EIP/HSF/DP.06.7

The Bolivian Health system and its impact on health care use and financial risk protection

DISCUSSION PAPER NUMBER 7 - 2006 Department "Health System Financing" (HSF) Cluster "Evidence and Information for Policy" (EIP)

World Health Organization 2006 © This research was funded by the National Institute on Aging (NIA), grant number 1-PO1-AG17625. The authors thank the National Institute of Statistics in Bolivia (INE) for providing the data used in this analysis. We are grateful as well to Dr Christian Darras and Dr. Alfredo Calvo in the country office of PAHO/WHO for their assistance and helpful suggestions . We also thank Anna Moore for editing the text and Cecilia Vidal for the inputs provided. Preliminary results of this paper were presented at the V International Health Economics Association Conference (IHEA) in July 2005 in Barcelona, and at the 2nd International Conference on Health Financing in Developing Countries in Clermont-Ferrand, France in November, 2005.

The views expressed in documents by named authors are solely the responsibility of those authors.

The Bolivian Health System and its impact on health care use and financial risk protection By

Ana Mylena Aguilar Rivera, Ke Xu and Guy Carrin

GENEVA 2006 2

3

Table of Contents Introduction..................................................................................................................6 I Background................................................................................................................7 1.1 Key morbidity and mortality data................................................................................................... 7 1.2 The health system in Bolivia ........................................................................................................... 7 1.3 The health system reform.............................................................................................................. 10

II Research questions, data and methods ................................................................11 2.1 Research questions ....................................................................................................................... 11 2.2 Data .............................................................................................................................................. 12 2.3 Methods ........................................................................................................................................ 13 2.3.1 Determinants of health utilization - Multinomial Logit analysis .......................................... 13 2.3.2 Estimation of catastrophic health payments and impoverishment......................................... 13 2.3.3 Determinants of catastrophic health expenditures - Logistic Analysis................................. 15

III Econometrical Results.........................................................................................17 3.1 Determinants of health care utilization ...................................................................................... 17 3.2 Determinants of catastrophic health payments............................................................................. 21

IV Discussion ..............................................................................................................23

4

Abstract One of the most important goals of a health system is to ensure that the population is provided with health services, and at the same time, protected from any financial burden due to health expenditures. Bolivia is one of the poorest and least developed Latin American countries and since 1996 has been going through a period of health system reform, characterized by a decentralization process along with the implementation of the national child and maternal insurance scheme. This paper explores the trend in health service utilization, out-of pocket payments, catastrophic expenditure and the socio-economic factors associated with them , particularly for the poor and non-poor, the young (under 5), the indigenous and the elderly (over 60) . Results show that the utilization of health services remained at about 50% during the four years studied, 1999-2002. Differences among socioeconomic groups are noteworthy Nationwide, the average health utilization rate among the poorest population barely reached 30 percent, compared with about 70 percent for the wealthier groups. In addition, although the rate of self-reported illness remained constant between 1999 and 2002, utilization of health care services decreased over this period in particular among the second and third quintiles. On average about 4 percent of households faced catastrophic health payments from 1999 to 2002, with significant variations across socio economic groups. Although the poor used fewer health services, they faced a greater incidence of catastrophic health expenditures than the non-poor. Catastrophic expenditures actually dropped from 1999 to 2002, possibly as a result of the decrease in the use of health services. Results suggest that it is the poor and the senior population who are more likely to face catastrophic expenditures. Those households with children under five years were more likely to access health services but less likely to face catastrophic health expenditures suggesting that health insurance coverage and social programs focusing on diarrhoea and acute respiratory illness (ARI) for children under 5 had a protective effect. However, the gap between poor and non-poor in the use of health services and catastrophic payments was not reduced during this period.

