Background
Headache is quite a tricky presentation in the ED. Whilst many of the patients we see have benign causes about 1:10 has a potentially life threatening diagnosis (tumour, bleed, infection etc.). We have some pretty good strategies to investigate those (well we do, and you should too), but then we are often left with the remaining 90% in whom we may or may not have a diagnosis. It seems that many clinicians have different approaches to this group. Some use dopamine antagonists which remain popular in some countries (not in my gaff), NSAIDs are widely used, paracetamol of course, and even steroids find their way into the treatment bundle. Some people use Codeine, but I don’t know why……, and of course there are those who think that Magnesium works for everything (it doesn’t) and so have started giving magnesium for these patients too.
I suspect this wild variation poitns to the fact that the evidence base is weak. I myself am guilty of this. I love a spheno-palatine block and use it frequently. My evidence – simply that 60% of the time it works every time * see references below . You can read/view more about those (very cool) blocks here.
There is a plausible biological rationale for magnesium and it is inexpensive and generally safe, but previous studies have been small and produced inconsistent results. Systematic reviews have been unable to make firm recommendations, largely because the available evidence has been underpowered and heterogeneous. So I was glad to see a well conducted RCT looking at ED patients with headache. The abstract is below, but as always please read the paper yourself and come to your own conclusions.
Abstract
The aim of this study was to determine whether intravenous MgSO4 offers additional analgesic benefit when administered simultaneously with paracetamol in emergency department patients presenting with acute nontraumatic headache.
Methods
In this randomized, double-blind, placebo-controlled clinical trial, adults presenting with acute nontraumatic headache received 1 g of oral paracetamol prior to randomization, then were assigned to receive either 2 g intravenous MgSO4 in 150 mL saline solution or placebo 150 mL saline solution alone, infused more than 30 minutes. The primary outcome was the treatment success, defined as a ≥30% reduction in the numerical rating s̶cale score at 30 minutes. Secondary outcomes were the need for rescue analgesia, patient satisfaction, and adverse events.
Results
We included 506 patients in MgSO4 group and 522 in placebo group. Success was more frequent in the MgSO4 (78.9 %) versus placebo group (65.1%) (difference 13.8%; 95% confidence interval [CI] 8 to 19); however, all timed numerical rating s̶cale differences were below the 1.3-point accepted threshold for clinical importance. In the MgSO4 group, rescue analgesia was required less frequently (7.1% versus 15.3%; difference -8.2%; 95% CI -12 to -4.3), patient satisfaction was greater (91.7% versus 85.1%; difference 6.6; 95% CI 2.7 to 10), and adverse events were more frequent (15.4% versus 11.1%; difference 4.3; 95% CI 0.1 to 8.4).
Conclusion
Adding intravenous MgSO4 to paracetamol achieved more frequent success in treating acute nontraumatic headache, but at levels below accepted thresholds for clinical importance. MgSO4 was associated with less frequent rescue analgesia and a modest increase in mild side effects.
What kind of study is this?
This is a multicentre, randomised, double-blind, placebo-controlled superiority trial in four university emergency departments in Tunisia, which is great to see. An RCT is the best way to test a therapy like this. Adults presenting with acute non-traumatic headache were randomised to receive either intravenous magnesium sulphate or placebo. All participants also received one gram of oral paracetamol before randomisation.
The methodology is appropriate, randomisation and allocation concealment were robust, outcome assessment was blinded and analysis was undertaken on an intention-to-treat basis. Overall this is a well-conducted pragmatic emergency medicine trial. Exactly the sort of thing we like to review at St Emlyn’s.
Tell me about the patients
More than one thousand patients (1028) were recruited, which I think makes this the largest emergency department trial of intravenous magnesium published to date. The patient groups are interesting. Past studies have often focused on migraine patients, but this was more wide ranging. Patients were enrolled with any acute non-traumatic headache. Around half were ultimately diagnosed with migraine, while the remainder had tension-type headaches or secondary causes such as viral illness and hypertension (whether hypertension is a cause of headache we can discuss later……, but I don’t think it does except in extreme circumstances).
They took headaches of up to 7 days duration, but excluded chronic headache patients (wise), trauma, pregnancy, those who had just taken OTC meds and a few other sensible exclusions.
This is a pragmatic approach and I’m OK with that. When we end up with patients with headache but no serious cause found I often see clinicians apply a benign label with little confidence about which one of the benign causes is the actual cause. Whilst I don’t think that’s great medicine I do get why people do it. Clinicians AND patients do love a label (diagnosis). Also, decisions about analgesia often precede a definitive/random diagnosis, and it is often more useful to know whether a treatment works across the range of patients we see than within a narrowly defined subgroup.
The study was set in Tunisia. I don’t know much about how EM is conducted in Tunisia, but it’s always good to learn about new places.
What did they do?
All patients got 1g paracetmol, and were then randomised to 2g Magnesium infusion or placebo.
What were the measured outcomes?
