LAUSR

This letter highlights a few concerns from a recent observational study, where authors found that extremely low birth weight infants receiving exclusive human milk (EHM) had a significantly lower intraventricular hemorrhage (IVH) or peri-ventricular leucomalacia (PVL) [1]. A confounder is a variable associated with the exposure and the outcome of interest and does not present in the causal pathway between the exposure and the outcome [2, 3]. Antenatal steroids satisfy the confounder’s criteria, and the authors appropriately controlled it while evaluating the association between EHM and IVH or PVL and closed the back-door path (Fig. 1). The authors also adjusted for necrotizing enterocolitis (NEC). However, NEC partially mediates the association between the EHM and PVL [4, 5]. Controlling NEC would be inappropriate as it would partially close the indirect causal path, thereby attenuating the observed association between EHM and IVH or PVL (overadjustment bias) [2, 3]. The authors incorrectly infer the association between EHM and IVH or PVL in the adjusted analysis. They report that “the incidence of severe IVH or PVL is significantly lower in the EHM group than the non-EHM group (odds ratio 2.7; 95% confidence interval 1.2, 6.0; P= 0.012).” The odds ratio is 2.7, indicating a higher risk for severe IVH or PVL in the EHM group. It is unlikely to be true, and I presume the analysis is flipped. The authors should carefully reconsider the statistical analysis and provide the true estimate keeping all the factors mentioned above. The authors should also present the baseline differences of important confounders, such as volume-guarantee ventilation and prophylactic indomethacin, that can influence IVH or PVL. The authors should also consider providing the adjusted analysis for NEC, an important and relevant outcome of this study. The NEC rate was 5 percent in the EHM group and 17 percent in the non-EHM group. However, in the abstract, the authors reported higher NEC in the EHM group. The error should be rectified.