CHapter  i  Introduction

Quality of human resources is a great concern to researchers and policymakers. Health improvement is considered as a part of human capital development in all countries in the world. People have to make a choice regarding utilization of health care services. Demand for health care is a major topic in health economics. It is also paid much attention to by researchers in empirical studies in health sectors. There are a lot of empirical studies on estimating determinants of demand for health care or health care expenditure that investigated some key factors determining health care expenditure such as age, gender, region, income, region, and type of diseases (David 1993, Hjortsberg 1999, Mocan 2000, Rous and Hotchkiss 2000, Lahga 2001).

In Vietnam, the Doi moi process started in 1986, of which health sector reform is a part. The most important health sector reforms are the introduction of user fees for health services at higher level public health facilities, legalization of private medical providers, liberalization of the pharmaceutical industry, and the introduction of a health insurance program in 1993. All of these reforms have had profound impacts on the health sector and on household health care-seeking behavior. However, one big problem of the health sector according to a report by the World Bank (2001) is that despite a reduction in the real price of drugs in recent years, health care costs, especially user fees at public hospital, are very large in relation to disposable income for the poor. Two Vietnam living standard surveys in 1993 and 1998 showed that most of the poor consider health care utilization with reluctance. The main reason for this is that expenditure for health care services is too high for the poor.  Moreover, there is a lack of confidence by people in the quality of local medial care services.

In Vietnam, few studies on health care seeking behavior have been carried out. Trivedi (2002) has studied pattern of health care utilization in Vietnam. In this study, Trivedi focused on analyzing the choice of provider types. He also showed separately the model for positive medical expenditure for individuals and households. In another study on health care expenditure in Vietnam, Do, Nguyen, Tran, and Vu (2001) pointed out many factors, such as demographic and geographic variables, income, insurance, and health status as determinants of health spending for sick people. All above studies were carried out for the country as a whole. In Vietnam, 80% of the Vietnamese population live in rural areas where health care systems are less developed compared to urban areas so improvement of rural health care system in order to enhance health status of people is always a target of government policy. As report by World Bank (WB), moreover, the perception of rural people of health care utilization is still very low (WB, 2001). Therefore, it is worth researching health care seeking  behavior in rural areas of Vietnam

This thesis will analyze the key determinants of health care expenditure in rural Vietnam. The thesis will explore the behavior of sick individuals with positive health care expenditure based on VLSS 97-98 data source. Explanation of individuals with zero health care expenditure is beyond the scope of this thesis.

The objective of the thesis is to answer the following research questions:

1.        What are the patterns of health care utilization in rural Vietnam?

2.        What are key factors that affect the health status of people in rural areas?

3.        What are the most important factors that determine whether sick people use medical care services?

4.        What are the most important determinants of health care expenditure in rural Vietnam?

The thesis will analyze the key factors that determine health care expenditure in Vietnam based on a framework of demand for health care. Effects of key factors on health care expenditure in Vietnam are examined by using multi-variable regression. The quantitative analysis is based on the Vietnam Living Standard Survey 1998 (VLSS 97-98). Descriptive analysis is also used to analyze the pattern of health care utilization in Vietnam.

The thesis is divided into four main chapters, excluding introduction. Chapter II introduces a theoretical framework that reviews theories and empirical studies related to the topic. Chapter III analyses the pattern of health care utilization in Vietnam based on a descriptive method. Chapter IV examines determinants of health care expenditure in Vietnam by using an econometric model. Chapter V summarizes all analysis and findings in previous chapters, and gives some policy recommendations.

Chapter ii  Theoretical  framework

The objective of this chapter is to answer the following two questions:

1. What is the relationship between utility and demand curve for health care services?

2.    What are factors that affect demand for health care services?

2.1 Concepts and Definitions

Health care is not an economic term and its definition is very broad. According to Appelbaum (1999), “health care” means any care, treatment, or procedure by a health care provider

(i) To diagnose, evaluate, rehabilitate, manage, treat, or maintain the physical or mental condition of an individual; or

(ii) That affects the structure or any function of the human body.

(iii) “Health care” includes participation in research that, considering the risks and benefits of participation, presents a reasonable prospect of direct medical benefit to an individual.

