Page - 5 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

Table 1. Characteristics of a useful predictive principle, examples and counter examples Characteristic Explanation Example predictive principle Counter example [explanation] Clear Worded so that the implications for system design, development or implementation are explicit Clinical systems that are problem focused, usable, and incorporate relevant technical standards will be well accepted by users High quality systems work best [too vague and imprecise to be useful] Predictive Applying a predictive theory correctly will result in a clinical information system that is superior in terms of usability, acceptability, effectiveness or cost effectiveness Clinical information systems that apply Schneiderman’s user interface principles will be better accepted by professional and public users The pricing of clinical information systems depends on the amount of business benefit they help organisations to realize [not predictive] Testable Can be readily tested for its relevance in predicting the usability, value or cost effectiveness of health informatics or clinical information systems User interface designs for clinical information systems that ignore portrait and landscape screen formats will be rejected by tablet computer users When people believe in a system, it will help them. [Not testable – and a self- fulfilling prophecy] Relevant Can be applied to the design, development or implementation of clinical information or clinical research systems Incorporating Michie’s behavior change taxonomy into the design of digital tools to influence health- related behaviours will make them more effective The development of infection control measures needs to focus on the source and vector of the infection and the nature of the pathogen [not relevant to design of clinical information systems] Generic Applicable across wide range of technologies, use cases, users, care settings and health systems Attention to usability and the balance of perceived benefits and costs, including time to use and privacy risks, will improve the engagement of health professionals with digital tools Use no more than 15 Cyrillic characters per prompt on ePrescribing app screens for use in Crete [too specific to be useful except in rare contexts] Enduring Not likely to be rapidly outmoded by changes in technology or clinical practice Data that is captured once in a neutral context and is accompanied by meta data is more likely to be reusable Avoid using batteries with less than 8 hours life in a wearable [will be obsolete once kinetic energy harvesting and thermoelectric generator technologies mature] 3. Testing the validity of a theory and link to evidence based health informatics One challenge with this new view on theory in health informatics as a predictive principle is that any self-appointed expert can formulate an apparently credible predictive principle that seems to comply with the five criteria listed above and then market it using a catchy acronym (eg. Include Technology When One Risks Knowledge Shrinkage, ITWORKS © ), resulting in a generation of clinical information systems that respect the new principle but are actually less usable and effective than those which ignore it. However, the fundamentals of evidence based informatics (EBHI) teach us that expert opinion and authority are not sufficient to provide valid theories, and that principles should be tested J.C.Wyatt /TheNeed forTheory to InformClinical InformationSystems 5