Article Text
PDF
Abstract
Objective This study examines trends, spatial distribution and determinants of age at death of under-5 children in Ethiopia.
Design This study used secondary data from the 2000, 2005, 2011 and 2016 Ethiopian Demographic and Health Surveys. A multilevel partial ordinal logistic regression model was used to assess the effects of variables on the age at death of children under 5 years.
Setting Ethiopia.
Participants The final analysis included a sample of 3997 deaths of newborns, infants and toddlers.
Results A total of 1508, 1054, 830 and 605 deaths of under-5 children were recorded in the 2000, 2005, 2011 and 2016 survey years, respectively. The death of newborns, infants and toddlers showed a significant decrease from 2000 to 2016, with reductions of 33.3% to 17.4%, 42.4% to 12.6% and 45.2% to 11.6%, respectively. The analysis using Global Moran’s Index revealed significant spatial autocorrelation in mortality for each survey year (p<0.05). The intraclass correlation of age at death of under-5 children within regions was substantial. Furthermore, the odds of newborn deaths among under-5 children (OR: 0.638, 95% CI: 0.535, 0.759) were lower for those delivered in health institutions compared with those delivered at home.
Conclusions Throughout the survey years from 2000 to 2016, newborn children had higher mortality rates compared with infants and toddlers, and significant spatial variations were observed across different zones in Ethiopia. Factors such as child’s sex, age of mother, religion, birth size, sex of household head, place of delivery, birth type, antenatal care, wealth index, spatial autocovariate, Demographic and Health Survey year, place of residence and region were found to be significant in influencing the death of under-5 children in Ethiopia. Overall, there has been a decreasing trend in the proportion of under-5 child mortality over the four survey years in Ethiopia.
- Aging
- Public health
- Community child health
- Child protection
- PUBLIC HEALTH
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This study used data from four successive surveys (2000, 2005, 2011 and 2016) and examined the spatial variation of age at death of under-5 children in Ethiopia.
The surveys were conducted at 5-year intervals, which restricted the assessment of age at death of under-5 children.
As the cross-sectional nature of the Demographic and Health Survey data introduced recall bias, further research using more recent survey data is recommended.
Introduction
Age at death for under-5 children is the time/age of death of under-5 children before even celebrating their fifth birthday. Death is one of the demographic phenomena that influence population development. The death of children is of interest to demographers, policymakers and researchers because it is one of the most critical indicators of the health conditions of a country.1 2 Globally, deaths of under-5 children contribute to approximately 5.2 million fatalities, with approximately 47% occurring within the first month of life, 28% between 1 and 11 months, and 25% between 1 and 4 years of age.3 The global under-5 death rates fell to 38 deaths per 1000 live births in 2019 from 93 in 1990 and 76 in 2000, which is a 59% and 50% decline, respectively. A 2013 report on child mortality states that about three-fourths of all child deaths happened in two regions: Africa (46%) and South-East Asia (28%). More than 50% of these deaths were observed in only six countries: China, the Democratic Republic of the Congo, Ethiopia, India, Nigeria and Pakistan. On average, 1 out of every 11 children born in sub-Saharan Africa dies before the age of 5 years.3 4
In sub-Saharan Africa, under-5 mortality rates remain high compared with other regions. In 2016, the average under-5 death rate was 79 per 1000 live births, with 1 in 13 children dying before their fifth birthday. This is 15 times higher than the average in developed nations. The risk of death for a child in the highest mortality countries is about 60 times higher than in the lowest mortality countries. Sub-Saharan Africa accounts for the highest child mortality rate globally, with an average of 76 deaths per 1000 live births in 2019. Over 80% of the 5.2 million child deaths worldwide occur in sub-Saharan Africa and Central and Southern Asia, with more than half happening in sub-Saharan Africa. Ethiopia, Nigeria and the Democratic Republic of the Congo, all located in sub-Saharan Africa, account for half of the world’s child deaths.5 6
