Promotion of Early and Exclusive Breastfeeding in Neonatal Care Units in Rural Rwanda: a Quasi-experimental Study

Background: Early initiation of breastfeeding after birth and ongoing exclusive breastfeeding for the rst 6 months improves child survival, nutrition and health outcomes. However, only 42% of newborns worldwide are breastfed within the rst hour of life. Small and sick newborns are at greater risk of not receiving breastmilk and often require additional support for feeding. This study compares breastfeeding practices in Rwandan neonatal care units (NCUs) before and after the implementation of a package of interventions aimed to improve breastfeeding, aligned with the Baby Friendly Hospital Initiative for small and sick newborns. Methods: A pre-post quasi experimental study was conducted at two District hospital NCUs in rural Rwanda from October 2017–December 2017 (pre-intervention) and September 2018–March 2019 (post-intervention). Only newborns admitted before their second day of life (DOL) were included. Data was extracted from patient charts for clinical and demographic characteristics, feeding throughout admission, and patient outcomes. Bivariate analyses were conducted using Fisher’s exact and Wilcoxon rank sum tests. Logistic regression was used to evaluate factors associated with exclusive breastfeeding at discharge following a backwards stepwise procedure. Results: Pre-intervention, 255 newborns were admitted in the NCUs and 793 were admitted in the post-intervention period. The percentage of infants who were exclusively breastfed on their day of birth, or day-of-life zero (DOL0) increased from 5.4% to 35.9% (p<0.001). For newborns discharged alive, the proportion exclusively breastfeeding increased from 69.6% to 87.0% (p<0.001). The mortality rate for all admitted newborns decreased from 16.1% to 10.5% (p<0.019). Factors associated with greater odds of exclusive breastfeeding at discharge included post-intervention time point (Odds Ratio (OR): 4.91, 95% Condence Interval (CI) 1.99-12.11, p<0.001), and admission for infection (OR 2.99, 95%CI 1.13-7.93, p=0.027). Home deliveries (OR 0.15, 95%CI 0.05-0.47, p=0.001), preterm delivery (OR 0.36, 95%CI 0.15-0.87, p=0.0260 and delayed rst breastmilk feed (OR=0.04 for DOL3 vs. DOL0, 95%CI 0.01, 0.35, p=0.004) address this, Partners In Health/Inshuti Mu Buzima (PIH/IMB) in collaboration with the Ministry of Health, implemented interventions for breastfeeding support for newborns and infants with feeding diculties in neonatal care units, which included training of health care providers, breastfeeding counselling for mothers and health system strengthening to promote early and exclusive breastfeeding. This study aims to compare breastfeeding rates of newborns admitted to district hospital neonatal special care units before and after the implementation of a package of interventions aimed to improve breastfeeding practices.


Background
Breastfeeding plays a paramount role in child survival, development and maternal health [1]. Early initiation of breastfeeding within one hour after birth and exclusive breastfeeding for the rst six months is recommended by WHO [2]. Immediate and early initiation of breastfeeding after birth is associated with better neonatal outcomes such as reduction of neonatal deaths, length of stay for sick infants and neonatal infections [3,4]. A meta-analysis showed that non-breastfed infants under 6 months of age have a 14 times higher risk of death, and partially breastfed infants have a 4.8-fold higher risk of death compared to predominantly breastfed infants [5].
Even though the majority of deliveries globally are attended by skilled health care providers, only 42% of newborns are breastfed within the rst hour of life worldwide [6]. Small and sick newborns requiring inpatient care after birth face unique challenges since the neonatal unit environment is not always conducive for initiation of early and exclusive breastfeeding, such as an infant requiring mechanical ventilation or being separated from the mother [7]. In addition, small and sick newborns may experience di culties breastfeeding, or are too immature or unstable to breastfeed immediately after birth, so mothers require specialized support to establish and maintain their milk supply [8,9]. Despite the strong recommendation from WHO that maternity and newborn services have trained and competent staff who can provide successful breastfeeding support to lactating mothers[10], they have noted that early breastfeeding is compromised by inappropriate procedures, such as infant-mother separation, performed by health care providers and outdated policies [6].
Many interventions exist at community and facility levels worldwide to promote early and exclusive breastfeeding. UNICEF and WHO launched the Baby-Friendly Hospital Initiative (BFHI) in 2009 to integrate breastfeeding with maternal and newborn care in hospitals. Through this multi-level approach, hospitals must implement the "Ten Steps to Successful Breastfeeding" to attain BFHI status, including having a breastfeeding policy, allowing rooming-in, and providing adequate training to staff [10,11]. Speci c recommendations for expanding BFHI to include the unique needs of small and sick newborns have also been developed [8,12]. Additionally, utilizing lactation consultants for breastfeeding support and lactation education has shown to increase rates of initiation, duration of any breastfeeding and exclusive breastfeeding compared to usual practices [13]. Peer counseling approaches have also demonstrated improved early and continuous exclusive breastfeeding [14].
In Rwanda, 81% of newborns are breastfed within one hour of life and 87% are exclusively breastfed up to six months [15], however little is known about rates among small and sick newborns. Data from a neonatal care unit (NCU) in rural Rwanda found a large proportion of children had breastfeeding di culties after discharge leading to use of infant formula and poor growth [16]. To address this, Partners In Health/Inshuti Mu Buzima (PIH/IMB) in collaboration with the Ministry of Health, implemented interventions for breastfeeding support for newborns and infants with feeding di culties in neonatal care units, which included training of health care providers, breastfeeding counselling for mothers and health system strengthening to promote early and exclusive breastfeeding. This study aims to compare breastfeeding rates of newborns admitted to district hospital neonatal special care units before and after the implementation of a package of interventions aimed to improve breastfeeding practices.

