Breast-feeding and child survival in Matlab, Bangladesh.

This study investigates the effect of both total and unsupplemented breast-feeding in conjunction with birth interval on early childhood mortality, using longitudinal data from Matlab, Bangladesh. A discrete hazard model approach shows that it is not the duration of total breast-feeding but the duration of unsupplemented breast-feeding which increases child survival. Unsupplemented breast-feeding appears as such as crucial determinant of early childhood mortality that its effect could not be substantially attenuated even when important demographic and socioeconomic factors were controlled. Each of the covariates--supplementation, previous birth interval and onset of a subsequent conception--has an independent influence on early childhood mortality.