Under- ve Child Survival and Asociated Risk Factors in Zambia

James Nilesh Mulenga Department of Economics, School of Social Science, Mulungushi University, Kabwe Bwalya Bupe Bwalya (  bwalya1983@gmail.com ) Department of Economics, School of Social Science, Mulungushi University, Kabwe Chabila Christopher Mapoma Department of Population Studies, School of Humanities and Social Sciences. The University of Zambia, Lusaka Erica Siyoto Department of Economics, School of Social Science, Mulungushi University, Kabwe James Banda Deparment of Clinical Services, JSI USAID Discover-Health, 22 Kariba Road, Kansenshi, Ndola


Introduction
There is global consensus that child health is a fundamental component of sustainable development. For this reason, "reducing infant mortality and under ve mortality" was set as a target for achieving the Sustainable Development Goal (SDG) number 3 on healthy lives and wellbeing for all. Target 3.2 aims to reduce neonatal and under-ve mortalities to 12 and 25 deaths per 1000 live births, respectively [1]. As one of the measures of child health and the health status of the population, under-ve mortality is de ned as the number of deaths between birth and the fth birthday per 1000 live deaths.
Globally, the rate of under-ve mortality had fallen from 93 deaths in 1990 to 39 per 1,000 live births in 2018 [2]. Even with this notable drop, under-ve mortality remains a global public health challenge faced by many developing countries, particularly in Africa. According to [2] 5.3 million children under the age of ve died in 2018 compared to 4.4 million in 2017. The highest under-ve mortality for the years 2017 and to 1 death in every 13 under-ve children. Majority of these deaths (47%) were among neonates (newborns). In the same way, projected estimates by [3] suggests that about 52 million children below the age 5 will die between 2019 and 2030 The decline in under-5 mortality in sub-Saharan Africa (SSA), has been uneven across countries and age groups [4]. As a whole, the SSA countries will not achieve the SDG target 3.2 at the current rate unless they accelerate their efforts. According the World Population Prospects of 2019, SSA will only reach 40 deaths per 1000 live births by 2030 [5], Population Division, 2019). Zambia has made strides in reducing the under-ve mortality in recent years. Figure 1.1 compares infant mortality rate for Zambia with the SSA country's average. It shows that Zambia has a lower rate in comparison with the SSA average. In addition While noting positive strides made to improve child health, under-ve mortality remains signi cantly high in Zambia posting a signi cant challenge to attain the SDG target as set by the UN Member States. In view of this therefore, this study aimed to provide some input into the policy debate on how Zambia can accelerate progress towards achieving the SDG target of between 12-25 deaths per 1000 live births by endeavouring to establish risk factors associated with the survival of children under the age of ve.

Literature
Under-ve mortality has been widely studied around the world. Despite this extent studies, focusing on the length of time it takes from birth to death among children under the age of ve are not so common in countries such as Zambia. Under-ve mortality reviews suggest that there are various factors that can be associated with under-ve mortality among which socioeconomic, biological and environmental factors play a pivotal role [6]. However, so far, there seems to be no consensus in literature on the actual risk factors associated with under-ve mortality [7] Suggest that factors such as family size, shorter birth intervals, duration of breastfeeding, water sources and mother's income are associated with under-ve mortality in Ethiopia. A similar study in Ethiopia however found some more extensions to the argument and suggested that modern contraceptive use, tetanus vaccinations, mother's age, child's sex, parity, postnatal check-up, marital status, and source of drinking water were instead more associated with under-ve mortality [8]. Within this complex web of factors relating to risks associated with under-ve mortality, a study by [9] found that maternal age, place of residence, household wealth index, level of education, employment, marital status, religious background, birth type, birth order and interval, sex and size of child, place and mode of delivery contributed to under-5 mortality rate in much of SSA. Cementing this idea, a study by [10] in Ghana also found that shorter birth intervals increased the risk of under-ve mortality; however, they also pointed out that sleeping under a mosquito net and increased labour force participation of mothers reduces the risk.
In Zimbabwe, [11] found that children whose mothers who had used contraceptives before and whose children had postnatal check-ups had lower likelihood of dying before the age of ve comparatively. The study further observed that small birth size and higher birth order increased the risk of dying. In another study by [12], it was found that low wealth status, source of drinking water, having an HIV positive mother were positively related to under-ve and infant mortality in South Africa.
In Zambia, a study by [6] found that malaria, diarrhoea and respiratory infections caused mortality among under ve children. The study further noted that increased frequency of visits to health centre signi cantly reduced mortalities in children by 3 out of a 1000 live births each year. Another study by [12], established that children with marasmus were more likely to die before the age of ve with HIV infected children having higher risks of dying compared to HIV negative children.
With all these studies in perspective, none was dedicated, at least in Zambia, to investigate the timing of the deaths at under ve although some of the risk factors have been highlighted. In view of this visible lacuna, this study therefore was positioned to determine under-ve child survival and associated risks in Zambia.

