Facility-based care for moderately low birthweight infants in India, Malawi, and Tanzania

Globally, increasing rates of facility-based childbirth enable early intervention for small vulnerable newborns. We describe health system-level inputs, current feeding, and discharge practices for moderately low birthweight (MLBW) infants (1500-<2500g) in resource-constrained settings. The Low Birthweight Infant Feeding Exploration study is a mixed methods observational study in 12 secondary- and tertiary-level facilities in India, Malawi, and Tanzania. We analyzed data from baseline facility assessments and a prospective cohort of 148 MLBW infants from birth to discharge. Anthropometric measuring equipment (e.g., head circumference tapes, length boards), key medications (e.g., surfactant, parenteral nutrition), milk expression tools, and human milk alternatives (e.g., donor milk, formula) were not universally available. MLBW infants were preterm appropriate-for-gestational age (38.5%), preterm large-for-gestational age (3.4%), preterm small-for-gestational age (SGA) (11.5%), and term SGA (46.6%). The median length of stay was 3.1 days (IQR: 1.5, 5.7); 32.4% of infants were NICU-admitted and 67.6% were separated from mothers at least once. Exclusive breastfeeding was high (93.2%). Generalized group lactation support was provided; 81.8% of mother-infant dyads received at least one session and 56.1% had 2+ sessions. At the time of discharge, 5.1% of infants weighed >10% less than their birthweight; 18.8% of infants were discharged with weights below facility-specific policy [1800g in India, 1500g in Malawi, and 2000g in Tanzania]. Based on descriptive analysis, we found constraints in health system inputs which have the potential to hinder high quality care for MLBW infants. Targeted LBW-specific lactation support, discharge at appropriate weight, and access to feeding alternatives would position MLBW for successful feeding and growth post-discharge.


Introduction
Globally, nearly 20 million infants are born with low birthweight (LBW) (<2500g) each year. LBW infants represent a heterogeneous group who may be preterm (<37 weeks gestation) and/or small-for-gestational age (SGA; weight for gestation <10th percentile) [1]. LBW infants are at increased risk of morbidity, infections, growth deficits, developmental delays, and mortality [2].
The global increase in rates of facility-based childbirth (currently >75% of births) [3] enables greater opportunity to identify LBW and vulnerable infants near the time of birth to provide early, high quality, and appropriate care. Although 91% of LBW infants are born in low-and middle-income countries (LMIC) and the majority are moderately LBW (MLBW) (1500-<2500g) [4][5][6]; research is predominantly from high-income countries and focuses on very LBW (<1500g) or preterm infants [7]. MLBW infants are at higher risk for complications than normal weight (>2500g) infants and have been shown to be at greater risk of sepsis, temperature instability, and low blood sugar [8]. Currently, there is limited knowledge of facilitybased feeding and discharge practices for MLBW infants, hindering proper management.
To help address these gaps, the Low Birthweight Infant Feeding Exploration (LIFE) study aimed to describe feeding and growth of MLBW infants in India, Malawi, and Tanzania [9]. In this analysis, we examined the health system inputs in the LIFE study facilities and described the current health facility care and feeding practices for MLBW infants from birth to discharge. The specific objectives were to: (1) examine the health system inputs available to support the care and feeding of MLBW infants; (2) understand the overall experience of facility care for MLBW infants with respect to the location and duration of their care and discharge practices; and (3) describe feeding practices from birth to facility discharge among MLBW infants.

Study design
LIFE is a formative, multi-site, observational cohort using a convergent parallel, mixed-methods design examining the feeding practices, growth, and health outcomes of MLBW infants (1500g� birthweight <2500g) in LMIC (Clinicaltrials.gov NCT04002908) [9]. Here, we present results from two quantitative data collection streams from LIFE: (i) a baseline facility needs assessment describing structural and service inputs of facilities where infants were enrolled, and (ii) a prospective in-facility observational cohort. Table 1 provides an overview of each data stream used in this analysis of the larger LIFE study (Table 1). Table 1. Overview of study objectives and data collection.

