General covariate-adaptive randomization targeting unequal allocation ratio

In clinical trials, a well-designed randomization procedure should have the potential to achieve compound objectives. Unequal treatment allocation is frequently designed, in that it has ethical and economic advantages. In this paper, we develop a general covariate-adaptive randomization procedure that can target any desired allocation ratio. To ensure the consistency of ratio at different levels, covariates are adjusted according to overall ratio, and a weighted imbalance measure is constructed with three types of imbalance under desired allocation ratio. Moreover, theoretical results are obtained under easily satisfied conditions. Finally, we investigate the performance of the proposed design and validate theoretical results through simulations.