https://orcid.org/0000-0002-5719-4294
Figures
Abstract
Background
Essential Newborn care (ENC) is a High-quality universal newborn health care devised by the World Health Organization for the provision of prompt interventions rendered to newborns during the postpartum period. Even though conducting comprehensive studies could provide a data-driven approach to tackling barriers to service adoption, there was a dearth of studies in Ethiopia that assess the geographical variation and predictors of missing ENC. Hence, this study aimed to identify geographical, individual, and community-level predictors of missing ENC messages at the national level.
Methods
This study used the 2016 Ethiopian Demographic and Health Survey, by using a weighted sample of 7,590 women who gave birth within two years prior to the survey. The spatial analysis was carried out using Arc-GIS version 10.7 and SaTScan version 9.6 statistical software. Spatial autocorrelation (Moran’s I) was checked to figure out the non-randomness of the spatial variation of missing ENC in Ethiopia. Six items of care used to construct a composite index.0of ENC uptake were cord examination, temperature measurement, counselling on danger signs, counselling on breastfeeding, observation of breastfeeding, and measurement of birth weight. To assess the presence of significant differences in the mean number of ENC items across covariates, independent t-tests and one-way ANOVA were performed. Finally, a multilevel multivariable mixed-effect negative binomial regression was done by using STATA version 16. The adjusted incidence rate ratio (aIRR) with its corresponding 95% CI was used as a measure of association and variables with a p-value<0.05 were identified as significant predictors of ENC.
Results
The overall prevalence of missing ENC was 4,675 (61.6%) (95% CI: 60.5, 62.7) with a significant spatial variation across regions. The majority of Somali, Afar, south Amhara, and SNNPR regions had statistically significant hotspots for missing ENC. The mean (±SD) number of ENC items received was 1.23(±1.74) with a variance of 3.02 indicating over-dispersion. Living in the poorest wealth quintile (aIRR = 0.67, 95%CI: 0.51, 0.87), lack of Antenatal care (aIRR = 0.52, 95%CI: 0.49, 0.71), birth at home (aIRR = 0.27, 95% CI: 0.17, 0.34), living in rural area (aIRR = 0.39, 95% CI: 0.24, 0.57) were significant predictors of ENC uptake.
Conclusion
The level of missing ENC was found to be high in Ethiopia with a significant spatial variation across regions. Hence, the government and policymakers should devise strategies for hotspot areas to improve women’s economic capabilities, access to education, and health-seeking behaviours for prenatal care and skilled delivery services to improve ENC uptake.
Citation: Hailegebireal AH, Kitila AT (2024) Geographical variation and predictors of missing essential newborn care items during the immediate postpartum period in Ethiopia: Spatial and multilevel count analyses. PLoS ONE 19(9): e0308104. https://doi.org/10.1371/journal.pone.0308104
Editor: Amanuel Abajobir, African Population and Health Research Center, ETHIOPIA
Received: September 19, 2023; Accepted: July 16, 2024; Published: September 6, 2024
Copyright: © 2024 Hailegebireal, Kitila. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data for this study were obtained from the DHS program with a reasonable request. Because the DHS office doesn’t allow to sharing of the data with other third parties, the one who needs the data supporting the findings of this study can get it in anonymized form from the DHS website at https://www.dhsprogram.com upon reasonable request. The authors did not have any special access privileges that others would not have.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AIC, Akaike’s information criterion; ANC, Antenatal Care; CSA, Central Statistical Agency; EDHS, Ethiopian Demographic and Health Survey; ENC, Essential Newborn Care; GWR, Geographically Weighted Regression; IRR, Incidence Rate Ratio; OLS, Ordinary Least Square; PNC, Postnatal Care; PPP, Postpartum Period; WHO, World Health Organization
Introduction
The neonatal period (the first month of life) is the most vulnerable and critical period for a child’s survival, growth, and development [1]. Approximately one million babies died during the first 24 hours of life in 2019, accounting for almost three-quarters (75%) of all neonatal deaths globally [2]. About 2.4 million newborns died during the neonatal period in 2020, with approximately 6700 newborn deaths every day, which represented nearly half (47%) of all under-5 deaths [2]. Sub-Saharan Africa(SSA) and Central and Southern Asia(CSA), with neonatal mortality rates (NNM) of 27 and 23 per 1000 live births, respectively, accounted for 43% and 36% of global neonatal deaths [2, 3]. A child born in SSA is ten times more likely than a child born in a high-income country to die in the first month [2]. Ethiopia has had significant achievement in reducing under-5 mortality rates by 71% over the last two decades, from 204 deaths per 1,000 live births in 1990 to 59 deaths per 1,000 live births in the 2019 mini EDHS report [4, 5]. However, the decline in NNM from 58 deaths per 1000 live births to 29 deaths per 1000 live births between 2000 and 2016 has not been satisfactory [4]. Even so, it has risen to 33 deaths per 1,000 live births in 2019 [4].
The postpartum period (PPP), which begins immediately after delivery and lasts up to six weeks (42 days), is an important time for women, babies, partners, parents, carers, and families [6]. The majority of neonatal deaths in developing countries including Ethiopia were due to a lack of quality essential newborn care (ENC), especially within the first two days of birth [2]. Thus, the World Health Organization (WHO) recently devised global recommendations on the timing and contents of postnatal care (PNC) for mothers and newborns, in resource-limited settings to ensure a positive postnatal experience [6]. One of the basic elements of PNC aimed at improving newborn survival is the provision of easy, low-cost, and less time-consuming ENC in the immediate PPP [7].
Essential Newborn care (ENC) is a WHO strategic approach for the provision of a constellation of interventions rendered to newborns during the immediate PPP [7, 8]. It is a High-quality universal newborn health care that entails prompt care at the time of birth and throughout the neonatal period, regardless of the place of delivery [8, 9]. It comprises measurement of body weight, temperature monitoring, hygienic cord care, early and exclusive breastfeeding, evaluation for danger signs, and preventive services like immunization [2, 7]. Findings showed that providing quality ENC is a cost-effective strategic approach that results in a considerable reduction in early and late neonatal death [10–12].
