Abstract:
|
Treatment comparisons in randomized clinical trials usually involve multiple endpoints. One common approach is to apply Bonferroni correction by multiplying P values with number of endpoints. However, the Bonferroni correction can be somewhat conservative if multiple test statistics are positively correlated. An alternative way is to use O'Brien global test by assigning weight to test statistics. O'Brien global test has been widely discussed and applied to many types of endpoints, such as all continuous endpoints, binary and survival endpoints. To the best of our knowledge, few articles discussed global test in presence of continuous and survival endpoint. We extended O'Brien global test to include both survival and continuous endpoints, and conducted simulation studies to compare with conventional methods.
|