# Effect of breastfeeding on gastrointestinal infection in infants: A targeted maximum likelihood approach for clustered longitudinal data

**Authors:** Mireille E. Schnitzer, Mark J. van der Laan, Erica E. M. Moodie, Robert W. Platt  
**arXiv:** [1407.8371](https://arxiv.org/abs/1407.8371)  
**Date:** 2014-07-31  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/1407.8371v1/

## Key points

- The PROBIT intervention successfully increased the duration of breastfeeding and lowered rates of gastrointestinal tract infections in newborns
- This study aimed to estimate the effect of breastfeeding duration on infection count
- The researchers encountered challenges due to confounding factors and used a double-robust method called Targeted Maximum Likelihood Estimation (TMLE)
- TMLE was compared with other causal methods for this example
- A simulation study was conducted to determine the effectiveness of TMLE in controlling for clustering indicators and the importance of using data-adaptive TMLE
- The findings provide valuable insights into estimating the causal effect of breastfeeding duration on gastrointestinal infection and can contribute to our understanding of how breastfeeding impacts infant health outcomes

## Summary

The PROmotion of Breastfeeding Intervention Trial (PROBIT) cluster-randomized a program encouraging breastfeeding to new mothers in hospital centers. The original studies indicated that this intervention successfully increased the duration of breastfeeding and lowered rates of gastrointestinal tract infections in newborns. There is still scientific and popular interest in determining the causal effect of longer breastfeeding on gastrointestinal infection. In this study, conducted by Mireille E. Schnitzer, Mark J. van der Laan, Erica E. M. Moodie, and Robert W. Platt, the researchers aimed to estimate the expected infection count under various lengths of breastfeeding in order to estimate the effect of breastfeeding duration on infection. They encountered challenges due to the presence of baseline and time-dependent confounding factors which required specialized "causal" estimation methods. To address these challenges, the researchers demonstrated the use of a double-robust method called Targeted Maximum Likelihood Estimation (TMLE) in the context of this application. They also reviewed other related methods and discussed adjustments required to account for clustering. The researchers compared TMLE (implemented both parametrically and using a data-adaptive algorithm) with other causal methods for this example. Additionally, they conducted a simulation study to determine two key aspects: (1) how effective it is at controlling for clustering indicators when cluster-specific confounders are unmeasured; and (2) how important it is to use data-adaptive TMLE. The findings from this study provide valuable insights into estimating the causal effect of breastfeeding duration on gastrointestinal infection using advanced statistical methods like TMLE which can contribute to our understanding of how breastfeeding can impact infant health outcomes and may have implications for public health interventions promoting breastfeeding practices.