Despite its importance in the reduction of NNM, 29.3%, 32.3%, 35.5%, and 31% of women in Nepal [13], India [14], Rwanda [15], and Tanzania[16] missed ENC, respectively. However, a recent systematic review meta-analysis in Ethiopia revealed that the uptake of ENC services was found to be low at 48.77% [17]. Inadequate intrapartum (skilled delivery service) and postpartum care (like ENC) accounted for a steady reduction in neonatal deaths in Ethiopia [4, 18]. The government has been striving to meet the third Sustainable Development Goal (SDG3), which is aimed at ensuring healthy lives and promoting well-being for all people through the provision of adequate maternal and newborn health services [19]. ENC service delivery at the health facility and community level is one approach to meeting this ambitious goal [20].
Despite several studies on ENC uptake in Ethiopia, none investigated the spatial distribution of missing service usage and its predictors at the individual and community levels by using a large population. Even though a recent study attempted to examine the spatial distribution, it primarily focused on women who received ANC [9] and may have failed to provide a clear picture of ENC. Thus, the current study used a relatively larger sample size of women who were eligible for PNC and employed spatial (a hot spot and a geographically weighted regression (GWR) analysis) and multilevel count analytic approaches. A hotspot analysis was performed to identify regions with a high proportion of women who missed ENC, followed by a GWR to identify potential predictors that lead to regional disparities in missing the service. In addition, a multilevel approach was done to identify factors at the individual and community levels that contribute to missing ENC items. Conducting such comprehensive analyses could provide a data-driven approach to comprehending and reducing the barriers to service uptake [21]. The findings will enable stakeholders to strengthen their efforts in addressing the impediments to service uptake by designing geographical, individual, and community-focused effective interventions to ensure a positive postnatal experience.
Methods and materials
Data source, study design, and period
This study used the 2016 Ethiopian Demographic and Health Survey (EDHS) report; a population-based, nationally representative survey collected from January 18 to June 27, 2016. The country is located at 30−150 N latitude and 330−480 E longitude. The country has a total surface area of 1,112,000 km2 and has boundaries defined on the west by Sudan, on the east by Somali and Djibouti, on the north by Eritrea, and on the south by Kenya. Nine regions (Tigray, Afar, Amhara, Oromia, Somali, Benishangul-Gumuz, Southern Nation Nationality and People’s Region (SNNPR), Gambella, and Harari) and two self-administrative cities (Addis Ababa and Dire Dawa) were included in the survey [22]. The data were obtained from the women’s (IR) file contained in the 2016 EDHS report.
Population of the study
The source populations were all women who responded to a query about receiving a postnatal check-up within two months. The study population, on the other hand, were mothers who had complete information on the uptake of each essential newborn care within the first two days following delivery. A total of 8,490 respondents were excluded from the study due to a lack of information on service usage (missing values).
Sampling procedure and data collection tools
Study participants were selected through a stratified two-stage cluster sampling technique, where each region was divided into urban and rural areas. In the first stage, 645 clusters or enumeration areas (202 urban and 443 rural areas) were selected randomly. Then, a fixed number of 28 households with eligible women were selected per cluster based on an equal probability systematic selection. The survey design and methodology were addressed as well in the 2016 EDHS [22]. The Global Positioning System (GPS) was used to obtain the geographic coordinates of each survey cluster [22, 23]. To assure the confidentiality of respondents, the geographical locations(latitude and longitude) were randomly displaced. The greatest displacement was two kilometers (km) for all urban and five kilometers (km) for 99% of rural clusters. The remaining 1% of rural clusters have been displaced up to a 10-kilometer distance [24]. Data were collected from all eligible women using the Woman’s Questionnaire, which includes socio-demographic and economic information, obstetric characteristics, and maternal health service usage.
Measurement of variables of the study
Outcome variable.
The outcome variable for this study was the uptake of essential newborn care services during the immediate postpartum period (PPP). The measurement was based on the receipt of six essential services: (i) cord examination (m78a_1), (ii) Temperature measurement (m78b_1), (iii.) Counselling on danger signs (m78c_1) (iv) Counselling on breastfeeding (m78d_1), (v) Observation of breastfeeding (m78e_1), and measurement of birth weight (m19_1). There were yes, and no response options for each question. For the sake of analysis, the categories were confined to Yes (= 1) and No (= 0). A composite index of essential newborn care (ENC) has been created based on the responses, which is a count of the number. The variable had a minimum and maximum value of zero and six, respectively. Finally, women with a value of ’0’ were considered as ’missing ENC service,’ whilst those who scored one or more were considered to have ‘received ENC’, and the spatial analyses were focused on those who did not receive the service [9, 25].
Explanatory variables.
After reviewing related and current literature, potential predictors of ENC were selected from the data set and nested at the individual and community levels [7, 25–27] (Table 1).
Data management and statistical analysis
STATA version 16, ArcGIS version 10.7, and SaTScan version 9.6 were used to analyze the data. The data were weighted to minimize under- or over-representation during strata-specific selection during the survey. This is vital to obtain a reliable estimate and draw proper inferences[30]. The weighted proportions of an ENC and its predictors were computed in STATA and prepared in Microsoft Excel 2016 (CSV format) before being loaded into ArcGIS 10.7 for further spatial analysis. Descriptive statistics such as frequency and percentage of different variables were estimated and displayed using texts, and tables.
Spatial analyses.
Initially, the data containing the variable of interest in CSV format were imported into ArcGIS version 10.8 and combined with the GPS data (shape file). After that, the event data was converted to a shape file and displayed in an XY plane (geographically coordinated system). Projections of the geographically coordinated data to the projected coordinate data were performed before the analysis.
Spatial autocorrelation (Global Moran’s I).
The global spatial autocorrelation was estimated using Moran’s index to decide whether missing ENC in Ethiopia was dispersed, clustered, or randomly distributed. In general, a Moran’s I value close to 1 suggests significant positive autocorrelation (missing ENC was clustered/non-random), whereas a Moran’s I value close to -1 shows significant negative autocorrelation (missing ENC was dispersed) across enumeration areas (EAs). On the other hand, the value near zero implies that the spatial distribution of missing ENC was random (independence between EAs or no spatial autocorrelation) [31, 32]. Accordingly, Moran’s I value was statistically significant (p<0.05) which indicates the spatial distribution for missing ENC is non-random (clustered) [32].
