UNCERTAINTY OF THE TIME OF FIRST SIGNIFICANCE IN RANDOM EFFECTS CUMULATIVE METAANALYSIS

We propose a process for evaluating quantities of clinical and statist ical interest in cumulative meta-analysis. We use Monte Carlo simulati on studies to assess the error variance of the time to significance in cumulative meta-analysis. For the specific cumulative meta-analyses t hat we simulated, the 95% confidence interval of the treatment effect, estimated by the random effects method at the time of earliest signif icance, appears to be approximately appropriate except when the hypoth esized treatment effect is near the null. The Monte Carlo approach tha t we used can also estimate the power of a cumulative meta-analysis wh en two treatments differ in efficacy. We illustrate these issues in th e context of five cumulative meta-analyses, each comparing two treatme nts for preventing mortality from myocardial infarction. These simulat ed meta-analyses demonstrate our main point, which is that the time of first significance, however parameterized, is itself a random variabl e with error variance. (C) Elsevier Science Inc.