Predictive models and significant predictors for vaginal birth after caesarean section: A scoping review

Caesarean section rates across the world have been rising steadily for decades. Women with a prior caesarean face uncertainty about whether to attempt vaginal birth in future pregnancies or undergo a repeat caesarean. Predictive models estimating VBAC success can help inform delivery decisions by clarifying risks and likelihood of success. This scoping review aims to investigate the current state of VBAC prediction modelling. We conducted a comprehensive scoping review of studies between January 1st, 2000 and December 31st, 2024 that have created and validated predictive models for VBAC. The results indicate that research on VBAC predictive modelling has primarily focused on identifying relevant predictors and validating models across different populations and demographics, often using limited datasets and relying predominantly on Area Under the Receiver Operating Characteristic Curve (AUC) as the performance metric. There is limited research into non-linear modelling, with current studies primarily utilising logistic regression. Four key challenges were identified: (1) the need to explore potentially more accurate non-linear models, (2) the need to incorporate influential VBAC predictors into model development, (3) the need for larger datasets to improve generalisability, and (4) the need to use multiple complementary metric to evaluate model performance. Future research should address the four challenges by exploring explainable, non-linear models that strike a balance between predictive accuracy and clinical interpretability.