Spatial interpolation.
The spatial interpolation method is used to predict the likelihood of missing ENC in unsampled areas based on sampled EAs. There are multiple geostatistical and deterministic interpolation techniques, and for this study, ordinary Kriging was performed since it optimizes the weight and has a low residual and mean square error [33, 34].
The spatial scan statistical (SaTScan) analysis.
A SaTScan analysis was carried out to show significant hot spots of missing ENC. The Bernoulli-based model was employed to detect the significant spatial clusters with no ENC since it uses a scanning window that moves across the study area [35]. To fit the Bernoulli model, women who missed ENC were treated as cases, while those who received the service were treated as controls. SaTScan statistics scanned gradually across the space to identify the number of observed and expected observations inside the window at each location. The default maximum spatial cluster size of 50% of the population was used as an upper limit, allowing both tiny and big clusters to be recognized. Using 999 Monte Carlo replications, the primary, secondary, and other significant clusters were identified and ranked based on the likelihood ratio test (LLR). A potential cluster, in the scanning window with the highest LLR and a significant p-value was selected as the high-performing cluster for being a case (missing ENC).
Hot spot analysis (Gettis-Ord Gi* statistics).
Getis-Ord Gi* statistics were carried out to identify significant hot spot and cold spot areas for missing ENC. The z-score was estimated to figure out the statistical significance of clustering, and the level of significance was set at p-value<0.05 with a 95% CI. Cold spot and hot spot were declared at z-score was less than -1.96 and greater than +1.96, respectively [36, 37].
Spatial regression analysis.
Spatial regression analysis incorporates global (ordinary list squares) and local (geographically weighted regression) techniques [38, 39].
Ordinary least square (OLS) analysis.
The Ordinary Least Squares (OLS) regression model is a global model that predicts only one coefficient per independent variable throughout the whole study area. The compliance of all assumptions was assessed to ensure the reliability of the findings in this model. First, the coefficients of the intercept and predictors have to be statistically significant having a positive or negative sign. Also, the presence of multicollinearity among explanatory variables should be confirmed by examining their variance inflation factors (VIF), and variables with VIF > 10 were considered multicollinear and iteratively excluded from the model [40, 41]. Furthermore, the Koenker Bp statistic was used to determine if the model could be employed for a geographically weighted regression (GWR) analysis. In the current study, the Koenker statistics were significant (p-value<0.001), and GWR analysis was required to examine the distribution of key variables.
Geographically weighted regression (GWR).
Unlike OLS, which fits a single linear regression equation to all of the clustered data, GWR selects data from neighbouring features, therefore the GWR coefficient has different values for each cluster [42]. Variables with p-values<0.05 in the OLS model were selected for GWR and discussed based on their coefficients.
Multilevel generalized linear model (GLM).
Due to the hierarchical nature of the EDHS data, where women were nested inside families and households were nested within clusters, ordinary one-level regression models were not appropriate, and hence multilevel modelling was used. Carrying out a multilevel analysis for such hierarchical data allows us to minimize biased parameter estimation [43].
Because the number of ENC items received is a non-negative integer (count), most recent thinking in the field suggests using GLM models with a Poisson link as a model of choice [44–46]. However, Poisson regression has an assumption, called the assumption of equidispersion, requiring the variance of the count response variable should be equal to its mean [45, 46]. The mean and variance of the count outcome variable in the current study were 1.23 and 3.02, respectively, indicating the presence of over-dispersed. Thus, a multilevel mixed-effect negative binomial regression model is appropriate for such an over-dispersed outcome variable [46–48].
Model building and selection.
Fixed effects. To look into the presence of statistically significant differences in the mean number of ENC items across each categorical variable, independent t-tests and one-way analysis of variance (ANOVA) were performed. Those variables with p-values<0.05 were incorporated into a multilevel negative binomial regression, where significant predictors of ENC were determined. Finally, the incident rate ratio (IRR) with a 95% confidence interval was reported, and statistical significance was determined at a p-value<0.05.
Random effects. Four distinct models were fitted using a multilevel approach. Model one (null model) is devoid of any explanatory variables. The second and third models comprised solely individual and community-level characteristics. The fourth (full) model included and accounted for all factors at the individual and community level factors. The Intraclass Correlation Coefficient (ICC), Median Odds Ratio (MOR), and Proportional Change in Variance (PCV) were estimated to measure the random effects (variability in ENC uptake between and across clusters).
ICC quantifies the degree of heterogeneity of ENC between clusters and estimated as:
Where Var(b) is the variance at the group level and Var(w) refers to the variance of the distribution used in the negative binomial link function, which is π2/3 ≈3.29.
The proportion of change in variance (PCV) measures the proportion of the total observed individual variation that is explained by the between-cluster variations and estimated as
where, Va is the variance of the initial model (null model), and Vb = variance of the subsequent models (models 2, 3, and 4). The value lies between 0 and 1 (0 < PCV < 1) or 0 and 100% (0 < PCV < 100%).
The median odds ratio (MOR) represents the change in odds of missing ENC items when moving from one cluster to another while holding individual-level predictors constant.
Model fitness. Deviance = -2 * (Log Likelihood (LL), Schwarz’s Bayesian Information Criterion (BIC), and Akaike’s information criterion (AIC) were used to determine the best model. Finally, the fourth model with the lowest deviance, AIC, and BIC values was chosen as the best-fit model for the current study (Table 5).
Ethical consideration and consent to participate. Following registration at the DHS Programme website with possible justification, ICF International provided written permission to use both the DHS and GPS datasets. The data obtained were used only for the registered research and were not shared with anyone other than the co-authors. The DHS also declared that informed consent was obtained from all subjects and/or their legal guardian during the primary data collection. Furthermore, as this is secondary data, the Institutional Review Board (IRB) of Wachemo University College of Medicine and Health Sciences declared that no formal ethics approval was required. However, the IRB ensured that the research approach was ethically compliant with national and international standards.