The primary outcome was the proportion of patients achieving a 30% reduction in pain score after thirty minutes. Secondary outcomes included rescue analgesia, patient satisfaction and adverse events. The authors also predefined a minimum clinically important difference of 1.3 points on the numerical pain scale (NRS pain scale – 11 point VAS). That decision is worth remembering when interpreting the results.
What are the main results?
The trial met its primary endpoint. Patients receiving magnesium were more likely to achieve a 30% reduction in pain at thirty minutes, required less rescue analgesia and reported higher levels of satisfaction. Adverse events were slightly more common, although these were generally mild.
The subgroup analyses showed a similar pattern across migraine, primary headache and secondary headache, suggesting that any treatment effect was not confined to a single diagnostic category. Taken together, these findings support the conclusion that magnesium has a measurable analgesic effect statistically, but whether that matters clinically is less clear.
What does this paper tell us?
At first reading the paper appears to provide a straightforward positive result. However, it’s worth digging into the details. Although more patients reached the predefined threshold for improvement (78.9% vs. 65.1%, Diff. 13.8 95% CI 8-19), the average difference in pain score between the groups at thirty minutes was only 0.43 points on a 10 point scale. That’s arguable very little, difference, and partly explained by a slightly higher baseline pain score in the magnesium group.
There isn’t a contradiction here. Binary outcomes and continuous outcomes answer different questions, and both are entirely valid. However, that can change what we think of the results. If we focus only on the primary endpoint, magnesium appears beneficial as more patients hit the 30% reduction in reduce pain scores, but they also reduced in the placebo group, such that the difference between the two approaches at 30 mins was minimal. So at 30 mins out and from an EM perspective I don’t think the difference is clinically important. The minimal difference in pain scores extended out to 90 mins. That said in 2 important subgroups, those with migraine and those with secondary headaches the results were better and in study participants with secondary headaches, the NNT was an impressive 3.2 (95% CI: 2.3, 5.1). So perhaps diagnosis does matter?
The apparent contradiction between binary and continuous data outcomes is probably due to some patients having a very significant impact on pain scores (interestingly I think we saw similar findings in some of the asthma/magnesium trials in the past). If that is the case and subgroups are the reason why we see this discrepancy then there may be work to be done to find out who is most likely to benefit (migraine and secondary in this study – but hypothesis generating).
Friedman’s accompanying editorial discusses this issue well and also raises the question of whether making an accurate headache diagnosis matters before acute treatment is started? The investigators deliberately enrolled patients with undifferentiated non-traumatic headache and found a consistent treatment effect across the major diagnostic groups. As he points out, it appears to be more effective in some groups, so maybe we can be selective (though this study did not test this directly).
So, I’m not convinced this means diagnosis is unimportant. It may simply reflect the fact that patients presenting to the emergency department with severe headache have more in common than the classifications suggest, at least when our immediate goal is symptom control. Either way, I think this aspect of the study deserves as much attention as the magnesium results themselves.
Should we start giving out magnesium to headache patients?
I don’t think so no. This study does not justify replacing established first-line therapies. The marginal difference in pain scores may be statistically interesting but I’m, no convinced that the differences found are clinically important enough to change routine practice. Even though the treatment appears safe, inexpensive and associated with a modest improvement in patient-centred outcomes. That said I might incoporate it in some patients who may be difficult to treat or who fail initial therapy. So, I don’t think this paper supports routine magnesium for every patient with headache. The observed benefit is real, but modest, and clinicians will inevitably judge that balance differently as clinicians and for different patients.
This is an well conducted pragmatic emergency medicine trial and it strengthens the evidence that intravenous magnesium can be used for acute non-traumatic headache, but whether it’s an important enough difference to change practice remains to be seen. The study also illustrates an important lesson in interpreting clinical trials. A statistically significant primary outcome does not automatically translate into a clinically important difference, and understanding both the size of the treatment effect and the way outcomes are reported remains essential.
The bottom line?
Use magnesium as an adjunct if you want to. It may make a small difference, but probably not for every headache patient you see.
Hopefully, we can learn more about which patients may particularly benefit from magnesium in future studies.
vb
S
References and more
- Spheno-palatine block: https://www.youtube.com/watch?v=-Ez7VCj_kgk
- Yaakoubi H, Jaballah R, Boukadida L, Zorgati A, Ben Soltane H, Mezgar Z, Khrouf M, Dlala I, Bettout S, Sghaier A, Jerbi N, Chaka A, Grissa MH, Boubaker H, Boukef R, Msolli MA, Nouira S. Intravenous Magnesium Sulfate for Acute Nontraumatic Headache in the Emergency Department. Ann Emerg Med. 2026 May 14:S0196-0644(26)00201-5. doi: 10.1016/j.annemergmed.2026.03.029. Epub ahead of print. PMID: 42138680.
- Friedman B. Treating Headache Without a Diagnosis: Lessons From Intravenous MagnesiumAnnals of Emergency Medicine, 2026
- Anchorman: 60% of the time…… https://youtu.be/IKiSPUc2Jck?si=qMECkUb75hEqcAHk&t=38
