Skip navigation

Making Sense Of Migraine

Treatments?


New triptans are becoming available for the treatment of acute migraine headaches. How effective are these at relieving acute pain due to migraine? Is one better than the others? Are there alternatives? What should we do?

All these questions and others will, of course, be answered when the National Institute of Clinical Excellence starts cranking out the guidelines and guidance, and there will already be many local prescribing policies in use. But Bandolier wanted the answer now! One way of trying to find it was to do a 'Saturday night special' review, using common entry criteria and common endpoints to try and see whether NNTs were about the same or very different, and to try to assess the weight of evidence for each possibility.

The trigger for this was a superb meta-analysis of sumatriptan [1] by Peer Tfelt-Hansen of Copenhagen. This review (of which more below) is a source of knowledge and wisdom about the conduct of migraine trials. Because sumatriptan is a modern medicine it has been subjected to many clinical trials of high quality, which makes for fertile ground for asking lots of different questions.

Strategy


Bandolier knows it is in a privileged position with regard to finding information, so the strategy was to approach the search much like an average GP connected to the Internet might do. We were looking for placebo-controlled RCTs of analgesics used for the treatment of acute migraine. We searched our own in-house databases (filing cabinet), PubMed for the last few years looking for meta-analysis and migraine, and wrote to companies requesting published randomised controlled trials of products already marketed.

To be included in the analysis, a trial had to be a single dose, randomised, double blind and placebo controlled comparison of a treatment for an acute migraine attack. If it was a multiple dose study we had to be able to extract single dose data. We were not interested (at this stage) in head-to-head comparisons of one antimigraine compound versus an active control.

All patients conformed to the diagnosis of migraine by International Headache Society criteria. Baseline pain was of moderate to severe intensity. We extracted dichotomous data for the primary outcome measure of headache response ( a successful outcome) at two hours. This was defined as a reduction of pain from moderate or severe to mild or none.

Search Results


We found RCTs and meta-analyses of seven treatments (counting effervescent aspirin and metoclopramide as different from non-effervescent):

♦ Sumatriptan
♦ Naratriptan
♦ Rizatriptan
♦ Zolmitriptan
♦ Excedrin (paracetamol 500 mg + aspirin 500 mg + caffeine 130 mg)
♦ Aspirin 900 mg with metoclopramide 10 mg (including an effervescent preparation)
♦ Tolfenamic acid.

References to all the papers found are here . References to meta-analyses are given below.

Analysis


The analysis followed the path set out in the NNT insert . The outcomes were graphed onto L'Abbé plots, and the combined data for any dose of drug were used to calculate the number needed to treat.

Sumatriptan


The review by Tfelt-Hansen [1] pulled together information on nearly 7,500 patients entered into studies of subcutaneous, oral and intranasal sumatriptan. It looked at efficacy and adverse effects, with a prodigious searching strategy which included hand searches and interrogation of trialists and their industrial sponsors.

For 6 mg subcutaneous sumatriptan, data on 3,127 patients in 12 studies (Figure 1 upper panel) showed that 69% (1337/1927) of patients treated with subcutaneous sumatriptan had a successful result (decrease in headache severity from severe or moderate to none or mild) at one hour compared with 19% (226/1200) of those given placebo. This gave an NNT for success of 2.0 (95%CI 1.9 to 2.1) at one hour after injection.


For 100 mg oral sumatriptan, data on 2,890 patients in 12 studies (Figure 1 middle panel) showed that 58% (1067/1854) of patients treated with oral sumatriptan had a successful result at two hours compared with 25% (256/1036) of those given placebo. This gave an NNT of 3.0 (2.8 to 3.4).

For 20 mg intranasal sumatriptan, data on 1,420 patients in six studies (Figure 1 lower panel) showed that 61% (563/917) of patients treated with intranasal sumatriptan had a successful result at two hours compared with 30% (149/503) of those given placebo. This gave an NNT of 3.1 (2.7 to 3.8).

Naratriptan


Two studies with naratriptan were found, in which the main outcome was headache relief at four hours or later. Extrapolation from a graph on one of these (249 patients) was possible, and showed that 42% of patients treated with oral naratriptan had a successful result at two hours compared with 30% of those given placebo. This gave an NNT of 8.8 (95%CI 4.3 to no benefit).

Rizatriptan


Several studies on rizatriptan were obtained, most of which examined a variety of doses between 2.5 mg and 40 mg. Most information was available for the 5 and 10 mg doses.

For 5 mg oral rizatriptan, data on 954 patients in two studies showed that 58% of patients treated with 5 mg oral rizatriptan had a successful result at two hours compared with 32% of those given placebo. This gave an NNT for success of 3.8 (3.1 to 4.9) at two hours.

For 10 mg oral rizatriptan, data on 1143 patients in three studies showed that 64% of patients treated with 10 mg oral rizatriptan had a successful result at two hours compared with 29% of those given placebo. This gave an NNT for success of 2.9 (2.5 to 3.5) at two hours.

Zolmitriptan


A number studies on zolmitriptan were obtained, most of which examined a variety of doses between 1 mg and 25 mg. Most information was available for the 2.5 and 5 mg doses, and most came from a single study.

For 2.5 mg oral zolmitriptan, data on 651 patients in two studies showed that 64% of patients treated with 2.5 mg oral zolmitriptan had a successful result at two hours compared with 35% of those given placebo. This gave an NNT for success of 3.5 (2.7 to 4.7) at two hours.

For 5 mg oral zolmitriptan, data on 407 patients in two studies showed that 67% of patients treated with 5 mg oral zolmitriptan had a successful result at two hours compared with 31% of those given placebo. This gave an NNT for success of 2.8 (2.2 to 3.9) at two hours.

