Risk Factors of Neonatal Mortality in Bangladesh

Introduction: To address United Nations Millennium Develop ment Goal 4 (MDG 4) on reducing childhood mortality rates by two-thirds by 2015, there is a need for better population-based data on the rates and causes of neonatal death. This study aims to identify the risk factors of neonatal mortality in Bangladesh. Materials and Methods: The study used data from the nationally representative 2007 Bangladesh Demographic and Health Survey. The survey gathered information regarding socioeconomic, demographic, environmental and maternal and child health care of 10,996 ever married women and 6,058 children. Both bivariate and multivariate statistical analyses were used to assess the relationship between neonatal mortality and contextual factors. Results: The prevalence of neonatal mortality was 37/1,000. The statistical analyses yielded quantitatively important and reliable estimates of neonatal death. The multivariate logistic regression analysis yielded significantly increased risk of neonatal mortality for children with mother who had no formal education, the Muslims, whose mother were adolescents of age 15-19, first ranked birth and twin babies. Conclusion: Emphasis should be given to improve female education in Bangladesh for a better chance of satisfying important factors that can improve infant survival: the quality of infant feeding, general care, household sanitation, and adequate use of preventive and curative health services.


Introduction
E ach year nearly four million newborns die during the fi rst 4 weeks of life and world-wide neonatal mortality makes up 40% of the total child mortality 1,2,3,4 .This means that the overall neonatal mortality rate is about 30/1,000 with an annual number of births in the world reaching 130 million 1 .Whereas developed countries maintain a neonatal mortality rate of 2-3/1,000, it is not uncommon that this rate reaches over 60/1,000 in the poorer segments of the world 5 .Of the deaths, 99% are accounting the developing countries, especially in sub-Saharan African and South Asia 1,5,6 .During the last 30 years, the reduction in neonatal mortality rates has been slower, compared to both under-fi ve and child mortality rates after the fi rst month of life 6,7,8 .
Millennium Development Goal 4 (MDG 4) calls for reducing the under-fi ve mortality rate by twothirds between 1990 and 2015.As global momentum and investment for accelerating child survival grow, monitoring progress at the global and country levels has become even more critical.The most recent Inter-agency Group for Child Mortality Estimation (IGME) estimates show that nearly 8.1 million under-fi ve children died in 2009, implies that more than 22,000 children die per day.These fi gures refl ect substantial progress.Globally, the under-fi ve mortality rate has fallen from 89 deaths per 1,000 live births in 1990 to 60 in 2009.But the rate of decline -a one-third reduction over 20 years -is insuffi cient to meet MDG 4, particularly in Sub-Saharan Africa, Southern Asia and Oceania. 9Neonatal and postneonatal mortality rates declined less 10 , 3.0% and 2.5% decrease respectively.As a consequence there is an increasing proportion of infant deaths occurring in the neonatal period worldwide, which now accounts for two-thirds of deaths in children less than one year old, and nearly four-tenths of all deaths in children less than fi ve years of age 8 .
Infant and child mortality rates refl ect a country's level of socioeconomic development and quality of life.They are used for monitoring and evaluating population and health programmes and policies.The rates are also important for monitoring progress towards the United Nations MDG to reduce child mortality as expected by the year 2015.It would be diffi cult to achieve MDG without reducing child mortality by two-thirds by 2015 11 .Thus the study on infant and child mortality; in particular, neonatal mortality is of great importance to monitoring the progress of children's health status.
Bangladesh has made substantial progress in the reduction of under-fi ve mortality.A recent estimate suggests that infant mortality reduced from 88/1,000 in 2000 to 55/1,000 in 2010.Absolutely, in Bangladesh, infant mortality reduced by 3.3% annually in the last decade.The neonatal and post neonatal mortality reduced from 65/1,000 and 38/1,000 in 1990 to 31/1,000 and 14/1,000 respectively in 2010. 5Besides, the Bangladesh Demographic and Health Surveys (BDHSs) reported that under-fi ve mortality reduced from 133/1,000 in 1993-1994 to 55/1,000 in 2007.In addition, the neonatal and post-neonatal mortality rates declined from 52/1,000 and 35/1,000 to 37/1,000 and 15/1,000 respectively during the same period 12,13 .The successive DHS surveys conducted in Bangladesh since 1993-1994 confi rm a declining trend in childhood mortality.During  1989-1993 to 2002-2006, infant mortality declined by 40% from 87/1,000 live births to 52/1,000.Even more impressive are the 72% decline in child mortality and the 51% decline in under-fi ve mortality over the same period.A comparison of mortality rates over the last three years shows that infant and child mortality declined by 20% and 42% respectively 13 .A recent study 14 conducted on rural areas at Matlab using the Health and Demographic Surveillance System (HDSS) data has shown gradual decline in under-fi ve mortality consistent with the reports of the BDHSs.
Whereas literatures on neonatal, post-neonatal, infant and child mortality are enormous in other developed and developing countries, those are limited in Bangladesh.Many studies carried out elsewhere on risk factors of neonatal mortality attempt to show causal relationship between neonatal mortality and utilization of maternal health care facilities, such as receiving antenatal care (ANC) services, place of delivery, skilled attendance during childbirth 15,16 .Caste, religion and standard of living index are also identifi ed as infl uential factors of childhood mortality in India 17 .A number of studies showed signifi cant relationship between mother's level of education 6,16,18 , maternal age 19,20,21,22 and early childhood mortality.Mother and child characteristics are analysed as determinants of child mortality and its components, socioeconomic characteristics, type of childbirth and birth defects have been extensively investigated 22 .Regional and social diff erences may infl uence the composition of population groups and the success of local health promotion programmes.Living in poor areas may have negative eff ects on children's health 23 .
Bangladesh is still far away to achieve the MDG 4. To achieve the goal, the infant mortality rate should be reduced at 31/1000 24 .It would be diffi cult to achieve MDG of reducing child mortality by two-thirds by 2015 without reducing neonatal mortality.Despite impressive progress, Bangladesh still has higher rate of infant and neonatal mortality.In Bangladesh, very few studies have reported the causes of early and late neonatal deaths 16 .Knowledge on some of the factors aff ecting child mortality is a fundamental requirement for devising appropriate policies and strategies to accelerate decline in neonatal mortality.This paper aims to investigate the risk factors aff ecting neonatal mortality in Bangladesh.The fi ndings of this study may help the policy makers, programme managers and donors in driving up understanding about the risk factors of early childhood mortality, particularly neonatal mortality of Bangladeshi children.

