Social Science & Medicine 111 (2014) 10e16

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Payment reform and changes in health care in China Chen Gao a, b, *, Fei Xu c, a, Gordon G. Liu d, a a

China Center for Health Economics Research, Peking University, China Novo Nordisk (China) Pharmaceuticals Co., Ltd, China c School of International Pharmaceutical Business, China Pharmaceutical University, China d National School of Development, Peking University, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 18 December 2012 Received in revised form 27 March 2014 Accepted 31 March 2014 Available online 1 April 2014

This paper is intended to assess the primary effects on cost, utilization and quality of care from payment reform of capitation and open enrollment in Changde city, Hunan Province of China. Open enrollment policy was introduced to deal with possible cream skimming associated with capitation. Based on the longitudinal Urban Resident Basic Medical Insurance (URBMI) Household Survey, this study analyses the URBMI data through a set of regression models. The original data included over five thousand inpatient admissions during the study period between 2008 and 2010. The study finds the payment reform to reduce its inpatient out-of-pocket cost by 19.7%, out-of-pocket ratio by 9.5%, and length of stay by 17.7%. However, the total inpatient cost, drug cost ratio, treatment effect, and patient satisfaction showed little difference between Fee-For-Service and capitation models. We conclude that the payment reform in Changde did not reduce overall inpatient expenditure, but it decreased the financial risk and length of stay of inpatient patients without compromising quality of care. The findings would contribute to the health care payment literatures from developing countries and open further research tracks on the ability of open enrollment to compensate for capitation drawbacks. Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: China Payment Capitation Policy impact Household survey data

1. Introduction By the end of 2011, the social health insurance programs covered 95% of the total Chinese population (Sun, 2011). The total payment from these programs was estimated to account for over 50% of provider revenues (Yao, 2011), and over 25% of total health expenditure (NHFPC, 2013; MoHRSS, 2013). Individual out-ofpocket payment, on the other hand, is on the decline as percentage of total health expenditure since 2001 (Fig. 1). As the health care safety net continues to grow even further in both depth and breadth, the payment system will play an increasingly important role in cost-containment and resource allocation of health care in China. China’s total health expenditure is about 24 times greater in 2009 than in 1980 after controlling for inflation, deflated using consumer price index (Pan et al., 2013). Improper provider payment incentives are largely responsible for these cost escalations (Eggleston et al., 2008). Chinese payers primarily use a “Fee-ForService” (FFS) payment method, which incentivizes providers to * Corresponding author. China Center for Health Economics Research, Peking University, Beijing, China. E-mail address: [email protected] (C. Gao). http://dx.doi.org/10.1016/j.socscimed.2014.03.035 0277-9536/Ó 2014 Elsevier Ltd. All rights reserved.

induce unnecessary demand at the expense of more cost-effective treatment, because a higher volume, especially on expensive drugs and equipment tests with high profit margins, means a higher profit. This overuse of expensive drugs and tests results in runaway cost inflation, waste resources, and may lower quality of care (Yip et al., 2012). Many countries including China have begun moving away from the FFS payment model and experimenting with alternative payment plans. The payment reform is on the political agenda for China’s health care system reform (CPC Central Committee and the State Council, 2009; The State Council, 2009). The two major payers in China, the Ministry of Human Resources and Social Security (MOHRSS) and the National Health and Family Planning Commission (NHFPC) have both issued official documents on payment reform, in 2011 and 2012, respectively, which identified the use prospective payment methods including capitation as apriority of payment reform (MoHRSS, 2011; NHFPC, 2012). In this paper, we focus on a local payment reform of capitation experiment for inpatient beneficiaries from Urban Resident Basic Medical Insurance (URBMI) program in Changde city, Hunan Province, China. Capitation system is thought to incentivize providers to contain cost, and if the contract is long-term, keep the population as healthy as possible. But it is vulnerable to cost-

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

11

Fig. 1. Dynamic trend of health care financing of health care from 2001 to 2010.

