Termination of breastfeeding after 12 months of age due to a new pregnancy and other causes is associated with increased mortality in Guinea-Bissau

Background As part of an assessment of breastfeeding and child health in Guinea-Bissau, we investigated the impact of mother's reason for weaning on subsequent child mortality. Methods Children were identified and followed by the demographic health surveillance system of the Bandim Health Project in Guinea-Bissau. Breastfeeding status and survival were ascertained by 3-monthly follow-up home visits. At termination of breastfeeding mothers were interviewed about her reasons for weaning. In all, 1423 children who terminated breastfeeding after 12 months of age were followed to 3 years of age. Results Median length of breastfeeding was 22 months. Following termination of breastfeeding, 66 children died before 36 months of age. in all, 62% (879/1423) were weaned because they were 'healthy'. Compared with the 'healthy' children, all other causes of weaning were associated with a higher mortality (mortality ratio [MR] = 2.97, 95% Cl: 1.54-5.73). For 237 children weaned due to a new pregnancy the MR was 3.25 (95% CI: 1.45-7.30). Seventy-five children weaned because of illness had a 2.98 (95% CI: 0.95-9.39) fold excess mortality compared with children considered healthy. Excess deaths in the 'non-healthy' group accounted for 44% (29/66) of post-weaning deaths. Median length of spacing between an index child and a new sibling was 28 months irrespective of whether the index child survived or died before 3 years of age. The majority of the deaths occurred before birth of the new sibling. Conclusion Popular rationalizations of abstinence during breastfeeding emphasizes, as we observed, that weaning due to new pregnancy of the mother is associated with higher mortality. This was not due to a shorter breastfeeding period of the child weaned due to a new pregnancy. Generally children weaned for other reasons than 'being healthy' had higher mortality. The mother's reason for weaning could potentially be used as screening criteria in child monitoring programmes in areas with high mortality.