BACKGROUND: Atrial fibrillation (AF) is a common complication after cardiac surgery. Previous meta-analyses have concluded prophylactic magnesium (Mg) prevents postoperative AF, although with a significant degree of heterogeneity among included studies. Recently, the largest randomized, controlled trial published to date (but not included in prior published meta-analyses) concluded that Mg sulfate is not protective against AF after cardiac surgery. The objective of this study was to conduct a new meta-analysis including the results of new Mg trials not included in previous meta-analyses, and to investigate the potential causes and effects of significant heterogeneity observed in previously published meta-analyses. METHODS: The MEDLINE, EMBASE, and CENTRAL databases were searched for relevant studies published up to March 31, 2012. Pooled odds ratios of occurrence of AF were calculated using random-effects models. Heterogeneity was assessed as significant using the I(2) statistic. RESULTS: Egger's and funnel plots demonstrated biases toward stronger and more positive effects of Mg in smaller studies. When the analysis was restricted to the five double-blind, intention-to-treat studies in which AF was the primary outcome (Mg arm, n = 710; control arm, n = 713), Mg did not prevent postoperative AF (odds ratio, 0.94; p = 0.77), and heterogeneity was no longer significant (I(2) = 40%; p = 0.15). CONCLUSIONS: This meta-analysis, restricted to well-conducted trials, does not support the prophylactic use of Mg to prevent AF after cardiac surgery. Prior meta-analyses have drawn conclusions from simple random-effects models with significant heterogeneity. However, this approach leaves important residual heterogeneity and overemphasizes the strongly positive effects of smaller studies.
BACKGROUND:Atrial fibrillation (AF) is a common complication after cardiac surgery. Previous meta-analyses have concluded prophylactic magnesium (Mg) prevents postoperative AF, although with a significant degree of heterogeneity among included studies. Recently, the largest randomized, controlled trial published to date (but not included in prior published meta-analyses) concluded that Mg sulfate is not protective against AF after cardiac surgery. The objective of this study was to conduct a new meta-analysis including the results of new Mg trials not included in previous meta-analyses, and to investigate the potential causes and effects of significant heterogeneity observed in previously published meta-analyses. METHODS: The MEDLINE, EMBASE, and CENTRAL databases were searched for relevant studies published up to March 31, 2012. Pooled odds ratios of occurrence of AF were calculated using random-effects models. Heterogeneity was assessed as significant using the I(2) statistic. RESULTS: Egger's and funnel plots demonstrated biases toward stronger and more positive effects of Mg in smaller studies. When the analysis was restricted to the five double-blind, intention-to-treat studies in which AF was the primary outcome (Mg arm, n = 710; control arm, n = 713), Mg did not prevent postoperative AF (odds ratio, 0.94; p = 0.77), and heterogeneity was no longer significant (I(2) = 40%; p = 0.15). CONCLUSIONS: This meta-analysis, restricted to well-conducted trials, does not support the prophylactic use of Mg to prevent AF after cardiac surgery. Prior meta-analyses have drawn conclusions from simple random-effects models with significant heterogeneity. However, this approach leaves important residual heterogeneity and overemphasizes the strongly positive effects of smaller studies.