shifting to the uninsured or services not covered by capitation, under treatment in the interest of cost-containment, or less responsiveness to population needs (Hu et al., 2008). Changde also introduces an open enrollment policy to incentivize providers to compete over both cost and quality to attract beneficiaries. Therefore, cost, utilizations and quality of care are important aspects when evaluating the reform policies in Changde. While it is generally accepted that the way providers are paid affects their performance, and that this response can be very large (McClellan, 2011; McGuire, 2010), the empirical studies on the impacts of changes in payment systems are limited in the following three ways: One, although there are several important exceptions, the existing literature is largely focused on the effects of payment reform within the United States. Some exceptions include evaluation of cost reduction associated with prospective payment reform in Hainan, China (Yip and Eggleston, 2001), study of the effects on hospital admissions and length of stay from DRGs reform in Hungary (Kroneman and Nagy, 2001), and experiment of payment reform in rural China to study its impact on efficiency and cost (Wang et al., 2011). Two, previous literates are limited in data and methodologies (Moreno-Serra and Wagstaff, 2010; Eggleston et al., 2008). There are only a few studies using survey data or claims data based on differences-in-differences (DID) model to control for the confounding effects. Three, the existing literature largely concentrates on provider-centric outcomes rather than patient-centric outcomes (Schmidt et al., 2011; Dafny, 2005; Shmueli et al., 2002). This paper tries to avoid the aforementioned limitations in the following three ways: one, we evaluate the policy impact of payment reform in China, adding to the limited payment literatures from developing countries; two, we employ household survey data for the empirical investigation, based on a couple of regression techniques to isolate the payment system effect from other confounding factors; and three, we use a set of variables to track down the effects of the payment change on cost, utilizations, and quality of care. This paper is organized as follows: Section 2 introduces the policy background of the evaluated local capitation experiment; Section 3 outlines the research design; Section 4 describes the study results; Section 5 is the conclusion and discussions.

2. Background 2.1. Reform rationales Prior to 2007, there were two social health insurance programs: Urban Employee Basic Medical Insurance (UEBMI) for urban employed and New Rural Cooperative Medical System (NRCMS) for the rural population. Urban Resident Basic Medical Insurance (URBMI) was created in 2007 to cover the third population cohort without formal employment in urban areas. Changde was one of the 79 cities chosen in 2007 to participate in a URBMI pilot project. Three main challenges emerged from the expansion of insurance in Changde city (Tan, 2009), including a disparity between the demand and ability to pay for health care from the URBMI population, the insufficient size of the Changde Health Insurance Bureau to manage URBMI under FFS model, and the continued inflation of city health care costs due to overprescribing. Cumulatively, the effects of these three challenges led Changde to pursue payment reform in order for a more smoothly implement of the URMBI policy. 2.2. Capitation policy In response to these challenges, Changde introduced capitation payment system to reimburse inpatient expenditure concurrently with the introduction of URBMI, while maintaining FFS for other insurance programs and for the uninsured. The insurance fund in Changde URBMI was divided into three parts to reimburse inpatient care: capitation fund, the equalization fund, and the preservation fund. The capitation fund makes up the large majority of the URMBI budget, accounting for 87% in 2008. The revenue and expenditure of URBMI insurance fund is operated by each districts in Changde, while the city bureau is responsible for management and regulations. Therefore, each district is responsible for paying hospitals, on a monthly basis, based on the same per capita base rate set by city bureau each year, but each payment differs according to the number of contracts. If the actual expenses are above the allocated budget, the hospital must bear the extra cost itself. If they are lower than the budget, the hospital may keep the surplus as a profit.

12

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

Therefore, under the Changde system, hospital would have a financial incentive to contain costs. The equalization fund is managed by city bureau to compensate the loss of small hospitals, such as mental health facilities with small risk pool. It differs essentially from risk-adjusted systems because the risk is managed ex post by equalization fund and at institution level. Risk-adjust system was not a viable option in Changde due to data limitations and management incapacities. The reservation fund was created to reimburse providers for performing special outpatient services. Special outpatient services reimbursement in Changde was designed to reimburse common chronic diseases with heavy disease burden at outpatient setting, such as diabetes or hypertension. In 2009, Changde extend its URBMI coverage to general outpatient services, but the reimbursement for general outpatient care is too minimal (about CNY 30 per month) to be considered in this paper. 2.3. Open enrollment and other supporting policies The major supporting policy for capitation in Changde is open enrollment. It means the URBMI beneficiaries could freely select a URBMI in-network provider as their first provider when seeking inpatient care. The insureds are allowed to change their enrollment once a year. Under the contract, beneficiaries cannot be reimbursed unless they seek care from their first providers or being referred from their first providers. The URBMI in-network providers are responsible for all inpatient costs incurred by their beneficiaries including referral to other providers. In 2008, there were 287 hospitals in Changde, 259 (90.24%) were listed as UEBMI in-network hospitals, while 154 (53.66%) were contracted with URBMI (Table 1). The enrollment is not evenly distributed across hospitals. As the URBMI fund is operated by each districts, so it is better to look at the distribution in districts. Wuling District is the central district which locates the only two tertiary hospitals in Changde city. Taking Wuling district as example, the enrollment concentrated in large hospitals, including 40.44% in tertiary hospitals and 55.70% in secondary hospitals in 2008 (Table 2). Open enrollment facilitates capitation in two ways. First, it makes calculating hospital budget possible. Usually Chinese patients may come to different hospitals seeking medical care without referral, so the number of patients a hospital covers is not fixed. With open enrollment, the hospital budget equals to the per capita base rate multiplied by the number of beneficiaries enrolled. Secondly, open enrollment creates incentives for hospitals to improve efficiency and quality in order to increase their profits, as the profits equal the average residuals for each enrollment resident multiplied by the number of contracts. Besides open enrollment, Changde does not use drug lists or deductibles as supplementary policies to capitation. Therefore, the present paper defines Changde’s payment reform project as a comprehensive set of policy measures including capitation, open enrollment, and other supporting policies, but with the capitation payment as the core component.

