Chapter 1: Introduction

In Vietnam, the mortality level, especially child mortality, has been reducing over years. The under-five mortality rate in 1960 was 89 per thousand live births has reduced to 42 per thousand in 2000 (UNICEF, 2001). However, differential of child mortality among regions is still high.

Since 1986, the renovations and introduction of a market economy have brought about crucial socio-economic changes in Vietnam. As Viet Nam continues its gradual transition to a market economy, besides many advantages, there are still some problems to be solved. One of those problems is the gap between the rich and the poor among regions.

Child mortality is one of the most important indicators to assess the socio-economic development of a nation; and lowering the level of child mortality is the first goal of UNICEF (2002). However, nearly 12 million under-five children around the world die every year, mostly from preventable causes. To date, there are many researches about child mortality in the world.

In Vietnam, previous studies about child mortality have suggested some recommendations regarding health policies, demography, and contributed to reducing child mortality, such as the Vietnam Population Census (1979, 1989, 1999), IDS (1994), DHS (1997), Nguyen (1998); Wagstaff and Nguyen (2001). However, there are not many researches.

According to Prime Minister's Decision on Approval of Strategy for People's Health Care and Protection 2001 - 2010, under-five mortality rate will reduce to about 32 per 1,000 live births. How can Vietnam reach the target in such limited time? These facts force and encourage researchers to study child mortality in Vietnam.

What are the major determinants of child mortality in Vietnam?

This research will concentrate on analysing the relationship between child mortality and socio-economic and demographic factors to find out weather they have any relation. These relationships are also analysed in multivariate way to give out some policy implications reducing child mortality in Vietnam. The objects of the study are under-five children.

To answer the research questions, this paper is a combination of method such as comparison, statistical methods, description and econometric method.

For satisfying the focuses of the analysis, this study will use the Vietnam Living Standard Survey (VLSS) 1997/1998. Although this data are not strong on the subject of mortality, they do provide a wealth of information on socio-economic and demographic factors which is not available in other data sets like DHS or Cencus.

The study consists of five chapters and a bibliography. Apart from Chapter I: Introduction, other chapters present the analysis and findings. Chapter II introduces the analytical framework. Chapter III provides a picture of child mortality in Vietnam through data analysis. Chapter IV analyses determinants of child mortality based on the econometric model. Chapter V provides some conclusions and policy implications.

Chapter 2: Analytical Framework

2.1. Concepts and Definitions

Child Mortality. First of all, we examine the concept of “death” before studying the concept of “mortality”. “Death” is defined by WHO (1989) as the forever loss of any indication of survival at any period of time after birth. According to The Dictionary of Medical (1999), “mortality is the incidence of death in the population in a given period”. So the concept of mortality is often used in demographic researches and surveys.

 In this paper, the phenomenon of mortality is referred to under-five children. There are 4 sorts of mortality in the age of five (Pandey, Choe, Luther, Sahu, and Chand, 1998):

Neonatal mortality is the mortality of children at 0 – 30 days from birth.

Postneonatal mortality is the mortality of children at 1-12 months from birth.

Infant mortality is the mortality of children at 0-12 months from birth.

Under-five mortality is the mortality of children at 0-5 years from birth.

Measures of Child Mortality.

Under-five mortality rate    U5MR = x 1,000

where Mc is the total number of deaths amongst children under five years of age and Pc is the total number of live births within five years dating backward from the survey.

Infant mortality rate   IMR =  x 1,000

where Do is the total number of deaths amongst children in the infanthood and Bo is the total number of live births in the year of the survey.

Causes of Child Mortality. According to WHO (1999), the leading cause of child mortality is a bad prenatal condition and it account for 20% of all children who died before age five. In addition, respiratory infections, diarrhoeal diseases and vaccine-preventable diseases are also important causes of child mortality and account for 18%, 17% and 15% respectively. Finally, malaria account for 7% and other causes account for 23% of child mortality in the world.

2.2. Theoretical Framework

Palloni Approach

This model consists of three levels of importance:

The primary level refers to policies and emphasizes socio-economic development and horizontal interventions as well (expenses on health care, malady prevention, treatment, maternal care services and family planning activities).

The second level is related to the characteristics of individuals and collectives. These factors have strong influence on the possibly of causing maladies and mortality. In this set of categories of factors, the author considers three smaller groups: the characteristics that are concerned with individuals, families, and community.

