Random assignment in clinical trials: Issues in planning (Infant Health and Development Program)

Various options available for the randomization of subjects into groups in a clinical trial are discussed, emphasizing the issues of logistics given less focus in more mathematical treatments. We discuss advantages and disadvantages of total randomization, of Zelen-type randomization procedures, of Efron-type procedures vs more classical blocking procedures to control the balance between groups, and of Simon-Pocock-type procedures vs more classical stratification for controlling possible biases in prognostic factors. Finally, we discuss issues related to choice and implementation of randomization procedures. The discussion is illustrated with the processes of decision-making in a national collaborative randomized clinical trial, the Infant Health and Development Program.