Table 1 Number (and percentage) of hospitals in Changde, UEBMI in-network list and URBMI in-network list (2008). Hospital level

Changde city

UEBMI in-network

URBMI in-network

Tertiary Secondary Primary All

2 36 249 287

2 36 221 259

2 30 122 154

(100%) (100%) (88.76%) (90.24%)

(100%) (83.33%) (49.00%) (53.66%)

Table 2 Distribution of enrollment in Wuling District, Changde city (2008). Hospital level

Number of hospitals

Number of contracts

Percentage

Tertiary hospitals Secondary hospitals Primary hospitals Total

2 8 10 20

31,837 43,854 3044 78,735

40.44% 55.70% 3.87% 100.00%

3. Method 3.1. Data The data used for this paper is from the URBMI household survey. In 2007, commissioned by the State council, the Peking University China Center for Health Economics Research (CCHER) conducted the URBMI household survey. Starting with the baseline survey in November 2007, the project has been scheduled for four consecutive years till 2010, covering nine representative cities. They are Baotou City of Inner-Mongolia SAR, Changde City of Hunan Province, Chengdu City of Sichuan Province, Jilin City of Jilin Province, Shaoxing City of Zhejiang Province, Xiamen City of Fujian Province, Xining City of Qinghai Province, Urumqi City of Xinjiang SAR, and Zibo City of Shandong Province. Using Probability Proportion to Size (PPS) sampling technique, there were 141 communities, 42 districts from nine cities amongst the 79 pilot cities in 2007 included in the final survey (Lin et al., 2009). The data in 2007 is not included in the final analysis, because the baseline survey was conducted in November 2007, immediately after URBMI starting its pilot. The period was too short for any policy intervention taking evident effect. The final analysis uses the data from 2008, 2009, and 2010, with original sample size of 5018 inpatient admissions. The objective of this survey is to understand the coverage and effects of the basic medical insurance. The survey includes information on the population demographics, individual health behaviors and status, health insurance status, utilization and satisfaction for outpatient and inpatient services, and household socioeconomic status etc. In addition, we also have the inpatient diagnosis coded according to ICD-10 by senior doctors, which provides solid database for research of this paper. All survey procedures were performed in compliance with the Statistics Law of China, and respondents’ private information was strictly confidential. By comparing inpatient bed per capita, doctors per capita and GDP per capita between Changde and the other cities, it seems the both economic development and health care supply capacities in Changde lagged behind of the other cities (Table 3). During the

Table 3 Comparison between Changde and other 8 cities in health and economic development (2008). City

Beds per 1000 population

Doctors per 1000 population

Per capita GDP (Yuan)

Changde Other 8 city average Baotou Chengdu Jilin Shaoxing Xiamen Xining Wulumuqi Zibo 9 city average

2.42 4.81 4.44 4.30 4.35 3.19 3.89 5.10 9.08 4.14 4.54

1.31 2.69 2.59 2.40 2.44 2.04 2.75 2.64 4.39 2.27 2.54

19,201.00 43,768.25 70,004.00 30,855.00 30,016.00 48,236.00 62,651.00 19,494.00 37,343.00 51,547.00 41,038.56

Source: health statistics year book of each province 2009.

