Predictive model for the determination of caesarean sections in the Catalan National Health System: an approach based on clinical factors

ObjectiveTo develop and validate a logistic regression model for analyzing the probability of caesarean section births, adjusted for clinical complexity, across public hospitals in Catalonia, and to identify deviations from expected caesarian section rates for benchmarking and quality improvement. MethodsThis retrospective cohort study analyzed data from the Catalan National Health Systems Minimum Basic Data Set (CMBD-AH), including all deliveries in public hospitals from January 2018 to June 2024. A logistic regression model was constructed using maternal and obstetric factors such as age, obstetric history, and clinical conditions. The model was validated through calibration plots and receiver operating characteristic (ROC) curve analysis, achieving an area under the curve (AUC) of 0.803. ResultsThe analysis revealed variability in observed-to-expected caesarean section ratios across hospital complexity levels. Level III hospitals aligned closely with expected rates, reflecting adherence to clinical standards for high-complexity cases. Level I hospitals demonstrated significant variability, with 59.1% performing more cesareans than expected; smaller hospitals with fewer than 1,000 births exhibited the greatest deviation. The model highlighted both underperforming and overperforming institutions, offering actionable insights for resource allocation and policy interventions. ConclusionsThe logistic regression model provides a robust framework for evaluating caesarean section practices, enabling fair comparisons between hospitals by adjusting for clinical complexity. It supports the identification of non-clinical factors influencing cesarean practices and offers a critical tool for quality improvement and optimizing maternal healthcare within Catalonias public health system.