The third level deals with the close variables to children health, morbidity exposition of children, frailty and susceptibility of individuals as well as their ability to resist maladies.

Mosley and Chen Approach

This framework provides a model with clear distinction between socio-economic determinants and proximate determinants of child survival in developing countries.

Mosley and Chen (1984) found that the major determinants of child survival are the proximate variables. The proximate determinants model was originally developed to study factors affecting child mortality, and is based on the idea that all social and economic determinants of child mortality operate through a set of biological or proximate determinants to affect a child’s probability of survival. Thus, this model combines social, economic, medical and biological explanations of child mortality. Mosley and Chen (1984) group the proximate determinants into five categories, namely, maternal factors; environmental contamination; nutrient deficiency; injury; and personal illness control.

Applying the Model in Analysing Child Mortality in Vietnam

After analysing the advantages and limitations of Palloni’s model and Mosley and Chen’s model, the study will use Mosley and Chen’s model for two reasons. First, it is the most frequently referenced in subsequent papers dealing with infant or child mortality determinants. Mosley and Chen’s model has identified variables that influence directly on child mortality so it is easier to apply. Second, it tries to integrate research methods employed by social and medical scientists and combine social, economic, medical and biological factors, which are not conducted in Paloni approach. However, it will be modified to overcome the limitations of the model and more fully analyse child mortality in Vietnam. Therefore, children age/sex and birth order, region and residence will be added in this model.

2.3. Empirical Evidence

Child sex and age. In almost all studies on child mortality, sex and age of the child are considered as important determinants. It is highlighted in Huq and Cleland, 1990; Kabir and Chowdhury, 1992. D'Souza and others (1980), Chen and his colleagues (1981) showed that, in the neonatal period, male mortality exceeds female mortality. Male mortality is often higher than female mortality (United Nations Secretariat 1988; Pandey, Choe, Luther, Sahu, and Chand, 1998). Many researches in Vietnam also state that child’s sex and age affects mortality level.

Regions. Besides child’s sex and age, child mortality rate is also different among regions. It is found by many research such as Wagstaff and Nguyen (2001), VPC (1979; 1989; 1999), IDS (1994), DHS (1997).

Residence. In developing countries, living conditions are generally worse in rural areas than in urban areas, and health-care facilities are less readily available and tend to be of poorer quality. These differences usually result in higher infant and child mortality in rural areas than in urban areas (Hobcraft, McDonald, and Rutstein 1984; Mosley and Chen 1984; United Nations 1985, 1991, 1998; VPC 1979, 1989, 1999; IDS 1994; DHS 1997).

Income. Economic status is also an important factor affecting child mortality, especially in developing countries (Hobcraft, McDonald, and Rutstein 1984; Mosley and Chen 1984; United Nations 1985; 1991; 1998; Wagstaff and Nguyen 2001).

Mother’s education. Mother’s education is widely regarded as an important determinant of child mortality and health outcomes in developing countries (Wolpin, 1997; Govindasamy and Ramesh 1997; Hobcraft, McDonald, and Rutstein 1984; Mosley and Chen 1984; United Nations 1985; 1991; 1998). Wagstaff and Nguyen (2001) state that mother’s education significantly improves survival prospects. And literate mothers are likely to have more influence in deciding to take sick children for treatment.

Maternal age. The age of mother at childbirth is very important because it related to mother’s knowledge and practice of giving birth to a child (Hobcraft, McDonald, and Rutstein 1985; Palloni and Milman 1986; Retherford et al. 1989; United Nations 1994; Nguyen 1998). Many research cited the increasing trend of child mortality when children are born to very young or to older mothers (Cantrelle and Leridon, 1971; Wolfers and Scrimshaw, 1975; Clark, 1981; Vanzo,1984; Pebley and Stupp, 1987; Palloni et al., 1994).

Birth order. Birth order also affects child mortality in developing countries where family-planning program is implemented (Hobcraft, McDonald, and Rutstein 1985; Palloni and Milman 1986; Retherford et al. 1989; United Nations 1994; Puffer and Serrano 1975; Omran 1981; Hobcraft et al. 1983)

Birth interval. Many researches found that birth interval has a effect on child mortality (WHO 2000; Hobcraft, McDonald, and Rutstein 1985; Palloni and Milman 1986; Retherford et al. 1989; United Nations 1994; Cantrelle and Leridon 1971; Wolfers and Scrimshaw 1975; Clark 1981; Rutstein 1983; Vanzo 1984; Pebley and Stupp 1987).