Results
Background characteristics of the respondents
The findings of this study were based on a total weighted sample of 7,590 women. The mean (±SD) age of women was 29.25 (±6.84), with over half (50.4%) of them belonging to the age group of 25–34 years. Three-quarters (75.8%) of women in the poorest wealth quintile and 85.5% of women without ANC missed ENC services. In addition, 68.2% of rural women and 72.0% of women with no formal education missed ENC service. Addis Ababa and Afar regions exhibited the highest (3.7) and lowest (0.41) mean number of ENC contents, respectively. In addition, there were statistically significant differences in the mean number of ENC services received across the educational level, wealth index, residence, region, place of delivery, and media exposure (p<0.001) (Table 2).
The level of missing ENC items and disparities across the regions
More than half, 61.6% (95% CI: 60.5, 62.7) of women missed at least one ENC item. The highest proportion of women who missed ENC was recorded in Oromia (73.6%), closely followed by Somali (73.1%) and Afar (72.7%) regions. The top three ENC items that the majority of women failed to receive were cord checks (90.5%), counselling about danger signs (89.3%), and temperature measurement (87.0%) (Table 3).
Spatial analysis results
Spatial distribution of missing ENC among respondents.
Somali, Afar, Western Oromia, and Gambella had a larger proportion of neonates who did not receive ENC. Tigray, Addis Ababa, and the Eastern border of Amhara, on the other hand, had a low rate of missing ENC (Fig 1).
Spatial autocorrelation of missing ENC.
The global spatial autocorrelation analysis revealed that the spatial distribution of missing ENC was non-random (i.e. there was significant spatial variation) across the country (global Moran’s I = 0.49, p<0.001). The clustered patterns (on the right sides) suggest that missing ENC occurred at a high rate throughout the study area (Fig 2).
Incremental autocorrelation.
To determine the average nearest neighbour and minimum and maximum distance band, the incremental spatial autocorrelation over a series of distances depicted by a line graph with a corresponding z-score was performed. With a starting distance of 121803 meters, a total of 10 distance bands were found, with the first highest peak (clustering) observed at 151367.66 meters (Fig 3).
Hot spot (Getis-Ord Gi*) analysis.
As evidenced by hot spot analysis, Somali, central and southwest Afar, the southern part of Amhara, southwest Oromia, and the north and eastern portions of SNNPR have a high rate of missing ENC. Tigray, Addis Ababa, and Dire Dawa, on the other hand, had a low percentage of missing ENC (Fig 4).
Spatial interpolation.
The spatial distribution of missing ENC for places where data were not collected was estimated using the ordinary Kriging technique. The highest predicted prevalence of missing ENC(red-shaded) was found in southern Somali, North and South Afar, Southwest and central Oromia, and parts of Addis Ababa and Gambella. In contrast, the expected proportion of high ENC uptake (green-shaded) includes the entire Tigray region, and central parts of Addis Ababa, Gambella, Dire Dawa, and Harari (Fig 5).
Spatial scan statistical (SaTScan) analysis.
The SaTScan spatial analysis identified fifteen statistically significant groups of SaTScan clusters with a high proportion of newborns missing ENC. This means that the prevalence of missing ENC was higher inside the SaTScan circular window than outside of it. In addition, the analysis identified a total of 333 significant clusters that accounted for missing ENC, of which 78 and 42 were found in the first and second most likely clusters respectively. The first most likely cluster located at geographical coordinates of (5.330795 N, 41.837597 E) with a 427.4 Km radius, and LLR of 82.86 at p<0.001 showed that newborn within the area had a 41% (RR = 1.41) higher risk to miss ENC than their counterparts outside the area (Table 4).
Results of spatial regression
The global ordinary least square (OLS) analysis results.
The OLS model is the first step toward choosing the appropriate predictors for the spatial variation of missing ENC. As a result, lack of formal education, giving birth at home, lacking ANC, and never watching television were found to be associated with missing ENC. There was no sign of multicollinearity among the selected predictors (mean VIF = 1.40, minimum VIF = 1.19, and maximum VIF = 1.94). Furthermore, the adjusted R2 = 0.718 from the OLS global model output indicated that the four predictors explained 71.8% of the variation in missing ENC. Jarque-Bera statistics with p-value>0.05 (p = 0.448) suggest that the model prediction was not biased (the requirement for residual normality was met). The Koenker statistics in the model, on the other hand, showed a statistically significant p-value (p<0.001), showing that the regression model is inconsistent across the study area, implying the necessity for the GWR model to estimate the model parameters properly (Table 5).
Geographically weighted regression analysis
To deal with this violation of the stationarity assumption of the global (OLS) model, the local (GWR) model was fitted to offer realistic estimates. The GWR analysis outperformed the global model (OLS) significantly. The AICc value in the GWR model reduced from -349.68 in the OLS model to -401.11. The adjusted R2 in GWR was greater than the OLS one, showing that the local model enhanced its ability to forecast hotspots of missing ENC. The highest range of R2 was shown in central Addis Ababa, central and western Oromia, and the northern part of SNNPR regions. As compared to other regions, the model is less explained by the predictors in the southern part of the Benishangul Gumuz region (Fig 6).
As the proportion of women who gave birth at home increased, so did the proportion of missing ENC in most parts of Afar and Somali, in some parts of Oromia and Gambella, and the eastern border of Benishangul Gumuz. On the other hand, the positive and weaker relationship between home birth and missing ENC was observed in most parts of Tigray, eastern and central Amhara, and some parts of Addis Ababa (Fig 7).
This study also highlights the space-dependent relationship between missing ENC and education status. As the proportion of women with no formal education increases, the likelihood of missing ENC also increases in the entire Somali, west Afar, most parts of SNNPR, and the Northern part of Gambella regions (Fig 8).
Women who did not have ANC had a strong link with women who did not have ENC. As the proportion of women who did not attend ANC rose, so did the number of missing ENC in Afar, Somali, Northwest Gambella, south of the SNNPR, and northern and southern Amhara regions (Fig 9).
Results of multilevel mixed effect negative binomial regression
Random effect (measures of variation).
