Determinants of Neonatal Mortality in Kenya: Evidence From Kenya Demographic Health Surveys, 2008 and 2014

Background: This paper analyses the contribution of socio demographic, neonatal, maternal and health facility related factors to the neonatal mortality in Kenya. Methods: Data from the Kenya demographic and health survey 2008 and 2014 was analyzed. Results: Logistic regression showed that in 2008, newborns with a short interval of <2 years (OR=1.938, p=0.000), very small babies, (OR 2.25, p=0.160), low birth weight (OR=6.677, p=0.000), male children (OR=1.243, p=0.028) and neonates not breastfed immediately after birth (OR=2.768, p=0.005) increased the risk of mortality. In 2014, being born in urban areas (OR=1.323, p=0.028), low birth weight (OR=2.354, p=0.008), birth interval of <2 years (OR=1.549, p=0.028, boys (OR=1.443, 0.014), mothers who did not attend ANC and those who had <3 ANC visits had (OR=4.668, p=0.000) and (OR=1.572, p=0.003) respectively as associated with neonatal mortality. Conclusion: This paper emphasizes on mother nutrition education, immediate initiation of breast feeding, attending 4 ANC and hospital delivery for better birth outcomes.


Background
Neonatal mortality remains a significant public health problem worldwide and accounts for 60% of the newborn deaths in the middle and lower income countries (UNICEF,2017). This is because the neonatal period is considered as the most vulnerable time for a child's survival. Globally, 18 children out of every 1000 live births. In 2017 alone,2.5 million children died in the first month of life in 2017 alone (UNICEF, 2017). Of these deaths, approximately 1,000,000 die within the first week of life. Decline in neonatal has been realized globally, but this decline is slower compared to mortality among children between 1-11 months and those 1-4 years. The neonatal mortality rates fell by 51% from 37 deaths per 1000 live births in 1990 to the current 18 deaths per 1000 live births in 2017 a small reduction compared to the other groups.
Despite the global reduction in the neonatal mortality, disparities exist in across the regions. Neonatal mortality is highest in in sub Saharan Africa (SSA) and in South Asia (SA) each with an estimated Neonatal Mortality Rate (NMR) of 27 per 1000 live births, and children born in these two regions are nine times more likely to die in their first month of life than a child born in a high income country. More than 50% of under-five Mortality rates (U5MR) occur in the neonatal period. In Aouth Asia, the proportion of children dying in the neonatal period is higher at 60% of the total U5MR. These statistics makes neonatal mortality an important public health concern which have been prioritized under the Sustainable Development Goals (SDGs) previously referred to as the Millennium Development Goals (MDGs). In Kenya, the neonatal mortality rate is 22 deaths per 1000 live births with the urban areas having a higher NMR of 26 per 1000 live births compared to 21 deaths per 1000 live births in the rural areas. (UNICEF, 2015) Although Kenya has made positive progress in the overall childhood indicators over time, the country continues to train in the neonatal mortality indicators which have only marginally reduced from 33/1000 live births in (KNBS.2003 to 31/1000 live births in 2008/9 (KNBS.2009) and 22 deaths per 1000 live births in 2014 (KNBS.2014). Studies in several countries have shown that neonatal mortality results from a complex chain of biological, socio economic, demographic and health care related determinants. However common causes of neonatal mortality include birth asphyxia, pre-term birth complications, intra partum related factors, infections such as pneumonia and tetanus, low birth weight congenital malformations and neonatal sepsis (Jehan et al, 2008;Khatun et al, 2012;WHO. 2011). There exists a relationship between maternal health and neonatal survival. Poor maternal nutritional and health status has been related to poor birth outcomes and this is influenced by elements such as socio economic, demographic and biological factors. The low uptake of contraceptives by women in reproductive age, especially those from the rural areas, advances the explanation that unplanned pregnancies and increased parity which research has shown are important risk factors to neonatal survival. Consequently, the intricate relationship between the mothers' health and that of the neonate's means that measures like the essential ante natal care (ANC), access to emergency obstetric care, access to skilled attendance at birth, adequate nutrition, post-partum care, neonatal care and early initiation of breastfeeding if adequately implemented can ensure neonatal survival. and 183 neonates died within the first 28 days Neonates born to mothers residing in the rural areas had a slightly lower NMR compared to their counterparts in the urban areas (NMR: 31vs 32) the poorest households had the highest NMR compared to the middle class households (NMR: 41 vs 37). Neonates whose mothers had no education had the highest NMR compared to those with primary education (NMR: 41 vs 27). Male neonates had higher NMR compared to females (NMR: 36 vs 25) respectively. Neonates whose mothers perceived them as small had a higher NMR of 89 compared to an average size child whose NMR was 24.
In 2014 DHS analysis the NMR in urban areas as higher compared to neonates born in the rural areas (NMR 25 vs 20). The 2014 KDHS however showed that poor households had lower NMR compared to the rich households (NMR: 20 vs 23). Neonates whose mothers had no education had the low NMR compare to mothers with higher levels of education (NMR: 20 vs 23). 46 cox hazards regression` using the formula below: The binary logistic regression model shows the independent (adjusted) effects of the socio demographic characteristic on neonatal mortality. The results showed in the table 3 below show that infants born from the urban areas had higher odds of dying compared to those born in the rural areas (OR 1.323 p 0.028) in the2014 KDHS. The 2008 results did no show significant relationship. Children born of poor households showed a high risk of dying within the first month of life however, these findings were not statistically significant. The same trend was observed in 2014.

