3. Association between menstruation duration or TRM and the clinical characteristics of the patients and the measurement data of CSD by univariate analysis.
Interestingly, we find that the thickness of the muscular layer of the lower segment of the uterus in the menstruation ≤7 days group is significantly lower than that in the menstruation >7 days group (22.67±26.88 vs. 31.77±27.84, p < 0.05). The width of the CSD via MRI of the menstruation ≤ 7 days group is 13.27±8.14 mm, which is significantly lower than that of the menstruation>7 days group (14.99±8.04 mm). 47.61% (409/859) of the patients from the TRM≥5.39 mm group has an anterior position of the uterus and only 23.46% (19/81) of patients from the TRM<5.39 mm group had an anterior position of the uterus. When TRM is≥5.39 in postoperative, the TRM is thicker in preoperative (2.79±1.87 vs. 2.16±0.93, p < 0.05) (eTable 3 in the
Supplement).
4. Multivariable linear regression models for the association of CSD parameters and optimal healing/menstrual improvement/TRM
The multiple linear regression results based on the entire sample are presented in
Table 1. There is a significant negative association between CSD width and myometrial
layer thickness of the lower uterine segment in the optimal healing group. Additionally,
there is a significant positive association between the length and depth of the CSD in
the optimal healing group. However, there is no significant association between CSD
width and TRM improvement; however, a significant positive association between
CSD depth and TRM is observed, and the uterine position is significantly associated
with TRM improvement. Even so, the width of the CSD is significantly associated with menstrual improvement. Additionally, there is a significant positive association
between menstrual cycle duration and CSD length or depth in favor of menstrual
improvement.
5. A prediction model for the optimal healing of CSD
Multivariable logistic regression analysis is used to assess the individualized
prediction model, and the final nomogram bases on logistic regression analysis in the
training cohort is built. Both the thickness and width of the CSD via MRI are negatively associated withoptimal healing (p<0.05). CSD depth,CSD length,the menstrual cycle after cesarean section, and the timing of surgery in the cesarean section was positively associated with optimal healing (p < 0.05). The model derived from the estimated β-regression coefficients of twelve variables is developed as a nomogram (Figure 2A, eTable 4 in the Supplement). The calibration curve of the nomogram shows adequate agreement between observation and prediction in the training cohort and internal validation cohort (Figure 2C-2E). The Hosmer‒Lemeshow test returned a nonsignificant value (P = 0.419 in the training cohort, P = 0.963 in the internal validation cohort). This result represent no departure from a perfect fit. The C-index for the prediction model is 0.790 for the training cohort. The calibration curve depicted the calibration of the model in terms of the agreement between the observed outcome of optimal healing and the predicted risk of optimal healing. Furthermore, the prediction nomogram yielded a C-index of 0.823 according to internal validation of the nomogram. The decision curve analysis for the nomogram is presented in Figure2B.The net benefit is comparable within this range based on the nomogram. With the nomogram model, we can provide patient treatment recommendations.
6. The predictive score model
The predictive score of each predictor is assigned by dividing the beta coefficient from the logistic regression model by 0.20 (eTable 5 in the Supplement). Thickness is given the highest score of 9.0, and CSD length is given the lowest score of 0.5. The menstrual cycle after C-sections is given a score of 3.0. CSD width is given a score of 6.0, and the timing of surgery in the cesarean section is given a score of 4.0. Total scores range from 0 to 25 points with a cutoff point of 16.5 (Table 2). When the score ≥16.5, the patient can be predicted to achieve optimal healing. This suggests that we can improve the surgical protocol or change the treatment plan if the score is <16.5.