Abstract
Background
Integrated care has become a global focus in efforts to improve health system performance and service delivery. China began exploring the construction of integrated care and the formation of county medical communities to improve county-level medical services in 2017. The Fujian–Sanming model has attracted extensive attention during this process. In this study, we used Sanming as the example, and the data envelopment analysis (DEA) model and Malmquist index were used to analyze the effectiveness and summarize the construction experience of advanced areas. We discuss the overall medical and health service efficiency in the county as it relates to the county medical community policy.
Method
We referenced previous studies to select three input indicators and two output indicators that represent the medical and health services in the county. We then conducted an efficiency analysis of 10 counties in Sanming from both the static and dynamic perspectives using a combination of the traditional DEA-BCC model and DEA-Malmquist index. In addition, we measured the efficiency of the primary medical institutions in Sanming.
Results
The average technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE) improved from 0.814, 0.907, and 0.898 to 0.917, 0.965, and 0.948, respectively, from 2017 to 2022. These improvements reflected a better use of resources, improved management practices, and more appropriate allocations of the service capacity. The DEA–Malmquist index results showed that the total factor productivity (TFP) of the medical and health services in the counties of Sanming was 0.957 from 2017 to 2022, slightly lower than 1. This result indicated that resource allocation had not yet reached an optimal state, possibly due to the COVID-19 pandemic impact. In addition, disparities between counties persisted, and the primary healthcare institutions generally exhibited lower efficiency levels.
Conclusions
The results of this study demonstrated that under the county medical community policy, the efficiency of medical and healthcare services in Sanming improved. The study results suggest that future efforts should focus on strengthening county hospitals, enhancing the capacity of primary care institutions, and promoting coordinated development across counties. The Sanming model offers valuable insights, particularly for developing countries, regarding strategies to advance integrated care reform.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s13561-026-00747-7.
Keywords: Integrated care, County medical community, Medical and health service efficiency, Sanming model, Data envelopment analysis
Introduction
Accelerated population aging, increases in the chronic disease prevalence, increases in disease economic burdens, and increases in the complexity of disease diagnoses and the population’s increased health awareness have caused countries around the world to propose the construction of integrated healthcare service systems [1] to improve the quality of healthcare services and reduce medical costs. For example, the United Kingdom has adopted a top-down approach led by the National Health Service (NHS), which integrates efforts from local councils, voluntary sectors, and healthcare providers to form Integrated Care Systems (ICSs). Within each ICS, Integrated Care Partnerships (ICPs) are responsible for developing long-term strategies for local health services [2]. Germany also follows a top-down model, in which Integrated Care Networks (ICNs) are formed by various stakeholders in healthcare and enter into Integrated Care Contracts (ICCs) with payers to deliver coordinated services [3]. In contrast, the United States has pursued a bottom-up integration path. Built on Medicaid and Medicare initiatives, integration is led by Accountable Care Organizations (ACOs), which voluntarily organize to provide a range of coordinated and community-based health services [4]. Singapore implements a national-level top-down model through the establishment of Regional Health Systems (RHSs), led by major public hospitals. In addition, targeted programs such as the CARITAS Integrated Dementia Care initiative provide population-specific services in designated regions [5].
China also faces a shortage of medical resources. To solve the problem of the fragmentation of medical services, establish an effective hierarchical diagnosis and treatment pattern, and enhance the efficiency of the use of medical resources, China issued a policy on the guidelines for promoting the construction and development of medical alliance in 2017. The policy formally proposed the integration of rural management and a comprehensive pilot construction of a compact county medical community [6]. A compact county medical community refers to a three-tier county healthcare service system that includes county hospitals, township healthcare centers, and village clinics that aim to further improve the county healthcare service system and enhance the allocation and efficiency of healthcare resources. Compared with OECD countries, China faces more severe constraints in both financial capacity and healthcare infrastructure. Therefore, China has developed a nationally distinctive model: a top-down policy framework led by central government mandates to establish medical communities, complemented by bottom-up innovations initiated by local governments. For example, the ‘Sanming model’ represents a flagship case under this background [7]. Sanming is in Fujian Province in southeastern China, and it ranks middle in terms of economic output and local revenue. As part of its health system reform, Sanming integrated all of the county, township, and village public healthcare organizations in the county into one entity in 2017, with the county hospital as the leader. This was done to construct a compact county medical community with shared benefits and responsibilities. Sanming provided a template for the construction of county medical communities in China [8, 9]. According to the statistics, primary medical institutions visits in Sanming had reached 57.36% of the total visits in 2020, an increase of 16.8% over the pre-implementation period (2016) of the implementation of the medical community. In terms of health management, by the end of 2020, more than 80% of patients with hypertension, type II diabetes, severe mental disorders, and tuberculosis had been under standardized management, and the premature mortality rate of major chronic diseases had decreased from 13.05% to 11.68% in 2017.
