Page - 115 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

1.5. Medical information provision through patient decision aids In order to make well-informed medical decisions, the patient and the healthcare provider need access to medical information about the options to which a patient is eligible, including risk information. $ 8 A classic example of how the boosting framework may help to make better decisions and involves boosts that help to better understand medical risk information. Generally, patients as well as healthcare providers have difficulty understanding conditional probabilities [e.g., 21]. For instance, consider the conditional probabilities for breast cancer. Let’s assume the base-rate (prevalence) to get breast cancer is one out of 100 women. The accuracy of a mammogram, an X-ray test to indicate whether a person has breast cancer, is about 80-90%. More specifically, the probability of the mammogram resulting in a positive test result when breast cancer is present (sensitivity) is 80%. The probability of the test result of the mammogram being negative when the disease is absent (specificity) is 90%. Now, a woman is tested positively on the X-ray test, what is the chance this woman has breast cancer? In other words, what is the positive predictive value of the mammogram, what is the probability that a patient has the disease when the test result is positive? Both healthcare providers and patients typically overestimate this chance and judge it to be around 75%, whereas the actual chance is much lower: It is only 7-8%. This is because people tend to neglect the base-rate. Gigerenzer and colleagues indicated that we could make these risks much more understandable for healthcare providers and patients by using natural frequencies rather than conditional probabilities. Risk information about breast cancer would then be explained in the following way: Out of 1000 women, 10 women will have breast cancer and 990 will not. Out of those 10 women who do have breast cancer, 9 will receive a positive result on the X-ray test and 1 will not (false negative). Out of the 990 women who do not have breast cancer, 99 will receive a positive result (false positive) and 891 women will receive a negative result. This way, it is more transparent to see the role of base-rates: many women without cancer are in fact tested positively. This approach to presenting risk information can be considered a boost, because by presenting risk information in terms of frequency information, the understanding of the information increases and therefore potentially the quality of decisions based on this information increases as well. In a similar way, illustrations, animations and videos can serve as boosts. Illustrations, in particular those supporting a text, are widely used to facilitate learning of information by improving comprehension and recall [22]. Adding videos to online texts, particularly personalized videos using a conversational narration style, also improves memory for medical information [22], and animations can even bridge the information processing gap between audiences with low and high health literacy [23]. M.deVriesetal. /FosteringSharedDecisionMakingwithHealth Informatics Interventions 115