Analysis of influencing factors of intrauterine adhesions after early missed abortion surgery and construction of predictive model

Objective To analyze the influencing factors of intrauterine adhesions after early missed abortion surgery and construct a predictive model.

Methods The data of 438 patients with missed abortion were retrospectively collected. A total of 307 patients were randomly selected as modeling set and 131 as test set. Univariate and multivariate Logistic regression analysis was performed on the modeling set data and the prediction model was constructed. The effectiveness of the model was evaluated using receiver operating characteristic (ROC) curve, calibration curve, and decision curve, respectively, and a test set to verify the effectiveness of the model was introduced.

Results A total of 58 patients with early missed miscarriage in the modeling set one month after curettage experienced intrauterine adhesions, with an incidence rate of 18.89% (58/307). Among them, there were 46 cases of partial adhesions in the uterine cavity and 12 cases of extensive adhesions in the uterine cavity. Multivariate Logistic regression analysis showed that times of previous induced abortions, previous uterine cavity procedures, methods of curettage, overexpressed transforming growth factor β1 (TGF-β1) and low expression of matrix metalloproteinase-9 (MMP-9) were influencing factors for postoperative uterine adhesions in early missed abortion (P < 0.05). The area under the ROC curve of the risk prediction model of intrauterine adhesion after early missed abortion based on the above indicators was 0.894, the optimal cutoff value (threshold probability) was 0.440, sensitivity was 0.890, and specificity was 0.848. The Brier index for calibration curve analysis was 0.068. The area under the ROC curve introduced into the test set for model verification was 0.877, the sensitivity was 0.875, and the specificity was 0.802. The Brier index for calibration curve analysis was 0.165. When the threshold probability value in the decision curve was set to 0.44, this predictive model can provide significant additional clinical net benefits.

Conclusion Based on the number of previous induced abortions, the number of uterine cavity procedures, and methods of curettage, TGF- β1 and MMP-9, the construction of a predictive model for the risk of intrauterine adhesions after early missed abortion has certain practical value.