A nomogram to predict pulmonary complications after gastrointestinal surgery: a retrospective study

We aimed at developing a nomogram able to predict postoperative pulmonary complications (PPC) after gastrointestinal surgery.

We retrospectively analyzed the clinical data of patients who underwent gastrointestinal surgery at Jiangnan University Affiliated Hospital from December 2017 to May 2022. Patients were randomly divided into training cohort and validation cohort at a 7:3 ratio. The training cohort is divided into PPC group and Non-PPC group. The Least Absolute Shrinkage and Selection Operator (LASSO) method and logistic regression were used to determine the independent risk factors. The identified risk factors were used to construct a nomogram model for predicting the risk of PPC after gastrointestinal surgery. The nomogram model was validated by the area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA).

A total of 563 patients were admitted. The incidence of PPC was 17.6% (99/563). In the training cohort, multiple logistic regression showed that age, hypertension, history of respiratory diseases, preoperative albumin, intraoperative blood loss, postoperative intensive care unit (ICU) time, postoperative arterial oxygen partial pressure (PaO2), and postoperative tracheal intubation time were identified as the influencing factors of PPC (P < 0.05). We constructed a nomogram model for predicting the PPC of the training cohort, with a C-index of 0.857 (95%CI 0.812-0.902). In the validation cohort, the C-index of the model is 0.936 (95%CI 0.890-0.982). The ROC curve of the training cohort is 0.875 (95%CI 0.832-0.918), similar with validation cohort 0.929 (0.876-0.982). The calibration curve indicates that the predicted results are correlated with the observed results.

The constructed nomogram model has certain predictive value, and can provide a scientific reference for predicting the occurrence of PPC after gastrointestinal surgery.