5

Introduction Every year, approximately 44 million households face catastrophic expenditures, and more than 100 million individuals are being impoverished throughout the world by paying for health services (WHO, 2005). In the absence of effective protection mechanisms, the most vulnerable groups in society face continuing risks of both financial hardship and ill-health. In order to improve population health and reduce inequalities it is essential to guarantee access to health services across the population. There are many barriers undermining access to health care in developing countries and it is generally the poor population who are most affected by these. Financial barriers can cause people to postpone or forgo medical attention or to become impoverished due to health payments because of a lack of financial resources. (Xu, Klavus, Kawabata, Evans, Hanvoravongchai, Ortiz de Iturbide et al. 2003; Xu, Evans, Kawabata, Zeramdini, Klavus & Murray, 2003a; Xu, Evans, Carrin & AguilarRivera, 2005). Bolivia is one of the poorest and least developed Latin American countries and since 1996 has been going through a period of health system reform. A decentralization process has taken place and different strategies aimed at reducing infant and maternal mortality rates have been implemented. Between 1997 and 2002 both the Seguro Nacional de Maternidad y Niñez (SNMN) and the Seguro Básico de Salud (SBS) 12 were introduced into the country. Both schemes aimed to cover mainly women of childbearing age and children under the age of five but the SBS also included interventions for sexually transmitted infections, malaria, cholera and TB for the whole population. This paper explores the trend (between 1999 and 2002) in health service utilization, out-of pocket payments, catastrophic expenditure and the socio-economic factors associated with them, particularly for the poor and non-poor, the young (under 5), the indigenous and the elderly (over 60). The study also considers whether the main programs introduced during this period had any effect on health care utilization and out-of pocket health payments, in particular for those groups of the population often excluded from insurance programs such as children over the age of five and the elderly. Section 1 of this paper provides background information on the Bolivian health system and its most recent reforms. Section two, describes the research questions, data and the methodology utilized for this study. Section three examines survey findings relating to health services utilization, out-of-pocket health payments, and the incidence and characteristics of catastrophic health payments. Section 4 discusses the results in the context of the reorganization of the health care system and outlines possible future research.

1

National Maternity and Child Insurance (SNMN) and Basic Health Insurance (SBS) respectively. The current health program, el Seguro Universal Materno Infantil (SUMI), launched in January 2003, was excluded from the analysis as the most recent household data available are from 2002.

2

6

I Background 1.1 Key morbidity and mortality data With at least 60 percent of its population, approximately 5.3 million people living in poverty, Bolivia has the third highest infant mortality in the region (World Health Organization, 2005c). This country is fighting against inequality and resource scarcity under difficult economic and political conditions (Government of Bolivia, 2001). Bolivia is characterized by its multiethnicity and its geographical dispersion which has hampered the provision of basic services such as education, health, drinking water, sanitation, and electricity to rural communities. Besides the lack of geographical access, financial, socioeconomic and cultural factors are the most important barriers to health services, the language barrier being one of the most significant (Landa Casazola, 2002). Bolivia is a "young" country with more than half of its population under the age of 25 and only about seven percent over 60. The total fertility rate remains high with a figure of 3.8 children per woman nationwide reaching 5 children in poor departments. The country has been marked by high infant and maternal mortality rates however and in 2003 the maternal mortality ratio was estimated as 420 deaths per 100,000 live births with women in the richest quintile making three times as many antenatal visits to a doctor than those in the poorest quintile (Oomman, Lule, Vazirani & Chhabra, 2003). The mortality rate among children under 5 years in 2003 reached 66 per 1,000, 2.5 times more than the average rate for the Latin-America region and unregistered child deaths are believed to be as high as 63 percent. Three-quarters of Bolivian infantile deaths were due to preventable diseases such as diarrhea, pneumonia or perinatal causes. These figures differ widely across Bolivia's administrative departments (Darras, 1998; Pan American Health Organization, 2002; World Health Organization, 2005b). Communicable diseases such as tuberculosis, malaria, yellow fever, and Chagas' disease are among the many significant public health problems in Bolivia (Pan American Health Organization, 2002). Socioeconomic and ethnic differences are preponderant. For instance, the indigenous population is relatively more exposed to communicable diseases. The incidence of tuberculosis is five to eight times higher than the national average, and cholera takes a particularly high toll among the Weenhayek (Mataco) and Guaraní communities. In addition, indigenous women suffer from poor reproductive and family planning conditions, characterized by early and multiple pregnancies among other factors (Pan American Health Organization, 2003). 1.2 The health system in Bolivia Most of the research regarding the health system in Bolivia has been focused on maternal and child health care (Bermudez, Salinas, Espada & Muñoz, 2001; Camacho, Schwab & Shaw, 2003; Vidal Fuertes, 2003). Evidence on household expenditures on health care is limited and only a few studies have explored the distribution of these payments by socioeconomic groups or have analysed the impact of out-of-pocket on health care utilization and poverty (Narvaez & Saric, 2004; Cardenas & Darrás, 1996).

