Problem statement
Global maternal health care is increasingly concerning, particularly in low-income countries The World Health Organization (WHO) reported a decrease in the global maternal mortality ratio (MMR) from 380 deaths per 100,000 live births in 1990 to 210 in 2013 However, the MMR in developing regions remains 14 times higher than in developed areas Despite the overall decline in maternal deaths worldwide, the target set by the Millennium Development Goal 5—to reduce the MMR by three-quarters between 1990 and 2015—has not yet been met.
Maternal deaths can be attributed to both direct and indirect causes Direct causes arise from complications during pregnancy, delivery, and postpartum periods, including issues like hemorrhage, infection, obstructed labor, unsafe abortion, ectopic pregnancy, and anesthesia-related fatalities In contrast, indirect causes stem from pre-existing health conditions not directly related to obstetrics, such as hepatitis, anemia, malaria, heart disease, and tetanus According to the World Health Organization (2005), direct causes account for approximately 80% of total maternal mortality rates, highlighting their significant impact on maternal health.
Antenatal care and delivery care, introduced by the WHO in the safe motherhood package in 1994, play a crucial role in preventing complications during pregnancy These services empower pregnant women and their families with essential information about maternal health and fetal development By monitoring the weight and growth of the unborn baby, antenatal care can help prevent low birth weights through improved maternal nutrition Additionally, regular check-ups identify potential risks and danger signs, allowing for timely interventions, such as tetanus immunization, which is vital for the health of both mother and child Effective management of conditions like high blood pressure during pregnancy is essential for ensuring maternal health and enhancing infant survival rates.
Delivery care is crucial for reducing maternal deaths, as recommended by the WHO, which advocates for childbirth to occur in health facilities attended by skilled healthcare professionals This approach ensures safe deliveries and the birth of healthy babies, as proper hygiene and adequate medical equipment can significantly reduce complications such as hemorrhage and obstructed labor Additionally, the presence of skilled health professionals in these facilities guarantees safe delivery practices and effective emergency management when necessary.
Vietnam is making significant strides in improving maternal health, aligning with Millennium Development Goal 5, as evidenced by a decline in the maternal mortality ratio (MMR) According to the World Bank, the MMR in Vietnam decreased from 81 deaths per 100,000 live births in 2000 to 54 per 100,000 in 2015 Access to antenatal care, crucial for the health of pregnant women and their babies, has also risen, with the Multiple Indicator Cluster Survey (MICS 5) in 2014 revealing that 95.8% of women aged 15-49 who had a live birth in the past two years received antenatal care at least once However, disparities in maternal mortality and healthcare utilization persist among different ethnic groups, residential areas, and regions.
Maternal mortality rates in mountainous regions are over three times higher than in lowland areas, as reported in 2015 The MICS5 data highlights significant disparities in prenatal care visits between women in rural and urban settings, particularly regarding those who receive more than four visits Ethnic minority groups face even greater challenges, with only 1% having one visit and 32.7% achieving at least four visits, compared to 99.2% and 82.1% of the Kinh group These growing disparities in health outcomes and healthcare access have created substantial challenges in recent years.
Challenges in maternal health care in Vietnam have prompted numerous studies, primarily examining the impact of demographic and socioeconomic factors on health outcomes (Sepheri et al 2008, Tran et al 2011, Goland et al 2012, Malqvist et al 2012).
Demographic factors influencing the utilization of health services include younger age and lower birth order, while separated or unmarried status and unintended pregnancies are associated with decreased usage Additionally, socio-economic factors significantly impact access to maternal health care, with a woman's education level emerging as the most crucial determinant according to various studies.
Lower household income significantly impacts the likelihood of utilizing maternal health care services (Sepheri et al 2008, Goland et al 2012) Research highlights disparities in maternal health care access between ethnic majority and minority groups (Malqvist et al 2012, Malqvist et al 2013) Additionally, Tran et al (2011) and Sepheri et al (2008) identified major inequities in maternal health care access between rural and urban areas, as well as regional disparities in Vietnam regarding the availability and accessibility of these services.
Many studies have neglected to consider community factors, with the exception of Sepheri et al (2008), who evaluated the impact of poverty rates This oversight can lead to biased conclusions regarding maternity healthcare utilization, particularly for disadvantaged women Furthermore, failing to account for community effects may distort the estimated influence of various factors, as highlighted by Singh et al (2014).
Community beliefs and norms significantly impact women's health care-seeking behaviors Economic development within a community can enhance access to health services, empowering women and fostering positive attitudes towards health service utilization (Stephenson et al 2006) Key community-level indicators influencing these behaviors include the poverty rate, the percentage of women with higher education, and the rate of women delivering in health facilities Notably, a higher poverty rate is linked to a lower likelihood of women seeking antenatal care (Gage & Calixte).
2006, Sepehri et al 2008, Ononokpono et al 2013, Singh et al 2014,) and facility delivery
Research indicates that a higher percentage of women with advanced education and those opting for facility-based deliveries are positively linked to increased utilization of maternal health care services.
Ononokpono et al 2013, Singh et al 2014)
This study emphasizes the importance of investigating the determinants of healthcare service utilization at individual, household, and community levels Utilizing the 2014 Vietnam Multiple Indicator Cluster Survey (MICS 5), it employs the Poisson Model to assess the influence of social determinants on prenatal care visits and the Multinomial Logistic Model to analyze the relationship between social factors and the selection of delivery care providers.
Research objectives
This thesis research aims to analyze the demand for prenatal health care by examining the factors influencing the number of antenatal care visits among women over the past two years, utilizing data from MICS 5 Additionally, it explores the choices pregnant women in Vietnam make regarding service facilities for delivery.
Research questions
To investigate the above objectives, the following questions need to be answered thoroughly:
Question 1: What are the determinants of the demand for antenatal care visits?
Question 2: What are the determinants of the choice on delivery care provider?
Structure
The paper is structured to first explore general theories on health care demand and provider choice in Chapter 2, alongside a review of previous studies on social determinants affecting maternal health care utilization It also outlines the conceptual framework and methodology, employing the Multiple Indicator Cluster Survey 2013-2014 (MICS5) dataset to analyze determinants of prenatal health care demand and delivery service facility selection Results and discussions are detailed in Chapter 4, while Chapter 5 concludes the paper with insights on policy implications.
The role of maternity health care
Motherhood is a rewarding journey for women, yet it comes with potential health challenges during pregnancy, childbirth, and the postpartum phase These issues can significantly affect both maternal and infant health, with three-quarters of maternal deaths occurring during childbirth and the postpartum period Fortunately, proper antenatal and delivery care can help prevent these complications.
