Sage Journals: Discover world-class research

Abstract

Improving the productivity of healthcare delivery and optimizing the allocation of regional healthcare resources are crucial for the health providers. The objective of this study is to evaluate the productivity dynamics of healthcare delivery at the regional (provincial) level in China, to provide evidence-based policy implications. After a review of literature, actual number of open beds, number of occupational or assistant doctors, number of registered nurses, and number of other staff were selected as input variables. The number of diagnostic visits and number of discharged inpatients were adopted as the output indicators. The panel data of 31 provinces in mainland China from 2010 to 2019 were extracted from Health Statistics Yearbook. Bootstrap-Malmquist Data Envelopment Analysis (DEA) model was used to measure the total factor productivity changes (TFPC) and its components. During the study period, the analysis of total factor productivity (TFP) in China revealed a declining trend with an average annual decline of 0.9% (ranging from 0.860 to 1.204). For each of the 31 provinces, the annual TFP scores varied from 0.971 to 1.029. On average, technical efficiency changes (TEC) had showed a downward trend from 2010-2011 (0.980) to 2013-2014 (0.982), and then an upward trend in 2014-2015 (1.029) and the following three consecutive years since 2016-2017 (1.000, 1.013, 1.009). Similarly, the trend in technological changes (TC) was consistent with the TEC from 2010-2019, which fluctuated between 0.969 and 1.011 on average per year at the provincial level. Notably, the point of inflection appeared at 2013-2014. Regional healthcare inputs and outputs in mainland China saw an upward trend from 2010 to 2019. However, TFPC, TEC, and TC decreased across all 31 provinces. TFP experienced a declining trend from 2010 to 2014, followed by growth until 2019. This may be related to the new healthcare reform being implemented since 2009, as service efficiency and capacity may undergo a reversal at the beginning of the reform.

Keywords

regional healthcare productivity Malmquist Bootstrap-DEA health policy efficiency

What do we already know about this topic?

Measuring the relative efficiency and productivity of the healthcare system is increasingly needed. The Bootstrap-Malmquist-DEA method has already been widely used for productivity measurement in many fields.

How does your research contribute to the field?

This research employs Bootstrap-Malmquist-DEA to measure the productivity of regional healthcare delivery, encompassing all kinds of different healthcare resources.

What are your research’s implications toward theory, practice, or policy?

First, the Bootstrap-Malmquist-DEA method can be used to measure the productivity of healthcare system. Second, the allocation of healthcare resources in some regions still need further optimization. Last, provinces with insufficient healthcare productivity should focus on improving their technical efficiency.

Introduction

China has established a universal healthcare insurance system in the past decade, accompanied with tremendous increase of demand for more healthcare services.^1,2 Despite the concerted efforts made by governments at all levels, meeting the increasing demand for healthcare from the population remains a significant challenge.³ There has been tremendous increase in investments directly toward primary care, secondary and tertiary hospitals over the years. However, despite these substantial investments, the quality of primary care providers remains low.⁴ Since primary care facilities typically serve as the initial point for patients seeking medical attention, the limited capacity and quality of these facilities contribute to a continuous influx of patients seeking care at higher-level hospitals.^5
-7 In the current mechanism design, healthcare providers in China are allowed to survive by selling drugs and conducting health examinations, which unexpectedly leads to issues such as over-prescription, excessive medical examinations, and unnecessary healthcare services, and finally these practices contribute to a heavy financial burden on patients and an inefficient healthcare delivery system.⁸ Unmet health care needs continue to increase, and health expenditure grows up sharply, which leads to problems such as poor doctor-patient relationship and a rapid growth of out-of-pocket payment.^9,10 According to Masri and Asbu,¹¹ improving efficiency of the healthcare system is an important approach to meet rapidly growing health care demand and constrain the growth of health care costs. For China, more and more research requires the use of the latest methods to measure the relative efficiency and productivity of the healthcare system at the local, regional, and national levels.

Data Envelopment Analysis (DEA) is a widely used method in healthcare organizations for estimating relative efficiency and productivity. It utilizes multiple inputs and outputs to evaluate the performance of healthcare systems. Previous literature showed that some traditional DEA models were widely used by Chinese researchers to measure the efficiency of healthcare organizations. These models include DEA-BCC (Banker, Charles, and Cooper)¹² or DEA-CCR (Charles, Cooper, and Rhodes),¹³ model for cross-sectional efficiency analysis and Malmquist-DEA model for longitudinal efficiency analysis of panel data. For example, Gai et al¹⁴ applied the DEA-BCC Model and DEA-CCR Model to calculate efficiency. Li et al⁸ applied Malmquist-DEA model to measure relative efficiency and productivity. Caballer-Tarazona et al¹⁵ used a DEA model to calculate the performance of 3 healthcare service units of 22 hospitals in East Spain. Sulku¹⁶ proposed a model based on DEA and the Malmquist index on the multiple inputs and multiple outputs of the ministry of health hospitals in Turkey. Also some researchers used related models of DEA to evaluate efficiencies of hospitals in Iran, Greece, US, Norway, Italy, Japan.^17
-19 Similar to international research, existing papers on the regional healthcare efficiency in China have obvious limitations. One prominent limitation lies in the flaws of traditional DEA models. The relative efficiency scores calculated by the traditional DEA models tend to be significantly higher than the real scores and are susceptible to environmental and random factors.^20
-22 To address this issue, Simar and Wilson, Daraio and Simar, among others, have introduced the Bootstrap method. This method improves the accuracy of DEA efficiency scores by correcting their bias and calculating confidence intervals. Based on the Malmquist index, the Bootstrap-DEA method has already been widely used for productivity measurement in the fields of banking,^23,24 transportation,^25,26 agriculture,^27,28 and more. However, its application in healthcare productivity analysis has been limited. Cheng et al²⁹ adopted Bootstrap Malmquist-DEA to measure the efficiency and productivity. Wang et al³⁰ utilized Bootstrap Malmquist-DEA to evaluate the productivity. To date, no study has been found that utilizes Bootstrap Malmquist-DEA for the comprehensive productivity measurement of regional healthcare delivery, encompassing various types of healthcare resources.

