Effects on the medical revenue of comprehensive pricing reform in Chinese urban public hospitals after removing drug markups: case of Nanjing

Abstract

Aims: The State Council of China requires that all urban public hospitals must eliminate drug markups by September 2017, and that hospital drugs must be sold at the purchase price. Nanjing-one of the first provincial capital cities to implement the reform—is studied to evaluate the effects of the comprehensive reform on drug prices in public hospitals, and to explore differential compensation plans.

Methods: Sixteen hospitals were selected, and financial data were collected over the 48-month period before the reform and for 12 months after the reform. An analysis was carried out using a simple linear interrupted time series model.

Results: The average difference ratio of drug surplus fell 13.39% after the reform, and the drug markups were basically eliminated. Revenue from medical services showed a net growth of 28.25%. The overall compensation received from government financial budget and medical service revenue growth was 103.69% for the loss from policy-permitted 15% markup sales, and 116.48% for the net loss. However, there were large differences in compensation levels at different hospitals, ranging from –21.92% to 413.74% by medical services revenue growth, causing the combined rate of both financial and service compensation to vary from 28.87–413.74%, There was a significant positive correlation between the services compensation rate and the proportion of medical service revenue (p < .001), and the compensation rate increased by 8% for every 1% increase in the proportion of services revenue.

Discussion: Nanjing’s pricing and compensation reform has basically achieved the policy targets of eliminating the drug markups, promoting the growth of medical services revenue, and adjusting the structure of medical revenue. However, the growth rate of service revenue of hospitals varied significantly from one another.

Conclusions: Nanjing’s reform represents successful pricing and compensation reform in Chinese urban public hospitals. It is recommended that a differentiated and dynamic compensation plan should be established in accordance with the revenue structure of different hospitals.

Keywords:

• Urban public hospitals
• pricing and compensation reform
• zero drug markups policy
• effects
• interrupted time series
• Nanjing
• China

Background

China’s newest round of healthcare reforms began in 2009, and has achieved significant success and global attention. However, the reform objectives (in various fields) have not yet resolved the problem of the excessive and rapid growth of public hospital costs^1,2. In September 2017, the State Council of China began to investigate the impact of the public hospital reform. This includes examining whether public hospitals have managed to comprehensively control the proportion of drugs in medical revenue, and whether the proportion of medical services revenue has increased. Historically, there was a special policy in Chinese public hospitals that allowed government-owned medical institutions to gain additional profit from 15% drug markups to compensate for insufficient financial investment and reimbursement for medical service provision^3–5. This policy later became an important reason for the rapid growth of medical expenses in China, the excessive proportion of drug revenue, the excessive use of drugs, and the abuse of antibiotics⁶. The historical reason for this policy lay in the fact that Chinese treasury was gradually empty after the founding of New China in 1949 because of both the central and local governments’ excessive pursuit of an economic development that relied primarily on industrial production. This led to a lack of financial investment in the medical and hygiene fields. This period of history can be divided into three stages. At the beginning of the founding of New China in 1949, the government was fully responsible for the profit and loss of public hospitals. The government was responsible for all investment regarding infrastructure, beds, personnel, and other related elements. In 1954, with a change of economic policies, the financial resources allocated to support the operation of medical institutions were no longer sufficient. At this time, the “Policy on Drug Markups” was introduced: the state allowed medical institutions to add a further 15% to the sale price of drugs so as to ease their financial pressure. It is worth mentioning that, at the same time, the medical security issue of farmers (accounting for 80% or more of the total population) was also financed and aided by self-founded collective organizations. Since the 1980s, the government has not been fully responsible for hospital revenue and expenditure, resulting in budget variances for hospitals. First, hospitals used their own revenue to offset the expenditure. If the revenue cannot offset the expenditure, then the balance is included in the national budget and is compensated with financial allocation. This management method is applicable to public institutions that have regular business revenue, but still cannot offset expenditure with revenue⁷. The financial subsidies can decrease without changing the ownership of the hospitals. Therefore, the proportion of the financial investment has inevitably and gradually decreased from 30% to less than 10% since that period. Until 2008, revenue from medications accounted for 46% of business revenue, far higher than the average around the world⁸.

Maintaining the operation of public hospitals by relying on drug markups presents serious issues. Allowing markups for medications led to an excessively rapid increase in medication costs and limited access. Chinese hospitals’ outpatient clinics and inpatient departments required prescriptions to be filled at their own institutions, resulting in near monopolies for medications. This resulted in substantial increases in overall medical expenses. Furthermore, the proportion of revenue from medications was out of line with other sources of revenue. Thus, hospitals became reliant on the pharmacy to generate revenue for the entire institution. The second negative aspect was the supplier-seduces demand from medical staff. Because of a lack of financial investment, the public welfare of hospitals was harmed and profitability was too high. To increase revenue and expand services, hospitals rated clinical departments by the quantity of drugs prescribed by doctors, encouraging medical staff to prescribe medications perhaps needlessly. This resulted in overuse of medications, including excessive antibiotic prescribing as well as other therapies. Thus, doctors’ behavior gradually deviated from the patients’ best interests.

The goals of the important comprehensive reform of pricing and compensation within public hospitals (referred to as the pricing and compensation reform) were to eliminate drug markups and to encourage appropriate use of medical services. The timeline of limiting medication markups in public hospitals was as follows: primary healthcare institutions → county-level public hospitals → public hospitals in medical reform pilot cities → all urban public hospitals⁹. During the first stage, when the new medical reform was launched in 2009, basic medical and health institutions (e.g. urban community healthcare centers, rural township hospitals, and village clinics) immediately implemented the policy of “zero profit for in-hospital drug sales”. This was done for medications listed in a National essential medicine list (with a total 307 classes) and local supplement lists (100–200 classes). Medications within these classes were required to be sold at the purchased price¹⁰. This round of reform was complete by the end of 2011¹¹. Urban and rural residents benefited greatly, but, because of the greatly compressed profit margins of pharmaceutical companies and medical institutions, medications were frequently out of stock, had limited distribution, or were not marketed. Residents were often still required to visit large hospitals because of drug shortages. In the second stage (from 2012–2013), county hospitals (the best public hospitals in rural areas) began to implement the zero-markup policy for drug sales¹². The policy was extended from public hospitals in the pilot counties to all county-level public hospitals in 2 years. During that period, the policy was successfully introduced in most rural areas. This had a number of effects: first, it reduced the speed of the increase of medical expenses, and the proportion of drug-selling revenue in total revenue declined; second, there was a reduction in orders for antibiotics; and, third, the medical consultation rate within county areas increased, up to 90% in some counties^13,14. In the third stage (from 2013–2014), a number of pilot cities were selected from various provinces to carry out reform of payment for urban public hospitals. In the fourth stage, the pricing and compensation reform of urban public hospitals was required to be fully implemented by September 2017.