In Ethiopia, there were many factors like the mother’s educational status, place of residence, birth order, number of children, preceding birth intervals, household size/family size, source of drinking water, mother’s marital status, types of birth and breastfeeding status that have statistically significant influences to control under-5 deaths.7–10 The influencing factors of under-5 child mortality are mainly socioeconomic, demographic and biological factors. Demographic factors affect both endogenic and exogenic deaths, like complications at birth which are difficult to prevent and control, and exogenous deaths which can be prohibited by public health actions, immunisation and antibiotic treatment.11
Ethiopia, with its diverse geographical landscape, was taken into account in this study by considering the spatial effect on under-5 child mortality based on their zone of residence. The country is characterised by various ethnic groups residing in different regions and city administrations, which are further divided into zones. These zones have distinct characteristics in terms of economy, culture and political landscape, reflecting the unique identities within Ethiopia.12 These different identities of an area lead to the different distributions of death of under-5 children in different neighbouring areas. In Ethiopia, the distribution of age at death of under-5 children was different in zones. Even if much research was done on spatial effects in Ethiopia, there was no research studied on spatial effects within zones of Ethiopia.13 Rather, most research was studied on regional variations. This was one of the cases that made this study possible, which will show in detail the spatial variation of distribution of age at death of under-5 children in Ethiopia.
The problem of dependencies between individual observations occurs in survey research, where the sample is not taken randomly but cluster sampling from geographical areas is used instead. Thus, the use of the standard logistic regression model is no longer valid and reasonable. Hence, to draw proper inferences and conclusions from multistage clustered data, it is better to use multilevel modelling. Likewise, this analysis allows us to examine both between and within-group variation.14 Several studies2 6 13 were conducted on deaths of under-5 children and determinant factors in Ethiopia, but they did not show the spatial distribution of the problem within zonal areas in the country and there are still serious problems concerning under-5 children in Ethiopia. The death of under-5 children varies in the country in space and time by changing magnitude. This indicator is significantly affected by geographical space, with wide variations and inequalities among areas.2 Therefore, this study aimed to investigate the spatial distribution, trends and determinants of age at death of under-5 children in Ethiopia.
Methods
Study area
The study was conducted in Ethiopia. Ethiopia is federally decentralised into nine regions and two city administrations, and these zones are also subdivided into 72 administrative zones.10 15
Data source and study population
The data for this study were obtained from four consecutive Ethiopian Demographic and Health Surveys (EDHSs) conducted in 2000, 2005, 2011 and 2016. These surveys were conducted in collaboration with the Federal Ministry of Health and the Ethiopian Public Health Institute with technical assistance from Inner City Fund (ICF) international and financial as well as technical support from development partners. They are designed to provide an estimate of the health and demographic variables.10 16–20
Inclusion and exclusion criteria
In this study, all under-5 children in Ethiopia who died within the 5 years preceding 2000, 2005, 2011 and 2016 EDHSs were included. Under-5 children who died with incomplete data and enumeration areas with longitude and latitude 0° for spatial data exploration were excluded from the analysis.
Study variable
Dependent variable
Age at death of under-5 children was the outcome (dependent) variable and was categorised into three: newborns, infants and toddlers.21 Newborn death is death within the first month of life, infant death is child death within the first birthday, and toddler death is the child’s death between the first and the fifth birthday.22
Independent variables
Variables from mothers, households and children were included in this study, obtained from each EDHS (see table 1).
Definitions and categorisation of variables in the study
Patient and public involvement
None.
