Analysing the performance of health systems in Asian countries: What Myanmar can learn from Bangladesh and Vietnam


 Background Advancements in medicine leads, among other things, to increasing life expectancy. However, at the same time, health care costs are increasing, and this may not be sustainable in the future. Governments and health care organizations need to implement efficiency measures in order to maximize health outcomes within available resources. This study aims to compare the technical efficiency of health systems in Asian countries, and to identify “efficient peers” for each “inefficient country”: in particular, for Myanmar. Methods A DEA variable returns to scale output-oriented model was used to evaluate technical efficiency in thirteen Asian countries. The input variables were current health expenditure per capita, the density of doctors, and the density of nurses and midwifery personnel. Two output variables, health adjusted life expectancy (HALE) and the infant mortality rate were (IMR) analysed separately. Myanmar may learn how to improve efficiency of its health care system through studying its efficient peers from DEA results. A review of relevant English language literature was used as a basis for informing a comparative analysis of the health systems of Myanmar and its efficient peers, Bangladesh and Vietnam. Results Among the thirteen Asian countries studied, 38.5% and 53.8% of countries were technically efficient when HALE and IMR were used as the measured output respectively. More countries were efficient at reducing IMR than increasing HALE. Myanmar is one of the most inefficient countries, and it should look at the health systems of its efficient peers, Bangladesh and Vietnam, to make its health system technically more efficient. Conclusions The results of this study suggested that countries with inefficient health systems can improve their health outcomes without increasing their health care resources. As DEA measures efficiency only, future studies should take into account equity to assess comprehensive health system performance.


Background
Advancements in medicine help people live longer. Life expectancy is increasing. According to data published by World Health Organization (WHO), life expectancy increased by 5.5 years between 2000 and 2016 globally (1). However, at the same time, health care is becoming more expensive. Between 2000 and 2016, health expenditure per capita rose from US$ 15 to 33 in low-income countries, from US$ 22 to 79 in lower middle-income countries, from US$ 103 to 455 in upper middle-income countries, and from US$ 2430 to 5180 in high-income countries (2). As such, increasing health care costs is a major challenge for almost every country in the world.
The significant amounts of health care spending are wasted worldwide. The WHO estimates that globally 20%-40% of total health spending, equivalent to a current monetary value around 1.5 trillion, is being wasted every year due to health system inefficiency (5). Inefficient health systems negatively impact on population health, which in turn affects productivity, education, and human welfare (3,4).
According to 2010 estimates from the Organization for Economic Cooperation and Development (OECD), life expectancy at birth would increase by more than two years without increasing health care spending, if the health care systems of all OECD countries performed at their maximum levels (6). These issues highlight the need for greater efficiency in health system.
In general, countries that spend more on health have higher life expectancies (7). For example, an average life expectancy at birth for high-income countries was 80.6 years whereas 62.9 years for lowincome countries in 2016 (8). However, countries that spend most on their health care do not necessarily have best health outcomes, best exemplified by the United States (7). Interestingly, countries with differing proportions of health care expenditure can have similar health outcomes. For instance, life expectancy in Uruguay and that in Paraguay are the same (75 year), but Uruguay's health expenditure per capita is four times higher than Paraguay's (6,9,10). Even though there are other determinants of health, there clearly is scope to improve the effectiveness of health care spending within countries (6).
Efficiency in the health sector means achieving maximum health within available resources or using minimum resources for the given level of outputs. This is "technical efficiency" in economic terms (5,11). Health care efficiency can be measured at micro level to compare performance in diagnostic and therapeutic procedures between departments and hospitals etc., (11,12) at meso-level to evaluate organizational efficiency (13), and at macro level to assess health systems between countries (3,14,15). Cross-country comparisons of health system performance can provide opportunities to identify whether countries' health systems are performing efficiently, and to take appropriate actions to improve efficiencies in under-performing countries (16).
In terms of health care spending, Asian countries spend much less than OECD countries. Current health expenditure (CHE), expressed as percentage of gross domestic product (GDP), for East Asia and the Pacific Region was 6.6% and that in the South Asia Region was 3.6% in 2016. In contrast, 2016 CHE for OECD countries was 12.6% (17). In addition, out-of-pocket payments (OOPs) as a percentage of CHE are very high in some Asian countries; OOPs were more than 70% of CHE in Myanmar and Bangladesh in 2016 (18). High OOPs are strongly correlated with catastrophic health expenditure and impoverishment and can exacerbate poverty (19,20). Moreover, South East Asian region countries are now facing a triple burden of diseases: communicable diseases, noncommunicable diseases, and injuries (21). All these factors demand efficient use of health care resources, especially in resource limited Asian countries, in order to improve health of their populations.
Some previous studies have analysed health system efficiency in OECD countries (3,14,22), but few have compared health system efficiency in Asian countries (15). To our knowledge, no studies have as yet undertaken comparative analysis of the health systems of inefficient Asian countries with the purpose of benchmarking so that inefficient countries might adopt best practices with regard to improving efficiency.
This study includes ten members of the Association of Southeast Asian Nations (ASEAN) (Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, the Philippines, Singapore, Thailand, and Vietnam) together with three neighbouring countries of Myanmar (Bangladesh, China, and India). The aim is to compare the technical efficiency of health systems in Asian countries and to identify "efficient peers" for each of the "inefficient" countries.