Antenatal care (ANC) has been established since the early 1900s to support the health of pregnant women and their unborn children, allowing for early detection of potential complications This care enhances women's understanding of fetal development and their own health, helping to prevent adverse outcomes such as low birth weight through improved nutrition Women receive crucial information about pregnancy and delivery risks, with the World Health Organization (WHO) recommending at least one visit to a skilled health provider or a minimum of four ANC visits WHO guidelines for ANC include assessments of both mother and fetus, such as body weight, height, blood pressure, and necessary blood and urine tests, along with medical provisions like tetanus vaccinations and iron and folate supplements, as well as health education and counseling.
The delivery process at health facilities is crucial for ensuring safe childbirth and healthy infants, as proper medical technology and hygienic conditions can significantly reduce complications and infections, thereby lowering maternal and child morbidity and mortality rates Skilled birth attendants, including midwives, doctors, and nurses, are essential in this process, as defined by WHO, as they are trained to facilitate normal childbirth, manage postnatal care, detect complications, and provide emergency interventions WHO recommends that countries with high maternal mortality rates ensure that at least 60% of deliveries are assisted by skilled birth attendants, a target that was met by 69% of women giving birth between 2005 and 2010 (Tran, 2012).
Overview of maternal health and health care in Vietnam
Vietnamese culture, particularly in the northern regions, is deeply influenced by Confucianism, which emphasizes the importance of sons inheriting family resources and honoring ancestors This tradition places a significant responsibility on sons to care for family members and ensure the continuity of the family lineage, leading to a societal pride associated with giving birth to a son Consequently, women in families with daughters often experience immense pressure to produce male heirs, contributing to a pronounced preference for sons that has resulted in an increasing sex ratio at birth.
In traditional family structures, male members are often viewed as the primary income earners and decision-makers, while females are seen as vulnerable, with their lives largely dictated by their parents Upon marriage, women typically move in with their husband's family, where their income and autonomy may be further restricted by in-laws and husbands The influence of Confucian values and existing hierarchies significantly limits women's independence and decision-making, particularly regarding their health For instance, the childbirth experiences of mothers and mothers-in-law can heavily impact young women's maternity care choices, potentially discouraging them from seeking essential maternal health services.
Vietnam's two-child policy, introduced in the late 1980s, limited the number of children per household and included family planning measures such as free birth control devices and abortion facilities Families that did not comply faced penalties, including fines and job repercussions for government employees, leading some women to hide pregnancies and neglect maternal healthcare The 2009 Population Ordinance, which allows couples to choose the timing and spacing of children while still limiting them to one or two, has contributed to a decline in the total fertility rate from 2.55 in 2001 to 1.99 in 2011, indicating the policy's effectiveness in managing population growth However, challenges remain, including ineffective contraceptive methods, with IUDs being popular despite their side effects, and a high abortion rate among youth due to a lack of contraceptive knowledge and socioeconomic factors.
2.2.3 Maternal mortality ratio and maternal health care in Vietnam
In Vietnam, the government advises that pregnant women attend a minimum of three prenatal visits to identify and mitigate health risks for both mother and baby Essential prenatal care includes monitoring blood pressure, urine and blood testing, and measuring weight and height Additionally, national guidelines recommend delivering babies at health facilities to ensure proper medical care and hygienic conditions, which help reduce complications during and after childbirth In cases where complications arise, a Caesarean section should be conducted by skilled obstetricians to guarantee a safe delivery Furthermore, during the postpartum period, at least two health checkups are recommended for both the mother and child.
The maternal mortality ratio (MMR) measures the number of women who die from pregnancy and childbirth-related causes within 42 days post-delivery, per 100,000 live births According to World Bank data, Vietnam has made significant progress in reducing its MMR, decreasing from 81 per 100,000 in 2000 to 54 per 100,000 in 2015, successfully achieving the Millennium Development Goal 5 target of 58.3 per 100,000 live births Despite this improvement, Vietnam still lags behind developed Asian nations like Singapore, Malaysia, and Thailand To ensure sustained population growth, Vietnam must intensify efforts to further reduce its maternal mortality rate.
Figure 1: MMR in Vietnam in the period of 2000 – 2015 Source: The World Bank
Figure 2: MMR of the Asian countries in the period of 2000 – 2015
Maternal mortality ratio (per 100,000 live births)
Lao PDR Malaysia Indonesia Thailand Cambodia Brunei Darussalam Singapore
To reduce maternal mortality rates (MMR) and infant mortality rates (IMR), the World Health Organization (WHO) recommends that each pregnant woman has at least four prenatal care visits or at least one visit with professional health staff Antenatal care is essential for informing women and their families about potential risks during pregnancy and childbirth Vietnam has made significant strides in antenatal care coverage, with 95.8% of pregnant women having at least one prenatal visit in 2014, up from 1997 However, only 73.7% received more than four visits, highlighting a challenge that requires further measures Additionally, there are notable disparities in maternal healthcare utilization among different ethnic groups, residential areas, and regions Women in rural areas have fewer prenatal visits compared to those in urban settings, and ethnic minorities face greater barriers, with only 79% having one visit and 32.7% having at least four, compared to 99.2% and 82.1% of the Kinh group, respectively.
Figure 3: Percentage of women having at least 1 visit and at least 4 visits during pregnancy
Source: Ministry of Planning and Investment, MICS4
At least 4 times by any providers At least 1 visit by skilled health worker
Figure 4: The percentage of the women taking antenatal care visits by residence in 2011 and 2014
Figure 5: The percentage of the women taking antenatal care visits by ethnicity in 2011 and 2014
Antenatal care visits by residence
Antenatal care visits by ethinicity
The demand for health care
Economists became interested in the health seeking behavior in the late 1960s and investigating the factors influencing the behavior The major contributions were by Grossman
(1972), Rosenstock (1974), Thaddeus and Maine (1994) and Andersen (1995)
Grossman (1972) posited that individuals seek good health rather than health care itself He developed a demand model for health, illustrating that individuals derive positive utility from consumption goods while experiencing negative utility from time spent sick (𝑡 𝑠 (𝐻)) This foundational concept highlights the distinction between health care services and the ultimate goal of achieving good health (Zweifel et al., 2009).
Health stock H evolves over time, resulting in a depreciation of health capital at a rate of 𝛿 Nevertheless, individuals have the opportunity to enhance their health capital through investments in I In a two-period model, current health can be defined accordingly.
