China’s new cooperative medical scheme and equity in access to health care: evidence from a longitudinal household survey

Countries across the world are looking to health insurance as a means of ensuring access to health care and protecting patients from financial risks. Health insurance has the potential to lower financial barriers of access to health care, since the financial risk of health care is shared among insurance participants and health cost will be reduced at the point of health care use[1]. One common way to organise insurance is to target its funds to either a group of the population, such as the vulnerable/disadvantaged socioeconomic population, or specific services that are most cost-effective and/or preferentially benefit the target population, such as primary care or outpatient care[1].

In China, a focus on health care for the rural population is gaining increased governmental attention in recent years. The government targets its public funds for health insurance by focusing on the rural population through the New Rural Cooperative Health Scheme (NCMS). Since “equitable access” has been officially declared by the State Council to be the principal aim of the rural health insurance reform[2], the main objective of the NCMS is to provide universal coverage and to improve equity and access to health care to the rural population regardless of individual characteristics such as job status, education, pre-existing condition, and level of wealth. For the past few decades, the state and enterprise funded health insurance only covered well-off urban employees, leaving the majority of the rural residents unprotected from health risks[3], the launch of the NCMS in 2003 is considered as a crucial step in closing the insurance gap and reducing inequity in access to health care for the rural population.

However, the real world experience actually tells us little about how far public health insurance can improve access to health care[3‐7]. One major concern is whether the insurance is able to reach vulnerable/disadvantaged socioeconomic groups[3‐5]. Evidence from developing countries also suggest that public voluntary insurance programmes, especially the ones that require substantial premiums and patient cost-sharing, may have little effect on improving the use of public financed health services of the poor. In Iran, despite the decent development of a few government health insurance schemes targeting the poor and catastrophic inpatient care in the last decades, co-payments still count for 58% of the health expenditures, and the proportion of people facing catastrophic health payment remained high even after the insurance reform[8]. In India, the newly developed insurance - Rashtriya Swasthya Beema Yojna(RSBY) which aimed to target the poor only allowed a limited rate of reimbursement for inpatient care. Studies found that expenditures on drugs claims which constituted around 75% of OOP payments and 80% of the spending on outpatient visits were not covered, and the impacts of the RSBY on protecting the poor against health payment-induced impoverishment were limited[9].

The launch of the New Rural Cooperative Medical Scheme (NCMS) in 2003 represents a major step of the Chinese government to move towards a more equitable and efficient rural health financing system, but it is not clear that the is sufficient enough to deliver equitable access in different types of health care. One major concern is that, under the NCMS, health care is provided in public health facilities through a fee-for-service reimbursement, and the reimbursement rates vary by different types of care and at different health facilities. Although the NCMS has extended its coverage to outpatient care since 2007, its emphasis is mainly on “catastrophic outpatient cost", and reimbursement is made either through participants' Medical Saving Account or pooled funds which requires substantial cost-sharing[10]. Further, the scheme only reimburses drugs listed on the National Essential Drug Reimbursement List, services covered by the insurance package, and care sought at state-owned public health facilities. The claimed reimbursement rates are the highest for care delivered at village/township health centres and the lowest at city/provincial hospitals, while care sought at the high level health facilities is usually associated with severe illness and high medical expenditures[11, 12]. Consequently, as argued by many scholars, despite the broad coverage, co-payments for the NCMS participants in general remained high even after the insurance claims were made, and this may impede a subpopulation of the rural poor from seeking care[13‐16]. In terms of outpatient care, scholars argued that the NCMS increased the use of outpatient care among the poor at village clinics, whereas the increased use in inpatient care overall and at the higher level health facilities was concentrated disproportionately only among people who were rich[17]. Liu et al.[3] and Yu et al.[18] also found that the NCMS only increased the use of inpatient care for the better-off, whereas it had no significant impact on outpatient use.

While the previous work is important, the investigation on how the NCMS impacts inequity in health use is subjected to little updated empirical research, which is the setting for this paper. Previous studies either limit their investigations to a given point in time[15, 17, 19], or a specific health service[16]. The NCMS has been implemented for a few years, but it is still not clear if the scheme has any impact on utilisation. If it were, as the reimbursement rates of the NCMS are set at different levels for different services, it is worth investigating how much it may influence the utilisation for different socioeconomic groups, and whether variations in use is a generalised phenomenon, or are observed only for some services. For instance, little information is available on the characteristics of the users of preventive care and folk doctors, consequently, how the NCMS impacts the use.

