The Journal

Migraine

Cyclical migraine: treat the map, not only the attack

Cyclical migraine mapped, not just medicated

Dr Amy Shacaluga·23 September 202614 min read

How I grade the evidence in this note

• 🟢 Green: strong enough to teach and act on with confidence, still inside a plan that fits you.
• 🟡 Amber: useful with caveats. I will say what would make it greener.
• 🔴 Red: interesting, or not a first move yet.

You already know the pattern can be real: attacks that cluster with bleeding, often harder to settle than your other migraines. This note is the next step. Lifestyle medicine comes first, because it is the only part of this picture that works on every cycle rather than one attack. Precision Health is the map: your window, your drivers, and which biology is actually speaking in you. Conventional UK acute advice belongs in the kit once an attack has started. It is not the evaluation.

Not another prevalence essay. Not PCOS or perimenopause by another name.

First, let conventional medicine clear the red flags

A diary plan is for the migraine you already recognise. It is not for a headache that is new, sudden, or different.

Seek urgent care (999, NHS 111, or same-day medical review) if this is the worst headache of your life, it comes on in seconds, you have fever or a stiff neck, weakness, speech change, confusion, a fit, a headache after a head injury, a headache that wakes you and keeps getting worse, or you are pregnant or recently pregnant and this does not feel like your usual migraine. New migraine after about fifty, or a known cancer or poorly working immune system, needs a proper medical look rather than a lifestyle note.

That screen is conventional medicine doing its job. Once those are out, we can treat the pattern in front of us.

Name the pattern before you name the hormone

UK guidance asks us to suspect menstrual-related migraine when attacks land predominantly between two days before bleeding and three days after it starts, in at least two of three cycles, and to confirm it with a headache diary kept over at least two cycles [1,2]. Pure menstrual migraine (that window only) is the minority. Menstrually related migraine, meaning the window plus other days, is far more common [2,3].

One honest caveat, because it changes how hard I push: these sit in the appendix of the international classification, not the main body [2]. Menstrual migraine is a well-described pattern, not a formally accepted separate diagnosis. Diary studies do suggest peri-menstrual attacks run longer and respond less well to acute treatment, around 23 hours against 16, with lower two-hour pain-free rates [3]. A 2023 meta-analysis found the differences smaller and the certainty low [4]. So I treat "these ones are worse" as your lived observation to be mapped, not a fact to be assumed.

Keep the diary for at least eight weeks: cycle day, attack start and duration, severity, aura yes or no, nausea, sleep (bedtime, wake time, and how broken the night was), food and alcohol, stress load, medicines taken, and life impact [1]. That is Precision Health in its plainest form. Assessment before treatment, not a boutique blood test. 🟢

The lifestyle floor, and why it is not the soft option

The week before a bleed is usually when sleep, food and stress come apart. That instability is a driver in its own right, not background noise. This is the part of the plan that tries to make the next cycle quieter, which no acute tablet can do.

Sleep continuity, more than sleep hours. The best prospective data here surprise people. One study followed 98 women and men with migraine for six weeks, using a wrist monitor to measure their sleep and a diary filled in twice a day. Two things did not bring an attack on: a short night, and simply feeling you had slept badly. What did matter was a broken night, the kind where you are in bed for eight hours but awake for a good part of them. After one of those nights, the odds of a migraine went up by roughly 39 per cent, and not the next morning but the day after that [5]. So the target in your luteal week is a protected, unbroken night: a consistent wake time, morning light, fewer late screens, alcohol out of the last few hours before bed. Chasing an extra hour matters less than protecting the continuity of the hours you get. 🟡, one good cohort, not a trial.

Movement, dosed properly. Aerobic exercise reduces migraine frequency and intensity across pooled trials, with a dose-response signal: benefit begins somewhere around 200 to 300 cumulative minutes and plateaus near 900, which in practice looks like roughly three 30-minute sessions a week sustained for ten weeks or so [6]. Yoga performs at least as well as moderate aerobic work in network comparisons [7]. The certainty is low, the trials are small and unblindable, and none of this is menstrual-migraine-specific. It is still the most reliably useful thing on this list, and it is free. 🟡, close to green for general migraine, unproven for the bleed window specifically. This costs nothing and a movable lever should you choose.

Food as a pattern, not a rulebook. The strongest dietary evidence in migraine is a 182-person randomised trial of fatty acid composition. Raising omega-3 to about 1.5 g EPA plus DHA daily while cutting linoleic acid produced four fewer headache days a month than control, and raising omega-3 alone produced two fewer [8]. Worth saying plainly: the diets were unblinded and intensively supported, and the headache-impact score did not improve significantly even though headache days did. Practically, that means oily fish two or three times a week, extra virgin olive oil in place of seed oils used for everything, and fewer ultra-processed foods carrying the linoleic load. Add regular meals so glucose does not crash. Skipping meals is reported as a trigger by around 57 per cent of migraine patients [9], although that is self-report from a headache clinic, and premonitory hunger can be mistaken for a dietary trigger. Eat regularly because it is a sane thing to do in a demanding week, not because a study proved it. Beware of the "fasting fads" especially during your luteal phase. 🟡 for fatty acids, 🟡 for meal regularity.