A definition of Health Care is found in the Dependent Adults Act in the United State. “Health Care” includes:

(i) any examination, diagnosis, procedure or treatment undertaken to prevent any disease or ailment,

(ii) any procedure undertaken for the purpose of preventing pregnancy

(iii) any procedure undertaken for the purpose of an examination or a diagnosis,

(iv) any medical, surgical, obstetrical or dental treatment, and

(v) anything done that is ancillary to any procedure, treatment, examination or diagnosis.”

The other definition is introduced by Filmer based on the Alma Ata declaration (WHO, 1988). Filmer, Hammer, and Pritchett (1997) defined health care as composed of three distinct elements: (i) simple curative care usually base in primary facilities, (ii) preventive activities aimed at health improvement, and (iii) public health campaigns.

Based on above definitions of health care, the term of health care used in the thesis is defined as all activities that satisfy the needs and wants of people to have good health.

2.2 Demand for health care

The thesis will analyze health care expenditure based on demand for health care theory that is derived from theories of microeconomics on consumer behavior. Consequently, determinants of health care expenditure will be pointed out based on determinants of demand for medical care.

2.2.1 The transformation of health care to health

The transformation of health care to health is process in which health as an output and health care as one of many inputs into the production of health. Grossman (1999) used the approach of household production function to account for the transformation of medical care to health. According this approach, health can be written as a function of medical care as follows:

H = f (m)

Where H is a health outcome and m is medical care services. It is normally assumed that more m produces more H, i.e. the marginal productivity of medical care is positive.

Figure 2.1 Health production functions for three diseases

2.2.2 Utility function for health care and other consumption

Consider a bundle of stock of health, H, and other goods x, an individual's utility function can be expressed as follows:

U = U(H,x)

or                                               U = U(m,x)

We can use the indifference curve as in figure 2.2 to represent preference over health care and other consumption goods.

Figure 2.2 Indifference curves for health care and other consumption

                                Source: Adapted from Jack (1999)

2.2.3 Optimal choice for health care and other consumption

In consumer theory, the bundle of goods that maximizes satisfaction must lie on the highest indifference curve that touches the budget line, i.e. the optimal choice is a point of tangency between the indifference curve and the budget line.

Figure 2.3  The consumption choice when people get sick (a) unchanged income,                     and (b) decrease income

2.2.4 Demand curve for health care

In microeconomic theory, a demand curve is derived based on considering how desired quantity of one good, that is represented by optimal choice of consumer, responses to change in its own price. The demand curve for health care is shown in figure 2.4 below

Figure 2.4a An increase in price of medical care

Figure 2.4b Demand curve for medical               care

2.3 Determinants of demand for health care

In this part, some key determinants of health care demand are pointed out based on standard theory of consumer behavior and on empirical studies on demand for health care.

2.3.1 Prices of health care

As demand for any goods and services, demand for health care is sensitive to the fluctuation in prices of health care services. In studying effects of the price on demand for health, most of researches focused on the difference of price elasticities for the poor and nonpoor (Akin, Guilkey and Denton 1995; Gertler, Locay and Sanderson 1987; Mocan 2000). In Vietnam, demand for health care and health care expenditure have been analyzed based on VLSS database in which information on price of health care services is inadequate. Therefore, price variables was omitted in the studies on health care in Vietnam (see Trivedi 2001; Lan 2001).

2.3.2 Income

The effects of income on the resource allocation between consumption and health care can be illustrated by income expansion paths as in figure 2.5.

Figure 2. 5  Income Expansion curves when health care is (a) a luxury  and (b) a necessity

In empirical studies, whether health care is a necessity or a luxury good is still a debate (see Newhouse 1977, Parkin et al. 1987, Gerdtham et al.1992). According to a report by World Bank on health sector in Vietnam, medical care is a luxury good for the poor in rural areas (World Bank, 2001). However, this argument is not improved by any researches in Vietnam (see Do et al. 2001; Trivedi 2001). Therefore, it is quite relevant to examine whether health care is luxury or necessity in the case of rural Vietnam.

2.3.3 Travel costs and time costs

In empirical studies, travel cost may be one of key determinants of demand for health care, especially in the case of developing countries. Phelps (1992) argued that travel to and from the doctor creates costs, both in terms of time and the direct travel costs. Travel costs to health care facilities are often higher in rural areas, especially in developing countries like Vietnam where infrastructure is underdeveloped. Moreover, the poor in rural areas does not own good vehicles that can increase the time costs of health care utilization.

