Breastfeeding education and support to improve early initiation and exclusive breastfeeding practices and infant growth: A cluster randomized controlled trial from a rural ethiopian setting

Although peer-led education and support may improve breastfeeding practices, there is a paucity of evidence on the effectiveness of such interventions in the Ethiopian context. We designed a cluster-randomized trial to evaluate the efficacy of a breastfeeding education and support intervention (BFESI) on infant growth, early initiation (EI), and exclusive breastfeeding (EBF) practices. We randomly assigned 36 clusters into either an intervention group (n = 249) receiving BFESI by trained Women’s Development Army (WDA) leaders or a control group (n = 219) receiving routine care. The intervention was provided from the third trimester of pregnancy until five months postpartum. Primary study outcomes were EI, EBF, and infant growth; secondary outcomes included maternal breastfeeding knowledge and attitude, and child morbidity. The intervention effect was analysed using linear regression models for the continuous outcomes, and linear probability or logistic regression models for the categorical outcomes. Compared to the control, BFESI significantly increased EI by 25.9% (95% CI: 14.5, 37.3%; p = 0.001) and EBF by 14.6% (95% CI: 3.77, 25.5%; p = 0.010). Similarly, the intervention gave higher breastfeeding attitude scores (Effect size (ES): 0.85SD; 95% CI: 0.70, 0.99SD; p < 0.001), but not higher knowledge scores (ES: 0.15SD; 95% CI: −0.10, 0.41SD; p = 0.173). From the several growth and morbidity outcomes evaluated, the only outcomes with significant intervention effect were a higher mid-upper arm circumference (ES: 0.25cm; 95% CI: 0.01, 0.49cm; p = 0.041) and a lower prevalence of respiratory infection (ES: −6.90%; 95% CI: −13.3, −0.61%; p = 0.033). Training WDA leaders to provide BFESI substantially improves EI and EBF practices and attitude towards breastfeeding. The design and methods used in this trial, the Breastfeeding Education and Support Intervention (BFESI), are described in detail elsewhere [20]. Briefly, the study involved a cluster randomized, parallel-group, single-blinded trial evaluating the efficacy of BFESI on EI and EBF practices, and infant growth in a rural Ethiopian setting. The study was conducted in the Manna district located in Jimma Zone in southwest Ethiopia, where there was no similar ongoing intervention or project. From the total of 78 sub-districts under Mana, 36 sub-districts were selected for the study. The 36 sub-districts selected for the study were randomly assigned to either an intervention group (n = 18) receiving the BFESI or a control group (n = 18) receiving the routine Ethiopian healthcare service. We used simple randomization with a 1:1 allocation to allocate sub-districts to either control or intervention. First, the 36 sub-districts were listed alphabetically and then they were sequentially numbered starting from 01 to 36. Then we generated 18 random numbers from those 01 to 36 using MS Excel 2010 and the districts with the selected random numbers were assigned to the intervention group, while the rest were assigned to the control group. The generation of the allocation sequence and the randomization of clusters were done by a statistician blinded to study groups and not participating in the research. Allocation concealment was not done for study participants, as they would know if they were in the intervention group or not. However, data collectors were masked to the sub-district allocation by not being informed of the allocation, not being part of trial implementers, and not being residents in any of the sub-districts. All pregnant women in the selected sub-districts were identified by reviewing the HEWs’ antenatal care logbook. Women in their second or third trimester of pregnancy, who were willing to participate with no intention of leaving the study area during the intervention period, were enrolled for the study between May and September 2017. Study exclusion criteria were the presence of severe mental illness that could interfere with consent and study participation, serious illness or clinical complications warranting hospitalization, the occurrence of maternal death, abortion, stillbirth, infant death, twin gestation, preterm birth (at <37 weeks gestation), or any child congenital malformation that could interfere with breastfeeding. The Ethiopian government introduced the health extension program in 2003 aimed at improving access to primary health care to rural communities through the expansion of health posts and training of HEWs [21]. After training and deployment to health posts, HEWs train model families on 16 health extension program elements over several weeks for 96 h. A woman who knows all the 16 packages and practices them is selected from the model family to lead other five women in her neighbourhood, supporting their adaptation of good practices, such as vaccinating their children, sleeping under mosquito bed-nets, building separate latrines, and using family planning [22]. Peer support is defined as the provision of emotional, appraisal, and informational assistance by a created social network member who possesses experiential knowledge of a specific behaviour or stressor, similar characteristics as the target population, and the ability to address a health-related issue [23]. With this in mind, for this trial, we selected WDA leaders who could read and write the local language, aged 24–39 years, with experience of motherhood as well as breastfeeding, being from the same community as the women they should support. The WDA leaders from the intervention communities were trained for five days as breastfeeding peer-supporters by a nutritionist and a nurse with prior training on breastfeeding. WHO/UNICEF/USAID manuals were used to develop Ethiopian training guidebooks in the Afan Oromo language [24,25,26]. Moreover, to equip WDA leaders with the ability to educate and support study participants, a handbook with counselling cards were translated and prepared from