Data Stream
Facility needs assessment In-facility observational cohort

Aim
To describe the health facility inputs as well as current facility care and feeding practices for MLBW infants from birth to discharge in LMICs in an effort to gather foundational knowledge and inform future interventions

Objectives
Health facility inputs Examine the health system inputs (e.g., equipment, supplies, and human resources) available to support the care and feeding of MLBW infants

Facility care practices
To understand the overall experience of facility care for MLBW infants with respect to the location and duration of their care and discharge practices Facility feeding practices To describe feeding practices from birth to facility discharge among MLBW infants

Study facilities
The LIFE study was conducted in 12 health facilities in four sites across three countries (Dares-Salaam, Tanzania; Lilongwe, Malawi; Belgaum and Davangere, Karnataka State, India; and Cuttack, Odisha State, India) [9]. Each site had two to five study facilities chosen based on delivery volume, capacity to care for LBW infants, and willingness of facility leadership to participate. Study facilities included secondary and tertiary-level public-sector hospitals located in urban areas; in India-Karnataka, three private health facilities were also included ( Table 2). The facility needs assessment was conducted between August-September 2019, prior to observational cohort enrollment.

Study participants
After the needs assessment was completed in each facility, MLBW infants were eligible for the in-facility observational cohort if they were born or presented at one of the 12 study facilities during the enrollment period from August 2019 to April 2020. Each site enrolled between 35-40 infants in order to describe in detail the care of infants while in the facility. Our sample size was determined by considering timeline and budget constraints to conduct the intensive infacility observation. Using a posthoc analysis with a sample size of 148 infants and 95% confidence, we were able to have a margin of error of 8% for common outcomes (50%) and 5% for an important feeding outcome of breastfeeding (90%). Infants were excluded if they had a birthweight <1500g or �2500g (n = 3); did not have maternal consent (n = 16); were born with congenital anomalies that could interfere with feeding (n = 3); were born to young mothers (n = 1) (i.e., <18 years in Tanzania and India; in Malawi, <16 years old if married or <17 years old if unmarried); or were born outside the facility (n = 10) [9]. Infants were also excluded if there was a maternal or newborn death, including stillbirth (n = 1) or death of a twin (n = 1).

Ethics & consent
The LIFE study was approved by 11 ethics committees in India, Malawi, Tanzania, and the United States [9]. For the facility assessment, facility leadership gave verbal consent; for cohort enrollment, women provided written informed consent.

Patient and public involvement
The LIFE study team involved clinicians, researchers, and community stakeholders who are familiar with the cultural context, study setting and populations in research design and study. In addition, study tools were piloted with mothers, community members, and health care providers as a way to capture culturally appropriate language, and ensure that surveys were acceptable.

