|
Abstract:
|
Sample size estimation is important in planning clinical trials. In cardiovascular clinical trials, time-to-event endpoints, such as major adverse cardiovascular endpoints (MACE) and endpoints of unstable angina, revascularization, and cardiovascular hospitalization are often used to evaluate the effectiveness of a new treatment and to plan the sample sizes. We used 24 NDA (new drug application) data submitted to the Division of Cardiovascular and Renal Products at FDA and evaluated in 1995 - 2015 to examine how the design elements for sample size planning are changing over time, e.g., hazard ratio (HR), event rates. The impact of the between-region heterogeneity in treatment effect on sample size planning was explored by simulation using both Bayesian and REML methods. The results suggest that the observed HR seems to be slightly decreasing and the postulated HR increases over years. The average incidence rate is constant for death/MI/stroke and decreasing for stroke/SEE. The sample size may be substantially underestimated if region heterogeneity in treatment effect is not taken into consideration, and the extent of impact is related to the variances between regions and HR.
|
