Age-stratified risk profiles for emergency colorectal cancer resection: A machine-learning analysis

Emergency colorectal cancer resection (ECCR) is associated with worse perioperative and oncologic outcomes than elective surgery. As the incidence of colorectal cancer rises among younger adults and disparities persist across the life course, we examined how clinical and sociodemographic factors intersect to shape ECCR risk across age groups.

We conducted a retrospective cohort study of adult colorectal cancer resection admissions in the National Inpatient Sample between 2018 and 2022. Multivariable logistic regression and classification and regression tree (CART) analyzes were used to identify independent predictors of ECCR and age-specific high-risk phenotypes.

Among 510,135 resection admissions, 64,175 (12.6%) were ECCR. ECCR admissions more often involved patients with compromised health (weight loss, anemia, coagulopathy, and electrolyte disorders), metastatic disease, and ≥3 comorbidities across all ages. Among younger and middle-aged adults, ECCR disproportionately involved patients who were on Medicaid or who were uninsured and those from lower-income communities. In multivariable models, female sex and higher income were associated with protection across most age groups, whereas being on Medicaid or uninsured status, having a higher comorbidity burden, and admission during the COVID-19 pandemic era were more frequently associated with higher odds of ECCR. CART analysis identified age-specific phenotypes, with insurance status predominating in patients younger than 75 years, and clinical complexity and pandemic-era timing predominating in those aged 76 years or older.

ECCR risk reflects different dominant mechanisms across age groups, suggesting that a one-size-fits-all approach is unlikely to reduce emergency presentations. Age-tailored strategies that address access barriers and optimize management of medically complex older adults may improve timely, equitable colorectal cancer care.