Data collection
Facility needs assessment. The facility assessment documented health facilities' policies and capacity to provide care for LBW infants, with a focus on infant feeding and discharge practices. It assessed vital statistics (i.e. volume of births, volume of LBW births), structural inputs (i.e. infrastructure such as number of NICU beds, facility type, electricity, backup generators), human resources, and equipment available for MLBW care and feeding for each study facility [10,11]. A study team member administered the assessment through direct observations, record/register reviews, and staff consultations, where needed. The level of care was defined based on recommendations from the American Academy of Pediatrics (AAP) (2012) [11], and guidance from the WHO/UNICEF Survive & Thrive report [12]. In our study, primary level facilities provide essential newborn care that includes immediate drying, skin to skin contact, early initiation and support of breastfeeding, outpatient care services and management and referral to higher level care. Secondary level facilities provide primary level care and address basic signal functions for small and sick newborns including the provision of extra warmth using incubators or radiant warmers in addition to Kangaroo Mother Care, administration of oxygen with continuous positive airway pressure (CPAP), resuscitation, detection and management of complications (such as neonatal encephalopathy (NEC), jaundice, infection, hypoglycemia) and access to a physician on call with specific neonatal skills in addition to specialized nurses/midwifery staff available at all times (24/7). Tertiary level facilities have a designated intensive care ward with secondary level capabilities as well as equipment for mechanical/assisted ventilation, specialized feeding equipment (e.g., total parenteral nutrition), nurses and doctors with specialized competencies in neonatal care available at all times (24/7), a neonatologist on call and access to other specialist doctors including pediatric surgery capabilities radiology, cardiology, neurology, and ophthalmology. In-facility observational cohort. The in-facility observational cohort closely examined current infant feeding and discharge practices for MLBW infants. Mother-infant pairs were screened, consented, and enrolled within six hours of birth by trained study personnel and followed during daylight hours (7:00am to 5:00pm, daily in India and Tanzania; on weekdays in Malawi) until facility discharge. Demographics, birth characteristics, infant feeding patterns, infant feeding competency, and neonatal weight changes were documented from birth until facility discharge.
Demographic information was extracted from medical records. Gestational age (GA) was based on ultrasound, last menstrual period (LMP), or fundal height recorded in a patient's chart with priority given to first trimester ultrasound followed by documented LMP. Using the WHO growth standards for term infants [13,14] and the INTERGROWTH-21st growth standards for preterm infants [15], infant LBW type were identified: preterm-small for gestational age (preterm-SGA), preterm-appropriate for gestational age (preterm-AGA), preterm-large for gestational age (preterm-LGA), and term-SGA [1]. Mother-infant separation, based on maternal report, was defined as the time the mother and infant were not sharing a room.
Check mark denotes present; blank cell denotes not present *Newborn level of care definition based on guidance from AAP [11] and WHO/UNICEF Survive and Thrive [12] **This facility does not have a formal NICU, but it has access to four warmers and basic resuscitation and oxygen facilities https://doi.org/10.1371/journal.pgph.0001789.t002 Trained study nurses directly observed feeding sessions and interviewed mothers about practices between observations regarding feeding patterns (e.g., direct breastmilk, expressed breastmilk, donor human milk (DHM), animal milk, formula, or another liquid), and lactation support (e.g., verbal advice, physical support, timing) at each visit. Observation started within six hours after birth and occurred every three hours for the first seven days, twice daily for days 8-14, and then once a day for unstable infants or every three days for stable infants until discharge. Both direct observation and maternal report were used to reduce the possibility of bias and allow for triangulation of data. The Preterm Infant Breastfeeding Behavior Scale (PIBBS), a 9-question validated tool, assessed feeding competency based on direct feeding observation by trained study nurses measuring infant rooting efforts, latch effort, latch duration, sucking duration, and swallowing; scores <15 (out of 20) signified poor feeding competency [16]. Timing of breastfeeding initiation was collected at the baseline observational visit within the first six hours after birth. Early initiation of breastfeeding, a validated Infant and Young Child Feeding indicators for feeding practices [17], was calculated as the frequency of infants who were breastfed within less than 1 hour.
Birthweight measured by clinical staff was used to assess MLBW infant study eligibility. After study enrollment and within six hours of birth, trained study staff measured study specific birthweight and other infant anthropometrics, in triplicate, using standardized and calibrated study specific equipment at birth. Measurements were repeated just prior to discharge [9]. The average of the two (out of the three) closest measurements was used for analysis. Birth and discharge weight were used to calculate absolute and proportional weight change [18,19].