7

The Bolivian health system is a mixed and segmented system organized in three subsectors (Pan American Health Organization, 2001): a) Public. The Ministry of Health (MoH) regulates and executes national health policies and strategies at central level. At regional level, the prefectures are responsible for human resources administration. At local level, the municipal governments are in charge of administering health facilities, aiming to provide health services to the low-income population. It is estimated that the coverage is still low, at around 30%; b) Social security. Salaried employees in the formal sector are eligible for social security. Nine cajas de seguridad social (social security funds) cover 27 percent of the population. The largest of these funds is the National Health Insurance, which provides eighty percent of the country’s social security coverage and for which the ultimate financing responsibility is borne by the Republic of Bolivia. c) Private sub-sector involving profit and non-for profit organizations. This sector provides services to 20 percent of the population financed mostly by households through out-of-pocket payments or prepayment private schemes, and private companies through private insurance. The role of the non-for profit organizations in the provision of health services is significant, in particular those administrated by the Catholic Church. They provide health care to 10% of the population mostly in marginal urban areas and extreme poor municipalities and are financed mostly by external funds. The number of traditional providers known as yatiri, curandero, callawaya, and naturista is significant, in particular in rural areas (Pan American Health Organization, 2003). In Bolivia social protection for health is still developing. Most of the labor force in urban areas belong to the informal sector. It is estimated that 80 percent of the population are outside the formal labor sector in the cities with this figure rising to 100 percent in rural areas. As a result, only around 15 percent of the population are covered by the social security scheme. An additional 3 percent are covered by private health insurance schemes (Pan American Health Organization, 2002). The economic crisis of the late 1990s led to a significant reduction in the number of people working on short term contracts and this period therefore saw a decline in the percentage of the population covered by social security. With regard to health system financing, total financial resources for health increased from five percent of the Gross Domestic Product (GDP) in 1998 to seven percent in 2002 (Box 1).On average, in that four year period, 22 percent of health expenditure was financed by the government, 38 percent by social security and 37 percent paid directly by households and non-governmental organizations (NGO's). The proportion of general government health expenditure in total health expenditure decreased while out-of-pocket health payments increased progressively from 27 percent in 1998 to 33 percent in 2002. In addition, around 7 percent of general government health expenditures came from external sources (World Health Organization, 2005c). Bolivia's annual per capita health spending in 1998 was $120 international dollars. This figure had risen to $179 by 2002 but even so only Honduras and Haiti ($int 83 and $int 154 respectively) presented lower health expenditure levels in the Latin-

8

American region than Bolivia. Private sector spending3 from 1998 to 2002 grew at a faster rate than general government sector spending. Whereas general government per capita health resources grew annually by 9 percent (from int$75 to int$107), out-of pocket per capita health expenses increased 15 percent per year, reaching $58 international dollars in 2002 (World Health Organization, 2005c).

Box 1: Health system financing indicators for Bolivia (1998-2002) Indicators Total population (1000) GDP per capita (int.$) Total health expenditure as a percentage of GDP (%) Total expenditure on health (THE) per capita (int.$) General government health expenditure* (excl.social insurance) % Social Security expenditure on health % THE Private health insurance premiums % THE Out-of-pocket payment % THE Non-profit institutions (NGO) % THE

1998 7,984 2,416 5.0 120 23.1 39.8 4.0 27.5 5.5

2000 8,317 2,468 6.1 150 22.8 37.3 3.2 32.6 4.1

2002 8,645 2,568 7.1 179 20.9 38.9 3.8 32.7 3.7

* Includes tax-based expenditure and external resources Source: WHR 2005, National Health Accounts data and WDI 2005.