Statistical method of data analysis
Spatial analysis
Spatial analysis involves analysing data concerning their location or spatial relationships. It includes techniques for visualising phenomena, assessing spatial autocorrelation and modelling spatial relationships.23
Spatial autocorrelation analysis
Spatial autocorrelation refers to the relationship between the proximity of observational units and the similarity of their values.24 The Global Moran’s I statistic is commonly used to test for spatial autocorrelation. In this study, Moran’s I was employed to assess the distribution of age at death patterns among under-5 children.25 A value close to −1 suggests a dispersed distribution, a value close to +1 indicates clustering and a value of 0 implies random distribution. A statistically significant Moran’s I (p<0.05) rejects the null hypothesis of random distribution, indicating the presence of significant spatial autocorrelation or dependence.26 Global Moran’s I was calculated as:27
(1)
where n is the number of observations in the whole cluster, x i and x j are the observations at locations of i and j, x̅ is the mean of x, and w ij is the spatial weight between locations of i and j.
The autocovariate variable is intended to capture spatial autocorrelation.28 The autocovariate variable (S i) at any site i may be calculated as:
(2)
The autocovariate variable (S
i) is a weighted average of the geographical units among a set of k
i neighbours of the geographical unit i, where y
j is the response value of y at site j among the site i’s set of k
i neighbours. The spatial weight between the geographical unit i and j is w
ij
=1/ℎ
ij, where ℎij is the Euclidean distance between the centroids of geographical unit i and j: 
, where a and b represent the coordinate of the ith and jth geographical unit. Introducing the spatial autocovariate variable reflects the first law of geography expressed as spatial autocorrelation.26
Hot spot analysis and spatial interpolation
Hot spots of mortality among newborns, infants and toddlers were identified using a hot spot analysis, which identifies areas with higher occurrence rates of child mortality. Cold spots, on the other hand, indicate areas with lower occurrence rates.29 To estimate values at specific locations, the study used the ordinary kriging technique, a geostatistical interpolation method that relies on surrounding existing values.30 31 It is calculated as follows:
where 
is the predicted location, 
is the unknown weight of the measured value of pairs of points at ith zone, 
is the measured value of pairs of points at ith location and N is the number of zones.
Ordinal logistic regression
The study employed an ordinal logistic regression model to examine the relationship between age at death (newborn, infant and toddler) and a set of independent variables. The logit of the cumulative probability, representing the likelihood of the child’s age at death falling within a specific category, was modelled as a linear function of the predictor variables. The model is given as:
where:
α j=threshold value, X i=sets of factors or predictors, Si=spatial covariate and ρ=spatial autocorrelation.
The study used a partial proportional odds model to accommodate the proportional odds assumption for certain covariates. For variables where this assumption was not met, the model incorporated specific parameters that varied across different comparison categories.32 33
Multilevel ordinal logistic regression
Multilevel analysis was used to estimate and account for variation in age at death within and between clusters. A multilevel ordinal logistic regression model is employed to model age at death, which is an ordinal categorical variable nested within higher levels.34 To determine if there are variations, the interclass correlation coefficient (ICC) is used. The ICC quantifies the proportion of outcome variation attributed to systematic differences between clusters. Measures of variation, including ICC and proportional change in variance, were presented. ICC measures within-cluster variation, reflecting variation between individuals within the same cluster.35 36 The ICC indicates the proportion of unexplained variance that is at the cluster level and is given by:
(5)
where, 
was the cluster or region variance and 
was the level 1 variance.37
Data management and analysis used SPSS IBM V.21, SAS V.9.4, Stata and ArcGIS V.10.8 statistical software.
Results
Descriptive statistics
Table 2 provides the prevalence of newborn, infant and toddler mortality in the EDHSs conducted in 2000, 2005, 2011 and 2016. A total of 1508, 1054, 830 and 605 deaths of under-5 children were recorded in these respective survey years. The frequencies of deaths among newborns, infants and toddlers are shown for each survey year. The analysis reveals a decreasing trend in under-5 mortality, with most deaths occurring during the newborn stage.
