A nomogram for predicting 3-month mortality in colorectal cancer patients requiring perioperative ICU admission: a retrospective cohort study.

backgroundColorectal cancer patients requiring perioperative intensive care unit (ICU) admission represent a high-risk subgroup with substantial short-term mortality. This study aimed to develop and internally validate a nomogram for estimating 3-month all-cause mortality in colorectal cancer patients requiring perioperative ICU care.

methodsThis single-center retrospective cohort study screened patients hospitalized for colorectal cancer surgery between January 2014 and December 2018. The final cohort included 415 patients requiring perioperative ICU admission. The primary endpoint was all-cause mortality within 3 months after ICU admission. The intended prediction time point was the end of the first 24 h after ICU admission. Candidate predictors were obtained from the preoperative period, the index surgery, and the first 24 h after ICU admission. Multivariable logistic regression was used to identify independent predictors and construct the nomogram. Model performance was evaluated using receiver operating characteristic curve analysis, bootstrap internal validation, calibration analysis, and decision curve analysis.

resultsAmong 415 included patients, 58 (14.0%) died within 3 months after ICU admission. The final nomogram incorporated preoperative albumin, APACHE II score, CRRT within the first 24 h after ICU admission, mechanical ventilation within the first 24 h after ICU admission, and laparoscopic surgery. Higher albumin level and laparoscopic surgery were associated with lower mortality risk, whereas higher APACHE II score, CRRT, and mechanical ventilation were associated with increased mortality risk. The nomogram showed good discrimination, with an AUC of 0.882 (95% CI, 0.829-0.927) and bootstrap internal validation yielded an optimism-corrected C-index of 0.872. Calibration analysis showed good agreement between predicted and observed risk, and decision curve analysis suggested potential clinical utility across clinically relevant threshold probabilities. The nomogram showed better discrimination than APACHE II alone.

conclusionsWe developed and internally validated a nomogram for estimating 3-month mortality at the end of the first 24 h after ICU admission in colorectal cancer patients requiring perioperative ICU care. The model may support early risk stratification in this high-risk population, but external validation is required before routine clinical implementation.