Breastfeeding is beneficial for the health of children in developing countries because breastmilk is (1) nutritionally appropriate, (2) provides immunological protection, and (3) is free of contamination. Breastfeeding also promotes improved health by lengthening birth intervals. This dissertation investigates the determinants of the incidence and duration of breastfeeding in 27 developing countries using data from the Demographic and Health Surveys, conducted in the late-1980s.
The incidence of breastfeeding is examined using logistic regression. Retrospective reports of age at weaning and current-status data (whether or not a child is being breastfed at the time of the survey) are both employed for the analysis of breastfeeding durations. Retrospective data are analyzed using proportional hazards models, but because age at weaning is misreported by many women, analysis focuses on current-status data. A methodology for analyzing these data in a multivariate framework is developed after comparison of a variety of parametric, semi-parametric and non-parametric specifications. The preferred methodology is a logistic regression in which the baseline log-odds of breastfeeding by child's age are represented by a natural cubic spline.
The results show that increasing levels of socio-economic factors have a strong negative association with breastfeeding incidence and duration, but that their effect operates primarily through demographic, attitudinal and economic factors. In multivariate models which simultaneously control for all determinants, five factors emerge as particularly important. A higher level of mother's education, use of oral contraceptives, first births, ownership of a refrigerator, and younger mothers are all associated with lower incidence and durations of breastfeeding.
Analysis of trends in breastfeeding durations between the late-1970s (based on the World Fertility Surveys) and the late-1980s (based on the Demographic and Health Surveys) show statistically significant increases in breastfeeding duration in eleven of the fifteen countries analyzed. A small portion of this positive trend is explained by improvements in child mortality. Changes in the distribution of women across population subgroups have generally exerted a downward pressure on the duration of breastfeeding, but within population subgroups, trends have been positive in nearly every country. The policy implications of these findings are also explored.
My Research