Risk factors and prediction of intensive care unit readmission after esophagectomy for cancer

Abstract OBJECTIVES Intensive care unit (ICU) readmission has been proposed as a metric for quality of surgical care. The current study investigated potential factors and developed a prediction model for ICU readmission in patients following oesophagectomy for cancer. METHODS A total of 3028 patients from 2019.1 to 2022.12 were retrospectively collated as training cohort, with 829 patients from 2023.1 to 2023.8 enrolled for validation, respectively. Univariable and multivariable analysis were performed to identify potential factors after which a nomogram based on results from multivariable analysis was constructed and validated. RESULTS In training cohort, the rate of ICU readmission was 3.6% (110/3028). Readmitted patients were associated with more reoperations, higher 90-day mortality and prolonged postoperative stay (all P < 0.001). Multivariable analysis demonstrated that older age ≥ 75 years, neoadjuvant therapy, preoperative albuminaemia, diffusing lung capacity for carbon monoxide (DLCO)%, longer operative duration and retention of endotracheal intubation when entering ICU were independently associated with ICU readmission. Based on these results, a nomogram for predicting readmission was constructed and validated. The Hosmer–Lemeshow test showed the model in training cohort was well calibrated (𝑥2=5.259, P = 0.73) and area under the receiver operating characteristic curve (AUC) was 0.739 (95% confidence interval [CI]: 0.691–0.787). Moreover, application of the nomogram in validation cohort showed an improved AUC of 0.780 (95% CI: 0.703–0.857). CONCLUSIONS ICU readmission after oesophagectomy although uncommon (3.6%) was associated with prolonged hospitalization and significant mortality. A nomogram based on six variables may assist intensivists to early identify patients at high risk of readmission.