gms | German Medical Science

Randomization in clinical trials is traditionally performed using randomization lists or sealed envelopes. In large and multi-centre clinical trials a trial coordinator centre is responsible for patient registration, data collection, and randomisation. These tasks are usually accomplished by fax and manually filled forms via telephone.

The "Randomizer", a web-based application for randomisation of multi-centre clinical studies, was developed at the Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Austria. Using the web reduces communication delays and expenses, provides a worldwide 24-hour-service, reduces transcription errors, and supports better auditing due to comprehensive logging of each transaction. Randomisation methods supported by the system are permuted blocks, completely at random, biased coin, minimization, big stick, and urn model.

The method of Pocock and Simon for sequential treatment assignments has been implemented [Ref. 1]. With this method, it is necessary to specify a function of imbalance which measures the amount of variation in treatment assignments, and an aggregation function, which measures the total amount of imbalance. For the different randomisation methods supported the algorithm by Pocock and Simon has been properly parameterised. The performance of these methods has been thoroughly tested and will be presented.

"Randomizer" is very flexible and has a user-friendly interface. Any user can generate trials and nominate participants for this trial. Different roles with predefined rights are assigned for each participant, e.g. biometrician, investigator, or monitor. Each role can receive additional rights or rights can be taken away, also new roles can be generated. The mapping between roles and rights can be defined by the trial coordinator. Effective management of trials and randomization of patients is well supported by this web-based application.