these manuals. The training involved classroom lectures, demonstrations, and role play. Use of the manual and the counselling cards was practised through role-playing in teams with feedback from peers. Follow-up and supervision were carried out monthly during scheduled visits, in addition to unannounced spot-checks. Every pregnant mother was given a form to tally the number and timing of the visits she received from the WDA leaders. During the supervision, the supervisors checked the tallied paper and collected it at the end of the intervention. Peer-supporters made home visits to women in the intervention clusters according to a pre-specified schedule [20]. During pregnancy, they made two home visits in the last trimester of pregnancy: during the 8th and 9th month. Visits after delivery were scheduled on the 1st or 2nd, 6th or 7th and 15th day, and thereafter monthly until the infant was five months. During the two antenatal visits, peer-supporters encouraged delivery at the nearby health centre, emphasized the importance of initiating breastfeeding within 1 h of delivery, feeding colostrum first, discouraging the use of traditional pre-lacteal foods (items given to newborns before breastfeeding is established such as raw butter, plain water and milk-other than breast milk), and post-lacteal foods in addition to advising them to eat one extra meal during pregnancy to support lactation. The discussions were combined with the use of educational materials and practical demonstrations on proper breastfeeding positioning and latching. During the first two weeks after delivery, peer-supporters emphasised frequent and on-demand breastfeeding, encouraged stopping any traditional pre-lacteal foods or post-lacteal food items if already given to the child. Besides, peer-supporters observed the positioning, latching, and feeding of the newborn, solving any breastfeeding problems and providing appropriate feedback, while encouraging the mothers to continue EBF for six months. During these visits, women were advised to eat two extra meals during lactation from a variety of foods available in their area to provide energy and nutrition for themselves and their babies as well as to secure sufficient breast-milk production. Starting from month one, in addition to the above components, peer-supporters emphasized techniques for preparing for work and management of breast-milk (breast-milk expression, storing breast-milk), discussed the lactation amenorrhea method, and other family planning options. Hands-on guidance was provided only when necessary. Personal cleanliness and domestic hygiene, hand washing before feeding, after going to the toilet, and after changing babies’ diapers, were promoted during each visit. The mothers were encouraged to ask questions related to any topic discussed. Peer-supporters also provided additional visits if women experienced breastfeeding problems such as engorgement, cracked nipple or insufficient breastmilk that prohibited them from continuing breastfeeding. In addition to the informational support described above, women also received an emotional, appraisal, and instrumental support (Table S1). The duration of each visit was typically 20–40 min. Women in the control group received the routine care offered by the HEWs and WDA leaders working in their cluster, similar to that received by women in the intervention group [22]. The current Ethiopian standard/routine prenatal and postnatal care by HEWs includes providing four focused prenatal visits, developing an individualized birth preparedness and complication readiness plan, accompanying a woman to a health facility during delivery, and conducting four postnatal visits [27]. Moreover, as part of the community-based nutrition program, HEWs are expected to deliver the following key breastfeeding and nutrition messages to mothers during the monthly growth monitoring sessions or during antenatal or postnatal care visits: the importance of antenatal care, maternal nutrition during pregnancy and breastfeeding, early initiation of breastfeeding, proper positioning and attachment, EBF for six months, breastfeeding on demand, and complementary feeding [24]. WDA leaders also support the HEWs by educating and mobilizing communities to use key available health services, including dissemination of essential health messages such as infant and young child feeding practices. The primary study outcomes were rates of EI and EBF for six months and infant growth. Secondary outcomes included maternal knowledge and attitude towards breastfeeding at the endline. We further included morbidity for common childhood illnesses over the past two weeks as an additional secondary outcome, although this was not considered a priori in the study protocol. Data were collected at three time-points including at study enrolment (May-September 2017), at around 1 month (±2 weeks), and 6 months (±2 weeks) postpartum. At baseline, data on demographic and socio-economic characteristics, information on various maternal factors, and maternal knowledge and attitude towards breastfeeding were assessed. At one month postpartum, information about pregnancy outcome and other study exclusion criteria, and maternal practice on early initiation of breastfeeding including information about colostrum and pre-lacteals feeding were gathered. Data collected at around six months postpartum included maternal knowledge and attitude towards breastfeeding, EBF practice, infant anthropometry measurements, and morbidity. Data were collected by trained nurses and all instruments used were Afan Oromo language translations of English versions. Gestational age was determined based on the last menstrual period (LMP). LMP was self-reported at baseline during enrolment. If women did not remember the exact date of the month, the 15th day of the month was used. First, we determined the estimated delivery date from the LMP and then subtracted the difference between the estimated delivery date and the actual delivery date from 280 days. Finally, we divided the total number of days by 7 to determine the gestational