2.3.4 Quality of health care service

In health economics, quality of health care service is a key determinant of demand. Phelps (1992) divided these aspects of quality into two facets. The first relates to the role of medical care in producing outcomes of good health. The second relates to the amenity aspect of medical care such as friendliness of staff, the convenience of working time and the like.

We should expect that the better the quality the higher demand for health care. But some improvements of quality can reduce the demand for health care because higher quality may improve the health status of individuals. Therefore, the effect of the quality on demand for health care is ambiguous.

2.3.5 Illness events

Demand for health care may vary depending on illness events. A series of demand cure for various illness events is shown in figure 2.6 where P_m denotes price of health care services.

Figure 2.6 Demand curves for various illness events

2.3.6 Age and Gender

Demand for health care changes in people's life cycle. When we become older, our demand for health care will increase. In empirical studies, age is mentioned as one of the key determinants of demand for medical care. It is founded that the relationship between age and health care utilization is positive. This means that people demand more health care as they become older.

Gender issues are paid much attention in health economics, especially in studies of demand for medical care. Most of empirical studies considered gender/sex of individuals as a key determinant of demand for health care (David, 1993; Mocan, 2000; Gupta and Dasgupta, 2000 and Sahn et al., 2002). Gender issues are also related to other aspect of health care choice that is boys have more favorable conditions in health care than girls.  

2.3.7 Education

A lot of researches on health care demand conclude that people with higher level of education demand more health care (see Mocan, 2000; Gupta and Dasgupta, 2000; Hjortsberg, 1999). In Vietnam, however, Do et al. (2001) found that education variables is insignificant in determining health care expenditure

Chapter IIi  Pattern of health care utilization in rural Vietnam

This chapter will analyze the pattern of health care utilization based on the social and economic characteristics of rural areas. First, the chapter aims to provide the whole picture of health care utilization in rural areas. Second, the chapter will focus on examining behavior of people in using health care services by a tool of descriptive and comparative analysis.

3.1 Overview of health status and health care expenditure in rural Vietnam

Calculation from VLSS 97- 98 shows that 42.8% of individuals in rural areas reported that they were ill or injured during the 4 weeks prior to being surveyed. In urban areas, this figure was 37.8%. The percentage of sick women is higher than that of men. In rural areas, 45.8% of women were ill compared to 39.7% of men. In urban areas, these figures are 41.2% and 34.2%, respectively.

For sick individuals, the percentage of types of illness is higher rural areas. The most common types of illness in rural are headache and dizziness (17.7%), and cough or cold (17.5%). There were substantial cases of fever (9.8%).

In Vietnam, per capita expenditure on health care is lower in rural areas. According to VLSS 97-98, rural per capita health care expenditure is equal to about 57% of urban areas. More specifically, per capita spending for health care in 4 weeks prior to survey is 45.51 and 79.93 thousand dong[1] in rural and urban areas, respectively.

The health status of people in rural areas is also analyzed based on the following aspects:

¨       Percentage of sick people by age groups and gender

¨       Percentage of sick people by regions

¨       Percentage of sick people by per capita expenditure quintiles

3.2 The choice of health care providers

In this section, the thesis will examine individual's behavior in choosing health care providers by comparing the average number of visits to different providers that is calculated based on VLSS 97-98. In Vietnam, there are eight types of medical care providers: government hospital, commune health center (CHCs), regional polyclinic, other government facility, private health facility, traditional medical care, doctor's home visits and pharmacy.

3.2.1 Choice of health care providers by sex and age group

Female in all group of age have higher number of visit to different health care provider than male. For children in the group of 0-4, however, the number of visits is higher for boys. For different types of providers, the visit number to regional polyclinic is lowest in all group of age for both male and female. The highest number of visit is observed for pharmacy.

3.2.2 Choice of health care providers by regions

The highest number of visit to government health care facility is Southeast. The lowest is in Northern Upland because most of people in this region live in mountain and remote areas. Consequently, it is difficult to have access to government hospital due to the distance to the nearest providers. As a result, the number of visit to CHCs is very high in Northern Upland, only after Mekong River Delta.

3.2.3 Choice of health care providers by per capita expenditure quintiles

Individuals in higher per capita expenditure quintile tend to use health care services more than lower quintiles. In VLSS 97-98, per capita expenditure is grouped in 5 quintiles that represent for 5 groups of economic situation as in table 3.7 given below. 

