, and it is the reason the famous list of trigger foods is dominated by things people crave rather than things people eat.

The practical defence is a labelling habit rather than extra effort. Write “possible warning, 2pm” instead of “trigger: bad sleep”. Same observation, two very different records, and only one of them stops your trigger data being quietly poisoned by your prediction data.

What to write down

Add one line to whatever you already keep:

• Warning noticed? Which one, and roughly what time
• Attack within 24 hours? Yes or no, including on days with no attack

Two marks. The clear days are half the measurement, so a record that only gets filled in when something happens cannot answer this question at all.

The free printable diary has a row per day, which is the shape this needs, and Migraine Journal timestamps entries as you make them so the order of events survives — which matters here more than anywhere else, because the entire question is what came before what.

None of this is medical advice, and a warning sign is not a diagnosis of anything. New neurological symptoms, or anything that comes on suddenly and severely, are a reason to be seen rather than logged. The other tracking guides build on the same daily record.