The decentralization process is one of the most important characteristics of the current system. Since the Law of Popular Participation of 1994 was implemented, 20 percent of central governmental spending has been allocated on a capitation basis to municipalities through tributary co-participation. In 1996, the implementation of the Seguro Nacional Materno Infantil (SNMN) earmarked 3.2 percent of these coparticipation funds to this program (Bossert, 2000)and three years after, in 1999, the Seguro Basico de Salud (SBS) was partly financed by 6.4 percent of the same funds (Narvaez & Saric, 2004). Municipalities have a basic control on investment and operational matters for the health sector, while the MoH through departmental prefectures retained control over personnel and salaries. Moreover, municipalities have the responsibility to formulate social and economic development plans along with the ownerships or infrastructure and all revenue generated by the services (Pan American Health Organization., 2002; Bossert, Ruiz Mier, Escalante, Cardenas, Guisani, Capra et al. 2000). The lack of human, financial and physical resources is one the main problems of the health system, particularly in the rural areas. In 1999, Bolivia counted 3.2 physicians per 10,000 habitants at national level and 1.3 at rural level. These figures were 8 times and 20 times respectively less than the average for the Latin American region (19.8). (Narvaez & Saric, 2004; Pan American Health Organization., 2002). According to 2004 data from the Bolivia Health Information System, physical resources in the system comprise of 1140 health centers, 168 hospitals, 1360 health posts, and 25 specialized institutes (Ministerio de Salud y Deportes, 2005).

3

Includes private prepaid plans, household out-of-pocket payments and non-profit institutions (NGO) spending.

9

1.3 The health system reform As with many of the Latin American countries, Bolivia has carried out health system reforms attempting to improve people's health and provide better access to health care since the last decade (Pan American Health Organization, 2002). The Strategic Health Plan, 1997-2002 outlined the primary aim of ensuring universal access to health care based on community primary health, social insurance and decentralization of resources. The main purpose of the reform was to reduce the severe infant and maternal mortality rate and eliminate the financial barrier of access to health care. These goals are coherent with the governmental commitment to work towards the achievement of the Millennium Development Goals (MDG) at national and local level and to fulfill the World Bank's Poverty Reduction Strategy (PRS) and HIPC objectives. (AIS, OPS & WEMOS, 2003; Government of Bolivia, 2001). In this respect, there have been significant improvements, such as a steady increase in immunization coverage and skilled birth attendance. However, the improvement is concentrated in urban areas and among the better off groups, the poor-rich gap is still considerably large. (Pan American Health Organization, 2003).

Since the mid-1990's there have been three health programs intended mainly for the very poor, and specifically for women of childbearing age and children under five years; the Seguro Nacional de Maternidad y Niñez (SNMN) was implemented in 1997 and consisted of 26 interventions. This program provided free pre- and peri-natal attention at public and participating non-governmental facilities and was co-financed by the central and municipal governments, the latter providing 3 percent of the above mentioned co-participation funds for the purchase of necessary medication and materials (Bossert et al., 2000; Dmytraczenko, 2000). Two years after the Seguro Básico de Salud (SBS)4 was put into operation the number of interventions was extended to 92, including treatment for post abortion care, complications of the newborn, sexually transmitted infections, and malaria, cholera and TB. The program also included regular visits by health professionals to rural communities without health facilities and it extended coverage beyond the public sector to social security, NGOs and church-dependent facilities. Providers are reimbursed on a fee for services basis. The SBS was financed by the national government (33%) covering salaries and equipment, international cooperation (10%) and 6.4 percent of municipalities co-participation funds assigned to cover drugs, supplies, hospitalization and laboratory exams, (Seoane, Equiluz, Ugalde & Arraya, 2003; Böhrt & Holst, 2000; Pan American Health Organization, 2002). Four years later, in 2003, a more comprehensive scheme was implemented. The Seguro Universal Materno Infantil Maternal (SUMI)5 comprised 400 interventions and in that year the infantile mortality rate decreased to 64 per 1000 births. This recent scheme covers antenatal care, labor and delivery (including caesarean sections and emergency care), postnatal care and newborn care. It is funded by central tax revenues, earmarked municipality resources (up to 10% of co-participation funds), and funds from the National Redistribution Fund (FSN) (financed gradually with up 4 5

Based on the Governmental Decree No. 25265. (Decreto Supremo No. 25265) Based on the Governmental Decree No. 2426. (November, 2002)