Table 3.7  Mean number of visits to different health care providers by quintiles

Source: Author's calculation based on VLSS 97-98

3.3 Health care expenditure in rural Vietnam

In this part, the thesis will analyze individual health care expenditure by factors including income, age, gender, education, and region. Health care expenditure is defined as a total cost for purchasing medicines, transportation, room and board, and care taking.

3.3.1 Individual health care expenditure by age group

In general, the older people become the higher health care expenditure they spend. When people become older they are more likely to get professional help in the event of illness. This may be because old people are more aware of the seriousness of their health status, or less capable of recovering naturally. This argument is illustrated in table 3.8 where the age group having the highest per capita health expenditure is over 60 group.

·         Health care expenditure by age group and per capita expenditure quintiles

Health expenditure in each group also differs by per capita expenditure quintiles. Individuals in higher quintile have higher health care expenditure, except the group of 0-4 in which the highest medical expenditure is mid-upper quintile.

Table 3.9 Health care expenditure by age group and expenditure quintiles ('000 dong)

Sources: Author's Calcualtion from VLSS 97-98

3.3.2 Individual health care expenditure by education

Individual behavior in health care utilization is also determined by education level. People with higher education level tend to spend more for health care. In the thesis, education level of an individual is classified into 5 levels: no education, primary, lower secondary, upper secondary, and university.[2] There is an unclear relationship between education level and health care expenditure.

·         Health care expenditure by education level and per capita expenditure quintiles

 The difference in health care expenditure by education levels can be observed more clearly by quintiles of per capita expenditure. Health care spending in each education level varies across expenditure quintiles. One surprise thing is that in the upper and mid-upper expenditure quintiles, health care expenditure of no education level is highest meanwhile university level spent least for medical care. This fact rises a question of whether higher level of education always associated with higher medical care expenditure, especially in the case of rural Vietnam.

3.3.3 Individual health care expenditure by gender

Sex of an individual is one important factor that influences people's behavior in health care utilization. In reality, women tend to spend more for health care because of nature physical characteristics, i.e. women are weaker than men. Moreover, they have a natural function is to be mothers so their health wears away faster than men. In rural areas, especially, most of women have to work hard in agricultural sector that affects adversely their health status.

·         Health care expenditure by gender and per capita expenditure quintiles

Medical care spending of female and male is different in each quintile of per capita expenditure. Poor women in rural areas have less access to health care services

3.3.4 Individual health care expenditure by regions

Individual health care expenditure also varies across different regions because each region has distinct features of geography, demography, and custom. For rural areas, individuals in Mekong River Delta had the highest expenditure for medical care and the lowest is in Northern Upland. This results from socio-economic characteristics of these regions. In Mekong River Delta, one of two main growing rice regions in Vietnam, farmers have relatively high income from growing rice for export and domestic consumption. Moreover, in a delta region people have more comfortable conditions (better transportation and health care systems) to access health care services[3]. Meanwhile, the Northern Upland is an isolated area where people are suffer from many difficult conditions such as harsh weather, underdeveloped production and backward infrastructure. In addition, the existence of old custom obstructs the access modern health care services of local people, especially ethnic minority people. 

Table 3.13 Health care expenditure by regions  ('000 dong)

                Source: Author's Calculation base on VLSS 97-98

3.3.5 Individual health care expenditure and Insurance

Health insurance is an important determinant of health care expenditure. In general, health insurance reduces the money payment for health care services so insured people have higher contact rates with health facilities.

Table 3.14 Health care expenditure by insurance ('000 dong)

                Source: Author's Calculation based on VLSS 97-98

Chapter Iv  determinants of health care expenditure in rural Vietnam

In order to answer the main question of "What are key determinants of health care expenditure in rural Vietnam?", chapter IV will use quantitative method to analyze significance and magnitude of each factor's effect on health care expenditure in rural Vietnam.

4.1 Data Set and Methodology

4.1.1 Data Set

This thesis mostly uses data source from The Vietnam Living Standard Survey 1997-1998, which is a large database related to many aspects of socio-economic life such as employment, education, fertility, and health. This is second survey carried out in Vietnam. The first survey was in 1992-1993. They were conducted by Vietnam General Statistical Office (GSO) with technical assistance of WB. Both surveys are nationally sampled of which the first survey was a sample of 4,800 households and the second one was 6,000 households.