Proportion of deaths of under-5 children from 2000 to 2016
Figure 1 illustrates the trends in the overall proportion of deaths among newborns, infants and toddlers across the survey years from 2000 to 2016. Χ2 tests were conducted to examine the associations between the age at death and each categorical independent variable. Significant covariates independently associated with age at death in the years 2000 and 2005 include mothers’ age, mothers’ education level, birth size, birth order, birth interval, antenatal care (ANC) visits and family size (see online supplemental table 1). Similarly, significant covariates independently associated with age at death in 2011 and 2016 include the place of residence, breast feeding, birth interval and mothers’ age (see online supplemental table 2).
Supplemental material
Trends of age at death of under-5 children in Ethiopia from 2000 to 2016.
Spatial analysis
Spatial distribution and autocorrelation
The analysis included a total of 533 clusters in the 2000 EDHS, 535 clusters in the 2005 EDHS, 597 clusters in the 2011 EDHS and 622 clusters in the 2016 EDHS. Each cluster was represented as a point on the map, indicating the proportion of age at death of under-5 children in that area. The map used green colour to represent areas with a low proportion of age at death, while red colour represented areas with a high proportion of age at death. Figure 2 shows the spatial distribution of age at death of under-5 children for newborns, infants and toddlers in the EDHSs conducted in 2000 (a), 2005 (b), 2011 (c) and 2016 (d).
Spatial distribution of newborn death per cluster from 2000 to 2016. EDHS, Ethiopian Demographic and Health Survey.
In the 2000 survey, the highest proportion of infant deaths per cluster was observed in Zone 1, Awi/Agew, Eastern Zone, West Shewa, Zone 5, Keffa, Gedio and KT Zones. Conversely, the lowest proportion of infant deaths was concentrated in Central, North Gondar, Waghimra, Southern, North Wollo, West Gojjam, East Gojjam, Kellem Wellega, West Wellega, Illubabur, Gurage, Selti, Hadiya, Liben and Assosa Zones (figure 3). Similarly, online supplemental figure 1 depicts the spatial distribution of toddler deaths among under-5 children across the survey years. The highest proportion of toddler deaths occurred in Western, North Gondar, Zone 3, East Gojjam, Assosa, North Shewa, West Shewa, East Wellega, Sheka and Guji Zones (see online supplemental figure 1).
Spatial distribution of infants’ death per cluster from 2000 to 2016. EDHS, Ethiopian Demographic and Health Survey.
The estimated values of the Global Moran’s Index were 0.060 (p<0.005) in 2000, 0.068 (p<0.027) in 2005, 0.076 (p<0.006) in 2011 and 0.068 (p<0.021) in 2016. These results indicate significant spatial autocorrelation in mortality rates across the study years.
Additionally, online supplemental table 3 presented the spatial autocorrelation of the independent variables. This analysis explored the spatial relationships and patterns within the data, providing insights into the spatial autocorrelation present in the study variables.
Hot spot analysis and spatial interpolation
In online supplemental figure 2, the red colour represents significant hot spots (high-risk areas) for child death, while the green colour indicates cold spots (low-risk areas) for the death of children under 5 years in Ethiopia. For newborns, the significant hot spot areas for death were observed in the Western, Central, North Western and North Gondar Zones. Cold spots were identified in Segen Peoples, Wolayita, South Omo, Benchi Maji, Jimma and Gamo Gofa Zones. For infants, hot spot areas were observed in Bale, Arsi, Segen Peoples, Alaba, Wolayita, South Omo, West Arsi, Dawuro and Jimma Zones, while cold spots were identified in Western, North Western, North Gondar, Central, Eastern, North Wollo, South Gondar and Awi/Agew Zones. In 2005, hot spot areas for infant death were observed in Fafan, East Hararge, Keffa and Selti Zones, while cold spots were identified in North Gondar, North Western, Southern, Central and Eastern Zones (online supplemental figure 3). For toddlers, a significant hot spot area for death was observed in Zone 1 of the Afar region, while cold spots were identified in Arsi, Horo Guduru and West Shewa Zones (online supplemental figure 4).
