Abstract: Low- and middle-income countries (LMICs) have been searching for effective strategies to reform their inefficient and wasteful public hospitals. Recently, China developed a model of systemic reform called the Sanming model to address the inefficiency and waste at public hospitals. In this paper, we explain and evaluate how the Sanming model reformed its 22 public hospitals in 2013 by simultaneously restructuring the hospital governance, altering the payment system to hospitals and realigning physicians’ incentives. By employing the difference-in-difference (DID) method and using hospital-level data from 187 public hospitals in Fujian province, we find that the Sanming model has reduced medical costs significantly without measurably sacrificing clinical quality and productive efficiency at its hospitals. The systemic reform, on average, has reduced the medical care cost per outpatient visit and per inpatient admission by 6.3% (p-value =0.0445) and 16.8% (p-value < 0.001) respectively. It is largely accomplished  through a decrease in drug expenditures per outpatient visit and per inpatient admission of more than 30% (p-value <0.001) and 80% (p-value < 0.001). These results show that the Sanming model has achieved at least a short-term success in improving the performance of the Chinese public hospitals. These findings suggest that such a systemic transformation of public hospitals could improve their performance; it holds critical lessons for China and other LMIC.

Keywords: Systemic Reform, Governance Structure, Incentives, Public Hospitals, China

 

 

 

 

 

 

 

 

 

 

Introduction

Restructuring public hospitals has been high on the agenda of many low- and middle-income (LMICs) countries since the 1980s (Preker and Harding, 2003). Authors have documented that public hospitals in many LMICs are inefficient and wasteful because of the rigid bureaucratic structure and the absence of incentives to improve efficiency and quality (Jakab et al., 2002). In the past decades, many nations introduced reform measures to improve the performance of public hospitals. Those reforms, however, were often not systemic and led to mixed consequences. For example, to address the slow responsiveness to patients’ demands, some LMICs such as China, Viet Nam, Indonesia and Ghana, automnomized or corporatized public hospitals. These reforms exacerbated the problem of  affordability of hospital services because autonomized or corporatized hospitals deviated from social objectives and engaged in high profit-margin services (Yip and Mahal, 2008; Wagstaff and Bales, 2012; London, 2013; Govindraj and Chawla, 1996). Those piecemeal reforms reflect our limited understanding of what actions can be taken to restructure the public hospitals effectively.

In 2009, China launched a comprehensive healthcare reform plan to fulfill its commitment to achieving universal health coverage by 2020 with affordable and equitable access to basic healthcare services (Central Committee of the Communist Party and State Council, 2009).   Extensive positive progress has been made[1] (Yip et al., 2012; Li, 2011). However, China is still confronted with major challenges in reforming its wasteful and inefficient public hospitals, which deliver more than 80% of outpatient and inpatient services in China. Like many other LMICs, China’s public hospitals suffer from distorted provider incentives and poor governance (Eggleston et al., 2007; Yip et al., 2012). As a result, Chinese public hospitals are impeding China’s movement toward affordable universal health coverage commitment by delivering inappropriate and less cost-effective hospital services that waste scarce national resources (Yip and Hsiao, 2014). Although China has experimented with dozens of pilot hospital reforms during the past five years, few have produced significant positive results (Barber et al., 2014; Zhang et al., 2016). The slow progress in reforming public hospitals is one of the least successful parts of China’s health system reform.

Among dozens of pilot public hospital reforms, the model developed by Sanming city stood out. The Sanming model took a systemic approach by simultaneously reforming the governance structure, payment system and physician compensation methods of its 22 public hospitals. The positive results of the Sanming model caught the attention of the top Chinese leaders. In February 2016, Chinese President Xi Jinping gave his endorsement by commenting that Sanming had achieved demonstrable progress in reforming the public hospital system. China’s State Council immediately issued a policy that the Sanming model would be replicated nationwide (State Council, 2016).

This paper explicates the Sanming model and provides a rigorous evaluation of the impacts of the systemic reform. We use the difference-in-difference (DID) method to evaluate the impacts of the Sanming model by comparing the performance of public hospitals in Sanming city with that of public hospitals in other cities of the same province that have not adopted these measures. We find robust evidence that the systemic reform in Sanming has reduced medical costs substantially without measurably sacrificing medical quality and productive efficiency. These results show that Sanming’s transformation has at least achieved a short-term success in improving the performance of public hospitals. Our finding implies that a systemic reform where the governance structure, payment system and physician compensation methods are aligned can significantly improve the performance of public hospitals.

Background on reforms

The causes of the performance of China’s public hospitals

The Chinese government vastly underfunded the public hospitals when it embarked on its economic reform in 1978. Government subsidies for public hospitals fell from 40-50% before the economic reform to a mere 10% of the facilities’ total expenditures in the early 1990s (Blumenthal and Hsiao, 2005). However, under its fee-for-service payment system, the government wanted to make the services affordable to patients, so it continued to set service charges below actual costs. Meanwhile, public hospitals had to survive financially. The government adopted a strategy to allow them to make profits from prescribing/dispensing drugs and providing high-technology tests[2]. Consequently, hospital directors managed their hospitals by tying physicians’ and staff’s compensation to the profits they generated, leading to the widespread over-prescription of drugs and overuse of diagnostic tests. Currently, the normal compensation (basic salary) to physicians in public hospitals only makes up one-third of their total compensation, and the other two-thirds comes from bonuses derived from profits generated (Li et al., 2015).

