China's rural health insurance in the international context

A growing number of developing countries are developing health insurance schemes that aim to protect households, particularly the poor, from financial catastrophe and impoverishment caused by unaffordable medical care. China, in common with other transitional economies, has experienced rapid economic and healthcare reforms over the past two decades. As a result of these reforms, the majority of households living in rural China face out of pocket charges for healthcare that are rising faster than their incomes. Prior to the reforms, over 90% of the rural Chinese population were covered by Co-operative Medical Schemes (CMS), which collapsed in the mid 1980s. In recent years the Chinese government has been reinstating health insurance, now known as the New Rural Co-operative Medical Scheme (NCMS).

There is a wide diversity of health insurance schemes in developing countries, ranging from 'community-based' schemes, which are often voluntary, mainly funded by the households and local communities, with limited government support, to national schemes with substantial government support, which aim towards comprehensive coverage, as is the case in China, although participation in the scheme is voluntary [1–3]. In theory, the primary purpose of any insurance scheme is to share risk between individuals and hence extend financial protection to members of the scheme [4]. The central stated goal of developing the rural health insurance scheme in China is to provide financial protection to individuals and households from catastrophic expenditures due to major illness [5].

In principle financial protection is achieved through pooling the contributions of all members and pooling risks to redistribute them horizontally between healthy and sick and vertically between richer and poorer members of the scheme. However, lack of cross-subsidisation across communities and inadequate cross-subsidisation within communities due to low coverage levels and adverse selection have been common reasons for scheme failure [6, 7]. Low coverage has resulted from a range of factors, including low capacity to pay premium contributions [1]. In all countries with (nearly) universal coverage, health care or insurance contributions of the poor are subsidized or directly paid by the government [3].

The NCMS has made rapid progress in increasing coverage to high levels since its inception in 2004. Recent estimates of the national average enrolment level have ranged from 82.7% [8] and 85.7% [9]. A high level of political commitment has been an important factor enabling this remarkable achievement [9]. Subsidies from national and local governments have enabled a relatively low individual premium level, which contributes to the high coverage rate. Premiums for the poorest are paid directly through by the government through the Medical Financial Assistance Scheme for the Poor [10].

The coverage of catastrophic risk versus primary healthcare services is a major dilemma facing many schemes. Given limited funding and the goal of protecting households from medical impoverishment, the majority of schemes focus on providing insurance against "catastrophic" inpatient expenses, with the rationale that most households can afford the expenses incurred by minor illnesses [11]. However, there is increasing empirical evidence that household spending on ambulatory services and drugs, rather than hospitalization, is the primary contributor to medical impoverishment [11–15]. The under-utilization of primary care, and higher costs related to the irrational utilization of hospital facilities for illnesses treatable at lower levels of the health system has also been recognised as a factor influencing the failure of some schemes [12]. Yip and Hisao [12] argue that coverage of primary care and outpatient services provides incentives for patients to use these levels of care, rather than hospital services.

Less than optimal mechanisms for provider payment (such as fee-for-service systems) resulting in supplier-induced demand have posed significant challenges to many HI schemes [1, 16–18]. Rapid increases in health expenditure in China in recent years have been linked to the fee-for-service payment modality and the need to control the high cost of healthcare has been an increasing concern in NCMS implementation [18–20].

Non-communicable disease burden in China

Chronic diseases account for an estimated 80% of total deaths and 70% of total disability-adjusted life-years (DALYs) lost in China [21]. According to the third National Health Services Survey in 2003, the morbidity rate of chronic diseases in the rural areas was 10.5%, and the most common diseases were: hypertension, gastroenteritis, rheumatoid Arthritis, chronic obstructive pulmonary disease (COPD), cerebrovascular disease, inter-vertebral disease disorders and peptic ulcers [22] (Ministry of Health 2004). Chronic diseases have become the leading cause of death in both urban and rural areas, with mortality rates of 85.3% and 79.5% respectively [23].

In 2003, the estimated economic burden of chronic diseases in China accounted for 71.5% of the total economic burden of all diseases and 7.3% of GDP [24]. The five diseases with the highest economic burden were malignancy, cerebrovascular disease, hypertension, coronary heart disease, and other types of heart diseases, accounting for 39.6% of the total economic burden of all chronic non-communicable diseases and 28.2% of the total economic burden of all diseases (including the direct health care expense and indirect expense that is the foregone income due to illness) [24].

An estimated 54% of total medical expenditure (by households) on chronic diseases was spent on outpatient services, rising to 67% for hypertension and 74% for rheumatoid arthritis [24]. Average inpatient expenditure on common chronic diseases, such as hypertension and diabetes, accounts for 1.5 times the annual per capita income of rural residents [25]. One study in Shandong and Gansu provinces estimated the annual per capita outpatient expenditure on chronic diseases at half of total annual medical expenditure [26]. The available evidence therefore suggests that chronic diseases pose a heavy financial burden for rural residents, and that outpatient expenditure is a significant part of this burden.