Maternal characteristics.
The model then analyzed the effects of maternal characteristics on the risk of neonatal deaths. The findings for both the 2008 and 2014 KDHS showed that infants born od mothers with low BMI had a lower risk of dying compare ho those with BMI above 25. These finds were inconsistent with existing literature on maternal BMI and the risk of neonatal mortality, where neonates born of mothers with low BMI had an increased risk of dying in the first month of life.
In regard to the preceding birth intervals, neonates born less than 2 years from the preceding birth had a higher odd of dying (OR =1.938, P=0.000) compared to those born within an interval of above 2 years. This was according to the KDHS 2008 study findings. The 2014 analysis showed a similar trend, with children within 2 years of the previous having 1.5-times likelihood of dying compared to those born with more than 2 years' birth interval. These associations showed statistical significance for both study periods.

Neonatal characteristics
The model that was considered in the analysis of child characteristics in relation to the risk of neonatal morality. In this model, the child related characteristics i.e. size of the child, birth weight of the child, sex of the child and the birth order of the child were all fit into the logistic regression model. The findings showed that first born (children of the first order) had the highest risk of dying compared to 4 th order and above. This was true in both the analyses of 2008 and 2014, however, these findings did not show statistical significance (OR=1.775, p=0.447 and OR=1.412, p=0.08) respectively. Children who were considered very small by their mothers at the time of birth were 2 times more likely to die compare to larger children. The 2008 findings showed that (OR=2.015, p=0.160) and the 2014 findings were OR=1.523, p=0.347). These findings were however not statistically significant. It was however noteworthy that children who were reported to be of average size at the time of birth in the 2014 KDHS study had the lowest risk of neonatal mortality with OR 0.446, p=0.031. Male neonates are more likely (OR 1.423, p=0.028) to die than female neonates in the first month of life. This was the case in the KDHS 2008 analysis. The scenario was similar in 2014 KDHS study with an OR 1.443, p=0.014). These associations were found to be statistically significant. Children who were not weighed birth and those born with low birth weight had the higher odds of dying compared to those of normal birth weight. In the 2008 KDHS study findings, LBW children were 6 times more likely to die in the neonatal period compared to normal weight children (OR=6.677, p=0.000). In 2014, LBW children were 2 times more likely to die (OR=2.354, p=0.008). these associations were found to be statistically significant.

Health care characteristics.
The analysis of the DHS findings showed that children born at the health facilities were at higher odds of dying within the neonatal period. However, these associations were not found to be statistically significant. In the analysis of the KDHS 2014 neonates of mothers who had not attended any ANC visits had a higher risk of death (OR=4.668, p=0.000) compared to mothers who attended more than 4 ANC visits. Similarly, mothers who attended between 0-3 ANC visits had 1.5 times more likely to lose their babes in the neonatal period compare to those who attended more than 4 ANC visits (OR= 1.572, p=0.003).
Neonates who were not immediately initiated to breastfeeding in 2009 KDHS were more likely to die (OR 3.142 CI (1.536-6.428), p=0.001 compared to neonates immediately initiated to breastmilk. A similar trend is also observed in 2014 with OR=2.496 CI 1.038-6.002 p=0.041.