Chinese public hospitals and its employed physicians suffer more than distorted incentives. Public hospitals operate in a poor governance structure, facing as many as sixteen governmental agencies that govern the hospitals; sometimes the governing rules are contradictory in purposes and goals. Consequently, what objectives the public hospitals should pursue are unclear, other than hospitals are self-motivated to survive and thrive. For example, the Ministry of Health wants public hospitals to be strong and expand, but the payers–three major social health insurance schemes–have different payment policies to contain hospital expenditures[3]. Under this circumstance, hospitals can manipulate their different payments to make the most profits and thrive (the same practice goes on in the USA under a multiple-payer system).

Additionally, the hospitals are given little autonomy to manage its own financial and human resources. For instance, civil service rules do not give hospital directors the power to hire and fire personnel. It inevitably results in the inability of hospital directors to optimize the skill mix of hospital staff. Under these circumstances, the government can’t establish any meaningful criteria for good performance and hold the directors accountable.

Since 2012, the Chinese government has given priority to reform its inefficient and wasteful public hospitals through various approaches, but the results of early appraisals of various approaches are not very encouraging. For example, some studies find that removing the 15% markup on drugs—the core strategy to reform the county-level public hospitals—has not reduced the total medical expenditures because it has led to rising diagnostic test costs and longer hospital stays (Xiao et al., 2013; Yi et al., 2015). One of the convincing explanations is that policies to reform public hospitals are piecemeal and uncoordinated, with the result that physicians and hospitals can circumvent the regulations and continue to generate profits to compensate themselves.

Systemic reforms in Sanming

Sanming, a prefecture-level city with six million residents in Fujian province, launched its systemic reforms at its 22 public hospitals in January 2013. The aim was to control health expenditure growth and ensure affordable health care services. Sanming developed a systemic reform plan that covered three crucial areas of public hospitals: the hospital governance structure, payment system to hospitals and physician compensation methods. Table 1 summarizes the key elements of these reforms.

[Table 1 inserted here]

First, Sanming reformed the governance structure.  It consolidated the dispersion of power among various departments to set health policies and govern public hospitals into one commission, chaired by the deputy mayor. This reform enabled the government to develop a set of coherent social objectives for public hospitals and hold hospital directors to be accountable for their performance. The commission also took charge of the strategic planning and strategic purchasing. This consolidation laid a foundation for creating a single-payer system with uniform payment policies for providers.

Equally important, hospital directors became more autonomous, especially in the area of managing human resources. They had more freedom to hire new staff, fire unqualified employees, and even appoint vice-directors of hospitals. At the same time, in order to hold directors accountable for the performance of public hospitals, the commission introduced a new performance measurement and reward system. The performance measurement system rated the performance of directors based on four categories with cost control given the greatest weight. The annual compensation to a director was solely based on hospital performance relative to the targets set by the commission in advance.

Next, Sanming altered the payment rate for hospital services. The distorted price schedule was modified to reflect more the cost of physician’s labor input for a service; prices for basic medical services were increased, prices for high-tech diagnostic tests were reduced moderately, and purchasing prices of drugs were reduced significantly, largely through negotiation between insurance plans and pharmaceutical companies. At the same time, the previously allowed 15% profit margin for drugs was removed so that the official linkage between drug sales and hospital surplus was disconnected. Additionally, a case-based payment method, involving about 30 diseases, was implemented to control cost inflation and assure medical quality.

Last but not the least, Sanming took actions to alter the physician compensation methods by delinking physician income from “profits”. Instead, physicians were paid a much higher basic salary plus a bonus based on performance[4]. The bonus is based on seniority, the quantity of services provided, the quality of medical services, and the achievement of strategic targets such as controlling cost inflation. Hence, the new compensation to physicians was divorced from drugs and tests that physicians would prescribe.

Methods

Study design and data sources      

Sanming introduced systemic changes in 2013. Thus, we regard hospitals in Sanming as our treatment group, while hospitals in other cities of Fujian province are treated as controls. Our basic study design is to compare changes in medical expenditures, the provision of medical services, and the medical quality at public hospitals in Sanming to the control group, where there were no systemic changes from 2013 to 2014.

The main data used in this paper is taken from three sources: Annual Statistical Report on China’s Public Hospitals, Annual Report on Medical Quality for Sanming’s Hospitals, and Fujian Statistical Yearbook (FSY). The Annual Statistical Report on China’s Public Hospitals contains extensive hospital-level information on health costs, the provision of medical services, and medical resources ranging from 2003 to 2014. We use total expenditures per outpatient visit, total expenditures per inpatient admission, the drug cost per outpatient visit, the drug cost per inpatient admission, drug sales as a share of total expenditures, outpatient visits, inpatient admissions, bed occupation rates, and days per inpatient admission as our dependent variables. All expenditure variables in the paper are converted to 2008 yuan (CN¥) using the CPI. In order to investigate the effect of Sanming’s reforms, we limit our sample to all 187 public hospitals in nine cities of Fujian province between 2008 and 2014. The Annual Report on Medical Quality for Sanming’s hospitals contains hospital-level information on the medical quality at Sanming’s public hospitals between 2008 and 2014. Measures of the patient satisfaction rate, the nosocomial infection rate, the surgical incision healing rate, the rescue success rate, the number of senior physicians per hospital, and the number of physicians with a master’s degree or higher per hospital are included. The Fujian Statistical Yearbook contains social and economic statistical data for each city or county from 2008 to 2014. The following indicators are selected as control variables: GDP per capita, year-end resident population, public revenue per capita, public expenditures on education per capita, respective output in the primary industry and secondary industry as a share of GDP. We match these control variables with our hospital-level data to construct our analysis sample, using the unique city or county code.