Study setting
We conducted this study in the Rwinkwavu District Hospital (RDH) and Kirehe District Hospital (KDH) neonatal care units (NCUs). RDH and KDH are Rwandan Ministry of Health public hospitals located in Kayonza and Kirehe Districts in the Eastern province of rural Rwanda. Both RDH and KDH have been supported by PIH/IMB, an international non-governmental organization, since 2005 and 2007, respectively. RDH supervises 8 health centers in its catchment area with a population of 215,555 and KDH supervises 16 health centers in its catchment area with a population of 384,776 [17], in addition to 2 health centers and over 60,000 people in a refugee camp in the catchment area [18]. The NCUs provide care for small and sick newborns and are equipped with incubators, radiant warmers, syringe pumps, phototherapy, oxygen and continuous positive airway pressure machines for the management of common neonatal conditions. They are staffed by nurses -with an average nurse to patient ratio of 1 to 8.5 [19] -and general practitioners, with mentorship by a pediatrician and midwife.

Intervention
Several inputs were introduced into the hospital neonatal care units to promote exclusive breastfeeding, including porridge for mothers and water lters to provide a high calorie, high protein supplement and ensure adequate hydration; pillows for more comfortable breastfeeding positioning; screens for mothers to breastfeed or express breastmilk privately; refrigerator and materials for storage of expressed breastmilk, and educational posters promoting exclusive breastfeeding and Kangaroo Mother Care (KMC). In addition to these inputs, a training was conducted in February 2018 called Working with Infants with Feeding Di culties delivered by two Speech and Language Therapists who are experts in infant feeding. A description of that training package and a case study from its implementation in Rwanda has been described elsewhere [20]. As a result of the training and in an effort to ensure sustainability of the skills learned during the training, each hospital hired two Expert Mothers to serve as peer counsellors to support mothers in assessing breastfeeding readiness, improve positioning and attachment, and create a breastfeeding-friendly, caring environment for mothers with a focus on one-to-one as well as group

Study Design and Population
We conducted a pre-post quasi-experimental study. We included all newborns admitted to the RDH and

Data Collection
Data from standardized neonatology patient charts is routinely collected by trained data collectors on a structured paper form and then entered in a Microsoft Access database. Types of data collected included infant's reason for admission, day of life on admission, relevant perinatal history, feeding data, weight gain, length of stay, and discharge outcomes.

De nition of Variables
Our primary outcome was exclusive intake of breast milk at the time of discharge among infants discharged alive. Day of life 0 (DOL0) referred to the child's day of birth. Newborns exclusively fed breastmilk was de ned as the feeding type recorded in the patient's chart as 'only breast milk', regardless of the method of feeding (i.e. via breast, via naso-gastric tube, etc.). Fed on breast was de ned as the method of feeding recorded in the patient's chart as 'only on the mother's breast' (i.e. not via cup, not via naso-gastric tube, etc.). Low birth weight (LBW) was de ned as any birth below 2,500 grams and premature births are births before 37 weeks gestation.