The stress and mind-body piece, honestly graded. In a randomised trial against active headache education, mindfulness-based stress reduction did not reduce migraine days more than education did. Both arms improved on migraine days; MBSR did improve disability and headache impact meaningfully [10]. That is not nothing. It is the difference between a cycle that floors you and one you can work through, which is usually what women actually come to me about. 🟡, for impact and function rather than attack count.

Alcohol and the pre-bleed week. Not a moral position. Alcohol fragments sleep exactly when sleep continuity is doing the most work for you, and wine sits on a migraine brain in its own right. Moving it out of the five days before a bleed is one of the cheapest experiments you can run, and your diary will tell you within two cycles whether it earned its place. 🟡

Supplements, kept in proportion. Magnesium is the one I discuss most often for this pattern. The only menstrual-migraine-specific trial randomised 24 women (20 completed) to 360 mg daily from ovulation to bleeding, and found fewer headache days and lower premenstrual symptom scores [11]. Twenty completers, in 1991. A later systematic review graded magnesium "possibly effective" overall and suggested higher daily doses [12], while a separate meta-analysis could not pool the data convincingly at all [13]. Riboflavin 400 mg and coenzyme Q10 each rest on one small positive trial in unselected migraine, with later evidence mixed or insufficient to pool cleanly, and neither tested in menstrual migraine [13]. So: a conversation, dose-limited by loose stools, not a protocol. 🟡 for magnesium in this window and the conversation is bigger than this blog. I always ensure I know your background before prescribing supplement support, this includes magnesium. 🔴 for anyone selling you a stack.

These are the first moves. They do not replace medicines once an attack has landed. 🟡 as sole therapy. 🟢 as the floor under everything else.

Which driver is speaking in you?

Precision Health here is not one blood result. It is matching your story to the most likely driver, then treating that, instead of handing every woman the same tablet.

Oestrogen withdrawal. Attacks cluster as oestrogen falls in the late luteal and early bleed days, and serotonin and pain threshold fall with it. It is the leading hypothesis and a useful working model. It is also less settled than it sounds: a 2023 critical review concluded the supporting evidence remains limited, with small studies and inconsistent case definitions [14]. In a double-blind crossover study, peri-menstrual estradiol gel reduced migraine in the window but attacks rose in the days after stopping [15]. If your diary is almost entirely bleed days, withdrawal is a fair hypothesis. If attacks also land at ovulation or scatter through the month, it is not the whole story. 🟡 for the hypothesis. 🔴 for gel as a routine UK strategy.

A GABA-sensitive luteal week. In the second half of the cycle, a progesterone metabolite (allopregnanolone) acts on GABA-A receptors, the same calming system alcohol and some sleep medicines touch. For some women that is soothing. For others it is the reverse: wired, broken sleep, a shorter fuse [16]. I want to be straight with you about the limits here. That paradoxical effect is described in PMDD, not proven in migraine, and the one direct study measuring it in menstrually related migraine found lower allopregnanolone, not higher [17]. There is no evidence based blood test that settles it and no treatment trial built on it. What it gives me is a clinical pattern worth recognising, not a mechanism I can claim. 🟡 as a pattern to watch for. 🔴 as a mechanism to claim.

Pillars off their pins. A clean hormone story still fails if you have slept four broken hours, skipped lunch, and used wine as a down-switch. Those are causes, not character flaws. They are also the only causes you can predict a week in advance. 🟢 as part of the map.

Contributors worth naming. Iron: iron deficiency anaemia was commoner in women with migraine than controls in a case-control study of 510 women, and independently associated with the menstrual pattern [18]. Direction of causation is unproven and no trial shows repletion cuts attacks, but if your bleeds are heavy, ferritin and a full blood count are a reasonable question, not a luxury. Metabolic health: the popular "insulin resistance drives migraine" line does not hold up cleanly. In the largest prospective analysis, an insulin-resistance marker did not predict incident migraine at all [19]. I still check fasting insulin and HbA1c when the wider picture warrants it, for your cardiometabolic health rather than as a migraine fix. Then the ordinary suspects: neck and jaw load, a new combined pill or stopping one, and medication overuse if painkillers have become the weekly plan [1,2].

What Precision Health is here, and what it is not

Is: a diary mapped against your cycle. Targeted bloods with a reason attached, such as ferritin and full blood count with heavy bleeding, thyroid function, HbA1c or fasting insulin where the metabolic picture warrants it. Wearable sleep data used to test one hypothesis (is the luteal week broken, or just short), not to generate anxiety. A plan written for your week, your work, and your household. 🟢 where each test answers a named question.

A nod to the conventional ladder

Brief, because this treats the attack in front of you rather than the pattern behind it.