Materials and Methods
Data for this study have been taken from 2007 BDHS.The BDHS is a nationally representative survey of 10,996 women age 15-49 and 3,771 men age 15-54 from 10,400 households and 6,058 under-fi ve children born during the last fi ve years preceding the survey date covering 361 sample points (clusters) throughout Bangladesh, 134 in urban areas and 227 in the rural areas.This survey is the fi fth in a series of national-level population and health surveys conducted as part of the global Demographic and Health Surveys (DHS) programme.It was designed to provide data to monitor the population and health situation in Bangladesh as a follow up to the past four BDHS surveys.The survey utilised a multistage cluster sample based on the 2001 Bangladesh Census and was designed to produce separate estimates for key indicators for each of the six divisions of the country -Barisal, Chittagong, Dhaka, Khulna, Rajshahi and Sylhet.Data collection took place over a fi ve-month period from 24 March to 11 August, 2007.
The survey obtained detailed information on fertility levels, marriage, fertility preferences, awareness and use of family planning methods, breastfeeding practices, nutritional status of women and young children, childhood mortality, maternal and child health, and knowledge and attitudes regarding HIV/AIDS and other sexually transmitted infections (STIs).research fi rm located in Dhaka.Technical assistance was provided by Macro International Inc. through the MEASURE DHS programme.Financial support for the survey was provided by the U.S. Agency for International Development (USAID/Bangladesh).However, the details of the survey are given elsewhere 13 .

Conceptual Framework
Many researchers such as Mahmood 25 and later on Titaley and his colleagues 6 adapted the conceptual framework proposed by Mosley and Chen 26 with some modifi cations based on the limitations and structure of the DHS data.Keeping in view the aforementioned frameworks, this study considered the "survival status" of children as the outcome variable.According to Mosley and Chen 26 framework, the socioeconomic factors aff ect the outcome variable through the four proximate determinants namely, socio-demographic factors, environmental factors, nutritional factors and health seeking behaviour of mother.Due to scarcity of information we also used a modifi ed model to investigate the risk factors for neonatal mortality in this study.The BDHS could not gather information of nutritional status of all living and death children such as height, weight or size of children at the time of their birth.Rather, the survey gathered information regarding nutritional status for the living children at the time of survey.Hence, we used mother's body mass index (BMI) as the proxy determinant.