Econometric analysis

We employ the difference-in-difference method to control for the unobserved time-invariant individual effects and common time-varying trends. Specifically, we use the following linear regression fitted by the least square approach:

(1)

The subscript i indicates the hospital and t indicates the year. Yit represents the dependent variables: medical expenditure indicators and service volume indicators. All the variables are estimated in logs except drug sales as a share of total expenditures because hospitals vary considerably in size. αi is a series of a hospital’s individual fixed effects that controls for the unobserved time-invariant individual heterogeneity across hospitals. γt represents a vector of year dummies that is used to control for flexible year effects. εit refers to the error term. Standard errors are clustered at the county level.

The key variable of interest is Reformit. It is a dummy variable indicating the reform status (i.e., it equals 1 for years after 2013 and equals 0 for years before 2013 if a hospital is in Sanming; it equals 0 for all years if a hospital is not located in Sanming). The coefficient, λ, captures average effects of reforms on outcome, Yit. Xit is a set of covariates, including GDP per capita, year-end permanent population, public revenue per capita, public expenditures on education, output in the primary industry as a share of GDP, and output in the secondary industry as a share of GDP.

Because we can only access data on the medical quality at Sanming’s hospitals, we are unable to implement the DID model to examine whether the reform has any side-effects on the medical quality. Instead, by adopting data only from Sanming, we investigate the time series pattern of the medical quality with linear regression, which can provide supplementary evidence on whether the medical quality at Sanming’s hospitals has been affected by the reforms. The time series pattern of coefficients on a vector of year dummy variables, λt, is of interest. The relative changes in λt provide suggestive evidence on whether reforms have decreased medical quality. In other words, if there is no turning point in the trend of λt before versus after 2013, then we can conjecture that the reforms may not reduce medical quality significantly.

(2)

 

Results

Summary Statistics

Figure 1 compares the time series patterns for six key variables between the treatment group and the control group. It is easy to observe that trends for Sanming’s hospitals in terms of medical expenditures drop sharply after the 2013 reforms, while their trends regarding the provision of medical services remain parallel with the trends for other cities in Fujian. Based on these patterns, we conjecture that the systemic reforms in Sanming may have impacts on reducing health expenditure.

[Figure 1 about here]

Table 2 provides detailed information on changes in hospital outcomes between the period before the reform (2008-2012) and after the reform (2013-2014), in which statistical information on hospitals in Sanming (treatment group) and other cities of Fujian province (control group) is reported respectively. The mean value of expenditure variables for hospitals in Sanming drops substantially before versus after 2013, while hospitals in other cities of Fujian experience a moderate increase in health expenditures during the same period. The magnitude of the reduction in drug costs is striking. For example, drug sales as a share of total expenditures decline from about 48% in the period of 2008-2012 to about 27% in the period of 2013-2014, while the value for hospitals in other areas remains above 45% in the period of 2013-2014. As for the provision of services, we find that all related variables increase for both Sanming and other cities, and the gap between them does not significantly change over time. Table 2 also shows the descriptive statistics for control variables. Based on the statistics, we find that there are not major changes in the gaps between Sanming and other cities regarding economic and social development.

Results from econometric analysis

In this section, we present the DID results based on linear regression models. Table 3 shows the estimated effects of Sanming’s reform on various medical expenditure variables: total expenditures per outpatient visit, total expenditures per inpatient admission, drug cost per outpatient visit, drug cost per inpatient admission, and drug sales as a share of total expenditures. The coefficients on these dependent variables clearly indicate that systemic reforms in Sanming have reduced medical costs substantially, both for outpatient and inpatient care. To be specific, the reforms, on average, have reduced medical care cost per outpatient visit and per inpatient admission by 6.3% (p-value=0.0445) and 16.8% (p-value<0.001) respectively, as shown in columns (1) and (3). Furthermore, column (2) and column (4) show that the drug cost per outpatient visit and per inpatient admission has been reduced by 34.6% (p-value<0.001) and 82.3% (p-value<0.001) respectively. As a result, drug expenditure as a share of total expenditures drops by about 17 percentage points (p-value<0.001). Not surprisingly, the magnitude of the drop in total expenditures is smaller than the drug cost because the prices for basic medical services have been increased. The net negative effects on total medical expenditures indicate that although the prices for basic medical services have been increased, reforms in Sanming have succeeded in slowing down the growth of medical expenses through controlling drug expenditures.