The overall effects of this reform are evident. Across the country, from 2009–2015, the average annual growth rate of total revenue in public hospitals has been maintained at 19.3%. However, the growth rate has slowed significantly since 2012, wherein the average annual growth rate was 24.6%. Since 2015, this has declined by 25% on a year-on-year basis. From 2010–2015, the proportion of drug costs in the average outpatient medical expenditure per patient decreased from 52.3% to 48.4%, while the proportion of drug costs in the average inpatient medical expenditure per patient decreased from 43.4% to 36.9%¹⁵. Compared to that in 2012, the proportion of drug revenue in public hospitals decreased from 45.0% to 40.9% in 2015. These results are largely consistent with the timing of the policy promotion in the first three stages¹⁶. However, the policy impacts in the fourth stage are still being evaluated. There are a few empirical studies on the compensation effect from county-level public hospitals, but relatively few in urban areas^17–19. The reform in Nanjing began on October 31, 2015. This was the first city in China to implement the reform, and the policy promoted faster and covered a wider area than other cities in China. This paper aims to evaluate the effects of eliminating the drug markups on urban public hospitals with comparing the revenue structure and also compensation structure of hospitals 1 year on after terminating the drug markup policy in Nanjing.

Methods

Setting

Located in a wealthy region of eastern China, Nanjing is the capital of Jiangsu Province, with 11 administrative regions. The city covers an area exceeding 6,500 square kilometers. As of the end of 2016, Nanjing’s total population was 8.27 million, with a per capita GDP of 127,001 RMB yuan (all values presented in this paper are in RMB yuan)²⁰. The average life expectancy was 82.19 years in 2016. At the end of 2015, Nanjing had more than 77 secondary hospitals with 46,643 beds and 78,882 health field practitioners (including 65,139 health workers). There were 6.36 hospital beds per 1,000 people, and 9.97 health technicians, including 3.41 doctors and 4.42 nurses. The average number of annual medical visits by residents was 10.94, with 18.63 inpatients per 100 residents²¹. Regarding its economy, Nanjing ranks fifth in all cities of China, behind Beijing, Guangzhou, Shenzhen, and Shanghai.

A pilot project introducing pricing and compensation reform in urban public hospitals in Nanjing was implemented in district-affiliated hospitals in 2013. This was extended to municipal hospitals on October 31, 2015. Within the scope of the administrative areas, all public hospitals in the city simultaneously implemented the zero markup policy for drug sales, with decreased prices of drug, examination, and tests, and increased prices of medical services, including the previously low priced items such as pathology, treatment, surgery, and nursing. Regarding the losses as a result of withdrawing the drug markup, the policy aimed to compensate 70–80% of these losses via an increase of medical service revenue, 20% by government financial support, and any remaining deficit met by improving the hospital operation efficiency.

Sampling

In Nanjing, a total of 57 hospitals participated in the reform, including 16 provincial and ministerial hospitals, 12 municipal hospitals, 23 district-affiliated hospitals, and six other hospitals (e.g. enterprise and public institution hospitals and psychiatric hospitals). Hospitals received the financial compensation from various levels of government and/or institution based on the hospital ownership. This study focuses on 12 municipal hospitals (Drum Tower Hospital, Nanjing First Hospital, the Second Hospital of Nanjing, Nanjing Maternity and Child Health Hospital, Nanjing Integrated Traditional Chinese and Western Medicine Hospital, Nanjing Children’s Hospital, Nanjing Center Hospital, Nanjing Hospital of Chinese Medicine, Nanjing Brain Hospital, Nanjing Stomatological Hospital, Nanjing Chest Hospital, Nanjing Prevention and Treatment Center for Occupational Disease [wherein the Prevention and Treatment Center for Occupational Disease belongs to the secondary hospitals and the rest are the tertiary hospitals]). Also included in this study were four non-municipal hospitals, including two provincial/ministerial hospitals (Jiangsu Province Tumor Hospital, Zhongda Hospital affiliated to Southeast University) and two non-municipal nor -provincial/ministerial hospitals (Nanjing Chenguang Hospital and Red Cross Hospital). Among the 12 municipal hospitals, Nanjing Integrated Traditional Chinese and Western Medicine Hospital, Nanjing Hospital of Chinese Medicine were considered another group of hospitals because the zero markup policy has not been applied to Chinese medicine with the purpose to protect traditional Chinese medicine. Also, Prevention and Treatment Center for Occupational Disease was considered a special municipal hospital because it ranked the only secondary-level hospitals in and all other municipal hospitals ranked first-level hospitals. Therefore, in total the hospitals were divided into four groups for comparison: (1) municipal hospitals (n = 9): Drum Tower, Maternity and Child Health, Chest, Nanjing First Hospital, Children, Stomatological, the Second Hospital of Nanjing, Brain, Central; (2) traditional medicine hospitals (n = 2): Chinese Medicine, and the Integrated Chinese and Western Medicine; (3) provincial and ministerial hospitals (n = 2): Provincial Tumor Hospital and Zhongda Hospital affiliated to Southeast University; and (4) the other type (n = 3): Prevention and Treatment Center for Occupational Disease, Red Cross and Chenguang Hospital.

Primary outcomes: policy enforcement and compensation effect

The markup ratio of drug sales and the growth rate of medical service revenue were compared before and after the reform. This was done to evaluate the effect of policy enforcement. The loss associated with drug mark-ups was calculated after the reform using estimated and actual expenditures. Medical service revenue changes were determined in a similar manner. The definition of the indexes and calculation methods are shown in Table 1. It should be noted that there is no widely accepted definition regarding medical services revenue. This paper adopts the definition used by the State Council of China in its public hospital reform and assessment: medical service revenue = medical revenue − drug revenue − medical material revenue − examination revenue − laboratory test revenue²².

Table 1. Definition of the evaluation indexes and calculation formula.

Index	Indication	Calculation/formula
Addition ratio of drug sales	Sales profit of hospital drug	Current actual drug balance between revenue and expenditure/actual drug expenditure * 100%
Net growth rate of medical services revenue	The increase in the revenue of the service projects after excluding the natural growth trend	(Actual service revenue after reform – the estimated service revenue after reform)/service revenue before reform * 100%
Actual service revenue after reform	The sum of actual medical service revenue 12 months after reform
Predictive service revenue after reform	The sum of predictive medical services revenue in 12 months after the reform according to the data of the first 48 months before reform
Loss on elimination of drug addition (by policy statement)	Loss on elimination of drug addition calculated according to the government’s assessment statement	Drug expenditure after reform * 15%
Net loss on elimination of drug addition	Decline of potential profit brought by reform to the hospitals	Actual balance between revenue and expenditure after the reform – the estimated balance between revenue and expenditure after the reform
Actual drug balance between revenue and expenditure after the reform	The sum of actual drug revenue and expenditure 12 months after the reform
Estimated balance between revenue and expenditure after the reform	Predict the sum of medical services revenue and expenditure in 12 months after the reform according to the data of the first 48 months before reform
Financial compensation rate	Government compensation degree	Financial special compensation revenue/net loss on elimination of drug addition * 100%
Service compensation rate	Self-sufficiency degree of service growth	Net increase in service revenue/net loss on elimination of drug addition * 100%
Comprehensive compensation rate	Benefit degree/loss degree of reform	Financial compensation rate + comprehensive compensation rate

