Seite - 45 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

3. How value chain analysis can assist in explaining health IT success or failure Value chain analysis has a role both in explaining what has already occurred, through retrospective evaluation, as well as in shaping the design of technology and the way it is embedded in the larger socio-technical system. It can be applied in a number of different circumstances, including: 1. Qualitative retrospective analyses. The overall evidence for the benefit of a specific technology is often patchy, and the choice of outcome measures for evaluations may not always be ideal. As we saw in Case Study 1, it is easy to pick intermediate process measures which give a false sense of success, just as it is easy to over- emphasize clinical outcomes when the real benefit of a technology is to optimize events earlier in the chain. Value chain analysis can provide a template to consider the real-world costs and benefits of a technology at different points in the chain, identifying gaps in knowledge about performance, as well as guiding the interpretation of success and failure [17]. 2. Quantitative retrospective analyses. When performance data are available for a specific system, as in Case Study 2, then value chain analysis can reveal specific problems in the design, implementation or use of a system. Event frequency data is ideally recorded automatically as part of system operation, and utility data can be obtained from system users, potentially even retrospectively. 3. Prospective quantitative studies: If a value chain can be provided with estimates of expected usage and benefit of an implemented technology, it can be used to provide predictions about overall system utilization and benefit. Such hypotheses can then be tested in prospective trials. 4. Technology design: Typically a digital service is built up of a bundle of separate elements. A decision support system bundle will actually require components that access the electronic health record, a user interface, and alerting strategy, and so on [18]. The overall performance of the bundle is thus dependent on the performance of individual components, and the dependencies between components. For example, if the electronic record component is suboptimal, then it does not matter how good the decision support engine might be, as the quality of recommendations will still be poor quality. System designers can estimate the necessary value profile for each element of a bundle, so that together the bundle performs as expected. 4. Discussion Value chain theory makes very few assumptions about the nature or purpose of technology, and so has broad applicability. The strongest assumption is that the purpose of technology is to improve specific decisions, and that there is a prospect that those decisions have a detectable outcome in the real world. By relying on standard tools such as probabilities and utilities, value chain analysis is strongly grounded upon well- accepted and proven analytic concepts and methods. One can consider a value chain to be the equivalent of a single path down a decision tree, but with some key differences. Most critically, in a decision tree we only calculate the utility of the final or terminal node. What is interesting about value chains compared to decision trees is that each node in a chain could be the terminal node, each with its own intrinsic and different utility in the world. One could stop a chain at reading an E.Coiera /AssessingTechnologySuccessandFailureUsing InformationValueChainTheory 45