[Table 3 about here]

We also investigate the effect of Sanming’s reforms on the spending of diagnostic tests per outpatient visit and inpatient admission. We find that the reforms have insignificant effects on the spending of diagnostic tests (see Supplementary Table 1). In other words, hospitals in Sanming did not provide a significantly rising number of diagnostic tests to make up for the loss from decreasing drug sales. In addition, the estimated coefficient on fiscal subsidies is negative and insignificant (also see Supplementary Table 1), which indicates that decreases in outpatient and inpatient cost are driven by systemic changes in provider incentives and governance structure rather than by the increase in public funding.

Given the remarkable shrinkage in medical expenses, it is necessary to examine the changes in the provision of services and medical quality. It is entirely possible that reforms might result in some negative effects like a reduced quantity of medical services provided or decreasing medical quality. Therefore, we cannot make a conclusion based simply on a reduction in health costs. Only if the reforms do not significantly lower the levels of service quantity and medical quality, can we conclude that reforms help ease patients’ burdens and make healthcare more affordable.

Table 4 presents the estimation results for the provision of medical services in public hospitals: the number of outpatient visits, the number of inpatient admissions, the bed occupation rate, and days per inpatient. As shown in column (1), the coefficient indicates that the reforms have reduced the number of outpatient visits moderately, by about 9.5% (p-value=0.0236). However, the estimation results in columns (2)-(4) show that there are insignificant effects on the changes in inpatient admissions, the bed utilization rate, and days per inpatient. These mixed results make it difficult to reach a conclusion on whether reforms in Sanming have led to any side-effect on the productive efficiency of public hospitals. It is unclear whether the reduction in outpatient visits is attributable to fewer repetitive visits or to a loss of patient base. To explore the underlying cause of the decrease in outpatient visits, we examine the effect of the reforms in Sanming on outpatient visits to private hospitals using the city-level aggregated data. If the decline in outpatient visits is due to lower productive efficiency of physicians, we would expect to find an increase in outpatient visits at private hospitals. As shown in Supplementary Table 1, the coefficient for private hospitals is small and insignificant. To sum up, we believe that the provision of medical services has not been affected by the reforms measurably. The underlying causes of decreasing outpatient visits to Sanming’s public hospitals after the reform need to be further explored.

[Table 4 about here]

Figure 2 shows the time trend of medical-quality measurements for Sanming’s hospitals by exhibiting the coefficients of the yearly variables estimated from the linear regression model. For convenience, we set 2008 as the reference year. Each graph in Figure 2 successively depicts the annual trends on the patient satisfaction rate, the nosocomial infection rate, the surgical incision healing rate, the rescue success rate. We observe that the rate of patient satisfaction and the rate of rescue success increase over time, but both of the coefficients are not significant in most years at the 5% significance level, which indicates that the patient satisfaction rate and rescue success rate have not experienced a significant decrease. The time trend for nosocomial infection rate decreases over time, and there is no turning point in the trend, which indicates that the nosocomial infection rate has not been affected by the reforms. As for the surgical incision healing rate, a drop is observed in 2011 and 2012, but it has been increasing since 2013. Figure 2 also exhibits the results for the number of senior physicians and the number of physicians with master’s degrees or above, which are indirect indicators of medical quality. The increasing trend in the number of senior physicians and the number of physicians with master degrees or above indicate that they are also not affected by the reforms. All these results suggest that there is no strong evidence to prove that clinical quality has decreased due to the reforms.

[Figure 2 about here]

Robustness

We implement two placebo tests to check the robustness of our findings. In the first test, we use observations of the 2008-2012 samples and classify hospitals to a treatment group or control group according to whether it is in Sanming and assume that the reforms happened in 2010, instead of 2013. The rationale of the test is that if the early regression results are driven by some unobserved, time-varying heterogeneities between the treatment and control groups, then these heterogeneities will lead to a placebo result similar to the early results(Bertrand et al., 2004). The second placebo test follows the idea of permutation test (Abadie et al., 2010), in which we reassign the treatment status to one of the other eight cities iteratively, keeping Sanming in the control group. That is, we proceed as if every other cities in the control group would have experienced systemic changes in 2013, instead of Sanming. As shown in Supplementary Table 2 and Table 3, we find that the relevant estimates are either insignificant or have conflicting signs, which suggests that our primary findings are driven by Sanming’s reform strategies rather than by some unobservable variables. In addition, we exclude hospitals in Fuzhou and Xiamen from our sample to test whether our early results are driven by the special trend in Fuzhou, which is the provincial capital, and Xiamen, which is the only deputy-provincial-level city in Fujian. Estimates in Supplementary Table 4 are similar to estimates in the early main regressions.

Conclusion and discussion

With the aim of delivering affordable high-quality health care to all people, Chinese government has spent additional trillions of dollars of public revenue to expand health insurance coverage since 2009. However, the people did not benefit that much because a large portion of the additional resources went to increasing the revenue of medical providers, the profit of pharmaceutical companies, and the income of physicians (Yip and Hsiao, 2014). Chinese inefficient and wasteful public hospitals are impeding China’s movement toward affordable universal health coverage commitment by delivering inappropriate and less cost-effective hospital services. Although Chinese government has tried many pilot programs to reform its public hospital system and control its soaring health expenditure since 2012, most of these pilot reforms have not yielded significant positive results because they are piecemeal remedies.

