Magnesium in acute myocardial infarctionBMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6995.1669c (Published 24 June 1995) Cite this as: BMJ 1995;310:1669
- Kent L Woods,
- D B Barnett
- Reader in therapeutics Professor of clinical pharmacology Department of Medicine and Therapeutics, University of Leicester, Leicester LE2 7LX
EDITOR,—Salim Yusuf and Marcus Flather1 offer no satisfactory explanation for the statistical incompatibility (P>0.0001 for heterogeneity) between the inefficacy of magnesium in the fourth international study of infarct survival (ISIS-4)2 and the clearly positive pooled results of the 10 other trials of intravenous magnesium in acute myocardial infarction. Magnesium has been reported to protect contractile function in a dozen experimental models of myocardial ischaemia and reperfusion and is now known to act during the first few minutes of reperfusion.3 4 ISIS-4 was unique in that most patients underwent thrombolysis before being randomised to magnesium or control (to accommodate the factorial design). The interval was not recorded but most typically have been three or more hours after randomisation took place at a mean of seven hours from the onset of symptoms. The 30% of patients in ISIS-4 who were not given thrombolytic treatment were randomised at a mean of 12 hours after the onset of symptoms, when spontaneous reperfusion of viable myocardium is unlikely to occur.
ISIS-4 was not designed to test for an antireperfusion effect of magnesium. Any subgroup analysis put forward to disprove such an effect requires (a) evidence that patients in the subgroup were likely to have experienced reperfusion in the presence of a raised magnesium concentration and (b) a power estimate for the analysis.
The second Leicester intravenous magnesium intervention trial in 2316 patients (LIMIT-2) was a placebo controlled trial to test rigorously the findings of the earlier magnesium trials using mortality end points, intention to treat analysis, and an explicit power calculation. Trial treatment was started immediately on admission at a median of three hours from the onset of symptoms. Thrombolysis was also given to 48% of all patients with confirmed acute myocardial infarction in the study, beginning immediately after the loading injection of trial treatment. The figure shows an update of mortality data already reported,5 with >99% complete follow up for a mean of 4.5 years and 529 deaths in the analysis.
The lesson to be learnt is that mechanistic insight has an essential role in the planning and interpretation of trials. Even a megatrial cannot test a hypothesis precluded by its design. The clinical message is that a patient admitted with suspected acute myocardial infarction should receive intravenous magnesium immediately and for as long as there is a likelihood of reperfusion (either induced by thrombolysis or spontaneous) taking place. This safe and simple intervention will reduce the risk of death from ischaemic heart disease over the next several years by 21% (95% confidence interval 5% to 35%).5