Statistical analysis

An interrupted time series (ITS) was used to predict the index levels of each month under a “pre-reform” state after the reform to give the reverse facts of scenario as “without reform”. Then, “actual values − predicted values” from all of the indexes after the reform were used to obtain the “net effect” of the reform for evaluation. A simple linear ITS model was used, which is constructed as follows: $$y={\beta }_0+{\beta }_1\times x_1+{\beta }_2\times x_2+{\beta }_3\times x_3+\epsilon ,$$ where y is the main observation index of this study, x₁ is the pre-reform time series, assigning 1, 2, 3 … x₂ as reform variables, before the reform x₂ = 0, after the reform x₂ = 1, and x₃ is the post-reform time series, assigning 49, 50 … 60. Furthermore, β₀ is a constant term, which represents the level of the observation index before the reform, and β₁ is the change trend of the observation index before the reform. The instantaneous change level of the observation index in the implementation of the reform is denoted as β₂, and β₃ is the change trend of the observation index after the reform. Therefore, after the reform, β₁ represents the predicted level of the original observation index in the pre-reform state, and β₂ + β₃ is the actual level of the observation index after the reform. The net effect of the reform can be obtained through the subtraction of the two indexes. The model diagram is shown in Figure 1.

Figure 1. Comparison of counterfactuals and actual comparison diagram via linear interrupted time series model.

Data analysis was performed using R software (V3.3.2). Anomaly values from 60 months are detected and substituted with average values, and then a simple linear ITS regression is followed for each index in a time series format, and the predicted value of the 12 months after the reform are then added up or averaged for comparison with the actual value.

Data collection

The revenue data for the 48 months (November 2011–October 2015) before the reform and the 12 months (November 2015–October 2016) after were collected from the hospital financial statements (see Table 2).

Table 2. Revenue data collection from financial reports of urban public hospitals.

Revenue account
1-1 Total revenue
1-2 Financial assistance (special compensation for drug price difference)
1-3 Medical revenue
1-3-1 Outpatient revenue
1-3-1-1Registered revenue
1-3-1-2Checkup revenue
1-3-1-3Inspection revenue
1-3-1-4Assay revenue
1-3-1-5Treatment revenue
1-3-1-6Surgery revenue
1-3-1-7Health material revenue
1-3-1-8Medicine revenue
1-3-1-8-1 Western medicine revenue
1-3-1-8-2 Chinese herbal medicine revenue
1-3-1-8-3 Chinese patent medicine revenue
1-3-1-9Medicine service fee revenue
1-3-1-10 Other outpatient revenue
1-3-2 Hospitalization revenue
1-3-2-1Bed revenue
1-3-2-2Checkup revenue
1-3-2-3Inspection revenue
1-3-2-4Assay revenue
1-3-2-5Treatment revenue
1-3-2-6Surgery revenue
1-3-2-7Nursing revenue
1-3-2-8Health material revenue
1-3-2-9Medicine revenue
1-3-2-9-1 Western medicine revenue
1-3-2-9-2 Chinese herbal medicine revenue
1-3-2-9-3 Chinese patent medicine revenue
1-3-2-10 Revenue of medicine service fee
1-3-2-11 Other hospitalization revenue

Results

Table 3 gives the drug revenue and expenditure margin before the reform (calculated as per calendar year) and the average value of groups of hospitals. Five years before the reform, the total average markup ratio of drug sales was 17.06%. The Traditional Chinese Medicine (TCM) hospitals group had the highest average surplus rate of 20.97%, followed by provincial-level hospitals of 18.01%. Of the 16 hospitals, six had an average drug surplus rate of more than 20.00%.

Table 3. Drug addition rate before reform: 2011–2015.

Hospital	2011	2012	2013	2014	2015	Averaged
Drum Tower	16.37%	14.31%	14.22%	13.88%	10.99%	13.65%
Maternity and Child Health	18.75%	18.19%	16.61%	17.12%	13.20%	16.56%
Chest	13.92%	13.64%	13.08%	14.35%	11.75%	13.26%
First Hospital	16.16%	17.26%	14.77%	14.95%	11.84%	14.79%
Children	20.41%	16.96%	14.18%	16.08%	13.44%	15.93%
Stomatological	/	14.84%	15.09%	12.00%	11.03%	12.99%
Second Hospital	/	12.13%	13.33%	13.74%	11.85%	38.68%
Brain	–0.77%	10.98%	14.52%	16.01%	13.11%	11.06%
Central	/	14.59%	15.36%	14.97%	11.57%	38.95%
Chinese Medicine	16.77%	22.31%	18.32%	15.97%	15.97%	17.73%
Integrated Medicine	36.62%	38.82%	33.60%	31.03%	19.92%	30.14%
Occupational Disease	/	24.16%	19.27%	22.37%	21.83%	44.25%
Chenguang	21.04%	20.74%	20.37%	26.08%	28.53%	23.29%
Red Cross	13.62%	6.64%	5.33%	9.97%	9.64%	8.93%
Provincial Tumor Hospital	12.33%	17.76%	11.95%	11.91%	25.16%	12.33%
Zhongda Hospital	19.26%	37.79%	14.01%	13.86%	13.03%	20.40%
Municipal level hospitals, averaged	30.67%	14.98%	14.38%	14.81%	11.92%	16.40%
TCM hospitals, averaged	20.94%	26.00%	21.97%	20.18%	17.28%	20.97%
Others, averaged	22.21%	10.35%	8.59%	12.95%	12.76%	13.03%
Provincial-level hospitals, averaged	15.03%	28.66%	12.76%	12.71%	19.92%	18.01%
Total, averaged	25.49%	19.68%	14.30%	14.48%	14.43%	17.06%

Table 4 gives the drug surplus rate after the reform. After the reform, the predicted average addition ratio of drug sales was 14.01%, while the actual average addition ratio of drug sales was 0.62%, which was a reduction of 13.39%. TCM hospitals (5.59%) and hospitals in other classes (4.09%) retained a high surplus rate, while the other hospitals sat around 0%.

Table 4. Actual and forecasted drug addition rate after reform: 2016.