DEA, a non-parametric method based on linear programing, has been widely applied to efficiency and productivity measurement. In the context of healthcare organizations with multiple inputs and outputs, DEA is often preferred over stochastic frontier analysis (SFA) as it does not require the construction of a specific function. However, there exist some limitations on the use of MPI based on the traditional DEA method,^31,32 as DMUs is more prone to be influenced by environmental and random factors. Although the 3-stage DEA model, proposed by Fried et al,³³ can isolate the influence of external environmental and random factors on efficiency scores, providing a better reflection of the real internal management level of DMUs, the results are still sample-sensitive and are susceptible to extreme values. Hence, Simar and Wilson proposed the Bootstrap-DEA model with bias correction to improve the accuracy of the efficiency and productivity scores.^21,22,34
-37 In this study, we used Malmquist productivity index together with Bootstrap-DEA method to evaluate dynamic changes in healthcare productivity across the 31 provinces in China.

Methods

Research Design

A longitudinal analysis was undertaken using a Bootstrap-Malmquist-DEA model. Total factor productivity changes (TFPC), technical efficiency changes (TEC) and technological changes (TC) were generated from the Bootstrap-Malmquist-DEA model for longitudinal comparisons, while the ranking of the average TFPC, TEC and TC of the 31 provinces was sorted for horizontal comparisons.

Input and Output Variables

In this study, input and output variables were selected by conducting an extensive literature review and analyzing available data from China Health Statistics Yearbook. However, selecting appropriate variables to measure healthcare efficiency remains a challenge in the current literature. Previous literature has suggested using statistical methods like cluster analysis and factor analysis for indicator classification and selection.³⁸ To address this issue, Our previous research conducted a statistical classification and analysis of published literature in Chinese on healthcare organization efficiency over the past 30 years, comparing it with peer international studies.³¹ We counted the top 5 input indicators applied in international literature were number of beds, number of physicians/doctors, number of nurses and/or nursing staff, operating expenses/supplies, and number of total FTEs/staff/labor size. While the top 5 output indicators in international studies were number of outpatient visits, number of total surgeries, number of inpatient days, total patient days, number of discharges. This study concluded that combining volume variables with monetary variables could lead to confusion between technical efficiency (TE) and allocative efficiency (AE). Therefore, only quantitative variables were included in the indicator system of this study to enhance the credibility of the results. Ng³⁹ emphasized that labor and capital are both significant input variables in healthcare delivery. In her research, the numbers of doctors, nurses, pharmacists, and other staff were selected as labor inputs and the number of beds was selected as capital input. Based on available data from China Health Statistics Yearbook, number of occupational or assistant doctors, number of registered nurses, and number of other staff (including other technicians, administrative staff, and handyman) were selected as labor input variables in our study. Regarding capital inputs, the number of beds was typically adopted as a proxy of capital in international research. However, Li et al⁸ argued that the dynamic nature of hospitalization in China is characterized by the use of numerous temporary beds. Thus, actual number of open beds was adopted as our capital input variable. For the output variables, we followed the approach of our previous research, adopting number of diagnostic visits and number of discharged inpatients as the output indicators.³²

Malmquist Productivity Index

In our study, we employed the MPI (Malmquist Productivity Index) to evaluate total factor productivity changes (TFPC) to reflect the longitudinal changes in the input-output efficiency of each decision-making unit (DMU). Initially introduced by Caves et al⁴⁰ for productivity analysis, the MPI was subsequently developed by Färe et al⁴¹ as a tool to explain efficiency change and technological change.

The Malmquist index can be decomposed into technical efficiency changes (TEC) and technological changes (TC), and TEC can also be decomposed into pure technical efficiency changes (PTEC) and scale efficiency changes (SEC) under the hypothesis of constant returns to scale. According to Fare et al,⁴² TEC measures the degree of catching-up of production decision-making units to the optimal production possibility boundary from period t to t + 1, also termed as “catching-up to the frontier effect.” If a DMU’s TFPC score is greater than 1, it indicates that DMU has experienced efficiency growth during that period; a TFPC score of 1 signifies unchanged efficiency, while a TFPC score less than 1 suggests efficiency deterioration. A TEC score greater than 1 indicates that the production of decision-making units has been closer to the production frontier and the relative TE has improved. TC shows the movement of the technological boundary from time t to time t + 1, also called “shift in the frontier effect.” A TC score greater than 1 implies that an improvement or innovation in technology, resulting in an upward movement of the production frontier.

Bootstrap-Malmquist-DEA Model

The basic idea of the Bootstrap method is to simulate the data generation process (DGP) through repeated sampling. The resulting sampling distributions and standard deviations therefore closely approximate the original ones. The bias corrected MPI with the confidence intervals on α level is produced based on the expanded simulated sample, which includes TFPC, TEC, and TC. The confidence interval excluding 1 indicates that it is statistically significant and the Malmquist index reflects the efficiency change, and vice versa.

Data Collection, Processing and Analysis

The data for the input and output indicators were extracted from the Health Statistic Year Book from 2011 to 2020. Descriptive analysis of the input and output indicators, including mean and standard deviation (SD), was conducted using SPSS 26.0 version statistical. The productivity and its decomposition indices of the provinces were measured using R 4.0.2 version statistical software, specifically employing FEAR package with 95% CIs.⁴³ The geometric mean was used to represent the average scores of productivity. A statistical significance was determined if the 95% CIs did not contain the value of 1.0000. The number of repeated sample selection was set at 2000. The statistical significance was set at α = 0.05.