Data analysis
Descriptive statistics included mean, standard deviation (SD), median and interquartile range (IQR) for continuous variables, and frequencies and percentages for categorical variables. Chisquared test was used to determine statistical differences between preterm birth and geographical region. Cochran-Mantel-Haenszel (CMH) chi-square test was used to compare differences of lactation support and counseling by sex, LBW type, and NICU admission after controlling for study site. We analyzed data by location of care (NICU or postpartum ward) [20]. A oneway analysis of variance was performed with a Bonferroni correction to compare differences in the mean gestational age by LBW type. Wilcoxon rank sum test was used to determine the difference between median birthweight by site, median length of stay differences by location of care (NICU vs postpartum ward), sex and LBW type. Crude binomial regression models, adjusted for clustering by health facility with compound symmetry and a log link function, were used to estimate relative risk ratios (RR) for the relationship between admission to the NICU and LBW type or birthweight bands (1500-<1800g, 1800-<2200g, 2200-<2500g) as well as the association between early initiation of breastfeeding (<1 hour) and mother-infant separation. Due to the small sample size, intended mainly for descriptive analyses, we did not adjust for any potential confounders in our analyses and only adjusted for clustering by facility. Exploration of the data demonstrated that missing data was rare; however, data were likely missing randomly or due to data collection difficulties; we excluded missing data from analysis. Data were analyzed using SAS 9.3 (SAS Institute. SAS Software Suite. Cary, NC: SAS Institute; 2020) and Microsoft Excel.

Facility characteristics
The facility assessment was completed prior to the in-facility observational cohort enrollment to understand currently available facility resources, demonstrating variation in access to general equipment, medications, infant feeding supplies, and level of care across the facilities ( Table 2). Weighing scales were mostly available, but length assessment tools were not.
All 12 facilities were accredited by the Baby-Friendly Hospital Initiative (BFHI) and prioritized breastfeeding [21]. All had supplies to support infant feeding including paladais and cups/spoons; most facilities (11/12) had nasogastric tubes. Infant formula was only used when a mother was absent or unable to express breastmilk. Half of the facilities had powdered infant formula in stock; two facilities had preterm formula; and none had pre-mixed liquid formula. Reportedly, formula was often purchased by families if required for their infant.
All 12 facilities followed formalized discharge policies based on national guidelines; 9/12 had facility-specific guidelines. At the time of discharge in all facilities, infants had to be feeding adequately, hemodynamically stable, and able to consume breastmilk (directly or expressed). Although not the sole criteria, policy-stated minimum discharge weight criteria varied by site: 2000g in Tanzania, 1800g in India, and 1500g in Malawi.

PLOS GLOBAL PUBLIC HEALTH
Facility-based care of moderately low birthweight infants

Feeding from birth to facility discharge
Practices. Of the 148 enrolled infants, one infant in Malawi did not have any feeds prior to discharge at two days postpartum; the remaining 147 infants had feeding observed from birth to discharge. The majority, 93.2% (137/147), were exclusively fed human milk [mom's own milk or DHM], 6.1% (9/147) were given mixed milk feeding [formula and breastmilk], 2.0% (3/147) were given other liquids (no specific information available), and 0.7% (1/147) was given only formula. Mixed milk feeding was more common in India-Odisha (6.1%, 6/35) compared to Tanzania (2.5%, 1/40), Malawi (2.9%, 1/34), and India-Karnataka (2.6%, 1/40). Provision of feeds other than human milk and formula (e.g., provision of other foods, teas, or liquids) were not observed. Based on maternal report, three infants were given other unknown liquids. Otherwise, feeding information (direct breastfeeding, milk expression, and/or infant formula use), obtained from direct observations was highly correlated with maternal selfreport (Fig 1).
Of the 82 mothers with available feeding initiation data, 81.7% (67/82) initiated breastfeeding within 1 hour of birth. Early initiation varied by infant LBW type: 90.0% of preterm-SGA, 59.1% of preterm-AGA, 25.0% of preterm-LGA, and 95.7% of term-SGA infants. This variation was likely related to mother-infant separation and infant prematurity with preterm-AGA and preterm-LGA infants averaging 33.5 and 30.1 weeks gestation respectively, preterm-SGA infants at 35.6 weeks and term-SGA at 38.9 weeks. Among infants separated, 69.0% (29/42) initiated early breastfeeding, whereas 95% (38/40) of non-separated infants initiated early breastfeeding. After adjusting for clustering by facility in a crude binomial regression model, infants who were separated from their mothers were significantly less likely to initiate early breastfeeding compared to those infants who were not separated [RR: 0.26, 95% CI: 0.10-0.42, p = 0.0013].
Lactation support and counseling. Overall, 81.8% of mothers reported receiving at least one lactation support and counseling session; yet, the frequency, provider, and type of counseling varied by site ( Table 4). Two or more sessions were provided to 56.1% of dyads, typically in a group setting. Support was primarily provided by healthcare providers (90.1%) or family members (27.3%). Type of support varied by site, including help with positioning of the infant, latch, milk expression, and feeding with a cup/paladai (Table 4). Counseling, as reported by mothers, did not vary by infant sex (χ 2 cmh = 2.85, p = 0.09), LBW type (χ 2 cmh = 0.004, p = 0.95), or NICU admission (χ 2 cmh = 0.87, p = 0.35). Mothers reported being the primary decision-makers for breastmilk feeding. Clinicians played an important role in influencing the mother to feed infants expressed breastmilk (68.1%), DHM (100.0%), and infant formula (100.0%). For the three infants whose mothers reported to feed 'other' unknown liquids, the mothers (n = 2) or family members (n = 1) made the decision.
Breastfeeding competency. Feeding competency assessments showed significant improvement from baseline to facility discharge across all sites (S2 Fig). The average PIBBS score at baseline was 11.9 (±4.3) and increased to 14.2 (±3.3) within 24 hours prior to discharge; 58.6% of infants were discharged with low feeding competency (PIBBS score �15). There were no significant differences in the mean discharge PIBBS scores by lactation support, sex, LBW type, size-for-gestational age, or GA.