NCMS development

The New Cooperative Medical Scheme (NCMS) was introduced on a pilot basis in 2003 and had expanded to 86% of China's rural counties by 2007, covering about 730 million rural people [26–29]. A key characteristic of NCMS is the increasing level of subsidies to the scheme from the Chinese government, with the aim of making the scheme fairer and financially more attractive to low-risk households. However, the 50 Yuan minimum level of financing per beneficiary represents only around one fifth of average per capita total health spending in rural areas [9] (at the time of data collection in May 2006, 1 US$ was equal to 8 Yuan). The new scheme operates at the county level, where there is considerable discretion in the design, including the benefit package, payment methods and reimbursement level.

As a result of the limited financial envelope, many services, and particularly outpatient care, are not covered or covered only partially, deductibles are high, ceilings are low, and coinsurance rates are high. There is, however, considerable heterogeneity in the benefit package across counties and coverage modes. One study of the 189 NCMS pilot counties in 2005, found that all counties covered inpatient care, only a quarter of counties covered outpatient expenses on a pooling basis, the rest did not cover them at all (10% of counties), covered only catastrophic expenses (10% of counties), or covered them through a so-called 'family account', which is not a pooled fund [9]. The bulk of reimbursement by NCMS is for inpatient expenses, even in counties where outpatient expenses are covered, and there are no special arrangements for reimbursing medical expenses resulting from the chronic diseases.

Since the establishment of NCMS in 2003, no study to our knowledge has investigated how far the new scheme contributes to alleviating the financial burden of chronic diseases. This paper aims to contribute to filling this gap by evaluating how far NCMS offers financial protection to rural residents with chronic diseases, particularly the poor, in Shandong Province and Ningxia Hui Nationality Autonomous Region, through an analysis of household survey data on healthcare utilization, costs and reimbursement by NCMS. Specifically the paper aims to answer the following research questions: 1) What is the financial burden of chronic disease in the study areas, especially for the poor? 2) How far does NCMS offer financial protection against chronic disease burdens, especially for the poor?

NCMS in Shandong and Ningxia

This study was conducted in Shandong province and Ningxia Hui Nationality Autonomous Region. Shandong Province, which is located in the Eastern coastal area, represents the more economically developed regions of China, with a GDP per capita of 23,546 Yuan in 2006 (about 3139.5 US$). Ningxia Hui Nationality Autonomous Region, a relatively poor area in north western China, represents the less developed inland regions. The region has a GDP per capita of 11,784 Yuan (about 1571.2 US$), and 36% of its population are minority ethnic groups.

Shandong Province was included in the NCMS pilot of 2003, and all 134 counties are now covered by the NCMS, with about 59 million rural residents enrolled in the new scheme, an enrollment rate of 90.08%. The financing level per beneficiary varied in different counties, with 81% of 134 counties reaching 50-55 Yuan per beneficiary in 2007, and other counties exceeding this level, with the highest at 100 Yuan per beneficiary (maintaining the 10 Yuan contribution per rural resident). The benefit package varied across counties, but generally inpatient services were covered by risk pooling in all counties. Some counties reimbursed outpatient expenditure through risk pooling, and others set up so-called 'family accounts', whereby a fixed amount of the per capita financing is allocated to each household to cover outpatient expenditure; in 2005 this was 8 Yuan out of the 10 Yuan premium per person. This is a pre-payment rather than a risk pooling mechanism, but households can pool their allocations and accounts 'roll over' from year to year so funds can be accumulated. Once the fixed sum has been exhausted there is no further reimbursement for outpatient services available. However, a few counties reimbursed outpatient medical expenditure for some chronic diseases through a risk pooling mechanism. Meanwhile a combination of deductibles, co-payment, ceilings, an essential medicine list and an essential medical services list were used to control the health expenditure. Reimbursement is very complicated, with variation in rates according to the level of health facilities, the type of services (outpatient or inpatient) and the level of medical expenditure. Information provided by the Shandong Provincial Department for Health showed that in general, the reimbursement rate for outpatient services ranged from 10-20%, and inpatient services from 10-80%. In October 2007, about 48% of enrollers were reimbursed by NCMS, and the actual reimbursement rate of outpatient and inpatient expenditure was 27.99% and 25.07% respectively. The average reimbursement of inpatient medical expenditure was 616 Yuan.

Ningxia was also included in the NCMS pilot of 2003. By 2007, all 18 counties were covered by the scheme, and the enrollment rate was 85.05%. At present the financing level per beneficiary was 50 Yuan, including 20 Yuan from the central government, 20 Yuan from the local government and 10 from individual members. The benefit package of NCMS has been uniform in Ningxia since 2006, when all the counties set up 'family accounts' for reimbursing outpatient expenditure and a risk pooling mechanism for reimbursing inpatient expenditure. According to the Ningxia Autonomous Regional Department for Health, for inpatient services, the reimbursement level varies according to the level of health institution and medical expenditure, ranging from 40-50% at the township health center, to 35-45% at county hospital and higher levels, based on the amount of medical expenditure. The real reimbursement rate at county hospital is currently about 35%, and 45% at township health center. Deductibles, co-payment, ceilings, an essential medicine list and an essential medical services list were also used to control the health expenditure.