Results

Descriptive Analysis of Input and Output Variables

Table 1 presents a descriptive overview of healthcare input and output variables of the 31 provinces in mainland China from 2010 to 2019. Data on all input and output variables showed a different levels of growth from 2010 to 2019. Notably, the number of discharged inpatients exhibited the highest average annual growth rate of 10.53%. On the other hand, the number of occupational or assistant doctors and the number of other staff showed a slower growth at average annual growth rates of 7.05% and 6.39% respectively. As shown in Table 1, the standard deviation of all input and output variables has been increasing year by year. This may indicate that the balance between regions was getting bad. Just like the result of the Healthcare Access and Quality (HAQ) Index. In 2016, China recorded the biggest range in subnational HAQ performance, which ranged from 91.5 score in Beijing to 48.0 score in Xizang (a 43.5-point difference).⁴⁴

Table 1.

Summary of Input and Output Variables.

	Input				Output
	I1	I2	I3	I4	O1	O2
2010
Min	5444	2502	1537	1537	3 390 468	108 864
Max	255 764	98 838	112 382	63 181	257 480 147	7 533 675
Mean	109 272	40 674	47 379	25 451	65 794 623	3 057 400
SD	64 117	24 450	28 614	14 858	52 901 572	1 958 193
2011
Min	6314	2457	1596	1618	3 998 564	119 792
Max	280 385	100 900	126 295	66 958	293 784 406	8 322 652
Mean	119 520	42 156	52 508	26 498	72 865 719	3 459 428
SD	70 994	25 575	31 990	15 567	60 142 155	2 208 198
2012
Min	5385	2341	1261	1264	3 172 647	111 872
Max	322 007	110 541	139 697	71 492	317 770 542	9 786 925
Mean	134 241	45 284	59 039	28 382	81 987 616	4 098 698
SD	81 534	27 969	36 353	16 749	66 727 310	2 632 814
2013
Min	7262	2977	1851	1611	4 324 188	159 276
Max	342 104	124 765	166 596	77 276	334 591 803	10 236 842
Mean	147 697	48 490	65 851	30 506	88 444 415	4 493 110
SD	89 222	30 356	41 616	18 339	71 157 910	2 843 829
2014
Min	7984	3151	2007	1715	4 767 555	187 251
Max	359 637	126 901	175 079	78 188	352 974 764	11 248 949
Mean	160 037	51 109	71 687	32 258	95 873 223	4 941 637
SD	96 846	31 733	44 826	19 509	76 122 087	3 138 216
2015
Min	9944	3617	2440	2396	5 486 127	244 036
Max	378 884	133 132	183 621	80 611	353 820 727	11 621 153
Mean	171 954	54 605	77 666	34 247	99 472 286	5 165 773
SD	104 484	33 704	48 364	20 505	77 726 161	3 292 714
2016
Min	10 397	3776	2947	2569	5 802 386	236 570
Max	399 427	141 053	198 249	83 899	370 428 821	12 924 743
Mean	183 512	58 176	84 302	36 357	105 469 643	5 623 433
SD	112 759	36 132	52 578	21 823	81 019 965	3 595 033
2017
Min	11 749	4278	3342	3141	6 261 344	276 583
Max	441 032	154 571	215 621	86 777	371 464 361	14 079 515
Mean	197 435	62 340	91 047	38 446	110 932 926	6 071 829
SD	121 577	38 996	56 706	23 056	83 974 935	3 902 616
2018
Min	12 604	4836	4076	3859	6 698 449	269 096
Max	460 703	166 567	234 242	91 355	374 831 839	14 567 847
Mean	210 314	66 243	97 446	40 196	115 399 200	6 438 051
SD	129 719	41 830	61 138	24 229	85 991 495	4 109 907
2019
Min	12 748	5032	4369	4442	6 466 630	279 105
Max	481 301	175 167	244 056	96 374	401 317 285	15 706 797
Mean	221 501	70 138	104 451	41 764	123 948 542	6 809 086
SD	136 247	44 084	64 855	25 261	91 984 132	4 358 650

I1 = actual number of open beds; I2 = number of occupational or assistant doctors; I3 = number of registered nurses; I4 = number of other staff; O1 = number of diagnostic visits; O2 = number of discharged inpatients.

Bootstrap MPI and Its Components

Table 2 shows the Bootstrap MPI (TFPC) and its decomposition indices (TEC and TC) of the 31 provinces in mainland China during 2010 to 2019. The provincial average TFPC fluctuated between 0.860 and 1.204 over this period. Notably, 83.15% of provinces exhibited statistically significant improvements in TFP. From 2010 to 2014, TFP experienced a decline, but it subsequently grew from 2014 to 2019. Furthermore, the provincial average TFPC occurred during the 2014-2015 period. However, 4 provinces (Heilongjiang, Shanghai, Chongqing, Xizang) experienced a decline in TFP during this period. In terms of individual years, only Ningxia exceeded a TFPC of 1 in 2010-2011, and only Yunnan achieved this in 2011-2012. On average, 8 provinces had improvements in TFP between 2010-2019, with Guangdong of the highest average annual growth rate of 0.98%. In addition, the annual TFPC among provinces varied over time. The provincial TE of healthcare in China exhibited a decline from 2010 to 2014, followed by an improvement from 2016 to 2019. The highest improvement in TE occurred in the period 2014-2015, with a growth rate of 2.92%. In that same period, only Chongqing and Heilongjiang do not experience TE growth. Shanghai remained unchanged, except in 2011-2012 and 2013-2014. Surprisingly, TEC was below 1 on average in 2015-2016. Among the 9 provinces analyzed, Guangdong exhibited the best TE improvement between 2010 and 2019, outperforming other provinces. On average, the TC first exhibited a positive trend, indicating a shift from technological regression to technological progress among hospitals in the 31 provinces of mainland China since 2013-2014. Prior to that period, the TC scores were below 1, although the minimum value of TC was greater than the corresponding TFPC and TEC values. The provincial average TC scores fluctuated between 0.942 and 1.169, with the highest score (1.169) observed in 2015-2016. The TC was below 1.025 but Xizang which was above 1.050 from 2015-2019, respectively. In general, all 31 provinces showed varying improvements in TC within the 9-year timeframe, with Xizang having the highest average growth rate.

