Predicting Cesarean Section in Indian Women: A Retrospective Analysis using Bootstrap Modeling

A bstract Background: Cesarean delivery rates are increasing globally, including in India, raising concerns about maternal and neonatal outcomes. Objective: To identify maternal, clinical, and sociodemographic predictors of cesarean delivery in an Indian cohort. Materials and Methods: This retrospective case-control study was conducted between March and September 2024 at a tertiary care hospital in Nagpur, India. The study enrolled 250 women—125 who underwent cesarean delivery (cases) and 125 who had vaginal deliveries (controls). Data were collected via structured interviews and medical records. Bootstrap modeling, a statistical resampling technique used to improve model reliability and minimize overfitting, was employed instead of the traditional logistic regression analysis. The model performance was evaluated using the receiver operating characteristic curve and the Hosmer–Lemeshow test. Results: Univariate analysis revealed significant associations with age >30 years, urban residence, short stature, obesity [body mass index (BMI) ≥ 30], comorbidities, fewer than four antenatal care (ANC) visits, and delayed ANC initiation. The bootstrap model identified urban residence adjusted odds ratio (AOR 4.0), short stature (AOR 2.6), obesity (AOR 2.5), comorbidities (AOR 2.8), inadequate ANC visits (AOR 2.2), and below poverty line status (AOR 2.0) as independent predictors. The final model demonstrated good discrimination (AUC = 0.842) and calibration (Hosmer–Lemeshow P = 0.687), with sensitivity of 83% and specificity of 79%. Conclusion: Urban residence, maternal obesity, comorbidities, and inadequate ANC are independent predictors of cesarean delivery in Indian women. The predictive model may be a useful clinical tool for risk stratification during pregnancy.