This paper explicates and evaluates a systemic reform of public hospitals in China called the Sanming model. By employing the DID method and time-series analysis, this paper finds robust evidence that reforms in Sanming have reduced medical costs substantially without measurably sacrificing medical quality and quantity of services. The Sanming model has reduced the medical care cost per outpatient visit and per inpatient admission by 6.3% (p-value =0.0445) and 16.8% (p-value < 0.001) respectively. It is largely accomplished through a decrease in drug expenditures per outpatient visit and per inpatient admission of more than 30% (p-value <0.001) and 80% (p-value < 0.001). These results show that the Sanming model has at least achieved a short-term success in improving the performance of public hospitals.

The Sanming model holds critical lessons for China and other LMICs. The inefficiency, waste and poor quality of public hospital services may be caused by systemic problems rather just underfunding which is often cited as the cause. While public hospitals have to be adequately funded, this does not assure the resources will be allocated efficiently and effectively. Existing studies find that the inefficiency rate of public hospitals in LMICs exceeds 15% (Masiye, 2007; Kathuria and Sankar, 2005; Chisholm and Evans, 2010).

Large hospitals are complex organizations.  Public hospitals’ performance depends on external and internal components.  Externally, governance and organizational structure influence the behaviors of hospital executives and hospital performance. Payment incentives to the hospitals and competition confronting the hospitals also influence hospital behaviors. Internally, the incentives to hospital physicians and medical staff influence clinical practices directly. The theory of managerial economics tells us that all these external and internal components have to be aligned properly (Brickley et al., 1995). Currently we pay more attention to incentives, but have not given adequate attention to the governance and organizational structure of public hospitals. Prominent health experts have attributed the failure of public hospitals in some LMICs to the inconsistent design that created dysfunctional organizations (McPake, 1996; Jakab et al., 2002).

The Sanming model demonstrates that a systemic reform can significantly improve the performance of public hospitals. China and LMICs can examine what components the Sanming model has reformed and aligned to achieve its results. These components at least include an effective governance structure that sets clear social goals for public hospitals, autonomous decision rights of directors to manage hospitals, accountability system for directors, rational payment method and rates to hospitals, and hospitals’ internal financial incentive that motivate physicians to deliver value-based care. An absence of anyone of these components may cause failure in the performance of public hospitals.

Acknowledgement                                                                                                              

We would like to thank Xiaoyan Lei, Tianyang Xi, Wei Huang and seminar participants at Peking University and Harvard University. We acknowledge the assistance of Health and Family Planning Commission of Sanming (HFPCS) in providing data analyzed in this paper. We have not accepted any financial support from HFPCS. The opinions expressed in this paper and any errors are those of authors alone.

Conflicts of interest

We declare that we have no conflicts of interest

The role of funding source

We declare no funding sources for this study. No agencies have the role in study design, data collection, data analysis, data interpretation, or writing of the paper. The corresponding author has full access to all the data in this study and has final responsibility for the decision to submit for publication.

 

Reference

Abadie, A., Diamond, A., Hainmueller, J. 2010. Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program. Journal of the American Statistical Association, 105, 493-505.

Arrow, K. J. 1963. Uncertainty and the welfare economics of medical care. The American economic review, 53: 941-973.

Barber, S. L., Borowitz, M., Bekedam, H., Ma, J. 2014. The hospital of the future in China: China’s reform of public hospitals and trends from industrialized countries. Health Policy and Planning, 29:367-378.

Bertrand, M., Duflo, E., Mullainathan, S. 2004. How Much Should We Trust Differences-In-Differences Estimates?. The Quarterly Journal of Economics, 119:  249-275.

Blumenthal, D. and Hsiao, W. 2005. Privatization and its discontents—the evolving Chinese health care system. New England Journal of Medicine, 353: 1165-1170.

Blumenthal, D. and Hsiao, W. 2015. Lessons from the East—China’s rapidly evolving health care system. New England Journal of Medicine, 372: 1281-1285.

Brickley, J., Smith, C., Zimmerman, J. 1995. The Economics of Organizational Architecture. Journal of Applied Corporate Finance 1995, 8: 19-31.

Central Committee of the Communist Party and State Council. People’s Republic of China. 2009. The standing conference of State Council of China adopted guidelines for furthering the reform of health-care system in principle. http://news.xinhuanet.com/ newscenter/2009-04/06/content_11138803.htm (in Chinese, accessed April 1, 2016)

Chisholm, D. and  Evans, D. B. 2010. Improving health system efficiency as a means of moving towards universal coverage. World health report. Background Paper(28)

Eggleston, K., Ling, L., Qingyue, M., Lindelow, M., Wagstaff, A. 2008. Health service delivery in China: a literature review. Health economics, 17:149-165.

Govindaraj, R. and Chawla. M. 1996. Recent Experiences with Hospital Autonomy in Developing Countries: What Can We Learn?. Boston, MA: Harvard School of Public Health.

Jakab, M., Preker,A. S., Harding,A. and  Hawkins, L. 2002. The introduction of market forces in the public hospital sector: From new public sector management to organizational reform. Washington, DC: World Bank.

Kathuria, V., and Sankar, D. (2005). Inter‐State Disparities in Health Outcomes in Rural India: An Analysis Using a Stochastic Production Frontier Approach. Development policy review, 23: 145-163.

Kesternich, I., Schumacher, H. and Winter, J. 2015. Professional norms and physician behavior: homo oeconomicus or homo hippocraticus?. Journal of Public Economics, 131: 1-11.