In Grossman's 1972 model, he posits that individual demand for health is driven by two key factors: first, health is viewed as an investment commodity, contributing to future productivity, and second, it serves as a consumption commodity, providing immediate satisfaction The model incorporates various elements such as utility (U), health capital (H), the depreciation rate (𝛿), wage rate (w), consumption goods (X), medical services (M), health investment (I), sick time (𝑡 𝑠), and time dedicated to health (𝑡 𝐼), highlighting the multifaceted nature of health in economic terms.
Basing on the function 𝐼(𝑀, 𝑡 𝐼 ) and 𝑡 𝑠 (𝐻 1 ) from (1.1) and (1.2), Grossman constructed the demand function for medical services in investment model: ln 𝑀 = 𝑐𝑜𝑛𝑠𝑡 − (1 + 𝛼 𝑀 (𝜀 − 1))𝑙𝑛𝑝 + (1 + 𝛼𝑀(𝜀 − 1))𝑙𝑛𝑤 − (1 − 𝜀)𝛼𝐸𝐸 (1.3)
Where, 𝛼 𝑀 is the production elasticity of medical services and 𝛼 𝐸 is the effectiveness of education E; 𝜀 is the marginal efficiency of health capital 𝐻 1
From the utility function (1.1) including sick time and consumption good, he constructed the demand function for medical services in consumption model: ln 𝑀 = 𝑐𝑜𝑛𝑠𝑡 − (1 + 𝛼 𝑀 (𝜅 − 1))𝑙𝑛𝑝 + (1 − 𝜅)(1 − 𝛼 𝑀 )𝑙𝑛𝑤 − (1 − 𝜅)𝛼 𝐸 𝐸 − 𝜅𝑙𝑛𝜆
The demand for health services is influenced by several factors, including the price of medical services, wage rates, education, and wealth Specifically, a decrease in the price of medical services leads to a reduction in the optimal quantity of health services (𝐻 1), while an increase in wage rates results in a higher quantity of health services demanded Additionally, in a multiple period model, age is incorporated into the demand functions, as the depreciation rate (𝛿) is positively correlated with age (Zweifel et al 2009).
The Health Belief Model, introduced by Rosenstock in 1974, explains individual health-seeking behavior by emphasizing the role of personal beliefs about health issues, perceived benefits and barriers, and cues to action This model suggests that demographic and psychosocial characteristics indirectly influence perceptions, with the perception of serious health risks significantly increasing the likelihood of engaging in health-promoting behaviors Individuals are more likely to take action when they perceive that the benefits of doing so outweigh the barriers, such as inconvenience or side effects Cues to action, which can be internal (like pain or symptoms) or external (such as information from friends or mass media), also play a crucial role in prompting health-related behaviors The model has been effectively used to design interventions aimed at changing health behaviors by addressing these key components.
Thaddeus and Maine (1994) introduced the three delay theory to highlight barriers to timely maternal health care utilization The first phase involves delays in seeking care, influenced by individual and family decisions, women's status, prior health care experiences, and financial or opportunity costs The second phase addresses delays in accessing health facilities, which are affected by facility availability, distance, transportation costs, and infrastructure Finally, the last phase pertains to delays in receiving adequate care, primarily due to insufficient equipment and a lack of qualified health personnel.
The Andersen Behavioral Model of Health Services Utilization (1995) identifies three key factors influencing individuals' access to and use of healthcare services: predisposing, enabling, and need factors Predisposing factors encompass demographic characteristics such as age, gender, and marital status, as well as social structures like education, occupation, and ethnicity, which affect an individual's likelihood of seeking care Health beliefs also play a crucial role, shaping perceptions and attitudes towards healthcare systems Enabling factors, both personal—such as income, health insurance, and travel times—and organizational, including the availability of health facilities, determine the actual ability to obtain services Lastly, need factors, which are the direct causes for seeking healthcare, involve self-assessment of health status and evaluations by healthcare professionals.
Prenatal care plays a crucial role in reducing maternal and infant mortality by providing essential measurements and counseling that help women understand their health and that of their babies This proactive approach aids in identifying risks, ensuring safer pregnancies and childbirth, and minimizing postpartum complications While research has examined the link between maternal health care and health outcomes, other studies have focused on the factors influencing the utilization of maternal health services This article highlights the key determinants affecting the demand for prenatal care visits and the selection of delivery locations based on previous studies.
Determinants of health and well-being are classified into three key categories: individual, household, and community levels Individual factors encompass education level, maternal age, marital status, religion, and ethnicity Household characteristics include household size and wealth At the community level, determinants consist of place of residence, regional variations, poverty rates, and illiteracy rates.
Individual level characteristics Mother’s education
Research consistently emphasizes the crucial role of maternal education in accessing maternal health care, indicating that women with higher education levels are more likely to utilize adequate antenatal care (ANC) (Arthur 2012, Bbaale 2011, Navaneetham & Dharmalingam 2002) Educated women tend to have greater decision-making power regarding health issues and are more inclined to seek better healthcare options (Navaneetham & Dharmalingam 2002) However, there is no significant difference in maternal health care utilization among educated women when comparing primary and secondary education levels (Navaneetham & Dharmalingam 2002) Furthermore, a study by Chen et al (2003) on Taiwan's National Health Insurance (NHI) revealed that while educational attainment had little impact on ANC utilization prior to NHI, it significantly influenced usage post-introduction This discrepancy remains unclear, but generally, educated women are more aware of the importance of maternal care, leading to increased frequency of ANC visits compared to their less-educated counterparts.
Marital status also is a key determinant of the use of maternal health care Sepehri et al
Research indicates that marital status significantly influences prenatal care utilization, with married women more likely to access maternal care compared to single mothers, who often face stigmatization in Vietnam (Sepehri et al., 2008) This stigma arises from societal beliefs that childbirth is a shared responsibility between the mother and her husband Supporting these findings, a study in Taiwan by Chen et al (2003) revealed that married women benefit from spousal support, leading to increased access to maternal care visits compared to their unmarried counterparts.
Maternal age significantly influences the utilization of health services during pregnancy As mothers age, their likelihood of seeking healthcare tends to decrease, highlighting a critical aspect of maternal health access.
Research indicates that experience and knowledge regarding maternal health significantly influence women's health-seeking behaviors (Chen et al., 2003) A study by Tsawe & Susuman (2014) found that women aged 15-39 are more likely to attend regular health check-ups compared to those over 40 Conversely, some studies, such as one conducted in Turkey, suggest that age may not have a significant impact on attendance at antenatal care (ANC) services.
