Birth weight and duration of breast-feeding: are the beneficial effects of human milk being overestimated?

The beneficial effects of breast-feeding on infant mortality and morbidity have been demonstrated in many studies. Few of these, however, have taken into account the possible confounding effect of birth weight. Several studies have shown that babies of low birth weight are less likely to be breast-fed. In some circumstances, this alone may account for a more than twofold excess in postperinatal infant mortality rates among non-breast-fed babies, even in the absence of any beneficial effect of breast-feeding. The association between birth weight and breast-feeding and the magnitude of the confounding effect is illustrated using data from a longitudinal study of infant mortality in Pelotas, Southern Brazil, and also using published results from other studies. It is concluded that studies designed or analyzed to relate breast-feeding to infant mortality should take the confounding effect of birth weight into account to avoid overestimating the beneficial effects of human milk.