Study design and sampling

The data used in this paper are derived from a cross-sectional household survey conducted in Shandong and Ningxia in 2006, as part of a project aiming to improve the design of rural health insurance in rural China and Vietnam. The study is entitled 'Bringing health care to the vulnerable - developing equitable and sustainable rural health insurance in China and Vietnam (RHINCAV)' and is funded by the European Community. For further details please see http://www.liv.ac.uk/lstm/research/rhincav/index.htm. The selection of the two provinces aimed to represent different levels of socio-economic development and geographic areas in China. Three counties were selected from Shandong (Zhangqiu, Changle, and Dong'e) and three from Ningxia (Zhongning, Yongning and Qingtongxia). The selection of counties was based on the following criteria: 1) all the counties have an existing New Rural Cooperative Medical Scheme (NCMS); 2) the County governments are willing and able to collaborate with the research project; and 3) the three counties should have different levels of socio-economic development and be located in a similar geographic area.

The household survey used a multistage sampling process. Three townships in each county and three villages in each township were selected using similar criteria to the selection of counties. In each village, a random sample of 100 households was selected based on the household registration list. The calculation of the sample size was based on the ability to compare the utilization rate for impatient services among 5 different income groups. A total of 6,147 households and 22,636 individual rural residents were sampled and investigated. Among them, 3944 individual patients responded that they had suffered from a (self identified) chronic disease in the previous year (May 2005 - April 2006). This paper presents an analysis of the sub-database of chronic disease patients.

Data collection and analysis

Structured questionnaires were used to conduct face to face interviews with people who reported suffering from one or more chronic diseases (in Chinese: man xing bing) to investigate: their basic social and economic characteristics, including income and expenditure levels and NCMS membership; and their health care utilization, associated healthcare costs and levels of reimbursement by NCMS where applicable. Master's students in Social Medicines were trained to conduct the questionnaire interviews. Data collected were for the past year - i.e. May 2005 to April 2006. Chronic disease was defined as an illness or symptom with at least 3 months continuous duration or intermittent presentation within the last 12 months. Interviewees were asked to give the name or main symptoms of each chronic disease they reported and, if relevant, where the disease was diagnosed (type and level of healthcare provider).

All the data were double entered in the database designed by Microsoft Office Access, and SPSS software was used to analyze the data. X² and T test were the main statistical methods used in the analysis for this paper.

To estimate the incidence of 'catastrophic' expenditure posed by chronic disease, we calculated the proportion of healthcare expenditure to household non-food expenditure. Estimates of expenditure were elicited from each household by asking the respondent to recall expenditure in response to a detailed list, encompassing production expenditure, consumption expenditure (including expenditure on: food, clothing and other daily living expenses; transportation and communication; housing, electricity, water and fuel; culture, education and entertainment; drugs, health care, and medical supplies) and other expenditures (e.g. paying back debt and interest, premiums, gifts, penalties and taxes). There is no international consensus on the threshold proportion of medical expenditure to household expenditure that constitutes 'catastrophic' expenditure. In previous studies, the threshold has varied from 5% to 40% of total household income [30–32]. We used a conservative threshold of 40% following Xu et al. [32].

Ethical approval

Ethical approval was obtained from the Research Ethics Committee at the Liverpool School of Tropical Medicine in UK and at School of Public Health of Fudan University in China (Approval No. IRB# 06-04-0061).

Limitations of the study methodology

The methodology used has several limitations for meeting our objectives, which are mainly due to the fact that the household survey was not designed solely to investigate the chronic disease burden. First, our estimations are based on self-reported chronic disease, rather than only those diagnosed by a physician. It is therefore possible that chronic illness and consequently its costs are either over or under-reported, due to the lack of a standard set of criteria used to distinguish chronic from acute illness. However, the question used to elicit self-reporting of chronic disease was drawn from the National Health Survey, which has been validated nationally. In addition, if we had included only diseases diagnosed by a physician, this could underestimate the disease burden given the barriers to accessing a diagnosis. Second, expenditure on healthcare for chronic disease was estimated by respondents over the year prior to the survey, which may have led to some recall bias or error. Third, it is difficult to assess the level of financial protection offered to NCMS members in comparison to non-members for several reasons. There are problems in interpreting our data on the difference between members and non-members in the proportion of medical expenditure to annual non-food expenditure in Shandong, because of skewed data due to very high expenditure in one case. There are further difficulties interpreting the data on the difference in levels of 'catastrophic expenditure' between members and non-members because of the confounding factors in terms of other differences between the two groups, such as income levels and chronic disease prevalence. Our paper has therefore focused on the degree of financial protection provided to NCMS members. Finally, it is not possible to definitively identify when costs are actually 'catastrophic' to households, since these are not always synonymous with high health-care costs [30]. However, even small costs for common illnesses can be financially disastrous for poor households [33]. Our analysis uses the term 'catastrophic' healthcare costs in line with norms in the international literature but we recognise that these costs are accurately termed 'potentially catastrophic'.

Sample characteristics and disease prevalence

Health care utilization by rural chronic disease patients in 2005-2006