Seite - 82 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

Use case 2: Safety around infusion devices on a haematology ward This study sought to evaluate safety around infusion devices on a haematology ward [18]. Using DiCoT-CL we explored determinant factors in the safety of infusions on the ward at the micro, meso and macro layers of the sociotechnical system. We present this as an example of how distributed cognition can be used in the late stages of the project lifecycle, post-implementation, to discover interactions that impact the performance of technology in practice. We spent 120 hours on the ward, shadowing and interviewing nurses, and observing infusion preparation and administration on the ward. DiCoT-CL was used for data gathering and evaluation; i.e., we attended to the five models, three different layers and the distributed cognition principles. Using DiCoT provided leverage to investigate the complex sociotechnical informatics that comprises infusion practice on the ward. We found that infusion safety was influenced by artefacts, social networks, the flow of information, the physical layout of the ward and interactions over time. Issues with design of infusion pump alarms, medication storage, prescription information, and hospital bed management systems were noted. Safety is affected by the co-evolution of structure (e.g. the design of pump alarms), agency and workarounds (e.g. telling some patients how to silence their alarms), and deviations (e.g. patients silencing the wrong alarms) [18]. 3. Explanation of success or failure of health information systems This section describes determinant factors in the success or failure for the above mentioned use cases and how distributed cognition theory has supported the investigation and explanation of such factors. 3.1. Use case 1: Preliminary results in what triggers the use of PHIMed DiCoT helped us start to think about the sociotechnical system that PHIMed may be embedded within. For example, early empirical feedback suggests that not all patients use PHIMed (i.e. they have different information requirements) but where it is used it can be recognised as an important artefact for supporting information flow around patient care. Before empirical work began we were already thinking about what contextual factors might be a ‘behavioural trigger’ to use of PHIMed (see application of DiCoT principles in Table 1). One conjecture was that patients who do not use PHIMed may perceive their general practitioner and/or medical record as an effective ‘information hub’ for their care (Figure 2.a), and that patients who do use PHIMed experience a more fragmented healthcare service that does not have an effective information hub. In this latter scenario the patient realises that they are the common feature in different consultations and act more like an ‘information hub’ in the absence of an effective hub on the healthcare side (Figure 2.b). Figure 2 shows two contrasting social networks inspired by the social model of DiCoT, which lead to different thoughts about what factors determine when PHIMed might be used and in what circumstances PHIMed might be most useful. These different patterns of distributed cognition are important for reflecting on who to target, when and in what circumstances to improve adoption and use in a future PHIMed intervention. We are continuing to test these and other conjectures in focus groups and interviews with healthcare professionals, patients and carers. D.Furniss etal. /DistributedCognition:UnderstandingComplexSociotechnical Informatics82