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

13

Table 4 URBMI initiation date, payment policies and changes in the 9 surveyed cities. City

Baotou

Changde

Chengdu

Jilin

Shaoxing

Xiamen

Xining

Urumqi

Zibo

URBMI Initiation Date UEBMI Payment (2008e10) URBMI Payment (2008e10) Dramatic Payment Change (2008e10)

2007.9 FFS* FFS* No

2007.10 FFS Capitation No

2007.1 FFS FFS No

2007.4 FFS FFS No

2007.1 FFS FFS No

2007.1 FFS* FFS* No

2007.7 FFS FFS No

2007.11 FFS FFS No

2007.10 FFS* FFS* No

Baotou, Xiamen, and Zibo use case-based payment for 10e20 diseases, but still with FFS as their major payment method, so the case-based payment has little overall impact in each city.

study period between 2008 and 2010, Changde is the only city among the 9 surveyed cities carrying out such thorough payment experiment through URBMI program as well as in UEBMI program. There was no drastic policy change regarding provider payments during the period between 2008 and 2010 in all 9 cities (Table 4). Table 5 presents the descriptive statistics of inpatient-related information of URBMI patients and UEBMI patients in Changde and other surveyed cities (2008e10). Student t-test was used to examine whether the difference of URBMI or UEBMI between Changde and other cities are statistically significant. We find that the total inpatient medical expenditure, length of stay, and out of pocket expenditure of Changde URBMI are all statistically lower than that of other city’s URBMI, and the inpatient total medical expenditure, out of pocket expenditure, and out of pocket ratio are all statistically lower in Changde UEBMI than other cities’ UEBMI. The rest differences are not statistically significant according to ttest. 3.2. Model specifications In October 2007, Changde implemented the URBMI pilot program and at the same time began using capitation to reimburse inpatient care for URMBI enrollees’ inpatient services. Therefore it is difficult to distinguish the effects of the payment reform from the insurance coverage expansion through a before-and-after comparison because both events happened simultaneously. It is possible though, to isolate the effect of just the payment reform on the dependent variables by comparing Changde’s URBMI program with capitation and other cities’ URBMI programs with mostly FFS, to examine the impact of payment reform. The straightforward approach is to calculate the difference in dependent variables between Changde URBMI and other cities’ URBMI. However, one of the potential problems is that along with the effects caused by payment reform, the dependent variables are also associated with some underlying differences between Change and other cities (such as the population health, health service consumption, and price). In an effort to control for the city fixed effects, we assume the differences between Changde’s UEBMI and other cities’ UEBMI are a good proxy for the magnitude of effects caused by inherent differences between cities. Changde’s UEBMI program,

like all other cities included uses a FFS reimbursement plan, produces similar incentives for both patients and providers. The difference between Changde’s URBMI program and other cities’ URBMI program, once the difference between cities has been controlled for, is considered as the policy effect of Changde’s payment reform in this paper. The Difference-in-Difference model is:

Yit ¼ a0 þ a1 Dit þ a2 Cit þ a3 ðDit  Cit Þ þ a4 Zit þ εit

(1)

where Yit is the inpatient medical expenditure, the out of pocket expenditure and its share as percentage of the total expenditure, drug share as percentage of the total expenditure, the inpatient length of stay, treatment effect (Donabedian, 1988), and patient satisfaction (Aharony and Strasser, 1993; Carey and Seibert, 1993) for patient i in the year t (2008e2010). Dit is the city dummy variable to distinguish the effect of cities where different payment policies adopted, with 1 for Changde with payment reform in URBMI and 0 for other cities without thorough reform in URBMI or UEBMI. Ct is the insurance dummy variable with 1 for URBMI and 0 for UEBMI. ðZit Þ is the characteristics of patient in the year t, including age, gender, household income, education, chronic conditions, self-rated health conditions and types of diagnosed diseases using ICD-10 classification. The coefficient of the interaction term of city and insurance dummy is the difference between Changde URBMI with payment reform and other cities’ URBMI without, which can be interpreted as the policy effect of payment reform in Changde URBMI. It is worth noting that the payment reform will have a proportional, not additive, effect on medical expenditure and length of stay. In other words, reducing the length of stay from 12 days to 11 days is easier than reducing to from 2 days to 1 day. The same applies to medical expenditure. This is a theoretical reason to take the logarithm of the dependent variable of total inpatient medical expenditure, out of pocket expenditure, and length of stay (Norton et al., 2002). In this paper, out-of-pocket (OOP) expenditure is calculated by subtracting reimbursement from total inpatient expenditure, same in both insurance schemes. Both reimbursement and total expenditure are documented in the household survey.

