Introduction

Neonatal deaths and essential newborn care

Globally, ~3.1 million children die each year in their neonatal period (first 28 days of life).1 Half of these neonatal deaths occur within the first 24 h of delivery, whereas 75% occur in the early neonatal period (0 to 6 days after delivery) because of preterm births, severe infections and birth asphyxia.1 Essential newborn care (ENC) practices can contribute to decreasing the incidence of neonatal morbidity and mortality.2 Such ENC practices, as recommended by the World Health Organization, include drying (wiping) and wrapping the newborn immediately after birth, initiating skin-to-skin contact, dry cord care (not applying any potentially harmful substance to the umbilical cord), immediate initiation of breastfeeding and delayed bathing (for at least 6 h).3

ENC practice in low and lower-middle income countries

A low coverage of ENC practices has been reported particularly for home deliveries with an unskilled birth attendant in low and lower-middle income countries (LLMIC).4, 5, 6 Based on home delivery data from Nepal, a study found that ~58% of newborns were immediately dried, 60% were immediately wrapped and 25% were bathed after the first 24 h of birth, while 64% were breastfed within 1 h of birth.5 A low uptake of ENC practices along with other sociodemographic factors has also been observed as a major cause of neonatal mortality in sub-Saharan Africa. For example, a study in rural southern Tanzania reported that ~41% of home-born babies were dried and 28% were wrapped within 5 min of birth, whereas only 19% were bathed after 6 h of birth.6

Rationale of this study

Examining the impact of specific ENC practices on neonatal mortality may provide key information for planning and implementing interventions that aim to improve newborn health outcomes in LLMIC. However, no studies have analyzed the extent to which each of the individual interventions of ENC can reduce neonatal mortality. A review of the literature also identified the limited availability of quality studies on the impact of ENC practices on maternal and newborn health outcomes in LLMIC.7 Against this background, this paper investigates the effect of ENC practices on neonatal mortality.

The focus of this study is on Bangladesh, which, similar to other LLMIC, exhibits a high neonatal mortality rate (32 per 1000 live births). Approximately 71% of births in the country are delivered at home, with more than half of the births assisted by traditional birth attendants who do not have any formal training.8 Only 2% of these newborns have been found to receive all elements of ENC.8 An incomplete or poor delivery of ENC interventions can contribute to high neonatal mortality in the country. Thus, this analysis may be useful in further embedding ENC in health workers’ practice and mobilizing community support. Such an analysis may also help to improve the design and implementation of ENC in Bangladesh, which may be transferrable to other LLMIC. Therefore, this study aims to analyze the impact of ENC practices on the mortality of home-delivered neonates in Bangladesh.

Materials and methods

Data source

Data were obtained from the 2011 Bangladesh Demographic and Health Survey; a periodic cross-sectional survey.8 Participants were recruited via a two-stage stratified sample design with the aim of collecting nationally representative demographic and health information. The Bangladesh Demographic and Health Survey was administered to a total of 17 141 households. In total, 17 842 ever-married women in these households aging from 12 to 49 years were interviewed. Male respondents were interviewed only from one-third of the households, which included 3997 male observations. The survey collected data using separate questionnaires for the household, for women and for men. All information obtained in the survey was self-reported and collected by face-to-face interviews. Informed verbal consent was taken from all surveys participants. Details of the design and data collection procedures were described in NIPORT et al.8

Study participants

Analyses were restricted to data on live-born infants who were delivered at home. Only the most recent birth by a mother within the preceding 3 years of the survey was considered because detailed information on ENC practices was available only for these births. A total of 3190 live-born infants were included in the sample, of which 60 (2%) cases experienced neonatal death.

Outcome variable

The primary outcome of interest in this analysis was neonatal mortality, defined as deaths occurring within the first 28 days of life.1 Neonatal mortality was identified from the complete birth history of mothers, which collected information on newborn’s gender, date of birth, survival status and age at death (if dead).