Li, F., Jin, C., Wang, L., Wang, C., Wu, B. 2015. Analysis on the Human Expenditure for Public Medical Institutions of Shanghai. China Health Economics, 1: 9-12. (in Chinese)

Li, L. 2011. The challenges of healthcare reforms in China. public health, 125: 6-8

London, J. D. 2013. The promises and perils of hospital autonomy: Reform by decree in Viet Nam. Social Science & Medicine, 96: 232-240.

Masiye, F. 2007. Investigating health system performance: an application of data envelopment analysis to Zambian hospitals. BMC Health services research, 7: 1-11.

McPake, B. I. 1996. Public autonomous hospitals in sub-Saharan Africa: trends and issues. Health policy, 35: 155-177.

Preker, A. S., and Harding, A. 2003. A conceptual framework for the organizational reforms of hospitals. In A. Harding & A. S. Preker (Eds.), Innovations in health service delivery: The corporatization of public hospitals (pp. 23-78). Washington, DC:World Bank.

Ramesh, M. Autonomy and control in public hospital reforms in Singapore. 2008. The American Review of Public Administration,  38: 62-79.

State Council. 2016. People’s Republic of China, priorities for economic reform in 2016, http://www.gov.cn/zhengce/content/2016-03/31/content_5060062.htm, (in Chinese, accessed April 4, 2016)

 

Wagstaff, A., Bales, S. 2012. The impacts of public hospital autonomization: evidence from a quasi-natural experiment. World Bank policy research working paper, (6137).

Xiao, Y., Zhao, K., Bishai, D. M., Peters, D. H. 2013. Essential drugs policy in three rural counties in China: what does a complexity lens add?. Social Science & Medicine,  93:220-228.

Yi, H., Miller, G., Zhang, L., Li, S., Rozelle, 2015. S. Intended and unintended consequences of China’s zero markup drug policy. Health Affairs, 34: 1391-1398.

Yip W, Hsiao W. 2014. Harnessing the privatisation of China’s fragmented health-care delivery. Lancet, 384: 805-818.

Yip, W. C. M., Hsiao, W. C., Chen, W., Hu, S., Ma, J., Maynard, A. 2012. Early appraisal of China’s huge and complex health-care reforms. Lancet, 379:833-842.

Yip, W., Mahal, A. 2008. The health care systems of China and India: performance and future challenges. Health Affairs, 27: 921-932.

Zhang, Y., Ma, Q., Chen, Y., Gao, H. 2016. Effects of Public Hospital Reform on Inpatient Expenditures in Rural China. Health economics, forthcoming

 

Figure 1: The time trend of six variables measuring medical costs and service delivery. The solid black lines refer to the time trends of relevant variables for Sanming, while the gray dash lines refer to the time trend for other cities in Fujian province.

 

Figure 2: The time trend of medical-quality variables for Sanming from 2008 to 2014. The graphs indicate the patterns of coefficients λt in equation (2) relative to the reference year (2008). The vertical lines indicate the 95% confidence interval of each coefficient relative to the reference yea

 

Table 1: An overview of China’s public hospital system and reform strategies in Sanming

Most Public Hospitals in ChinaReform Strategies in Sanming
Governance Structure
The dispersion of power among Government Departments

 

Public hospitals are governed by as many as 16 ministries. Strategies on reforming public hospitals cannot be coordinated because of the dispersion of power among competing ministries.A commission chaired by the deputy mayor was established to address the dispersion of power between government departments. This consolidation laid a foundation for creating a single payer system and clearly identifying social objectives of hospitals.
The Role of Hospital DirectorsDirectors have freedom concerning the discretionary use of profit but very limited staffing power.

 

An evidence-based accountability system has not been established to hold directors accountable for hospital performance.

Directors became more autonomous in human resource management. They had more freedom to hire new staffs, fire unqualified employees, and even appoint vice-directors of hospitals.

A performance measurement system was introduced to rate the performance of directors based on four categories with cost control given the greatest weight. The annual payment to a director was based solely on hospital performance relative to targets set in advance.

Financial Incentives
Payment to HospitalsPayment methods are dominated by fee-for-service. Prices for basic medical services are set far below cost, but the prices for high-tech diagnostic tests and drugs are set above cost.Prices for basic medical services were adjusted upward on the basis of cost. Prices for high-tech diagnostic tests and drugs were reduced.

A case-based payment method involving about 30 selected diseases was introduced to assure medical quality and to control cost growth.

Physician Compensation MethodsPublic hospitals are under great pressure to bring in revenues for their financial survival because of few subsidies from the government. Physicians’ personal income and promotions are heavily tied to the profits they produce.Physicians’ income was unlinked from “profit” that they produced. Physicians were paid a basic salary plus a bonus based on performance. The bonus is based on seniority, the quantity of services provided, the quality of medical services, and the achievement of strategic targets such as controlling cost inflation.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2: Summary Statistics

