Layperson Views about the Design and Evaluation of Decision Aids: A Public Deliberation

Abstract

Purpose: We carried out the first public deliberation to elicit lay input regarding guidelines for the design and evaluation of decision aids, focusing on the example of colorectal ("colon") cancer screening. Methods: A random, demographically stratified sample of 28 laypeople convened for 4 days, during which they were informed about key issues regarding colon cancer, screening tests, risk communication, and decision aids. Participants then deliberated in small and large group sessions about the following: 1) What information should be included in all decision aids for colon screening? 2) What risk information should be in a decision aid and how should risk information be presented? 3) What makes a screening decision a good one (reasonable or legitimate)? 4) What makes a decision aid and the advice it provides trustworthy? With the help of a trained facilitator, the deliberants formulated recommendations, and a vote was held on each to identify support and alternative views. Results: Twenty-one recommendations ("deliberative conclusions") were strongly supported. Some conclusions matched current recommendations, such as that decision aids should be available for use with and without providers present (conclusions 1-4) and should support informed choice (conclusion 9). Some conclusions differed from current recommendations, at least in emphasis-for example, that decision aids should disclose cost of screening (conclusion 11) and should be kept simple and understandable (conclusion 14). Deliberants recommended that decision aids should disclose the baseline risk of getting colon cancer (conclusions 15, 17). Limitations: Single location and medical decision. Conclusions: Guidelines for design of decision aids should consider putting a greater focus on disclosing cost and keeping decision aids simple, and they possibly should recommend disclosing less extensive amounts of quantitative information than currently recommended.