While many scholars have acknowledged that integrated care can, to some extent, improve hospital operational efficiency, reduce healthcare costs, and enhance chronic disease management [10–12], others have questioned its actual effectiveness in practice. Evidence from high-income countries suggests that the outcomes of integrated care reforms have not always met expectations, and in some cases, have even led to a decline in care quality [13–15]. China as a developing country with relatively limited economic and healthcare resources, the effectiveness of integrated care policies, especially the compact county medical community model, deserves critical evaluation after five years of implementation. Previous studies have provided some insight into the effects of these reforms. For example, Yuan J et al. used an interrupted time series to explore the changes in healthcare efficiency indicators after the implementation of the healthcare community policy. They found that indicators such as the number of outpatient and emergency room visits and the number of discharged patients in county-level hospitals increased [16]. Several scholars have explored indicators such as the county consultation rate, primary care consultation rate, and flow of health insurance funds. These scholars pointed out that the construction of county medical communities would enhance the level of medical services in a county [17–20]. However, the current discussion regarding the overall county healthcare services remain incomplete as there exists a lack of exploration of the impact of the county medical community on the overall efficiency of county healthcare services. Therefore, it is necessary to conduct a comprehensive empirical study that addresses the effect of county medical community construction on the overall efficiency of county medical and health services. This would provide both theoretical and practical insights into the implementation of integrated care, and offer valuable empirical evidence on such reforms in developing countries, where existing research remains limited, particularly in light of the conflicting findings observed in high-income countries.
Data Envelopment Analysis (DEA) is a well-established quantitative method for assessing the relative efficiency of decision-making units (DMUs) with multiple inputs and outputs. Originally developed by Charnes, Cooper, and Rhodes in 1978, DEA uses linear programming to construct a production frontier and determine the efficiency of each unit relative to peers [30]. Given its suitability for analyzing multi-input, multi-output systems, DEA has been widely applied in evaluating the performance of healthcare systems since the 1980 s [21]. Whatmore, this approach also aligns with theoretical perspectives from public economics, where healthcare services are treated as quasi-public goods and resource allocation efficiency is a key criterion in policy evaluation. Two classic DEA models are commonly used: the CCR model, assuming constant returns to scale, and the BCC model, which accounts for variable returns to scale and decomposes efficiency into scale and technical components. To address the limitations of traditional DEA—such as the inability to control for environmental influences or statistical noise—Fried et al. introduced a three-stage DEA model that integrates stochastic frontier analysis (SFA), enabling a more accurate reflection of intrinsic efficiency [22]. In addition, Färe et al. developed the DEA–Malmquist index to capture changes in productivity over time, offering insights into dynamic efficiency across different periods [23].
Therefore, in this study, we use the DEA to measure the role and impact of the construction of a compact county medical community in Sanming on the overall efficiency of medical and health services in the county. The aim is to help the compact county medical community further develop at the practical and theoretical levels, enhance the level of county medical and health service supply capacity, and perform policy optimization. Moreover, by shedding light on the effectiveness of integrated care reforms in a developing country, this study contributes to the broader global discourse on health system integration and offers valuable insights for other low- and middle-income countries exploring similar reform paths.
Methods
Study design and data collection
This study was conducted to assess and compare the technical efficiency of the national healthcare system of Sanming, Fujian Province, in China that includes 10 counties: Sha County, Ninghua County, Qingliu County, Mingxi County, Datian County, Jianning County, Youxi County, Yongan City (county-level city), Taining County, and Jiangle County. We used a DEA-based Malmquist approach that covered the time-period from 2017 to 2022. Each of the 10 counties was considered to be decision making units for conducting the DEA. The data were collected through on-site visits to ten county medical communities in Sanming (Jan–May 2022), primarily from institutional administrative records and local health statistics reports provided by the County Health Bureaus.
Measures
Input and output indicators
In a DEA, the set of variables (inputs and outputs) that need to be included in the model should meet the following criteria: the number of decision-making units (DMUs) should be greater than twice the product of the input and output indicators [24]. Since there were 10 DMUs in this study, the number of input and output indicators should not be more than 5. Moreover, the input and output variables should constitute a set of factors common to all of the units under evaluation [25].
Thus, it was critical to review current published research for evidence of indicators choice. In accordance with published research [25–28], the actual variables selected for this study primarily considered the capacity of the healthcare institutions that are the most commonly used in hospital efficiency evaluations [29]. A summary of the input/output variables with the selection rationale is provided in Table 1. In addition, Charnes et al. pointed out that economic-based indicators are not applicable when discussing nonprofit decision-making units [30]; hence, in this study, economic and financial indicators, such as healthcare revenues, were excluded.
Table 1.
DEA input and output indicators
| Variable | Name | Description | Justification |
|---|---|---|---|
| Inputs | |||
| I1 | Actual open beds | Total operational hospital beds |
Core resource indicator reflects the service capacity and are standard in healthcare DEA studies |
| I2 | Employees on board | Full-time clinical/non-clinical staff | Measures human resource input and are essential for the productivity analysis |
| I3 | Equipment (> $10,000) | High-value medical devices (e.g., MRI, CT scanners) | Captures the capital investment and aligns with WHO resource tracking standards |
| Outputs | |||
| O1 | Inpatient admissions | The number of inpatient admissions within the county | Primary service volume metric adopted from the NHCE evaluation frameworks |
| O2 | Outpatient & emergency visits | The number of outpatient and emergency room visits within the county | Reflects accessibility and service utilization; validated in county-level DEA |
Model selection
BCC model
The DEA model includes the number of DMUs (n), and under the assumption that each DMU contains the number of input variables (m) and output variables (s), Eq. (1) represents the DEA model.
