Drug Information Journal - March 2009 - (Page 136) 136 MEDICAL INFORMATION Cohen essary data capture modules. Instead, if a tool were used to create a semantically modeled protocol and the protocol tool and EDC tool could communicate with each other, no human interpretation by the data domain expert or anyone else would be needed to set up the EDC system. It would be necessary only to transfer (or clone) the structured protocol data into the EDC system. We use the term “extensible protocol” to describe such semantically modeled protocols, owing to their intrinsic ability to extend the utility of information created for the purpose of preparing a protocol document beyond the protocol document itself. Practically, the automated transfer of protocol information downstream to other systems and documents is made possible when both an extensible-protocol authoring tool and a downstream information system or documentauthoring tool speak the same structured language. The most likely candidate for a universally spoken structured language is XML (extensible markup language—a general purpose specification for creating structured language) operating within rules imposed by a universally accepted data standard. An open-source data standard for clinical research has been promulgated and supported by the Clinical Data Interchange Standards Consortium (CDISC) organization. The specific CDISC standard governing the interchange of protocol data among information systems is known as ODM (operational data model) (10). Conceptually, ODM can be considered an XML representation of an annotated case report form (CRF), which itself is simply a blank CRF that documents the location of the data with the corresponding names of the data sets and the names of variables included in them (1 1,12). Note that the above-described benefit of extensible protocols to interchange data with downstream systems is theoretical only in the sense that the magnitude of net benefit that can be achieved is unknown when such an interchange is performed routinely on the enterprise scale. Proof-of-concept demonstrations of this interchange have been performed publicly at several CDISC-organized events, whereby a sin- gle extensible protocol authoring tool transferred data via an ODM-formatted XML file to several EDC system vendors simultaneously, and these data populated relevant data capture screens. In addition to greatly reducing the impact of or perhaps eliminating the information interpretation step, routine use of extensible protocol tools could also improve knowledge transfer at all other steps of the information reuse cycle, as described below. At information creation (step 1), structured information, characterized by its explicitly defined conceptual relationships, can be subjected to logical data checks that can greatly reduce operational ambiguities (7). For instance, a computer could check whether each study objective has a corresponding outcome measure and associated activities in the schedule of events. Such checks could, in large part, supplant human quality assurance of protocol content. At the storage step (step 2), the definitive instance of a concept can be stored within the working version of a protocol model at all times, ensuring that authors are always working with the definitive version. Furthermore, variably restrictive controls on content can be instituted (eg, locking content by concept area), further reducing leaks at this step. For the first time, it also becomes practical for large organizations to determine the state of protocols-in-process directly (concomitant with the storage of active versions), enabling portfolio tracking at the study planning stage. At the retrieval step (step 3), use of semantically modeled content provides an opportunity to create semantic knowledge repositories, distinguished from typical document or content repositories by enhanced search and retrieval and analysis functionality. With enhancements made possible by the use of extensible protocols and related content, information separated by time, space, and knowledge domain can be logically associated, searched, and retrieved by users in ways that are intuitive. In the past, such functionality has required the manual population of document-metadata fields to implement.
For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.