Table 5 Descriptive statistics of inpatient information from Changde and other cities (2008e11). Changde

Other cities

URBMI

Inpatient (Yuan) Drug Ratio (%) Length-of-Stay (Day) Out-of-Pocket (Yuan) Out-of-Pocket ratio (%)

UEBMI

URBMI

UEBMI

Mean

Standard deviation

Mean

Standard deviation

Mean

Standard deviation

Mean

Standard deviation

4598.26*** 72.95 11.63*** 3330.32*** 66.44

9088.33 30.03 11.06 8779.36 29.24

7222.42*** 72.07 17.53 2870.13*** 51.72***

15,319.60 30.64 18.41 4581.27 34.93

7213.58 72.16 15.91 5169.84 69.24

11,424.91 30.41 20.65 9202.94 28.17

9706.50 70.19 19.04 4370.49 43.70

15,178.80 36.67 24.02 9641.56 33.50

*** means the difference between Changde and other cities is significant at 1% level. The comparison is made between the Changde URBMI and other cities’ URBMI, and between Changde UEBMI and other cities’ UEBMI.

14

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

4. Results 4.1. Regression results Table 6 presents the parameter estimates for the policy impact on outcome variables of our interests, including inpatient medical expenditure, out of pocket expenditure and it share as the total inpatient medical expenditure, drug-to-total expenditure ratio based on OLS model, and treatment effect and patient satisfaction based on Order Probit model. Firstly, the coefficients of key independent variable (the policy effect, i.e. the interaction term of city and insurance dummy variables) from model (1) to model (5) show interesting results. The coefficients of total inpatient expenditure and the drug cost’s share are 0.032 and 0.020 respectively; both are negative but are not statistically significant even at 10% level. The coefficient for inpatient length of stay is 0.177 and is significant at 1% level. The coefficients of treatment effect and patient satisfaction are 0.033 and 0.054 respectively, though neither of them is significant even at 10% level. Secondly, although the coefficients of key independent variable is not statistically significant as in model (1), the coefficients of city and insurance dummy are 44.6% and 13.1%, all significant at 1% level, indicating that the total inpatient medical expenditure is lower in Changde, a city with less developed economy, than other surveyed cities by almost a half, meanwhile, the same indicator is lower for URBMI patients than UEBMI patients in all nine surveyed cities, possibly due to limited financing level and different enrollee characteristics. As for the controlled variables, the results show that the male, highly educated, elders, respondents with higher household income, and those self-rated poor in health conditions have higher inpatient medical expenditures. Thirdly, as seen from model (3), the inpatient length of stay in Changde is shorter than the rest surveyed cities by 10.5%, and the

same indicator is shorter for URBMI patients than UEBMI patients from all nine surveyed cities by 12.3%, both are statistically significant at 1% level. The men, elders, those with chronic conditions and with poor self-rated health conditions have longer length of stay. Last but not the least, as in model (6) and model (7), the coefficients of out-of-pocket payment (after controlling for total medical expenditure) and its share as the total expenditure are 0.197 and 0.095 respectively, both are significant at 1% level. In model (6), the coefficients of city and insurance are 0.128 and 0.396 respectively, and are both significant at 1% level, indicating that out-of-pocket payment for Changde inpatient patients is higher than their counterparts in other cities by 12.8%, at the same time, the same indicator is 39.6% higher for URBMI patients than their UEBMI counterparts in all nine cities. Among the coefficients of controlled variables, gender, age, household income, and selfrated health conditions have significant impact on out-of-pocket payment. Model (7) shows similar results in spite of differences in magnitude. 4.2. Robust test According to the above regression results, the self-rated health conditions have evidently significant impact on all outcome variables of interests. People with different health status might also differ in their medical seeking behaviors, health service utilizations, and medical expenditures; on the other hand, in-network providers might select patients according to their health status. Therefore, we divide the full sample into two groups according to their self-rated health conditions, with the first group including those with fair, poor, and very poor health conditions, and the second cover the rest people with good and very good self-rated health conditions. Separate regressions are conducted based on the two subsamples to examine the potential differences in coefficients.

Table 6 Parameter Estimates for the policy impact of payment reform in Changde.

City Insurance Interaction Gender (Female ¼ 1) Education (years) Age Ln (household income) Chronic conditions Self-rated health Year 2010 Year 2009

Model (1)

Model (2)

Model (3)

Model (4)

Model (5)

Model (6)

Model (7)

Ln (total cost)

Drug cost ratio

Ln (length of stay)

Treatment effect

Patient satisfaction

Ln (out-ofpocket)

Out-of-pocket ratio

0.446*** [0.060] 0.131*** [0.043] 0.032 [0.089] 0.079** [0.032] 0.015*** [0.004] 0.007*** [0.001] 0.119*** [0.024] 0.057 [0.041] 0.048*** [0.018] 0.083** [0.038] 0.011 [0.039]

0.014 [0.024] 0.023 [0.017] 0.020 [0.036] 0.002 [0.013] 0.001 [0.002] 0.000 [0.000] 0.001 [0.009] 0.017 [0.016] 0.020*** [0.007] 0.036** [0.015] 0.024 [0.015]

