Temporal–spatial analysis and driving factors of coupling coordination between population and health services systems in China, 2010–2022

Abstract

Background

Demographic transitions are reshaping health needs and challenging the sustainability of health governance, with the coordination of population and health services systems closely connected to these processes. This study investigates China’s population and health services systems from 2010 to 2022, aiming to explore the spatiotemporal evolution and key driving factors of their coupling coordination.

Methods

Data were obtained from the China Statistical Yearbook and China Health Statistics Yearbook, covering 31 provinces from 2010 to 2022. An evaluation model was developed to assess the coupling coordination degree (CCD) between the population and health services systems. Spatial autocorrelation analysis was employed to examine spatial dependence, and a Spatial Durbin Model (SDM) was applied to identify key driving factors.

Results

From 2010 to 2022, the development indices of China’s population and health services systems rose steadily, with the CCD increasing from 0.407 to 0.835, indicating a progression from initial imbalance toward advanced coordination. Regionally, the pattern of “east–strong, central–rising, west–weak” persisted, though disparities have diminished as central and western regions steadily caught up. Spatial analysis confirmed significant dependence and stable in provincial CCD, with patterns dominated by “high–high” clusters in the Yangtze River Delta and Beijing–Tianjin–Hebei region, which gradually extended into central provinces, and by persistent “low–low” clusters in the west and northeast, alongside several outliers. Spatial econometric results identified government health expenditure, health insurance coverage, digital and transport infrastructure as significant positive drivers, with government health expenditure and digital development showing stronger cross-regional spillovers.

Conclusion

Coordination between the population and health services systems improved substantially, although regional disparities persisted. Further progress requires closer alignment between demographic change and health service capacity. Differentiated regional strategies are therefore needed: economically and resource-advantaged regions could harness agglomeration and spillover effects through cross-regional collaboration; central provinces should consolidate their catch-up momentum and strengthen their regional service capacity; and western and other low-coordination regions, facing population mobility and ageing pressures, should prioritise targeted fiscal transfers, workforce retention incentives and improved service accessibility to promote more equitable and sustainable development.

Instruction

Profound demographic transitions are reshaping health needs. These changes are posing challenges to the sustainability of health governance in China and worldwide. Population ageing, declining fertility, rapid urbanization, and changing morbidity profiles have collectively increased the demand for more diverse and continuous health services [1]. These demographic and epidemiological shifts manifest in rising burdens of chronic diseases, mental health conditions, and multimorbidity, as well as widening inequalities in service utilization across regions and population groups [2–4]. At the same time, the erosion of traditional family support networks and increasing dependency ratios have added further strain to formal health systems [5, 6]. In this context, the health service system serves as the core supply framework for meeting population health needs and ensuring equitable access to essential services [7, 8]. Its performance and sustainability, however, are also inherently influenced by demographic dynamics, as population change is closely associated with variations in the scale and structure of health service demand [9–11].

From a social-ecological perspective, the population and health services systems can be regarded as interacting subsystems [12, 13]. The population system, defined by size, structure and spatial distribution, generates health service demand, while the health services system constitutes the institutional supply side that allocates resources and delivers care [9, 11, 14]. Their relationship can thus be conceptualised as a dynamic demand–supply interaction [11, 15], providing the basis for analysing how demographic change aligns with service capacity in the pursuit of sustainable health governance [16, 17]. Building on this view, previous studies have examined the links between population dynamics and health-related subsystems from several perspectives. One line of research has focused on older populations, analysing the alignment between their health service needs and utilisation patterns, and assessing the adequacy of delivery systems in meeting age-related demands [18–21]. Other studies have analysed shifts in population health status and disease burden under conditions of growing mobility and migration, underscoring the emerging challenges faced by both medical and public health services [22–25]. A further strand of research has emphasised the role of preventive, integrated and coordinated health services in improving health outcomes [26, 27]. While these studies provide important insights, analyses of how population structural characteristics are embedded within health service systems remain insufficient, and comprehensive assessments of their coupled development are still lacking. This issue is particularly pertinent in China, where a large population base, rapid demographic transitions in ageing and urbanisation, and relatively constrained health resources have raised critical questions about the extent to which the health services system can adapt to demographic changes and maintain effective coordination with the population system.