Table 2.

Bootstrap MPI and its Components by Region in Period 2010-2019.

	2010-2011			2011-2012			2012-2013			2013-2014			2014-2015
DMU	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC
Beijing	0.914**	0.960**	0.952**	0.944**	0.947	0.997	0.951**	0.953	0.998	0.996	0.996	1.000	1.043**	1.043	1.000
Tianjin	0.870**	0.903**	0.963	0.972**	0.976	0.995	0.981**	0.986	0.995	0.964**	0.964	1.000	1.044**	1.044	1.000
Hebei	0.972**	0.990	0.982	0.951**	0.949**	1.002	1.026**	1.024	1.002	1.005**	0.996	1.009	1.071**	1.063**	1.008
Shanxi	0.940**	0.951**	0.989	0.935**	0.931**	1.004	0.994	0.992	1.001	0.972**	0.965	1.007	1.059**	1.052**	1.007
Neimenggu	0.970**	0.982	0.987	0.964**	0.952**	1.013	1.016**	1.012	1.004	1.001	1.005	0.997	1.060**	1.061**	0.999
Liaoning	0.955**	0.970**	0.985	0.913**	0.912**	1.001	0.957**	0.957**	1.000	0.963**	0.963	1.000	1.022**	1.020	1.002
Jilin	0.940**	0.954**	0.986	0.934**	0.922**	1.013	0.981**	0.969	1.012	0.990**	0.993	0.997	1.028**	1.027**	1.001
Heilongjiang	0.960**	0.975**	0.985	0.931**	0.927**	1.004	0.955**	0.955	1.000	0.988**	0.980	1.008	0.993**	0.983	1.010
Shanghai	0.944**	1.000	0.944	0.988	0.988	1.000	0.997	1.000	0.997	0.986**	0.986	1.000	0.999**	1.000	0.999
Jiangsu	0.946**	0.989	0.956**	0.985**	0.988	0.998	0.972**	0.974**	0.998	0.980**	0.980	1.000	1.013**	1.013	1.000
Zhejiang	0.942**	1.000	0.942**	0.966**	0.962	1.004	1.005**	1.001	1.004	1.000	1.003	0.997	1.046**	1.048**	0.998
Anhui	0.920**	0.948**	0.971	0.941**	0.954**	0.987	0.971**	0.984	0.987	0.973**	0.973	1.000	1.006	1.006	1.000
Fujian	0.982**	1.000	0.982	0.918**	0.920**	0.998	1.005	1.007	0.998	1.006	1.006	0.999	1.070**	1.070**	0.999
Jiangxi	0.935**	0.969	0.965	0.895**	0.904**	0.991	0.979**	0.999	0.980	0.976**	0.976	1.000	1.012	1.012	1.000
Shandong	0.966**	1.012	0.954**	0.946**	0.946**	1.000	1.067**	1.067	1.000	0.953**	0.953**	0.999	1.013**	1.013	1.000
Henan	0.977**	0.994	0.982	0.889**	0.889**	0.999	0.990**	0.992	0.998	0.969**	0.966	1.003	1.028**	1.026	1.002
Hubei	0.966**	0.991	0.975	0.907**	0.907**	1.000	0.994	0.995	1.000	0.968**	0.968	1.000	1.050**	1.050	1.000
Hunan	0.950**	0.999	0.952**	0.899**	0.939**	0.958**	0.970**	1.017	0.953**	0.990**	0.990	1.000	1.069**	1.063**	1.006
Guangdong	0.969**	1.000	0.969	0.994	0.989	1.004	1.026**	1.022	1.004	1.001	1.003	0.998	1.049**	1.050**	0.999
Guangxi	0.957**	0.994	0.962**	0.929**	0.929**	0.999	0.956**	0.957**	0.999	0.977**	0.977	1.000	1.040**	1.040	1.000
Hainan	0.985**	1.010	0.976	0.907**	0.909**	0.999	0.991**	0.992	0.999	0.984**	0.984	1.000	1.022**	1.022	1.000
Chongqing	0.938**	0.980	0.957**	0.913**	0.914**	0.999	0.960**	0.961**	1.000	0.973**	0.973	1.000	0.963**	0.963**	1.000
Sichuan	0.978**	1.010	0.968**	0.896**	0.896**	1.000	0.989**	0.990	0.999	1.007**	1.006	1.001	1.007	1.007	1.000
Guizhou	0.977**	1.014	0.963	0.860**	0.887**	0.969	0.910**	0.954	0.953**	1.020	1.000	1.020	1.060**	1.060**	1.000
Yunnan	0.949**	1.000	0.949	1.009	1.000	1.009	1.017	1.017	1.000	1.002	1.000	1.002	1.033**	1.033	1.000
Xizang	0.917**	0.928	0.988	0.992	0.993	0.999	1.019	1.018	1.001	0.971**	0.971	1.000	0.966**	1.000	0.966
Shaanxi	0.949**	0.980	0.968**	0.888**	0.889**	0.999	0.948**	0.953**	0.994	0.953**	0.953**	1.000	1.011**	1.011	1.000
Gansu	0.922**	0.943**	0.978	0.876**	0.870**	1.006	0.963**	0.961	1.002	1.012**	1.000	1.012	1.017**	1.006	1.011
Qinghai	0.956	0.982	0.974	0.962**	0.952**	1.010	1.067**	1.064**	1.003	0.950**	0.943**	1.007	1.062**	1.054**	1.007
Ningxia	1.002**	1.023**	0.979	0.883**	0.884**	0.999	1.027**	1.028	0.999	0.998**	1.002	0.997	1.059**	1.062**	0.997