0.105*** [0.038] 0.123*** [0.028] 0.177*** [0.058] 0.099*** [0.021] 0.001 [0.002] 0.005*** [0.001] 0.004 [0.015] 0.111*** [0.026] 0.049*** [0.011] 0.016 [0.025] 0.002 [0.025]

0.101 [0.067] 0.013 [0.051] 0.033 [0.103] 0.040 [0.037] 0.001 [0.004] 0.007*** [0.001] 0.170*** [0.028] 0.585*** [0.048] 0.323*** [0.021] 0.046 [0.045] 0.012 [0.045]

0.084 [0.062] 0.074 [0.046] 0.054 [0.094] 0.100*** [0.034] 0.002 [0.004] 0.005*** [0.001] 0.018 [0.025] 0.064 [0.043] 0.174*** [0.018] 0.053 [0.040] 0.014 [0.040]

0.072*** [0.020] 0.211*** [0.014] 0.095*** [0.029] 0.018* [0.011] 0.001 [0.001] 0.002*** [0.000] 0.039*** [0.008] 0.003 [0.014] 0.020*** [0.006] 0.066*** [0.013] 0.048*** [0.013]

3985 0.147

3296 0.009

4934 0.134

4612

4673

0.128*** [0.045] 0.396*** [0.032] 0.197*** [0.066] 0.044* [0.024] 0.003 [0.003] 0.003*** [0.001] 0.083*** [0.018] 0.002 [0.031] 0.039*** [0.013] 0.113*** [0.029] 0.083*** [0.029] 0.935*** [0.012] 3620 0.658

Ln (total cost) Observations R-squared

3956 0.140

All models controlled for disease types using ICD-10; t-statistics in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1 Model (1) (2) (3) (6) (7) are based on OLS while Model (4) (5) are based on Order Probit.

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

15

Table 7 Parameter Estimates based on the subsample of fair-, poor-, or very poor- self-rated health conditions.

City Insurance Interaction Observations R-squared

Model (1)

Model (2)

Model (3)

Model (4)

Model (5)

Model (6)

Model (7)

Ln (total cost)

Drug ratio

Ln (length of stay)

Treatment effect

Satisfaction

Ln (out-of-pocket)

Out-of-pocket ratio

0.445*** [0.071] 0.105** [0.050] 0.001 [0.103] 2842 0.166

0.020 [0.031] 0.030 [0.022] 0.060 [0.045] 2374 0.016

0.118** [0.046] 0.111*** [0.033] 0.136** [0.067] 3412 0.154

0.090 [0.086] 0.023 [0.065] 0.063 [0.131] 3261

0.083 [0.078] 0.092 [0.057] 0.056 [0.116] 3304

0.145** [0.058] 0.386*** [0.040] 0.258*** [0.081] 2555 0.626

0.072*** [0.024] 0.225*** [0.018] 0.124*** [0.036] 2818 0.151

All models controlled for types of disease using ICD-10; t-statistics in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1; Model (1) (2) (3) (6) (7) are based on OLS while Model (4) (5) are based on Order Probit; the same control variables were used as in Table 6. Not all results were reported here due to page limits but were available upon requests.

Table 7 presents the parameter estimates based on the first sample of fair, poor, or very poor health conditions. The results from the relatively poor health subsample present a similar pattern with the results based on the full sample, that the out-of-pocket payment, out of-pocket ratio and length of stay are all reduced while the total inpatient medical expenditure remain unchanged. As for the population cohort with good and very good self-rated health conditions, the payment reform in Changde has little impact on either providers or patients. Results are not presented but are available upon request to authors. 5. Conclusion and discussions According to literatures, capitation is associated with changes in costs (Manning et al., 1984; Dickey, 1997; Bloom et al., 2002); other types of prospective payment reforms are also associated with changes in length of stay and quality of care (Kroneman and Nagy, 2001; Wang et al., 2011; Shmueli et al., 2002; Norton et al., 2002). In this paper, we found that the payment reform in Changde was associated with a 19.7% reduction of the inpatient out-of-pocket cost, 9.5% reduction of out-of-pocket ratio, and 17.7% reduction of length of stay. However, the total inpatient costs, drug cost ratio, treatment effect, and patient satisfaction exhibit little difference between the Fee-For-Service and capitation models. We conclude that the payment reform in Changde did not reduce overall inpatient expenditure, but it decreased the financial risk and length of stay of inpatient patients without compromising quality of care. As for cost-shifting, we added to examine the behavior change of shifting to outpatient care which continues to be paid by FFS. We found outpatient expenditure remain unchanged while outpatient utilizations increased, indicating providers respond to capitations by shifting to outpatient utilizations. The lack of decrease in overall in-patient medical expenditure is not consistent with the theory surrounding the use of capitation or much of the previous empirical literature (Manning et al., 1984; Dickey, 1997; Bloom et al., 2002), as capitation has been widely thought to be associated with under-utilization of health services and thus less medical expenditure. Three explanations are responsible for the lack of decline. First, URBMI is relatively new and providers have not yet adapted behavior. So we further examined how the policy effect varies across years but found no dynamic trend over time from 2008 to 2010. It might take even longer time to observe the possible impact on cost-containment, which remains a research topic in the future. Secondly, URBMI only account for a small share of entire social health safety net and reform only happen in URBMI have not yet produced overall impact. In 2011, the total revenue of URBMI was CNY 59.42 billion in China, as compared with CNY 494.5 billion for UEBMI and CNY 204.85 billion for NRCMS (MoHRSS, 2013; NHFPC, 2013). It