Given the new normal of population growth and the emerging requirements for high-quality population development, examining the coupling and coordination between the population system and the health services system provides a useful approach to assess how well current health services adapt to demographic transitions and to identify their remaining limitations. Building on existing research, this study adopts a demand–supply perspective to examine the spatiotemporal evolution of coupling coordination between the two systems in China from 2010 to 2022 at national, regional, and provincial levels. It further explores spatial spillover effects to reveal inter-provincial dynamics and identifies the driving factors associated with the alignment between demographic change and health service capacity. The findings are expected to provide empirical evidence on the interaction between demographic dynamics and the health services system, enrich understanding of system coordination, and offer insights to support the optimisation of health governance.

Theoretical and conceptual framework

Population system

From a social-ecological perspective, the population system constitutes a key social subsystem that interacts continuously with the economy, society, and health [12, 13]. Demographic transition theory views population change as a dynamic process shaped by fertility, mortality, and migration [5]. Research on population dynamics typically emphasises three dimensions: size, structure, and spatial distribution [9, 11, 21, 28]. These dimensions not only describe the internal functioning of the population system but also establish how demographic change translates into health service demand.

Population size is closely associated with the overall volume of health needs [29]; population structure, including age composition and dependency ratios, influences the type and intensity of demand; and spatial distribution, reflected in density, rural–urban patterns and migration, affects the equity of resource allocation [29, 30]. This tripartite framework has been widely applied in demographic and public health studies, offering a clear basis for linking demographic dynamics with health service demand. Accordingly, the population system is generally regarded as the demand-generating side of health governance.

Health services system

Within the health governance framework, the health services system is commonly understood as the institutional supply side responsible for organising and delivering services in response to population health needs [14, 15, 31]. More broadly, the World Health Organization defines a health system as encompassing all organisations, people and actions whose primary purpose is to promote, restore, or maintain health [32]. This definition extends beyond service delivery to include governance, financing, pharmaceuticals, technologies, and community-based initiatives.

In some studies, the term “health services system” is used more narrowly, referring either to the institutional arrangements through which preventive, curative, and rehabilitative services are delivered, or to the organisational structures of service provision across different administrative levels [10, 20, 29, 33]. Despite differences in scope, these perspectives converge in conceiving the health services system as the organised supply dimension most directly connected to population health demand.

Building on these definitional distinctions, existing frameworks further characterise the health services system as inherently multidimensional. The WHO “six building blocks” approach identifies service delivery, health workforce, information systems, essential medicines and technologies, financing, and governance as the core elements of effective functioning [32]. Other frameworks emphasise stewardship, financing, resource generation, and provision, or focus on outcomes such as efficiency, quality, and equity [34]. While differing in emphasis, these approaches consistently highlight the health services system as a mechanism that integrates resource mobilisation, effective service provision, and equitable coverage, thereby linking institutional arrangements to population health needs.

Interaction between the population system and the health services system

From a demand–supply perspective, the population system represents the source of health service needs, while the health services system constitutes the institutional supply side that responds to these needs [9, 11, 15]. Population size influences the overall volume of demand; structural shifts such as ageing or changing dependency ratios modify its type and intensity; and spatial patterns of density, urban–rural distribution, and migration generate regional disparities in access and utilisation [29, 30]. On the supply side, the health services system responds through resource allocation, service delivery and the assurance of accessibility and equity [15–17, 20, 21].

When demand and supply are aligned, the system functions more efficiently and equitably; when misaligned, shortages, inefficiencies, or spatial inequalities may arise [18, 29, 35]. This dynamic interaction provides the theoretical foundation for analysing the coupling and coordination of the two systems, underscoring the practical importance of aligning demographic change with service capacity to achieve sustainable health governance.

Building on this perspective, this study develops a conceptual framework to operationalise the interaction between the population and health services systems. The population system is defined as the demand side, characterised by size, structure, population density, and spatial distribution, whereas the health services system is defined as the supply side, characterised by health resources, health service provision and efficiency, and health management and coverage. To evaluate the alignment between demographic change and service capacity, the coupling coordination degree model is applied. Figure 1 presents this conceptual framework.

Fig. 1.

Conceptual framework of the interaction between the population system and the health services system

Materials and methods

Data source

The data for this study were obtained from statistical yearbooks covering the period from 2010 to 2022, including the China Health Statistics Yearbook and the China Statistical Yearbook [36, 37]. Administrative data at the national, regional, and provincial levels were compiled into a balanced spatiotemporal panel dataset. Following the classification defined in the China Health Statistics Yearbook, regions were divided into the three major spatial units: East, Central and West. The eastern region comprises 11 provinces and municipalities: Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong and Hainan; the central region includes Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei and Hunan; and the western region includes Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang. Taiwan, Hong Kong and Macao, China, were not included in the analysis because relevant comparable data were unavailable.

