ABSTRACT
Objective: To explore the related factors of surgical treatment of patients with corpus luteum rupture and establish a risk prediction model of surgical treatment of corpus luteum rupture. Methods: 222 patients with corpus luteum rupture treated in Jingzhou First People’s Hospital from January 2015 to March 2022 were analyzed retrospectively, including 45 cases of surgery and 177 cases of conservative treatment. The training set and validation set were randomly assigned according to 7:3. We collected the basic information, laboratory and ultrasonic examination data of 222 patients. Logistic regression analysis was used to determine the independent risk factors and combined predictors of surgical treatment of corpus luteum rupture. The risk prediction model was established and the nomogram was drawn. The discrimination and calibration of the prediction model were verified and evaluated by receiver operating characteristic (ROC) curve, calibration curve and Hosmer-Lemeshow goodness of fit test; Decision curve analysis (DCA) was used to evaluate the clinical effectiveness of the prediction model. Results: Univariate logistic regression showed that whole abdominal pain (OR: 2.314, 95% CI: 1.090 - 4.912), abdominal muscle tension (OR: 2.379, 95% CI: 1.112 - 5.089), adnexal mass ≥ 4 cm (OR: 3.926, 95% CI: 1.771 - 8.266), hemoglobin < 12 g (OR: 11, 95% CI: 4.724 - 25.616), pelvic effusion depth ≥ 3 cm under ultrasound (OR: 10.606, 95% CI: 4.602 - 24.445) and positive cervical lifting pain (OR: 3.960, 95% CI: 1.831 - 8.563) were suspected risk factors for surgical treatment of corpus luteum rupture; Multivariate logistic regression analysis showed that hemoglobin < 12 g (OR: 5.398, 95% CI: 1.985 - 14.682), pelvic effusion depth ≥ 3 cm under ultrasound (OR: 6.256, 95% CI: 1.607 - 24.354) and positive cervical lifting pain (OR: 2.995, 95% CI: 1.19 - 7.538) were independent risk factors for surgical treatment of corpus luteum rupture (P < 0.05). The nomogram is drawn according to the prediction variables, and the prediction model is constructed. The prediction model predicted that the area under the ROC curve (AUC) of patients with corpus luteum rupture in the training set was 0.841, 95% CI (0.759, 0.922), and the area under the ROC curve (AUC) of patients with corpus luteum rupture in the validation set was 0.919, 95% CI (0.821, 0.999). Conclusion: The nomogram prediction model containing three predictive variables (hemoglobin, depth of pelvic effusion under ultrasound and cervical lifting pain) can be used to predict the risk of surgical treatment in patients with corpus luteum rupture.