Medwise: development and evaluation of a highly user-configurable ('web 2.0') electronic health record platform

A primary problem in the acceptance and use of electronic health records has been the perception and experience that they often do not reflect clinician concepts and priorities and consequently do not meet user needs. This has resulted in slow uptake of electronic health records, and frequently, their underuse or outright rejection. At the same time in the public internet, new approaches (‘web 2.0’) have greatly increased ease of use, flexibility, and user participation in software and content creation and design, as well as sharing of user-created resources, with far-reaching effects. This dissertation describes the creation of a system, MedWISE, which aims to give clinicians user far greater control of the EHR than current systems, including the ability to create and share information resources, and configure the (widget-based) interfaces, using these approaches. First, we formulated a new model of how these approaches might be used to create a healthcare adaptive architecture and address some of the current problems with healthcare information systems. Then, we applied these approaches to create a new electronic health record platform, MedWISE, which embodies part of this model. This was tested with a mixed methods approach including cognitive usability studies with typical clinical users at a major academic medical center. We also conducted logfile analyses of the current legacy system in order to prepopulate interfaces with appropriate combinations of clinical data. In addition, we conducted an experiment to evaluate whether MedWISE could be rapidly configured to provide decision support for initial H1N1 screening and treatment. The new approaches significantly decreased repetitious navigation compared to use of the legacy system for the same patient cases by clinicians in the same roles (p<.0001). They also facilitated clinician mental processes through several mechanisms, and showed promise for facilitating communication and collaboration. The Keyhole effect, a term describing the problem in human computer interaction that volumes of information must be accessed through a very small window, was also reduced. We demonstrated the technical feasibility of developing an EHR platform with this approach. Subjects learned and used the system to review real and complex clinical cases in a very short period of time, rated ease of use and usefulness highly, and predicted time savings and other advantages. In the logfile analysis study, we were successful in identifying suitable clusters of laboratory tests to pre-populate interfaces. The system was successfully configured for H1N1 clinical support. A concern is the possibility of inducing errors given the great latitude users can have to configure their displays. Further work in controlled deployment settings and laboratory studies are necessary to establish the parameters under which these approaches may be used. The system can serve as a research platform for exploring extensible and configurable EHR applications and for examining clinical cognition, communication, and technology design and adoption.