Caesarean delivery and early childhood caries: Estimation with marginal structural models in Brazilian pre-schoolers

Abstract

Objective: This study analysed the association between caesarean section and early childhood caries (ECC), estimating the effects using regression and causal inference models. Methods: This was a historical cohort study of 697 mother-child dyads, conducted in São Luís, Brazil. The caesarean section was the exposure, and the severity of ECC (dmft) was the outcome. Covariates household income, maternal schooling, maternal hypertension, maternal obesity and birth weight were adjusted for in the models. The effects were estimated by Poisson regression (Means Ratio-MR) and causal inference using a marginal structural model (MSM) (MR and Average Treatment Effect-ATE coefficients), weighted by the inverse probability (IPW) of exposure. Results: Caesarean section was protective against caries in the bivariate (MR 0.81; CI 0.70-0.94; P = 0.005) and multivariate (MR 0.78; CI 0.67-0.91; P = 0.002) models. In MSM analyses, the caesarean section had no effect on ECC (ATE = −0.35; P = 0.107), controlling for IPW of exposure. Conclusion: The apparent association between caesarean section and ECC severity seems spurious, as it did not persist after employing a superior approach to estimating causality.