Another issue that is largely neglected in previous studies, but is central to the measurements of equity in use is that studies tend to overlook the association between health use and health needs. Inequity in health use may be driven by factors as legitimate/fair inequity, such as health needs, and illegitimate/unfair inequity that arise from circumstances beyond individual’s control, such as per capita income. Policy may concern less with legitimate inequities because inequity arising from legitimate factors, e.g. health needs, is usually reasonable and acceptable. Therefore, a measure of socioeconomic related health inequality should control for legitimate differences, in this case, factors associated with health needs[20].

To shed light to above issues, this paper measures the extent to which the NCMS affects health care utilisation on the rural population in China, considering two types of formal health care (outpatient care and prevention care), and one type of informal health care (folk doctor care[1]). This paper first compares the magnitude of inequities in health use in 2004 (before the national rollout of NCMS) and 2009 (after the expansion of NCMS across the rural China). The Concentration Indices for utilisation, which compares the cumulative distribution of health use with the cumulative distribution of the population ranked by individual wealth, is used[21, 22]. It then investigates the determinants of patterns of health care use and the characteristics of the users for different services, taking into account the contribution of the NCMS to equity in health use. Data are drawn from China Health and Nutrition Survey 2004 and 2009.

The empirical findings derived from the study are expected to feed back into the policy making process. In particular, we are concerned that the expansion of the NCMS does not necessarily lead to equal access to care. Drawing from the literatures discussed above, this paper identifies situations in which at least one of them is expected to dominate. First, the launch of the NCMS is a means of improving the equitable access to formal care and discouraging the use of informal care/folk doctor care. Folk doctor care is not covered by the insurance scheme, whereas outpatient and inpatient care are included in the insurance package. Since the reimbursement rate set for formal care is relatively low, co-payments is likely to become one of the barriers to impede access to formal care among the poor. The NCMS may have positive impacts in reducing the use of informal care; however, the impact may be limited since unmet health need may still lead to increased use of informal care, which is less costly and widely accessible compared with formal care. Second, the NCMS may also exert some positive influence on use of preventive care, which historically required more cost-sharing and is now partially covered by the NCMS (e.g. general physical examination, blood pressure screening, and prenatal examination). As the co-payments are still high, preventive care use may still appear concentrated among the rich, but the level of inequity may become less pronounced.

In the subsequent sections, a background of the NCMS is provided; followed by methods, results, and a discussion of the policy implications.

Income-related inequity in health is estimated by pooled Probit Model and well-established methods based on the Concentration Indices[21, 28]. The method involves three basic steps: (1) estimate pooled Probit Models on the determinants of health use, and predict need (indirectly) standardized health for each health variable, and for each year separately; the computation of variance inflation factors (VIF) indicates that multicollinearity is not a problem. Ramsy RESET tests are performed, and results show the models have no specification problems; (2) calculate the Concentration Indices for actual use EI (the inequity driven by the actual health care utilisation), the horizontal equity indices HI (the inequity driven by socioeconomic factors); (3) decompose the socioeconomic factors that contribute to the inequities for the each year to see whether contributions have changed over time.

The multivariate regression models of health variables for step (1) and (2) above are central to the methods. The nature of the health use variables (binary variables) formally calls for a non-linear estimation. However, the disadvantage of this procedure is that the certain components of the equity analysis, such as decomposition analysis are difficult to implement and interpret. Studies have shown that equity measurements calculated by Linear Probability Model (LPM) do not differ importantly from the non-linear estimation[28, 29]. Therefore, the paper will use LPM instead of non-linear regression to standardize the health variables and to decompose the Concentration Indices. Results from Probit Model are nonetheless presented in the appendices for comparisons. Instead of using the Concentration Indices, Erreygers’s Concentration Index, which has been recently developed and has proved to be a better estimation for binary variable, will be used[30‐33].

The study reveals a mixed picture in terms of the variation of health care utilisation and how the NCMS has influenced the level of inequity. Although the data used in this study represent only two points in time, it covers the whole period of the expansion of the NCMS from 2004 (before the national rollout of NCMS) and 2009 (after the expansion of NCMS across the rural China). The study has yielded some interesting findings. First, the study found that the level of inequity remained the same for outpatient care. In terms of preventive care, a pro-rich inequity was observed both in 2004 and 2009, but the level of inequity had decreased. However, an enlarged gap between the poor and the rich in terms of folk doctor use was observed. Decomposition analysis showed that the NCMS had helped to reduce income-related health inequity in folk doctor care and preventive care, but the contribution was rather limited. Commercial insurance had contributed towards the inequity in preventive care use; this suggested that the better-off were more likely to participate in commercial insurance and to use preventive care. Other socioeconomic factors including income had contributed positively to inequity in health use.