Antenatal care. The importance of antenatal factors in contributing to infant and child mortality is well recognized. (Puffer and Serrano, 1975; Omran, 1981; Hobcraft et al., 1983; Andersen 1968; Fiedler 1981; Kroeger 1983). Prenatal health services improve the survival and quality of life for mothers and children (UNICEF 1989).

Breastfeeding. Children who are never breastfed in infanthood are at a much higher risk of becoming ill and of being malnourished. Because of the synergism between disease and mortality as mentioned in causes of child mortality, they are at a much higher risk of dying (Briend, Wojtyniak, and Rowland 1988; Cabigon 1997; Habicht, Vanzo, and Butz 1986; Huffman and Lamphere 1984; Jelliffe 19; Palloni and Tienda 1986; Yoon 1996).

Water. Water is an important factor of living environment. Thus it also affects children health and mortality (Woldemicael, 1999).

Injury. Injury is rarely included in analysing child mortality in previous researches. But the results often show the significant effect on level of child mortality, such as Riddle (1998).

Chapter 3: Child Mortality in Vietnam

3.1. The Vietnamese Context

Geography and Natural Resources. In the North of Vietnam, monsoon climate is the reason for respiratory and diarrhoeal diseases, especially for children. In the Southern of Vietnam, Mekong Delta region is the leading rice producer in the country. The Central of Vietnam has harsh climate and favorable for cultivation of tropical and cash crops. However, in such harsh climate areas, drought and flood often happen and affect child living condition. The high fertile land gives Vietnam favourable condition to develop agriculture, forestry and provide abundant food for children. Thus it contributes to reducing child mortality rate in Vietnam. However, 25 per cent of Vietnam’s population reside in highland and mountainous regions (UNDP, 1998). And 9.9 million of Vietnamese people belong to 53 ethnic minority groups contributed to the high level of child mortality.

      Economic Growth. In the centrally planned economy, the child mortality rate at high level because of the low level of living standard and lack of knowledge and information about health care and nutrition. However, there was not a big gap between the rich and the poor. There was almost no differential of child mortality between the rich and the poor. Since doimoi, the Vietnamese economy has moved to a more market-oriented one and achieved many great successes in economic reform. These successes have significantly lower the child mortality in Vietnam. Despite these successes, the economy of Vietnam faces many challenges. There is also a huge disparity in socio-economic development between urban and rural areas and among regions.

Education. Despite being a poor country with low per capita income, Vietnam has made great achievements in education. Education level and health care knowledge of parents are increased and leads to the improvement of children’s health. In addition, the higher parity in education between male and female has contributed much to reducing child mortality.

Health Care. Prior to doimoi, people could access what the health system would provide for free. So the gap of child mortality between rich and poor families was not a problem.  However, medical and health equipment was provided with limited quantity. That is why many diseases cannot be treated, especially for the maternals and children. Under doimoi, Viet Nam has made significant advances in ensuring children’s right to survival such. Basic health education and knowledge are improved. As a result of the decrease in fertility trend, parents can reserve money and time for educating and caring their children. Thus, children health care is also improved. However, these activities and programs are still not available enough in ethnic minority and disadvantageous areas.

3.2. Child Mortality in Vietnam

Trend in Child Mortality

All of data sources show that children’s health has been improved remarkably and child mortality level in Vietnam has been reducing. Child mortality in Vietnam experienced very high level in the past, 89 per thousand in 1960. But in 1990 and 2000 under-five mortality rate is only 50 per thousand and 42 per thousand respectively, much lower than 1960 (UNICEF, 2001). The annual average rate of reduction in U5MR from 1960 to 1965 was 2.0% up to 2.6% from 1995 to 1996 (UNICEF, 1999). This decreasing trend in child mortality rate resulted from many reasons of the socio-economic conditions as analysed in the above section. UNICEF (2001) ranked under-five mortality rate of all countries in the world from 1 for the highest rate to 187 for the lowest one. And under-five mortality rate in Vietnam is ranked 89^th. It is not very high in comparison with other countries in the region. But ranking 89^th means that child mortality in Vietnam is still at a high level in the world.  