Data
In order to achieve set objectives, this study used the 2018 Zambia Demographic and Health Survey (ZDHS) Children's dataset) and World Bank Development Indicators. The 2018 ZDHS is a cross-sectional nationally representative survey which uses a two-staged strati ed sampling method to obtain household sample. Various questionnaires were used to collect data from selected households included among them are the Woman's questionnaire, Man's Questionnaire, Biomarkers Questionnaire and Household Questionnaire. The data used in this study was obtained from birth records obtained from each of the women interviewed. A total of 13,683 out of the identi ed 14,189 women age 15-49 were interviewed translating to a response rate of 96%. The data is freely obtainable from the DHS Program website (www.DHSprogram.com)

Statistical Methods
Microsoft Excel 2013 and Stata 13 software were used to analyse the data. The rst step of analysis involved establishing percentage distributions and graphical presentations of variables of interest; once this was done, data was then prepared to perform survival analysis. The Survival analysis modelling involved the production of the Kaplan -Meier outputs with associated log-rank tests; further, the Cox proportional hazard regression model was used to produce estimates for the risk factors associated with under ve mortality. Kaplan Meier was used to estimate survival probabilities for children under the age of ve while the log-rank test was used to compare two or more survival functions in the model [14]. For this study, the log-rank test was used to test the equality of two or more Kaplan Meier survival curves.
The Cox Proportional Hazard Model was used to determine factors associated with the survival of children under the age of ve in Zambia. All variables were weighted to adjust for differences in probability of sample selection. Variables used were selected based on reviewed empirical literature based on studies undertaken by [7,12,15]. The dependent variable was the relative risk of death occurring between age zero and 59 months. While independent variables included sex of the child, access to water, sanitation, type of residence, birth interval, place of delivery, mode of delivery, marital status, duration of breastfeeding, education level of parents, wealth index, immunization status, family size, mother's age at birth and unmet need for contraception for mothers. Table 1.1 presents sample characteristics. About 51% of children were male with 78% of them having a birth interval of two years and above. Majority (21.5%) of interviewed mothers were aged between 20 to 24 years while only 6.4% were aged between 15 and 19. By education level, about 53% of mothers had primary education while only 3.3% had tertiary education. About 65% of the respondents lived in rural areas. A quarter of these children belonged to households classi ed as poorest (25.3%) and 78% of them resided in male headed households. Most of the respondents (81%) used open well water sources and 75% used pit and traditional toilets respectively.