The period 2010–2022 was selected for three reasons. First, considering the availability and quality of data, this timeframe provides access to a consistent and comprehensive set of high-quality national statistics, supporting the reliability, comparability, and accuracy of analysis at both national and provincial levels. Secondly, the policy background was considered. Since the launch of China’s new round of healthcare reform in 2009, a series of measures have been implemented to expand coverage, strengthen primary care, and improve resource allocation. The chosen period captures the long-term demographic and health service transformations occurring within this reform framework. Thirdly, the relatively long timespan allows for the identification of both short-term fluctuations and long-term trends, such as accelerated population ageing, fertility decline, and structural adjustments in the health services system. By extending the dataset to 2022, the analysis incorporates the most recent demographic and health service developments, thereby enhancing the timeliness and policy relevance of the findings.

Data analysis

To evaluate the coordination between the population system and the health services system, this study established a comprehensive analytical framework. First, an indicator system was developed on the basis of existing literature and data availability, and was aligned with the theoretical framework proposed in this study. Second, the data of each indicator selected in each system were standardized and the entropy value method is used to determine the weight of each indicator; factor analysis was further conducted as a robustness check to verify the stability of the weighting scheme. Third, the coupling coordination degree (CCD) model was applied to assess the coupling and coordination level between the population system and the health services system. Finally, spatial dependence and the underlying driving mechanisms were assessed using global and local spatial autocorrelation analysis and further investigated through the Spatial Durbin Model (SDM).

Methodology

Selection of indicators

This study follows the principles of scientific validity, comprehensiveness, and data availability to construct an integrated evaluation indicator system for the population system and the health services system, as shown in Table 1. The weights of each indicator were determined using the entropy method.

Table 1.

Evaluation indicator system for the population and health services systems

System	Sub-dimension	Indicators	Unit	Indicator direction	Average weight
Population system	Population size	Total population at year-end	10,000 persons	+	0.0998
		Natural population growth rate	‰	+	0.0657
	Population structure	Proportion of population aged 0–14 years	%	+	0.2188
		Proportion of population aged 15–64 years	%	+	0.1120
		Proportion of population aged ≥ 65 years	%	+	0.1551
		Total dependency ratio	%	+	0.1285
	Population density and distribution	Population density (per sq. km)	Persons/km²	+	0.0998
		Urban–rural population ratio	–	+	0.1200
Health services system	Health resource	Number of hospitals	piece	+	0.0742
		Hospital beds per 1,000 population	Beds/1,000 persons	+	0.0521
		Hospital health professionals per 1,000 population	Persons/1,000	+	0.0748
		Number of primary health care institutions	–	+	0.0809
		Primary beds per 1,000 population	Beds/1,000 persons	+	0.0664
		Primary health professionals per 1,000 population	Persons/1,000	+	0.1047
		Number of public health institutions	–	+	0.0586
		Public health beds per 10,000 population	Beds/1,0000 persons	+	0.0564
		Public health professionals per 1,000 population	Persons/1,000	+	0.0646
	Service provision and efficiency	Outpatient visits per physician per day (hospitals)	Visits/day	+	0.0433
		Inpatient bed-days per physician per day (hospitals)	Days/day	+	0.0796
		Outpatient visits per physician per day (primary)	Visits/day	+	0.0769
		Inpatient bed-days per physician per day (primary)	Days/day	+	0.0855
	health management and coverage	Child health management rate (< 7 years)	%	+	0.0368
		Systematic maternal health management rate	%	+	0.0441

The population system was assessed across three dimensions: population size, population structure, and population density and distribution. Population size captures both the overall stock and dynamic changes of the population; population structure is reflected in the proportions aged 0–14, 15–64, and ≥ 65 years, along with the total dependency ratio, which together indicate demographic composition and ageing pressures; population density and the urban–rural ratio represent spatial distribution and settlement patterns.

The health services system was evaluated across three dimensions: health resources, health service provision and efficiency, and health management and coverage. Health resources encompass the availability of medical institutions, bed capacity, and health personnel, constituting the essential material and human foundation of service delivery. Health service provision and efficiency reflect both the intensity of healthcare utilization and the effectiveness with which resources are transformed into services, thereby indicating the system’s operational capacity to meet population health needs. Health management and coverage embody the accessibility and coverage of essential health services, with child and maternal health management serving as representative indicators of the system’s preventive and long-term health protection functions.