10

to 10 percent of the Special National Dialogue Account of the HIPC debt-relief initiative).(Bermudez et al., 2001; World Health Organization, 2005c) The evaluation of the programs (SNMN and SBS) show that regardless of the increase in the utilization of health services for child and maternal care(Dmytraczenko, 2000; Böhrt & Holst, 2000) the differences between rich and poor across municipalities are still considerable. The improvements are concentrated in urban areas and among better off groups. (Vidal Fuertes, 2003; Pan American Health Organization, 2001). Among some of the drawbacks of the Seguro Básico de Salud (SBS) was the limited promotion of the program in rural areas (which meant that people did not know that they were entitled to program services); the delay in the payment of providers by municipalities; the reduced benefit package, and the frequent drug stock-outs causing an increase in out-of-pocket health payments so as to cover the cost of medicines supplied in the private sector (Narvaez & Saric, 2004) . An additional program is the SMGV (Seguro Médico Gratuito de Vejez) aimed at those over the age of 60. This scheme aims to provide free medical services to the elderly. The services are offered through the social security network leading to the exclusion of those living in rural areas who are, in fact, the most in need. The elderly represent half of the population in the rural communities since the majority of young people have migrated to the cities. The waiting lists and the quality of the services associated with this program have deterred people from using it. For those living in the rural areas the main barrier is having to travel to the city, sometimes without even receiving the service (Narvaez & Saric, 2004).

II Research questions, data and methods 2.1 Research questions One of the main challenges for Bolivia's health system is to improve health services access and protect the population from the negative effect of out-of-pocket health payments. This paper aim to answer the following questions; a) Who uses health services and where do they go?; b) Who pays how much and for what kinds of health services?; c) How do these payments affect a household’s financial situation? and d) Which are the determinants of households catastrophic health payments. Based on the existence of the two main programs focused on infant and maternal health we would expect that a household with children under five years is more likely to use health services (mostly public facilities) and less likely to face catastrophic expenditures than a household with children under 14 years or with a senior member. We also believe that health insurance coverage (employment-based and private) is an important factor in enhancing the probability of using health services and protecting people from excessive health expenses. We consider that due to the lack of an extensive and comprehensive program focused on the elderly population, this group would be specially exposed to the negative effect of out-of-pocket payments. Finally, since no specific policy has been implemented to regulate the effect of the user charges at public facilities with the exception of the child and maternal schemes we expect that households could face catastrophic health expenditures not only by using private facilities but also using public ones.

11

2.2 Data The analysis uses data from the Bolivia Living Conditions Survey (Encuesta Continua de Hogares, Condiciones de Vida). This survey is part of the IDB-World Bank regional program to improve the household surveys to measure living conditions and poverty in Latin America (MECOVI). The dataset is representative at national, regional and urban-rural levels. It contains a comprehensive individual socioeconomic section, including one on health status and service utilization. In addition, the household questionnaire covers expenditure patterns including health care expenditure. For the purpose of this study, four survey years were included (1999, 2000, 2001 and 2002). The MECOVI average survey sample over the years analysed comprised 5,000 households and 21,000 individuals. All variables related to expenditures utilized in the analysis were transformed to a monthly value if they were reported weekly such as food expenditures, or quarterly or yearly for items such as education and health. Data was collected in the national local currency, Bolivianos, at nominal value. The expenditure section includes information on medicines and inpatient and outpatient services during the last 12 months. Out-ofpocket refers to payments made by households at the point of receiving health services. Typically, these include doctors’ consultation fees, purchases of medication and hospital bills. The constructed variable "out-of-pocket expenditure for health" (OOP) included in the analysis consisted of payments for outpatient services, diarrhea and respiratory illness, and drugs expenditure in the previous month as reported in the health section plus the household inpatient payments collected in the expenditure section adjusted to a monthly basis. It was considered that since inpatient episodes are usually occasional, a one-year recall period for hospitalization services would capture more cases than a one-month period.

The health section included questions on self-reported illness for the overall population, and diarrhea and acute respiratory sickness for those under 5 years. Those who reported either illness or injury during the previous four weeks were asked whether they sought medical attention and asked for details of the provider and the cost. Accordingly, the variable on health service utilization falls into four categories: public, private self treatment and non use of services a) Users of public health facilities included those people who visited state hospitals, health centres or health posts where the service was provided by a physician, nurse or health promoter; b) Users of private health facilities included private hospitals, clinics, or consultations with a doctor or nurse at the patient's home. Traditional healers were also included in this group since the number of observations were very small; c) Self treatment included purchasing the medicines directly from a pharmacy. In the case of children with diarrhea and ARI, it comprised those children who did not seek medical care but received non-prescription treatments such as oral rehydration salts (ORS), oral rehydration solutions or home based solutions from a pharmacy or from a

12

member of the family. Since it was not possible to know whether the person at the pharmacy was qualified to prescribe or not, we decided to include them in this category. d) Non-users included those reporting sickness but not seeking medical attention.