4.1.2 Methodology

In this chapter, the thesis will use the econometric models to explain behavior of individuals in health care utilization in rural areas. A three-step analysis will be used. In the first step, a logistic model will be used to examine what factors cause individuals to have sickness. In the next step, a logistic model is also used to point out what factors determine whether a sick individual will use medical care services. The last step is to answer the main question of the thesis is what determinants of health care expenditure are. In this step, the thesis will use Heckman method to estimate determinants of health care expenditure in rural Vietnam.

·         Dependent Variables: There are three dependent variables:

-A dummy variable is used in the first step that has a value of 1 if an individual get sick or injury in the 4 weeks before survey, and of 0 otherwise.

- In the logistic model of second step, the dependent variable is also a dummy variable that is set equal to 1 if a sick individual get medical care services, and to 0 otherwise.

- In the last model, the dependent variable is logarithm of expenditure on health care

·         Independent variables

- Gender: This is a dummy variable that is set equal 1 if an individual is a male and 0 otherwise.

- Age: This variable will used in all three models. In the thesis, the age of an individual calculated in years.

- Insurance: In econometric models, insurance is a dummy variable that has value 1 if a person is insured and 0 otherwise.

- Education: Education is classified into 5 level as descriptive analysis in chapter III. In the models, 4 dummy variables will be used to present education level of an individual in which no education is omitted.

- Household size: Household size is the number of people in a household.

- Expenditure per capita: Expenditure per capita is a good proxy for an income variable. In the econometric model, expenditure is usually used rather than income variable.

- Regions: In Vietnam, the country is comprised of seven regions that are Northern Upland, Red River Delta, North Central Coast, Central Coast, Central Highlands, Southeast and Mekong River Delta. Northern Upland is used as a benchmark and six dummy variables are used for remaining regions.

- Health status: Based on VLSS 97-98, we can have information on health status of a person by considering pattern of illness that includes 10 types of disease as presented in figure 3.1. Each type of diseases is represented by one dummy variable that takes value 1 if a person have that disease.

4.2 Estimated Results and Explanations 

The estimated results that are presented in this part are the best result. They are obtained after dropping insignificant variables. The basic criterion for dropping variables is p-value. If one variable has p-value higher 0.1 then it will be dropped. Therefore, some variables that are described in part 4.1 will not appear in estimated results .

4.2.1 The logistic model for who was ill

The result of logistic model for estimating factors determining the probability of falling ill is showed in table 4.1. This table indicates that the probability of illness is determined by gender, age, age squared, household size, expenditure per capita, and regions.

According to table 4.1, gender variable has a expected sign that men have lower probability of falling ill. The age variable has a significant impact on health status. A negative sign of the age variable means that when age of an individual increases the probability of illness will reduces.

The household size variable has an unexpected impact on probability of illness. The estimated result shows that a person living in a larger household is less likely to fall ill, holding other variables unchanged.

Table 4.1 Logistic model: Probability of falling ill in the 4 week before survey

Notes: Based on 20,858 observations. Pseudo R²=0.045. Omitted region is Northern Upland

(*) Marginal effect is for discrete change of dummy variable from 0 to 1

Source: Author's Estimation from VLSS 97-98

Expenditure per capita is another important variable. It is no surprise that expenditure is negatively associated with probability of getting sick.

The probability of getting sick varies across difference regions. The probability of illness is highest in Central Highlands. The lowest probability of illness is in Central Coast while the probability of getting sick is relative high in the region of Red River Delta. This is a unexpected result because rural people in delta regions have better living condition compared to other rural areas but this result is anticipated by descriptive analysis. 

4.2.2 Logistic model of getting medical care

In the previous logistic model, significant factors determining the probability of illness have been pointed out. In this part, a logistic model will be used to estimate what factors determining whether a sick person get medical help or not that is shown in table 4.2.

The gender variable has a significant negative effect on probability of getting medical attention for sick individuals.

Age is a variable that has a significant negative effect on the probability of getting medical care but its magnitude is very small. This negative effect is a contrast to result of descriptive analysis in chapter III. Insurance variable has a positive effect on probability of getting health care services.

