Medication Overuse Headache: When Rebound Becomes the Problem

You take something for the attack. It works. Next time you take it a little sooner, because waiting only made the last one worse. None of those decisions looks like a mistake from the inside. But past a certain number of days per month, in some people, the treatment starts feeding the pattern it was meant to interrupt.

That is medication overuse headache, and the short answer is a number rather than a dose. ICHD-3, the International Headache Society's classification, defines it as headache on 15 or more days a month in a person with a pre-existing headache disorder, developing as a consequence of regular overuse of acute or symptomatic headache medication on 10 or more, or 15 or more, days a month depending on the medication, for more than 3 months.

ICHD-3 also states the part worth holding on to: it usually, but not invariably, resolves after the overuse is stopped.

Why the old name misleads

ICHD-3 lists rebound headache under "previously used terms" for this diagnosis, alongside drug-induced headache and medication-misuse headache. Rebound suggests a single dose wearing off and the pain bouncing back the same afternoon, which is not what the diagnosis describes. The criteria are about a pattern that has to hold for more than three months before anyone can call it that.

ICHD-3 is careful about blame, too. It calls medication overuse headache an interaction between a therapeutic agent used excessively and a susceptible patient. Two things have to be true at once, and susceptibility is not something a person chooses. Among people with a previous primary headache diagnosis, ICHD-3 notes, most have migraine or tension-type headache or both, so the people at risk are largely the people whose headaches were already frequent enough to need treating.

The day counts that define it

Thresholds differ by drug class, and the split is the thing to memorize. Ergotamine, triptans and opioids each carry a threshold of 10 or more days a month for more than 3 months. So do combination analgesics, which ICHD-3 defines as formulations combining drugs of two or more classes, each with analgesic effect or acting as adjuvants, and which it gives a separate code to because they are so commonly implicated.

Plain painkillers sit at the higher line. Paracetamol has a threshold of 15 or more days a month for more than 3 months, and so do NSAIDs including acetylsalicylic acid. ICHD-3 treats non-opioid analgesics as one class for counting purposes, which it makes explicit when it says that a product combining only aspirin and paracetamol is not a combination analgesic, because both drugs are in the same class. Paracetamol on some days and ibuprofen on others is one running total, not two.

The rule most people have never heard of

There is a fifth pattern that catches people who are under every individual limit. ICHD-3 code 8.2.6 covers medication overuse headache attributed to multiple drug classes not individually overused, and its criterion is regular intake of any combination of ergotamine, triptans, non-opioid analgesics and opioids on a total of 10 or more days a month for more than 3 months, without overuse of any single drug or class alone.

Work that through with real numbers. A triptan on 6 days, ibuprofen on 5 days. Neither crosses its own line: the triptan is under 10, the ibuprofen nowhere near 15. But the total is 11, and 11 is over the threshold for 8.2.6. Nobody arrives at that combination carelessly; it is what sensible rationing looks like when you are trying to spread your triptans across a bad month.

One honest caveat belongs here. ICHD-3 says outright that the specified numbers of days considered to constitute overuse are based on expert opinion rather than on formal evidence. These are working lines drawn by a committee, not biological cliffs, and crossing one by a day is not a verdict about you.

How common it is

Common enough that the thresholds get a lot of use. A 2016 review by Westergaard, Munksgaard, Bendtsen and Jensen of the Danish Headache Center, published in Therapeutic Advances in Drug Safety, reports prevalence often cited as 1 to 2% but running as high as 7% overall, with higher proportions among women and among people in a low socioeconomic position.

One figure reframes the problem, and it comes from ICHD-3 itself: epidemiological evidence from many countries indicates that more than half of people with headache on 15 or more days a month have medication overuse headache. If you are in the chronic range, the odds that acute medication is part of what keeps you there are better than even.

That does not make it the whole story, and it says nothing about any one person, but it is the first thing a headache clinic will want to rule in or out. Our guide to how many migraine days count as chronic covers where that 15-day line comes from.