Xinjiang	0.921**	0.955**	0.964	0.920**	0.932**	0.987	1.024**	1.052	0.973	0.999	0.978	1.022	1.036**	1.014	1.022
	2015-2016			2016-2017			2017-2018			2018-2019			2010-2019
DMU	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC	TFPC	TEC	TC
Beijing	0.969**	0.944**	1.026	1.109**	1.083**	1.024	1.000	0.990	1.010	1.008**	0.998	1.010	0.991	0.989	1.002
Tianjin	1.024**	1.000	1.024	1.080**	1.057	1.022	1.046**	1.036	1.010	1.004**	0.994	1.010	0.997	0.995	1.002
Hebei	0.962**	0.959	1.004	1.021**	1.020	1.001	1.026**	1.006	1.019	1.042**	1.022	1.019	1.008	1.003	1.005
Shanxi	0.962**	0.956**	1.006	0.975**	0.966**	1.009	0.973**	0.974	0.999	1.001	1.001	0.999	0.978	0.976	1.002
Neimenggu	0.978**	0.972**	1.007	0.981**	0.972**	1.008	0.985**	0.987	0.998	1.055**	1.058**	0.997	1.001	0.999	1.001
Liaoning	1.001	0.993	1.008	1.006**	0.997	1.009	1.028**	1.029	0.999	1.039**	1.041**	0.998	0.986	0.986	1.000
Jilin	1.006	1.004	1.001	0.981**	0.978	1.003	1.057**	1.050**	1.006	1.006	1.005	1.001	0.991	0.988	1.002
Heilongjiang	0.956**	0.953**	1.003	0.975**	0.971	1.004	1.008**	1.008	1.000	0.997	0.997	1.000	0.974	0.972	1.002
Shanghai	1.013	1.000	1.013	0.989	1.000	0.989	1.013	1.000	1.013	0.992	1.000	0.992	0.991	0.997	0.994
Jiangsu	1.026**	1.022	1.004	1.004**	1.001	1.004	1.030**	1.037**	0.993	1.006**	1.013	0.993	0.995	1.002	0.994
Zhejiang	1.016**	1.019	0.997	0.977**	0.985	0.992	1.012	1.022	0.990	1.009**	1.020	0.990	0.997	1.006	0.990
Anhui	0.984**	0.974**	1.010	0.988**	0.978**	1.011	1.015**	1.016	1.000	0.997**	0.998	0.999	0.977	0.981	0.996
Fujian	0.996	0.996	1.000	1.027**	1.031	0.996	1.026**	1.033**	0.994	1.003	1.009	0.994	1.003	1.007	0.996
Jiangxi	0.998	0.992	1.006	1.002	1.000	1.002	0.972**	0.972	1.000	1.002	1.002	1.000	0.974	0.980	0.994
Shandong	0.962**	0.953**	1.009	1.004**	0.995	1.010	1.053**	1.055**	0.998	1.006**	1.008	0.998	0.996	0.999	0.996
Henan	0.997	0.984	1.013	1.008**	0.997	1.011	0.973**	0.974	0.999	0.984**	0.984	1.000	0.979	0.978	1.001
Hubei	0.979**	0.974	1.005	0.985**	0.983	1.002	0.995**	0.996	0.999	0.965**	0.966**	0.999	0.978	0.980	0.998
Hunan	1.004**	0.995	1.009	1.013**	1.005	1.009	1.017**	1.017	1.000	1.066**	1.066**	1.000	0.996	1.009	0.987
Guangdong	1.025**	1.031	0.994	1.017**	1.020	0.997	1.023**	1.034	0.989	0.986**	0.996	0.990	1.010	1.016	0.994
Guangxi	1.004	1.008	0.995	0.992**	0.997	0.994	1.007**	1.009	0.997	0.986**	0.989	0.998	0.982	0.989	0.994
Hainan	0.996**	0.990	1.006	0.986**	0.985	1.001	1.012**	1.016	0.996	0.991**	0.996	0.995	0.986	0.989	0.997
Chongqing	1.014**	1.016	0.998	1.009**	1.012	0.997	1.055**	1.056**	0.999	0.985**	0.986	0.999	0.978	0.984	0.994
Sichuan	0.996	0.981	1.015	0.981**	0.970**	1.011	1.030**	1.031**	0.999	0.977**	0.978	0.999	0.984	0.985	0.999
Guizhou	1.030**	1.024	1.006	0.981**	0.980	1.000	0.987**	0.987	1.000	1.045**	1.045**	1.000	0.983	0.993	0.990
Yunnan	1.008	1.000	1.008	0.996**	1.000	0.996	0.975**	0.975	1.000	1.057**	1.058**	1.000	1.005	1.009	0.996
Xizang	1.169**	1.000	1.169**	1.000	0.882**	1.133**	1.204**	1.106	1.089**	1.056**	0.981	1.077**	1.029	0.985	1.045
Shaanxi	0.991**	0.996	0.995	0.999**	1.006	0.993	0.991	0.991	1.000	1.019**	1.019	1.000	0.971	0.977	0.994
Gansu	0.965**	0.965	1.000	1.044	1.042	1.002	0.994	0.991	1.002	1.083**	1.083**	1.000	0.984	0.983	1.001
Qinghai	0.976**	0.972**	1.005	1.059**	1.050**	1.009	1.012**	1.013	0.999	0.996**	0.997	0.999	1.003	1.002	1.001
Ningxia	1.002	1.002	1.000	1.055**	1.050**	1.005	0.992**	0.997	0.995	1.005**	1.010	0.995	1.001	1.005	0.996
Xinjiang	1.102**	1.006	1.006	0.998**	0.992	1.006	1.006**	1.006	1.000	0.980**	0.980	1.000	0.997	0.990	0.998