In the null model, the value of ICC 0.261 which implies that 26.1% of the total variability in the receipt of ENC items was explained by the difference between clusters. Furthermore, variation at the individual and community levels accounted for 11.1% (ICC = 0.111, p<0.001) and 14.1% (ICC = 0.141, p<0.001), respectively, of the variation in the mean number of ENC items. Individual, and community-level factors together accounted for 51% of the mean variation seen in the null model (PCV = 73.2%). The values of AIC, BIC, and Deviance decreased as we progressed from model 1 (the empty model) to model 4 (the full model), indicating that the final model fitted throughout the study had adequate goodness of fit. Finally, the fourth model with the lowest deviance (17708.0) was chosen as the best model fit (Table 6).
Fixed effects: Predictors of receipt of ENC items.
In the multivariable multilevel negative binomial regression analysis, wealth index, educational status, frequency of ANC, residence, and region were identified as significant predictors of receiving ENC.
The incidence rate of getting ENC was lowered by 40% (aIRR = 0.60, 95%CI: 0.46, 0.79) and 33% (aIRR = 0.67, 95%CI: 0.51, 0.87) among women in the poorest and poorer wealth quintiles, respectively than women in the richest wealth quintile. Women without formal education were 27% less likely to receive items of ENC than those who attend secondary or higher education (aIRR = 0.73, 95%CI: 0.60, 0.89). As compared to women who received 4 or more ANC visits, the chance of receiving ENC items fell by 48% among those who did not receive any ANC visits (aIRR = 0.52, 95%CI: 0.49, 0.71). Women who gave birth at home had a 73% lower likelihood of receiving ENC items (aIRR = 0.27, 95% CI: 0.17, 0.34) than women who gave birth in health facilities. Region and residence were the two community-level factors that were identified as significant predictors of ENC. Women who lived in the peripheral regions had 44% less chance of receiving ENC items (aIRR = 0.56, 95% CI: 0.47, 0.66) as compared to women who lived in metropolitan one. Similarly, women who resided in the rural part of the country were 61% less likely to receive ENC items as compared to their urban counterparts (aIRR = 0.39, 95% CI: 0.24, 0.57) (Table 6).
Discussion
It is vital to investigate the spatial distribution and predictors of essential newborn care utilization at the national level to design a targeted intervention to reduce neonatal mortality and increase satisfaction and utilization of maternal and neonatal health services by women [26]. In this study, the prevalence of missing ENC was found to be 61.6% (95% CI: 60.5, 62.7) which is higher than a systematic review and meta-analysis conducted in Ethiopia (51.2%) [17], and similar studies conducted in Nepal (29.3%) [13], India (32.3%) [14], Rwanda(35.5%) [15], and Tanzania(31%) [16]. The disparity could be attributed to differences in socioeconomic and sociocultural characteristics, availability and accessibility of health service infrastructure, and maternal health service coverage, between countries.
In addition, spatial global Moran’s analysis found that the percentage of women who didn’t receive ENC varied geographically. As per hot spot analysis, statistically significant hotspot areas for missing ENC were Somali, central and southwest Afar, south Amhara, and the eastern border of SNNPR regions. Previous studies revealed that maternal and neonatal health service uptake was low in those regions [49–53]. This clustering might be due to a variety of factors. To begin, as compared to other regions, those two regions, particularly Afar and Somali, are known to have a shortage of healthcare facilities, limited healthcare providers, and a lack of medical supplies and equipment to provide ENC services [54, 55]. In addition, they are located in remote and arid areas with inadequate road infrastructure and transportation services, making it difficult for both women and health care providers to access and deliver immediate postnatal care services, potentially leading to high coverage of missing ENC [56]. Furthermore, because the majority of the people in these regions live pastoral or partly nomadic lives, it may be difficult for postpartum women to access adequate ENC services because their places of residence change over time.
In multilevel mixed effect negative binomial and Geographically weighted regression analyses, having no formal education, giving birth at home, and lack of ANC were identified as significant predictors of not receiving ENC in Ethiopia. In addition, living in the poorest wealth quintile, being a rural resident, and living in a peripheral region were also identified as significant determinants of the uptake of ENC items in a multilevel regression model.
Not receiving ANC for the last pregnancy was found to be strongly associated with missing ENC items. This was supported by studies conducted in Bangladesh [57], Nigeria [58], Ghana [59], Uganda [60], East Africa [61] and Ethiopia [62]. In addition, studies conducted in Asia revealed that information on newborn care practices provided to pregnant women and their families during ANC visits resulted in enhanced ENC practices such as complete cord care, complete thermal care, and breastfeeding initiation [63]. In addition, ANC links pregnant women with skilled healthcare providers who can assist with safe deliveries and immediate newborn care [64]. Lack of ANC visits could result in missed opportunities for newborn care education and information, reduced access to skilled birth attendants, and inadequate birth preparedness and complication readiness, all of which can contribute to a higher likelihood of missing ENC, which is critical for ensuring the health and survival of newborns [58].
Home birth was also identified as a significant predictor of missing ENC packages. This was in tandem with the findings of studies conducted elsewhere [65–67]. This might be due to a variety of reasons. First, giving birth at home potentially can lead to difficulty in accessing skilled healthcare providers in the immediate postpartum period, and this results in missing ENC [65]. As compared to births at health facilities, those at home have a lower or no likelihood of receiving postnatal care, education, and counselling, and being closely monitored by healthcare providers resulted in missing vital ENC components. Thus, efforts should focus on increasing access to skilled birth attendants through improving healthcare and transportation infrastructure, educating communities about the importance of facility deliveries, and addressing cultural and socioeconomic factors that hinder delivery choices [68, 69].
The current study revealed that the risk of receiving ENC items was lower among women who never attended formal education. This was supported by studies conducted in LMICs [70], Pakistan, Ghana [59], and Ethiopia [71]. This could be due to, women without formal education may have limited access to information about ENC and its importance, which can lead to suboptimal care. Furthermore, a lack of formal education may result in a lower socioeconomic status and accompanying financial constraints, as well as a lack of autonomy in decision-making, all of which can be linked to limiting access to the resources needed and healthcare services for ENC. As a result, improving access to formal education is essential for addressing those challenges, particularly in areas where formal education is limited.