Entropy weight method

The entropy method, was derived from the concept of information entropy in thermodynamics and evaluates the degree of variation in data to determine indicator weights [38]. Indicators that exhibit greater variability across regions and over time are assigned higher weights, reflecting their stronger influence on the comprehensive development level. Compared with subjective weighting methods, the entropy method reduces bias and enhances comparability, making it particularly suitable for constructing composite indices in large-scale spatiotemporal studies. In this study, the entropy method was applied to objectively assign weights to the indicators of the population and health services systems. The comprehensive development indices for the two systems were then calculated by weighted summation [39, 40]. The calculation steps are as follows:

(a) Data normalisation

1

2

(b) Normalising the indicators/proportions

3

(c) Calculating the information entropy of each indicator

4

(d) Determining the weight of each indicator

5

(e) Measuring the comprehensive development level of each subsystem

6

Here, X_ij and X_ij’ are the raw and normalised values of indicator j in region i, respectively. X_max and X_min denote the maximum and minimum values of the indicator across all regions and years. P_ij is the proportion of indicator j in region i. E_j and W_j represent the entropy value and weight of indicator j, respectively; and Ui indicates the comprehensive development index of the subsystem. The dataset is a balanced spatiotemporal panel covering 31 provinces in China from 2010 to 2022. The rightmost column of Table 1 reports the average weights of each indicator over the study period.

Coupling coordination degree model

The coupling coordination degree(CCD) integrates the development levels of both systems and reflects the extent of their interaction. It is widely applied to measure the coordinated development between multiple systems or subsystems [41]. Accordingly, this study adopts a two-system CCD model, expressed as follows:

7

8

9

where C denotes the coupling degree. U₁ and U₂ represent the comprehensive evaluation values of the population system and the health services system, respectively. α₁and α₂ are the weights reflecting the contribution of each system. To reflect the theoretical symmetry between demand and supply, this study adopts an equal-weight specification in the baseline model (α₁ = α₂ = 1/2). From a governance perspective, the population system represents the demand side of health services, while the health services system constitutes the supply side; both are indispensable for achieving sustainable and equitable health development, and thus merit equal consideration. Nevertheless, to address the limitations of a purely normative assumption, factor analysis was employed to derive empirical system-level weights. The analysis extracted two factors with eigenvalues greater than 1, which together explained 76.3% of the total variance. Based on factor loadings, the empirical weights were estimated as α₁ = 0.431 for the population system and α₂ = 0.569 for the health services system. Robustness checks show that results under both weighting schemes yield broadly consistent conclusions, thereby supporting the appropriateness of the baseline equal-weight assumption while providing empirical validation.

T shows the comprehensive evaluation index of the two systems, and D represents the coupling coordination degree. A larger value of D indicates a higher overall development level of the subsystems and a stronger degree of coordination between them. Referring to previous studies [39, 40, 42], the types of coupling coordination development are classified as shown in Table 2.

Table 2.

Classification of coupling coordination degree

Coordination type	Value range	Coordination rank
Dysfunctional recession	0 ≤ D < 0.1	Extremely dysfunctional recession
	0.1 ≤ D < 0.2	Severely dysfunctional recession
	0.2 ≤ D < 0.3	Moderate dysregulation recession
	0.3 ≤ D < 0.4	Mild dysregulation recession
Overcoordination stage	0.4 ≤ D < 0.5	On the verge of coupled coordination
	0.5 ≤ D < 0.6	Barely coupled coordination
Coordinated development	0.6 ≤ D < 0.7	Primary coupling coordination
	0.7 ≤ D < 0.8	Intermediate coupling coordination
	0.8 ≤ D < 0.9	Good coupling coordination
	0.9 ≤ D ≤ 1	Quality coupling coordination

Spatial autocorrelation

Global and local spatial autocorrelation analyses were applied to examine the spatial clustering characteristics of the CCD between the population and health services systems across China’s provincial-level regions. The spatial weight matrix was constructed using a contiguity-based approach, defining provinces with shared boundaries as neighbors to capture direct geographic interactions. Global spatial autocorrelation is measured by Moran’s I, which quantifies the degree of spatial aggregation or dispersion among spatial units [43]. The specific formula is as follows:

10

where , . Z_i denotes the CCD of region i, n is the number of spatial units, and W_ij is the spatial weight. Moran’s I ranges from − 1 to 1: a value between 0 and 1 indicates positive spatial correlation, while a value between − 1 and 0 indicates negative spatial correlation.

Local spatial autocorrelation is measured by the local Moran’s I (or LISA), defined as:

11

Based on LISA results, spatial relationships can be classified into five types: high–high, low–low, high–low, low–high, and insignificant clusters. A positive I_i indicates spatial clustering, where high (or low) values are surrounded by similar values. A negative I_i implies spatial mismatch, where a high value is surrounded by low values, or vice versa. In this study, local spatial autocorrelation is applied to reveal the spatial distribution patterns and heterogeneity of the CCD index.