Hospital	Forecasted revenue	Forecasted expenditure	Actual revenue	Actual expenditure	Forecasted balance	Actual balance
Drum Tower	149,782.79	131,993.07	131,072.13	131,169.33	13.48%	–0.07%
Maternity and Child Health	18,038.53	15,528.22	17,441.23	17,139.44	16.17%	1.76%
Chest	20,420.08	17,855.07	15,857.44	15,871.59	14.37%	–0.09%
First Hospital	94,458.58	83,273.63	78,993.50	78,564.99	13.43%	0.55%
Children	52,867.16	45,269.01	45,196.53	44,409.91	16.78%	1.77%
Stomatological	2,958.55	2,648.68	2,543.51	2,575.31	11.70%	–1.23%
Second Hospital	25,837.56	22,219.66	25,430.66	25,550.24	16.28%	–0.47%
Brain	47,492.05	40,829.69	41,644.76	42,154.03	16.32%	–1.21%
Central	13,321.00	11,924.79	12,384.85	12,334.29	11.71%	0.41%
Chinese Medicine	28,624.77	24,252.18	30,497.91	28,596.47	18.03%	6.65%
Integrated Medicine	15,879.99	12,723.40	13,436.64	13,011.91	24.81%	3.26%
Occupational Disease	755.80	589.75	637.91	594.41	28.16%	7.32%
Chenguang	2,351.50	1,881.07	2,370.00	1,902.62	25.01%	24.56%
Red Cross	13,213.28	11,849.18	10,545.71	10,524.33	11.51%	0.20%
Provincial Tumor Hospital	99,234.61	90,464.48	90,438.43	90,478.45	9.69%	–0.04%
Zhongda Hospital	78,521.64	68,898.79	69,052.52	69,053.70	13.97%	0.00%
Municipal level hospitals, averaged	47,241.81	41,282.42	41,173.85	41,085.46	14.44%	0.22%
TCM hospitals, averaged	22,252.38	18,487.79	21,967.28	20,804.19	20.36%	5.59%
Others, averaged	5,440.19	4,773.33	4,517.87	4,340.45	13.97%	4.09%
Provincial-level hospitals, averaged	88,878.13	79,681.64	79,745.48	79,766.08	11.54%	–0.03%
Total, averaged	41,484.87	36,387.54	36,721.48	36,495.69	14.01%	0.62%

Figure 2 gives an intuitive reflection of the impact of the reform on the drug surplus. During the period covering the 48th and 49th months, the drug surplus ratio in hospitals shows a sharp decline, while the predicted payment ratio shows an upward trend.

Figure 2. Fitting chart of 16 hospitals’ drug revenue and expenditure balance: November 2011–October 2016.

Table 5 gives the growth rate of medical service revenue before reform. Revenue growth of medical services in 2013–2015 was calculated. Before the reform, the revenue of medical service showed a year-on-year increase. The average growth rate in 2013–2015 was 7.10%, and the revenue of medical services showed an overall increase. Municipal hospitals showed a slower growth rate of 3.10%, while other hospitals (26.74%) and Provincial-level (23.94%) showed faster growth.

Table 5. Growth rate of medical service revenue before reform: 2013–2015.

Hospital	2013	2014	2015	Average rate
Drum Tower	–0.91%	16.31%	16.32%	7.35%
Maternity and Child Health	8.36%	16.86%	14.00%	12.53%
Chest	26.39%	14.80%	21.36%	20.46%
First Hospital	–55.27%	–1.88%	26.63%	–33.75%
Children	6.06%	–13.76%	10.95%	–4.36%
Stomatological	16.33%	10.51%	8.45%	13.38%
Second Hospital	–16.45%	13.23%	5.29%	9.73%
Brain	7.17%	14.57%	12.94%	10.81%
Central	9.89%	21.41%	27.57%	15.51%
Chinese Medicine	–2.29%	7.41%	7.18%	2.45%
Integrated Medicine	12.24%	24.79%	32.20%	18.35%
Occupational Disease	7.81%	9.66%	–5.61%	8.73%
Chenguang	20.62%	–1.22%	24.69%	20.62%
Red Cross	–8.10%	100.89%	22.03%	35.87%
Provincial Tumor Hospital	32.93%	15.59%	26.14%	23.95%
Zhongda Hospital	23.43%	21.47%	25.05%	22.45%
Municipal level hospitals, averaged	–11.71%	7.27%	15.71%	3.10%
TCM hospitals, averaged	0.98%	11.76%	14.17%	8.82%
Others, averaged	–2.65%	73.78%	20.34%	26.74%
Provincial-level hospitals, averaged	27.81%	18.65%	25.56%	23.94%
Total, averaged	–5.72%	10.78%	17.61%	7.10%

Table 6 provides the net growth rate for medical services after the reform. Pre-reform, the predicted average growth rate of medical services revenue was 8.27%, the actual growth rate was 36.51%, and the net growth rate was 28.25%. Among the different hospital groups, TCM hospitals had the fastest growth (38.64%), and the Provincial-level hospitals had the slowest (21.27%). The Chest hospital was the only one which showed a negative growth rate of –6.63%.

Table 6. Actual and forecasted growth rate of medical service revenue after reform: 2016.

Hospital	Actual revenue pre-	Actual revenue post-	Forecasted revenue post-	Actual growth	Forecasted growth	Net growth
Drum Tower	53,441.52	75,982.82	58,058.08	42.18%	8.64%	33.54%
Maternity and Child Health	27,952.13	39,276.97	30,139.44	40.52%	7.83%	32.69%
Chest	8,531.55	9,365.92	9,931.29	9.78%	16.41%	–6.63%
First Hospital	28,641.39	36,646.25	30,897.49	27.95%	7.88%	20.07%
Children	25,511.85	38,321.14	25,094.31	50.21%	–1.64%	51.85%
Stomatological	22,138.81	27,497.20	24,596.65	24.20%	11.10%	13.10%
Second Hospital	4,644.29	6,778.47	4,906.25	45.95%	5.64%	40.31%
Brain	14,331.91	20,761.21	15,627.48	44.86%	9.04%	35.82%
Central	4,884.11	6,000.24	5,284.65	22.85%	8.20%	14.65%
Chinese Medicine	10,384.43	14,389.88	11,046.97	38.57%	6.38%	32.19%
Integrated Medicine	4,759.94	7,659.82	5,150.76	60.92%	8.21%	52.71%
Occupational Disease	469.88	558.89	407.96	18.94%	–13.18%	32.12%
Chenguang	1,207.64	809.12	983.35	840.3%	21.53%	3.85%
Red Cross	6,785.73	9,286.79	7,671.31	36.86%	13.05%	23.81%
Provincial Tumor Hospital	25,266.16	29,054.60	26,759.27	14.99%	5.91%	9.08%
Zhongda Hospital	27,930.49	41,225.73	32,099.01	47.60%	14.92%	32.68%
Municipal level hospitals, averaged	21,119.73	28,958.91	22,726.18	37.12%	7.61%	29.51%
TCM hospitals, averaged	7,572.19	11,024.85	8,098.87	45.60%	6.96%	38.64%
Others, averaged	2,688.24	3,609.68	2,973.19	34.28%	10.60%	23.68%
Provincial-level hospitals, averaged	26,598.33	35,140.17	29,429.14	32.11%	10.64%	21.47%
Total, averaged	16,655.21	22,736.83	18,031.95	36.51%	8.27%	28.25%

Figure 3 intuitively reflects the continuing growth trend of medical services revenue under the original price system, and the revenue and sum of medical services in each hospital after the price change show a fault-type increase.