Exposure variables

ENC practices were the primary exposures in our study. As a benchmark, we used the National Neonatal Health Strategy and Guidelines for Bangladesh, which provided advice on the following six practices: (1) the use of instruments disinfected using boiling water to cut the umbilical cord; (2) cord care, which did not involve the application of any substances to the umbilical cord stump after being cut; (3) drying of the newborn within 5 min of birth; (4) wrapping of the newborn within 5 min of birth; (5) bathing of the newborn at least 72 h after birth; and (6) the initiation of breastfeeding within 1 h of delivery.9 The information on the use of disinfected instruments and the umbilical cord care were collected in binary form (yes, no). The data on the timing of drying after delivery was categorized as a binary variable (<5 min, ⩾5 min/not dried) in our models. Similarly, information on the timing of wrapping after delivery was categorized as a binary variable (<5 min, ⩾5 min/not wrapped). The ENC practices of delayed bathing and immediate initiation of breastfed were also categorized as binary variables (yes, no).

Potential confounding variables

In identifying other factors that might potentially be associated with neonatal deaths, 12 factors were considered in this analysis. These included administrative divisions, urban/rural status of the interviewees usual place of residence, seasonal variation, socioeconomic status of the participants, mother’s education, mother’s age at birth, number of antenatal care visit, gender of the child, whether the selected child was first birth, size at birth perceived by mother, assistance received during delivery and the timing and the provider of a newborn’s first postnatal check-up. The socioeconomic status of participants was represented by a wealth index provided in the data that was constructed using the household asset data and the principal components analysis method.8, 10 The seasonal variation was generated from child's month of birth and, following previous literature, categorized as Winter (December–February), Summer (March–May), Rainy (June–August) and Autumn (September–November).11 Birth weights were usually unknown for babies born at home in LLMIC. Therefore, following earlier studies, we considered a baby’s size at birth perceived by mother as a proxy for birth weight.8, 11, 12

Statistical analysis

The differences in the exposure and other variables associated with the survival status of the newborns were examined using χ2 and t-tests. Then, an adjusted logistic regression model was used to examine the effect of the exposures after controlling for the confounding variables (listed above).

Code availability

All statistical analyses were performed using STATA version 11.2 (College Station, TX, USA). The STATA survey commands were used to adjust for the sampling weights and design. All codes, used for this analysis, are available from the corresponding author upon request.

Results

Descriptive statistics indicated statistically significant differences for dry cord care and delayed bathing between alive and deceased neonates (Table 1). Specifically, neonatal mortality was higher if nothing was applied to the umbilical cord (P<0.001) and bathing was not delayed (P=0.004).

Table 1 Association between ENC practices and child survivala
Full size table

With the exception of two administrative divisions, mother’s demographic characteristics were similarly distributed across the survival status (Table 2). Higher neonatal mortality was observed in Barisal and Sylhet divisions (P=0.006).

Table 2 Association between mother’s demographic characteristics and child survivala
Full size table

Table 3 provided details of the association between child’s demographic characteristics and neonatal deaths. Only the timing of first postnatal check-ups for newborn showed a statistically significant difference between the alive and the deceased children. Specifically, neonatal mortality was more likely for the newborn who had postnatal check-ups within 2 days of delivery (P=0.002).

Table 3 Association between child’s demographic characteristics and their survivala
Full size table

The logistic regression model showing the association between ENC practices and neonatal death was presented in Table 4. Neonatal mortality was significantly lower for ENC practices related to the deferment of bathing for 72 h after delivery. Specifically, the odds of neonatal deaths decreased by 86% for the children who had delayed bathing compared with the children either bathed before 72 h of delivery or not bathed (odds ratio=0.14; 95% CI: 0.03, 0.68; P=0.015).

Table 4 Logistic regression model output of the associations between neonatal mortality and the ENC practices
Full size table

With regard to the avoidance of applying any substances to the umbilical cord, the odds of neonatal mortality for those who applied nothing to the umbilical cord were nearly four times more likely compared with those who applied substances to the cord (odds ratio=3.81; 95% confidence interval: 1.75, 8.29; P=0.001). Note that a detailed investigation revealed a reduction in neonatal mortality for those who applied antiseptic in the umbilical cord though the relation was statistically insignificant (data not shown).