Similarly, living in a household with the poorest wealth quintile was revealed to be a statistically significant predictor of not receiving ENC. This was supported by studies conducted in India [72], Nepal [73], and Ethiopia [74]. Those women in the poorest wealth quintile are more likely to experience financial barriers in accessing maternal healthcare, especially ANC, skilled delivery, and PNC services all of which are vital entry points to get ENC services. Furthermore, poverty and financial constraints can put a strain on women and families by limiting access to information and the ability to make health-care decisions [75].
The study has the following strengths. First, the findings were based on an analysis of nationally representative data, making the findings more generalizable. This study is also enriched by the findings of Hot Spot, geographically weighted regression, and multilevel mixed-effect negative binomial regression analysis. As a result, the findings can assist government and program planners in designing geographical, individual, and community-focused public health interventions based on the identified predictors for tackling barriers to ENC. On the other hand, the finding also should be interpreted in light of limitations. To begin, to safeguard the confidentiality of respondents or the community, the geographical coordinates of clusters were displaced by up to 2km in urban areas, 5km for most rural clusters, and 10km for 1% of rural clusters; this may alter estimated cluster effects in the spatial regression. Second, because of the cross-sectional nature of EDHS data, it is hard to infer a temporal/causal relationship between the variables. Furthermore, as the data were based on the retrospective interviews of women who responded to receipt of PNC for their last birth, the findings may be subject to recall bias. Finally, the data were based on self-report and this tends to face recall- and social desirability bias which may underestimate or overestimate the true association.
Conclusion
The level of missing ENC during immediate PPP was found to be high in Ethiopia with a significant spatial variation across regions. Statistically significant hotspot areas for not receiving ENC were northern Somali, central and southwest Afar, the southern part of Amhara, and the eastern border of SNNPR regions. Living in the poorest wealth quintile, being a rural resident, having no formal education, giving birth at home, and lack of ANC were identified as significant predictors of not receiving ENC. Hence, the government and policymakers should devise strategies for hotspot areas to improve women’s economic capabilities, access to education, and health-seeking behaviours for prenatal care and skilled delivery services in order to improve ENC uptake.
Acknowledgments
We are grateful to ICF macro (Calverton, USA) for providing the 2016 DHS data of Ethiopia.
References
- 1. Sharrow D., et al., Global, regional, and national trends in under-5 mortality between 1990 and 2019 with scenario-based projections until 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation. The Lancet Global Health, 2022. 10(2): p. e195–e206. pmid:35063111
- 2.
World Health Organization, Newborn Mortality,Key facts, 28 January 2022, https://www.who.int/news-room/fact-sheets/detail/levels-and-trends-in-child-mortality-report-2021.
- 3. Hug L., et al., National, regional, and global levels and trends in neonatal mortality between 1990 and 2017, with scenario-based projections to 2030: a systematic analysis. The Lancet Global Health, 2019. 7(6): p. e710–e720. pmid:31097275
- 4. Shibre G., Idriss-Wheeler D., and Yaya S., Inequalities and trends in neonatal mortality rate (NMR) in Ethiopia: evidence from the Ethiopia demographic and health surveys, 2000–2016. PloS one, 2020. 15(6): p. e0234483. pmid:32520940
- 5.
Ethiopian Public Health Institute (EPHI) [Ethiopia] and ICF. 2021. Ethiopia Mini Demographic and Health Survey 2019: Final Report. Rockville, Maryland, USA: EPHI and ICF.
- 6.
Organization, W.H., WHO recommendations on maternal and newborn care for a positive postnatal experience. 2022: World Health Organization.
- 7.
Organization, W.H., Essential newborn care course. 2010, World Health Organization.
- 8.
World Health Organization. (2014). Early essential newborn care: clinical practice pocket guide. https://iris.who.int/bitstream/handle/10665/208158/9789290616856_eng.pdf
- 9. Asratie M.H., et al., Spatial distribution and determinants of newborn care within 2 days of postpartum period among women with at least one antenatal care visit in Ethiopia: Mixed effect multilevel analysis. Plos one, 2023. 18(2): p. e0282012. pmid:36854013
- 10. Zamboni K., et al., Effect of collaborative quality improvement on stillbirths, neonatal mortality and newborn care practices in hospitals of Telangana and Andhra Pradesh, India: evidence from a quasi-experimental mixed-methods study. Implementation Science, 2021. 16(1): p. 1–17.
- 11. Carlo W.A., et al., Newborn-care training and perinatal mortality in developing countries. New England Journal of Medicine, 2010. 362(7): p. 614–623. pmid:20164485
- 12. Walker D., et al., Effect of a quality improvement package for intrapartum and immediate newborn care on fresh stillbirth and neonatal mortality among preterm and low-birthweight babies in Kenya and Uganda: a cluster-randomised facility-based trial. The Lancet Global Health, 2020. 8(8): p. e1061–e1070. pmid:32710862
- 13.
Tuladhar, S., The Determinants of Good Newborn Care Practices in the Rural Areas of Nepal. 2010.
- 14. Castalino F., Nayak B.S., and D’Souza A., Knowledge and practices of postnatal mothers on newborn care in tertiary care hospital of Udupi district. Journal of Health and Allied Sciences NU, 2014. 4(02): p. 098–101.
- 15. Batamuriza M., Uwingabire E., and Oluyinka A., Essential newborn care among postnatal mothers at selected health centers in eastern province, Rwanda. Rwanda Journal of Medicine and Health Sciences, 2020. 3(2): p. 139–151.
- 16. Stensgaard C.N., et al., Essential newborn care practices for healthy newborns at a district hospital in Pemba, Tanzania: a cross-sectional observational study utilizing video recordings. Global Health Action, 2022. 15(1): p. 2067398. pmid:35674453
- 17. Alamneh Y., et al., Essential newborn care utilization and associated factors in Ethiopia: a systematic review and meta-analysis. BMC Pregnancy and Childbirth, 2020. 20(1): p. 124. pmid:32093648
- 18. Doherty T., et al., Reduction in child mortality in Ethiopia: analysis of data from demographic and health surveys. Journal of global health, 2016. 6(2). pmid:29309064
- 19. Ayele A.A., Tefera Y.G., and East L., Ethiopia’s commitment towards achieving sustainable development goal on reduction of maternal mortality: There is a long way to go. Women’s Health, 2021. 17: p. 17455065211067073. pmid:34913391
- 20.