Research indicates that as the number of children born to a mother increases, the likelihood of utilizing maternal health care services decreases, primarily due to time and resource constraints faced by women with larger families (Navaneetham & Dharmalingam, 2002) First-time mothers are more inclined to seek antenatal care, often due to their inexperience However, previous negative experiences with maternal health services can deter women from seeking care in subsequent pregnancies (Arthur, 2012) This notion is supported by Tsawe & Susuman (2014), who emphasize that positive experiences with health care encourage more frequent use of services Additionally, policies such as the two-child limit and associated penalties can further reduce the utilization of maternal health care among families with more than two children (Sepehri et al., 2008).
A study by Wado et al (2013) in Southwestern Ethiopia revealed a significant correlation between pregnancy intention and the use of antenatal care, while its connection to delivery care remained unclear The researchers suggested that women experiencing unwanted pregnancies may not adequately prepare emotionally or financially for childbirth, leading to less attention to their health and that of their unborn child One contributing factor is that these women often recognize their pregnancies later, resulting in missed early antenatal care visits Specifically, Wado et al found that women with unintended pregnancies detected their condition approximately one month later than those with intended pregnancies Overall, the study highlighted the strong association between pregnancy intention and maternal care utilization, although the relationship with delivery care warrants further investigation.
The choice of health care provider
In the late 1980s, Gertler, Locay, and Sanderson introduced the concept of health-seeking behavior in choosing healthcare providers They proposed a two-stage decision-making process where individuals first determine whether to seek healthcare and then select providers that offer the greatest utility The utility function for an individual receiving care from a specific provider is a key element of this framework.
In which, 𝑈 𝑖𝑗 is the utility of the individual i after receiving health care from provider j, ℎ 𝑖𝑗 is expected health status of the individual after receiving health care from provider j and
𝐶 𝑗 is other consumption expenditure after paying provider j
The health status after receiving health care from provider j for an individual i depends on the quality of provider j’s health care
Where ℎ 0 is the health status before receiving health care from provider j
The quality of health care varies significantly among different providers and individuals This variation is influenced by the specific characteristics of health care providers (𝑍𝑖) and the individual patients (𝑋𝑖).
After receiving healthcare, the remaining consumption expenditure \(C_j\) reflects the leftover income \(Y_i\) after settling payments to healthcare provider \(j\) The price \(P_{ij}\) for alternative \(j\) encompasses both direct costs, like consultation and medication expenses, as well as indirect costs, including transportation fees and waiting time.
Assuming that the individual has j alternatives and would like to maximize the utility so the utility maximization is expressed as:
Where 𝑈 ∗ is the maximum utility and 𝑈 1 , … , 𝑈 𝑗 is the individual utility with alternative of health care provider 1, …., j
Researchers have found it challenging to measure individual utility directly; instead, they focus on analyzing the characteristics of both the alternatives available and the individual (Train, 2009) Consequently, the utility function for an individual can be articulated as follows:
In which, 𝑉 𝑖𝑗 is observed characteristics and 𝜀 𝑖𝑗 is unobserved characteristics
The individual characteristics that can be measured include gender, age, education, income, and insurance status, while unobserved traits encompass perceptions of healthcare quality and preferred medical administration For healthcare providers, observable factors consist of pricing and the distance from the patient's home, whereas unobserved elements include the provider's reputation and prestige.
Childbirth in health facilities significantly reduces maternal mortality and morbidity rates, yet many women, particularly in developing countries, still opt for home births Research has explored the factors influencing this choice, examining the individual characteristics of women as well as the household and community contexts in which they reside.
Prenatal care visits play a crucial role in influencing healthcare decisions during pregnancy Research by Stephenson et al (2006) indicates that prenatal care significantly impacts the choice of facility-based delivery by educating expectant mothers about the advantages of institutional care Similarly, Sepehri et al (2008) emphasize that timely and sufficient prenatal visits enhance awareness of the necessity for proper delivery care.
Educational attainment significantly influences the choice of childbirth location, with studies indicating that women with higher education levels are more inclined to opt for facility-based deliveries over traditional home births This trend suggests that educated women tend to make more informed and independent decisions regarding their healthcare options Additionally, they are better equipped with knowledge about the advantages of facility-based deliveries, which further encourages their preference for such services.
With respect to the birth order, the previous studies reported that it has strong association with the choice of delivery at health facilities Navaneetham & Dharmalingam
Research from 2002 indicates that women having their first child are more likely to deliver in healthcare institutions compared to those having subsequent children This disparity can be attributed to various factors, including the time and resource constraints faced by women with larger families, which hinder their access to healthcare services Additionally, negative experiences from previous childbirths may lead these women to underestimate the need for facility-based deliveries.
Research on the impact of maternal age on delivery choices presents mixed findings According to Celik & Hotchkiss (2000), there is no significant correlation between a woman's age at her last childbirth and her preferred delivery location.
On contrary, Stephenson et al (2006) argued that the age of the interviewed women had significant association with the choice of facility delivery in the study of six Africa countries
Research indicates that women aged 40-49 and 30-39 are more inclined to give birth in health facilities compared to those aged 20-29 Nevertheless, this trend is not significantly observed in certain countries, including Burkina Faso, Ivory Coast, and Ghana.
Marital status influences the choice of birth delivery location, with mixed outcomes observed According to Stephenson et al (2006), women in polygamous marriages or those who are separated are less likely to deliver their last child in health institutions across all six studied areas.
Africa countries However, Sepehri et al (2008) found that marital status had no significant impact in the choice of delivery location
The choice of delivery location is significantly affected by household characteristics, including wealth, ethnicity, and religion Research by Stephenson et al (2006) indicates that women from higher-income households are more inclined to opt for institutional delivery compared to those from lower-income backgrounds This trend is largely due to the financial burden of childbirth and transportation costs that limit access to health services for poorer women Additionally, previous studies have highlighted the substantial impact of ethnicity on the preference for facility-based deliveries.
Hotchkiss (2000) highlighted that both urban and rural women face similar challenges in accessing healthcare, which may stem from cultural and economic barriers Additionally, religion plays a significant role in women's childbirth decisions A study by Stephenson et al (2006) across six African countries revealed that in Ghana, Muslim women are less inclined to give birth in healthcare facilities compared to their Catholic counterparts, while Protestant women are more likely to utilize these services than Catholic women.
There are also remarkable regional differences in the choice of delivery location Celik
Research by Hotchkiss (2000) indicates that urban women are more inclined to opt for facility delivery compared to their rural counterparts Additionally, Celik & Hotchkiss (2000) highlight that women in Eastern Turkey face more challenges in accessing facility delivery due to the region's disadvantaged status relative to the Western and Northern regions Gage (2007) further emphasizes that these regional disparities reflect the unequal accessibility and availability of healthcare facilities.