2.3 Methods 2.3.1 Determinants of health utilization - Multinomial Logit analysis In order to investigate differences across socio-economic groups in the use of public, private or self-treatment health care we estimated a multinominal logit model using an independently pooled cross- section data for 3 years. The subsample for the year 2001 was excluded since it does not contain health insurance information. This model is used for analysing unordered categorical response variable (Powers & Xie, 1999). The sample comprises 10,075 individuals. The dependant variable assesses whether an individual who reported illness during the previous four weeks sought medical attention or not. It includes four possible answers. y= 0= no use, 1 =use of public facility, 2 =use of private facility, 3 =self-treatment. The regressors include age, gender, health insurance, region (urban-rural), year, education and income level. Table 1 shows a summary of the variables used in the analysis. The model assumes that there exists independency among alternatives and probabilities relative to a baseline group. The probability of an individual with characteristics xi of falling into J particular categories with j=1 as reference group is given by (Powers & Xie, 1999): Pr( yi = j | xi ) = Pij ==

(

exp xi' β j

(

)

1 + ∑ j = 2 exp x β j J

' i

)

'

, for j > 1

and Pr( yi = 1 | xi ) = Pi1 ==

1

(

1 + ∑ j = 2 exp x β j J

' i

)

'

For the baseline category

2.3.2 Estimation of catastrophic health payments and impoverishment Low-income governments face the challenge of avoiding the regressive burden of out-of-pocket payments while at the same time assuring equal access to health care services. Reducing out-of-pocket payments is critical to achieving equal access. (World Health Organization, 2005a; World Health Organization, 2000).Different approaches have been developed to measure the effect of excessive out-of-pocket payments on households. (Wagstaff & van Doorslaer, 2003; Russell, 1996; Xu, Evans, Kawabata, Zeramdini, Klavus & Murray, 2003b) Some studies have analysed the situation in developing countries such as Burkina Faso, Mexico, Thailand and Vietnam (Sesma-Vazquez, Perez-Rico, Sosa-Manzano & Gomez-Dantes, 2005; Su,

13

Kouyate & Flessa, 2006; Ensor & San, 1996; Pannarunothai & Mills, 1997; Knaul, Arreola-Ornelas, Mendez & Martinez, 2005).These studies have concluded that socioeconomic determinants are the key factors in explaining the negative effect of out-of-pocket payments in low income countries. This study is based on the WHO's methodology on Distribution of Health Payments and Catastrophic Expenditures (Xu, 2005). The purpose of this methodology is to identify the extent to which households are being affected by health expenses. The consequences of out-of-pocket health expenditures ranged from "not going to health services because it is not affordable" to "being impoverished because of the financial burden associated with health expenditures". It is also possible that households are forced to reduce their basic needs expenditure in order to cope with health payments. This methodology is based on the estimation of out-of-pocket health payments share of household capacity to pay (oopctp). Household capacity to pay or non-subsistence expenditure (ctp) is based on total household expenditure minus subsistence expenditure (se), and adjusted by a scale factor6. However, some households may report food expenditure that is lower than subsistence spending (seh>foodh)7. In this particular case the non-food expenditure is used as non-subsistence spending. ctp h = exp h − seh

if se h foodh

Household subsistence expenditure (se) is defined as the minimum consumption required for subsistence and it is estimated as the average food expenditure of those household that allocate between 45 percent and 55 percent of their expenditure to food. Household expenditure is used instead of an income variable in order to resolve for the measurement error of the variable of income that previous studies have found underreported, subject to more variance and therefore, less reliable than household consumption (Murray et al., 2003; Deaton, 1992). The following are the main variables used to identify the effect of out-of-pocket payments on households: Catastrophic health payments— those expenditures that involve a significant burden on a household's well being — as those equal to or greater than 40 percent of the household's capacity to pay. The variable on catastrophic health expenditure is constructed as a dummy variable with value 1 indicating that a household faces catastrophic expenditure, and 0 otherwise.