age in weeks. Maternal knowledge and attitude towards breastfeeding were assessed using Afan Oromo (AO) versions of the Breastfeeding Knowledge Questionnaire (BFKQ) and the Iowa Infant Feeding Attitude Access Scale (IIFAS), which were culturally adapted and validated in the same population. Details of the adaptation process and psychometric properties of both tools are reported previously [28]. Both the BFKQ-AO and the IIFAS-AO had an acceptable level of internal consistency with Cronbach alpha values of 0.79 and 0.72, respectively. Since our breastfeeding knowledge questionnaire adopted from Malaysia does not have a cut-off point suggesting an optimal knowledge level, we used a cut-off of ≥the median for a good level of knowledge. For attitude, we used the recommended cut-off of ≥70 scores for a positive attitude towards breastfeeding [29]. EI of breastfeeding and EBF practices were defined according to the WHO Infant and Young Child Feeding indicators [30]. Accordingly, mothers were asked how soon after delivery they put their newborn to the breast with responses ≤1 h considered as optimal EI practice. EBF practice was defined as feeding infant no other food or drink, not even water, except breast milk for the first six months of life, but allowing the infant to receive oral rehydration solution, drops, and syrups (vitamins, minerals, and medicines). The following questions were asked to evaluate EBF practice: (i) For how many months did you exclusively breastfeed (name); (ii) Are you currently giving your infant any food/drink other than breast milk? (iii) If yes to ii, we asked the age at which the food/drink was started. Thus, we used questions i, ii, and iii to determine if the child had been exclusively breastfed for six months since birth. Anthropometry measurements of infant length, weight, and mid-upper arm circumference (MUAC) were done in duplicate by two independent teams of data collectors and recorded on separate forms so that the first measurement could not influence the second. Then, a supervisor compared the duplicate measurements, and both teams repeated the measurement whenever there was a difference of ≥0.5 kg for weight, ≥1.0 cm for length, or ≥0.5 cm for MUAC. Recumbent length was measured using a length board (SECA 417) with a precision of 0.1 cm. Weight was measured together with the mother using an electronic scale (SECA 876) to the nearest 1.0 g. MUAC measurement was taken on the left arm to the nearest 0.1 cm using flexible non-stretchable measuring tapes (SECA 212). Instruments were calibrated before each measurement session. The average value of the duplicate anthropometry measurements was used for analysis, and length-for-age (LAZ), weight-for-length (WLZ), and weight-for-age (WAZ) z scores were calculated based on the WHO 2006 Child Growth Standards using the Stata zscore06 command [31]. Child stunting, wasting, and underweight were determined from the respective z-score values using a cut-off <−2 SD from the median. Mothers were asked to recall infant morbidity during the two weeks before the endline follow-up. Diarrhoea was defined as three or more liquid or semisolid stools within a day. Fever was determined by the mother’s report. Acute respiratory infection was defined as a combination of fever and cough. Serious illness was the occurrence of an illness that required medical attention, i.e., hospital or health centre visits. Household wealth status was assessed using a 16-item household asset questionnaire adapted from the Ethiopian Demographic and Health Survey [11], and principal components analysis was used to generate a household asset score. Household food security status was assessed using the Household Food Insecurity Access Scale from the Food and Agriculture Organization [32]. The sample size was calculated by taking into account an assumed intracluster correlation coefficient, the expected effect, and the power of the study [20]. In Ethiopia, the current overall rate estimate of EBF was 58%, according to the EDHS 2016 report [11]. We hypothesized that the BFESI would increase the prevalence of EBF to 78% in the intervention group. To detect a 20% difference in the rates of EBF, with 80% power, and 5% type I error, assuming an intra-cluster correlation of ρ = 0.1 [9], 10 pregnant women were needed per study cluster. We inflated the sample size from 346 to 432 to accommodate for a potential 20% attrition rate. Data were entered using Epi-data version 3.1 (EpiData Association) and consistency checks and statistical analysis were conducted using Stata version 13.0 (StataCorp LLC, College Station, TX, USA). Descriptive statistics were summarized as frequency and percentage for the categorical variables, and mean and standard deviation for the continuous variables. Maternal knowledge and attitude scores were standardized to z scores based on the distribution of the data. The effects of the intervention were estimated using linear regression models for the continuous outcomes, and linear probability models for the binary outcomes. The use of linear probability models for binary outcomes is well established and allows for a straightforward interpretation of the average intervention effect expressed as a risk difference using percentage points [33]. However, for the rare outcomes of child stunting, wasting, and underweight, we fitted logistic regression models. In all models, we applied a robust standard error estimation taking into account the clustering of subjects by sub-districts. Both unadjusted and adjusted group differences were estimated with covariates used for adjustment, including maternal age, educational status, wealth index, parity, household food insecurity status and IIFAS score at baseline. Analyses were done following the intention-to-treat principle. For this purpose, we conducted multiple imputations of missing data using chained equations under the assumption of missing at random. To estimate the regression coefficients, we ran a hundred imputations of missing data for cases lost-to-follow-up. Statistical significance was declared at p-value < 0.05.