Data Analysis
We described socio-demographic characteristics of infants and their mothers, and clinical and feeding characteristics of infants using frequencies and percentages for categorical data and median and interquartile ranges for continuous data. We conducted bivariate analysis using Fisher exact test to compare the pre-and post-intervention periods for all categorical socio-demographic, clinical and feeding characteristics described for infants with data recorded. Wilcoxon Ranksum test was used for bivariate analysis of continuous variables for infants with data recorded. Then, we used multivariable logistic regression models to identify predictors associated with the outcome 'exclusive breastfeeding on discharge', built using backward stepwise procedures for all variables signi cant at α = 0.20 in bivariate analyses. All factors signi cant at the α = 0.05 signi cance level were retained in the nal model. The data were analyzed using Stata v.15.1 (Stata Corp, College Station, TX, USA).

Ethics
The study received ethical approval from the Rwanda National Ethics Committee.

Results
In total, 255 newborns were admitted in the neonatology units during the pre-intervention and 793 were admitted in the post-intervention periods (Table 1). There were no signi cant differences in admissions for prematurity at pre-intervention compared to post-intervention (40.0% vs. 40.6%, p = 0.940), maternal age (50.0% age 25-34 years vs. 43.7%, p = 0.25) maternal gravidity (27.1% primigravida vs 31.5%, p = 0.396), and age on admission (83.1% admitted on DOL0 vs. 85.4%, p = 0.422). Overall, in both time periods, infants delivered in health facilities was greater than 88%; however, there were slightly more nonfacility births in the post-intervention period (11.6%) compared to pre-intervention (7.1%, p = 0.024). Compared to the pre-intervention period, infants were also more likely than in the post-intervention period to be delivered by caesarean section (31.5% vs. 41.1%, p = 0.007), be male (39.5% vs. 58.9%, p < 0.001), weigh more than 1,500 grams (86.6% vs. 91.9%, p = 0.023), be born at 32 or more weeks of gestation (86.7% vs. 92.5%, p = 0.048) and be born to a primiparous mother (32.4% vs. 43.6%, p = 0.007). The percentage of infants who were fed on the breast on DOL0 increased from 5.8-35.6% (p < 0.001) and exclusively breastfed on DOL0 increased from 5.4-35.9% (p < 0.001, Table 2). On DOL1, feeding on the breast increased from 30.5-68.6% (p < 0.001) and exclusive feeding on breastmilk increased from 37.4-70.2% (p < 0.001). For newborns discharged alive, the proportion fed on the breast increased from 59.8-84.7% (p < 0.001) and the proportion exclusively feeding on breastmilk increased from 69.6-87.0% (p < 0.001). Introduction of rst breast milk feeding on DOL0 increased from 7.1% pre-intervention to 31.9% in post-intervention (p < 0.001). The median length of stay among infants admitted to the neonatal unit was reduced from 8 days in the pre-to 7 days in the post-intervention periods (p < 0.001) ( Table 3). The overall mortality rate for all newborns admitted decreased from 16.1% in pre-to 10.5% in post-intervention periods (p < 0.019). Mortality rate for LBW newborns reduced from the pre-intervention to post-intervention period (23.6% vs. 15.0%, p = 0.039), but remained similar among babies diagnosed with other conditions.

