High prevalence of under-ve child mortality in the pastoralist communities of Ethiopia: Evidence from the 2016 Ethiopian demographic and health survey data

Objective: The aim of this study was to assess the prevalence and association of child mortality in the pastoralist regions of Ethiopia. The study is a further analysis from 2016 Ethiopian Demographic and Health Survey data. Results: The prevalence of under-ve child mortality in the pastoralist’s regions was 23.2%, 95%CI (21.4%, 24.6%). The prevalence of mortality among daughters was 15.4%, 95%CI (14.2, 16.6%), and sons 16.8%, 95%CI (15.6, 18.1%).In logistic regression, wealth index, head of household, Khat chewing, type of child birth, husband education, and child age in months were associated with under-ve mortality irrespective of the deceased children’s gender. The prevalence of under-ve child mortality in the pastoralist regions of Ethiopia was high, which was far highest in relative to the national under-ve mortality prevalence. In assessing the effect of variables on under-ve child mortality by gender, almost all the variables that have an effect on female or male child are similar. The government should emphasize on the pastoralists’ regions to decrease the high prevalence of under-ve child mortality.


Introduction
Worldwide, nearly 15,000 children die daily from preventable and treatable diseases such as diarrhea, malaria and fever (1). However, the world has made substantial progress in reducing child mortality over the last four decades (2)(3)(4). Despite such global progress in reducing child mortality over the past decades, an estimated 5.4 million children under the age of ve years were died in 2017, and half of those deaths occurred in sub-Saharan Africa (5). Under-ve mortality is a leading indicator of child health and overall development of a nation, since it re ects the social, economic, and environmental conditions in which children live (6). Sub-Saharan Africa countries continue to confront signi cant challenges of the highest child mortality rates in the world. In this risky region, 98 deaths per 1,000 live births were recorded in 2012. Sixteen countries that had under-ve mortality rate above 100 deaths per 1,000 live births werealso in sub-Saharan Africa (7).The 2015 UN Inter-Agency Group for Child Mortality Estimation (IGME) report, Ethiopia reached its target for Millennium Development Goal 4 for child survival with an estimated under-ve mortality rate of 59 per 1000 live births in 2015. This represents an average reduction in mortality of 5% per year which was higher than the average for sub-Saharan Africa (2.9%) (8). However the pastoralist regions had a highest prevalence of under-ve mortality. For example the child mortality rates in Afar region were very high. In this region, both under-ve mortality rate and infant mortality rate were the highest in Ethiopia. It is worrying that during the past ten years no progress has been made to reduce early childhood deaths in this region. However childhood mortality rates are decreasing in Ethiopia, but in Afar region the rates are still increasing (9). The reason for such persistent child mortality in Afar region might be as a result of highest deprivation rate in health (80%) and nutrition (85%) compared to the other regions (10). The other three regions were also grouped in thetop fourregions in having thehighest rates of child mortality including Afar region. According to the 2011 Ethiopia Demographic and Health Survey data, under-ve mortality rates per 1000 live births in Benishangul-Gumuz was 169, in Afar was 127, in Gambelawas 122, and in Somalia123 (11). However, the sustainable development goal's target is to reduce under-ve mortality rate to below 25 under-ve deaths per 1000 live births at the end of 2030 (12).The aim of this study wasto assess the prevalence and association of under-ve child mortality in the pastoralist regions of Ethiopia.

Methods
Data collection period, study design, and data collection The data collection period for the 2016 EDHS was from January 18 to June 27, 2016. The 2016 Ethiopian Demographic and Health Survey (EDHS) data was used for this further analysis. The 2016 EDHS data was the fourth survey conducted in Ethiopia. The survey collected information on household's and respondent's characteristics, child health, infant and child mortality, malaria, maternal health, maternal mortality, nutrition, tobacco use, women's empowerment, anemia, domestic violence, environmental health, family planning, fertility and fertility preferences, and etc. The purpose of the EDHS is to provide up-to-date estimates of the key demographic and health indicators of the population (13). The survey included reproductive age group women, under-ve children, and productive age group men (aged 15-59 years) (13,14) Sampling technique and study population The 2016 EDHS data collected using a strati ed two stage sampling method to select a representative sample. All the regions of the country were strati ed into urban and rural areas. From the total 11 administrative states, 21 sampling strata were yielded. The samples of the enumeration areas (EAs) were selected independently in each stratum in two stages. The implicit strati cation and proportional allocation were achieved at each of the lower administrative levels by sorting the sampling frame within each sampling stratum before sample selection according to the administrative units at different levels, and by using a probability proportion to size selection at the rst stage of sampling. The 2016 EDHS data collected from 645 EAs with a probability proportional to the EA size and with independent selection in each sampling stratum.
The EA size is the number of residential households in the EA that was determined in the 2007 Ethiopian Population and Housing Census. According to the 2016 EDHS procedures, a household listing operation was implemented in the selected EAs, and the resulting lists of households served as the sampling frame for the selection of households in the second stage. All the under-ve children, who were usual members of the selected households or who spent the night before the survey in the selected households were eligible for the child survey (14). After managing the missing data, 3527 respondents that had under-ve children were included for this frther analysis. However, the 2016 EDHS data collected data from 10641 children. But, the original EDHS data were undergoningthrough rigorous phases of data re ning until we get the nal sample size.