 |
1 |
The BCC model was more appropriate for this study due to the varying scale effects during the construction and development process of healthcare communities. Based on the CCR model, the BCC model divides the total technical efficiency (TE) into the pure TE (PTE) and scale efficiency (SE). The efficiency of healthcare in a county is measured on a scale of zero to one. A value closer to 1 indicates a higher efficiency. The BCC model defines θ as the technical efficiency of health services in the county medical community. ε is a non-Archimedean infinitesimal, while S–S + are the slack variables (where S + is the output term difference variable, and S- is the input term difference variable). X0 is the slack-adjusted input variable, and Y0 is the slack-adjusted output variable. If θ is equal to 1 and both S- and S + are equal to 0, the DEA is effective in the county and maximizes the use of healthcare resources. The BCC model Eq. (2) is as follows:
 |
2 |
Malmquist index
The DEA–Malmquist index was chosen in this study to complement the DEA–BCC model for the overall dynamic analysis and the cross-period comparisons of the panel data. The equation for the Malmquist index is shown in Eq. (3):
 |
3 |
where Mt represents the effect of technical efficiency on total productivity from t to t + 1, Xt, and Xt+1, and Yt and Yt+1 represent the inputs and outputs, respectively, during the period from t to t + 1. Dt(Xt, Yt) and Dt(Xt+1, Yt+1) indicate the efficiency of the DMU during period t and period t + 1 at the technology level of period t and at the production frontier.
The MPI is the advantage of dividing the total factor productivity (TFP) into its two components: the technical efficiency change (EFFCH) and the technological change (TECHCH). Further, the EFFCH can be subdivided into the pure technical efficiency change (PECH) and the scale efficiency change (SECH) [31]. The final equation can be expressed as Eq. (4):
 |
4 |
The Malmquist index is a measure of dynamic change. If M is greater than 1, the DMU experiences an improvement in efficiency from period t to period t + 1, and this indicates an improvement in efficiency of county health services, and vice versa, the efficiency decreases. In addition, the PECH, SECH, and TECHCH represent the operational and management level, scale efficiency, and technological innovation level, respectively, of the county medical community.
Statistical analysis
We used DEAP2.1 software for the DEA and Malmquist index analysis to quantitatively analyze the effectiveness of healthcare services in the county after the construction of the integrated medical community.
Results
Descriptive input and output
The baseline characteristics of the inputs/outputs across the 10 counties are summarized in Table 2, and the longitudinal trends (2017–2022) in health services in county-level medical institutions and primary medical institutions are visualized in Fig. 1. It should be noted that fixed output weights were not imposed, as there is no clear policy-based rationale or established weighting scheme to prioritize one indicator over the other.
Table 2.
Descriptive inputs and outputs in Sanming
| DMUs | Input | Output | |||||
|---|---|---|---|---|---|---|---|
| I1 (Bed) |
I2 (Person) |
I3 (Equipment) |
O1 | O2 | |||
| County-level | Primary | County-level | Primary | ||||
| Sha County | 1040 | 1200 | 965 | 33,847 | 10,871 | 1,558,731 | 908,003 |
| Ninghua County | 1057 | 1361 | 1359 | 41,095 | 15,360 | 1,283,170 | 721,393 |
| Qingliu County | 604 | 815 | 779 | 18,759 | 7278 | 609,267 | 349,362 |
| Mingxi County | 385 | 674 | 910 | 9980 | 1063 | 574,842 | 273,983 |
| Datian County | 1110 | 1213 | 1090 | 45,956 | 12,602 | 1,841,323 | 1,337,961 |
| Jianning County | 440 | 631 | 423 | 18,067 | 6163 | 534,530 | 285,755 |
| Youxi County | 1424 | 2051 | 1694 | 58,025 | 15,997 | 2,624,751 | 1,866,497 |
| Yongan City | 1917 | 2380 | 1724 | 61,415 | 10,099 | 2,178,672 | 1,054,185 |
| Taining County | 688 | 720 | 1174 | 17,475 | 2923 | 892,755 | 504,038 |
| Jiangle County | 838 | 971 | 787 | 19,554 | 3708 | 1,139,255 | 638,028 |
Fig. 1.
Changes in health services in county-level medical institutions and primary medical institutions in Sanming City from 2017 to 2022
Comprehensive efficiency results for the data envelopment analysis of 10 county-level medical communities
Technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE) results for 2017 and 2022 are detailed in Table 3, highlighting changes in return-to-scale patterns. The comprehensive efficiencies of most of the county medical communities were less than 1, meaning that the maximum utilization of resources could have been more effective. The overall medical and health service efficiency level of the Sanming Medical Community was low, with the mean technical efficiency value being only 0.814. After five years, three county medical communities’ technical efficiencies reached 1 in 2022. In addition, the previous two counties achieved the maximum utilization of resources. The mean of the technical efficiencies increased from 0.814 to 0.917. Additionally, we tested the results using the CCR model, which assumes constant returns to scale. The efficiency scores were nearly identical to those derived from the BCC model, supporting the robustness of our model specification. We further conducted a Bootstrap–DEA analysis to assess the robustness of the efficiency scores, and the results are provided in Appendix A.
Table 3.