Discussion
Overall the aim of the study was to identify risk factors associated with neonatal mortality in Kenya comparing data from KDHS 2008 and KDHS 2014 using a nationally representative sample. This analysis showed that several factors were significantly associated with neonatal mortality after adjusting for confounding variables. The analysis shows a decline in the neonatal mortality rates from 31 deaths per 100,000 to 23 deaths per 100,000 between 2008 and 2014. The decline in the neonatal mortality rates is an indicator towards the combined interventions by different sectors, including health and social awareness programs to support child survival. The analysis of the KDHS survey data showed that male children had a significantly higher risk of dying during the neonatal period compared to female neonates. This finding is consistent with other study findings which have had similar outcomes (Ezeh et al, 2014) who analyzed the determinants of neonatal mortality with evidence from the Nigeria demographic health survey under taken in 2008.
The findings also showed that mothers who perceived their neonates to be very small had a greater risk of dying in the first month compared to those who perceived their neonates as large. These findings are consistent with other similar studies in India and Asia have shown the same findings, it is important to note that the rationale for the perceived size of the child is subjective. However, birth size was an important proxy for birth weight considering than more than 50% of the neonates were not weighed at birth. Other studies have also shown correlation between birth size and actual birth weight. Low birth weight children had the greatest risk of dying within the first month of life. Neonates of mothers who had not attended any ante natal clinics were 4 times more likely to die in the first month of life compared to those whose mother attended ANC clinics. These findings are consistent to studies that have shown the protective role of ANC against neonatal mortality. In the meta-analysis of data across regions, Doku & Neupane (2017), findings showed that ANC attendance was protective against neonatal mortality.
Early initiation of breast feeding was a significant determinant of neonatal survival. Neonates who were initiated to breastfeeding after one hour had their risk of death increase compared to initiate to breastfeeding breastfed immediately. These findings were similar to other studies and reviews that showed that early initiation to breastfeeding was associated with increased risk of neonatal mortality (Khan et al, 2015). There was a significant association between birth weight and neonatal deaths. Children with low birth weight in 2008 were 6 times more likely to die in the first month of life compared to normal weight children. This was also the case in the 2014 study. Infants of mother with less than 2 years spacing had a higher risk of dying compared to those born with more than 2 years spacing.
The findings in this study show that there are association between various demographic, maternal, neonatal and health related characteristics with neonatal mortality.
The strengths and weaknesses of this study need to be considered. This study is nationally representative and a multi stage cluster sampling. Additionally, recall errors arising from the dates of birth and death given by the women. One of the weaknesses of this study is that only surviving women were interviewed, therefore missing on other neonatal deaths that may have also occurred as a result of the death of the mothers. Other factors such as gestational age that are known to affect the neonatal survival were not assessed in this study.

Conclusions
Analysis on the determinants of neonatal mortality show that sex of the child (being male), birth weight of the child, preceding birth interval living in urban areas and initiation of breast feeding were important determinants of neonatal mortality. In the 2008 KDHS, birth weight, birth spacing being male and initiation of breastfeeding one hour after birth. In the 2014 analysis, the most significant determinants of neonatal mortality include being born in the urban areas, children born with low both weight, children born less than 2 years after the previous birth and children whose mothers did not attend any ANC clinics and those who attended less than 4 clinic sessions.
Based on these findings, further social advocacy on the adoption of the recommended minimum of 4 ANC visits during pregnancy is imperative for expectant mother to enhance opportunities for care during pregnancy, improve nutrition support through supplementation of Iron and folic acid, nutrition education for dietary diversity and awareness creation on any danger signs during pregnancy. In addition, further sensitization on the need adopt appropriate family planning options by women should be scaled, so that they are able to make better decisions on when to have children after healing. This will go hand in hand in demystifying the misconceptions that are preventing uptake of contraceptive options among the women. At the community level, operationalization of the community level strategies especially through the community health front line workers to incentivize mothers to deliver their babies in health facilities especially in key in ensuring there is an increase in the number of expectant women delivering their babies under the supervision of a skilled health care provider. Finally, there should be further strengthening of baby friendly hospital initiatives (BFHI) and complimenting it with the baby friendly community initiative (BFCI) will ensure that more women are initiating their infants on breast milk immediately after birth and continue to breastfeed exclusively for 6 months, to increase the survival of the neonates since breastfeeding exclusively and on demand has been identified among the high impact nutrition interventions that have improved neonatal outcomes.