indicates payment reform should be parallel among three insurance programs, if the three insurance programs are not integrated into one. Thirdly, the lack of autonomy and independence of public hospital makes it less responsive to economic incentives (Pan et al., 2013). All these reasons above could compromise the incentives built into capitation system, which eventually lead to the null finding in cost-containment. Apart from the lack of reduction in overall expenditure, it is also worth noting that the inpatient length of stay fell significantly after the introduction of capitation. There are previous literatures documenting how reimbursement affects length of stay (Norton et al., 2002). In this paper, we found from our field trips that hospital managers dealt with financial risk (from capitation) by imposing strict requirements in average length of stay and average inpatient expenditure. The former approach turned out to be more effective according to our regression results. In response to the requirements, doctors tend to reduce the amount of defensive medicine at inpatient settings such as rechecks. These behavioral changes eventually led to decreased LOS identified. In this paper, we also focus on the impact of payment reform on patients, a perspective lacking in much of the previous capitation literature. The data show that the financial burden of seeking medical care is reduced for patients, so the problem of “kanbinggui” was alleviated under Changde’s capitation reform. Equally important, treatment effect and satisfaction showed no significant difference from cities using the fee-for-service system, which were mostly contributed by its complementary policies of open enrollment (Mills et al., 2000). Open enrollment might prevent cream skimming when it is associated to capitation. It could be hypothesized that patient choice promoted by open enrollment induces a competition between providers, which guarantees quality of care whiling reducing financial risk. This finding could open future research tracks on the ability of open enrollment to compensate for capitation drawbacks. The study is also not without limitations. Firstly, there have been only 6 years since the earliest payment reform in Changde, and we were limited to looking at only the first 3 years of the reform. In the future, it will be necessary to further study dynamic trends over a longer period of time. Secondly, due to limitations in dataset and policy setting, we were unable to direct elicit the impact of just payment reform. Our identification strategy uses UEBMI differences to capture city differences, which might overlook some nonUEBMI factors that could affect outcomes of interest. As time passes by, there will be more local payment reforms in China to provide fertile ground for research. Acknowledgments The authors are grateful to Sam Krumholz for editing early draft, to three anonymous referees for valuable comments, and to the

