Statistical Analysis

[摘要] For trials reporting time-to-event outcomes with hazard ratios(HRs), we compared the hypothesized effect size from the sample sizecalculation in the research protocol with the observed effect size in thepublished article to calculate the ratio of observed-to-expected HRs. AllHRs were standardized for a reduction in adverse events relative to thestandard control group, such that HRs less than 1 indicated a benefit toexperimental therapy. In this case, a ratio of observed-to-expected HRsgreater than1 indicates effect size overestimation. For trials that did notreport the observed HR, we calculated the HR as a ratio of the mediansurvival time between arms based on the assumption of exponentialsurvival and constant hazards.