Conceptual framework
Prenatal care visits and the choice of delivery care providers are influenced by various factors at the individual, household, and community levels Individual characteristics such as education level, access to mass media, employment status, marital status, pregnancy intention, and birth order play a significant role Household factors, including wealth index, size, ethnicity, and the religion of the household head, also impact maternal health decisions Additionally, community-level characteristics, such as residence location, poverty and illiteracy rates, and the prevalence of women giving birth in health facilities, further shape maternal health-seeking behavior This comprehensive framework highlights the interconnectedness of these factors in influencing prenatal care.
Figure 6The association between individual level, household level and community level characteristics with the utilization of maternal health care services
Empirical framework
This study aims to evaluate the impact of social determinants on maternal health care services, specifically focusing on the frequency of prenatal care visits and the selection of delivery facilities To analyze the number of prenatal check-ups, the researchers employed the Negative Binomial Model, while the Multinomial Logit Model was utilized to assess the decision-making process regarding delivery facilities.
The Poisson Model is utilized for dependent variables that are non-negative integers, employing the Poisson distribution to quantify the probability of an event occurring k times within a specified timeframe This distribution function effectively captures the likelihood of such events.
With the condition that 𝜆 is non-negative and mean equal to variance 𝐸(𝑌) = 𝑣𝑎𝑟 (𝑌) 𝜆.The meaning is that when X changes, how the expected value of y changes
The Poisson Model has a key limitation where the variance is equal to the mean, which can be unrealistic for certain count data To address this issue, Negative Binomial regression serves as a viable alternative, allowing for variance to differ from the mean While Negative Binomial regression shares the same mean structure as Poisson regression, it includes an additional parameter to account for over-dispersion In Stata, the command nbreg is utilized to estimate Negative Binomial regression and to conduct tests for over-dispersion.
The study examines the number of maternal check-ups taken by pregnant women as the dependent variable, while independent variables are categorized into individual, household, and community level characteristics Notably, healthcare prices and income are excluded from the analysis due to data limitations, with further details provided in the following section.
3.2.2 Choice of birth delivery facility
The Multinomial Logit Model is utilized to assess the relationship between categorical variables and various explanatory factors, with the assumption that mothers select the facility that maximizes their utility Each individual's choice of facility, denoted as \(Y_i\), can be represented by the alternatives 1, 2, 3, …, J The probabilities associated with each choice are represented as \(p_{i1}, p_{i2}, …, p_{iJ}\) The logit function is defined accordingly.
Log-likelihood function: log 𝐿 = ∑ 𝑁 𝑖=1 ∑ 𝐽 𝑖=1 𝑦 𝑖𝑗 ln 𝑝 𝑖𝑗 where 𝑦 𝑖𝑗 = 0 𝑖𝑓 𝑗 𝑖𝑠 𝑁𝑂𝑇 𝑐ℎ𝑜𝑠𝑒𝑛
The study examines the log odds of individual i choosing delivery alternatives 2 (public hospitals) and 3 (private hospitals or clinics) compared to the reference alternative of home delivery Alternative 2 serves as the base outcome The analysis focuses solely on chooser-related characteristics due to the limitations of the MICS 4 data, which does not include information on healthcare providers Independent variables are categorized into three groups: individual, household, and community, similar to the approach used in analyzing the demand for prenatal care visits.
The mlogit command in Stata is used to estimate multinomial logistic models, incorporating independent variables categorized into three distinct groups: individual level, household level, and community level characteristics, which will be elaborated on in the following section.
Data
The Multiple Indicator Cluster Survey (MICS) in Vietnam, conducted by the General Statistics Office in collaboration with UNICEF, aimed to provide national-level estimates on the status of children and women across urban and rural areas and six geographic regions The survey utilized three sets of questionnaires: one for households to gather information on economic status and household members, another for female household members aged 15-49, and the last for children under 5 years and their caretakers MICS5 included a sample of 10,018 households, with 9,827 women and 3,316 children interviewed To minimize recall errors, the study focused on 1,479 women who had given birth to a live child within the two years preceding the survey.
Variables definition
The study evaluates maternal health care service utilization through two key indicators: antenatal care coverage and place of delivery Antenatal care coverage is defined as the percentage of women aged 15–49, who had a live birth within the two years prior to the survey, that received care from skilled health personnel at least once The World Health Organization recommends that pregnant women have a minimum of four antenatal care visits during their pregnancy Early attendance at these visits is vital for preventing and identifying potential health issues for both the mother and baby, and antenatal care should be maintained throughout the entire pregnancy.
“Skilled personnel” includes accredited health professionals such as midwives, physicians and nurses, but not traditional birth attendants Therefore, the antenatal care visits are integer variables
The place of delivery refers to whether childbirth occurs in a health facility, such as public or private hospitals, or outside the health system, like home births Increasing the number of births that take place in health facilities significantly lowers health risks for both mothers and babies Ensuring proper medical care and maintaining hygienic conditions during delivery can mitigate complications and infections that may lead to morbidity or mortality Consequently, delivery locations are classified into three categories: home births, births in public hospitals, and births in private hospitals or clinics.
The study utilizes individual-household and community-level explanatory variables derived from existing theoretical and empirical literature to assess the use and availability of maternal healthcare services Two regressions will be conducted for the distinct research objectives, employing a consistent set of independent variables across both analyses A comprehensive description of the selected variables is provided below.
The study examines various individual-level factors influencing maternal health, including the mother's age at childbirth, which ranges from 15 to 49 years, and birth order represented as a continuous variable Maternal education is categorized into five levels: no education, primary, lower secondary, upper secondary, and tertiary Marital status is assessed through dummy variables, indicating whether a woman is divorced or never married Additionally, exposure to mass media is evaluated using four dummy variables based on access frequency to mobile phones, newspapers, radio, and television Lastly, pregnancy intention is measured, with a value of 1 indicating an unintended pregnancy and 0 for intended pregnancies.
Household-level factors influencing living standards include the ethnicity and religion of the household head, as well as the household's wealth status Ethnicity is categorized as Kinh or Hoa (1) versus non-Kinh/Hoa (0) Despite being the sixth largest minority group in Vietnam, their living standards are comparable to those of the Kinh majority The religion variable is represented as a dummy variable, with 1 indicating no religion and 0 otherwise Additionally, household wealth status is also a dummy variable, where 1 signifies that the household belongs to the poorest or poor quintiles and 0 indicates otherwise.
Community-level factors, alongside individual and household factors, play a crucial role in assessing overall effects The analysis categorizes place of residence into urban (coded as 1) and rural (coded as 0) Additionally, six regional dummies are utilized to represent various socioeconomic regions, including the Red River Delta and Northern Midlands and Mountain areas.
