Distributed cognitive resources: The analysis of human interaction with a complex clinical information system
Health information technology (HIT) has been designed in part to expedite clinical work and to reduce the frequency of preventable medical errors. However, after several decades of development and implementation, HIT is yet to reach its full potential. Interface usability problems and difficulties related to human-computer interaction (HCI) have been identified among the key barriers to high-quality performance. This work presents a multifaceted cognitive methodology, the Distributed Cognitive Resources (DCR) framework, for the characterization of cognitive demands of complex clinical information systems. The research was informed by the theory of distributed cognition, a novel approach to HCI that describes the characteristics of user interfaces that introduce extraneous cognitive complexity into the interaction. The new framework integrates elements of Norman's Theory of Action and dual-task theory of cognitive processing with distributed cognition, and characterizes the relative distribution of cognitive resources active in the interaction along two explanatory axes: external-internal, and system-medical. Methods of evaluation include several modified techniques from HCI and cognitive science research such as the cognitive walkthrough and think-aloud interaction protocols. The DCR framework was used to explain variation in user performance and to characterize the relationship of resource distribution and ordering errors in a study of computer-based clinical ordering. Specifically, the analysis suggested that the configuration of cognitive resources embodied in a clinical ordering application placed extraneous cognitive demands on users, especially on those who lacked a robust conceptual model of the system. The DCR framework also provided insight into interactive strategies employed by clinicians and patterns of associated errors. Results supported the claims that human cognition is routinely distributed across technology and artifacts, that user performance is related to the relative proportion of internal and external resources and affects the rate of errors, and that systems of greater interface complexity that provide few external resources impose significant cognitive demand on users and necessitate longer training. The DCR framework and associated analytical methods can suggest better interface design alternatives to software developers and identify crucial aspects of user training.