Comprehensive efficiency results of the DEA of the DMUs
| DMUs | Year | TE | PTE | SE | Return to Scale |
|---|---|---|---|---|---|
| Sha County | 2017 | 0.837 | 0.870 | 0.961 | irs |
| 2022 | 0.914 | 0.939 | 0.974 | irs | |
| Ninghua County | 2017 | 0.810 | 0.821 | 0.986 | irs |
| 2022 | 0.968 | 1.000 | 0.968 | drs | |
| Qingliu County | 2017 | 0.895 | 0.966 | 0.927 | irs |
| 2022 | 0.759 | 0.845 | 0.899 | irs | |
| Mingxi County | 2017 | 0.644 | 0.987 | 0.653 | irs |
| 2022 | 0.827 | 1.000 | 0.827 | irs | |
| Datian County | 2017 | 1.000 | 1.000 | 1.000 | — |
| 2022 | 1.000 | 1.000 | 1.000 | — | |
| Jianning County | 2017 | 0.992 | 1.000 | 0.992 | irs |
| 2022 | 1.000 | 1.000 | 1.000 | — | |
| Youxi County | 2017 | 1.000 | 1.000 | 1.000 | — |
| 2022 | 1.000 | 1.000 | 1.000 | — | |
| Yongan City | 2017 | 0.734 | 0.738 | 0.995 | drs |
| 2022 | 0.703 | 0.863 | 0.814 | drs | |
| Taining County | 2017 | 0.605 | 0.919 | 0.658 | irs |
| 2022 | 1.000 | 1.000 | 1.000 | — | |
| Jiangle County | 2017 | 0.623 | 0.769 | 0.810 | irs |
| 2022 | 1.000 | 1.000 | 1.000 | — | |
| Mean | 2017 | 0.814 | 0.907 | 0.898 | / |
| 2022 | 0.917 | 0.965 | 0.948 | / |
Irs Increasing return to scale, Drs decreasing return to scale
The pure technical efficiencies of seven counties were less than 1, indicating deficiencies at the technical level. In 2022, seven county-level medical communities reached 1 in the pure technical efficiency. Pure technical efficiency increased from 0.907 to 0.917 between 2017 and 2022.
The scale efficiencies of the seven counties are increasing, indicating a positive trend where their output growth rate is greater than the input growth rate. Moreover, the scale efficiencies of two medical communities remain unchanged, and the efficiency of one is decreasing. The pure technical efficiencies have reached 1 for Ninghua County and Jianning County, but the technical efficiencies were still ineffective due to the ineffective scale efficiencies. An ineffective scale efficiency means the scale is relatively large, and the growth rate of input is greater than the output growth rate. However, there were five county medical communities that showed upward trends in the scale efficiency that increased from 0.898 to 0.948, offering a glimmer of hope.
Productivity change for the Malmquist index during a six-year period
Temporal decomposition of productivity changes (2017–2022) is presented in Table 4 and Fig. 2, revealing three distinct evolutionary phases. The mean of TFP among the 10 DMUs was 0.957. The TE increased by 2.6%, and the PTE increased by 1.3%. These results indicated the progress of the technical efficiency during the process of building a county medical community. The scale efficiency increased by 1.3% during the same time, reflecting a trend of gradually moving closer to the optimal scale. However, it should be noted that the technological change dropped by 6.7%, which meant that less than 1 needed to improve. In that case, there is still room for progress in the quality of healthcare services related to technological development in the construction of county medical communities. Overall, the current technological progress within the medical communities was the primary driving factor for the development of healthcare services in the county, while the level of management of the medical communities, scale efficiency, and other factors primarily led to the fluctuating TFP development.
Table 4.
Malmquist indices during a five-year period
| Year | EFFCH | TECHCH | PECH | SECH | TFP |
|---|---|---|---|---|---|
| 2017–2018 | 1.014 | 0.932 | 1.051 | 0.965 | 0.945 |
| 2018–2019 | 1.018 | 1.000 | 1.007 | 1.010 | 1.018 |
| 2019–2020 | 1.030 | 0.857 | 1.018 | 1.012 | 0.882 |
| 2020–2021 | 1.052 | 0.943 | 0.994 | 1.059 | 0.992 |
| 2021–2022 | 1.017 | 0.937 | 0.998 | 1.020 | 0.953 |
| Mean | 1.026 | 0.933 | 1.013 | 1.013 | 0.957 |
EFFCH Technical efficiency change, TECHCH Technological change, PECH Pure technical efficiency change, SECH Scale efficiency change, TFP Total factor productivity changes
Fig. 2.
Malmquist indices during a five-year period
The index changes could be divided into three major stages from a time perspective, as shown in Fig. 2. The first stage (2017–2019): During this period, the Malmquist index rose from 0.945 to 1.018 with an increase of 7.3%. The composition of the Malmquist indices all showed upward trends, demonstrating the improvement in the efficiency of county medical and health services during the first two years with the construction of the medical community. The second phase (2019–2020): The Malmquist index declined from 1.018 to 0.882 during this phase. Although the PECH and the SECH still showed upward trends, the TECHCH fell by 14.3%, which ultimately led to a decline in the TFP primarily due to the severe impact of the COVID-19 pandemic. The third phase (2020–2022): The Malmquist index during this phase primarily showed a trend of rising volatility. It was affected by factors such as epidemic fluctuations and national policy, and the medical resources within counties were often involved in anti-epidemic efforts. Therefore, indicators such as the management level of the medical community and the scale of marginal remuneration all rebounded slowly.
Productivity changes of the Malmquist index by the DMUs
Table 5 shows that in the five years since the construction of the county medical community was implemented, the TFP values of Mingxi County, Taining County, and Jiangle County were more significant than one, which indicated a better level of healthcare service efficiency. Among them, all of the efficiency indicators of Mingxi County were greater than one, reflecting that its technological development, operation management, marginal control, and other aspects were at the leading level in Sanming. The Taining County and Jiangle County technical efficiencies and scale efficiencies values were greater than 1; however, their technological changes were only 0.945 and 0.925. Except for the above three counties, the TFP values of the other counties were less than 1. The TECHCH values of Sha County, Jianning County, Datian County, and Youxi County were all less than 1, and the other efficiency indicators were all greater than 1. These results indicated that in these four counties, due to interference from external factors such as the epidemic, they experienced slow technological development changes that had directly restricted the progress of the TFP. In addition to the counties mentioned above, Ninghua County and Yongan City also experienced issues of scale efficiency. As mentioned above, Yongan City, the only county-level city in Sanming, may have received large early investments but experienced slow current returns and diminishing returns of scale.
