Page - 40 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

information system will yield information, or that the information will lead to a decision change. For example, the number of times a clinician reads a patient record is always going to be greater than the number of times that reading leads to a change in decision. Similarly, not every computer generated alert will result in a change in decision. The number of times a decision is changed is also going to be greater than the number of times any such change leads to a measureable improvement in patient care. Additionally the value of events early on in the value chain will often be lower than for events later on. For example, optimizing user interaction with medication alerts is likely to be of much lesser value than reducing the number of unsafe medication prescriptions, which in turn is of lesser value than reducing the number of adverse outcomes from medication errors. Similarly, the time saved in optimizing a user interaction with an EHR is likely to be of lesser value than improvements to the way tests are ordered, and these are often of lesser value than patient outcome changes such as improved survival or QALYs based on more appropriate investigation of patients. This typical increase in value of events as we move down the chain is driven by increased value associated with real world health outcome changes compared to the value of improvements in process alone. It is however quite possible that in some settings that it is the early stages in the chain that are of higher value. For example, if human resources are scarce and expensive, then using information tools to optimize human efficiency and effectiveness might have very great value. Figure 2: The number of events is typically higher earlier in the value chain, whilst the value of individual events tends to be higher further down the chain. Combining event frequency (or probability) with event value (or utility) provides the expected utility at each point in the chain (from Coiera, 2015). Recall that by combining event frequency (or probability) with event utility, we arrive at an expected utility. We can thus calculate the expected utility of using a given system along the different steps in the value chain. The resulting value profile of expected utility will not necessarily be constant across the different steps. For example, a telecare system may be designed to maximize expected utility at the interaction stage by reducing face-to-face interactions, but with no expectation of changing clinical outcomes. A decision support system would be designed specifically to improve decision- making and outcomes, while an EHR is typically designed to improve record keeping, E.Coiera /AssessingTechnologySuccessandFailureUsing InformationValueChainTheory40