Effect of Patient Portal Messaging Before Mailing Fecal Immunochemical Testing Kit on Colorectal Cancer Screening Rates: A Randomized Quality Improvement Trial

Colorectal cancer (CRC) is the second leading cause of death from cancers affecting both men and women in the United States. One in 17 Americans will suffer from CRC during his/her lifetime. Early detection by screening has been shown to reduce CRC mortality. Despite screening recommendations, the U.S. screening rate remains well below the national benchmark of 80% as established by the National Colorectal Cancer Roundtable. The U.S. Multi-Society Task Force (MSTF), however, does recommended FIT and colonoscopy as first line screening modalities for CRC in 2017. To help address the issue of suboptimal CRC screening, the investigators implemented an intervention utilizing the principles of behavioral economics to improve screening rates. Studies in behavioral economics and psychology indicate that how information or choice is framed impacts behavior in predictable ways, which has applications in health and medicine, including the design of CRC screening strategies. This project builds upon our institutions continued quality improvement efforts utilizing behavioral economics principles to improve CRC screening, through improving the choice architecture, framing and salience of information to incentivize routine screening.

For this project the investigators will leverage our electronic health records (EHR) patient portals to improve CRC screening. Specifically, the investigators developed an electronic primer within the EHR patient portal to alert patients due for CRC screening before arrival of a mailed FIT Kit. The investigators randomized implementation of the primer at the patient-level to determine whether the electronic primer improved CRC screening completion in patients enrolled in our mailed FIT program.

For our analysis, after summarizing our demographic data, the investigators plan to compare screening completion in the two study arms using an intention-to-treat analysis and t-tests. The investigators then plan a logistic regression and Cox proportional hazards model to compare time to screening utilization in the two study arms, controlling for age, sex, race, and ethnicity. Following this the investigators use Fisher's exact tests to compare completion of individual screening modalities in the two study arms. Lastly, the investigators plan a secondary, analysis to determine the impact of opening the portal message on screening utilization, using randomization arm as an instrumental variable. In this analysis the investigators compare the subset of patients in the intervention group that opened the portal primer message to the control group. P-values less than 0.05 are considered statistically significant.