Spatial interpolation analysis identified high-risk areas (red colour) and low-risk areas (green colour) for child mortality in the spatial distribution maps. Online supplemental figures 5–7 show the interpolated mortality patterns for newborns, infants and toddlers, respectively.
Parameter estimates of partial proportional odds model
Among the predictor variables, the child’s sex, age of the mother, region, birth order, birth interval, residence, child size at birth, sex of the household head and spatial autocovariate violated the parallel line assumption (Χ2=143.26, p<0.0001). This suggests that each explanatory variable has a different effect on each category of age at death of under-5 children. On the other hand, predictor variables such as place of delivery, birth type, ANC, religion, wealth index and DHS year satisfied the test of the proportional odds assumption (p=0.623). This indicates that each explanatory variable had the same effect on each category of age at death of under-5 children.
Significant factors associated with newborn deaths (p<0.05) include the child’s sex, place of delivery, birth type, age of the mother, ANC visits, region, religion, wealth index, birth order, residence, child size at birth, DHS year and spatial autocovariate. The odds of newborn death were 0.687 times lower for infants delivered in health institutions compared with those delivered at home. The odds of newborn death were 2.216 times higher for multiple births compared with single birth. Newborns whose mothers received ANC had odds of death 0.711 times lower compared with those without ANC. Newborns from the Muslim religion had odds of death 0.714 times lower compared with those from the Orthodox religion.
The spatial variable showed a positive correlation of 3.846 with newborn deaths, indicating that clusters with a high incidence of newborn death were typically surrounded by clusters with a low incidence. Survey time also had a significant effect on the age at death of newborns. The odds of newborn deaths in 2005, 2011 and 2016 were 0.750, 0.721 and 0.638 times, respectively, compared with the reference survey year.
Multilevel partial proportional odds model analysis of the data
For newborns versus infants and toddlers, 
showed that 10.14% of the total variation in death of newborns versus (infants and toddlers) explained or accounted for by regions reveals that multilevel ordinal logistic regression is an appropriate model and the remaining 89.856% of the variation in death for children is explained by lower-level units within clusters. For (newborns and infants) versus toddlers, 
showed that 4.12% of the total variation in death of (newborns and infants) versus toddlers explained or accounted for by level two units reveals that multilevel ordinal logistic regression is an appropriate model and the remaining 95.88% of the variation in death for children is explained by lower-level units within clusters.
The estimated probabilities for newborns and infants become 
and 
, respectively. This indicates that the estimated likelihood of newborn mortality compared with infant and toddler mortality was 40.06%. Similarly, the average chance of mortality for newborns and infants compared with toddlers was 14.27%.
Parameter estimation using spatial multilevel partial proportional odds model
The variance component of the random effect representing variation between clusters was included, and covariates such as place of birth, birth type, ANC visit, wealth index and survey years were found to be significant and met the parallel line assumptions. On the other hand, covariates including the child’s sex, mother’s age, birth interval, birth size, household head’s sex and the spatial covariate were significant covariates that violated the parallel line assumptions (see online supplemental table 4).
The odds of newborn deaths among under-5 children with an average-sized child during birth were 0.637 times lower compared with those with low birth weight, while the odds of newborn deaths among under-5 children with high birth weight were 1.475 times higher. Similarly, the odds of newborn and infant deaths among under-5 children with high birth weight were 0.916 times lower compared with those with low birth weight, and the odds of newborn and infant deaths among under-5 children with high birth weight were 1.069 times higher. These comparisons were made within the same clusters while controlling for other variables.
The positive estimated spatial effects (0.347) suggested that areas with low rates of newborn death were typically surrounded by areas with similarly low rates, and vice versa. The likelihood of newborn deaths for children under 5 years old delivered at a healthcare facility was 0.638 times lower compared with those delivered at home. Likewise, the likelihood of newborn and infant deaths for children under 5 years old delivered at a healthcare facility was 0.638 times lower compared with those delivered at home. These comparisons were made within the same clusters while accounting for other variables. The OR had a 95% CI ranging from 0.535 to 0.759.