Spatial Durbin Model

Given the potential spatial spillover effects in the CCD across provinces, this study adopts a spatial econometric approach. The commonly used spatial econometric models include the Spatial Lag Model (SLM), the Spatial Error Model (SEM), and The Spatial Durbin Model (SDM). Among them, the SDM is widely recognised as a foundational framework, as it captures spatial dependence in both the dependent and independent variable [44].

Model selection. To determine the appropriate model form, the Lagrange Multiplier (LM) test and the robust LM test, as well as the Likelihood Ratio (LR) and Wald tests, were conducted. Under the inverse-distance spatial weight matrix, the test statistics were all significant at the 1% level, rejecting the null hypothesis of no spatial effects and supporting the choice of the SDM over the SLM or SEM. In addition, the Hausman test indicated that the fixed-effects specification was preferred to the random-effects model. Therefore, the empirical analysis employs a two-way fixed-effects SDM.

Model specification. The SDM is formally expressed as follows:

12

Where D_it denotes the CCD in province i at time t; C₀ is the constant term; ρ is the spatial autoregressive coefficient of the dependent variable; Wij are elements of the row-standardised spatial weight matrix W; X_it is the vector of explanatory variables with coefficients β; θ indicates the coefficients of the spatially lagged explanatory variables, capturing their spillover effects; µi and λi denote province and year fixed effects, respectively; and ε_it is the error term.

Spatial weights. The spatial weight matrix W was specified as a row-standardised first-order contiguity (Rook) matrix with zero diagonal elements, in which provinces sharing a common border are considered neighbors. This specification is widely used in spatial econometric analysis as it captures direct geographic interactions across adjacent administrative units. Robustness was further assessed using alternative matrices, including a k-nearest neighbors (k-NN) specification.

Endogeneity and identification. To mitigate simultaneity (e.g., provinces with higher CCD may subsequently attract greater fiscal inputs or digital infrastructure upgrades), potentially endogenous covariates were lagged by one period in the baseline specification to mitigate reverse causality concerns.

Explanatory variables. Four principal explanatory variables were included to capture the institutional and structural determinants of CCD (see Table 3). Government health expenditure reflects the scale and intensity of public investment in health services [45, 46]. Health insurance coverage represents financial protection and access equity [47]. Digital infrastructure measures the availability and penetration of information technologies [48], and transport infrastructure reflects regional accessibility and connectivity [49]. Each of these variables was constructed as a composite index using the entropy-weight method, based on multiple sub-indicators in the China Health Statistical Yearbook and China Statistical Yearbook. By integrating diverse sub-indicators, the entropy-weight method provides a more balanced and representative measure than reliance on any single indicator.

Table 3.

Explanatory and control variables: description and measurement

Variable	Indicators	Unit	Direction	Average weight
Explanatory variables
Government health expenditure	Government health expenditure	100 million CNY	+	0.5181
	Health expenditure as % of GDP	%	+	0.4818
Health insurance coverage	Number of insured persons	10,000 persons	+	0.6362
	Health insurance coverage rate	%	+	0.3637
Digital infrastructure	Number of internet users	10,000 persons	+	0.4771
	Broadband access subscribers	10,000 households	+	0.5228
Transport infrastructure	Road density	km/ km2	+	0.4406
	Railway density	km/ km2	+	0.5593
Control variable
Economic development	GDP per capita	CNY	+	–
Ageing level	Proportion of elderly population	%	+	–
Urbanisation rate	Proportion of urban population	%	+	–
Disposable income level	Per capita disposable income	CNY	+	–

Note: Average weights were calculated using the entropy-weight method based on temporal and provincial variation

Control variables. To account for socio-economic background conditions, additional covariates were included: economic development, ageing rate, urbanisation rate, and disposable income level. Province and year fixed effects were also incorporated. All continuous variables were standardised prior to estimation to ensure comparability and numerical stability across specifications.

Analysis of the comprehensive evaluation index and coupling coordination degree between the population and health services systems from 2010 to 2022

From 2010 to 2022, the comprehensive evaluation indices of China’s population system and health services system exhibited an overall upward trend. The population system index increased from 0.140 to 0.674, captured long-term shifts in population size, structure, and spatial distribution. The health services system index rose from 0.196 to 0.722, reflected continuous growth in health resources and service provision (Fig. 2a). Correspondingly, the coupling coordination degree (CCD) between the two systems advanced from 0.407 to 0.835, indicated a progressive enhancement in the coordination of population dynamics and health service development (Fig. 2b).

