A Case Study in Translating Medical Evidence into Mobile Decision Support



Background: Combination oral contraceptives are prescribed 10 million times a year in the United States alone. Prescribers have more than 70 oral-contraceptive choices, but individual women's medical conditions may make some contraceptives inappropriate or dangerous. Traditional sources of prescribing information are impractical at the point of care. Mobile apps are revolutionizing clinical workflow, but none have been deployed for oral-contraceptive selection. Objective/Purpose: to design of an evidence-based interactive tool for selecting oral contraceptives, as a model for dissemination of clinical practice guidelines to clinical settings via mobile applications. Methods: I indexed the practice guidelines for hormonal contraception from the World Health Organization and the American Congress of Obstetricians and Gynecologists, and expert recommendations from two contraceptive textbooks. I gathered the rules into IF-THEN use cases, annotated with evidence quality and prioritized by likelihood and importance, and classified contraceptive choices into specific recommendation groups. Results: The sources yielded 32 guidelines and 24 expert recommendations using a total of 73 data elements, mostly TRUE/FALSE choices or numeric values. Organizing these into 22 use cases evaluating one to eight data elements each, few enough to fit on a smartphone screen, permits the entire questionnaire to be navigated in 22 or fewer screens. Discussion/Conclusions: Handheld clinical applications can facilitate evidence-based medicine, and reduce medical errors, by moving the best available medical knowledge from laborintensive traditional resources to prescribers' fingertips. They may create privacy and data-security risks, but limiting apps to anonymous single-use queries obviates these. Potential enhancements include additional object-oriented treatment domains, automatically updating rules engines using the Guideline Elements Model, and allowing user communities to add rules.