Figure 3. Fitting chart of medical service revenue: November 2011–October 2016.

Table 7 shows the changes to the medical revenue structure before and after the reform. Before the reform, drug sales accounted for the highest proportion of the total revenue (45%), followed by services (20%), materials (17%), image examinations (9%), and tests (9%). After excluding any natural increases, the ratio changed (in order of magnitude): services (5% increase), drugs (5% decrease), tests (1% decrease), examinations (0.26% increase), materials (0.16% increase), and others (0.12% decrease) (values rounded off in Table 7).

Table 7. Proportion of total revenue after reform: 2016 (%).

Hospital	Drug			Image examination			Test			Material			Service			Other
	Actual	Forecast	Net	Actual	Forecast	Net	Actual	Forecast	Net	Actual	Forecast	Net	Actual	Forecast	Net	Actual	Forecast	Nets
Drum Tower	0.36	0.41	–0.05	0.12	0.12	0.00	0.08	0.08	0.00	0.24	0.25	–0.02	0.21	0.16	0.05	0.01	0.01	0.00
Maternity and Child Health	0.17	0.19	–0.02	0.13	0.13	0.00	0.22	0.23	0.00	0.11	0.11	0.00	0.38	0.32	0.06	0.02	0.03	0.00
Chest	0.43	0.46	–0.03	0.06	0.05	0.00	0.16	0.15	0.01	0.11	0.08	0.02	0.25	0.22	0.03	0.06	0.08	–0.02
First Hospital	0.38	0.43	–0.05	0.10	0.09	0.01	0.08	0.09	–0.01	0.25	0.25	0.00	0.18	0.14	0.04	0.02	0.02	0.00
Children	0.38	0.45	–0.07	0.04	0.05	–0.01	0.12	0.12	0.00	0.13	0.17	–0.03	0.32	0.22	0.11	0.03	0.04	0.00
Stomatological	0.07	0.09	–0.02	0.01	0.01	0.00	0.01	0.02	0.00	0.14	0.13	0.00	0.77	0.76	0.01	0.04	0.05	0.00
Second Hospital	0.58	0.59	–0.01	0.05	0.06	–0.01	0.13	0.13	–0.01	0.09	0.11	–0.02	0.15	0.11	0.04	0.01	0.01	0.00
Brain	0.49	0.58	–0.09	0.07	0.06	0.01	0.09	0.09	0.00	0.11	0.09	0.02	0.25	0.19	0.05	0.07	0.07	0.00
Central	0.53	0.57	–0.05	0.11	0.08	0.03	0.08	0.08	0.00	0.02	0.02	0.00	0.26	0.23	0.03	0.05	0.04	0.01
Chinese Medicine	0.54	0.57	–0.03	0.06	0.05	0.01	0.08	0.08	0.00	0.07	0.08	–0.01	0.26	0.22	0.03	0.04	0.05	–0.01
Integrated Medicine	0.43	0.51	–0.07	0.11	0.12	–0.01	0.09	0.09	0.00	0.12	0.11	0.00	0.25	0.16	0.08	0.02	0.02	0.00
Occupational Disease	0.30	0.38	–0.08	0.23	0.23	0.00	0.18	0.25	–0.07	0.02	0.02	0.00	0.26	0.21	0.06	0.12	0.14	–0.02
Chenguang	0.67	0.68	–0.02	0.01	0.00	0.01	0.03	0.03	0.00	0.02	0.01	0.01	0.28	0.23	0.04	0.17	0.14	0.03
Red Cross	0.45	0.54	–0.09	0.08	0.08	0.00	0.05	0.06	–0.01	0.02	–0.02	0.03	0.40	0.32	0.08	0.04	0.02	0.01
Provincial Tumor Hospital	0.59	0.59	0.00	0.08	0.09	–0.01	0.04	0.04	0.00	0.11	0.11	0.00	0.19	0.16	0.03	0.00	0.01	0.00
Zhongda Hospital	0.39	0.44	–0.06	0.09	0.08	0.00	0.07	0.07	0.00	0.22	0.21	0.01	0.23	0.18	0.05	0.01	0.01	0.00
Municipal level hospitals, averaged	0.36	0.42	–0.04	0.09	0.09	0.00	0.10	0.10	0.00	0.18	0.19	0.00	0.26	0.20	0.05	0.03	0.03	0.00
TCM hospitals, averaged	0.50	0.55	–0.05	0.08	0.08	0.00	0.08	0.09	0.00	0.09	0.09	0.00	0.25	0.20	0.06	0.03	0.04	–0.01
Others, averaged	0.47	0.55	–0.06	0.09	0.08	0.00	0.06	0.07	–0.03	0.02	–0.01	0.02	0.37	0.30	0.06	0.06	0.05	0.01
Provincial-level hospitals, averaged	0.48	0.52	–0.03	0.08	0.08	0.00	0.06	0.05	0.00	0.17	0.16	0.01	0.21	0.17	0.04	0.01	0.01	0.00
Total, averaged	0.40	0.45	–0.05	0.09	0.09	0.00	0.09	0.09	–0.01	0.17	0.18	0.00	0.25	0.20	0.05	0.02	0.02	0.00

Table 8 shows the losses resulting from the removal of drug markups. After the reform, 16 hospitals had a predicted total loss of 181,906,800 yuan, which was lower than the policy loss of 218,994,300 yuan (accounting for 15% of the drug expenditure). Hospital losses were related to hospital level and scale. Provincial-level hospitals had a higher average loss of 92,170,900 yuan, and the remaining groups had a lower average loss of 5,018,000 yuan.

Table 8. Loss of removing drug-addition after reform: 2016.

Hospital	15% drug addition loss	Net loss compared to counterfactuals
Drum Tower	19,675.40	17,886.92
Maternity and Child Health	2,570.92	2,208.52
Chest	2,380.74	2,579.16
First Hospital	11,784.75	10,756.44
Children	6,661.49	6,811.53
Stomatological	386.30	341.67
Second Hospital	3,832.54	3,731.04
Brain	6,323.10	7,171.62
Central	1,850.14	1,345.65
Chinese Medicine	4,289.47	2,471.14
Integrated Medicine	1,951.79	2,731.85
Occupational Disease	89.16	122.54
Chenguang	285.39	40.12
Red Cross	1,578.65	1,342.72
Provincial Tumor Hospital	13,571.77	8,810.15
Zhongda Hospital	10,358.05	9,624.03
Municipal level hospitals, averaged	3,832.54	3,731.04
TCM hospitals, averaged	6,323.10	7,171.62
Others, averaged	1,850.14	1,345.65
Provincial-level hospitals, averaged	11,964.91	9,217.09
Total, averaged	21,899.43	18,190.68

Table 9 shows the financial compensation rate, service compensation rate, and combined compensation rate. Twelve months after the reform, 16 hospitals accepted special financial subsidies totaling 155.56 million yuan. Maternity hospitals and dental hospitals were not eligible for compensation according to the reform regulations. Vocational training hospitals, Chenguang Hospital and Zhongda Hospital Affiliated to Southeast University, have received no financial compensation. The financial compensation rate within the scope of municipal compensation was 21.56%, which exceeds the policy design target. The compensation rate of other hospitals (5.31%) was insufficient. Overall, the financial compensation rate was 17.76% for the 15% markup removal loss and 19.95% for the net loss.

