Conducting a randomized trial in rural and urban safety-net health centers: Added value of community-based participatory research

Background: Colorectal cancer (CRC) is the second most common cancer in the US. Despite evidence that screening reduces CRC incidence and mortality, screening rates are sub-optimal with disparities by race/ethnicity, income, and geography. Rural-urban differences in CRC screening are understudied even though approximately one-fifth of the US population lives in rural areas. This focus on urban populations limits the generalizability and dissemination potential of screening interventions. Methods: Using community-based participatory research (CBPR) principles, we designed a cluster-randomized trial, adaptable to a range of settings, including rural and urban health centers. We enrolled 483 participants across 11 health centers representing 2 separate networks. Both networks serve medically-underserved communities; however one is primarily rural and one primarily urban. Results: Our goal in this analysis is to describe baseline characteristics of participants and examine setting-level differences. CBPR was a critical for recruiting networks to the trial. Patient respondents were predominately female (61.3%), African-American (66.5%), and earned <$1200 per month (87.1%). The rural network sample was older; more likely to be female, white, disabled or retired, and have a higher income, but fewer years of education. Conclusions: Variation in the samples partly reflects the CBPR process and partly reflects inherent differences in the communities. This confirmed the importance of using CBPR when planning for eventual dissemination, as it enhanced our ability to work within diverse settings. These baseline findings indicate that using a uniform approach to implementing a trial or intervention across diverse settings might not be effective or efficient. Keywords: Colorectal cancer screening, Community-based participatory research, Health disparities, Medically underserved populations, Dissemination and implementation, Randomized tria