The probability of getting medical attention also depends on expenditure per capita. An individual with higher expenditure is more likely to get medical care services when s/he falls ill. This is an expected result. There are eight types of illness that have significant effects on probability of seeking medical care. They are cold, vomiting, respiratory, fever, infection, diarrhea, other diseases, and injury.

Table 4.2 Logistic model: Probability of getting Medical care

Note: Number of observation 8,934. Pseudo R²=0.068 This regression is for sick individuals. The omitted region is Northern Uplands.

    (*) Marginal effect is for discrete change of dummy variable from 0 to 1

                Source: Author's Estimation from VLSS 97-98

4.2.3 Model of determinants of health care expenditure

The estimated result shows that older people spend more for health care expenditure because older individuals often have more serious disease than young individuals. However, the effect of age variable on health care expenditure is very small.

Table 4.3 Model for determinants of health care expenditure (Heckman Method)

Note: Number of observations: 20,858 of which Censored observations: 13,425 and Uncensored observation: 7,433. The omitted region is Northern Uplands.

Source: Author's Estimation from VLSS 97-98

There is one point should be noted that the gender variable is not significant in determining health care expenditure.

Another important factor determining individual health care expenditure is the living standard that is measured by expenditure per capita. In rural areas, if expenditure per capita rises by 1%, then per capita health care expenditure will increase 0.6%. This result indicates that health care is not a luxury good for rural people.

Some points on geographic effects derive from the estimated result. The lowest health care spending is in Red River Delta region. While the highest health care expenditure is in Central Highlands where health care system is very poor and subsidy program was implemented late.

In regression result, five types of disease determine health care expenditure, they are vomiting, respiratory, fever, injury, and other. There are differences in spending for each type of disease. This reflects seriousness of different types of illness.

Chapter V Conclusion and Recommendations 

This final chapter will give some conclusions that summarize the main findings and result in the thesis. Based on these conclusions, policy recommendations will be suggested.

5.1 Conclusion

Health status of rural people is worse than urban people, as shown by the higher rate of illness in rural areas.

There is a difference in the number of visits to different types of health care providers. In rural areas, contacted rates of pharmacy, private facilities, and CHCs appear to dominate other types of providers. The ability of rural people to access government hospital is limited because of distance, especially in remote areas. Self-medication, the common problem in health care utilization of Vietnamese in general, is highlighted in the case of rural areas.

There are two main results related to gender issues in seeking health care. First, women are more likely to be ill and spend more for health care. Second, parents pay more attention to the health status of boys than girls in a household.

In the model for estimating the determinants of probability of getting medical care, there are 4 important findings. First, age has a negative impact on the probability of getting medical help. Second, the effect of insurance variable is positive (in the whole country, this variable has a negative effect). Third, per capita expenditure, which does not appear in the same model for the whole country, is significant in determining the probability of seeking medical care. Fourth, there are 8 types of illness that affect the probability of getting medical attention while in the country there are only 3 types of illness.

                Health care expenditure is determined by living standards measured by per capita expenditure. Medical care spending varies across expenditure quintiles. Individuals with high expenditure spend more for health care. Meanwhile, low expenditure people are more likely to get sick but their expenditure for health care is lower.

Behavior of seeking health care is affected by geographic characteristics. People in the Central Highlands suffer from the worst health status (the highest rate of illness) and high costs of utilization due to the lack of government subsidy for health care[4].

5.2 Policy Recommendations

In order to improve the quality and reduce costs of medical care in rural areas, the government should focus more on the health care system in rural areas by equipping CHCs with health care equipment and training professional staff for rural health care facilities.  Improvement of CHCs is also a way to solve the overload problem in high level hospitals.

Programs of using mobile health workers should be enhanced to provide medical care service for people in isolated regions. In addition, developing rural infrastructure, especially the transport system, is a key policy to promote health care utilization in rural areas.

The highest number of visit to drugs vendors without prescription of professional practitioners indicates that there is an excessive reliance on self-medication. Self-medication is very dangerous so it is necessary to propagate the importance of safe drug usage, especially in rural areas where access to health care information is limited. 

The new voluntary health insurance program should be expanded because it can reduce the heavy reliance on self-medication that is the main problem in behavior of seeking medical care not only in rural but also in urban areas.

5.3 Suggestion for further study

In order to have more comprehensive analysis of determinants of health care expenditure more information on health care should be collected such as information on quality and price of medical care, and individual's insurance status.