How Long Before a Migraine Preventive Works?

Two months, roughly, for a tablet. Three for a monthly injection. Those are the published numbers, and the more useful fact is that the clock does not start when you swallow the first dose.

The American Headache Society’s consensus statement on integrating new migraine treatments recommends giving oral preventive treatments an adequate trial of at least eight weeks at a target or usual effective dose, and switching if there is no response after that. For the injectable CGRP monoclonal antibodies it sets a different bar: assess clinical benefit after at least three months of monthly treatment, or at least six months for the quarterly ones. It also notes that people with a partial response should be told cumulative benefit may accrue over six to twelve months of continued use.

Read the oral one again, because the qualifying phrase is doing all the work.

The clock starts at the target dose

Many oral preventives are titrated: you start low to limit side effects and step up over several weeks. If it took six weeks to reach the target and you gave up at week ten feeling like you had given it two and a half months, you actually ran a four-week trial.

That is the single most common way a working drug gets discarded. The fix is a question at the point of prescribing rather than a realisation afterwards: what dose are we aiming for, and how fast do we get there? Write the answer down with a date next to it, and the eight weeks becomes something you can count from rather than estimate.

The second, quieter version of the same failure is a titration that stalls. The step-up was meant to continue at the next review, the review got moved, and four months later you are still on the starting dose with everyone assuming you are at target. Nobody is being careless; it just needs one person tracking it, and you are the only person in the arrangement guaranteed to be at every appointment.

The baseline you did not take

Here is the measurement problem underneath all of it.

At the end of the trial someone asks whether it helped. To answer, you need two numbers: migraine days per month before, and migraine days per month now. Almost nobody has the first one, so the comparison ends up being against a memory — and memory of pain is compressed, weighted toward the worst stretches, and quietly rewritten by whatever you have come to expect.

A baseline is not elaborate. One month of marks before you start: a tick per day for headache yes or no, a second mark for whether it was migrainous, and a note of days you took anything acute. If you are about to start a preventive next week, that month has already gone, so use whatever you have and say so — “about twelve, but I wasn’t counting” is honest and still better than nothing.

If you are reading this before the prescription rather than after, start the count today. It is the cheapest thing on this page and the only one you cannot do retrospectively. What to record in a headache diary covers the six fields worth keeping, and the free printable diary is a monthly grid laid out to produce exactly this figure.

What a response actually looks like

Preventives are not usually judged on attacks stopping. They are judged on a reduction in monthly migraine days, and the reduction that counts as meaningful is smaller than most people expect.

Sixteen days down to nine is a substantial response. From inside it, it can feel like failure — you are still having attacks, still cancelling things, and the drug has side effects you did not have before. The count is what tells you the difference between a drug that did nothing and a drug that removed a week of your month, and those two situations lead to opposite decisions.

This is also why “it doesn’t feel like it’s working” is worth converting into a number before acting on it. The feeling is real information about whether the trade is worth it to you. It is not, on its own, information about whether the drug had an effect.

The confound that ruins short trials

Attack frequency moves around on its own. A good fortnight happens for no reason, and so does a bad one.

Start a new preventive in an unusually bad month — which is when people start them, because that is what prompts the appointment — and the following month will often be better regardless of what you took. That is the pull of the average, not a drug effect, and it is why eight weeks at a proper dose exists as a threshold rather than two.

It cuts the other way too. A trial that begins in a quiet stretch can make a genuinely useful drug look like it changed nothing, because there was little to remove.

Both problems have the same defence: enough weeks, and a count on either side rather than an impression.

The other things to write down during a trial

Three, none of them onerous.

Side effects, with dates. Some settle in the first fortnight and some do not, and by week six you will not remember which week the fog was worst. Ask at the outset which side effects mean stop and which mean push through — without that answer you can neither abandon nor persist with any confidence.

Acute medication days per month. A preventive that leaves your attack count unchanged but halves how often you need something acute has still done something, and this is the number most likely to show it. It is also the number with its own threshold worth watching.

What you can do again. Days worked, plans kept. It is the least precise of the three and the one people care most about at the end.

Take all of it to the review rather than a verdict — what to bring to a neurologist appointment covers the one-page format, and the treatment history section is where a trial that failed becomes useful information rather than a wasted quarter. A drug abandoned at a low dose after two weeks and a drug that failed at full dose after four months point at entirely different next steps.

Nothing here is medical advice, and dose, duration and whether to continue are decisions for whoever prescribes for you. Stopping a preventive on your own initiative is its own risk, and the sensible use of a count is to bring it to that conversation rather than to settle it in advance. The rest of the tracking guides start from the same monthly record.