Table 5.
Malmquist indices by the DMUs
| DMUs | EFFCH | TECHCH | PECH | SECH | TFP |
|---|---|---|---|---|---|
| Sha County | 1.018 | 0.900 | 1.015 | 1.003 | 0.916 |
| Ninghua County | 1.036 | 0.939 | 1.040 | 0.996 | 0.973 |
| Qingliu County | 0.968 | 0.945 | 0.974 | 0.994 | 0.915 |
| Mingxi County | 1.051 | 1.007 | 1.003 | 1.049 | 1.059 |
| Datian County | 1.000 | 0.918 | 1.000 | 1.000 | 0.918 |
| Jianning County | 1.002 | 0.912 | 1.000 | 1.002 | 0.913 |
| Youxi County | 1.000 | 0.955 | 1.000 | 1.000 | 0.955 |
| Yongan City | 0.991 | 0.886 | 1.032 | 0.961 | 0.878 |
| Taining County | 1.106 | 0.945 | 1.017 | 1.087 | 1.046 |
| Zongle County | 1.099 | 0.925 | 1.054 | 1.043 | 1.017 |
| Mean | 1.026 | 0.933 | 1.013 | 1.013 | 0.957 |
Technical efficiency of primary medical institutions-case study
Primary healthcare improvements are critical issues that require attention during the medical community construction process. As a pioneer of reform, the Sha County medical community has become a typical model in Sanming’s healthcare reform. Therefore, we chose to study Sha County as an example to focus on the technical efficiency progress at a more micro level. We further selected two indicators, the “chronic disease management rate” and the “occupancy rate of hospital beds,” to explore the TE of primary medical institutions in Sha County, as shown in Table 6. Further, Table 7 shows the DEA-Malmquist analysis conducted in 12 township health centers in Sha County.
Table 6.
Descriptive inputs and outputs in Sha county
| DMUs | Number of actual open beds | Number of employees | Fixed assets | occupancy rate of hospital beds | Chronic disease management rate |
|---|---|---|---|---|---|
| Xiamao | 76 | 65 | 15,091,192.70 | 63.23% | 92.00% |
| Gaoqiao | 31 | 35 | 9,439,754.19 | 60.44% | 82.16% |
| Gaosha | 23 | 29 | 5,218,778.90 | 44.50% | 70.66% |
| Fukou | 30 | 25 | 5,351,492.15 | 67.40% | 44.64% |
| Nanxia | 15 | 22 | 2,932,426.96 | 43.95% | 55.18% |
| Daluo | 25 | 24 | 4,890,789.53 | 29.37% | 51.84% |
| Huyuan | 15 | 16 | 3,835,455.13 | 55.86% | 83.54% |
| Nanyang | 20 | 22 | 4,636,926.05 | 22.42% | 78.13% |
| Qingzhou | 20 | 25 | 3,699,982.83 | 34.30% | 70.71% |
| Zhenhu | 26 | 20 | 3,363,884.42 | 40.38% | 65.55% |
| Chengqu | 57 | 91 | 13,699,404.30 | 71.26% | 93.53% |
| Langkou | 11 | 28 | 4,974,529.47 | 35.71% | 93.15% |
Table 7.
Malmquist index in Sha county during a five-year period
| Year | EFFCH | TECHCH | PECH | SECH | TPF |
|---|---|---|---|---|---|
| 2017–2018 | 0.962 | 0.949 | 0.996 | 0.966 | 0.912 |
| 2018–2019 | 1.061 | 0.865 | 1.014 | 1.046 | 0.918 |
| 2019–2020 | 0.927 | 1.014 | 1.010 | 0.917 | 0.940 |
| 2020–2021 | 0.898 | 1.006 | 0.927 | 0.969 | 0.903 |
| 2021–2022 | 1.117 | 0.858 | 1.073 | 1.041 | 0.959 |
| Mean | 0.989 | 0.936 | 1.003 | 0.987 | 0.926 |
The TFP in Sha County was 0.926, less than 1, meaning it has not yet reached complete efficiency optimization. Among the indicators, the PTE was greater than 1, indicating improvements in the technical efficiency within five years of implementing the medical community reform. Similar the trends of the county-level changes, during the early stages of the construction of the medical community, the TFP developed during this period due to factors such as “close linkage among primary, secondary, and tertiary healthcare” and county-level medical institutional increases in manpower, material, and financial resources to support the primary healthcare. With the recurrence of the epidemic, more resources within the medical community have been deployed to support the fight against the epidemic. This has resulted in the interruption of primary healthcare, and daily management work such as health testing has been affected. These factors have led to a decrease in the overall medical and health service efficiency. Since then, the country has implemented new post-transition policies, and primary healthcare institutions have assumed a larger share of COVID-19 admissions at the end of 2022. The overall trend is shown in Fig. 3.
Fig. 3.
Malmquist index in Sha county in five-year period
Discussion
Sanming, China was used as a case study to examine the efficiency of county-level healthcare services under the compact county medical community policy. Results from the DEA–BCC model indicated a clear improvement in the overall healthcare service efficiency across the region. After five years of policy implementation, half of the counties achieved overall efficiency, 70% reached optimal pure technical efficiency, and 80% showed increasing returns to scale. These outcomes reflect notable advancements in the technical capacity and service delivery at the county level, suggesting that the policy has facilitated both the rational use of resources and the optimization of healthcare workflows.