Table 9. Compensation rate after reform: 2016.

Hospital	From government			From service growth		Combined
	Amount	Proportion of 15% loss	Proportion of net loss	Proportion of 15% loss	Proportion of net loss	Proportion of 15% loss	Proportion of net loss
Drum Tower	3,262	16.58%	18.24%	91.10%	100.21%	107.68%	118.45%
Maternity and Child Health	0	0.00%	0.00%	355.42%	413.74%	355.42%	413.74%
Chest	1,310	55.02%	50.79%	–23.75%	–21.92%	31.28%	28.87%
First Hospital	4,150	35.22%	38.58%	48.78%	53.44%	84.00%	92.03%
Children	426	6.39%	6.25%	198.56%	194.18%	204.95%	200.44%
Stomatological	0	0.00%	0.00%	750.85%	848.93%	750.85%	848.93%
Second Hospital	630	16.44%	16.86%	48.55%	49.87%	64.99%	66.95%
Brain	450	7.12%	6.27%	81.19%	71.58%	88.31%	77.86%
Central	616	33.29%	45.78%	38.68%	53.18%	71.97%	98.96%
Chinese Medicine	590	13.75%	23.88%	77.93%	135.28%	91.69%	159.15%
Integrated Medicine	418	21.42%	15.30%	128.55%	91.84%	149.97%	107.15%
Occupational Disease	0	0.00%	0.00%	169.28%	123.17%	169.28%	123.17%
Chenguang	0	0.00%	0.00%	50.12%	356.52%	50.12%	356.52%
Red Cross	80	5.07%	5.96%	102.33%	120.31%	107.40%	126.27%
Provincial Tumor Hospital	3,624	26.70%	41.13%	16.91%	26.05%	43.61%	67.19%
Zhongda Hospital	0	0.00%	0.00%	88.11%	94.83%	88.11%	94.83%
Municipal level hospitals, averaged	10,844.00	19.55%	20.53%	101.13%	106.16%	120.69%	126.68%
TCM hospitals, averaged	1,008.00	16.15%	19.37%	93.76%	112.47%	109.91%	131.85%
Others, averaged	80.00	4.10%	5.31%	97.76%	126.84%	101.86%	132.16%
Provincial-level hospitals, averaged	3,624.00	15.14%	19.66%	47.73%	61.96%	62.88%	81.62%
Total, averaged	15,556.00	17.76%	19.95%	85.93%	96.53%	103.69%	116.48%

In terms of service compensation rate, municipal hospitals (106.16%), TCM hospitals (112.47%), and other hospitals (126.84%) had a higher rate of compensation, and the rate of compensation of Provincial-level (61.96%) was relatively low. The overall service compensation rate was 85.93% for the 15% markup removal loss and 96.53% for the net loss. Six hospitals did not get fully compensated: Thoracic Hospital (–21.92%), Nanjing First Hospital (53.44%), the Second Hospital of Nanjing (49.87%), the Brain Hospital (71.58%), the Center Hospital (53.18%), and the Tumor Hospital (26.05%).

In terms of combined compensation rate, municipal hospitals (126.68%), TCM hospitals (131.85%), and other hospitals (132.16%) had a high rate of comprehensive compensation, and the provincial-level hospitals (81.62%) were relatively low. The overall comprehensive compensation rate was 103.69% for the 15% markup removal loss and 116.48% for the net loss. There were four hospitals that “suffered loss” in the reform: Chest Hospital (28.87%), the Second Hospital of Nanjing (66.95%), the Brain Hospital (77.86%), and the Provincial Tumor Hospital (67.19%). In addition, there were three hospitals which were close to insufficient compensation: Nanjing First Hospital (92.03%), the Center Hospital (98.96%), and Zhongda Hospital Affiliated to Southeast University (94.83%). In conclusion, the Chest Hospital, which was very drug-dependent, suffered the most serious damage after the reform, followed by the Brain Hospital and the Second Hospital of Nanjing, Nanjing First Hospital, and other general hospitals that were more reliant on drug sales.

Table 10 shows the goodness of fit of all prediction index models. Except for a few indicators, most have a goodness of fit higher than 0.5. The accuracy of the model prediction is the basis of the analysis, which shows that the goodness of fit is high.

Table 10. Fitness of forecast using interrupted time series model: F-value and adjusted R².

Hospital	Forecasted drug revenue		Forecasted drug expenditure		Forecasted service revenue
	F-value	Adjusted R^2	F-value	Adjusted R^2	F-value	Adjusted R^2
Drum Tower	84.24	0.8142	124.00	0.8662	110.50	0.8633
Maternity and Child Health	11.99	0.3706	23.77	0.5408	136.70	0.8790
Chest	30.42	0.6161	31.04	0.6125	59.92	0.7761
First Hospital	24.70	0.5594	38.96	0.6743	56.95	0.7670
Children	29.99	0.6169	34.31	0.6409	30.00	0.6041
Stomatological	25.09	0.5815	36.82	0.6614	49.78	0.7305
Second Hospital	2.96	0.0950	8.24	0.2832	68.97	0.7906
Brain	54.90	0.7394	36.20	0.6415	118.00	0.8645
Central	30.38	0.6290	57.39	0.7701	109.90	0.8559
Chinese Medicine	8.61	0.3010	26.77	0.5979	73.74	0.8016
Integrated Medicine	83.72	0.8351	126.50	0.8665	193.70	0.9088
Occupational Disease	7.11	0.2644	1.96	0.5820	7.63	0.2807
Chenguang	0.79	0.0130	1.14	0.0074	2.53	0.0839
Red Cross	42.38	0.6778	34.29	0.6287	115.61	0.8642
Provincial Tumor Hospital	29.27	0.6023	5.17	0.1938	94.84	0.8416
Zhongda Hospital	109.30	0.8463	0.39	0.0317	77.58	0.7957

Figure 4 shows the cluster results of the ward linkage level. The rate of service compensation can be generally divided into the following classes:

Figure 4. Service compensation rate level clustering.

• Class I, service compensation ≥100%, which is divided into two sub-classes:

  1. Sub-class 1 is represented by women and children’s hospitals and oral health hospitals, which have service compensation rates exceeding 100%; therefore, no special financial compensation is granted; and

  2. Sub-class 2 is represented by the Drum Tower Hospital, which earns its own service compensation; thus, government financial compensation can be viewed as an extra benefit.