Methods
Data Envelopment Analysis (DEA) DEA is one of the most widely used methods to measure the technical efficiency of operating units (called decision making units, DMU). It is a non-parametric, linear programming technique that measures the relative technical efficiency of several decision-making units by calculating the ratio of the weighted sum of outputs to the weighted sum of inputs. The efficiency frontier is plotted based on the combinations of inputs and outputs of the best performing DMUs and the efficiency of each other unit can be measured by comparing it with an identified frontier of efficiency. The efficiency ratio ranges from zero to one, with the efficiency frontier being scored 1 (100%) and the inefficient units falling below the frontier scored less than 1 (0-99%). DMUs located on the frontier are known as "benchmarks or peers". These are the best practice examples which allow comparison with less efficient DMUs. DEA is also considered as a powerful benchmarking technique (23).
The DEA method was originally introduced by Charnes, Cooper, and Rhodes in 1978 assuming constant returns to scale. It should be used when all DMUs are operating at an optimal scale. In 1984, Banker, Charnes, and Cooper developed a second DEA model assuming variable returns to scale. This is appropriate when DMUs are not operating at an optimal scale, i.e. DMUs facing imperfect competition, government regulations, etc. Increasing returns to scale (or economies of scale) is a situation where changes in input of 1% results in greater than 1% changes in output. Conversely, decreasing returns to scale (or diseconomies of scale) refers to a situation where changes in input of 1% results in less than 1% changes in output. The DEA model can be input or output oriented. An output-oriented DEA model maximizes output for a given level of input, whereas, an input-oriented DEA minimizes input for a given level of output (23).
This study uses variable returns to scale, an output-oriented model. Variable returns to scale are used because the health care systems in the studied Asian countries (DMUs) differ in terms of per capita incomes, health financing policies and socio-economic characteristics. An output-oriented model is used as most of the health system objectives refer to maximizing health outcomes by using available resources as efficient as possible. Moreover, inputs such as health expenditure and health care professionals are already assigned, so that it would be inappropriate to advise to reduce health care inputs by using an input-oriented model.

Data and Variables
Data were collected for the year 2016 from WHO Global Health Expenditure Database and WHO Global Health Observatory Data Repository (24)(25)(26)(27)(28). Where 2016 data were not available, the closest earlier data were used. WHO data are comparable across countries which is crucial for DEA. The WHO uses the System of Health Accounts (SHA 2011) framework to provide comparable national health expenditure statistics internationally (29,30).
For the analysis, it is important that the selected health resource inputs have significant impact on the selected outputs. Three selected input variables are current health expenditure per capita in international dollars (Int$), density of doctors, and density of nursing and midwifery personnel for thirteen Asian countries (18,26,27). The output variables, HALE and IMR were used in analysis 1 and 2 respectively (25,31). The IMR was converted to the infant survival rate (ISR) for consistency with the DEA assumption that more output is better. The ISR is 1-IMR/1000 (32). The selection of three inputs and one output meets the DEA model requirement in that the number of decision-making units or countries (n = 13) exceeds three times the number of input and output variables (23).

Comparative Analysis
Information on the health systems in Myanmar and its benchmark countries, Bangladesh and Vietnam, was extracted from the published scientific literature in online databases, PubMed and Google Scholar etc. Search terms are "health system in Bangladesh", "health system in Myanmar", "health system in Vietnam" and "health system changes in Asia".

DEA Results
The descriptive statistics of the input and output variables for thirteen Asian countries are shown in     The country should look at best practice of peers associated with the highest weight value so that the role models for Myanmar will be Bangladesh and Vietnam for improvement in both HALE and ISR.

Comparative analysis of health systems of Myanmar, Bangladesh and Vietnam
Myanmar should learn successful strategies of health systems from its benchmark countries, Bangladesh and Vietnam, to improve efficiency. "A system framework for analysing the efficiency of health care resource use" indicates that health system efficiency can be influenced by a set of policy instruments: financing, provider payment methods, organization and regulation (10). The policy makers should target these instruments to reform their health system to be more efficient. In the following sections, the health systems of three countries are compared and analysed in terms of policy instruments. However, provider payment methods are not used owing to a lack of comparable data.  (38). The private, for profit, sector has expanded rapidly especially in major cities and it is estimated to provide 75%-80% of ambulatory care (34). Vietnam's Ministry of Health is responsible for setting rules and regulations in health system and carried out health care provision under direction of health care activities (DOHA) (37). In 2014, the Health Insurance Law was revised, aiming to cover 100% of population with health insurance. The

International aid agencies used to have difficulty in engaging Myanmar
Pharmaceutical Law was revised in 2015 ensuring access to affordable quality essential medicines (43).

