Intravenous streptokinase for myocardial infarction: a reanalysis of the historical cumulative evidence using Trial Sequential Analysis

IntroductionCumulative meta-analysis of intravenous streptokinase for myocardial infarction has been used as a text-book example to show how the megatrials GISSI and ISIS-II were redundant and wasteful. We decided to reanalyse the dataset with Trial Sequential Analysis to account for statistical heterogeneity and the risk of bias of the historical trials to reassess whether GISSI and ISIS-II were justified or redundant. MethodsWe extracted data from overviews published in 1982 and 1985 and trial reports on mortality outcomes. For the five largest trials conducted before GISSI and ISIS-II, we also extracted information on the used comparator, randomisation, blinding, dropout proportions, and the use of intention-to-treat analyses. We did random-effects cumulative meta-analyses with Trial Sequential Analysis considering diversity. ResultsThe largest trials conducted before GISSI and ISIS-II had serious methodological differences and high risks of bias making a cumulative meta-analysis invalid by todays standards of evidence synthesis. The Trial Sequential Analysis showed that the monitoring boundary for a mortality benefit of streptokinase was reached during the ISAM trial. However, both GISSI and ISIS-II were launched before the ISAM trial was published. Focusing only on the cumulative assessment, the megatrials were potentially futile. Sensitivity analyses corroborated these results. ConclusionOur Trial Sequential Analysis of the historical dataset of streptokinase for myocardial infarction found that conclusive evidence favouring streptokinase was established after the megatrials were launched. However, considering the methodological differences and risks of bias, such cumulative meta-analysis seems invalid. Accordingly, the megatrials were not wasteful.