• Class II, service compensation <100% is also divided into two sub-classes:

• Sub-class 1, where the service compensation rate is roughly 70% or below, such as Nanjing First Hospital, the Second Hospital of Nanjing, Center Hospital as the representative, which are mostly hospitals which rely comprehensively on medicine and service provision. The comprehensive compensation rate is still insufficient under the current financial compensation level; and

• Sub-class 2, where the service compensation rate is below 0%, namely the Chest Hospital. This hospital suffered “serious harm” after the price change because its revenue was very dependent on drugs sales; thus, the proportion of financial compensation should be higher.

Table 11 illustrates the relationship between the rate of service compensation and the ratio of medicine and medical service revenue. These tables explore the relations among the policy service compensation rate, net loss service compensation rate, and all ratios. The model included the “pre-reform drug ratio” and “service revenue ratio before reform”, and the model parameters are all significant (all p < .001). The R² coefficients are 0.907 and 0.916, respectively, which shows that the service compensation rate is significantly related to the structure of a hospital’s revenue. The service compensation rate is significantly correlated with the revenue ratio of medical services before the reform. A 1% increase in the revenue ratio of medical services will result in an 8% increase in the net services compensation rate.

Table 11. Association between compensation proportion of service revenue and revenue structure.

					Co-linearity
Compensation proportion of service revenue growth	Independent variable	Standardized coefficient	t-value	p-value	Allowance	VIF
Based on 15% loss	Constant	1.63	1.59	.1380	/	/
	Proportion of drug revenue pre-reform	–4.61	–2.98	.0115	0.4356	2.2959
	Proportion of service revenue pre-reform	7.98	5.27	<.001
Based on net loss	Constant	1.04	0.93	.3700	/	/
	Proportion of drug revenue pre-reform	–4.32	2.56	.0250	0.4356	2.2959
	Proportion of service revenue pre-reform	10.10	6.13	<.001

Discussion

The growth of medical services revenue is significant, but has not yet achieved the policy objective

After the reform, the predicted average markup ratio of drug sales was 14.01%, while the actual average markup ratio of drug sales was 0.62%. This represents a 13.39% decrease, showing a satisfactory policy impact. Except for TCM hospitals (5.59%) and hospitals in the remaining classes (4.09%), all other hospitals reach ∼0%. The removal of drug markups has been sufficiently implemented.

After the reform, the predicted average growth rate of medical service revenue is 8.27%, and the net growth rate is 28.25% on average (less than the year-on-year growth rate 36.51%). Revenue from medical services shows a considerable increase. In addition, the proportion of services revenue increased by 4.99%, the proportion of drug revenues decreased by 4.63%, and test revenue accounts for 0.59%. Furthermore, the examination revenue accounts for 0.26%, materials revenue for 0.16%, and all other revenue accounts for 0.12%. Overall, the effect of the price adjustment policy is remarkable. The relationship between drug revenue and service revenue has changed considerably, but that between image examinations, materials, and tests still needs further improvement.

The degree of benefit among hospitals is substantial, and the price adjustment policy requires further expansion

In terms of service compensation, municipal hospitals (106.16%), TCM hospitals (112.47%), and other hospitals (126.84%) have a higher services compensation rate, when Provincial-level hospitals are relatively low (61.96%) in terms of net loss. The overall policy services compensation rate is 85.93%, and the predictive services compensation rate is 96.53%. Looking at the policy rates and predictive compensation rates, all exceed 90% and 100%, except for Provincial-level hospitals.

Regarding the growth rate of pre-reform medical services revenue, this revenue source for most hospitals was increasing, showing that the supply of technical services is gradually expanding. This trend accords with the law of social economic development. Post-reform, the increase in the price of medical services has resulted in a significant increase in services revenue for most hospitals, except chest hospitals. However, there is still a big gap with the 2016 reform target (>35%). The goal of the services compensation policy plan has been basically achieved, with the net compensation rate of urban hospital services achieving “self-sufficiency”, but the services compensation level of some general hospitals (Nanjing First Hospital, The Second Hospital of Nanjing) and specialized hospitals (chest, brain) remains low.

Overall, there is still room for further price adjustments. During the keynote interview, we understood that urban public hospitals in Nanjing had only conducted one round of price adjustments as at May 2017. Although the price adjustment scheme included more than 5,900 items, some items were not adjusted. Thus, the range and extent of the price adjustment should go further.

The need to establish a dynamic and differentiated fiscal compensation mechanism

Regarding fiscal compensation, this has been fulfilled at the municipal level, and has surpassed the designed policy goal, but provincial, ministerial, and enterprise and public institution hospitals are clearly receiving inadequate compensation. Furthermore, in terms of comprehensive compensation, four hospitals are still “struggling” after the reform, and three are close to entering this category. This indicates that, in a general sense, almost half of the hospitals have failed to achieve the reform goal. In conclusion, the level of comprehensive compensation in all hospitals (excluding provincial and ministerial levels) more than meets the policy goal, yet there is imbalance within. Thus, nine hospitals can be classified as “benefitting” from the reform, and seven can be considered as “struggling”.

Combined with the clustering and regression results, the service compensation rate shows a significant correlation with the proportion of medical services revenue before the reform. If the proportion of medical services revenue increases by 1%, the service compensation rate will increase by ∼8%. Therefore, it is suggested that the public hospitals be divided into three types (or four groups) to allocate differentiated financial compensation: (1) “Service-dependent” hospitals, represented by maternity, children’s hospitals, and oral health hospitals. These hospitals are specialist hospitals with a high dependence on technical services. Thus, they did not experience significant revenue loss after the reform, and they have no need for fiscal specialist compensation; (2) “Drug-dependent” hospitals, represented by thoracic hospitals. These hospitals are also specialized. In contrast to the above, they are highly dependent on drug prescription to achieve therapeutic effects for their patients. Therefore, their losses after the reform are serious, with less room for improvement in the area of medical services revenue. It is suggested that the financial departments focus on compensation for such hospitals; (3) Comprehensive hospitals can be divided into two groups: one type can achieve self-sufficient services revenue with little need for fiscal subsidies. Most of the hospitals in this group had a relatively stable revenue structure and high operating efficiency before the reform. Thus, their losses from the removal of the drug markups are balanced by the growth of services revenue and management improvements. The other groups received a service compensation rate below 70%, and would have some losses according to the original financial investment compensation rate of 20%. These hospitals did not have a high proportion of services revenue before the medical reform, and, therefore, it is important to increase their services sooner rather than later. The finance department needs to improve the compensation rate for these hospitals. It is also suggested that dynamic adjustment and regular monitoring should be performed according to the actual annual compensation situation and medical revenue structures.

Comparison with reforms in other regions of China

A comparison of the pilot reform projects at Nanjing municipal and district-level hospitals in 2013 shows similarities regarding price range and compensation targets. Elsewhere, a comparison showed that the compensation results of public hospitals in five non-central urban areas have increased²³.

