Migraine Prophylaxis: preventing you a headache

Migraine headaches are common.  They can be really disabling for patients and if they occur with any frequency they can make life miserable.  Fortunately there are many options out there for the longer term management and prophylaxis of migraine and other primary headaches.

The only problem for us is that there are so many, seemingly effective remedies – it can be a real headache deciding which to try with your patient.

A dive into the literature reveals a broad range of medications and classes of medications that may be suitable – but the data is difficult to get one’s head around.  There are many ways that headache prevention can be measured and weighed – making it hard to compare apples with apples.  And to make things worse – a lot of newer agents come with a hefty price tag if you are prescribing off-schedule…   so how to choose?

Well I have been diving into the literature and have tried to summarise the data around migraine prophylaxis as cleanly as I possible.  How?  The NNT and the NNH for these meds are a fair place to start – you can weigh the risks and benefit with your patient if you know the rough numbers.  And then it is a matter of individualising care to your patient’s other needs and medical problems.

So here we go a whirlwind tour through Migraine Prophylaxis !

A quick explanation of the statistics is required!  When researchers look into migraine prophylaxis interventions – they tend to use a slightly odd, but useful measure of “effect”.  Clearly if you wanted you could measure the number of headaches per month / year etc and have a scale of effect – but that is a bit messy.  So instead they use the arbitrary bar of “50% reduction in headache frequency”.  So when I tell you that the NNT for drug X is 7 – that means that you would have to treat 7 patients with that drug for whatever time period in order to reduce their headaches from 12 to 6, or 8 to 4 – or whatever you like.  Hope that makes sense.

Now usually when we look at drugs in medicine – they get NNTs of 10 or 20 – which most folk would regard as useful.  [NB: aspirin has an NNT of 42 for mortality in acute STEMI.  Its NN to harm is about 167.]

When we look at migraine prophylaxis we are talking about NNTs in the range of 3 – 10.  Now that may sound like these drugs are really great!  But recall we are using “50% reduction in headache frequency” as the end point.  That is a reasonably low hurdle to jump over.  If we were to ask for “complete prevention, absolute prophylaxis” then the NNT would be much higher- I am guessing in the 100s.

So who needs prophylaxis?

  • Most Neurologists would consider 2 – 3 severe / disabling headaches per month a reasonable threshold for considering prophylaxis
  • This may vary depending on the patient’s choice
  • Some patient’s will feel strongly about preventing any headaches due to the severity; interference with life, work, family etc.
  • This needs to be explored with each migraine sufferer  on a case-by-case basis.  That is what they pay us the big bucks to do!
  • Remember that this is an “optional” intervention – so do not be too precious with particular drugs – it is a fine balance between potential harms and benefits – very much in the realm of “shared-decision making” and being pragmatic.

Most of the agents used for abortive therapy / acute management of migraine run the risk of causing rebound if used more than 10 – 15 times a month.  So another way of looking at it would be to use prophylaxis as a means to limit the amount of “acute abortive care” required and therefore maintain its efficacy.

OK – so that is the boring biostats explained.  Lets look at the candidates available to you and your patient to choose from.  I have lifted this very useful table from the Med Journal of Australia 2008 – link is here – it is a really well written piece.

Migraine prophylaxis (Click for the summary table from the MJA 2008 paper)

The Players:

This is a really old agent dating from the 60s. There is not a lot of quality evidence – but it has been the benchmark and default drug for a lot of doctors for many years. It is the oldest 5HT antagonist – the main side effects are weight gain and drowsiness. It is cheap and we know it. Specifically it is touted to help more with vestibular migraines.

Propranolol is the 2nd oldest agent out there. Usually dosed at 20 mg BD (not great for a long term med). It has been well studied. A NNT of 4, and a NNH of 40. Dirt cheap Some benefit for the anxious, tremulous or hypertensives – but can cause exacerbation of asthma / COPD in some punters.

Old TCA well known to most GPs. Going out of favour in Psych but still widely used.  Has it share of anticholinergic side effects and we all worry about OD! There is consistent evidence that amitriptyline works over a long period.  Of course, if the patient has depression as a comorbid diagnosis – this is a good option. None of the SSRIs etc have been shown to help.  Generally the dose required is at the lower end of the spectrum

This is a tough one – these agents have been looked at and I find mixed signal when I read the papers.   Some say it works, but the Forest plot is inconsistent!  At best I would say that there are better / more efficacious agents out there.  Also there was a pretty strong harm signal – dizziness, somnolence and cognitive dysfunction all being  significantly more common with these drugs!  So I would imagine a reasonably high rate of drop out after a period taking these drugs.  Here’s the Cochrane review from 2013.

