Uncovering the Red Flags: A Cross-Sectional Retrospective Case-Control Study on Predictors of Early Onset Colorectal Cancer in a Multi-State Community Health System

Abstract The increasing prevalence and aggressive, potentially fatal, nature of early onset colorectal cancer (EOCRC) makes it critical to identify risk factors that can facilitate earlier screening and detection. With electronic medical record data, we compared adults aged 20 to 49 years diagnosed with EOCRC (2017-2023) to control patients without EOCRC who were 1:2 exact matched based on age and sex. We extracted data on patients’ sociodemographics, comorbidities, hereditary syndromes, family history, symptoms, specialty visits, and medications within 12-24 months prior to or on their diagnostic visit. Using a model-building approach, we analyzed data with univariate then multivariate logistic regressions to predict the odds of EOCRC diagnosis. We constructed three models using high-risk background characteristics as predictors (Baseline and High-Risk) and exclusion criteria (Sporadic), respectively. Our final analysis included 684 EOCRC cases and 1,368 controls. The mean age was 42 years, with 53% male, 72% White, and 72% non-Hispanic/Latino. Across all models, several predictors were significantly associated with higher EOCRC odds, including alcohol use history, higher number of comorbidities, abdominal pain, rectal bleeding, constipation, iron deficiency anemia, and prescriptions for metformin, non-steroidal inflammatory drugs, and multivitamins. Significant predictors of lower EOCRC odds were employment and Medicare/Medicaid insurance. By concurrently including symptoms, medical history, and sociodemographic characteristics, we constructed and validated well-fitting models with good discrimination that replicated and extended prior case-control research. To facilitate earlier screening and detection, these EOCRC risk factors can be used to identify patients who would benefit from screening earlier than 45 years of age.