Discussion
Our study showed that a multi-level intervention, aimed at improving rates of exclusive breastfeeding in a hospital neonatology unit in rural Rwanda, increased early and exclusive breastmilk feeding, and also reduced length of stay and decreased mortality among small and sick newborns.
We found that from the pre-to post-intervention period, signi cantly more infants were fed exclusively on the breast and were exclusively fed breastmilk. We also observed earlier initiation of breastfeeding in the post-intervention and this earlier initiation increased the odds that an infant was discharged exclusively feeding on breastmilk. This is consistent with other studies as it is well known that if milk removal does not occur either by infant suckling or expression by hand or pump, milk secretion will start to decline around day 3 postpartum [21,22]. A study comparing milk expression within 1 hour after delivery to within 1-6 hours after delivery showed that the earlier expression group had earlier lactogenesis stage II (transition from colostrum to copious breast milk production) and resulted in higher milk volume [23].
Another study showed that milk volume on postpartum day 4 is predictive of having an adequate milk supply at 6 weeks [24]. These studies demonstrate the critical need for early expression of breast milk after delivery, whether the infant is able to breastfeed on the breast or if the mother expresses breastmilk and the infant receives breastmilk through other enteral feeding routes (i.e., cup, naso-gastric tube).
Mortality among newborns decreased from the pre-to post-intervention period, particularly among infants born LBW. The association between decreased mortality and exclusive breastfeeding has been well established in the literature and is often promoted as strong support for initiation of early and exclusive breastfeeding [1,[3][4][5][6]25].
Overall length of stay showed a signi cant reduction from the pre-to post-intervention period. While hospital neonatology units are meant to be an environment for infants to improve from various illnesses or conditions, long length of stay in hospitals also increases an infant's chance of contracting hospital acquired infections [26]. Therefore, the ability to reduce the length of stay for newborns may have an impact on the overall morbidity of the infant. While we did not measure morbidity in this study, early initiation of breastfeeding has been shown to reduce morbidity in newborns [3], which likely has a positive impact on the total length of hospitalization. Reducing the time families, particularly mothers, spend in hospitals can also have a signi cant impact on the mother's stress, and the family's economic situation [27,28]. Reduced length of stay was likely a secondary outcome of improved breastfeeding rates in the post-intervention period. The overall larger number of admissions with a birthweight greater than 1,500 grams and gestational age over 32 weeks in the post-intervention period may have also contributed to lower mortality and shorter length of stay.
We found that the location of where the infant is born was associated with whether they are discharged from the hospital exclusively breastfeeding or not. Infants born at health centers or in the home were less likely to be discharged exclusively breastfeeding, compared to those born in the hospital which has also been seen in other studies [29,30]. There may be many reasons for this. Infants born at home need to rst be transferred to the health center, and subsequently to the hospital which may delay the introduction of breastmilk for those infants, and subsequently impact whether the infant is discharged exclusively breastfeeding. The clinical staff at health centers may also be less experienced in caring for high-risk newborns, and may not follow essential newborn care practices and delay introduction of breastmilk since the infant needs to be clinically stabilized and then transferred to the hospital.
Infants born preterm were less likely to be discharged exclusively breastfeeding compared to infants born at term. Infants born preterm have unique feeding needs that require specialized interventions and management, and other studies have demonstrated similar ndings of reduced exclusive breastfeeding rates even among moderate to late preterm newborns compared to term newborns [31]. Similarly, infants admitted for infection risk or neonatal infection in our study had much higher odds of exclusive breastfeeding. These ndings are not surprising, as these infants are often term and feed easily. But notably, even when considering all of these factors in multivariate analysis, admission during the postintervention period was the strongest predictor of exclusive breastfeeding at the time of discharge with nearly double the odds of exclusive breastfeeding compared to the pre-intervention period. Other factors in the neonatal care unit environment may also interfere with early and exclusive breastfeeding, including delayed initiation of KMC, or skin-to-skin contact, especially for sick newborns [32]. We were unable to measure timing or duration of KMC but this is an area that warrants further attention to reduce breastfeeding barriers.
Our study has some limitations. First, we used routinely collected data for the study, which results in some missing data and reliance on clinician skills in completion of medical les. In addition, precise measurement of gestational age is a challenge in Rwanda like in other low-and middle-income countries where availability of ultrasound dating is limited. Due to the use of routine data, it was not possible to reliably discriminate between newborns with infection or those with risk of infection and so we included all of these newborns in our sample. We also had a small sample size of patients born with HIE and those born with extremely low birth weight, which prevented measurement of the impact of interventions on these subsets.

Conclusions
A multi-level breastfeeding intervention improves earlier initiation of breastmilk feeding, exclusive breastfeeding on discharge, reduced length of stay, and decreased mortality among infants admitted to hospital neonatal units in a rural African setting. Adoption of evidenced-based guidelines such as the Baby Friendly Hospital Initiative aimed at the unique needs of small and sick newborns and innovative interventions should be expanded and adapted in similar settings to improve outcomes for these infants. This study was approved by the Rwanda National Ethics Committee. Data was captured through review of routine records and so informed consent was not required.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was funded by the Primates World Relief and Development Fund and Global Affairs Canada, who provided the funding for the interventions and evaluations. MAITS provided funding for the Infants with Feeding Di culties trainers.