The factors considered were the Bishop score, vaginal childbirth history, neonatal weight, maternal age, and BMI.
Model Properties
The evaluation of the predictive models includes discrimination and calibration [39]. Among the 26 studies included, 25 used the area under the receiver operating characteristic (ROC) curve to assess the model's discriminative ability. Among these, 3 models had a discriminatory capacity of less than 0.7, 17 had a discriminatory capacity between 0.7 and 0.9, and 5 had a discriminatory capacity greater than 0.9. This indicates that most of the prediction models had good discrimination, and only 1 study did not report the model’s discriminatory capacity [12]. The Hosmer-Lemeshow test was used in 9 studies to verify the models' calibration; the models in 6 studies showed a reasonable degree of calibration.
Model Validation
Model validation can be either internal or external [39]. Of the 26 models included, 4 were internally validated, 6 were externally validated, and only 1 was validated both internally and externally. Most models (11 studies) are presented in the form of regression equations; other formats include nomograms (7 studies), machine learning models (2 studies), scoring systems (5 studies), and web-based calculators (1 study).
Discussion
In this review, we summarized 26 predictive models for vaginal birth after cesarean (VBAC). Due to differences in medical and cultural backgrounds across regions, there is a wide variety of VBAC predictive models. However, most of these models are presented in the form of regression equations and have not been visualized, which increases the complexity of clinical use. Regarding model validation, the majority of models lack external validation, indicating deficiencies in the verification aspects of existing VBAC predictive model studies. In terms of study populations, 15 of these predictive models originate from China. The predominance of Chinese literature can be attributed to the surge in second-child pregnancies in China following the implementation of the two-child policy in 2016. This demographic shift included a significant number of women with a history of cesarean delivery, who were faced with the decision to attempt a trial of labor after cesarean (TOLAC). This may explain the abundance of related Chinese literature. Since most of the included studies are concentrated in China, these predictive models may primarily be applicable to the medical environment and population characteristics in China. Differences in medical resources, cultural backgrounds, and maternal health status across regions may affect the generalizability of these models.