Federal Ministry of Health, Ethiopia, Essential Care for Every Baby Training Participants’ Manual, 2016
- 21.
Bolarinwa, O.A., et al. Spatial patterns and multilevel analysis of factors associated with antenatal care visits in Nigeria: Insight from the 2018 Nigeria Demographic Health Survey. in Healthcare. 2021. MDPI.
- 22.
Csa, I., Central statistical agency (CSA)[Ethiopia] and ICF. Ethiopia demographic and health survey, Addis Ababa, Ethiopia and Calverton, Maryland, USA, 2016. 1.
- 23. Eng J.L.V., et al., Use of handheld computers with global positioning systems for probability sampling and data entry in household surveys. The American journal of tropical medicine and hygiene, 2007. 77(2): p. 393–399. pmid:17690421
- 24.
ICF International. 2013. Incorporating Geographic Information into Demographic and Health Surveys: A Field Guide to GPS Data Collection. Calverton, Maryland, USA: ICF International https://dhsprogram.com/pubs/pdf/DHSM9/DHS_GPS_Manual_English_A4_24May2013_DHSM9.pdf.
- 25.
Habte, A., K. Lukas, and T. Tamirat, The level of Community-Based Essential Newborn Care utilization and associated factors among rural women in Southern Ethiopia, 2020: Based on the updated Community-Based Essential Newborn Care guideline. SAGE open medicine, 2022. 10: p. 20503121211067690.
- 26. Peven K., et al., Evaluating implementation strategies for essential newborn care interventions in low-and low middle-income countries: a systematic review. Health policy and planning, 2020. 35(Supplement_2): p. ii47–ii65. pmid:33156939
- 27. Alamneh Y., et al., Essential newborn care utilization and associated factors in Ethiopia: a systematic review and meta-analysis. BMC pregnancy and childbirth, 2020. 20: p. 1–9. pmid:32093648
- 28. Asabu M.D. and Altaseb D.K., The trends of women’s autonomy in health care decision making and associated factors in Ethiopia: evidence from 2005, 2011 and 2016 DHS data. BMC Women’s Health, 2021. 21: p. 1–9.
- 29. Tesema G.A., et al., Spatial clusters distribution and modelling of health care autonomy among reproductive‐age women in Ethiopia: spatial and mixed‐effect logistic regression analysis. BMC Health Services Research, 2021. 21: p. 1–12.
- 30.
Rutstein S.O. and Rojas G., Guide to DHS statistics. Calverton, MD: ORC Macro, 2006. 38: p. 78.
- 31. Kissling W.D. and Carl G., Spatial autocorrelation and the selection of simultaneous autoregressive models. Global Ecology and Biogeography, 2008. 17(1): p. 59–71.
- 32. Ord J.K. and Getis A., Local spatial autocorrelation statistics: distributional issues and an application. Geographical analysis, 1995. 27(4): p. 286–306.
- 33. Gribov A. and Krivoruchko K., Empirical Bayesian kriging implementation and usage. Science of the Total Environment, 2020. 722: p. 137290.
- 34. Lam N.S.-N., Spatial interpolation methods: a review. The American Cartographer, 1983. 10(2): p. 129–150.
- 35. Block R., Software review: scanning for clusters in space and time: a tutorial review of SatScan. Social Science Computer Review, 2007. 25(2): p. 272–278.
- 36. Songchitruksa P. and Zeng X., Getis—Ord spatial statistics to identify hot spots by using incident management data. Transportation research record, 2010. 2165(1): p. 42–51.
- 37.
Nikitopoulos, P., et al. Hot spot analysis over big trajectory data. in 2018 IEEE International Conference on Big Data (Big Data). 2018. IEEE.
- 38.
Scott L.M. and Janikas M.V., Spatial statistics in ArcGIS, in Handbook of applied spatial analysis: Software tools, methods and applications. 2009, Springer. p. 27–41.
- 39. Koo H., Chun Y., and Griffith D.A., Integrating spatial data analysis functionalities in a GIS environment: Spatial Analysis using ArcGIS Engine and R (SAAR). Transactions in GIS, 2018. 22(3): p. 721–736. pmid:30828255
- 40. Wang H. and Zhu J., Variable selection in spatial regression via penalized least squares. Canadian Journal of Statistics, 2009. 37(4): p. 607–624.
- 41. Chi G. and Zhu J., Spatial regression models for demographic analysis. Population Research and Policy Review, 2008. 27: p. 17–42.
- 42.
Charlton, M., S. Fotheringham, and C. Brunsdon, Geographically weighted regression. White paper. National Centre for Geocomputation. National University of Ireland Maynooth, 2009. 2.
- 43. Opdenakker M.-C. and Van Damme J., The importance of identifying levels in multilevel analysis: an illustration of the effects of ignoring the top or intermediate levels in school effectiveness research. School Effectiveness and School Improvement, 2000. 11(1): p. 103–130.
- 44. Kianifard F. and Gallo P.P., Poisson regression analysis in clinical research. Journal of biopharmaceutical statistics, 1995. 5(1): p. 115–129. pmid:7613557
- 45.
Saputro, D.R.S., A. Susanti, and N.B.I. Pratiwi. The handling of overdispersion on Poisson regression model with the generalized Poisson regression model. in AIP Conference Proceedings. 2021. AIP Publishing.
- 46. Maxwell O., et al., Modelling count data; a generalized linear model framework. Am J Math Stat, 2018. 8(6): p. 179–183.
- 47. Green J.A., Too many zeros and/or highly skewed? A tutorial on modelling health behaviour as count data with Poisson and negative binomial regression. Health Psychology and Behavioral Medicine, 2021. 9(1): p. 436–455. pmid:34104569
- 48.