The article discusses key regions in Vietnam, including the Central area, Central Coastal area, Central Highlands, South East, and Mekong River Delta It highlights the illiteracy rate among women in the community, measured by the percentage of illiterate women Additionally, it examines the poverty rate, defined as the percentage of women in the lowest wealth quintile within the commune Lastly, it addresses maternal health by noting the percentage of women delivering their babies in hospitals.
ANC Number of antenatal care visits visits
Where the women give birth 1- At home
2- Government hospital or commune health centre 3- Private hospital or clinic categories
AGE Age in years years
NOEDU Dummy variable indicating the individual has not finished primary school Yes = 1, No = 0
PRIMARY Dummy variable indicating the individual finished primary school Yes = 1, No = 0
The article discusses three educational attainment variables: LOWSECOND, which indicates whether an individual completed lower secondary school (1 for Yes, 0 for No); UPSECOND, which signifies completion of upper secondary school (1 for Yes, 0 for No); and TERTIARY, representing those who have completed college or higher (1 for Yes, 0 for No).
MARITAL Whether the women is separated or never married Yes = 1, No = 0
CEB Number of children ever born children
WORKING Whether the woman has been working in the last two years Yes = 1, No = 0
MOBIPHONE Whether the woman reads or writes SMS messages everyday Yes = 1, No = 0
NEWSPAPER Whether the woman reads Newspaper or Magazine everyday Yes = 1, No = 0
RADIO Whether the woman listens to radio everyday Yes = 1, No = 0
TV Whether the woman watches TV everyday Yes = 1, No = 0
UNWANTED Whether the woman wanted the last child No = 1, Yes=0
POOR Whether women belong to the poorest and poor quintiles group Yes = 1, No = 0
HHSIZE Number of household members persons
ETHNIC Whether the household head belong to the ethnic minority group Yes = 1, No = 0
NORELI Whether the household head has no religion Yes = 1, No = 0
RURAL Whether the women live in rural area Yes = 1, No = 0
RRD Red River Delta Yes = 1, No = 0
MN Northern Midlands and Mountainous Area Yes = 1, No = 0
NC North Central and Central Coastal Area Yes = 1, No = 0
CH Central Highlands Yes = 1, No = 0
SE South East Yes = 1, No = 0
MD Mekong River Delta Yes = 1, No = 0
POVERTY Percentage of women with poorest and the 2nd quintile in the commune percentage (%)
ILLITERACY Percentage of women in the commune with no education certificates percentage (%)
HOSPDELIRATIO Percentage of women in the commune giving birth the last child at hospitals percentage (%)
CHAPTER IV RESULTS AND DISCUSSIONS
This chapter outlines the findings of a study examining the relationships between various determinants and the demand for prenatal care visits, as well as the selection of delivery care providers Initially, it presents descriptive statistics for both dependent and independent variables Next, it analyzes the bivariate associations between each dependent variable and the independent variables Finally, the chapter concludes with a regression analysis of the demand for prenatal care visits and the choice of delivery care, comparing these results with those from previous studies.
Descriptive Results
Table 2 and Table 3 present the summary statistics, revealing that 1,479 women reported giving birth in the last two years On average, these women attended 6 prenatal care visits, exceeding the World Health Organization's guideline of at least 4 visits for a safe pregnancy and fetal development The average age of the participants is 28, with ages ranging from 15 to 47 Consistent with Vietnam's two-child policy, the average number of children per woman is 2, although some women in rural areas have as many as 11 children Alarmingly, 100% of women from the poorest and poor household wealth quintiles belong to communities where basic needs are unmet, and over half of the women in certain areas are illiterate, often lacking even primary education.
Table 3 presents descriptive statistics for dummy variables related to childbirth choices among women A majority opted for delivery in government hospitals or community health centers to ensure safety, although 136 women chose to give birth at home Most participants had completed lower secondary education, accounting for 35% of the total, followed by upper secondary, tertiary, primary, and a mere 6% with no education Additionally, 3% of the interviewed women reported being previously married or never married Notably, nearly 20% experienced unintended pregnancies, and many were unemployed In terms of media exposure, women predominantly engaged with television and SMS, rather than newspapers or radio, reflecting the widespread use of these mediums in Vietnam for accessing current affairs and maintaining communication.
Over 40% of women reside in the poorest wealth quintiles, with approximately 24% belonging to ethnic minorities Many of these women live in households that do not adhere to any religion, although ancestor worship is prevalent in Vietnamese families Buddhism is the most practiced religion, followed by Catholicism The residential location significantly impacts the utilization of maternal health care, as most interviewed women live in rural areas, which often lack adequate health care facilities and have less developed economic conditions compared to urban regions Notably, there is no significant difference in the number of women across six socio-economic regions.
Table 2:Descriptive Results – Numeric Variables
Variable Obs Mean Std Dev Min Max
Table 3 :Descriptive Results - Dummy Variables
Home 136 9.2 Not using every day 1,299 87.83
Public Hospital 1,280 86.54 Using every day 180 12.17
Education level Not using every day 245 16.57
No certificate 89 6.02 Using every day 1,234 83.43
Martial status Kinh/ Hoa Group 1,129 76.34
Analysis of Demand for prenatal care
Table 4 illustrates the bivariate relationship between the number of prenatal care visits and various social determinants Notably, women without an educational certificate have significantly fewer prenatal care visits—three less on average—compared to those with higher education Additionally, disparities are evident among women from rural areas, low-income households, and different ethnic groups, with disadvantaged women facing greater barriers to accessing prenatal care Furthermore, regions like the Red River Delta show a marked contrast in prenatal care accessibility.
In the Mekong River Delta and Southeast regions, the likelihood of utilizing prenatal care is higher compared to less developed areas like the Central Highlands and Northern Midlands However, there is minimal difference in maternal health care demand between working women and religious women As depicted in Figure 9, the use of maternal health care declines with increasing age, the number of children a woman has, and the community's illiteracy rate.
Table 4: Bivariate analysis in the demand of prenatal care visits
ANC Observation Mean Std Dev Min Max
T ime s re ce ive d a n te n a ta l ca re
T ime s re ce ive d a n te n a ta l ca re
Figure 7: The association between the demand of maternal care visits and numerical independent variables
4.2.2 Analysis of Negative Binomial Model
The regression analysis begins by testing the likelihood ratio to determine if the dispersion parameter alpha is equal to zero A test statistic with a p-value close to zero indicates that the response variable is over-dispersed, making the Negative Binomial model more suitable than the Poisson model, as detailed in Appendix 6 Subsequently, the regression is conducted using robust methods to address heteroscedasticity The findings from the Negative Binomial model are presented in Table 5.
