Name: Lan

Firstname: Gordon

Title: Distinguished Scientist

Institution: Pfizer Inc. MS6025-A5148

Street: 50 Pequot Avenue

City: New London

Zip-Code: CT 06320

Country: USA

Phone: 860-732-1983

Fax: 860-732-7036

Email: gordon_lan@groton.pfizer.com

Authors: Gordon Lan

Title: Repeated significance tests in adaptive sequential designs

Abstract: We consider the control of the alpha level under the following situations: 1.In a sequentially designed clinical trial, if the frequency of monitoring is data-driven, then the alpha level cannot be preserved exactly. We considered conditions and situations where this effect is minimal. 2.The measure used to compare treatments may involve multiple endpoints and require a complex definition of superiority. However, the control of the alpha level usually applies only to the primary endpoint with a specific test statistic. Consequently, some people have been hesitant to document a sequential boundary in the protocol. Moreover, it is not clear what to do if the boundary was crossed but the Data Monitoring Board did not want to stop the study. We consider the conditions under which the sequential boundary may be overruled along with other conditions where overruling may inflate the alpha level. 3.Repeated significance tests when the sample size is re-estimated.

References: 1.Lan and DeMets, "Changing frequency of interim analysis in sequential monitoring", Biometrics 1989, pp1017-1020. 2.Lan and Wittes, "Data monitoring in complex clinical trials", JSPI 1994, pp241-255.