Discharge procedures
Based on national and facility policy, infants met certain criteria around feeding, stability, and weight to be discharged. Of the 140 infants assessed at discharge, 97.1% were fed only breastmilk (69.3% direct, 7.1% expressed only, 20.7% both direct and expressed), 2.7% fed formula only, and 0.7% fed a mix of formula, direct and expressed milk. Weight change between birth and discharge was minimal ( Table 5). On average, infants lost 2.4% (± 5.8%) of their birthweight, varying by length of stay. Overall, 5.1% (7/138) lost >10% of their birthweight at the time of discharge. However, 18.8% of infants were discharged with weight below the site specified discharge criteria: 20.6% (7/34) of infants in India-Karnataka and 5.7% (2/35) with weights <1800g; 40.0% (16/40) in Tanzania had <2000g; (3.5%) (1/29) in Malawi <1500g. All infants were stable at the time of discharge.

Discussion
MLBW infants represent a heterogeneous group with different needs and implications for care and utilization of limited resources. Across the 12 facilities in 3 countries, we observed high adherence to EBF and relatively short lengths of stay (average 3 days), yet notable levels of mother-infant separation and gaps in universal and repeated lactation support or counseling [22][23][24].
Uptake and adherence to EBF was high, aligning with global recommendations and facility standard protocols [22,25]. More than 80% of infants had early initiation of breastfeeding, which is higher than the global average (57.6%) [26][27][28]. Further, 97.1% of infants were  discharged EBF, again higher than expected [29]. In our study, EBF was supported with utilization and promotion of breastmilk expression. Additional methods to aid with lactogenesis and support milk expression, including safe storage and increased availability of pumps in the facility, may be beneficial. Nearly 30% of infants were fed expressed breastmilk at discharge, which may be challenging to sustain at home given the need for clean water to wash breast pump materials to safely feed expressed breastmilk [30]. While uptake of EBF was excellent, health system constraints (e.g., bed capacity, human resources, limited specialized LBW training) likely led to misalignment of practice and policy and limited optimization of in-facility care, demonstrated by <60% receiving 2+ lactation support sessions, >66% of dyads experiencing maternal-newborn separation, and discharge of 18.8% of infants below minimum weight thresholds. Nurse-to-patient ratio guidelines from medical and nursing societies range from 1:3 for the lowest acuity care (level 1) to 1:1 for the highest acuity care (level 5) [31][32][33]. Only three facilities met nurse-to-patient guidelines for the lowest acuity care at level 1 facilities (1:3); some facilities had ratios as low as 1:35 infants. Increasing health worker capacity has been identified as a core standard for improving quality of maternal and newborn care in health facilities [34], meeting global recommendations [22], and improving neonatal outcomes [35,36].
Lack of universal and hands-on provision of lactation support and counseling was possibly driven by the high levels of mother-infant separation, staff shortages, and lack of specialized training on LBW lactation support among providers, which are well-documented barriers to support of EBF [37][38][39][40][41]. Unresolved breastfeeding problems, highlighted by poor feeding competency scores, can lead to growth faltering, reduced duration of EBF, subsequent hospitalizations, and mortality [42,43]. Group counseling and support in the facility and in-home peercounseling interventions have shown improved EBF initiation and duration [41]; these strategies could be adapted specifically for LBW infants to improve in-facility support.
