Does randomization matter in rare disease clinical trials

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background Conducting Rare Disease Clinical Trials poses a wide variety of research challenges. Many techniques that are usually considered the best standard for evaluating new therapies are more difficult to implement here. This also holds in terms of randomization in clinical trials. The question arises how to randomize and, especially, which randomization procedure (RP) should be used to prevent the influence of bias focussing on allocation bias as efficiently as possible. Methods We present guidance that supports users in the selection and implementation of an RP based on the ERDO framework, which is a template for the evaluation of RPs for design optimization. For this purpose, we use the freely available R-package randomizeR, which is an easy-to-use tool to evaluate different RPs for various study designs while quantifying different types of bias. The evaluation is based on different metrics, such as the mean type-1 error probability or the go-no-go criterion, and brings these together with indirect evaluation criteria such as predictability and the number of correct guesses. Results For the evaluation of RPs and the comparison of RPs with each other, both the presented indirect and direct criteria should be considered together. Examining the criteria individually might be misleading and may lead to different decisions regarding the RP. Different study designs may result in different assessments, so there is no universal solution or choice for RPs. Conclusion Randomization is and will continue to be a key methodology for rare disease clinical trials. Quantification of bias in particular with respect to the test decision is an essential issue, and the use of the ERDO framework should be a routine practice in the design of a trial.