Internet-based assessment tools for predicting colorectal cancer risk: How do they communicate individualized risk to the young adult?

Abstract

72 Background: Online prediction tools for cancer risk can inform individuals about their risk and potentially encourage adoption of risk-modifying health behavior. The Colorectal Cancer Risk Assessment Tool (CCRAT) is an interactive tool developed by the National Cancer Institute for estimating colorectal cancer (CRC) risk for individuals between the ages of 45 and 85 without high risk predisposing conditions. Given the rising incidence of CRC among young adults under age 50, we aimed to investigate the performance of CCRAT in estimating their CRC risk. Methods: An institutional protocol prospectively enrolled 563 patients newly diagnosed with non-hereditary CRC and administered a health behavior questionnaire at baseline. Self-reported demographics, diet and physical activity, medication use, and family history were extracted and entered into the CCRAT ( https://ccrisktool.cancer.gov/calculator.html ) to calculate their predicted 5-year and lifetime CRC risks. Health behaviors and the predicted CRC risks in reference to the population average were compared among three groups: youngest- (aged 18-44, N = 276, 49%), young- (aged 45-50, N = 178, 32%), and older-onset (aged > 50, N = 109; 19%). Results: The patient groups did not significantly differ in gender or in family history of CRC, but the youngest group had significantly more non-White patients (27, 22, vs. 19%; p = 0.008) and distal tumors (80, 76 and 65%; p = 0.05). The youngest and young groups had a higher prevalence of morbid obesity (18 and 15% vs. 12%; p = 0.002) and active smokers (6 and 8% vs. 3%; p = 0.005). There were no significant differences in vegetable intake, moderate exercise, NSAID use, or female hormone use. The 5-year CRC risk was correctly predicted as “higher than average population” for a significantly smaller proportion of the youngest (42%) and young (45%) patients, when compared to their older counterpart (71%; p < 0.001). Similarly, the tool communicated “higher than average” lifetime risk for CRC in only 56 and 60% of the youngest and young patients, but 72% of the older ones (p = 0.021). Conclusions: The existing risk prediction tool inadequately communicated CRC risk for adults younger than 50 years of age. Risk factors underlying young-onset CRC likely differ from and extend beyond those as currently assessed. New tools capturing age-specific risk factors are needed to accurately communicate individualized risk and potentially motivate risk-modifying lifestyles.