Variables2008-20122013-2014
OverallSanmingOther regionsSanmingOther regions
MeanStd.MeanStd.MeanStd.MeanStd.MeanStd.
Panel 1: Medical care expenditures (CN)
Total expenditures (10K)14321260095349707011965218829772110442363336214
Medical care cost per outpatient visit
Total cost122.660.7997.9625.41120.662.08106.423.513865.69
Drug cost63.9738.3347.7913.3364.2739.8739.6812.172.1139.67
Medical care cost per inpatient admission
Total cost4645335436411321460134343500113752523706
Drug cost204417571723656.821261855781.3318.521331813
Drug cost as a share of total cost (%)46.5411.5547.687.09348.3311.1527.24.5144.5111.66
Panel 2: Medical service utilization
Outpatient visits369607466474183153142440362279458584233464191507469362551107
Inpatient admissions159811589495918187152141481713377122342029919382
Bed occupation rate (%)88.3321.487.2615.1389.4922.7581.3716.7386.5920.76
Days per inpatient10.4810.119.4762.34110.6810.068.391.3610.5912.28
Panel 3: Medical care quality
Patient satisfaction rate

93.824.659

95.053.744

Nosocomial infection rate1.8981.3411.2030.860
Surgical incision healing rate91.0422.9991.4323.18
Rescue success rate89.417.14988.2612.32
Number of senior physicians7.89510.9211.6014.04
doctors with master’s degree or above4.30514.937.21122.07
Panel 4: Control Variables
GDP per capita (¥)43343172113777114962392531611754616148775344715815
Permanent resident population (10K)104.6108.624.5910.98111.7103.824.1210.80125.8127.4
Public revenue per capita2563192916881010248919872325847.730721998
Educational expenditure per capita866.7341.5853.6268.2761.9270.41289439.21063370.1
Primary industry output as a share of GDP13.747.14017.641.81412.506.86321.025.45214.367.952
Secondary industry output as a share of GDP48.186.53448.821.91848.636.71146.365.94847.137.039

Notes: Data for Panel 1 and 2 are taken from Annual Statistical Report on China’s Public Hospitals. Data for Panel 3 comes from Annual Report on Medical Quality for Sanming’s Hospitals. All variables in Panels 1, 2, and 3 are measured annually at the hospital level. Data on control variables are from the Fujian Province Statistical Yearbook.

 

 

Table 3: Impact of reforms on medical expenditures

(1)(2)(3)(4)(5)
VariablesLog (total cost per outpatient visit)Log (drug cost per outpatient visit)Log (total cost per inpatient admission)Log (drug cost per inpatient admission)Drug sales as a share of total health cost
Reform-0.0628**-0.346***-0.168***-0.823***-16.89***
(0.0311)(0.0477)(0.0305)(0.0451)(0.992)
Individual fixed effectsYYYYY
Yearly fixed effectsYYYYY
Covariates includedYYYYY
R-square0.5590.2880.4210.5450.615
Observations1,2431,2431,2421,2421,243

Notes: Standard errors clustered at the county level are reported in parentheses. Covariates include indicators of GDP per capita, public revenue per capita, public educational expenditures per capita, permanent resident population, the output in the primary industry as share of GDP, and output in the secondary industry as a share of GDP. All control variables are measured at the county-level. ***, **, and * denote the significance at the 1%, 5%, and 10% level respectively.

 

 

Table 4: Impact of reforms on provision of medical services

(1)(2)(3)(4)
VariablesLog (outpatient visits)Log (inpatient admissions)Log (bed occupation rate)Log(days per inpatient)
Reform-0.0948**-0.0817-0.0536-0.456
(0.0410)(0.0510)(0.0408)(0.537)
Individual fixed effectsYYYY
Yearly fixed effectsYYYY
Covariates includedYYYY
R-square0.6190.5900.0570.066
Observations1,2431,2421,2421,243

Notes: Standard errors clustered at the county level are reported in parentheses. Covariates and variable definitions are the same as those in Table 3. ***, **, and * denote the significance at the 1%, 5%, and 10% level respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Supplementary Table 1: Impact of reforms on other hospital performance variables

(1)(2)(3)(4)(5)
VariablesLog (diagnostic test cost per outpatient)Log (diagnostic test cost per inpatient)Log (fiscal subsidy)Log (outpatient visits for private hospital)Log (inpatient visits for private hospital)
Reform-0.05660.0516-0.04880.0360.027
(0.0469)(0.0701)(0.164)(0.129)(0.191)
Individual fixed effectsYYYYY
Yearly fixed effectsYYYYY
Covariates includedYYYYY
R-square0.0790.4410.3410.920.88
Observations1,2421,2421,2306262

Notes: Standard errors clustered at the county level are reported in parentheses. Covariates include indicators of GDP per capita, public revenue per capita, public educational expenditures per capita, permanent resident population, the output in the primary industry as a share of GDP, and output in the secondary industry as a share of GDP. All control variables are measured at the county-level. ***, **, and * denote the significance at the 1%, 5%, and ten%t level respectively.