The results of the Malmquist index revealed that although the efficiency of healthcare services within Sanming’s county medical communities exhibited an overall upward trend over the five-year period, notable fluctuations occurred. Notably, changes in the total factor productivity were closely aligned with variations in technological progress. During the early stages of the medical community reform, the local government in Sanming made significant investments in personnel, funding, and equipment. Sanming also improved integrated management mechanisms [8, 32–34]. These efforts contributed to initial efficiency gains. However, with the global spread of the COVID-19 pandemic, both developed and developing countries have been impacted, inevitably leading to a conflict between health equity and efficiency [35, 36], ultimately resulting in a decline in overall efficiency within counties. As the epidemic situation stabilized and national policies shifted, the focus of reform returned to internal system building within the medical communities. This resulted in a subsequent rebound in overall efficiency.
We also identified notable efficiency disparities among different counties within the Sanming jurisdiction in this study. Geographic location appeared to influence the performance of medical institutions to varying degrees [37]. Counties located in the northwest of Sanming demonstrated relatively higher total factor productivity compared to those in the south. One possible explanation is that southern counties, such as Yongan that have larger populations and land areas, invested heavily during the early stages of reform to establish leading hospitals. However, such intensive investments may have resulted in temporary mismatches between inputs and outputs, thereby undermining scale efficiency. Furthermore, the “resource siphoning” phenomenon where patients are drawn from less-developed areas to nearby cities with superior medical resources was particularly relevant for southern Sanming, given its proximity to major urban centers [38, 39], such as Xiamen, Quanzhou, and Fuzhou. Moving forward, it is recommended that health authorities conduct more accurate scale estimations to avoid over-expansion [40, 41], and place greater emphasis on cost-efficiency evaluations. In addition, efforts should be made to enhance inter-county coordination by encouraging the development of specialty services with comparative advantages, fostering cluster-based synergies, and optimizing the allocation of regional healthcare resources.
We also examined primary healthcare institutions in Shaxian County as a case study to evaluate their service efficiency. The findings revealed that the development level of primary care institutions lagged behind the overall county-level performance. Prior research by Li et al. highlighted long-standing challenges faced by China’s primary healthcare sector, including shortages of healthcare professionals, a predominant focus on treatment over prevention, inadequate funding, and limited service capacity [42–45]. Some of these issues have been partially alleviated under the county medical community policy. For example, in Shaxian, initiatives, such as community-based health management and the advancement of telemedicine, have been introduced. However, primary care institutions still trail behind county-level hospitals in terms of service quality and technical performance [39]. It is recommended in the future that primary care providers within medical communities adopt a health-oriented approach, enhance care quality and digital capabilities, and promote integration between medical services and public health. In addition, leading hospitals within the community should take a proactive role in offering technical support and facilitating resource sharing [46], thereby improving the overall capacity and efficiency of healthcare service delivery within the county medical community.
Limitations
In this study, we measured the role and impact of the construction of a compact county medical community in Sanming on the overall efficiency of healthcare services. However, there were some limitations to this study. First, due to the limited number of decision-making units (10 counties), the number of input and output indicators in the DEA model was constrained. This may have resulted in a partial representation of healthcare service efficiency. In addition, because of this limitation, this study did not include patient-level outcomes, such as health improvements, quality of care, or satisfaction. Future research should incorporate such indicators to comprehensively assess whether efficiency gains translate into actual improvements in population health. Second, the COVID-19 pandemic had a substantial impact on service delivery in Sanming between 2020 and 2022. As COVID-19 settles down, longitudinal studies with more balanced data will be valuable in reassessing efficiency trends. Third, due to the inherent limitations of DEA in causal identification, we were unable to apply methods such as difference-in-differences (DID) to isolate policy effects. Moreover, the study may be subject to potential endogeneity caused by unobserved confounding factors (e.g., differences in local governance capacity). Future research should consider broader datasets and incorporate causal inference techniques to better control for such influences.
Conclusions
Despite these limitations, this study offers substantial academic and policy relevance. Drawing on data from 10 counties in Sanming City, China, from 2017 to 2022, we evaluated changes in the efficiency of county-level medical and health services under the background of the county medical community policy. Shaxian County was selected as a case to further examine the efficiency performance of primary healthcare institutions. The results indicated that, although the efficiency fluctuated in some years due to exogenous shocks such as the COVID-19 pandemic, the overall trend remained upward. In addition, noticeable efficiency disparities were observed across counties, and primary healthcare institutions consistently exhibited lower efficiencies.
Sanming’s reform of the county medical community that was characterized by organizational restructuring, enhancement of county hospital capacity, and a focus on primary healthcare management provides practical insights for building integrated care systems in other countries or regions, particularly in low- and middle-income settings. However, successful replication requires careful adaptation to local contexts, including national institutional capacities and health policy priorities.
Methodologically, this study integrated the DEA–BCC model and the Malmquist index to assess healthcare service efficiency from both the static and dynamic perspectives. This dual approach not only extends the application scope of DEA in integrated care research but also offers a robust quantitative tool for policymakers to monitor reform outcomes and guide resource optimization.
Supplementary Information
Acknowledgements
The authors would like to express their sincere gratitude to the Sanming Municipal Government and the Sanming Health Commission for their generous support and assistance during the data collection and fieldwork phases of this study. Their collaboration was essential to the completion of this research.
