Magnesium for AF in the ED?

Looking at the website and my relatively infrequent posts over the past 6 months it makes me realise that sometimes we need to say NO at work, as something always gives, and its usually the thing you do for fun, or the thing you do for free….so given that Christmas trees have been up since September in some places, consider this my new years resolution….2019 shall be different. I will say No, I will say No….

This paper was found for me by one of our awesome SpRs as part of his QIP work up (thanks Gordon Fuller – I have no doubt future RCEM Prof) and it looks at the use of supplementary Magnesium Sulphate in the management of Atrial Fibrillation with a fast ventricular response (VR).

Title of the Paper: Low Dose Magnesium Sulfate versus High dose in the early management of rapid atrial fibrillation: randomised controlled double blind study.
Clinical Question: Does adding Magnesium to standard care for ‘fast AF’ improve rates of return to normal ventricular rate compared with placebo.
Population: Adults with AF with a ventricular response >120.
Intervention: Magnesium sulphate IV (there were two arms to intervention group ‘low’ dose at 4.5grams and ‘high’ dose at 9grams)
Comparitor: Placebo (in this case N/Saline)
Outcome: Reduction in ventricular rate to less than 90bpm, or a reduction in rate by 20%.

Patients were given IV magnesium in addition to any anti-arrhythmic drugs (clinician choice) if they had a clear onset time within the previous 48 hours.
The dosing strategy for magnesium on the surface appears ‘big’ to me.
‘Standard’ dosing in our place is 2 grams, for acute severe asthma, for poisoning with a long QTc, and by default that’s the dose I’ve used when treating ventricular arrhythmia (I’ve used Mag since being a cardiology SHO in the olden days, as an adjunct in VT, and in resistant arrhythmia, however will hold my hands up that this practice has been ‘evidence light’ and based on advice and teaching from my bosses at the time).
So is 4.5 grams really low dose? Is 9 grams super high and likely to just give you more side effects? Whats your experience? let me know below.

Some good things in this trial design were that all 3 trial meds were prepared to appear identical, they were produced by a pharmacist not involved in the trial and blinded to the allocation group. This pharmacist also undertook the randomisation via random number tables in blocks of 3. This isn’t the ‘distant randomisation’ gold standard of telephone or internet based randomisation schedule, especially in blocks of 3, it would be possible for a clinician to talk to their local pharmacist about the last patient who received a trial drugs and experienced some side effects, and this in turn could lead to inadvertent or open unblinding of the next trial drug. Which in turn, one could argue, may affect the clinician deciding if that HR that’s flickering up and down around 90, is actually less than 90 or not…….

Shall we talk about the elephant in the room? Well the elephant in the paper – and that big grey lump is….what defines ‘standard care’ for AF?
Whats your current approach?
Digoxin for elderly-frail patients?
Beta blockers for elderly less frail patients?
Flecanide or electricity or amiodarone for rhythm control. When do you go for rhythm control?
Rate control for young people or old people. what defines old? What defines rate controlled? Is it 90. Is it 110?
Is it the mean BPM, or perhaps the lowest you see on the monitor?

I guess the point I’m trying, in a ham-fisted way to make, is that the recipes that are used to manage what is, lets face it, one of the most frequent cardiac complaints in an ED are hugely variable. From department to department, and I’d hazard from consultant to consultant.
My personal preferences – Flecanide or electricity for rhythm control – in most patients where the AF is the primary and only problem to be dealt with, and if no significant concerns around structural heart disease and if they have awareness of the act they’re in fast Af (i.e. there is definite onset time). I tend to let the patient choose which they get – anecdotally electricity has yet to fail me in ED, flecanide has, although infrequently.
I tend to use beta blockers for the vast majority of my other patients.

What about the patients in this trial?

18 years plus with AF at >120bpm. Mean age was around 66 years, with around 60% f the patients in each arm being female.
The patients were recruited at one of 3 participating EDs in Tunisia between 2009 and 2014.
The most commonly used ‘heart drug’ in the trial? Digoxin. About 45-50% depending on which arm the patient was in (the drug choice was down to the treating clinician which is entirely pragmatic and reflects the real world as previously discussed).
The remaining 45-50% of patients received diltiazem or a beta blocker.
Give the primary outcome measure was VR at 4 hours I can understand that patients receiving the above agents are of interest, however, I cant understand why none of the patients in any arm were given a rhythm controlling agent/or a shock.

Some good design points are that they undertook a power calculation, estimating an effect of 50% in the placebo group, a 15% effect from the MgS04 and having an 80% power to detect that difference. Patients were analysed as randomised with an intention to treat analysis which is great, it helps maintain the confounders generated in the randomisation. Have a look at this awesome review of ITT from Sketchy EBM – they’re one of my go-to places for all things EBM.

What did they find?

They managed to recruit around 150 patients in each group, with as previously described a mean age of (66ish) and around 60% female patients.
They found an absolute difference in favour of Magnesium compared to placebo of 20.5%, and interestingly 15.8% when comparing high dose and placebo.
They were also surprisingly accurate with their estimation of effect with 43.6% of placebo patients meeting primary outcome (ventricular rate of 90 or 20% reduction) at 4 hours, and 64.2% in the low dose Mag group.
Interestingly the conversion to SR was reported too and ran at 6.7% in placebo and 12.1 and 7.8% in the low and high dose groups respectively.
The harms of flushing, hypotension and bradycardia were also recorded and were only significantly different clinically in terms of episodes of flushing (18patients in the high dose group, compared with 6 and 1 in the low and placebo groups).

So what does this mean?

I appreciate there are some shortcomings in the trial, largely around drug choice, the dosing of the ‘high’ dose group, a lack of true rhythm control strategy and the fact that Yorkshire isn’t in Tunisia and I would love to see this trial replicated in a UK setting (with less digoxin) BUT I do think this trial has the potential to influence my practice – perhaps those patients where I am giving a second dose of beta blocker as the first hasn’t reduced the rate significantly. I’ll go with ‘low dose’ as the number of harm events in this group was low, warning the patient of ‘flushing’ in a small number of patients in that group.
What do you think?
Please let me know in the comments section.

For another spin on this paper check out the REBEL-EM review here

Thanks for reading!
Keep fighting the good fight, winter is here.



LOMAGHI Trial: Magnesium Sulfate for Rapid Atrial Fibrillation?

Leave a Reply

Your email address will not be published. Required fields are marked *