Characteristics of Child Mortality

First of all, the study considers child mortality rate in 1998 VLSS in comparision with 1993 VLSS. Infant mortality rate and under-five mortality rate reduce remarkably from 1993 to 1998.  Infant mortality rate and under-five mortality rate in 1998 was 36.97%o and 45.05%o respectively, while these numbers in 1993 was 38.52%o and 48.27%o. But the reducing level of under-five mortality is likely to larger than infant mortality.

Second, it considers sex differences in comparison with 1993 VLSS. It indicates that the probability that a boy died before age five is higher compare to girls but the trend in poorer health for boys is reducing.

Thirdly is considering the changes of infant and under-five mortality between rural and urban. It is noted that child mortality rate, especially the infant mortality rate, in rural areas is much higher than urban areas.

Differential of Child Mortality

In going further to understand child mortality in Vietnam, this section concentrates on examining the differential of child mortality such as: differential of child mortality by regions, residence, income, mother’s literacy, mother’s age at child birth, birth order and birth interval.

This section found that there are differentials of child mortality between those factors. For example, the highest mortality levels are observed in the Central Highlands where the probability for a child dying before age five in 1998 is 71.43%o, however, the Red River Delta and the South East have the lowest mortality rate, only 32.68 %o and 27.55 %o respectively.

It also compares with VLSS 1993 to find out the trend of these differentials and the reason for this trend. For example, although child mortality reduced from 1993 to 1998, there was not a remarkable fall in disadvantageous regions such as the Central Highlands, South Central Coast and Northern Upland. It is due to the health care programs are not efficient enough in those regions. Thus poor living and health care conditions are the reasons for the lower mortality rate in those regions compared to other regions.

After combining above analysis, the fact should be noted that child’s sex, child’s age, mother’s literacy, income, regions, residence, mother’s age at childbirth, birth order and birth interval are very important to determine child mortality in Vietnam.

Chapter 4: Determinants of Child Mortality in Vietnam

4.1. Methodological Approach

Methodology

The maximum likelihood estimation of the dichotomous dependent variable died or not is performed using the probit analysis model. We are interested in the factors affect the probability P_i that the child i being died. The model is formulated with the dependent variable, yi as a probability function for yi:

F(yi ) = P_i ^yi (1-P_i )^{1- yi}, where yi = 0,1

The dependent variable yi depends on unobservable index or the probit score yi *, is presented by:

yi * = ’x_i + ui

yi = 1 if yi * > 0

yi = 0 if yi *   0

where the dependent variable in the equation, yi, is observed only if ’x_i > 0; yi is indicator for whether or not the parents have lost a child,  yi  is a binary selection indicator variable coded “1” for those cases observed and “0” otherwise, x_i  is a vector of covariates, and ’ is vector of parameters. Here, ui is random error terms, it is assumed that ui ~ N(0,²_u).

The likelihood function is:

L = ’x_i)]^yi [1 - ( ’x_i)]^{1- yi}

where is the distribution function of the standard normal.

Data Set

The research makes extensive use of the Vietnam Living Standard Surveys 1997/1998. The survey is nationally representative and covered 5,999 households. All observations in the sample were constructed by a multi-stage sampling procedure. The sample includes 10 strata and 194 communes. And each commune contains about 9.5 children. Individual weights were calculated in the datasets to preserve the representative

From the fertility history section of VLSS 97/98, the analysis obtained 2444 observations of under-five children. However, only 1,749 observations are included in the empirical analysis since 695 observations were dropped when merging data and calculating variables from different sections of VLSS 97/98.

Model Specification

The first model is performed to analyse the determinants of child mortality with all under-five children. Thus all 1,749 observations are included in this model.

The second and the third models are performed to examine the differences in determinants of first-born and later-born children. Thus, only under-five first-born children are selected to analyse in the second model and it includes 503 observations. In the third model, later-born children (from birth order 2 to 12) are selected to analyse, thus 1246 observations are included in the third model. From these models, comparison of determinants between first-born and later-born children will be made to find out reasons for the differences.