Discussion Of Findings
Results from this study has established that under-ve mortality is affected by various risk factors which include mother's age, mothers education, birth interval, wealth status, type of place of residence and sex of the household head. The hazards for under-ve mortality increase with age of the mother, with highest hazards observed among women aged 45 to 49. These ndings are similar to a study by [16] who undertook to investigate under ve mortality patterns and associated maternal risk factors in SSA countries where it was observed that children in older women were more likely to die before attaining age 5. According to [16], older women lose their children mainly due to either being less educated of have higher parity comparatively. However, [18] found that under-ve mortality was in fact lower among children born from older mothers in Bhutan.
Apart from the signi cant effects of age on child survival, education of a mother, both at secondary (20%) and tertiary (55%) level, has a protective effect against under-ve mortality.. This nding is not in isolation as several other studies con rm this relationship [17,19,20,21]. These ndings underscore the importance of education to the health of children since educated women may be more enlightened on the importance of the health of children and able to engage in health seeking behaviour such as better child care, use of modern health facilities and curtailing of known cultural practices that may inadvertently promote IMR and U5M.
As observed by other studies [19,22] in other countries the wealth status of a household plays a very important role in child survival. In this study, results show that children in households classi ed as richest had lower (32%) hazards compared to those from households classi ed as poor. This is because wealth gives women the ability to access and utilize the health care services for themselves and their children. However, a study by [23] found that children born in households classi ed as rich and richer had no signi cant impact on child survival in Sierra Leone.
This study further found that hazard for under-ve mortality reduces as birth interval increases. These results are similar to those by [7,10]. On the contrary however, [11] observed that higher birth order increased the risk of under-ve mortality in Zimbabwe.
Contrary to expectations and norms in child health, this study found that children from rural areas had lower hazards of under-ve mortality, compared to those living in urban areas. This may be explained only in speculative forms; that various community based health care programmes [24], such as the Community Health Assistant (CHA) Program, being undertaken in rural areas of Zambia may already be yielding positive results in helping reduce under-ve mortality in Zambia. Other similar studies in other countries had mixed ndings; for example [17] found that place of residence had no signi cant effect on under-ve mortality in Ghana. A study by [25] however found contrary results where they observed that living in rural areas increase the hazards of under-ve mortality.
Finally, female headed households had 31% higher hazards of under-ve mortality, compared to male headed households. A similar study by [26] made similar conclusions. These could be a result of a mixture of aspects including general disadvantages that women, especially single women, face in society which are also comingled with power, access to resources and other traditional and cultural undertones that affect women, and by extension affect the survival of their children as well.

Policy Implication
In view of the ndings, the study has important policy implications at national level in monitoring the implementation of various infant and under ve interventions aimed at reducing IMR and U5M.
Determinants affecting under-ve mortality are multifaceted and a mixture of both demographic and socio-economic factors. In order for Zambia to achieve SDG 3.2, all interventions should have a demographic (access to reproductive services for older women) and socio-economic (enhance income and education levels to female headed households) lens at all levels of service provision and implementation. Nonetheless, these should be implemented in line with already known high impact under-5 interventions already being implemented if maximum results are to be achieved.

Conclusion
In our study, we explored various factors likely to be associated with the risk of under-ve mortality in Zambia. Based on our ndings, it is clear that Under-ve survival is greatly in uenced by various biodemographic and socio-economic factors. The Cox regression model established that mother's education, higher birth order, being from a household classi ed as rich and residing in rural areas reduces under-ve mortality. However, older mothers and female headed household increase the risk of under-ve mortality. It is clear that Under-ve mortality remains high in Zambia and in uenced in a signi cant way by two major factors -older ages at birth for mothers and female headed households. While the latter is a result of society and community settings and how they interact with various cultural and traditional norms as well as power relations and the gender perspective, the former could be associated with access to comprehensive reproductive health services coupled with access to education. Although this data included human beings, this data analysis was secondary and all methods were carried out in accordance with relevant guidelines and regulations based on the permission granted to us by DHS program and ZamStats for us to use the ZDHS dataset in particular the women's (ZMIR71FL) stata data le. All experimental protocols for survey methodology, biomarker measurements, and all instruments prior to data collection were approved by National and international Institutional Review Boards (IRBs) that is Tropical Diseases Research Centre (TDRC) in Zambia and at ICF in the USA. During data collection, informed consent was obtained from all subjects and/or their legal guardian(s) (ZamStats, MoH and ICF 2019). In addition, no potentially identifying information is part of this dataset.

Consent for publication
Not Applicable

Availability of Data and Materials
The datasets generated and/or analysed during the current study are not publicly available due the fact we had to get permission from DHS and Zamstats for us to use the dataset as such it requires us to get permission to allow others to use it, hence them being, available from the corresponding author on reasonable request.