6

This adjustment considers the effect of an extra member in the household, but it does not take into account whether the additional member is a child or an adult. The factor is estimated from a 59 country regression using household survey data. For further reference please see (Murray, Xu, Klavus, Kawabata, Hanvoravongchai, Zeramdini et al. 2003) 7 This may indicate that reported food expenditure does not consider food subsidies, coupons, selfproduction and other non-cash means of food consumption.

14

cata h = 1 if

oop h

cata h = 0 if

oop h

ctp h ctp h

≥ 0.4 < 0.4

Impoverishment. A non-poor household is impoverished by health payments when it becomes poor after paying for health services. The variable is defined as 1 when household expenditure is equal to or higher than subsistence spending but is lower than subsistence spending net of out-of-pocket health payments, and 0 when both are equal to or higher than subsistence spending. impoorh = 1

impoorh = 0

if exp h ≥ seh and exp h − oop h < seh if exp h ≥ seh and exp h − oop h ≥ seh

2.3.3 Determinants of catastrophic health expenditures - Logistic Analysis For the analysis of determinants of catastrophic health payments we used the method of multiple logistic regression. Table 1B shows a summary of the variables used in the analysis. The dependant binary variable in this case has a value of one if household health expenditures are greater than or equal to 40 percent of its capacity to pay, and 0 otherwise. The following variables were included as potential explanatory factors: region (urban/rural), presence of senior member, presence of children under 14 years, household expenditure quintile, insurance coverage, sex and labor condition of household head, as well as private and public health services utilization . The analysis was done at household level using an independently pooled cross- section data for 3 years8 including 13,525 observations. The logistic transformation of the success probability p is given by (Powers & Xie, 1999): ⎛ p ⎞ Logit ( pi ) = log⎜⎜ i ⎟⎟ = βX ⎝ 1 − pi ⎠

8

Idem.

15

Table 1A. Summary of variables used in the health services analysis ( Individual level) Variable

Variable label

Note

Mean

Dependent variable (categorical 4 values)*

1999 2001 2001 2002

public private self treatment nothing

36.9 13.5 3.6 46.1

35.4 11.6 3.3 49.7

35.3 9.2 4.2 51.3

34.0 11.5 5.0 49.6

6.5

6.7

6.4

6.5

12.3

13.3

13.2

12.9

80.8

80.2

78.7

80.2

64.7

67.5

65.7

65.4

15.4

17.0

n.a

12.0

4.2

3.7

n.a

3.2

0.4

0.1

n.a

0.5

62.8

63.6

62.3

62.3

Use of public facility Use of private facility Self treatment Non use (base category)

Independent variables 1. yes 0. no 1. yes z 0.935 0.018 0.247 0.235 0.695 0.042 0.000 0.001 0.002 0.000 0.000 0.000 0.151 0.045 0.000

3.2 Determinants of catastrophic health payments In 1999, Bolivian households spent an average of 87 bolivianos per month ($15) on health services. This figure represents, on average, 4 percent of their total expenditure and 7 percent of their disposable income. Three years later the mean out-of-pocket health payments had decreased to 82 bolivianos ($7.2) and amounted to 3 percent and 5 percent of household monthly expenditure and household capacity to pay respectively. Results show that out-of-pocket payments (OOP) and out-of-pocket expenditure to total expenditure (oopexp) increased as household income level rose. A household in the richest quintile spent on average 30 times more than that spent by a household in the poorest quintile; likewise compared to families in the lowest quintile, households in the highest quintiles almost doubled the share allocated to health care from their total expenditure (oopexp) (Table 6). In 2002, households spent on average 5% percent of their capacity to pay on health care, down from 7 percent in 1999. In addition, the positive association of OOP as a percentage of capacity to pay with income level in 2002 indicates that out-of-pocket payments is a current practice across all income groups. For instance, while poorest households allocate 4.4% of their nonsubsistence expenditure to health, richer households did so up to 5.8%. Out-of-pocket payments on health care as a share of household capacity to pay ranged from less than 10 percent to over 40 percent. Whereas in 1999, 12 percent of households spent more than 20 percent of their capacity to pay on health expenditure, only 8.6 percent did so in 2002. Moreover, approximately 5 percent of households had catastrophic expenditures. By 2002, this proportion of households had fallen to 4 percent. Across expenditure quintiles, the proportion of households with catastrophic health payments varied from 2.9 % to 6.8% in 1999 and 2.6% to 4.9% in 2002 (Table 6). The poorest two quintiles showed the highest percentage of households with catastrophic health payments in 1999 (5.4% and 6.8% respectively). However, it was the highest income groups who faced the higher proportion of households with catastrophic payments in 2002 (4% and 5% respectively). Between 1999 and 2002 fewer households in the first two quintiles suffered from catastrophic expenses while the proportion of those households in higher income groups slightly changed. High out-of-pocket health payments had an impact on poverty in Bolivia. In 2002, more than 100,000 people were impoverished due to health payments, around 2 percent of the overall population. Poor households were the group most affected; seven percent of the households in the second poorest quintile were impoverished in 1999. By 2002, it declined to four percent. The fact that there was no presence of impoverishment in the first quintile was because this group was already below the poverty line.