Fig. 2.

(a) Trends in the comprehensive development indices of the Population system and the Health services system, 2010–2022. (b) Trends in the coupling coordination degree (CCD) between the Population system and the Health services system, 2010–2022

At the regional level, both subsystems have shown steady improvement, though their development trajectories varied across regions (Fig. 3a and b). In the east, the CCD increased from 0.682 in 2010 to 0.885 in 2022, with both the population and health services indices remained at relatively high levels and advanced steadily. In the central region, the CCD rose from 0.484 to 0.754, with a pronounced gap between the two subsystems during 2010–2012, followed by gradual convergence in later years. In the west, the CCD grew from 0.474 to 0.699, with an initial gap between the two subsystems, had followed by steady growth and narrowing differences over time.

Fig. 3.

(a) Trends in the comprehensive development indices of the Population system (U₁) and the Health services system (U₂) across eastern, central, and western China, 2010–2022 (Regions follow the China Statistical Yearbook classification; values are region-level composites, not provincial averages.). (b) Trends in the coupling coordination degree (CCD) between the Population and Health services systems across eastern, central, and western China, 2010–2022

At the provincial level, CCD values ranged from 0.560 to 0.715 in 2010, with eastern coastal provinces concentrated at the upper end of the distribution (e.g., Shanghai at 0.715), while several inland provinces stayed closer to 0.56 (Fig. 4a). By 2016, the range shifted upward to 0.652–0.709, with most provinces clustered above 0.67 (Fig. 4b). In 2022, CCD values further increased to 0.680–0.752, with leading provinces such as Shanghai, Jiangsu, Zhejiang, and Beijing exceeding 0.72, while some western provinces remained below 0.70 (Fig. 4c).

Fig. 4.

Provincial variation in the coupling coordination degree (CCD) across China, (a) 2010, (b) 2016, and (c) 2022

Spatial autocorrelation analysis of the coupling coordination between population and health services system from 2010 to 2022

Global spatial autocorrelation of the coupling coordination degree (CCD) between the population and health service systems was assessed using Moran’s I. As reported in Table 4, the global Moran’s I values ranged from 0.184 to 0.238 during 2010–2022, with all values statistically significant. This reflects a stable and positive spatial autocorrelation, indicating persistent clustering and spatial spillover effects among provinces.

Table 4.

Global Moran’s I values for the coupling coordination of the Population and Health services systems, 2010–2022

Year	Moran’s I	Sd(I)	Z	Year	Moran’s I	Sd(I)	Z
2010	0.184*	0.106	2.047	2017	0.251**	0.103	2.757
2011	0.195**	0.086	2.651	2018	0.268**	0.117	2.573
2012	0.203**	0.082	2.878	2019	0.273**	0.114	2.684
2013	0.216**	0.087	2.862	2020	0.262**	0.108	2.731
2014	0.237**	0.091	2.967	2021	0.254**	0.109	2.633
2015	0.254**	0.103	2.786	2022	0.238*	0.120	2.258
2016	0.263**	0.111	2.667

Note: * P < 0.05, ** P < 0.01, *** P < 0.001

To analyze the evolution of the spatial distribution of the coupling coordination degree (CCD) across provinces, the years 2010, 2016, and 2022 were selected as cross-sectional time points. The Local Indicators of Spatial Association (LISA) were applied to these years, and the spatial clustering patterns of CCD values for 31 provincial units in mainland China were presented in Figs. 5; Table 5.

Fig. 5.

Moran scatterplots of the coupling coordination degree (CCD) between the Population system and Health services system in China for the years 2010, 2016, and 2022

Table 5.

Provincial distribution of Local Moran’s I cluster types for the coupling coordination degree in China, 2010, 2016 and 2022

Year	High–high (HH)	High–low (HL)	Low–high (LH)	Low–low (LL)	Not significant
2010	Beijing, Tianjin, Liaoning, Shandong, Shanghai, Fujian [6]	Hubei, Guangdong [2]	None	Xinjiang, Qinghai, Sichuan, Yunnan, Guangxi, Jiangxi [6]	Hebei, Shanxi, Inner Mongolia, Jilin, Heilongjiang, Jiangsu, Zhejiang, Anhui, Henan, Hunan, Chongqing, Guizhou, Tibet, Shaanxi, Gansu, Ningxia, Hainan [17]
2016	Beijing, Tianjin, Hebei, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian [8]	Sichuan [1]	Liaoning [1]	Xinjiang, Tibet, Qinghai, Jilin, Heilongjiang [5]	Shanxi, Inner Mongolia, Henan, Anhui, Jiangxi, Hubei, Hunan, Guangdong, Guangxi, Hainan, Chongqing, Guizhou, Yunnan, Shaanxi, Gansu, Ningxia [16]
2022	Tianjin, Hebei, Shanxi, Shandong, Henan, Hubei, Shanghai, Zhejiang [8]	Sichuan [1]	Liaoning, Anhui, Fujian [3]	Xinjiang, Tibet, Qinghai, Jilin, Heilongjiang [5]	Beijing, Inner Mongolia, Jiangsu, Jiangxi, Hunan, Guangdong, Guangxi, Hainan, Chongqing, Guizhou, Yunnan, Shaanxi, Gansu, Ningxia [14]