Double asterisk (**) denotes significant differences at 95%. Significant results are in bold.

A score>1 indicates growth, a score = 1 signifies stagnation and a score <1 indicates decline or deterioration.

Ranking of TFPC, TEC, and TC

Hebei, Qinghai and Neimenggu had higher ranks in TFPC, TEC and TC standings. While Guangdong, Yunnan and Fujian had higher ranks in both TFPC and TEC standings but lower ranks (18th, 20th, and 25th) in TC standings. Xizang secured the top rank in both TFPC and TC standings but ranked 21st in TEC. Shaanxi, Heilongjiang and Jiangxi occupied the bottom 3 positions in TFPC standing. For Shaanxi and Heilongjiang, the particular low TEC rankings (29th and 31st) contributed significantly to their low TFP growth. Meanwhile, Guangxi and Guizhou had low rankings primarily due to their disappointing TC standings.

Discussion

Our results indicated that overall regional healthcare inputs and outputs in China kept ascending from 2010 to 2019, driven by growing demands for healthcare services and supported by healthcare organizations’ development strategies and national policies. Du et al⁴⁵ demonstrated that there were no obvious productivity fluctuations in public hospitals during the recent observing years, indicating satisfactory performance in meeting the rapidly growing medical demand. However, variations in healthcare inputs (such as open beds and healthcare professionals) and outputs (such as diagnostic visits and discharged inpatients) were observed across provinces each year. This highlighted the imbalanced spatial distribution of healthcare resources in China, as discussed by Yin et al,⁴⁶ leading to inadequacy and limited access to affordable healthcare services. To address this issue, it is crucial to redistribute high-quality healthcare resources from urban areas to grassroots and communities, aiming to balance and optimize the allocation of healthcare resources between developed and underdeveloped regions.

Total factor productivity referred to the output growth rate brought about by all production factors except for labor and capital. The core of the Malmquist index was to analyze the reasons for total factor productivity changes, identify whether its growth was based on technological progress, efficiency improvement, or scale efficiency. Technological change referred to a change in the overall production technology, which can be divided into technological progress and technological regression. Technological progress comes from technological innovation, while technological innovation comes from the introduction of new technologies and equipment, as well as the proficient use of new technologies and equipment by professionals. Efficiency changes were caused by the development level of existing technology, the level of operation and management, and the suitability of hospital production scale.

The results indicated that, on average, TFPC, TEC, and TC of all 31 provinces showed a slight decline in 10 years. Al-Hanawi and Makuta⁴⁷ also indicated TFP for healthcare services has decreased on average by 5.6% per year primarily due to a technical regression from 2006 to 2018. Base on the overall situation of each province, the decline in total factor productivity may be mainly attributed to changes in technical efficiency. According to Li et al,⁸ technical efficiency changes had trivial contributions to TFP growth, which is a common problem among Chinese public hospitals. But between every 2 years, the TFPC, TEC, and TC showed fluctuations from decline to improvement. Cheng et al²⁹ reported similar results that the township hospitals experienced a positive productivity shift from 2008 to 2012 but experienced deterioration from 2012 to 2014, and that TFPC of the township hospitals declined by 2.14% during the 7-year period. Li et al⁴⁸ also noted the efficiency of the county-level public hospitals in Anhui Province has not improved significantly, and the total productivity factor had declined during 2010-2015. To be specific, technology growth started in 2013- 2014, while TFP and TE have increased in 2014-2015. This may be due, in part, to the new healthcare reform being implemented since 2009.⁴⁹ The reform aimed to establish a relatively comprehensive healthcare system and has lead to substantial improvement in overall service capacity and medical technology across all levels of the healthcare organizations. However, initially, service efficiency and capacity may have experienced setbacks before progressing. The reform also emphasized the provision of basic public health services, chronic disease management, and mental health services by primary healthcare organizations. As the development of primary healthcare organizations continues, the extensive needs of urban and rural residents for public health services can be meet. In addition, the basic healthcare insurance system improved in the reform covers both urban and rural areas universally. On the one hand, it helps minimize individual healthcare expenses and improve access to healthcare services for both urban and rural residents.²⁹ On the other hand, third-party purchasers of healthcare insurance institutions are playing an increasingly significant role in intervening and supervising the operating practices of healthcare organizations, contributing to the continuous improvement of regional healthcare service capacity, such as reducing hospitalization lengths and increasing bed turnover rates.⁵⁰

Our study shows that, for some provinces such as Xizang, Hebei, and Tianjin, technological improvement is the major contributor to TFP growth. The introduction of new drugs, updated medical technologies and high-tech medical equipment has proven crucial in improving regional healthcare productivity. This suggested that the provinces with insufficient healthcare productivity growth, such as Guizhou and Hunan, should focus on improving their TC. Guizhou Province located in the southwest of China, medical technology may lag behind eastern cities. Guizhou had many highland and mountainous areas, and it is a province where multiple ethnic groups coexist. Hunan is the most populous province in China, medical needs is directly proportional to population. They can introduce new technologies, introduce or update devices, construct clinical specialties, strengthen county hospitals and so on. Their emphasis should be placed on improving the utilization of resources and fully leveraging technological capabilities. Similarly, Nghiem et al⁵¹ revealed technical changes are the driving force of productivity growth in Queensland health in the study period 1996 to 2004. Al-Hanawi and Makuta⁴⁷ indicated addressing existing technological challenges can unlock potential gains in healthcare service productivity. However, with the technological advancement and innovation, doctors’ profit-driven behaviors can increase healthcare expenditures, thus undermining the efficiency of healthcare system (it is important to mitigate profit-driven behaviors among doctors that can lead to increased healthcare expenditures and undermine system efficiency).⁵² With the new healthcare reform being implemented, this unhealthy mechanism was seen to gradually diminish. The results of Cordero et al⁵³ showed that the productivity growth of hospitals belonging to the SSF has been much higher than those belonging to the Ministry of Health during the period between 2005 and 2015, primarily due to superior technological advancements, particularly in the later years of the evaluation period. Based on such reasons, the provinces with relatively low TFPC and TEC, such as Heilongjiang, should focus on improving their TEC. This can be achieved through optimizing health personnel training and management to stimulate the scale efficiency, introducing third-party agencies for performance evaluation and adopting benchmarking and cross-organizational learning mechanism to help improve the pure technical efficiency and internal management levels.

