A model to control data flow in multicenter clinical trials.

A model for the data management procedures of multicenter clinical trials is proposed. Focus will be restricted to those trials in which data are collected over time depending on therapy and follow-up schedules. The requirements for the model are structured data collection forms and a theoretical data flow governing the submission of materials by the participating institutions. The existence of a coordinating center is assumed, but the model presented here can be applied irrespective of the organizational set up of the research environment. It is suggested that each data collection form (or any other piece of documentation) is seen as an independent information holder, irrespective of the actual data it contains. The status of each form, from when it is expected by the coordinating center to when it is accepted for statistical analysis, can be summarized by a single variable, maintained by the data manager. Depending on its value, certain activities can be initiated, such as requesting missing forms or corrections. The implementation of the described model within a database system as well as its relation to the requirements for reliable statistical evaluation of the study are discussed with reference to a real example.