Document Type


Degree Name



Dept. of Medical Informatics and Clinical Epidemiology


Oregon Health & Science University


Objective. This study set out to use evaluate how well risk information was presented within a mobile patient decision aid app, Mammopad, and to provide a detailed characterization of benefits and shortfalls in quantitative risk communication scenarios, including the Mammopad graphics, where they occur. Methods. Risk presentation quality was evaluated in three pieces: 1) A risk scenario question where participants answered a word problem with their quantitative estimate of the positive predictive value of screening mammograms for women in their forties (administered before and after using the decision app); 2) A thematic analysis of transcripts from interviews with a subsample of Mammopad participants about what they found valuable about quantitative risk information; and 3) A thematic qualitative analysis of transcripts from the same interviews about the interpretation of risk communication diagrams, including misperceptions/misinterpretations of data. Results. Estimates of positive predictive value of mammography by participants (n=71) were more accurate after using the aid than they guessed before using it, although that was partly due to confusion between different risk statistics presented in the decision aid. Twenty-one participants enrolled in the interview phase: twelve participated in an interview session within an hour of using Mammopad, and nine others were recalled weeks or months after first using Mammopad. Negative themes concerning numeric risk presentation were: 1) lack of gradations in perception of uncertainty based on numbers, 2) numbers are sometimes provided as explanation, instead of a tool for explaining, 3) skepticism about the value of numeric information in light of forgetting, and 4) confusion about different statistics. Positive themes were: 1) valuing grounding in real-world groups, 2) valuing a connection to medical research, and 3) valuing transparent enumeration of outcomes. Conclusions. The numerical risk graphics in Mammopad were well received and informative, although not memorable on a long-term basis. It may be helpful to administer Mammopad at multiple times during a patient’s forties to refresh her memory. The more complicated of the two risk graphics in Mammopad was less often completely read or comprehended by participants, but cognitive tools such as animation could ease the cognitive burden of comprehending this information and improve understanding.




School of Medicine

Available for download on Friday, March 09, 2018