Data collection tools and procedure
The EDHS usually use ve groups of questionnaires in collecting the data. Those questionnaires are the Household questionnaire, the woman's questionnaire, the man's questionnaire, the biomarker questionnaire, and the health facility questionnaire. The questionnaires were adapted from the DHS program's standard demographic and health survey questionnaires in a way to re ect the population and health issues relevant to Ethiopia. Questions that stated about children were integrated to woman's questionnaire.

Variables
The outcome variable of this further analysis was under-ve child mortality. The independent variableswere socio-demographic variables of both children and mothers, health services provided to children and the wider community, substance use such as maternal chat chewing and Cigarette smoking in considering the availability of the variables in the 2016 EDHS data.

De nition
Under-ve child mortality: The report of mothers that they lost children in their life timebefore two weeks of the 2016 EDHS data collection takes placed Under-ve children: Children from birth to 59 completed months at June 27/2016 Pastoralist regions: Afar, Somalia, Benishangul, and Gambella regions were considered as pastoralist communities in this further analysis.

Socio-demographic characteristics
In this study, 2385(67.6%) husbands did not have professional works, but 1641(46.5%) mothers involved in agricultural jobs. More than eighty ve percent 3008(85.3%) of mothers were rural residents, 2711(76.9%) mothers were gave birth in home, and 3505(99.4%) mothers were married. Regarding religion 2574(73.0%) mothers were a follower of Muslim, and 2593(73.5%) were unable to read and write.

Association of under-ve child mortality
In this study, the associations of under-ve child mortality and potential determinates were measured in considering the gender of deceased children. This means, the factors of mortality for under-ve female, and under-ve male children were computed separately. In the nal logistic regression model,wealth index, head of household, Khat chewing, type of child birth, husband education, and child age in months were associated with under-ve child mortality in the pastoralist regions of Ethiopia regardless of the deceased children's gender (Table3).The association of under-ve child mortality and potential factors were also assessed by gender. In the nal logistic regression model, wealth index, religion, head of household, wontedness of the child, husband education, and short and rapid breathing were associated with under-ve sons' mortality in the pastoralist regions of Ethiopia (Table4).In the nal logistic regression model, wealth index, religion, head of household, type of child birth, child age in months, husband education, and size of child at birth were associated with under-ve girls' mortality in the pastoralist regions of Ethiopia (Table5).