What happens when people stop

Much of what we know comes from a randomized trial by Carlsen and colleagues at the Danish Headache Center, published in JAMA Neurology in 2020. They randomized 120 patients with medication overuse headache to one of three outpatient strategies: withdrawal plus preventive treatment from the start, preventive treatment without withdrawal, or withdrawal with optional preventives two months later. Of those, 102 completed the six-month follow-up.

Headache days per month fell by 12.3 in the withdrawal plus preventive group, 9.9 with preventives alone and 8.5 with withdrawal alone, and those differences were not statistically significant. Where the groups did separate was on getting out of the chronic range: 74.2% reverted to episodic headache with withdrawal plus preventives, against 60.0% with preventives alone and 41.7% with withdrawal alone. The authors' conclusion was that all three strategies worked, and that combining withdrawal with preventive medication from the start is what they recommend.

Carlsen's group followed 96 of those patients for a year, reporting in Headache in 2021. Headache days per month across the whole cohort fell from 24.6 to 15.0, and 11% relapsed. But 41% still had chronic headache at one year, which is the number that keeps this honest. Stopping the overuse is not a reset button, and the trial's own predictors of still being chronic a year later were higher headache frequency, more days with acute medication, higher pain intensity and depression.

A word on how, because this is the part that belongs with a doctor rather than an article. Withdrawal in these trials was a supervised plan with preventive treatment alongside it, not a decision to stop cold on a Tuesday. ICHD-3 notes that patients overusing opioids have the highest relapse rate after withdrawal treatment, which is one of several reasons the class you are stopping changes the plan. Bring this to the person who prescribes for you.

The number nobody remembers

Everything above turns on a count of medication days, and that is the count almost nobody can produce on request. Headache days at least announce themselves. Medication days are a blister pack in a bag, a tablet at work, a dose at 2am that by morning has merged with the night before.

ICHD-3 has a code for exactly this problem. 8.2.7 covers patients who are clearly overusing multiple drugs but cannot give an adequate account of their names or quantities, and the classification's comment is blunt about the fix: in almost all cases, this necessitates diary follow-up. The diagnosis is frequently unreachable without a record made at the time.

I started building for my mom and my wife, who both get migraines, and this is the gap I did not expect to find. Their attacks were vivid and easy to describe. Medication days were invisible, and undercounted every time, never in the other direction.

Some tracking apps can at least carry the other half of the picture. MigrAid, for example, logs an attack in one tap and exports a dated history as PDF or CSV over a range you choose, so the headache-day count is written down before the appointment. Medication days are a count you keep yourself, on paper or in your phone, and keeping it at all puts you ahead of the 8.2.7 problem.

What to bring to your doctor

Two numbers do most of the work: headache days per month and days per month on which you took anything for it, separated by drug where you can manage it. A rough count written at the time beats a precise-sounding guess made in the waiting room, and the 2021 follow-up found days with acute medication to be an independent predictor of still being chronic at one year, which is why clinicians ask.

It is also worth saying the awkward part out loud, because the consequence of not saying it is a preventive plan built on the wrong picture. ICHD-3 is unambiguous that this diagnosis matters clinically: the majority of patients improve after discontinuation of the overused medication, and so does their responsiveness to preventive treatment. Our notes on preparing for a neurologist appointment and on the MIDAS disability score cover how to present both in the ten minutes you get.

One nearby thing that is not this: the headache that follows a missed morning coffee is caffeine withdrawal, a separate ICHD-3 entry with its own timing, and we cover it in the piece on caffeine and migraines. Caffeine matters here too, since it is a common adjuvant in the combination analgesics that carry the 10-day line, but the two problems are not the same problem.

The bottom line

Medication overuse headache is defined by days, not by doses or willpower. Ten days a month for triptans, ergotamine, opioids and combination analgesics; 15 for paracetamol and NSAIDs; 10 for any mix of classes even when each one is under its own limit. Sustained for more than three months, in a person who already had headaches.

The trial evidence says the situation is workable: three different strategies all reduced headache days, three quarters of patients on withdrawal plus preventives came back out of the chronic range within six months, and only 11% relapsed over a year. What none of it works without is the count. Write down the days you medicate, take that number to your doctor, and let the plan be built on what actually happened rather than on what either of you can remember.