I really, really, wanted this study to work. Few things are more disheartening on an overnight observation shift than needing to place someone on a diltiazem drip after an inability to rate control. Ergo, I have been known to give a few grams of magnesium to try to decrease the likelihood of that happening. Therefore, on the surface, this study seems promising – it looks at standard of care for Atrial fibrillation with rapid ventricular response (dealer’s choice, metoprolol, diltiazem, or digoxin) given in combination with either one of 3 treatments: placebo, 4.5g of magnesium, or 9g of magnesium; with about 150 patients in each arm. This study took 5 years (!) over 3 academic centers in Tunisia, who’s ED’s service 90,000-110,000 patients per year. Patients needed to have a heart rate over 120 bpm, have a systolic BP >90 mmHg, and without: renal impairment, wide-complex tachycardia, decompensated CHF, acute myocardial infarction, or an impaired level of consciousness. All seems fair.
The results, at face value, seem great if you’re a magnesium believer: rate control at 24h of 83.3% for placebo, 97.9% for 4.5g MgSO4, and 94.1% for 9g of MgSO4. This is a great example of completely reading a paper before you start to fight about giving magnesium.
First, all groups used digoxin around 50% of the time for rate control. This clearly does not mimic US practice. Nor does giving 4.5-9g of magnesium over 30 minutes. Then the authors sneak this one in:
“In a secondary analysis including only patients receiving beta blockers and calcium channel blockers, the obtained results were not significantly different compared to those found in the overall group.”
This is sandwiched between mentions of adverse drug reactions (4% flushing in the 4.5g arm vs 12% in the 9g arm vs <1% in the placebo arm, and otherwise there was no significant difference between the 3 arms), and the discussion of 24h rate control. I am not 100% certain what they meant by this statement – were they referring to ADRs? Were they implying that there was no difference between metoprolol and diltiazem treated patients and placebo at 24 hours? With only about 50% of patients per arm (~75 patients in total/arm) being treated with these agents, it would be hard to show a meaningful improvement. Not to mention the fact that the actual data for this secondary analysis is nowhere to be found in this paper. Nor have the authors responded to my email asking for it.
Then, of course, there are the prior trials with less than 60 patients / arm comparing diltiazem to metoprolol showing >90% efficacy with diltiazem.
And, of course, there is the next question, of are we doing any good? Since rate control has not always shown to be in the patients best interest – a 6 fold higher rate of adverse events– and none of the ED AFib RVR magnesium studies look further out than 24 hours, perhaps we should cautiously, if at all, recommend magnesium, or even suggest waiting until long term outcomes are further elucidated. Since this study took 5 years to complete, I do not see the desired study happening anytime soon.