The value of ultrasound in the diagnosis and prediction of the severity of abdominal adhesions

Objective: The aim of this study was to construct a preoperative ultrasound prediction model, and compare its diagnostic performance with the existing sliding sign to predict the severity of abdominal adhesions in order to reduce the occurrence of intraoperative complications and shorten the duration of surgery. Methods: Between June 2020 and June 2022, 100 patients with a history of abdominal surgery were included in this retrospective study. All participants underwent ultrasound examination of five sites on the anterior abdominal wall before surgery. Eighteen sites were excluded because of surgical stomas, etc. Finally, 482 sites were examined by ultrasound, of which 138 (28.6%) were severe adhesions. Based on the intraoperative findings, the patients were divided into two groups: patients with severe adhesions and patients with non-severe adhesions. Finally, the data of 482 abdominal sites examined were randomly divided into a training cohort (70%) and a validation cohort (30%). The least absolute shrinkage and selection operator (LASSO) regression and multivariate binary logistic regression were used to determine the independent influencing factors, thereby constructing a clinical ultrasound prediction model for severe abdominal adhesions. The sensitivity, specificity, positive and negative predictive value (PPV and NPV) and accuracy of the model were calculated. Prediction models were established, and the area under the receiver operating curve (AUC) between models and compared with sliding sign (MA) using the DeLong test to determine the optimal model. At the same time, intraclass correlation coefficient (ICC) was used to evaluate the inter-observer agreement. Results: The LASSO regression showed that wall syndrome, traction sign, deformation of abdominal organs, two-layer peritoneal bright lines, sliding sign (head–foot direction mobility), and examination site were associated with severe adhesions. Multivariate analysis showed that traction sign, two-layer peritoneal bright lines, head–foot direction mobility and examination site (all P < 0.05) were the independent predictors of severe abdominal adhesion. Based on these predictors, the improved prediction model (referred as MB) was established. It showed that the diagnostic performance of MB was better than MA [AUC MB = 0.943 (95% CI: 0.920–0.967) vs AUC MA = 0.827 (95% CI: 0.920–0.967), P < 0.001] in the training cohort. The results were validated [AUC MB = 0.873 (95% CI: 0.818–0.928) vs AUC MA = 0.803 (95% CI: 0.755–0.851), P = 0.005] in the validation cohort. In the training cohort, the MB improved the specificity (22.5%), PPV (22.6%) and the accuracy (13.6%). In the validation cohort, the MB improved the specificity (19.2%), PPV (10.8%) and the accuracy (6.9%). At the same time, the ICC showed that the ultrasonic parameters had high consistency. Conclusions: In the presence of complex adhesions in the abdominal cavity, the new diagnostic model (MB) significantly improved the diagnostic performance compared with the conventional model (MA). The MB improved the AUC, specificity, PPV, and accuracy. Therefore, this model has the potential to reduce severe intraoperative complications.