Note: Values in parentheses denote the number of provincial-level regions classified within each Local Moran’s I cluster category. HH represents high-value regions surrounded by neighbouring high-value regions; HL represents high-value regions surrounded by neighbouring low-value regions; LH represents low-value regions surrounded by neighbouring high-value regions; and LL represents low-value regions surrounded by neighbouring low-value regions. ‘Not significant’ denotes regions in which local spatial autocorrelation was not statistically significant at the specified level. Taiwan, Hong Kong and Macao, China, were not included in the analysis because comparable data were unavailable

In 2010, the LISA results indicated that high–high (HH) clusters were concentrated in several eastern coastal and northern provinces, including Beijing, Tianjin, Liaoning, Shanghai, Fujian, and Shandong. These provinces, located in the Bohai Rim and the Yangtze River Delta, benefited from relatively strong economies, better governance capacity, and a closer alignment between demographic structures and healthcare resources. In contrast, low–low (LL) clusters were mainly distributed in central and western provinces such as Jiangxi, Guangxi, Sichuan, Yunnan, Qinghai, and Xinjiang, where economic constraints, demographic imbalances, and limited healthcare capacity corresponded with relatively low coordination levels.

By 2016, spatial clustering became more evident. HH clusters expanded to include Beijing, Tianjin, Hebei, additional provinces in the Yangtze River Delta such as Jiangsu and Zhejiang, as well as Fujian and Shandong. Liaoning appeared as a low–high (LH) outlier, exhibiting relatively low internal coordination but surrounded by provinces with higher values, while Sichuan emerged as a high–low (HL) outlier, showing higher internal coordination but neighboring provinces with lower levels. At the same time, Heilongjiang, Tibet, Qinghai, and Xinjiang formed LL clusters, indicating that relatively low coordination persisted in parts of the Northeast and the western regions.

By 2022, the HH clusters had shifted somewhat, now concentrated in Tianjin, Hebei, Zhejiang, Shandong, Henan, and Hubei. This indicates that, in addition to the eastern coastal provinces, some central provinces such as Henan and Hubei had joined the high-value clusters, reflecting the rising role of the Central Plains in population–health coordination. Liaoning, Fujian, and Anhui were identified as LH outliers, while Sichuan remained a HL outlier. LL clusters persisted in Tibet, Qinghai, and Xinjiang, and also included the northeastern provinces of Jilin and Heilongjiang. These regions continued to show relatively low coordination levels, likely influenced by a combination of geographic, economic, and demographic factors.

Factors influencing the coupling coordination between the population and health services system

Spatial econometric analysis was applied to investigate the driving factors of the CCD. To determine the appropriate econometric model specification, a series of diagnostic tests were performed, including LM, LR, Wald, and Hausman tests. The results rejected the null hypothesis of no spatial dependence, and both LR and Wald tests supported the SDM as the optimal specification. The Hausman test favored fixed effects, justifying the use of a two-way fixed effects SDM to account for spatial and temporal heterogeneity (Table 6).

Table 6.

Results of specification tests for spatial econometric models

Specification Test	Test Statistic	P
Moran’s I	137.426	< 0.001
LM-lag test	15.136	< 0.001
LM-error test	12.478	< 0.001
Robust LM-lag test	10.112	0.006
Robust LM-error test	7.903	0.004
LR_Spatial_lag	82.17	0.002
LR_Spatial_error	96.44	< 0.001
Wald_Spatial_lag	38.29	< 0.001
Wald_Spatial_error	24.36	0.001
Hausman text	58.75	< 0.001

In the baseline OLS model (Table 7), without accounting for spatial effects, government health expenditure (GHE) (β = 0.322, P < 0.01) and digital infrastructure (DI) (β = 0.281, P < 0.05) were positively associated with CCD. Health insurance coverage (HIC) (β = 0.102, P < 0.01) and transport infrastructure (TI) (β = 0.157, P < 0.05) also showed significant positive associations. In the SDM, these associations were stronger. GHE (β = 0.434, P < 0.01), DI (β = 0.503, P < 0.001), HIC (β = 0.149, P < 0.001), and TI (β = 0.325, P < 0.05) all remained statistically significant. The spatial autoregressive coefficient was significant (ρ = 0.493, P < 0.001), confirming spatial dependence and spillover effects across provinces. Robustness checks using a K-nearest neighbor (K = 4) weight matrix and a lagged SDM produced results consistent with those from the baseline SDM based on Rook’s contiguity.