Discussion

Our study, which is the first to report the impact of different components of ENC practice on neonatal mortality, revealed a number of interesting and important findings. Specifically, the practice of deferment of bathing (for 72 h after delivery) significantly contributed to reducing neonatal mortality, but a counterintuitive relation was observed between the avoidance of applying any substance to the umbilical cord and neonatal mortality. The remaining ENC practices did not show any statistically significant association with neonate deaths.

Impact of deferment of bathing

Delayed bathing is an essential component of thermal care practice that has a crucial role in reducing neonatal hypothermia (body temperature below 36.5 °C).3, 13 A review of recent research found that a significant proportion of neonates born at home in LLMIC experienced hypothermia, which contributed as a comorbidity to major causes of neonatal deaths such as infection, asphyxia and preterm birth.14 Therefore, delayed bathing, by reducing the incidence of hypothermia, might significantly contribute in reducing neonatal mortality. The findings of our study on neonatal mortality in Bangladesh provided support to this hypothesis.

Impact of dry cord care

Sepsis or inflammation is a major cause of neonatal deaths in LLMIC. Immediately after cutting the umbilical cord, traditional practices in some countries may dictate the application of substances such as ash, oil, mud, butter and spice pastes that can be harmful in causing omphalitis or infection of the cord stump and lead to newborn sepsis and therefore death.15, 16 However, our findings indicated that avoiding the application of any substances to the umbilical cord significantly increased neonatal mortality. Interestingly, this was consistent with the findings of another study in South Asia where dry cord care was significantly associated with increased odds of death.2

On the other hand, applying chlorhexidine as an antiseptic to the umbilical cord stump in community and primary care settings in LLMIC had been found to prevent sepsis and reduce neonatal mortality.17 Thus, along with the recommendation of clean and dry cord care, World Health Organization additionally recommends the application of topical antiseptics to the cord stump where the risk of infection is high.18 Until recently, sepsis was a major cause of neonatal deaths in Bangladesh.8 Therefore, the application of antiseptic to the umbilical cord in the context of home births might be effective in reducing neonatal deaths compared with the application of no antiseptic.

Impact of immediate wrapping

As with delayed bathing, the ENC practice of immediate drying and wrapping the neonate is an important intervention to prevent hypothermia and thus reduce neonatal mortality.5, 13 However, we observed an insignificant impact of immediate wrapping on neonatal deaths in Bangladesh. An important point here was that while World Health Organization emphasizes on the use of a soft dry clean cloth to wrap a newborn after birth, a number of studies reported that a significant proportion of newborns delivered at home were either not properly covered or wrapped with dirty fabric.19, 20 Such practices, either by a failure to prevent rapid heat loss or by increasing the risk of sepsis through contact with unhygienic cloth, might increase the likelihood of neonatal mortality. This could account for the insignificant result of immediate wrapping on neonatal deaths in our study.

Impact of immediate drying

We also observed an insignificant and counterintuitive relationship between neonatal mortality and the practice of the immediate drying of newborns. Studies on newborn skin-wiping indicated that wiping the newborn with chlorhexidine solution, thereby reducing the risk of early neonatal sepsis, reduced neonatal mortality for low birth weight babies.21 An unhygienic method of drying of newborns, in contrast, could be responsible for sepsis and thus neonatal mortality. The discussion in the review by Darmstadt et al.,19 concerning the practice of wrapping neonates with dirty fabric in Bangladesh, could also indicate the possibility of improper drying of newborns in our study. Such an unhygienic practice might explain the increased mortality of children who were dried immediately compared with those who were not.

Impact of boiled instrument

Unlike our study, where we observed a statistically insignificant relationship between the use of boiled instruments to cut the cord and neonatal mortality, other research had noted a reduction in the risk of sepsis and neonatal deaths associated with use of a clean delivery kits that included a boiled blade to cut the cord.2 In contrast, Blencowe et al.,22 reviewing four studies on Pakistan, Tanzania and Senegal, did not find any impact of the use of clean birth kits (including boiled/sterile/new blades) on neonatal sepsis or tetanus, and thus on neonatal deaths. However, the review could not provide any reasonable explanation for such findings.

