The characteristics of the children lost to follow-up were compared with those of the children who remained in the study using inferential analysis, as were the characteristics of the control and intervention groups. Means and proportions were compared using Student’s t or Mann-Whitney’s test and Pearson’s chi-square or Fisher’s exact test, respectively. Multivariable Poisson regression model with robust variance was used with a model that included variables showing differences between the control and intervention groups as a result of the losses (p < 0.20). After the use of a non-adjusted model, different cumulative models were tested, progressively including variables of the previous model(s). The first adjusted model considered only group and propensity score (Seeger et al. [20]). The propensity scores were estimated using logistic regression, modeling the probability of an individual being allocated to the intervention group and considering the following predictors: maternal age, educational attainment, skin color, and parity; infant weight and mode of delivery; and parental cohabitation. Significance was set at 5 % (p ≤ 0.05). All analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 21.0.