Limitations

The Bootstrap-Malmquist-DEA model was used in this study to evaluate the intertemporal healthcare productivity changes in 31 provinces of mainland China. However, this study can be improved in the following aspects. First, the DEA method used in this study has some limitations, such as the inability to assign appropriate weights to the selected input-output indicators. For future efficiency research, it would be beneficial to explore alternative methods like SFA and compare their effectiveness. Second, further studies can consider employing 3-stage DEA combining Bootstrap-DEA for intertemporal data analysis to reduce the impact of some environmental and random factors on efficiency scores, for a more accurate measurements of healthcare productivity changes.

Conclusions and Recommendations

The results indicated that both regional healthcare inputs and outputs in all the provinces have different degrees of growth after the implementation of the new healthcare reform in 2009, but the allocation of healthcare resources in some regions still need further optimization. Moreover, all 31 provinces had experienced healthcare productivity growth during the study period, for which substantial improvement in TC was the main contributor.

Observed differences in the productivity improvements between different provinces suggest that those with insufficient healthcare productivity should focus on improving their TE. On the one hand, the development of hospital scale should be guided by regional health planning and optimized allocation of health resources, adhering to the principle of moderation. Strengthening cooperation and coordination among medical institutions was crucial for optimizing the allocation of medical resources. Implement a graded diagnosis and treatment system was a favorable means to achieve the rational utilization of medical resources. Improve institutional construction was the guarantee. On the other hand, increase government investment to comprehensively improve the efficiency of hospitals from various dimensions. Focus on leveraging technological and specialized advantages.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Key Research and Development Program of China (Grant No. 2020YFC2006000). This study was supported by the National Natural Science Foundation of China (Grant No. 71573061).

Ethics Approval

This paper does not contain data collected from human subjects requires ethics approval.

Consent for Publication

All the authors listed in this paper have already read this manuscript and agreed to submit the paper and for later publication.

ORCID iD

Meng Li

References

, et al. Transformation of potential medical demand in China: a system dynamics simulation model. J Biomed Inform. 2015;57:399-414.

Xiong

Zhang

Ren

, et al. Impact of universal medical insurance system on the accessibility of medical service supply and affordability of patients in China. PLoS One. 2018;13(3):e193273.

Wang

Rao

Liu

Health Care in China. Chest. 2013;143(2):524-531.

Zhang

Ren

Sun

Liu

Inequality in the distribution of health resources and health services in China: hospitals versus primary care institutions. Int J Equity Health. 2017;16(1):42.

Lam

TP.

Underuse of primary care in China: the scale, causes, and Solutions. J Am Board Fam Med. 2016;29(2):240-247.

Huang

Liang

Chu

Rutherford

Geng

The emerging role of private health care provision in China: a critical analysis of the current health system. Soc Sci Electron Publ. 2009;10. doi:10.2139/ssrn.1658564

Lam

Zhou

Sun

KS.

Challenges to healthcare reform in China: profit-oriented medical practices, patients' choice of care and Guanxi culture in Zhejiang province. Health Policy Plan. 2017;32(9):1241-1247.

Dong

Liu

Relative efficiency and productivity: a preliminary exploration of public hospitals in Beijing, China. BMC Health Serv Res. 2014;14:158-169.

Tang

Tao

Bekedam

Controlling cost escalation of healthcare: making universal health coverage sustainable in China. BMC Public Health. 2012;12(S1):S8.

10.

Deng

Yang

, et al. Doctors’ job satisfaction and its relationships with doctor-patient relationship and work-family conflict in China: a structural equation modeling. Inq J Heal Care Organ Provis Financ. 2018;55:1-11.

11.

Masri

Asbu

EZ.

Productivity change of national health systems in the WHO Eastern Mediterranean region: application of DEA-based Malmquist productivity index. Glob Health Res Policy. 2018;3(1):22.

12.

Charnes

Cooper

Rhodes

Measuring the efficiency of decision making units. Eur J Oper Res. 1978;2(6):429-444.

13.

Banker

Charnes

Cooper

WW.

Some models for estimating technical and scale inefficiencies in data envelopment analysis. Manage Sci. 1984;30(9):1078-1092.

14.

Gai

Zhou

, et al. Health resource allocation and productive efficiency of Chinese county hospitals: data from 1993 to 2005. Biosci Trends. 2010;4(5):218-224.

15.

Caballer-Tarazona

Moya-Clemente

Vivas-Consuelo

Barrachina-Martínez

A model to measure the efficiency of hospital performance. Math Comput Model. 2010;52(7-8):1095-1102.

16.

Sulku

SN.

The health sector reforms and the efficiency of public hospitals in Turkey: provincial markets. Eur J Public Health. 2012;22(5):634-638.

17.

Gholami

Añón Higón

Emrouznejad

Hospital performance: efficiency or quality? Can we have both with it?

Expert Syst Appl. 2015;42:5390-5400.

18.

Fragkiadakis

Doumpos

Zopounidis

Germain

Operational and economic efficiency analysis of public hospitals in Greece. Ann Oper Res. 2016;247:787-806.

19.

Kawaguchi

Tone

Tsutsui

Estimation of the efficiency of Japanese hospitals using a dynamic and network data envelopment analysis model. Health Care Manag Sci. 2014;17(2):101-112.

20.

Simar

Wilson

PW.

Sensitivity analysis of efficiency scores: how to bootstrap in nonparametric frontier models. Manage Sci. 1998;44:49-61.

21.

Simar

Wilson

PW.

A general methodology for bootstrapping in non-parametric frontier models. J Appl Stat. 2000;27(6):779-802.

22.

Daraio

Simar

Advanced Robust and Nonparametric Methods in Efficiency Analysis: Methodology and Applications. Springer; 2007.

23.

George Assaf

Barros

Matousek

Productivity and efficiency analysis of Shinkin banks: evidence from bootstrap and Bayesian approaches. J Bank Finance. 2011;35(2):331-342.

24.

