Design and analysis of the ESVEM Trial

Inadequate understanding of the design and statistical approach of a clinical trial and the failure to recognize subjective aspects of the analysis often result in misinterpretation of trial results. This is exacerbated by the push to shorten publications and the wish for a simple message that summarizes the outcome of the trial. The purpose of this review is to critically review the design and statistical analyses of the results, to evaluate the assumptions underlying the statistical tests, and to examine the results of exploratory analysis on the interpretation of major findings of the Electrophysiologic Study Versus Electrocardiographic Monitoring (ESVEM) trial. The trial was unusual because its primary objective was to compare testing methods instead of treatments. This necessitated using a subset of the original randomized groups for sensible analysis of the clinical question. Nevertheless, the two groups appeared to be well balanced. The absence of a difference in outcome could be verified by several analyses. In addition, confidence intervals were narrow, indicating the high precision and reliability of the findings. However, the comparison of antiarrhythmic drugs is problematic because the trial was not designed to address this issue. There were differences in the distribution of clinical characteristics between the groups who received different antiarrhythmic drugs. Nevertheless, using both univariate analyses and a variety of adjustments for important prognostic variables, treatment with sotalol appeared to be a significant predictor of reduced arrhythmia recurrence, and sotalol was consistently associated with a trend for nearly a 50% reduction in sudden death and all-cause mortality as compared with the other drugs administered in the trial. In conclusion, the ESVEM trial raises a number of interesting and instructive issues about clinical trial design and analysis.