The associations between maternal body composition and breast milk output: An explorative causal inference approach

Introduction: Several scholars have found an association between maternal body composition and breast milk output. Heavier mothers tend to have lower breast milk output. Other maternal and child factors might affect breast milk output and/or body composition as well. Examples are parity, weight of the child, and even non-biological factors like socioeconomic status or smoking. In order to obtain the desired unconfounded effect estimates for the association between body composition and breast milk intake by the child, knowledge of the underlying causal mechanisms is necessary, taking into account all possible confounding background variables. The aim of this study was to explore these causal mechanisms. Methods: Causal inference search algorithms were used to explore the underlying mechanisms, by determining which causal graphs were compatible with the given data. With this technique, not only the causal effects between exposure and outcome variables can be determined, but also the effect of background variables on both exposure and outcome variables can be explored. Structural equation modelling was then applied to the models to find the values for the regression coefficients. Data for this study was obtained from a database in which milk intake data from 16 studies was pooled together (n=188). All studies used the dose-to-the-mother deuterium-oxide turnover method for measuring breast milk transfer from mother to child. Results: For each of the three measures of maternal body composition (BMI, FMI, FFMI), and height the model best fitting the data was selected. Preliminary findings of the model for BMI showed a direct negative effect of BMI on milk output, i.e. a higher BMI results in lower breast milk output. This direct negative effect was also found in the FMI model, but not in the FFMI model. Education was found to have a confounding effect in the association between body composition and milk output through parity and smoking. Firstly, higher maternal education was associated with lower parity, and subsequently lower fat mass, and higher milk output, Secondly, higher maternal education was associated with lower smoking prevalence, which in turn resulted in higher milk output. Conclusion: Even though results are still preliminary, this exploratory study identified several confounders in the association between body composition and breast milk output, like parity, smoking, and education. When decomposing BMI into FMI and FFMI, we found that the effect of body composition on milk output is caused by maternal fat mas and not by her lean mass. (author)