Results ①The proportions of patients with spontaneous labor, history of lateral episiotomy/laceration and without Kegel training in study group were significantly higher than those in control group, and the differences were statistically significant (χ2=10.03, 8.57, 6.47; P=0.002, 0.003, 0.039). ②The bladder neck rotation angle (BNRA), posterior urethrovesical angle (PUA), bladder neck inclination angle (BNIA), bladder neck distance (BND), anal constriction-levator hiatus area (AC-LHA), rest-levator hiatus area (R-LHA), Valsalva-levator hiatus area (V-LHA) in study group were significantly higher than those in control group, and the differences were statistically significant (t=13.82, 15.36, 10.35, 6.86, 5.41, 3.38, 6.98; all P<0.001). ③The ROC curve analysis results of 7 parameters of three-dimensional pelvic floor ultrasound (BNRA, PUA, BNIA, BND, AC-LHA, R-LHA, V-LHA) for predicting the occurrence of POP in parturients showed that the AUC of these 7 parameters alone and the combination of these 7 parameters for predicting the occurrence of POP in parturients were 0.745 (95%CI: 0.611-0.879, P<0.001), 0.796 (95%CI: 0.692-0.906, P<0.001), 0.720 (95%CI: 0.593-0.847, P=0.001), 0.701 (95%CI: 0.561-0.840, P=0.002), 0.679 (95%CI: 0.545-0.813, P=0.005), 0.653 (95%CI: 0.520-0.786, P=0.008), 0.712 (95%CI: 0.577-0.845, P=0.001) and 0.851 (95%CI: 0.743-0.959, P<0.001), respectively. ④Multivariate unconditional logistic regression analysis with whether the subjects were postpartum POP as dependent variable, and mode of delivery, lateral episiotomy/laceration, Kegel training, the above 7 ultrasound parameters as the independent variables showed that spontaneous labor, history of lateral episiotomy/laceration, BNRA>30°, PUA>135°, BNIA>54°, BND>18 mm, AC-LHA≥13 cm2, R-LHA≥17 cm2, and V-LHA≥18 cm2 all were independent risk factors for the occurrence of POP in parturients (OR=3.800, 2.238, 5.689, 5.992, 5.199, 4.288, 3.301, 2.723, 4.732; 95%CI: 2.814-6.596, 2.110-4.176, 3.556-9.631, 3.707-10.087, 3.362-8.846, 3.009-7.376, 2.592-5.819, 2.291-4.964, 3.179-8.095; all P<0.05); Kegel training had a protective effect on postpartum pelvic floor (OR=0.596, 95%CI: 0.518-0.674, P=0.008). AUC of the multivariate unconditional logistic regression model for predicting the occurrence of POP in parturients was 0.822 (95%CI: 0.759-0.885), with a sensitivity of 94.2% and specificity of 60.1%. The C-index of this model was 0.809 (95%CI: 0.744-0.874), indicating that the model was well differentiated, and P-value of Hosmer-Lemeshow test for this model was 0.870, indicating that the model was well calibrated. Internal data validation of the model showed that the corrected C-index was 0.814 (95%CI: 0.750-0.878) and P-value of the Hosmer-Lemeshow test was 0.894; meanwhile, the validation of calibration of the prediction model showed that the ideal curve in calibration plot for predicting postpartum POP matched good with the actual curve, and the bias calibration curve had a small bias, indicating good agreement between its prediction and actual of POP occurrence in parturients.
Conclusions The accuracy of combined application of all three-dimensional pelvic floor ultrasound parameters to predict the occurrence of POP in parturients is high. Spontaneous delivery, history of lateral episiotomy/laceration and BNRA>30°, PUA>135°, BNIA>54°, BND>18 mm, AC-LHA≥13 cm2, R-LHA≥17 cm2, and V-LHA≥18 cm2 are independent risk factors for the occurrence of POP in parturients, while Kegel training has a protective effect on the postpartum pelvic floor.