Discussion
The aim of this study was to assess the prevalence and association of under-ve child mortality in the pastoralist communities of Ethiopia. The present study reported 23.2%, 95%CI (21.4%, 24.6%) of under-ve child mortality in the pastoralist regions, Ethiopia.This current prevalence is similar with the global prevalence of under-ve mortality 21.6% (15), and the UN Inter-Agency Groupprevalence of under-ve mortality for Latin America and the Caribbean regions, 27.0% (16).The similarity might be as a result of comparable socioeconomic status between Ethiopia, and Latin America or the Caribbean regions. The present study's prevalence of under-ve child mortality is also similar with a study that reported a 22.9% of under-ve mortality from Afar region (17).The prevalence of the current study is higher than the prevalence of U5M in the west Africa countries that varied from the highest 11.1%in Sierra Leone, Nigeria 10.0%, Mali 10.6%, to the smallest prevalence of mortality in Cape Verde 1.7% (18). The difference might be as result of socio-economic difference. The former studies were a nationwide studies but the current study consider the poorest regions in Ethiopia that might contribute for the highest prevalence of under-ve child mortality. The prevalence of under-ve mortality in the pastoralist regions arehigher than the 2016 EDHS based studythat reported 6.7% (5). The difference might be as a result of socio-demographic difference between the pastoralist regions and the wider community, Ethiopia. The current study also reported farhigher than the other reports from Afar region that were 12.3% in 2005, and 12.7% in 2011 (17).The present study is higher than a study that conducted in Somalia, Benishangul, and Gambella regions (17). The difference might be as a result of study population and data collection methods difference. The former studies were a passive survey among motherchild pair who visited health facility that contradicts with the present active community survey. The current study is lower than the UN Inter-Agency Group report, 31.4% in 2016 at South-Eastern Asia (16). The difference might be as a result of thatthe current study is a proximate prevalence that might be biased because of recall bias, and social desirability bias.
In addition to the above general description, we considered the total prevalence of mortality by gender. In the 2016 EDHS data collection, the total number of children died in the house despite the year that children died were considered. This might create discomfort for mothers who lost many children. Thus, we analyzed the mortality data by sex independently.Accordingly, the prevalence of under-ve mortality among daughters was 15.4%, 95%CI (14.2, 16.6%), and among sons was 16.8%, 95%CI (15.6, 18.1%). The current prevalence among sons and daughters were almost similar by consensus but in 95%CI, the mortality among boys was high.
However, we discussed the prevalence of under-ve mortality in considering the lower con dence interval of daughters, and the upper con dence interval of sons (14.2, 18.1). The separate prevalence among daughters and sons might be more precise than the aggregate prevalence reported despite the sex of children. Accordingly, the current prevalence of U5M is higher than a study conducted in west Africa and that reported 11.1% in Sierra Leone, Nigeria 10.0%, Mali 10.6%, and in Cape Verde 1.7% (18). The difference might be as a result of study population, and socio-economic dissimilarity. The former study considered children despite their sex, however this study's analysis focus on male or female children from poorest regions of Ethiopia separately. Thepresent study's prevalence is consistent with a decomposition analysis conducted in Sub-Saharan Africa in considering Benin, Chad, Congo, Côte d'Ivoire, Ethiopia, Gabon, Malawi, Mozambique, Rwanda, Sierra Leone, Uganda, and Zambia. The analysis indicated the presence of inequalities between under-ve boys' and girls' mortality that reported high boy mortality (19), which is true for this study, even the difference is not too ample. This variation in boys and girls mortality also supported by other study (20), in which all the reports indicated that boys were more affected. The justi cation for such variation between girls and boys might be because of biological factors (boys lower resistance to infection, and higher risk of premature birth), and gender discrimination (differential feeding and medication) (21,22). There is also a study that supported the difference of U5M among sons and girls in Ethiopia (23). In contrary, a study in Ethiopia indicated a lack of signi cant difference of under-ve death between males and females (24).The variation might be as result of study design difference, in which the former study used a case-control design.
The case control design is not appropriate to report magnitude than cross-section design.
In assessing the association of under-ve mortality with predictors, we consider the gender of children. Initially, the study tries to assess the association of U5M irrespective of the deceased child's gender. Beyond this, we try to assess the association of U5M among daughters and among sons independently. The assumption we considered is that the factors for daughters and sons might have difference. Therefore contradict with a study that reported a preventive association (26). But this association is consistent with a study that reported from India (27,35).History of short, and rapid breathing ((1.74), (1.22, 2.48) increase the probability of under-ve sons' death than children who had not it. This association is supported by WHO report that indicated respiratory infections were one of the causes of U5M (36). Unlike the previous study conducted in Ethiopia (31), none of the pastoralist region have no association with U5M. Since the causes of high under-ve mortality in resource poor settings are complex and merit concerted efforts to clarify their implications to improve child survival(37), we try to evaluate the prevalence of U5M by selecting the most susceptible areas and identi ed factors through triangulating the data set by sex as mentioned above. The government of Ethiopia and the pastoralist regions will utilize this evidence to draft police as per the following recommendation we stated in the conclusion section.

Conclusion
The prevalence of under-ve child mortality in the pastoralist regions of Ethiopia was high, which was far highest in relative to the national under-ve mortality prevalence. The proportion of male to female child mortality was nearly 1. In assessing the effect of variables on under-ve child mortality by gender, almost all the variables that have effect on female and male children are similar. For example, wealth indexand husband education were associated with daughters, sons, as well as both male and female children aggregate mortality irrespective of sex. Multiple birth and age of children were associated with U5M, and increases the odds of U5M among daughters in assessing the association by sex. Religion was associated with U5M in the pastoralist regions. Thus, the government should emphasize on the pastoralists' regions to decrease the high prevalence of under-ve child mortality.

Limitations
The 2016 EDHS collected a crude child death data that was not limited to de nite years. Thus, the prevalence of child mortality in this study indicated a total death of children among mothers who enrolled in this study who have also under-ve children at the time of data collection. This indicates that the prevalence of underve mortality in this study might not be a precise indicator of child mortality. But, it indicates how many of mothers lost female children or and male children in their lifetimes up to the 2016 EDHS data collection takes placed. The other limitation of this study was the repetition of mothers' report while they complain for both male and female child deaths. However, we try to analyze the data separately for male and female child death.

Declarations
Ethics approval and consent to participate Not applicable Consent for publication Not applicable

Availability of data and materials
The raw materials that support the conclusions of this research can uploaded as a supplementary le together with the manuscript documents.

Competing interests
The authors declare that they have no con ict of interests