Abstract: Objective To compare different diaphragmatic ultrasound functional indices, diaphragmatic excursion during quiet breathing(DE-QB), diaphragmatic excursion during deep breathing(DE-DB), thickening fraction of diaphragm(TFdi), in predicting postoperative pulmonary complications(PPCs)after abdominal surgery. Methods A total of 154 patients scheduled for elective laparoscopic abdominal surgery under general anesthesia were selected. DE-QB, DE-DB, and TFdi were measured by bedside ultrasound before and 24 hours after surgery. Postoperative pulmonary complications were recorded for 7 days. Patients were divided into two groups according to the occurrence of pulmonary complications: the pulmonary complications group(PPCs group, n=48)and the non-pulmonary complications group(n-PPCs group, n=106). The differences in diaphragm function indices between the two groups were compared, and those with significant differences were selected to plot ROC curves. The ROC curves were compared to assess the predictive value of three ultrasound diaphragmatic functional indices for postoperative pulmonary complications after abdominal surgery. A predictive model for postoperative pulmonary complications was created using Lasso Logistic regression analysis and compared with individual diaphragmatic functional indices. Results The differences in preoperative DE-QB, DE-DB, and TFdi between the two groups were not statistically significant(P>0.05). However, the postoperative DE-QB, DE-DB, and TFdi in the PPCs group were lower than those in the n-PPCs group, with statistically significant differences(P<0.001). The areas under the ROC curve(AUC)for postoperative DE-QB, DE-DB, and TFdi were 0.780, 0.779, and 0.713, respectively, with no significant differences. Patients were divided into high-risk PPCs and low-risk PPCs groups according to the optimal cut-off value of the ROC curves of the three functional indices. When patients were differentiated by postoperative DE-QB(optimal cut-off value of 0.93 cm), the incidence of pulmonary complications differed between the high-risk and the low-risk groups(14.1% vs. 61.8%, P<0.001); when patients were differentiated by postoperative DE-DB(optimal cut-off value of 2.41 cm), there was a difference in the incidence of pulmonary complications between the high-risk and the low-risk groups(13.3% vs. 56.3%, P<0.001); when patients were differentiated by postoperative TFdi(optimal cut-off value of 0.23), there was a difference in the incidence of pulmonary complications between the high-risk and the low-risk groups(16.5% vs. 52.4%, P<0.001). By Lasso Logistic regression analysis, the final risk factors included in the prediction model were age, smoking history, duration of surgery, DE-QB(postoperative), and DE-DB(postoperative), and the Lasso Logistic regression model(AUC=0.851)outperformed any single index of diaphragm function. Conclusion The DE-QB, DE-DB and TFdi measured at 24h post-operatively alone can predict the occurrence of post-operative pulmonary complications(PPCs)to some extent when used individually, however the measurement of diaphragmatic excursion(DE)provided better predictive value for PPCs than TFdi. The predictive model established by Lasso Logistic regression can better predict the incidence of PPCs than any single diaphragm function index.

Key words: Diaphragm excursion, Fraction of diaphragm thickening, Abdominal surgery, Ultrasound, Pulmonary complications