4.2. Results of Estimation

Table 4.2 presents probit result for the probability of dying for under-five children in Vietnam 1997-98. Instead of reporting the parameter estimates, which are difficult to interpret the results, the marginal effects will indicate the range of variation of the changes in probability of dying responding to the changes in explanatory variable

It is noted from Table 4.2 that breastfeeding has the most important effect on child mortality. As presented in the result, breastfeeding variable has negative impact on child mortality and significant at 5 percent level. Because of its important nutrition for children at the childhood, nowadays, breastfeeding for children is encouraged.

The effect of mother’s age at childbirth on child mortality is considered as a very important determinant. It reflects truly the role of mothers in child survival. This result draws out the U-shaped relationship between mother’s age at childbirth and mortality rate. It can be explained that if they give birth at such young ages (less than or equal 20), they will face high risk because of lacking basic knowledge and experience in prenatal and delivery care. And if they have children at over 30, they have to face the risk of biological factors although they have more experience than younger mothers. Birth order is of great importance for a child survival and could draw the U- shaped relationship between birth order of the child and child mortality also.

Table 4.2 also tells us the important role of residence and regions variables. If the child lives in rural area, his probability of dying is higher than the child in urban area and  child mortality is higher in the Northern Uplands than in the Red River Delta, North Central Coast, South East and Mekong Delta. Mother’s literacy is also important for child mortality because it is considered as a key to improving the health, nutrition and education of children.

As shown in Table 4.2, income variable seems quite important in determining child mortality in Vietnam. It can be seen that child’s sex and child’s age are less important variables compared with the above ones. It is shown in the model result since a child’s sex has a positive relationship with mortality. In addition, child’s age has negative relationship with mortality at 5 percent significant level. It means that infants face a higher probability of dying than other ages. And antenatal visits of mothers have the smallest relationship with child mortality.

Although water sources variable is positively associated with child mortality, they do not appear to statistically significant at 10 percent level. Injury of women in pregnancy seems not statistically significant at 10 percent level. It indicates that, in general, injury does not affect child mortality.

Table 4.1 - Variable definitions

Table 4.2 - Determinants of the probability of dying for under-five children (Probit)

Number of obs 1749 

Prob> chi2 0.0000

Pseudo R²  0.2067

(*)     dF/dx is for discrete change of dummy variable from 0 to 1;

    Source: Author’s calculations based on data of Vietnam Living Standards Survey 1997-1998

4.3. Difference in Determinants of First-Born and Later-Born Children

In this section, we go further to understand the difference in determinants of first-born and later-born children on child mortality in multivariate way.

Firstly, injury has a strong relationship with mortality of first-born children while this variable does not affect later-born children. It can be explained that mothers of the first-born are often lack of practice and knowledge in improving health after injury happened. And in fact, injury is likely to occur with young mothers of the first-born. Mothers of later-born children get more fertility experience to protect their child from mortality risk.

Secondly, besides many decisive factors of mortality of first-born children, later- born children are affected by birth interval. Birth interval is negatively associated with child mortality at 5 percent significant level. It means that if the birth spacing between two children is over 24 months, the probability of dying for the child is 2.3 percentage point lower than in the case of less than 24 months.

Finally, the results also indicate that the important level of determinants is also different between first-born and later-born children. For example, with the first child of mothers, breastfeeding and injury are most important determinants and mother’s age seems not very important. But with later-born children, breastfeeding and mother’s age are most important determinants while injury seems not to affect child mortality. This difference is due to the fact that, in Vietnam, mothers often give birth the first time in the age of 20 - 30 years. These ages were found not to face high risk of mortality. Thus effect of mother’s age on child mortality does not as strong as breastfeeding and injury. But with later births, if women give birth at over 30 she will face higher risk not only for her but also for her children. In addition, looking at antenatal visits variable, we can see that it is quite important with the first birth of mothers but it has the smallest importance with later births. It can be explained that health care services are very important with mothers at the first birth since they are lack of fertility knowledge and experience in the period of pregnancy and giving birth.

Table 4.3 - Determinants of the probability of dying for first-born children (Probit)

            Number of obs      503

            Prob> chi2       0.0000

        Pseudo R²        0.3591

(*)    dF/dx is for discrete change of dummy variable from 0 to 1

    Source: Author’s calculations based on data of Vietnam Living Standards Survey 1997-1998

Table 4.4 - Determinants of the probability of dying for later-born children (Probit)

Number of obs    1246

Prob> chi2       0.0000

Pseudo R² 0.2672