A review of the National Neonatal Health Strategy and Guidelines for Bangladesh reveals that the recommendation for clean cord-cutting (new/sterile blade or surgical blade) also includes the use of sterile thread for tying the cord.9 However, the use of sterilized thread to tie the cord was less common in Bangladesh and other South Asian countries where dirty thread (cotton thread from used quilts or jute fiber) was often used to tie the cord.19, 20, 23, 24 On the other hand, different components of the delivery kit (i.e. the use of a boiled blade to cut the cord, use of boiled thread to tie the cord and using a clean plastic sheet as the delivery surface) were all shown to be associated with significant relative reductions in mortality in South Asia.2 Unfortunately, such information was limited in our data, which being unaccounted for in our analysis could be responsible for a statistically insignificant effect of boiled instrument on neonatal mortality in our model.

Impact of breastfeeding within the first hour of birth

Earlier studies found that the risk of neonatal mortality increased with the delay in initiation of breastfeeding, indicating a dose–response relationship.25 Studies in Bangladesh and other developing countries, on the other hand, observed a common practice of premature breast milk supplementation such as honey and sweet water, which might expose newborns to pathogens.19, 23, 26 Newborns in our study, who started breastfeeding within first hour of birth, might have also consumed other supplements that could be responsible for the statistically insignificant result of the effect of breastfeeding (in the first hour) on neonatal mortality.

Some points are worth mentioning to contextualize our results. First, ENC practice is multifactorial issue where different components may influence each other. Therefore, a comparison of the cases who reported providing all six ENC practices with those provided different five or no practices could be insightful. However, owing to the size of our sample, such analysis was not possible. Second, our analysis showed a higher neonatal mortality for those who had early postnatal check-ups. This finding could be because of the higher likelihood of morbidity or disease. An earlier study on the utilization of postnatal care among rural women in Nepal observed a similar outcome.27 Finally, our analysis did not find any effect of birth size on mortality. This is counterintuitive, given that birth size is associated with birth weight and would therefore also be associated with mortality, as confirmed in earlier studies.12 A further review of previous studies indicates that this association may relate to intrauterine growth restriction and extents of prematurity.28, 29 Unfortunately, these information were unavailable to us preventing an investigation of these factors.

Limitation

This study had several potential limitations. The data relie on self-reporting by the participants. Self-reported information can be biased in a certain direction, whereas recall bias questions the accuracy of information, particularly about the timing of drying and wrapping. The relatively small sample size, specifically the number of neonatal deaths, may have resulted in low statistical power. Nevertheless, similar to previous studies,5, 30, 31 such limitations are outstripped by the insights gained through analyzing data obtained from such a large, nationally representative sample of newborns.

Conclusion

We investigated the effect of ENC practices on reducing the mortality of neonates delivered at home in Bangladesh. Our analysis offers some important insights for the improvement of ENC policy and health worker practice. First, neonatal mortality may be reduced through emphasizing delayed bathing. Second, the ENC guidelines in Bangladesh should include the use of topical antiseptics such as chlorhexidine to the cord stump. Such a recommendation may also apply to LLMIC where the incidence of sepsis is high. Third, the insignificant findings in our study associated with immediate drying and wrapping of newborns may be related to the reportedly common practice of using unclean cloth in Bangladesh. The National Neonatal Health Strategy and Guidelines for Bangladesh do not include any instruction relating to the cleanliness of the cloth needed to dry and wrap newborns. Specific guidelines on these practices, as well as community-based interventions, are required to improve hygiene at baby’s birth. Finally, more emphasis is required on the use of clean delivery kits and initiation of immediate and exclusive breastfeeding. However, we recommend such surveys to collect information on the different components of the delivery kit used at birth and premature breast milk supplementation, as such data may provide more precise insight into the impact of a boiled instrument (to cut the umbilical cord) and immediate initiation of breastfeeding on neonatal death.