For 10 mg oral zolmitriptan, data on 369 patients in one study showed that 66% of patients treated with 10 mg oral zolmitriptan had a successful result at two hours compared with 34% of those given placebo. This gave an NNT for success of 3.1 (2.4 to 4.6) at two hours.

Excedrin (paracetamol 500 mg + aspirin 500 mg + caffeine 130 mg)


A review of three large studies [2] give results on a US over-the-counter migraine product called Excedrin Extra-Strength, consisting of paracetamol, aspirin and caffeine. This may not be available in the UK, but it is roughly equivalent to one paracetamol plus one aspirin taken with a strong cup of coffee (given that nausea is not too much of a problem).

Combining three similar studies with 1,220 patients, they found that 59% of patients had a successful result with Excedrin compared with 33% of those with placebo. The NNT was 3.9 (3.2 to 4.9).

Aspirin 900 mg with metoclopramide 10 mg


Three studies (including one with effervescent aspirin - Migravess) studied 765 patients. Combined they showed that 62% of patients had a successful result with aspirin/metoclopramide compared with 31% in those with placebo. The NNT was 3.3 (2.7 to 4.2).

Tolfenamic acid


A single study with 84 patients compared tolfenamic acid rapid release 200 mg. In the placebo group 29% of patients had a success at two hours, compared with 77% with tolfenamic acid. This equated to an NNT of 2.1 (1.5 to 3.5).

Summary


Table 1 gives the number of trials and patients in each comparison, and NNT for each drug and dose analysed. Where there were at least two studies or at least 400 patients studied a league table of NNTs (Figure 2) shows the relative efficacy of the oral antimigraine analgesics for successful treatment at one or two hours.
Table 1: Summary of randomised comparisons with placebo for pain relief at one or two hours (no pain or mild pain only)
Drug Route Time (hr) Dose (mg) Number of studies Number of patients NNT (95%CI)
Sumatriptan Subcutaneous 1 6 12 3,127 2.0 (1.9 to 2.1)
Sumatriptan Oral 2 100 12 2,890 3.0 (2.8 to 3.4)
Sumatriptan Intranasal 2 20 6 1,420 3.1 (2.7 to 3.8)
Naratriptan Oral 2 2.5 1 249 8.8 (4.3 to no benefit)
Rizatriptan Oral 2 5 2 954 3.8 (3.1 to 4.9)
Rizatriptan Oral 2 10 3 1,143 2.9 (2.5 to 3.5)
Zolmitriptan Oral 2 2.5 2 651 3.5 (2.7 to 4.7)
Zolmitriptan Oral 2 5 2 407 2.8 (2.2 to 3.9)
Zolmitriptan Oral 2 10 1 369 3.1 (2.4 to 4.6)
Excedrin Oral 2 500/500/130 3 1,220 3.9 (3.2 to 4.9)
Aspirin/metoclopramide Oral 2 900/10 3 765 3.3 (2.7 to 4.2)
Tolfenamic acid Oral 2 200 1 84 2.1 (1.5 to 3.5)

Comment


All this means that there is a lot of information out there on pharmacological treatments of acute migraine attacks. The great mass of evidence, on 7,500 patients, is for sumatriptan, and relatively little yet on the newer triptans. For some common treatments Bandolier could find no comparable data - reflecting perhaps the fact that some have been around for a long time. A paper with much wisdom [3] comments upon unpublished meta-analyses for some triptans.

Effectiveness is not the whole story, and adverse effects are often well reported. Tfelt-Hansen [1] analyses these for sumatriptan, but a much more detailed analysis of the event rates of individual adverse effects with the numbers needed to harm could be done.

Clearly this is a fertile area for more analysis, especially as a hatful of new triptans could be coming our way. Peter Goadsby's thoughtful paper [3] makes the case for the need for more treatments, for more analysis of those treatments, and for decisions of how to treat. Should we use a stepped approach (analgesic to ergots to triptans), or a stratified care model in which we analyse the severity of the problem and move immediately to the most appropriate care?

Goadsby also reminds us that 'the natural history of a migraine attack is to stop. The use of four-hour time points has led to flatteringly high headache response data, which translate into unrealistic expectations for clinicians and disappointment for patients.' The ideal end-point is the proportion of patients pain free at two-hours. So few studies give this end-point that it is sidelined for now, especially for comparison with older studies.

Other issues


As well as stopping acute headache and adverse effects of medicines that do that, there are many other issues not touched on here. These include headache recurrence, and the whole issue of prophylactic measures for dealing with migraine.

Summarising information like this is no more than a snapshot. There is probably loads more information on newer treatments just waiting to be published, and as well as thanking pharmaceutical companies for sending information on their products, Bandolier invites them to send more information as it becomes available.

Then there is the issue of relative efficacy from indirect comparisons with placebo. Perhaps the gold standard is one enormous randomised trial with all the treatments examined head-to-head, but that's just baying for the moon. The point is that the information presents itself not like a boxing competition, with products knocking each other out, but like a sprint world record. As long as conditions are fair (same distance, no following wind), then the time the runner takes is the time the runner takes.

References:

  1. P Tfelt-Hansen. Efficacy and adverse events of subcutaneous, oral, and intranasal sumatriptan used for migraine treatment. A systematic review based on number needed to treat. Cephalgia 1998 18: 532-8.
  2. RB Lipton, WF Stewart, RE Ryan et al. Efficacy and safety of acetaminophen, aspirin and caffeine in alleviating migraine headache pain. Archives of Neurology 1998 55: 210-17.
  3. PJ Goadsby. A triptan too far? Journal of Neurology, Neurosurgery, and Psychiatry 1998 64: 143-7.



previous or next story in this issue