Technologies for Automating Randomized Treatment Assignment in Clinical Trials
- PDF / 342,203 Bytes
- 5 Pages / 504 x 719.759 pts Page_size
- 39 Downloads / 222 Views
0092-8615/98
Copyright 0 1998 Drug Information Association Inc.
TECHNOLOGIES FOR AUTOMATING RANDOMIZED TREATMENT ASSIGNMENT IN CLINICAL TRIALS UWE HAAG,PHD Heidelberg University, Institute for Medical Biometry and Informatics, Heidelberg, Germany
Randomly assigning patients is considered the choice in clinical trials when comparing different treatments. Automatic tieamtent assignment systems have been developed in order to reduce costs, increase reliability, or to provide 24-hour service. Communication devices such as telephone, telefkx, and computers can be usedfor this purpose. Advantages and disadvantages of the diflerent technologies are discussed. It is demonstrated that the newly proposed central treatment assignment via standard computer networks is feasible, comfortable. and efficient. Key Words: Treatment assignment; Randomization; Patient registration; Automation
INTRODUCTION
imization’ method as an example (2). In such cases one would prefer to use a comVARIOUS METHODS OF randomized paputer. tient assignment have been developed. PoIt is now assumed that there is a study cock provides a practical guide to the most center in which all patients should be regiscommonly applied (1). On the one hand, a tered. From an organizational point of view basic random-based method should be suffiit is important to distinguish between centralcient to reduce bias in the comparison of ized and decentralized randomization methtreatment groups. On the other hand, there ods. An assignment method is referred to as are good reasons to ensure that important being decentralized if, once an eligible paprognostic factors are distributed homogetient is identified, the local researcher can neously between treatment groups. This can determine the assigned treatment without be achieved by introducing stratification into contacting the study center. Centralized asthe methods of patient assignment. Assignsignment, on the contrary, requires that the ment methods used in practice are well-destudy center is contacted to determine the fined simulations of random processes. This actual treatment group. Centralized assignis generally accepted as long as the next asment has three main advantages: signment is not predictable for the individual researcher. A method is called adaptive if the current assignment depends on the previous 1. It ensures that the patient is registered at the study center before he is assigned to a ones. Some methods require quite complex treatment. This makes the data monitoring calculations or algorithms. Consider the ‘minprocedures much easier, 2. The risk of inadvertent or intended assignment errors can be reduced dramatically. Reprint address: Uwe Haag, Heidelberg University, InThis issue will be discussed in more detail stitute for Medical Biometry and Informatics. Im Neuenheimer Feld 305. D-69120 Heidelberg, Germany. throughout the paper, and 7
Downloaded from dij.sagepub.com at UNIV OF GEORGIA on May 31, 2015
uwe Haag
8
3. Any randomization algorithm can be used, including global stratification over all
Data Loading...