The study reveals that age significantly impacts prenatal care visits, with a coefficient of 0.016 and a p-value close to zero Specifically, an increase of one year in a woman's age correlates with an additional 0.08 prenatal care visits, indicating that older women tend to seek maternal care more frequently This finding contrasts with previous research by Arthur (2012) and Tsawe & Susuman (2014), suggesting that older pregnant women may be more aware of the potential risks associated with pregnancy.
Research indicates that educational attainment plays a crucial role in socioeconomic outcomes, as evidenced by studies (Arthur 2012, Bbaale 2011, Navaneetham & Dharmalingam 2002) showing significant disparities among individuals with no education, primary education, and lower secondary certificates These findings highlight the clear differences in opportunities and outcomes associated with varying levels of education.
T ime s re ce ive d a n te n a ta l ca re
T ime s re ce ive d a n te n a ta l ca re
T ime s re ce ive d a n te n a ta l ca re
T ime s re ce ive d a n te n a ta l ca re
Research indicates that individuals without any educational certificates are expected to have 1.67 times fewer prenatal care visits compared to those with tertiary education, while other factors remain constant Furthermore, individuals with primary and lower secondary education are projected to have 0.76 and 4.98 times fewer prenatal care visits, respectively, than tertiary certificate holders Notably, there is minimal difference in the number of prenatal care visits between upper secondary certificate holders and those with tertiary education.
The study reveals that exposure to mass media significantly influences prenatal care visits, with a notable finding that individuals who read newspapers and magazines daily have 0.39 more prenatal care visits compared to those who read less or not at all In contrast, the frequency of reading SMS messages, listening to the radio, and watching TV does not show a statistically significant impact on prenatal care visits.
On the contrary to Wado et al (2013), the coefficient of unwanted pregnancy is not statistically significant Similarly, the coefficient of working status also is not statistically significant
The significant negative coefficient of non-union in marital status aligns with the findings of Sepehri et al (2008) and Chen et al., highlighting its importance in understanding marital dynamics.
In 2003, it was found that women living with their husbands are 1.16 times more likely to attend prenatal care visits compared to single expectant mothers This suggests that single mothers may experience stigma when seeking maternal checkups in public, as childbirth is often viewed as a consequence of marriage.
The birth order coefficient is statistically significant, with a p-value nearing zero, indicating that as women have more children, their number of prenatal visits decreases by 0.52 This decline is likely due to the increased responsibilities associated with caring for multiple children, which limits their ability to attend check-ups These findings align with previous research conducted by Navaneetham & Dharmalingam (2002), Sepehri et al (2008), Arthur (2012), and Tsawe & Susuman (2014).
The study reveals that living in a poor household, belonging to an ethnic minority group, and not adhering to any religion significantly impact women's access to prenatal care, while household size does not have a notable effect Notably, women from the poorest quintiles experience 1.1 fewer prenatal visits than those from higher quintiles Financial constraints hinder these women from affording prenatal care and transportation, supporting the findings of previous research by Bbaale (2011) and Tsawe.
& Susuman 2014) that the household wealth income negatively affects the demand of prenatal check-ups
Similar to these studies of Celik & Hotchkiss (2000), Navaneetham & Dharmalingam
Women from ethnic minority households experience 0.6 fewer prenatal care visits compared to those from the Kinh or Hoa groups, likely due to language barriers and cultural differences Additionally, women whose household heads do not adhere to any religion show a positive influence on prenatal care demand, with a marginal effect of 0.3, suggesting that religious norms may hinder access to these services This study supports findings from Bbaale (2011) and Tsawe & Susuman (2014).
The study indicates no statistically significant difference in prenatal care demand between rural areas and communities with higher poverty rates, contradicting Sepehri et al (2008) However, significant community-level factors include the proportion of women with no education and those delivering in hospitals Similar to Gage (2007), expectant mothers in areas with higher illiteracy rates have 0.05 more prenatal care visits compared to those in lower illiteracy regions Conversely, a higher hospital delivery ratio positively influences antenatal care demand, with women in such areas having 0.03 more visits This suggests that community practices significantly affect women's attitudes and healthcare-seeking behavior, supporting findings by Ononokpono et al (2013).
There is significant regional disparity in the utilization of prenatal care visits, consistent with findings from Celik & Hotchkiss (2000) and Sepheri et al (2008) Residents in underdeveloped areas, such as the Central Coast, Central Highlands, and Northern Midlands and Mountainous Area, are 1.05, 0.76, and 0.71 times less likely to frequently attend prenatal care visits compared to those in the Mekong Delta However, no statistically significant difference exists between residents of the Southeast region and those in the Mekong River Delta.
It implies that there are remarked differences in the implementation of health care program, the availability and accessibility of the health care services among the regions
Table 5: Negative binomial regression for the demand of prenatal care visits
Independent Variables Individual-level Characteristics
MOBIPHONE (using mobile phone every day=1) 0.047 (0.030) 0.239 (0.157) NEWSPAPER (reading newspaper every day=1) 0.075* (0.035) 0.385* (0.184)
RADIO (listening radio every day=1) -0.062 (0.040) -0.308 (0.190)
TV (watching television every day=1) -0.014 (0.046) -0.071 (0.236)
POOR (living in the poor and poorest households=1) -0.217*** (0.047) -1.081*** (0.231)
ETHNIC (being in ethnic minority group=1) -0.117* (0.052) -0.572* (0.248)
RURAL (living in rural areas =1) -0.0482 (0.033) -0.245 (0.169)
Northern Midlands and Mountainous Area -0.148* (0.059) -0.716* (0.270) North Central and Central Coastal Area -0.227*** (0.049) -1.059*** (0.215)
*** 1% significance, ** 5% significance, * 10% significance - Robust standard errors in parenthesis
Analysis of Choice in the delivery care providers
Table 6 highlights the correlation between delivery facility choices and various independent variables It indicates that women with a higher birth order are more inclined to opt for home childbirth compared to those with a lower birth order Furthermore, a higher average poverty rate and illiteracy rate are associated with women choosing to give birth at home, while those delivering at home typically reside in communities with a lower rate of facility-based deliveries.
