Selection Bias in Studies with Unequal Allocation

• Activity Number: 124 - Special Topics in Clinical Trial Design and Analysis
• Type: Contributed
• Date/Time: Monday, July 31, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Olga M Kuznetsova*
• Companies: Merck & Co., Inc.
• Keywords: unequal allocation ; randomization ; selection bias ; brick tunnel randomization ; directional strategy ; permuted block

Abstract

In an open-label single-center trial with restricted randomization the investigator might try to guess the next treatment assignment and allocate a patient best suited for the guessed treatment. This introduces the selection bias in the study results by making the treatment groups incomparable. Selection bias is well studied for equal allocation procedures but is less studied for unequal allocation. For 2-arm equal allocation selection bias depends on the overall proportion of correct guesses while for unequal allocation it depends on the sum of expected proportions of correct guesses in each arm. In this talk we will justify the directional guessing strategy for trials with unequal allocation. We will quantify the selection bias with unequal allocation Permuted Block Randomization (PBR) under the directional strategy and the strategy where the investigator biases the randomization only when the expected pay-off is sufficiently large. We will also describe the advantage of using the Brick Tunnel Randomization in a study with unequal allocation when the investigator relies on the biasing strategy with a pay-off threshold and expects the PBR to be used in the study.