The odds of newborn death among under-5 children whose mothers had ANC during pregnancy were 0.633 times lower than the odds of newborn death among under-5 children whose mothers had no ANC during pregnancy. The wealth index of a family significantly affected the odds of newborn death among under-5 children. The odds of newborn death among under-5 children from medium and rich families were 0.798 and 0.776 times lower, respectively, than the odds of newborn death among under-5 children from poor families.
Furthermore, online supplemental table 5 used multilevel multinomial modelling to estimate the impact of covariates on the age at death of under-5 children. This approach allowed for a comprehensive analysis of the covariate effects and their association with different age categories of child mortality. The estimated marginal (total) effects of covariates were depicted in online supplemental table 6.
Discussion
The main objective of this study was to analyse the variations, trends and determinants of age at death among under-5 children in Ethiopia using data from the 2000–2016 EDHSs. The study employed spatial analysis, ordinal logistic regression and multilevel ordinal logistic regression models. The determinant variables considered in the study included the place of residence, child’s sex, sex of household head, wealth index, religion, type of birth, place of delivery, mother’s occupation, birth size, marital status, birth order, birth interval, maternal age, family size, ANC usage, spatial autocovariate and DHS year.
This study differs from previous research,38 39 by examining the spatial distribution of newborn, infant and toddler deaths within zones, providing a more granular analysis compared with regional-level studies. The findings reveal a combination of hot spots and cold spots across different zones within regions, offering a more nuanced understanding of the clustering patterns. Furthermore, the study highlights that newborn deaths among under-5 children are lower for girls compared with boys, and a similar trend is observed for newborn and infant deaths compared with newborn and toddler deaths. This aligns with previous studies conducted in Ethiopia,40 suggesting a potential biological advantage as a contributing factor.41
Throughout the survey years from 2000 to 2016, newborn children had higher mortality rates compared with infants and toddlers, and significant spatial variations were observed across different zones in Ethiopia. Factors such as child’s sex, age of mother, religion, birth size, sex of household head, place of delivery, birth type, ANC, wealth index, spatial autocovariate, DHS year, place of residence and region were found to be significant in influencing the death of under-5 children in Ethiopia.42 The difference can be attributed to disparities in rural and urban service accessibility and delivery. In rural areas, mothers do not have adequate knowledge about childcare during and after pregnancy. Especially during pregnancy, they often engage in strenuous work until childbirth, which can affect child nutrition and increase the risk of injuries, potentially leading to child mortality. These findings align with previous studies.6 43
This study also demonstrated that the odds of death of newborns among multiple births were more likely than newborns from single birth type in the same clusters. This finding is consistent with studies conducted in Ethiopia.40 44 45 Multiple births are associated with a high risk of various negative birth outcomes, which can contribute to a higher infant mortality rate. Additionally, these outcomes can pose risks to the health of the mother.28
This study also demonstrated that the odds of death of newborns among birth intervals 25–36 and ≥37 months were less likely than newborns from birth interval ≤24 months in the same clusters, respectively. This finding is consistent with studies conducted in Ethiopia.40 44 Multiple births are at high risk of numerous negative birth outcomes, and these outcomes contribute to a higher rate of infant mortality.45 In line with this, newborn children from middle and rich families had a lower likelihood of death compared with newborn children from poor families.45 46
The birth weight of under-5 children was found to be significantly associated with their likelihood of death. The study revealed that newborns with average and high birth weight had lower and higher chance of death, respectively, compared with newborns with low birth weight, while considering other variables constant within the same clusters. These findings align with previous studies conducted in Ethiopia.43 46 A possible justification might be birth size reflects the quality of care given to the mother, the health status of the mother during pregnancy, and poor nutritional status may influence size at birth and thereby could affect the risk of death of under-5 children.1 Newborns under female household heads had higher death rates compared with those under male household heads. Similarly, the death rate for newborns and infants was higher when the household head was female, even after controlling for other variables in the same clusters. This was in line with a study done by Santos et al.45 Additionally, the mother’s age was a significant predictor of death among children under-5 years and consistent with studies by Tadesse and Abate et al.43 46