Further, breastmilk alone may be insufficient to meet the nutritional needs of MLBW infants [42,[44][45][46]; most facilities did not have alternatives readily available. Only one facility had access to parenteral nutrition; one site supplemented infants with micronutrients; protein supplementation and fortification of human milk were not used at any site. When appropriate and safely administered, DHM and formula can provide a bridge to breastfeeding [46][47][48]. However, a human milk bank was only at one facility and formula availability was limited; a potential barrier of formula use was the cost to families. Finally, of note, we found a high prevalence of discharge among infants still demonstrating postnatal weight loss (Table 5) and/or discharge prior to reaching minimum weight benchmarks as stipulated by facility discharge policies. Weight loss among LBW and vulnerable infants is of critical importance and requires additional in-facility care measures including daily monitoring of anthropometrics, clinical documentation, and plans for continuity of care for post-discharge monitoring to ensure LBW infants do not become further growth-restricted.
Our study addresses key knowledge gaps as the majority of published research focuses on very LBW/preterm infants in high-income settings despite the predominance and risk of MLBW infants [7]. We present detailed data from three diverse countries and twelve hospitals. Multiple data collection methods, including facility needs assessment, direct observations, and maternal reports, allowed for triangulation of information.
We do have some notable limitations. We focused on MLBW infants, which preselected a preterm-AGA population that is less mature than a preterm-SGA group, (i.e., most early preterm-SGA infants have birthweights <1500g and were therefore ineligible). In addition, our study population included infants born at secondary and tertiary facilities and thus cannot be generalized to infants born at lower-level facilities or at home. The study sample size allowed for a deep understanding of the feeding experiences and care of LBW infants and resulted in a rich descriptive data at each site. However, this sample does not represent all LBW infants and, since this was intended largely as a descriptive paper, our modeling analyses are limited. Statistically significant findings should be interpreted cautiously as we could not control for potential confounders, and observed associations may be spurious. Further studies need to confirm our findings. Additionally, GA source differed by site which could cause LBW type misclassification; however, we use prioritization criteria for choosing the source of GA, namely ultrasound at first trimester when available over other sources followed by LMP in medical charts. Finally, observation by study staff may have altered participant behaviors; however, we had multiple observations over time and had specific research nurses gather information rather than clinical staff.

Conclusion
With increases in facility-based childbirth, critical opportunities exist to set up MLBW infants, their mothers, and families to thrive post-discharge. Given the renewed investment and recent guidelines in the care of small and sick newborns [49,50], the global community must identify gaps and at-risk groups, and modify strategies to support the adherence to policies like BFHI, zero-separation, and minimum discharge weights. Our study fills a critical gap in foundational knowledge on the health system inputs and current feeding and care practices for MLBW infants in resource-constrained settings. This is a group for whom evidence is limited, but the potential to survive and thrive is high [4,51,52]. Our findings highlight areas of focus to improve the quality of care delivered to MLBW infants including health systems improvements to increase human capacity, staffing and training, to improve breastfeeding alternatives availability, and to support adherence to discharge policies, which would improve motherinfant experience of care by reducing mother-infant separation, initiating breastfeeding early, and providing lactation support and counseling specialized for MLBW.