Azanaw, M.M., et al., Determinants of Numbers on Antenatal Care Visits in Rural Ethiopia: A Multilevel Negative Binomial Regression Analysis. 2021.
- 49. Abdu H., et al., Knowledge and practice of immediate newborn care among midwives and nurses in public health facilities of Afar regional state, Northeast Ethiopia. BMC pregnancy and childbirth, 2019. 19(1): p. 1–10.
- 50. King R., et al., Barriers and facilitators to accessing skilled birth attendants in Afar region, Ethiopia. Midwifery, 2015. 31(5): p. 540–546. pmid:25745841
- 51. Wester K.C., et al., Best practices for addressing socio-cultural barriers to reproductive, maternal and neonatal health service utilization among women from pastoralist communities of Afar, Ethiopia. The Ethiopian Journal of Health Development, 2018. 32(Special Is).
- 52. Jalu M.T., et al., Exploring barriers to reproductive, maternal, child and neonatal (RMNCH) health-seeking behaviors in Somali region, Ethiopia. PLoS one, 2019. 14(3): p. e0212227. pmid:30875382
- 53. Oladeji M.T.O. and Oladeji B., Strengthening quality of maternal and newborn care using catchment based clinical mentorship and safe delivery app: A case study from Somali region of Ethiopia. International Journal of Midwifery and Nursing Practice, 2022. 5(1): p. 13–18.
- 54. Eba K., et al., Mobile health service as an alternative modality for hard-to-reach pastoralist communities of Afar and Somali regions in Ethiopia. Pastoralism, 2023. 13(1): p. 1–12.
- 55. Germossa G.N., et al., Availability of comprehensive emergency obstetric and neonatal care in developing regions in Ethiopia: lessons learned from the USAID transform health activity. BMC Health Services Research, 2022. 22(1): p. 1–10.
- 56. Medhanyie A.A., Alemayehu M., and Mulugeta A., Triad limiting the provision and uptake of reproductive, maternal and neonatal health services in the pastoralist communities of Afar, Ethiopia. The Ethiopian Journal of Health Development, 2018. 32(Special Is).
- 57. Sen K.K., Jamee A.R., and Bari W., Women’s multidimensional empowerment index and essential newborn care practice in Bangladesh: The mediating role of skilled antenatal care follow-ups. Plos one, 2023. 18(2): p. e0281369. pmid:36749769
- 58. Omotosho A., et al., How effective is antenatal care in preparing mothers for newborn care? An exploratory survey of lactating women in a rural Nigerian district. Heliyon, 2022. 8(11).
- 59. Saaka M. and Iddrisu M., Patterns and determinants of essential newborn care practices in rural areas of northern Ghana. International Journal of Population Research, 2014. 2014.
- 60. Waiswa P., et al., Differences in essential newborn care at birth between private and public health facilities in eastern Uganda. Global health action, 2015. 8(1): p. 24251. pmid:25843495
- 61. Amsalu E.T., et al., The effects of ANC follow up on essential newborn care practices in east Africa: a systematic review and meta-analysis. Scientific Reports, 2021. 11(1): p. 12210. pmid:34108559
- 62. Tafere T.E., Afework M.F., and Yalew A.W., Does antenatal care service quality influence essential newborn care (ENC) practices? In Bahir Dar City Administration, North West Ethiopia: a prospective follow up study. Italian journal of pediatrics, 2018. 44(1): p. 1–8.
- 63. Darmstadt G.L., et al., Evidence-based, cost-effective interventions: how many newborn babies can we save? The Lancet, 2005. 365(9463): p. 977–988. pmid:15767001
- 64. Shiferaw K., et al., The effect of antenatal care on perinatal outcomes in Ethiopia: A systematic review and meta-analysis. PloS one, 2021. 16(1): p. e0245003. pmid:33444374
- 65. Chichiabellu T.Y., et al., Essential newborn care practices and associated factors among home delivered mothers in Damot pulasa Woreda, southern Ethiopia. Reproductive health, 2018. 15(1): p. 1–11.
- 66. Målqvist M., Pun A., and Kc A., Essential newborn care after home delivery in Nepal. Scandinavian Journal of Public Health, 2017. 45(2): p. 202–207.
- 67. Kim E.T. and Singh K., The state of essential newborn care by delivery location in Bangladesh. Maternal and child health journal, 2017. 21: p. 2078–2085. pmid:28712021
- 68. Denktaş S., et al., An urban perinatal health programme of strategies to improve perinatal health. Maternal and child health journal, 2012. 16: p. 1553–1558. pmid:21870042
- 69. Coxon K., Sandall J., and Fulop N.J., To what extent are women free to choose where to give birth? How discourses of risk, blame and responsibility influence birth place decisions. Health, risk & society, 2014. 16(1): p. 51–67.
- 70. Akter T., Dawson A., and Sibbritt D., What impact do essential newborn care practices have on neonatal mortality in low and lower-middle income countries? Evidence from Bangladesh. Journal of Perinatology, 2016. 36(3): p. 225–230. pmid:26633143
- 71. Mersha A., et al., Essential newborn care practice and its predictors among mother who delivered within the past six months in Chencha District, Southern Ethiopia, 2017. PloS one, 2018. 13(12): p. e0208984. pmid:30533040
- 72. Krishnamoorthy Y., Majella M.G., and Rajaa S., Equity in coverage of maternal and newborn care in India: evidence from a nationally representative survey. Health Policy and Planning, 2020. 35(5): p. 616–623. pmid:32236550
- 73. Sunny A.K., et al., Out of pocket expenditure for sick newborn care in referral hospitals of Nepal. Maternal and Child Health Journal, 2020. 24: p. 57–65. pmid:31981065
- 74. Wuneh A.D., et al., Wealth-based equity in maternal, neonatal, and child health services utilization: a cross-sectional study from Ethiopia. International Journal for Equity in Health, 2019. 18: p. 1–9.
- 75. Dahab R. and Sakellariou D., Barriers to accessing maternal care in low income countries in Africa: a systematic review. International journal of environmental research and public health, 2020. 17(12): p. 4292. pmid:32560132