Wijesiri

Meoli

Productivity change of microfinance institutions in Kenya: a bootstrap Malmquist approach. J Retailing Consum Serv. 2015;25:115-121.

25.

Delfín-Ortega

Navarro-Chávez

CL.

Technical efficiency in the container terminals in Mexico, 1982-2010: through data envelopment analysis (DEA). iBusiness. 2013;05(04):154-160.

26.

Örkcü

Balıkçı

Dogan

Genç

An evaluation of the operational efficiency of Turkish airports using data envelopment analysis and the Malmquist productivity index: 2009–2014 case. Transp Policy. 2016;48:92-104.

27.

Balcombe

Davidova

Latruffe

The use of bootstrapped Malmquist indices to reassess productivity change findings: an application to a sample of Polish farms. Appl Econ. 2008;40(16):2055-2061.

28.

Baležentis

Dynamics of the total factor productivity in Lithuanian family farms with a statistical inference: the bootstrapped Malmquist indices and multiple correspondence analysis. Economi Res Ekonomska Istrazivanja. 2016;29(1):643-664.

29.

Cheng

Cai

Tao

, et al. Efficiency and productivity measurement of rural township hospitals in China: a bootstrapping data envelopment analysis. BMJ Open. 2016;6(11):e011911.

30.

Wang

Fang

Tao

, et al. Bootstrapping data envelopment analysis of efficiency and productivity of county public hospitals in Eastern, Central, and western China after the public hospital reform. Curr Med Sci. 2017;37(5):681-692.

31.

Dong

Zuo

Guo

, et al. Data envelopment analysis for relative efficiency measurement of Chinese hospitals: a systematic review. Res Health Sci. 2017;2(2):79-103.

32.

Dong

Measuring and benchmarking technical efficiency of public hospitals in Tianjin, China. Inq J Heal Care Organ Provis Financ. 2015;52:1-5.

33.

Fried

Lovell

CAK

Schmidt

Yaisawarng

Accounting for environmental effects and statistical noise in data envelopment analysis. J Prod Anal. 2002;17(1/2):157-174.

34.

Efron

Bootstrap methods: another look at the Jackknife. Ann Stat. 1979;7(1):1-26.

35.

Atkinson

Wilson

PW.

Comparing mean efficiency and productivity scores from small samples: a bootstrap methodology. J Prod Anal. 1995;6(2):137-152.

36.

Simar

Wilson

PW.

Estimating and bootstrapping Malmquist indices. Eur J Oper Res. 1999;115(3):459-471.

37.

Simar

Wilson

PW.

Estimation and inference in two-stage, semi-parametric models of production processes. J Econom. 2007;136(1):31-64.

38.

Method and Empirical Study of Three-Stage DEA Model. Peking Union Medical College; 2013 [in Chinese].

39.

YC.

The productive efficiency of Chinese hospitals. China Econ Rev. 2011;22(3):428-439.

40.

Caves

Christensen

Diewert

WE.

The economic theory of index numbers and the measurement of input, output, and productivity. Econometrica. 1982;50(6):1393-1414.

41.

Fare

Grosskopf

Lovell

CAK

. Production Frontiers. Cambridge University Press; 1994.

42.

Färe

Grosskopf

Norris

Productivity growth, technical progress, and efficiency change in industrialized countries: reply. Am Econ Rev. 1994;87(5):1040-1044.

43.

Wilson

PW.

FEAR: a software package for frontier efficiency analysis with R. Socioecon Plann Sci. 2008;42(4):247-254.

44.

GBD 2016 Healthcare Access and Quality Collaborators. Measuring performance on the healthcare access and quality index for 195 countries and territories and selected subnational locations: a systematic analysis from the Global Burden of Disease Study 2016. Lancet. 2018;391(2):2236-2271.

45.

Cui

Gao

Assessing productivity development of public hospitals: a case study of Shanghai, China. Int J Environ Res Public Health. 2020;17:6763-6772.

46.

Yin

Chen

Zhuo

Tao

Efficiency comparison of public hospitals under different administrative affiliations in China: a pilot city case. Healthcare. 2021;9:437-458.

47.

Al-Hanawi

Makuta

IF.

Changes in productivity in healthcare services in the Kingdom of Saudi Arabia. Cost Eff Resour Alloc. 2022;20:3-11.

48.

Wang

Efficiency and productivity of county-level public hospitals based on the data envelopment analysis model and Malmquist Index in Anhui, China. Chin Med J. 2017;130:2836-2843.

49.

The State Council of China. Opinions of the State Council on Deeping the Reform of the Medical and Health Care System. 2009. Accessed April 9, 2023. http://www.gov.cn/zhengce/content/2009-07/23/content_6221.htm.

50.

Wang

Yue

A study on analysis of the dynamic changes of general hospital efficiency: evidences based on Guangzhou city panel data. Chin J Soc Med. 2015;6:437-439.

51.

Nghiem

Coelli

Barber

Sources of productivity growth in health services: a case study of Queensland Public Hospitals. Econ Anal Policy. 2011;41(1):37-48.

52.

Yip

Hsiao

China's health care reform: a tentative assessment. China Econ Rev. 2009;20(4):613-619.

53.

Cordero

García-García

Lau-Cortés

Polo

Efficiency and productivity change of public hospitals in Panama: do management schemes matter?

Int J Environ Res Public Health. 2021;18:8630-8649.

The Regional Total Factor Productivity Changes of Healthcare Delivery in China: A Bootstrap Malmquist Data Envelopment Analysis

Abstract

Keywords

Introduction

Methods

Research Design

Input and Output Variables

Malmquist Productivity Index

Bootstrap-Malmquist-DEA Model

Data Collection, Processing and Analysis

Results

Descriptive Analysis of Input and Output Variables

Bootstrap MPI and Its Components

Ranking of TFPC, TEC, and TC

Discussion

Limitations

Conclusions and Recommendations

Footnotes

Declaration of Conflicting Interests

Funding

Ethics Approval

Consent for Publication

ORCID iD

References