Beyond Binary ICU Admission: Duration-Sensitive Evaluation and Patient-Level Reclassification of Textbook Outcome After Colorectal Cancer Surgery.

Background/

Objectives

Textbook outcome (TO) is an all-or-none composite in which a common, practice-sensitive component may exert disproportionate influence on patient classification. We evaluated the classification impact of binary postoperative intensive care unit (ICU) admission and the additional information retained by actual ICU duration after colorectal cancer surgery.

Methods

This retrospective single-center cohort included 241 patients undergoing colorectal cancer resection between 2020 and 2026. Comprehensive TO required no major postoperative complication, reoperation, 30-day mortality, or postoperative ICU admission, together with microscopically margin-negative (R0) resection of the colorectal primary and postoperative length of stay ≤14 days. Core TO omitted ICU admission and length of stay and was used as an analytical comparator rather than as a newly validated TO definition. Patient-level reclassification was the primary outcome. ICU duration was evaluated continuously in the principal concurrent-association analysis; the 0-, 1-, 2-, and ≥3-day categories were exploratory.

Results

Comprehensive and core TO were achieved in 27 (11.2%) and 169 (70.1%) patients, respectively, resulting in reclassification of 142 patients (58.9%; exact McNemar p < 0.001). ICU admission alone accounted for reclassification in 118 patients (49.0% of the cohort). Each additional ICU day showed a concurrent association with higher odds of core TO failure (adjusted odds ratio (OR), 1.43; 95% confidence interval (CI), 1.25-1.63) and major postoperative complications (adjusted OR, 1.63; 95% CI, 1.40-1.91; both p < 0.001). Categorical estimates for 1- and 2-day stays were imprecise and did not establish equivalence with no ICU admission. Duration-sensitive representations showed better internal model fit, discrimination, and probabilistic performance than binary ICU admission.

Conclusions

In this center with frequent postoperative ICU utilization, binary ICU admission dominated comprehensive TO classification. ICU duration retained additional clinical information, but the findings do not establish the intrinsic invalidity of the ICU component, equivalence of short stays with no ICU admission, or a universal duration cutoff. Multicenter external validation across different ICU policies is required.