[Randomization for controlled therapeutic trials].

The choice of an adequate randomization procedure, the organization of the randomization process and its correct implementation are questions that may influence the quality of a controlled therapeutic trial. Starting with a brief critical review of randomization procedures now available, this paper presents in greater detail one method which is thought to be optimal for clinic trials of typical sample sizes. The "adaptive randomization" procedure of Pocock and Simon [3] keeps all 2-8 factors of prognostic importance in satisfactory balance-as has been confirmed by trials in the past 4 years. An interactive computer program, RANDOM, is introduced, whereby easy and safe use of the procedure is enabled via a video terminal. Furthermore, standardized recording of the randomization process by a clerk for medical documentation in the central randomization office is described. The possibility of randomizing multi-factorial trials and the admissibility of eliminating already randomized patients from a trial are discussed. The working group on documentation and biometry of the 1st Department of Surgery, Vienna University Medical School, offers the program to any centres that may be interested and is also willing to undertake randomizations (by phone) for clinical trials within Austria.