Authors’ contributions
LZ was responsible for the overarching research goals (Conceptualization), designing the methodology (Methodology), drafting the manuscript (Writing–original draft). TX and XW contributed to data analysis (Formal analysis) and participated in drafting the manuscript (Writing–original draft). LZ and YL performed data cleaning (Data curation) and polished the manuscript (Writing–review & editing). HT and ZZ managed project logistics (Project administration) and oversaw the research process (Supervision). All authors read and approved the final manuscript.
Funding
This research received no external sponsorship or funding.
Data availability
No datasets were generated or analysed during the current study. All data used for the DEA analysis are presented within the article. The original administrative data were provided by the Sanming municipal authorities and are not publicly available due to institutional restrictions.
Declarations
Ethics approval and consent to participate
This study was not a study on human beings, and the data collected had no relationship with patient medical records data; therefore, an ethics statement was not needed.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Hong Tan, Email: 1434563692@qq.com.
Zongjiu Zhang, Email: zhangzongjiu@tsinghua.edu.cn.
References
- 1.Araujo de Carvalho I, Epping-Jordan J, Pot AM, Kelley E, Toro N, Thiyagarajan JA, et al. Organizing integrated health-care services to meet older people’s needs. Bull World Health Organ. 2017;95(11):756–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Thomson LJM, Chatterjee HJ. Barriers and enablers of integrated care in the UK: a rapid evidence review of review articles and grey literature 2018-2022. Front Public Health. 2024;11(4):1286479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Milstein R, Blankart CR. The health care strengthening act: the next level of integrated care in Germany. Health Policy. 2016;120(5):445–51. [DOI] [PubMed] [Google Scholar]
- 4.Sandhu S, Sharma A, Cholera R, Bettger JP. Integrated health and social care in the United States: a decade of policy progress. Int J Integr Care. 2021;21(4):9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nurjono M, Yoong J, Yap P, Wee SL, Vrijhoef HJM. Implementation of integrated care in Singapore: a complex adaptive system perspective. Int J Integr Care. 2018;18(4):4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yip W, Fu H, Chen AT, Zhai T, Jian W, Xu R, et al. 10 years of health-care reform in China: progress and gaps in universal health coverage. Lancet. 2019;394(10204):1192–204. [DOI] [PubMed] [Google Scholar]
- 7.Liu T, Shen J, Zhang X, et al. A scoping review of Chinese national policies to promote the three-medicals linkage reform since 2014. NPJ Health Syst. 2025;2:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tu WJ, Zhong SF, Liu YK, Zhan J, Liu Q. The sanming three-in-one model: a potentially useful model for China’s systemic healthcare reform. J Am Geriatr Soc. 2019;67(10):2213–5. [DOI] [PubMed] [Google Scholar]
- 9.Zhong Z, Yao Q, Chen S, Jiang J, Lin K, Yao Y, et al. China promotes sanming’s model: a national template for integrated medicare payment methods. Int J Integr Care. 2023;23(2):15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Murtagh S, McCombe G, Broughan J, Carroll Á, Casey M, Harrold Á, et al. Integrating primary and secondary care to enhance chronic disease management: a scoping review. Int J Integr Care. 2021;21(1):4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Peng Z, Zhu L, Wan G, et al. Can integrated care improve the efficiency of hospitals? Research based on 200 hospitals in China. Cost Eff Resour Alloc. 2021;19:61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rocks S, Berntson D, Gil-Salmerón A, Kadu M, Ehrenberg N, Stein V, et al. Cost and effects of integrated care: a systematic literature review and meta-analysis. Eur J Health Econ. 2020;21(8):1211–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Baxter S, Johnson M, Chambers D, et al. The effects of integrated care: a systematic review of UK and international evidence. BMC Health Serv Res. 2018;18:350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Billings J, de Bruin SR, Baan C, et al. Advancing integrated care evaluation in shifting contexts: blending implementation research with case study design in project SUSTAIN. BMC Health Serv Res. 2020;20:971. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Brekke KR, Siciliani L, Straume OR. Competition, quality and integrated health care. J Health Econ. 2024;95:102880. [DOI] [PubMed] [Google Scholar]
- 16.Yuan J, Ye L, Chen XJ, et al. Evaluation of health service efficiency of county medical alliance reform based on interruptedtime series analysis. Chin Hosp Manage. 2021;41(12):69–72. [Google Scholar]
- 17.Dai WY, Hu L. Exploration and practice of primary healthcare during the construction of county-level medical alliance. Chin Gen Pract. 2021;24(1):23–9. [Google Scholar]
- 18.Shen LJ, Huang CF, Li LL, et al. Exploration and practice of county medical community model——taking Tianchang City, Anhui Province as an example. Health Econ Res. 2018;12:7–11. [Google Scholar]
- 19.Yu JX, Tu YX, Wu C. Exploring the Chinese approach of an integrated medical and health service system-a comparative study of medical community for the county from Anhui,Shanxi and Zhejiang Province. Govern Stud. 2020;36(01):5–15+2. [Google Scholar]
- 20.Zhao MJ, Jia M, Wang F, et al. Measures and effects of the pilot of county-based healthcare network in Deqing county of Zhejiang province. Chin J Health Policy. 2019;12(11):53–8. [Google Scholar]
- 21.Sherman HD. Hospital efficiency measurement and evaluation. Empirical test of a new technique. Med Care. 1984;22(10):922–38. [DOI] [PubMed] [Google Scholar]
- 22.Fried HO, Lovell CAK, Schmidt SS, Yaisawarng S. Accounting for environmental effects and statistical noise in data envelopment analysis. J Prod Anal. 2002;17(1):157–74. [Google Scholar]
- 23.Farrell M. The measurement of productive efficiency. J R Stat Soc Ser A Gen. 1957;120(3):253–90. [Google Scholar]
- 24.Khezrimotlagh D, "How to Deal with Numbers of Decision-Making Units and Number of Variables in Multiple Input-Output Production Functions", Applications of Management Science, Kenneth D. Lawrence, Dinesh R. Pai. 2020.
