Page - 84 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

member of staff that patients may have inadvertently silenced a low power warning thinking it was a pre-alarm, leading to the infusion pump stopping. DiCoT-CL helped to describe the artefact, physical layout and social interactions that were contributing to this issue. There were also factors at play in different layers of the sociotechnical system. The main issues were being experienced at the micro and meso layer on the ward. Nurses complained that it was the way the pumps were designed, and there was nothing that could be done about it at the macro layer. However, further investigation revealed that this was a design configuration issue that could be adjusted by the hospital, at a different place in the macro layer. The analysis showed misunderstandings about the determinant factors contributing to this issue, which had different downstream consequences for staff and patients. Haematology services have now moved to a different hospital, but it seemed clear, through application of distributed cognition theory, that the configuration of the pre-alarms and barrier nursing was not working for staff and nurses. However, the pumps could have been reconfigured and tested using staff and patient satisfaction scores; efficiencies could have also been gained in the reduction of alarms. Again, distributed cognition can lead to testable hypotheses. 4. Discussion Distributed cognition theory is relevant for the understanding of complex sociotechnical informatics. Cognitive concepts can be applied to sociotechnical systems to elucidate how information flows across artefacts, social networks, over different physical configurations and different spans of time. The theory had been around since the nineties, but more detailed support to facilitate its application has only been developed more recently. Many applications of distributed cognition theory provide description and explanation of how information flows in a sociotechnical system. However, it also lends itself to considering determinant factors for the success and failure of technology, and how the technology might fit (or not) in context. This can lead to design ideas and hypotheses that can be tested. For example, the use cases presented here show how distributed cognition can inform the design of PHIMed and in what circumstances patients might be most receptive to it; and how distributed cognition can be used to evaluate systems that have been already been deployed. Like all theories, distributed cognition has strengths and weaknesses. Distributed cognition encourages a level of description about a system or process that lends itself to developing design ideas, but it may not readily emphasise the role of individuals or emotions as it focuses on systems and more observable functional issues [19]. This potential limitation is partially dependent on how rigidly one applies the theory. For example, more inductive ethnographic techniques can include interesting phenomena that might not readily fit the theory. Similarly, DiCoT can be applied in more or less rigid ways. Those unfamiliar with DiCoT may like to treat its models and principles as a check list, but we recommend its use in a semi-structured way to enrich what can be seen in context but not be too limited by the theory. Even with semi-structured use, DiCoT adds structure that draws attention to certain features and away from others, which should be reflected on by those who use it. Halverson [19] says that it is not clear whether success from the theory is from the theory itself or its commitment to ethnographically collected data. However, at least in the PHIMed example above we have seen that distributed cognition concepts and constructs can inspire thoughts and questions before empirical work has begun (Table 1), which suggest value lies in the theory itself. Since the theory D.Furniss etal. /DistributedCognition:UnderstandingComplexSociotechnical Informatics84