MAGNAM Trial, Magnesium Versus Amiodarone in Atrial Fibrillation in Critical Care

This is a Multi-centre, randomised controlled, clinical trial that is comparing a Stepwise Strategy of Magnesium Followed by Digoxin, With an Amiodarone Backup vs a Strategy of First-line Amiodarone to See Which is More Effective in Returning the Heart to Normal Rhythm After Experiencing Rapid Atrial Fibrillation in General ICUs. It will take place in Critical Care Units in Toronto Health Sciences Centres over a course of 2 years. The sample size is 200 patients. Investigational Product and Planned Use Magnesium sulphate The trial intervention will be Magnesium sulphate followed by digoxin as second line therapy with Amiodarone as third line. The Standard of care intervention will be Amiodarone as first line treatment in the and then no more than 2g MgSO4 be delivered. Data will be collected retrospectively at the outcome timepoints. Statistical Analysis: This analysis will be conducted using the intention to treat principle, therefore all randomised patients will be included in the main analysis. Crossovers and protocol violations will remain in their original study group. Baseline data will be summarized per group for continuous variables using means and standard deviations or medians and interquartile ranges as indicated the distribution and for discrete variables using frequencies and percentages. For the rate / rhythm co-primary outcome the authors will use a sentinel time point analysis at the 6 and 24 hour time points assuming there no / very low competing risk for death using a multivariate regression model to test for differences between groups and adjusting for baseline variables such as age, hospital site, shock status (requirement for inotropes Y/N), mechanical ventilation (Y/N) and known chronic atrial fibrillation. For the ICU length of stay co-primary outcome the authors recognise the competing risk of death and for this outcome and the authors propose to use Fine and Gray models adjusting for the stratification variables (as above). Estimates will be presented as sub-distribution hazards and 95% confidence intervals. The authors will test for interaction between sub-group and treatment and present the estimates per sub-group. All secondary outcomes are binary and differences between groups will be tested using Chi square test or Fisher exact test as appropriate. Missing data will be uncommon for our key outcome data considering the nature of this data. The authors do not propose to use imputation for missing data in these primary or secondary analyses.