Study Variables
The primary outcome variable of this study is "neonatal death", which is defi ned as the death of a live born infant in the fi rst four weeks of life.In the descriptive analyses, the neonatal mortality rate has been expressed as percentage.The outcome variable "neonatal death" was recorded as a binary variable.The explanatory variables included in this study have been defi ned in Table 1.

Statistical analyses
Both bivariate and multivariate statistical analyses have been used in this study.Bivariate analyses namely chi-square tests have been applied to examine the association of various independent variables and neonatal mortality.This was followed by multivariate logistic regression analysis.The logistic regression model used in this study is as follows: 1 1 e âX P − = + β where, P is the probability of death of a child within the fi rst four weeks of his/her born, β is a vector of unknown coeffi cients and X is the vector of covariates that aff ect the outcome variable.Thus the general multivariate logistic regression model can be expressed as: which expresses the log odds of the outcome variable as a linear function of the independent or explanatory variables.The results of the logistic regression analysis have been presented by regression coeffi cient (β), standard error (SE), odds ratios (OR) (expβ) with 95% confi dence interval (CI).The data of this study have been analysed by means of SPSS v17 software.

Background characteristics of the children
The characteristics of the study children are presented in Table 2 with the prevalence of neonatal mortality by socioeconomic, demographic, environmental and maternal health care services strata.As shown in the table, slightly over two-fi fths were from poor households and a slightly less than two-fi fths were from rich households.Four in fi ve children were from rural areas.The more children were from Dhaka division, followed by Chittagong, Rajshahi, Khulna, Sylhet and Barisal divisions respectively.With regards to mother's education, slightly over one-fourth had no formal education and over two-fi fths had at least secondary level of education.The vast majorities were Muslims.29% of the mothers were adolescents during childbirth, almost one-third were aged 20-24, one-fi fth were aged 25-29 and the rests were of age 30-49.One-third of the children were the fi rst ranked children, one-fi fth were the second ranked children and the rest were third or higher ranked children.Half of the children were male and half were female.Almost 99% of the children were singleton birth and slightly over 1% was twin or multiple in terms of foetus type.Approximately one-third of the mothers were thin, three in fi ve were normal and 8% were overweight or obese.One-fourth of the children used to defecate in the unhygienic places, only 8% children were from households with less polluted cooking fuel and only 3% had no facilities of safe drinking water.Over half of the mothers received ANC services, whereas less than one-fourth sought skilled birth assistance (SBA) and only 15% births were delivered in medical facility places.

Prevalence of neonatal mortality
The third broad column of Table 2 shows the percentage of neonatal mortality by various characteristics under socioeconomic, demographic, environmental and maternal health care factors.Primarily sixteen variables were included in the analyses to examine their association with neonatal mortality.Overall, the neonatal mortality was 37/1000.Of the variables included in the study wealth index, region, As expected, the prevalence of neonatal mortality was 1.1% higher in rural than urban areas (urban 2.8% vs. rural 3.9%).Although, the administrative regions did not show signifi cant association with neonatal mortality, the prevalence varied 30/1,000 in Khulna division to 52/1,000 in Sylhet division.Maternal education and maternal age showed negative association with neonatal mortality.The Muslims had higher rate of neonatal mortality than their peer non-Muslims.The prevalence of neonatal mortality was higher for the fi rst and fi fth or higher ranked births than that of the the second, third or fourth.The prevalence of neonatal mortality among twin or multiple births was 325/1,000 as against of 33/1,000 for singleton birth.Households with unhygienic toilet facility and polluted cooking fuel showed higher prevalence of neonatal mortality than those with hygienic toilet facility and less polluted cooking fuel.