The findings were consistent with some of the previous research. Zhou et al.[16] suggested that inpatient care use was concentrated among the better-off, but the inequity indices decreased from 0.224 in 2003 to 0.115 in 2008. In terms of outpatient care use for mid-aged and elderly people, Wang et al.[15] found that in more affluent provinces like Zhejiang, outpatient use was concentrated among the better-off, while in provinces with low economic development, such as Gansu, use of health care was equally distributed across income groups. The study also suggested that this may be because of the difference in terms of health care provision and coverage of insurance between these two provinces. In terms of folk doctor care, the growing inequity between the rich and the poor is troubling, and such a problem is particularly severe for low income groups. Similar findings were demonstrated in studies conducted in other developing countries. These studies suggested that demand of lower social classes for care was highly price-elastic and usually exceeded that of the rich[50‐52]. Hence, the poor were more likely to use more informal and less qualified providers, or resorted to self-treatment when they were ill[53].

To compare the level of inequity in health use of China with other countries, Van de Poel et al.[54] showed that the Erregyers’ Concentration Indices of all health care use was 0.1 in India, 0.018 in Malaysia and 0.018 in Bangladesh, which seemed comparable with the indices from China. This suggested that, in a comparative sense, China was in a similar level of equity in health utilisation as other low- and middle-income countries.

The study has a few policy implications. The extension of the NCMS coverage reduces inequitable access in formal care, but does not eliminate them. One important constraint of the NCMS is the low reimbursement rate and the high co-payment at visit. Reported average reimbursement rate for outpatient care under the NCMS was only approximately 10%[10]; it is argued that even though out-of-pocket payments for outpatient care may be easy to cope with in a short term, a large amount of outpatient costs in aggregate may still be excessively high from a social standpoint and may have substantial effects on household[9]. Similarly, the use of preventive care is unequally distributed and related to the unequal distribution of income level. A more comprehensive coverage in terms of outpatient care and preventive care is needed because outpatient care is the most commonly used for effective and efficient treatment for many health problems, especially chronic diseases, and preventive care is equally important in terms of allowing for early detection of diseases.

The NCMS aimed to achieve equity in the contribution through co-payments regardless of income levels of the participants; however, among the NCMS participants, there existed a wide gap in financial status. Low income participants are already burdened with a premium, while substantial co-payments due to the limited coverage further aggravate HI in health care access[55]. A possible solution is to implement well-designed and regulated health insurance with comprehensive coverage to provide the low income participants with better financial protection. Successful examples include - Universal Coverage scheme of Thailand and Seguro Popular of Mexico for the poor and uninsured[56‐58].

It is also worth mentioning that more and more rural Chinese, especially the better-off, are seeking for financial protection from participating in commercial insurance. The NCMS with the objective of protecting the vulnerable groups from financial barriers to care had a slightly favouringpoor effect, whereas commercial insurance contributed to the favouring-rich distribution of health use. The movements towards a more generous benefit package of the public funded insurance - the NCMS - is crucial in improving access to care and to better align health use with need. For services not being covered by the NCMS, as exemplified by the study results, resorting to commercial insurance to improve equity in access might be one option for the rural Chinese.

The study has a few limitations. The first concern is the dataset. The dataset used is probably by far the most comprehensive ever used in studying health inequality in the Chinese context; however, only nine provinces are included. Most of these provinces are situated in the eastern and coastal part of China, where the levels of economic development are high. Hence, any further generalization should be made with caution. As all the survey information is self-reported, this can be biased because of problems in reporting (e.g. inaccurate recall, misreporting). However, these are the limitations of using self-assessed morbidity measurements in the absence of other possible objective variables, such as biomarker. Second, the difference between what is officially called informal care and what happens in practice needs further refinement in future studies. In this dataset, all informal care providers are evaluated at the same standard, and are specified as “folk doctor care”; however, it is possible that folk doctor use may relate to the use of traditional Chinese medicines and healer, which are widely accepted and even recommended in some medical settings[59‐61]. Therefore, the dataset needs further refinement in the definition of folk doctor care in order to make inference on equity of use.