16

C. Gao et al. / Social Science & Medicine 111 (2014) 10e16

Human Resources and Social Security Bureau of Changde city for cooperation in field interviews. References Aharony, L., Strasser, S., 1993. Patient satisfaction: what we know about and what we still need to explore. Medical Care Review (1), 49e77. Bloom, J.R., Hu, T.W., Wallace, N., Cuffel, B., Hausman, J.W., Sheu, M.L., Scheffler, R., 2002. Mental health costs and access under alternative capitation systems in Colorado. Health Services Research 37 (2), 315e340. Carey, R.G., Seibert, J.H., 1993. A patient survey system to measure quality improvement: questionnaire reliability and validity. Medical Care (9), 834e845. CPC Central Committee and the State Council, 2009. Opinions of the CPC Central Committee and the State Council on Deepening the Health Care System Reform. http://www.china.org.cn/government/scio-press-conferences/2009-04/09/co ntent_17575378.htm. Accessed on 30.01.12. Dafny, L., 2005. How do hospitals respond to price changes? American Economic Review 95 (5), 1525e1547. Dickey, B., 1997. Assessing cost and utilization in managed mental health care in the United States. Health Policy 41 (suppl.), S163eS174. Donabedian, A., 1988. The quality of care how can it be assessed? The Journal of the American Medical Association, 1743e1748. Eggleston, K., et al., 2008. Health service delivery in China: a literature review. Health Economics 17, 149e153. Hu, S.L., et al., 2008. Reform of how health care is paid for in China: challenges and opportunities. The Lancet (372), 1866e1953. Kroneman, M., Nagy, J., 2001. Introducing DRG-based financing in Hungary: a study into the relationship between supply of hospital beds and use of these beds under changing institutional circumstances. Health Policy, 19e36. Lin, W.C., Liu, G.G., Chen, G., 2009. The urban resident basic medical insurance: a landmark reform towards universal coverage in China. Health Economics 18, S83eS96. Manning, W.G., Leibowitz, A., Goldberg, G.A., Rogers, W.H., Newhouse, J.P., 1984. A controlled trial of the effect of a prepaid group practice on use of services. New England Journal of Medicine 310 (23), 1505e1510. McClellan, M., 2011. Reforming payments to healthcare providers: the key to slowing healthcare cost growth while improving quality? Journal of Economic Perspectives 25 (2), 69e92. McGuire, T.G., 2010. Physician agency. In: Culyer, A.J., Newhouse, J.P. (Eds.), Handbook of Health Economics. Elsevier, Amsterdam (Chapter 9). Mills, A., et al., 2000. The response of providers to capitation payment: a case-study from Thailand. Health Policy 51, 163e180.

MoHRSS, Ministry of Human Resources and Social Security, 2013. 2012 China Labor Statistics Yearbook. MoHRSS, Ministry of Human Resources and Social Security, 2011. Opinions of MOHRSS on Advancing the Health Insurance Payment Method Reform. Document No. 63 (in Chinese). http://www.mohrss.gov.cn/page.do?pa¼40288020240500280124088 2b84702d7&guid¼bfa0c812467e4687be4740b80bd75b4a&og¼40288020246f91 830124766485391b1c. Accessed on 30.01.12. Moreno-Serra, R., Wagstaff, A., 2010. System-wide impacts of hospital payment reforms: evidence from Central and Eastern Europe and Central Asia. Journal of Health Economics, 585e602. NHFPC, National Health and Family Planning Commission, 2012. Notice on Promote the Work of 2012 NRCMS (in Chinese). http://www.moh.gov.cn/publicfiles/busi ness/htmlfiles/mohncwsgls/s3581/201205/54885.htm. Accessed on 1.06.12. NHFPC, National Health and Family Planning Commission, 2013. 2012 China Health Statistics Yearbook. Norton, et al., 2002. Does prospective payment reduce inpatient length of stay? Health Economics 11, 377e387. Pan, J., et al., 2013. How does separating government regulatory and operational control of public hospitals matter to healthcare supply? China Economic Review 27, 1e14. Schmidt, H.H., et al., 2011. How payment systems affect physicians’ provision behavior-An experimental investigation. Journal of Health Economics 30, 637e 646. Shmueli, A., et al., 2002. The effects of introducing prospective payment to general hospitals on length of stay, quality of care, and hospitals’ income: the early experience of Israel. Social Science & Medicine 55, 981e989. Sun, Z.G., 2011. Interview with the Head of State Council Health Care System Reform Office (in Chinese). http://www.sdpc.gov.cn/shfz/yywstzgg/ygdt/t20111220_ 451627.htm. Accessed on 30.01.12. Tan, G.J., 2009. Initial Exploration of the Market First Admission Policy. China Health Insurance (in Chinese). The State Council, 2009. Essential Implementation Schemes on Health Care System Reform (2009e2011) (in Chinese). http://www.gov.cn/zwgk/2009-04/07/conte nt_1279256.htm. Accessed on 30.01.12. Wang, et al., 2011 Dec. An Experiment in payment reform for doctors in rural China reduced some unnecessary care but did not lower total costs. Health Affairs 30 (12), 2427e2436. Yao, H., 2011. The National Insurance Spending is Estimated to Account for More Than 50% of the Medical Institutions Revenue (in Chinese). http://finance.ifeng. com/stock/roll/20110515/4023588.shtml. Accessed on 30.01.12. Yip, W., Eggleston, K., 2001. Provider payment reform in China: the case of hospital reimbursement in Hainan province. Health Economics, 325e339. Yip, W., Hsiao, W., et al., 2012. Early appraisal of China’s huge and complex healthcare reforms. Lancet 379, 833e842.

Payment reform and changes in health care in China.

This paper is intended to assess the primary effects on cost, utilization and quality of care from payment reform of capitation and open enrollment in...
330KB Sizes 0 Downloads 3 Views