Relationship of human milk feeding with infant health outcomes: A longitudinal analysis of BINGO and pairings study data

Objectives: To assess the relationship between intensity of human milk feeding and the occurrence of infant illness during the first year of life. Methods: Data on infant feeding were collected by interviewing the mother at 1, 3, and 6 months postpartum. Four levels of ‘human milk feeding intensity', the exposure variable of interest, are defined as: (1) exclusively formula fed, (2) <20% human milk, (3) >20% and <80% human milk, (4) >80% human milk based on a previously published approach. Infant health data were collected using (a) medical records and (b) mother-report of infant illness. Marginal structural models were used to estimate the relationship between human milk feeding intensity and infant illness. Inverse probability of treatment weights were applied to the data to adjust for confounding. A gamma sensitivity analysis was used to assess whether the results were sensitive to unmeasured confounding. Results: Exclusive formula feeding, compared with >80% human milk feeding, was associated with an increased odds of respiratory infection, as reported on medical records. Exclusive formula feeding, compared with >80% human milk feeding, was associated with increased odds of maternal reporting that her infant had a cough, eczema, cold, and any illness. <20% human milk feeding, compared with >80% human milk feeding, was associated with increased risk for maternal report of infant cough and report of any illness. Conclusions: The results show that reduced human milk feeding is associated with increased infant illness during the first year of life.