Discussions of DEA results, strengths and weakness
Out of thirteen countries, five (38.5%) were technically efficient for the output HALE, while seven (53.8%) were technically efficient for the output IMR. The inefficient countries could improve HALE and IMR by 6.88% and 0.92% respectively (compared to efficient countries in this study) without using more resources. The findings suggest that more countries are efficient in reducing IMR than increasing HALE so that inefficient countries have more room for improvement with regard to HALE rather than IMR.
The findings from a previous study of OECD countries were mixed (14). Similar to this study, more countries were efficient for IMR; 19 (70.4%) out of 27 countries were efficient for IMR whereas 13 (48.1%) were efficient for life expectancy (LE). However, inefficient OECD countries on average could increase their LE by only 2.1% but could reduce IMR by 14.5% with existing resources so that inefficient countries had more room for improvement in IMR than in HALE. A possible reason why Asian countries are more efficient at reducing IMR is that there may be more policy emphasis on maternal and child health given that the IMR was one of the important targets under the Millennium Development Goals and is a current target under the Sustainable Development Goals (44,45).
The DEA's identification of peers provides an important and useful policy information for inefficient countries to reform their health systems. The selected sample in this paper included Asian countries only. Geographic proximity makes it easier for country to relate within their region. As HALE and IMR were analyzed separately, an inefficient country could focus more on peer countries with higher reference weights, depending upon the area of improvement that is being targeted.
Although DEA is very useful, it has its own limitations. Bangladesh has been recognized as an example of "good health at low cost" and praised for exceptional health achievements (46).
First thing that Myanmar should learn from Bangladesh is its extensive community-based health service delivery system. It reached almost all the households in rural areas and achieved high coverage. For decades, both government and NGOs use large-scale community-based health workers to address the shortage of human resources in health sectors. One successful TB community program (more than 90% cure rate) was already adopted by South Africa for treatment of TB and HIV (47).
Another thing to learn is Bangladesh government's willingness to create an environment for pluralistic health system. Government of Bangladesh has partnered with NGOs, private sectors, informal providers and international donors etc., to address limitation of health care resources in government in order to improve health outcomes of population (47). In 2014, the Health Insurance Law was revised aiming to cover entire population with health insurance (43). Health insurance coverage had reached 77% of population in 2015 (37).
Myanmar health care system is still weak in above mentioned practices; therefore, it will be very useful for Myanmar to adopt these approaches in order to reform its health system. These examples show how non-health approaches such as economic development, education and women's empowerment as well as health system related strategies such as health insurance system, largescale community-based programs, creating an environment for donors play very important role in health system improvement.

Concluding Remarks
Health system efficiency is crucial to improve health outcomes of populations especially in the current situation of increasing health care costs resulting from aging populations and expensive medical care.
Knowing the efficiency of a country's health system provides the opportunity to analyse health systems and to take appropriate actions to improve performance. Our cross-country comparison of health system performance by using DEA provides empirical evidence of the technical efficiency of health systems in thirteen Asian countries. The findings suggest that countries with inefficient health systems can improve their health outcomes without increasing health care resources. Results also highlight that more countries in this group are performing better with regard to reducing IMR rather than increasing HALE. Moreover, DEA identifies most comparable benchmark countries for every inefficient country so that inefficient countries can potentially emulate policies and practices of their efficient peers. It will be very beneficial for Myanmar to learn useful practical approaches and policies of its peers, Bangladesh and Vietnam, from our comparative analysis of health systems of these countries.
However, efficiency alone is not enough to meet the needs of increasing demand especially in resource-constrained countries. The amount of resources governments spend on health care is also important to provide basic health care that their citizens need. One paper argues that government health expenditure should be more than 5% of GDP and more than US$86 per capita in the pursuit of universal health coverage (49). In addition, DEA cannot measure equity of health care. According to the economic theory, a situation where there is pareto efficiency; allocation of resources in which no individual can be made better off without making another individual worse off; there may be no equity (10). Some studies suggest that instead of health gain and health equity as two separate outcomes, combination of these two should be the one combined goal of the health system that we should try to achieve as efficient as possible (50). Another suggestion that is based on the phenomenon called relative income hypothesis is that income inequality is inversely related to health outcomes, meaning that individual living in a more equal income society will have better health status (51). Therefore, targeting at reducing income inequality together with improving health outcomes of a society efficiently will be a way to achieve both efficiency and equity. Future research should be aimed at developing methodologies to measure efficiency and equity together to assess the comprehensive health system performance.

Consent for publication
Not applicable

Availability of data and materials
The datasets used and analyzed during the current study are publicly available from World Health Organization (WHO) and World Bank.