From a province-wide perspective, other cities in Jiangsu Province simultaneously carried out pricing and compensation reform on October 31, 2015, with similar reform measures of “four decrease, one increase and a trial”, namely, to remove the drug markups, reduce examinations costs and consumable prices, improve the price of surgery and medical treatment projects, and withdraw market regulation project prices. The following summary is based on a literature review for this issue. First, the hospital revenue has re-structured after the reform: the total cost of medical care has increased by ∼10% with the decline of drug expense and its proportion, and outpatient consultation and treatment fees have increased significantly²⁴. Second, patient out-of-pocket money has decreased: the level of the individual burden of insured workers decreased slightly^25–27.

To compare the measures, progress, and the effects of pricing and compensation reform, Beijing²⁸, Shanghai^29,30, and Chongqing³¹ are selected as representative municipalities, and, in terms of regions, Anhui (Central China)^32,33, Fujian (southeast)³⁴, Guangdong (south China)^35–37, Shaanxi (northwest)^38–40, Zhejiang (east)⁴¹, and Sichuan (southwest)⁴² are selected as representative provinces. The steps taken by the cities are detailed below:

1. Beijing established medical service fees to directly compensate the medical visits.

2. Anhui implemented a systematical reform, including cost pricing and grade pricing, introduced a reference price system of medical insurance payments, and negotiated consortium procurement with target quantities, and bulk purchases. Furthermore, it changed the payment method for medical insurance to global budget payment combined with clinical pathways, and reduced the burden on patients by including a price adjustment in the reimbursement range.

3. Fujian implemented differentiated pricing and compensation: provincial, city, and county medical service prices were compensated by 90%, 85%, and 82%, while financial departments provide compensation of 0%, 10%, and 15%, respectively, compensation was provided for medical services at 100%, 100%, and 82%, respectively. For county-level public hospitals, financial departments offer 15% compensation.

4. Guangdong attached importance to controlling medical expenses: it stipulated that the growth rate of medical expenses must not exceed 9% in 2018, secondary public medical institutions and those below must not exceed 9% in per capita outpatient cost growth rate and in the per capita hospitalization fee growth rate. For tertiary public medical institutions, the per capita outpatient cost growth rate and per capita hospitalization fee growth rate shall not exceed 7%, and the proportion of personal healthcare expenditure in total healthcare expense must fall to below 27%.

Regarding impact, there are different performances regarding medical expenses, patients’ burden, and the flow of medical consultations. However, the price relations of the medical services system have improved with raised medical services revenue, the reflected value of the technical services of medical personnel, and patients’ greater degree of acceptance.

In addition, because limited empirical articles have been published, evidence of the impact of the reform is scarce and cannot support further comparison. However, as estimated by domestic scholars, the price increases must be applied to 2,000 items, and this can then motivate the significant increase of medical services revenue and satisfy the policy goal to “provide 70% or higher compensation via services” in most regions⁴³. In summary, the effect of the reform in Nanjing should be affirmed, but further consultation and improvement are needed.

Innovation and limitations

Counterfactual construction plays a very important role in policy evaluation and other types of research design, and determines the credibility and interpretation methods of intervention effects^44–46. In policy evaluation, because it is difficult to find a control group in the strict sense, it is, therefore, often conducted in a self-controlled manner. In the traditional sense, a before-and-after contrast often directly analyzes the indicators in terms of year-on-year rate of change, thus neglecting the influence of the trends of the indicators themselves. This can result in an under-estimation or over-estimation of the intervention effects. The ITS model can return the time series data with breakpoints, and is often used to construct counterfactual impacts in policy evaluation^47,48. Compared with many time series models, the ITS model is simple in its structure and easy to interpret, so its efficiency is higher. This study uses historical data to predict the level of indicators in the “pre-reform” condition within the same period of time after reform, so as to directly refer to the pre-reform level for evaluation. This method can exclude the impact of natural growth trends, and better estimate the “net effect” of the policy.

This study isn’t without limitations. It is usually necessary to discount the revenues to a yearly basis for a study whose observation period is longer than a year. However, we didn’t discount based on the following reasons: first, the reason why we collected a 60-month period of data was because it would provide more robust analysis based on the model we used, other than to observe all the policy changes in a study period. Also, before using the regression model, we assumed the confounders from policy term might be the same for all the study objects; second, the study objects were sampled from the same region and shared the same CPI, thus allowing us to exclude the impact of inflation to different hospitals; third, although the CPI varied month-to-month, it has stayed at ∼2% during the past 5 years. So, it might either introduce too much work if we counted the slightly changed CPI on a monthly basis or it might be just unnecessary to not use the same CPI for the 60 months.

Also, although the simple linear ITS model adopted by the hospital financial data department is more accurate for most of the indicators, there are several indicators with a poor degree of fit. The following are some potential explanations: ITS default indicators linearly change over time and, therefore, the seasonal fluctuations and non-linear changes of the indicators are ignored. In addition, ITS has higher requirements for the length of historical data and limitations for applying short-term data. In addition, the degree of mining for some data is far from sufficient and only skims the surface. For example, regarding medical revenue structure and medical services revenue structure, a longitudinal analysis is often carried out to determine the average medical expenditure structure per patient from the perspective of a before-and-after contrast. This method lacks a horizontal comparison between hospitals. In addition, many longitudinal and horizontal comparisons and analyses lack verification, and it can, therefore, be difficult to draw conclusions.

Conclusions

One year after the pricing and compensation reform in Nanjing, a number of significant effects are evident: the drug markups are considered basically removed, the number of prescriptions and examinations has declined, the service revenue enjoys a substantial increase, and there is no significant increase in the burden on patients. Furthermore, the average net growth rate in medical services revenue reached 28.25% after excluding the natural growth trend, showing a larger potential. Regarding compensation level, in addition to provincial and ministerial-level hospitals, policy and predictive comprehensive compensation rates in all hospitals are over 100% and 120%, enjoying a high level of compensation. In all, the reform has achieved significant success, and also exceeds baseline expectations.

However, the reform does have some limitations that are mainly reflected in the imbalance between the level of financial compensation and the growth of services revenue among hospitals. This has resulted in the unequal distribution of comprehensive compensation, and some comprehensive and specialist hospitals still have room for service revenue growth. If the proportion of medical services revenue increases by 1%, the services compensation rate will increase by 8%, as suggested in this study. It is then concluded that a differential compensation plan with a dynamic adjustment mechanism be established according to the type and revenue structure of the hospital. Furthermore, the implementation of price reform should be improved in terms of the amount and degree of price items.

Transparency

Declaration of funding

This project is funded by the Chinese National Natural Science Foundation/Youth Program (Grant NO: 71603278) and Jiangsu Philosophy and Social Science Foundation (Grant NO: 2016SJD630007).

Declaration of financial/other relationships

The authors declare no competing interest in this publication.

Acknowledgments

We thank Katie Stallard, LLB, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript.