Effect of Breastfeeding Education and Support Intervention on Timely Initiation and Exclusive Breastfeeding

Overview

The purpose of this study is to develop and test Breastfeeding Education and Support Intervention for improving timely initiation and Exclusive Breastfeeding rate. Infant mortality rates are still high in Ethiopia. Breastfeeding (BF) is regarded as the simplest and least expensive strategy for reduction of infant mortality rates. Ethiopia does not meet the international recommendation for exclusive breastfeeding (EBF) for the first six months of life. Community-based educational and support interventions provided prenatally and postnatally are effective in increasing BF rates. However, there is paucity of such information in Ethiopia.

Sample size was calculated using Sample Size Calculator (SSC) a Windows based software package (93) with the following assumptions: to detect an increase in exclusive breastfeeding for 6 month from 52% to 72% (44), with 95% Confidence Intervals and 80% power, assuming an intra-cluster correlation coefficient of 0•1 from Ugandan study (94) for a cluster size of ten, it is calculated that we will need thirty six clusters. Adding 20% of the sample size for loss to follow-up, the final sample size is 432 pregnant women (216 in intervention, and 216 in control). Standard operating procedure will be prepared and used for data collection. Data will be entered into Epi-data version 3.1 and analyzed using STATA version 12.0. Frequency distributions will be run to identify outliers. In relation to the trial hypotheses, intervention group will be compared to the routine education using intention to treat analysis. For the outcomes, proportions of women who initiated BF timely and who exclusively breastfed their babies at 72 hours and at 6 month will be compared using odds ratios derived using logistic regression between intervention and control groups. In all analyses, adjustment will be made for clustering at the zone level since randomization was done at cluster level rather than individual level. Generalized linear models will be used to calculate prevalence ratios (PR) for the categorical anthropometric outcomes. In the longitudinal analysis, the correlated nature of the data from the repeated measurements will be taken into account by adjusting for repeated measurements in the same individual as well as the above-mentioned design effect resulting from the cluster randomization. A linear mixed effects regression model will be used in the longitudinal analysis of z-scores with cluster as a random effect and the repeated measurements in each child accounted for through a first order autocorrelation structure. Least Squares Means (LSM) are reported at 3, 6 months, corresponding to the scheduled data collection visits.