Table 6 : Bivariate analysis in the choice of delivery care providers - numeric independent variables
Obs Mean Std Dev Min Max
Home 136 39.73 26.78 0 77.78 Public facility 1280 4.35 10.05 0 77.78 Private Facility 63 4.36 7.53 0 28.57
RATE OF WOMEN GIVING BIRTH IN HEALTH FACILITY
Home 136 29.81 28.14 0 85.71 Public facility 1,280 96.93 11.26 7.69 100 Private Facility 63 98.12 7.29 70.00 100
Table 7 illustrates the relationship between delivery care provider choices and various independent variables Delivery locations are categorized into three options: home delivery, public hospitals, and private hospitals Home delivery is the most economical but comes with lower hygiene standards and limited medical equipment Public hospitals, funded by the government, offer low costs but often face overcrowding In contrast, private hospitals provide superior services and advanced technology at a higher price The data indicates that women opting for home births have less exposure to mass media compared to those using health facilities Additionally, the likelihood of home births increases among women in disadvantaged areas, such as the Central Highlands and North Mountainous regions, where access to healthcare is hindered by poor infrastructure Notably, women who delivered at home had typically been engaged in low-paying farm work for two years prior to the interview, limiting their ability to afford healthcare costs Conversely, women with higher living standards and those from ethnic majority households are more inclined to give birth in private hospitals, while rural women are less likely to do so, likely due to the concentration of private facilities in urban areas and their associated costs.
Table 7: Bivariate analysis in the choice of delivery care provider – dummy independent variables
Not using every day 135 99.26 900 70.31 40 63.49 Using every day 1 0.74 380 29.69 23 36.51
Not using every day 136 100.00 1,013 79.14 45 71.43 Using every day - - 267 20.86 18 28.57
Not using every day 134 98.53 1,111 86.80 54 85.71 Using every day 2 1.47 169 13.20 9 14.29
Not using every day 74 54.41 162 12.66 9 14.29 Using every day 62 45.59 1,118 87.34 54 85.71
The middle quintiles and more 2 1.47 804 62.81 50 79.37
4.3.2 Analysis of Multinomial Logistic Model
The estimated multinomial logistic regression coefficients are detailed in Table 8, with the marginal effects presented in Table 9 The reference group consists of individuals who opted for delivery in public hospitals The coefficients indicate the positive or negative relationships between independent variables and the likelihood of selecting home or private hospital deliveries compared to public hospital deliveries Additionally, the marginal effects illustrate how changes in independent variables influence the probability of choosing home or private clinic deliveries, while keeping other variables constant Note that the variable hospdeliratio has been omitted due to collinearity issues.
Childbirth at home relative to Childbirth at Public Hospital
Research indicates that households characterized by poverty and belonging to ethnic minority groups have a significantly higher likelihood of home childbirth, with a p-value approaching zero This suggests that financial constraints, along with language and cultural barriers, hinder pregnant women from opting for health facility deliveries These findings support previous studies by Celik & Hotchkiss (2000), Stephenson et al (2006), and Wado et al (2013) Interestingly, the study found no significant association between the religion of the household head and the decision to deliver at home, contrasting with findings from Stephenson et al (2006).
Living in poverty and belonging to an ethnic minority group are associated with an increase in the likelihood of home births by 0.08% and 0.07%, respectively; however, these findings are statistically insignificant.
Individual characteristics significantly influence the choice of delivery location Research indicates that regular exposure to mass media, such as listening to the radio or reading newspapers, is negatively associated with the likelihood of home childbirth, as it enhances awareness of maternal health services Specifically, these media habits decrease the probability of home delivery by approximately 0.03 percentage points Additionally, each prenatal care visit reduces the likelihood of home delivery by 0.02 percentage points, a finding supported by multiple studies Conversely, having more children slightly increases the probability of home delivery by 0.01 percentage points, suggesting that women with higher birth orders may underestimate the necessity of facility-based care Interestingly, factors such as age and education level do not show a significant correlation with the choice of delivery location.
Living in rural areas significantly increases the likelihood of home delivery, with a marginal effect of 0.0003, indicating a 0.03 percentage point rise in the probability of giving birth at home Additionally, a higher concentration of women without educational certificates is positively associated with the choice of home childbirth over public hospital delivery, aligning with findings from previous studies by Celik & Hotchkiss (2000), Gage (2007), and Stephenson et al (2006).
Childbirth at private hospital relative to Childbirth at Public Hospital
The analysis reveals that working status, rural residence, and living in a poor household are significant factors influencing the choice between public and private hospitals, with p-values close to zero Specifically, individuals from poor households have a 0.14 percentage point lower probability of delivering in private hospitals, aligning with findings from Thind et al (2008) that suggest lower living standards correlate with a preference for public hospitals Additionally, residing in rural areas decreases the likelihood of private hospital delivery by 0.12 percentage points Regional disparities also exist, as living in the North Central and South East regions reduces the probability of opting for private hospitals by 0.12 and 0.1 percentage points, respectively This indicates that private hospitals are predominantly located in urban and developed regions, where delivery costs are higher than those at public hospitals.
Table 8: Multinomial Logistic Regression for the choice of delivery care provider
At home vs At Public Hospital
At Private Hospital vs Public Hospital
Independent Variables Individual-level Characteristics
TERTIARY (college above =1) Reference Reference
MOBIPHONE (using mobile phone every day =1) -1.358 (1.102) 0.197 (0.326)
RADIO (listening radio every day =1) -1.219* (0.689) 0.0936 (0.383)
TV (watching television every day =1) -0.442 (0.315) -0.325 (0.413)
POOR (living in the poor and poorest households=1) 1.953* (0.885) -0.917* (0.479)
ETHNIC (being in ethnic minority group=1) 1.357** (0.521) -0.121 (0.462)
RURAL (living in rural area =1) 0.873* (0.485) -0.640* (0.298)
Northern Midlands and Mountainous Area 1.827* (0.976) -17.78*** (0.346)
North Central and Central Coastal Area 1.868* (1.076) -0.968* (0.425)
Mekong River Delta Reference Reference
*** 1% significance, ** 5% significance, * 10% significance - Robust standard errors in parenthesis
Table 9: Marginal effects for the choice of delivery care provider
At Public Hospital At Private Hospital
Marginal effect Marginal effect Marginal effect
Independent Variables Individual-level Characteristics
TERTIARY (college above =1) Reference Reference Reference
MOBIPHONE (using mobile phone every day =1) -0.0003
(0.0006) NEWSPAPER (reading newspaper every day =1) -0.0033 (0.0017) 0.0033
(0.0006) RADIO (listening radio every day =1) -.00030
TV (watching television every day =1) -0.0002
(0.0001) POOR (living in the poor and poorest households) 0.0008
(0.0008) ETHNIC (being in ethnic minority group=1) 0.0007
RURAL (living in rural area =1) 0.0003
Northern Midlands and Mountainous Area 0.0012
North Central and Central Coastal Area 0.0013
Mekong River Delta Reference Reference Reference
*** 1% significance, ** 5% significance, * 10% significance - Robust standard errors in parenthesis