Abstract
BACKGROUND: Suboptimal breastfeeding is associated with higher mortality among infants and young children in the developing world. We describe patterns in 'exclusive breastfeeding' and 'any breastfeeding' rates and quantify exposure to suboptimal breastfeeding among children aged two years or younger in developing countries. METHODS: We reviewed nationally representative surveys that collected data on breastfeeding rates in 94 developing countries. Surveys were categorized by completeness and comprehensiveness of data. Complete and comprehensive data were analysed with minimum chi-square regression. With a fitting procedure, estimated parameters were used to impute missing observations for incomplete or non-comprehensive surveys. Breastfeeding indicators were calculated and are reported for 135 developing countries by UN region. RESULTS: Amongst infants aged six months or younger in the developing world, the prevalence of exclusive breastfeeding is 39% and the prevalence of no breastfeeding is 5.6%. The prevalence of continued breastfeeding is 86% and 68% for infants and children aged 6-11 and 12-23 months, respectively, in the developing world. Imputation expands population coverage of indicators, especially for infants. Breastfeeding trends are highly linear and estimated parameters defining the age-specific attrition hazard are robust. Survey-reported rates, particularly for exclusive breastfeeding, appear to have systematic upward bias, and exposure estimates must be considered conservative. CONCLUSIONS: Compliance with breastfeeding recommendations in developing countries is low, and more attention should be given to increasing breastfeeding - especially exclusive breastfeeding - and to monitoring trends. Although the introduction of more standardized and better validated survey instruments is desirable, since data coverage, completeness and comprehensiveness are extensive, global exposure assessment is relatively robust. Moreover, the regularity of breastfeeding patterns show existing survey data capture real biological and social phenomena. Our method for the analysis of breastfeeding rates provides a potent tool for summarizing trends, validating observations, translating and extrapolating indicators (as well as projecting and imputing estimates when necessary) and should support more effective child health monitoring.