In this study, it was observed that under-5 children born to mothers aged 30–34, 35–39 and 45–49 years had a lower likelihood of death as newborns compared with those born to mothers aged 15–19 years. Similarly, the odds of death for newborns and infants born to mothers aged 35–39, 40–44 and 40–49 years were lower and higher, respectively, compared with those born to mothers aged 15–19 years. These findings align with previous studies conducted in Ethiopia.47 48 The observed differences in vaccination rates among older mothers may be due to factors such as having a larger number of children and a higher workload in caring for their children. These factors could potentially limit the time and resources available for vaccination activities, leading to lower vaccination rates among older mothers.46
This study revealed that place of delivery is significantly related to the death of under-5 children. The likelihood of death of newborns and infants who were delivered at a health institution was less likely as compared with those delivered at home. This finding is consistent with studies conducted in Ethiopia.2 40 49 50 Mothers living in urban areas may have better access to health institutions for ANC follow-ups and media exposure. This improved access can positively influence their knowledge and caregiving practices, particularly during pregnancy, which can benefit their children.
Antenatal visits during pregnancy were found to be significantly associated with child death. The study revealed that newborn children whose mothers received ANC had a lower risk of death compared with children whose mothers did not receive any ANC during pregnancy. This finding suggests that attending ANC plays a protective role in reducing early child mortality by improving the nutritional status of mothers during pregnancy. Additionally, it highlights the financial barriers faced by poor women in accessing adequate care during pregnancy, contributing to wealth-related inequality in child health outcomes. These findings are consistent with previous studies conducted in Ethiopia,51 52 emphasising the importance of allocating resources to support vulnerable households.
The findings of this study indicate a significant positive spatial effect, where clusters with a low incidence of infant mortality were typically surrounded by clusters with a high incidence of infant mortality. Similarly, clusters with a high incidence of infant mortality were often surrounded by clusters with a high incidence of infant mortality. This pattern aligns with studies conducted in Brazil.45 Furthermore, the likelihood of newborns and infants dying during the period 2005–2016 was lower, respectively, compared with the newborns and infants in 2000.
This study used data from four successive surveys and examined the spatial variation in antenatal and delivery care utilisation. The findings can contribute to improving awareness of maternal healthcare utilisation and guide policymakers in implementing appropriate policies. However, this study has several limitations. The surveys were conducted at 5-year intervals, limiting the assessment of delivery and ANC within that time frame. Additionally, the cross-sectional nature of the DHS data introduced recall bias. Further research using more recent survey data is recommended to address these limitations. Additionally, a qualitative study is suggested to gain insights into how healthcare utilisation affects the likelihood of deaths among children under 5 years.
Conclusions
The study found that newborns had a higher death rate compared with infants and toddlers from 2000 to 2016, and there was significant variation in death rates across different zones in Ethiopia. Several individual-level factors such as child’s sex, age of mother, religion, birth size, sex of household head, place of delivery, birth type, ANC, wealth index, spatial autocovariate, DHS year and place of residence were identified as significant factors influencing the death of under-5 children in Ethiopia. Overall, the proportion of under-5 child deaths has decreased over the past four survey years in Ethiopia.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Acknowledgments
The authors acknowledge the Ethiopian Demographic and Health Survey for data availability.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors BTA proposed the first draft, conducted data analysis and interpretation, and wrote the manuscript. DBB and LMT edited, revised and wrote the manuscript. LMT was the guarantor and accepts full responsibility for the work and/or the conduct of the study, had access to the data, and controlled the decision to publish. Finally, all authors read and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.