East Multi-Arm Multi-Stage (MAMS)

Multi-Arm Multi-Stage (MAMS) is an East module that offers the ability to design and monitor multi-arm multi-stage studies with options for early stopping, dose selection, and sample size re-estimation. It has two features:

  • The group sequential theory extended for multi-arm setup (Gao et al., 2014) has been implemented for normal endpoint.
  • The multi-arm two-stage design using p-value combination approach (Posch et al., 2005) has been implemented for both normal and binomial endpoints.

East MAMS guides users through the complexities of multiplicity adjustment in order to construct reliable trials with multiple arms and multiple stages. A crucial regulatory requirement for multi-arm trials is establishing strong control of the family-wise error rate. East provides a variety of multiplicity adjustment measures to confirm strong control, along with advanced calculations that demonstrate a trial’s statistical power.

Resources:

Watch - Multi-Arm Multi-Stage (MAMS) module in East with Pantelis Vlachos, Ph.D & Cyrus Mehta, Ph.D

Read -Design and monitoring of multi‐arm multi‐stage clinical trials

Features:

  • Multi-arm Multi-stage designs (MAMS)
    • available for normal and binomial endpoint trials: analytical design, simulations and interim monitoring
    • can compare up to six treatment arms in five stages
    • treatment selection at any interim stage; drop the loser; stopping for futility
    • different rules for treatment selection
  • Two-stage Treatment selection design using the p-value combination method.
    • available for normal, binomial, and survival endpoint trials: design by simulations and interim monitoring
    • can compare up to twelve treatment arms in two stages
    • well-known multiplicity adjusted methods available; “Inverse Normal” for p-value combination
    • treatment selection, drop the loser(s), stop for futility at stage 1
    • different rules for treatment selection; can also integrate R functions to define customized rules
    • adaptive sample size re-estimation; promising zone can be a continuous or step function
    • Multi-arm multi stage designs (combining p-values approach) for survival endpoint
    • Multi-arm multi stage designs (group sequential approach) for discrete endpoint