 

 

Supplementary Table 2: Placebo test I implemented to exclude the possibility that the main results are driven by some unobserved, time-varying heterogeneities between the treatment and control groups

Panel A(1)(2)(3)(4)(5)
VariablesLog (total cost per outpatient visit)Log (drug cost per outpatient visit)Log (total cost per inpatient admission)Log (drug cost per inpatient admission)Drug sales as a share of the total cost
Pseudo-reform0.000404-0.001110.02820.02650.0229
(Assumed in 2010)(0.0233)(0.0318)(0.0197)(0.0292)(0.787)
Individual fixed effectsYYYYY
Yearly fixed effectsYYYYY
Covariates includedYYYYY
R-squared0.0910.2360.3530.2770.130
Observations696696696696695
Panel B(1)(2)(3)(4)
VariablesLog (outpatient visits)Log (inpatients admissions)Log(Bed occupation rate)Log (Days per inpatient)
Pseudo-reform-0.0259-0.0680**0.04550.0348
(Assumed in 2010)(0.0360)(0.0288)(3.295)(0.0430)
Individual fixed effectsYYYY
Yearly fixed effectsYYYY
Covariates includedYYYY
R-squared0.5030.6290.0540.052
Observations696696696696

Notes: Standard errors clustered at the county level are reported in parentheses. Covariates and variable definitions are the same as those in Supplementary Table 1. ***, ** and * denote the significance at the 1%, 5%, and 10% level respectively.

 

 

 

 

 

 

 

 

 

 

 

 

Supplementary Table 3: Placebo test II by assigning reform status to each one of other eight cities

(1)(2)(3)(4)(5)
Treatment GroupLog (total cost per outpatient visit)Log (total cost per inpatient admissiondrug sales as a share of the total costLog (outpatient visits)Log (inpatients admissions)
Sanming-0.0628**-0.346***-16.89***-0.0948**-0.0817
(0.0311)(0.0477)(0.992)(0.0410)(0.0510)
Fuzhou0.221***-0.004684.190***0.03810.0182
(0.0558)(0.0224)(0.999)(0.0301)(0.0443)
Xiamen-0.01050.0230-2.079-0.110**-0.0892*
(0.0649)(0.0295)(1.547)(0.0467)(0.0529)
Putian-0.0672**0.0408**1.680-0.0808**-0.0629*
(0.0264)(0.0186)(1.820)(0.0378)(0.0316)
Quanzhou-0.0549-0.01572.5210.06610.0849
(0.0597)(0.0279)(1.711)(0.0401)(0.0521)
Zhangzhou

 

-0.05340.0340-1.1580.0401-0.0668
(0.0540)(0.0373)(1.102)(0.0571)(0.112)
Nanping

 

-0.04780.05717.043***-0.0702-0.0231
(0.0670)(0.0356)(1.920)(0.0684)(0.0850)
Longyan

 

0.04290.0662*2.5320.03070.0902**
(0.0859)(0.0377)(2.295)(0.0626)(0.0437)
Ningde-0.01700.05183.888**0.101**0.131**
(0.0537)(0.0313)(1.673)(0.0429)(0.0581)

Notes: To save space, the estimation results for drug cost per outpatient, drug cost per inpatient, bed occupation rates, and days per inpatients are not shown here. The results for these dependent variables are similar to the results are shown here. Standard errors clustered at the county level are reported in parentheses. Hospital individual fixed effects and yearly fixed effects are controlled. Covariates and variable definitions are the same as those in Supplementary Table 1. ***, **, and * denote the significance at the 1%, 5%, and 10% level respectively.

 

 

Supplementary Table 4: Excluding observations in Fuzhou & Xiamen

Panel A(1)(2)(3)(4)(5)
VariablesLog (total cost per outpatient visit)Log (drug cost per outpatient visit)Log (total cost per inpatient admission)Log (drug cost per inpatient admission) Drug cost as a share of total cost
Reform-0.0683**-0.350***-0.171***-0.818***-16.90***
(0.0338)(0.0497)(0.0316)(0.0459)(1.059)
Individual fixed effectsYYYYY
Yearly fixed effectsYYYYY
Covariates includedYYYYY
Observations892892891891892
R square0.6720.3050.4680.6080.663
Panel B(1)(2)(3)(4)
VariablesLog (outpatient visits)Log (inpatient admissions)Log (bed Occupation rate)Log (days per inpatient)
Reform-0.0883*-0.0827-0.0454-0.997*
(0.0443)(0.0573)(0.0384)(0.525)
Individual fixed effectsYYYY
Yearly fixed effectsYYYY
Covariates includedYYYY
Observations892891891892
R square0.6020.6000.0610.046

Notes: This table reports the results of regressions that exclude observations in Fuzhou and Xiamen. Standard errors clustered at the county level are reported in parentheses. Covariates and variable definitions are the same as those in Supplementary Table 1. ***, **, and * denote the significance at the 1%, 5%, and 10% level respectively.

 

[1] Progress includes the coverage of more than 95% of the Chinese population by three major social health insurance schemes, the strengthening of primary care delivery and the equalization of access to public health services across both rural and urban areas.

[2] Legally, hospitals could mark up drugs by 15% for profit. This irrational approach required the physicians to prescribe $70 of drugs to generate $10 profit to finance their compensation. The higher the drug price is, the greater the profit. Thus hospitals and physicians have the incentives to use the most expensive drugs.

[3] They are the rural new cooperative medical scheme (NCMS), the urban resident basic medical insurance (URBMI) and the urban employee basic medical insurance (UEBMI).

[4] On average, the new total compensation to physicians may be two or three times as much as their previous legal income.

發佈留言

發佈留言必須填寫的電子郵件地址不會公開。 必填欄位標示為 *

這個網站採用 Akismet 服務減少垃圾留言。進一步了解 Akismet 如何處理網站訪客的留言資料