This is the drug I would take if I needed migraine prophylaxis.  Unfortunately, it is expensive and the PBS in Australia only covers it if you have failed on the other agents. But it was in the top performers in terms of NNT = 4.  Also it seems to be effective at lowish doses 50 – 100 mg – so less side effects potentially.  The NNH is quoted as 3 – 25, a big range! And it depends on the dose.  But one of the “harms” counted is weight loss.  So this is probably a welcome effect for many patients!  In fact this makes Topiramate the only agent that seems to not be associated with weight gain.  The other common SE was taste disturbance (maybe related??)  Biggest downside is the potential for birth defects – so you need to use with care in fertile women!

Another effective drug – with an NNT of 4. But the NNH a little harsher at 7 – 14. The major problem with Valproate is weight gain and the metabolic syndrome. There are whole websites devoted to this syndrome! Valproate 400 mg has been trialled head-head against topiramate 50 mg…. and Topiramate came out on top

Another epilepsy med that has been looked at. Not a lot of great data – small trials showed a NNT of 2 . Was effective – but had a NNH of just 2!  Nausea, vomiting, drowsiness, dizziness… not easy to sell as a long term solution to migraine I think.

There have been a few ACE inhibitors and ARBs tried – so I have chosen candesartan as it the most widely used in Australia for hypertension. In head to head against propranolol – it was equally efficacious and well tolerated. Clearly this is a viable option for a large group of patients who might get benefit from BP control or renal / diabetic disease.

Yes Botox! I am including this in my review as it is out there and used. Your patients will find it when they Google their options. In summary: there is not a lot of quality evidence. Some Neurologists believe it helps with some subtypes (“imploding”) of migraine. And it is sometimes used in patients where other options have failed with a modest 10 – 30 % rate of response. So it is certainly not first line! Maybe 3rd or 4th line! Of course, in patients with coexisting wrinkles or narcissism there maybe additional benefits (Jokes ;-))


OK – that is it for my quick review of the literature around migraine prophylaxis.  Would love to hear your experience or views on the data.  I stingily recommend reading the great MJA piece that is linked above.



  1. Pik Mukherji says:

    “Grew up” with propranolol and Ca antagonists without too much success. Am more enamored of amitriptyline bedtime for prophylaxis at the moment. Just added steroid at time of D/C for acute migraine and haven’t done occipital injection, though I’m looking to get experience.

    • Hi Pik
      Calcium channel blockers -- specifically verapamil have some low-quality evidence -- not stellar.
      My reading is that verapamil is used first line in patients with a prominent aura / visual symptoms in their prodrome.

      The Discharge Dexamaethasone dosing for recurrence is a reasonable short term preventative strategy -- I believe the NNT is around 10 for preventing recurrence.
      There is a review on this.

  2. Andy Neill of wrote the bestBET on dexamethasone for prevention of recurrence -- I helped him out with a lit check. The link is here:

  3. That’s great Casey. Which one do you recommend after looking at the literature? And more interesting question, which one do you actually use?

    I still find this issue currently has that old chestnut -- “If there are many ways to treat something it is likely that none of them are great”

  4. This is a great summary of the evidence!

    A couple of pedantic points from me:
    NNTs are highly context dependent so great care needs to be taken when comparing NNTs in different therapeutic settings (to avoid the numbers being misleading). For instance, an NNT of 50 to avoid death at a month (the asprin for major STEMI) I would argue to be an enormously beneficial effect. We need to remember that the event rate in the placebo group is only about 12% at a month, so aspirin at most can at most only benefit 1 in 8 individuals. That is, even if aspirin worked perfectly and prevented all deaths, the best possible NNT in that study setting is 8.

    The setting for migraine prophylaxis, however, is that these individuals by definition ARE going to have migraines. This was how these individuals were selected. A perfectly working drug (meets the outcome for all patients) would have an NNT of 1. Even an NNT of 2 (as compared to placebo) means that only half your patients are going to get benefit (as per the defined outcome) from the drug.

    My medication of choice would be propranolol. At the typical doses, it has a low side-effect burden, and the side-effects are predictable. If there is a problem, switch to something else!

    The ACE-I and ARBs are possible choices though are likely less effective. They might be worth a go given their low burden of side-effects, and how common comorbid hypertension is.


    • Rob Dickson says:

      Pedantic maybe but thanks for a stellar NNT interlude. Seems like a good way to compare the different drugs but if the choices meds with a NNT of 4 are only helping 25% of patients it says a lot about the uphill fight against migraines!

      Appreciate the education

  5. Christine says:

    I agree topiramate is highly efficacious, however it has distinct effects on cognition that are unbearable for many patients. I have a colleague (an RN) who’s migraines decreased markedly on topiramate therapy, but in her own words she became so “dumb” she couldn’t function in her job (she actually felt unsafe to practice). Have certainly had this complaint from patients as well.

  6. Great work Casey,

Speak Your Mind