Results of multivariate logistic regression analysis
The results of the multivariate logistic regression analysis have been shown in Table 3.All the variables identifi ed to have signifi cant association with neonatal mortality were included in the multivariate analysis to assess their net eff ect.After controlling for other confounding factors, the environmental factors and maternal health care services appeared to have no signifi cant eff ect on neonatal mortality.The variables showed to have net eff ect on neonatal mortality were maternal education, religion, maternal age, birth order and type of foetus.
Maternal education showed negative relationship with neonatal mortality.For instance, the odds of neonatal death was signifi cantly 28%, 33% and 85% lower among children whose mother had primary, secondary and higher education than those who had mothers with no formal education.The non-Muslim children had 40% lower risk of neonatal death as compared to the Muslim children.Maternal age showed negative relationship with neonatal mortality of the children.As compared to fi rst ranked birth, the second ranked birth had signifi cantly 50% lower risk of neonatal death, whereas the diff erence of likelihood of neonatal death between third and higher ranked births and the fi rst ranked had no signifi cant eff ect.The twin or multiple births were at 15.2 times as likely as to be dying at neonatal period than that of the singleton birth.

Discussion
To address United Nationas MDG 4 on reducing childhood mortality, there is a need for better populationbased data on the rates and causes of neonatal death 27 .In this study it was our aim to identify the risk factors of neonatal death in Bangladesh using the nationally representative 2007 BDHS data.Both bivariate and multivariate statistical analyses were employed to examine the factors aff ecting neonatal death.Findings showed that the prevalence of overall neonatal death was 37/1,000.Our analyses revealed that socioeconomic and demographic factors rather than environmental factors and maternal health care services had signifi cant net eff ect on neonatal death.The variables infl uencing neonatal death are maternal education, religion, maternal age, birth order and type of foetus.
Consistent with earlier studies 18,28,29,30 , the fi ndings of this study revealed that the higher the mother's education the lower the risk of neonatal mortality.A review of the literature shows that while the higher socioeconomic status of better educated women explains about half of the magnitude of the relationship between maternal education and child survival 31 , the domestic health practice of individual women is probably the new most salient mechanism in the maternal education child mortality relationship.The fact that mother's education is a more important determinant of child survival than father's education: explains that there is greater maternal involvement in child-health related care 32 .The mother's education infl uences her choices and skills in health care practices 32,33 .For instance, both educated and illiterate mothers recognize when their child is sick but an educated mother more frequently will take action "without waiting for (her) husband or mother-in-law to notice the child's condition too" 32 .This is partly because illiterate women do feel a lack of capability when dealing with the modern world.
We found that being born to a mother who practiced Islam was signifi cantly associated with an increased risk of neonatal death.The infl uence of religion on neonatal mortality may be linked to the beliefs and myths they may have concerning child birth.This was the case with the Faith Assembly people in Indiana, U.S., who believed that child bearing was an act of God not to be interfered with 34 .It is evident that Muslim women have higher fertility, lower use rate of contraceptive method, early age at marriage and motherhood and lower access to maternal health care services than their peer non-Muslim sisters in Bangladesh 35,36 .Hence, we explored this association further by introducing an interaction term between religion and maternal age (results not shown).We found that children who were born to Muslim adolescent mothers had a higher risk of neonatal deaths compared to children born to non-Muslim mothers (results not shown).Despite this, it is diffi cult to say whether or not this fi nding refl ects the early motherhood among the subgroup of traditional Muslim women and thus further qualitative research may be useful.
Our fi ndings showed signifi cant increased risk between adolescent motherhood and neonatal mortality.This result is comparable to numerous studies conducted elsewhere which found higher risk of neonatal mortality in younger adolescents than older mothers 19,20,22 .This is partially explained by diff erences in socioeconomic factors in younger versus older women and is mediated primarily through preterm delivery, small for gestational age (SGA) and low birth weight (LBW) or some interaction of these variables 37 .The signifi cance of young and old maternal ages at childbirth as risk factors for adverse neonatal health outcomes remains largely socioeconomic, environmental and cultural context dependent.Previous research that aimed at providing an explanation to the frequently observed association between mother's age and various quality-of-life measures, including perinatal, neonatal and infant mortality 38,39 , preterm delivery 38,40 , and low birth weight 41,42 have yielded inconsistent results.Birth order of a child showed controversial results for neonatal mortality.Some studies showed that fi rst or lower ranked births were at higher risk of neonatal mortality, whereas some others showed that higher ranked births were at increased risk of neonatal mortality.For instance, in a study to determine the impact of maternal and child health (MCH) services on child survival in a socio-economically poor rural Pondicherry, India, infants of fi rst birth had higher risk of neonatal mortality than fourth or higher ranked births.The study further showed that second ranked births had the least mortality risk 42 .A study conducted in Taiwan 44 showed that children with fi rst and fi fth ranked births were at higher risk of early neonatal deaths, while in Nigeria 45 , the children with sixth or higher order births were at increased risk of neonatal mortality.It is argued that, in most of the developing countries, higher mortality risks were found for the fi rst-born children compared to birth order two through six 46 .Our study showed that children with fi rst ranked birth were signifi cantly at increased risk of dying at early infancy than the second or higher ranked births.Social and cultural contexts may be partly attributed to the infl uence on birth order on neonatal mortality.However, our studies are consistent with most of the earlier studies as mentioned above.
Multiple births are relatively rare events, but contribute substantially to mortality in both neonatal and post-neonatal periods 28 .Our study showed a reduced risk of neonatal mortality for singleton baby compared to twins or multiples.Twins are more likely to be born with low birth weight and biological immaturity.Possibly, the use of the curative services, which are needed in emergency situations, could be underutilized by mothers of twins 47 .However, one possible reason for this observed association is that multi-foetal pregnancy and multiple births including twins and higher order multiples such as triplets and quadruplets are high-risk pregnancy and birth.These high-risk births are frequently accompanied by a number of associated foetal and neonatal complications that require special and expensive medical care 48 .In addition, multiplebirth children are at greater risk of birth defects and/ or disabilities and accounted for larger percentage of prenatal deaths 49 .Therefore, mortality of these highrisk groups contributes to the higher rate of childhood mortality especially during the early period of life.This fi nding is consistent with earlier studies conducted elsewhere 28,47,48,49,50 .
It is important to discuss the limitations of this study as well as its strengths.The fi rst potential limitation of this study is the cross-sectional nature of the analyses.Second, the information used in this study is based on the retrospective birth history of children and reported characteristics of mothers and households by the mother's of the children, which may cause underreporting in age.Third, only surviving women were interviewed, which may have lead to an underestimate of the neonatal mortality rate, because of the association of neonatal deaths with maternal deaths.This could also have lead to an underestimate of the eff ect of some of the associated factors, such as delivery complications 6 .Fourth, as a secondary data set, we could not include weight of child at the time of birth which is an important predictor in the study of neonatal mortality.Another limitation with household wealth indices derived from DHS is that those are based on current status data so that the variable might not capture the true level of household wealth during the infancy of children born several years before the survey.However, since these analyses are restricted to births within fi ve years of the surveys, this biasness will not be substantial.Despite these limitations, the strength of the study is that it dealt with a large nationally representative sample size with stratifi ed random sampling gathered by international standardized method, which are mostly used in public health research.