The acute plan is the same as for your other migraines: a triptan when appropriate, with or without an anti-inflammatory or paracetamol, plus an antiemetic if you need one. Opioids and ergots are not the plan [1]. Have it within reach before the window opens, and take it early in the headache phase. 🟢

Ask yourself calmly: triptans on ten or more days a month, or simple painkillers on fifteen or more? Extra peri-menstrual triptan days stacked on mid-cycle doses raise the risk of medication-overuse headache [1,2]. That is a reason to review the plan, not a scolding. 🟢

If cycles are predictable and acute treatment is not enough, UK guidance supports short-course frovatriptan 2.5 mg twice daily, or zolmitriptan 2.5 mg twice or three times daily, across the expected days [1,20]. It works: pooled numbers needed to treat sit around four for frovatriptan twice daily [21]. It is also off-label in the UK, and the pivotal trials were manufacturer-funded [22]. 🟡, and a poor fit on irregular cycles. 🔴 as a forced first move when cycles are unpredictable.

If you need contraception and you have migraine with aura, combined hormonal contraception is not the method, at any age [23]. The relative risk of ischaemic stroke rises roughly six-fold against no migraine and no combined pill, although the absolute risk at these ages stays small [24]. 🟢 as a stop sign. Keep HRT for symptoms and risk, not as a migraine shortcut. 🔴 as migraine therapy. Where disability is high mid-cycle too, conventional prevention and later specialist CGRP pathways can join the plan [1]. 🟡, after the map, not instead of it.

Precision Health pearl

The diagnosis is a diary aligned to the bleed window, not a hormone level. The treatment is whichever driver that diary points to, supported by a lifestyle floor built in the one week of the month you can predict in advance.

Lifestyle takeaway

For the next two cycles:

• Run the diary: cycle day, aura, sleep (including how broken), food, alcohol, medicines, impact.
• Defend sleep continuity, regular meals, and three 30-minute walks or sessions a week, in the week you can already see coming.
• Move alcohol out of the five days before bleeding and let the diary judge it.
• Have the conventional acute kit ready before day minus two.
• Bring one question to clinic: which driver looks most like me, and would any tablet earn its place without tipping me into medication overuse?

Clinician note

If you live with cyclical migraine and you are approaching your luteal phase, prepare. Do not burn the candle at both ends. Meal-prep so you are not choosing between a sugar crash and an unplanned fast, and hydrate.

A food and sleep diary over two cycles can be extremely enlightening, if it serves you. Skip it, or stop it, if tracking food makes nutrition a problem or brings old eating difficulties back to the surface. The diary is a tool, not a test you pass.

If you would like to take this further

If you want the map treated properly — diary, sleep and food week, which driver is speaking, and whether any tablet has a job — that is what a consultation is for.

Consultations and fees

This article is general education, not personal medical advice. Not for emergencies. Urgent care: 999 or NHS 111. Video consultations are UK based.

References

1. NICE. Headaches in over 12s: diagnosis and management. CG150, 2012 (updated December 2021; last updated June 2025).
2. Headache Classification Committee of the IHS. ICHD-3. Cephalalgia 2018;38:1-211 (appendix A1.1.1, A1.1.2; 8.2.2).
3. Pinkerman B, Holroyd K. Cephalalgia 2010;30:1187-94.
4. Wang M et al. J Neurol 2023;270:1249-65.
5. Bertisch SM et al. Neurology 2020;94:e489-96.
6. Ogrezeanu DC et al. Headache 2026;66:40-52.
7. Reina-Varona A et al. Headache 2024;64:873-900.
8. Ramsden CE et al. BMJ 2021;374:n1448.
9. Kelman L. Cephalalgia 2007;27:394-402.
10. Wells RE et al. JAMA Intern Med 2021;181:317-28.
11. Facchinetti F et al. Headache 1991;31:298-301.
12. von Luckner A, Riederer F. Headache 2018;58:199-209.
13. Okoli GN et al. Can J Neurol Sci 2019;46:224-33.
14. Raffaelli B et al. J Headache Pain 2023;24:131.
15. MacGregor EA et al. Neurology 2006;67:2159-63.
16. Backstrom T et al. Prog Neurobiol 2014;113:88-94.
17. Rustichelli C et al. Cephalalgia 2020;40:1355-62.
18. Gur-Ozmen S, Karahan-Ozcan R. Pain Med 2016;17:596-605.
19. Wu J et al. J Oral Facial Pain Headache 2025;39:165-75.
20. SIGN 155: Pharmacological management of migraine.
21. Hu Y et al. J Headache Pain 2013;14:7.
22. Silberstein SD et al. Neurology 2004;63:261-9.
23. FSRH. UK Medical Eligibility Criteria, 2025 summary tables (migraine with aura, CHC = UKMEC 4).
24. Champaloux SW et al. Am J Obstet Gynecol 2017;216:489.e1-7.