21

Table 6. OOP as share of expenditure, OOP as share of capacity to pay, catastrophic expenditure and impoverishment, by quintile (1999-2002) Year

poorest

second

Quintile third

fourth

richest

total

oop%exp

1999 2002

2.5% 1.8%

4.0% 2.4%

3.2% 2.9%

3.9% 3.9%

5.5% 5.0%

3.8% 3.2%

oop%ctp

1999 2002

7.0% 4.4%

8.8% 5.6%

5.8% 5.2%

5.8% 5.7%

6.4% 5.8%

6.8% 5.3%

Catastrophic exp.

1999 2002

5.4% 3.6%

6.8% 3.7%

2.9% 2.6%

3.7% 4.0%

5.1% 4.9%

4.8% 3.7%

Impoverished

1999 2002

0.0% 0.0%

6.8% 4.5%

1.3% 1.0%

0.0% 0.1%

0.4% 0.4%

1.7% 1.2%

Source: Encuesta de Mejoramiento de condiciones de Vida 1999-2002 In terms of the structure of out-of-pocket payments, drug spending accounted for almost half of household monthly health expenditures. Diarrhoea and ARI consultations to children under 5 years was the second category with 31 percent, followed by outpatient services with 11 percent. Hospital charges make up 9 percent and other services such as ambulance, appliances and tests account for 4 percent of total OOP. The structure of health payments did vary across expenditure level. Drug spending plus diarrhoea and ARI expenses represents 80 percent of the out-of-pocket expenditures for the poor population across the four years whereas the non-poor population spent more on outpatient services and other items such as appliances and tests. Socio-economic characteristics of households that might influence catastrophic payments were explored using logit regressions. This method is used to examine the effect of any indicator, keeping the effects of all other indicators constant. Our interest in this econometric analysis is to identify those households at greater risk of bearing a financial catastrophe due to health services. The independent variables are: children, ageing, sex , region, indigenous, utilization public, private and selftreatment, health insurance and income quintile. Table 7 shows that the principal factors protecting households against catastrophic health payments were health insurance coverage, either public or private, being located in an urban area and the work status and education level of the household head. Being employed and having a higher level of education could be translated into more opportunities to cope with the financial burden such as borrowing money or selling assets. The results from the regression analysis indicated that households with a senior member, independently of their income level are more likely to face catastrophic expenditures. On the other hand, households with children under five years are less likely to face catastrophic expenditure. This could indicate that the public programs focused on childcare have had a positive impact. The findings show, however, that households with children between the ages of 5 and 14 years were at the same risk of facing catastrophic health expenditures as the overall population. People using either public or private health services had a considerably higher probability of catastrophic payments than those that opted for self-treatment or no care at all. Households using public facilities were at the same risk of facing catastrophic health payments as those visiting private facilities. 22

The analysis also found that households with high income were 60% more likely to face financial burden due to health care than the poorest households. This disparity could be explained by the fact that either they were able to finish the treatment or because poor people opt to seek lower quality private health services. The findings of the econometric analysis revealed that after controlling for the above mentioned factors the proportion of households with catastrophic health payments increased from 1999 to 2000 but then returned to the1999 level in 2002. Table 7 Logistic Regression on determinants of catastrophic health expenditures (1999-2002 ) Dependable Variable: Dummy (1= oopct>40%, 0=othercase) Covariables

OR

Coeff.

s.e

P>z

Socioeconomic member +60 years member