Conclusion
In summary, we found that twin or multiple births are more likely to be dying during the fi rst four weeks of life as compared to singleton children.Maternal education, religion, maternal age and birth order are also important determinants of neonatal mortality in Bangladesh.This evidence suggests that improving maternal education may be key factor to improving child survival in Bangladesh.A well educated mother has a better chance of satisfying important factors that can improve infant survival: the quality of infant feeding, general care, household sanitation, and adequate use of preventive and curative health services.Adolescents should be discouraged to initiate childbearing at the teen ages not only for the reduction of neonatal mortality, also for their physical and psychological maturity that may bring overall wellbeing to mother and child health.
The 2007 BDHS was conducted under the authority of the National Institute for Population Research and Training (NIPORT) of the Ministry of Health and Family Welfare.It was implemented by Mitra and Associates, a Bangladeshi Risk Factors of Neonatal Mortality in Bangladesh

Table 1 :
Operational defi nition and categorization of the variables used in the analysis of child, mother's BMI, drinking water facility and maternal health care factors did not show to have signifi cant association with neonatal mortality.The variables identifi ed to have signifi cant association with neonatal mortality were place of residence, maternal education, religion, maternal age, type of foetus, toilet facility and cooking fuel.

Table 2 :
Background characteristics of under-fi ve children and prevalence of neonatal death, BDHS-2007

Table 3 :
Multivariate logistic regression analysis showing the risk of neonatal mortality among under-fi ve children of Bangladesh, BDHS-2007Note: level of signifi cance *** p<0.001; ** p<0.01; * p<0.05; and † p<0.10.§The fi gures do not round to 6,058 due to exclusion of non-dejure residents.‡ The fi gures do not round to various missing information.
Risk Factors of Neonatal Mortality in Bangladesh