Measuring patient preferences for colorectal cancer screening using a choice-format survey.

OBJECTIVE: Colorectal cancer (CRC) screening uptake remains poor. Until we understand patient motivation and preferences for undertaking screening, it is unlikely the uptake will be optimal. Our objective is to examine patient preferences for CRC screening modalities and uptake rates using utility-based methods. METHODS: The preference survey was mailed to a random sample of Canadian subjects aged 40 to 60 years from a primary care network. A fractional factorial experimental design maximized D-efficiency and included four blocks with 12 choice tasks in a conditional two-step design, two-alternative discrete choice format with five screening attributes (process, pain, preparation, sensitivity, and specificity). Bivariate probit regression analysis was used to estimate patient preferences for attributes, choice probabilities for alternative modalities and expected rates of uptake. RESULTS: Five hundred forty-seven of 1047 surveys were returned. Almost 30% of respondents preferred no screening. The most preferred test attribute levels were noninvasive process (e.g., CT), no preparation, no pain, 100% specificity, and 90% sensitivity. Accuracy-related attributes were more important than test process-related attributes. Virtual colonoscopy was the most preferred, followed by colonoscopy, barium enema, sigmoidoscopy, and fecal DNA testing, based on simulated choice probability estimates. Fecal occult blood testing (FOBT) was least preferred. Adjusted screening uptake rate estimates showed the greatest impact (42% increase) would be achieved if all CRC screening modalities were available rather than FOBT alone. CONCLUSIONS: Our findings emphasize the important role of patient preferences for no screening and in selecting alternative CRC screening modalities. CRC screening implementation in Canada should consider patient preferences to optimize uptake.

Duke Scholars

Author F. Reed Johnson Population Health Sciences