- 25.Flokou A, Aletras V, Niakas D. A window-DEA based efficiency evaluation of the public hospital sector in Greece during the 5-year economic crisis. PLoS One. 2017;12(5):e0177946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Dimas G, Goula A, Soulis S. Productive performance and its components in Greek public hospitals. Oper Res. 2012;12(1):15–27. [Google Scholar]
- 27.Lu D, Yang X. Analysis the operation efficiency of county medical community based on DEA in Wujin District Medical Alliance Model. Jiangsu Health Syst Manag. 2022;2:149-152,163. [Google Scholar]
- 28.Wei JM, Gao QL, Huang MZ, Dong HJ. Study on the operation efficiency of county medical community based on DEA Model. Chin J Health Policy. 2021;14(2):23–7. [Google Scholar]
- 29.Imani A, Alibabayee R, Golestani M, Dalal K. Key indicators affecting hospital efficiency: a systematic review. Front Public Health. 2022;10:830102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Charnes A, Cooper WW, Rhodes E. Measuring the efficiency of decision making units. Eur J Oper Res. 1978;2(6):429–44. [Google Scholar]
- 31.Iliyasu A, Mohamed ZA, Hashim M. Productivity growth, technical change and efficiency change of the Malaysian cage fish farming: an application of Malmquist Productivity Index approach. Aquacult Int. 2015;23(4):1013–24. [Google Scholar]
- 32.Lin K, Liu X, Xiang L, Luo F. Sanming model of medical service integration: impact on medical expenditures, service provision, and resource allocation in China. Risk Manag Healthc Policy. 2025;18:205–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Qi B, Xin L, Li Y. Towards a more globally inclusive academic discussion on local government: is more attention to Asian countries the answer? Local Gov Stud. 2024;50(6):1153–65. [Google Scholar]
- 34.Zhao L, Li Y, Zhu L, Wang X, Xin T, Zhang Z. Construction and application of an evaluation index system for compact county medical community development. Chin J Hosp Adm. 2025;41(03):172–7. [Google Scholar]
- 35.Vrabková I, Lee S. Approximating the influence of external factors on the technical efficiency score of hospital care: evidence from the federal states of Germany. Health Econ Rev. 2023;13:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Selamzade F, Ersoy Y. Health efficiency measurement of Turkish Provinces using DEA, TOPSIS and EDAS methods. In: Emrouznejad A, Thanassoulis E, Toloo M, editors. Advances in the theory and applications of performance measurement and management. DEA45 2023. Lecture notes in operations research. Cham: Springer; 2024. [Google Scholar]
- 37.Alatawi AD, Niessen LW, Khan JAM. Determinants of technical efficiency in public hospitals: the case of Saudi Arabia. Health Econ Rev. 2020;10:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Li L, Cui X, Feng W. Enhancing patient satisfaction in cross-regional healthcare: a cross-sectional study in the knowledge-based healthcare landscape. J Knowl Econ. 2024;15:14172–98. [Google Scholar]
- 39.Yang X, Chen Y, Li C, Hao M. Effects of medical consortium policy on health services: an interrupted time-series analysis in Sanming, China. Front Public Health. 2024;12(24):1322949. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Giancotti M, Guglielmo A, Mauro M. Efficiency and optimal size of hospitals: results of a systematic search. PLoS One. 2017;12(3):e0174533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Xiao X, Shen K, Zheng X, Wu D, Pei T, Lin XH, et al. Resource allocation efficiency in 68 county-level traditional Chinese medicine hospitals in China: a data envelopment analysis. BMJ Open. 2024;14(10):e088462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ding S, Zhou Y. County medical community, medical insurance package payment, and hierarchical diagnosis and treatment-empirical analysis of the impact of the pilot project of compact county medical communities in Sichuan Province. PLoS One. 2024;19(4):e0297340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Jakovljevic M, Chang H, Pan J, Guo C, Hui J, Hu H, et al. Successes and challenges of China’s health care reform: a four-decade perspective spanning 1985–2023. Cost Eff Resour Alloc. 2023;21(1):59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Li X, Krumholz HM, Yip W, Cheng KK, De Maeseneer J, Meng Q, et al. Quality of primary health care in China: challenges and recommendations. Lancet. 2020;395(10239):1802–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Zeng Z, Luo Y, Tao W, et al. Improving access to primary health care through financial innovation in rural China: a quasi-experimental synthetic difference-in-differences approach. BMC Prim Care. 2024;25:195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Dong C, Liu J, Mi J. Information-driven integrated healthcare: an analysis of the cooperation strategy of county medical community based on multi-subject simulation. Healthcare. 2023;11(14):2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
No datasets were generated or analysed during the current study. All data used for the DEA analysis are presented within the article. The original administrative data were provided by the Sanming municipal authorities and are not publicly available due to institutional restrictions.