Table 7.

Regression results of driving factors for the Coupling Coordination Degree (CCD) between population and health services systems: OLS and SDM Models

Variable/Model	OLS	SDM(Rook)	SDM (KNN = 4)	SDM_LAG
GHE	0.322**(0.110)	0.434**(0.168)	0.412**(0.160)	0.276*(0.134)
HIC	0.102**(0.039)	0.149***(0.045)	0.125**(0.048)	0.127**(0.049)
DI	0.281*(0.142)	0.503***(0.151)	0.578***(0.159)	0.492**(0.173)
TI	0.157*(0.072)	0.325*(0.160)	0.291*(0.145)	0.302*(0.147)
ED	0.094(0.070)	0.115*(0.057)	0.135**(0.052)	0.168**(0.065)
AGE	-0.011(0.027)	-0.026(0.051)	-0.031(0.054)	-0.014(0.052)
UR	0.028(0.031)	0.035(0.243)	0.046(0.251)	0.029(0.239)
DIL	0.154(0.141)	0.171(0.194)	0.252(0.189)	0.248(0.191)
Constant	0.563***(0.073)	0.298***(0.090)	0.301***(0.091)	0.274**(0.105)
ρ(rho)		0.493***(0.095)	0.441***(0.110)	0.488***(0.097)
sigma2_e		0.006	0.0018	0.003
R2	0.723	0.831	0.829	0.772

Notes: * P < 0.05; ** P < 0.01; *** P < 0.001; Standard errors in parentheses

To further assess spatial effects, the SDM estimates were decomposed into direct, indirect, and total effects. As shown in Table 8, GHE (total effect = 0.489, P < 0.01) and DI (total effect = 0.569, P < 0.001) were significantly associated with CCD, mainly through direct effects. TI also showed a significant total effect (β = 0.356, P < 0.05), primarily from the direct effect. HIC was significant as well (total effect = 0.168, P < 0.001). Notably, both DI (indirect effect = 0.084, P < 0.01) and GHE (indirect effect = 0.069, P < 0.05) displayed significant positive spillover effects, indicating that improvements in these areas can enhance coordination in neighbouring regions. By contrast, the indirect effects of TI and HIC were not statistically significant, suggesting their influence is mainly localised. Among the control variables, ED was positively associated with CCD (direct effect = 0.109; total effect = 0.131, P < 0.05), whereas AGE showed a negative direct effect that was not statistically significant. UR and DIL likewise did not show significant direct or indirect effects.

Table 8.

Impact decomposition of the Spatial Durbin Model with two-way fixed effects

Variable	Direct Effect	Indirect Effect	Total Effect
GHE	0.420**(0.162)	0.069*(0.035)	0.489**(0.166)
HIC	0.142***(0.043)	0.026(0.027)	0.168***(0.051)
DI	0.485***(0.147)	0.084**(0.032)	0.569***(0.150)
TI	0.305*(0.155)	0.051(0.033)	0.356*(0.158)
ED	0.109*(0.055)	0.022(0.020)	0.131*(0.059)
AGE	-0.024(0.052)	-0.005(0.013)	-0.029(0.054)
UR	0.041(0.237)	0.012(0.025)	0.053(0.239)
DIL	0.163(0.189)	0.033(0.050)	0.196(0.202)

Notes: * P < 0.05; ** P < 0.01; *** P < 0.001. Standard errors are reported in parentheses. Estimates are derived from the partial impact decomposition of the SDM, employing a Rook contiguity spatial weight matrix and two-way fixed effects

Discussion

Coordinated development of the population and health services systems from a comprehensive perspective

Over the past decade, socio-economic development has exerted a notable influence on China’s population and health services systems. Between 2010 and 2022, the comprehensive development indices of both subsystems rose steadily, from relatively low to higher levels, suggesting gradual improvement in demographic dynamics and health system performance [50, 51].

On the population side, accelerated ageing, rapid urbanisation, and large-scale migration reshaped demographic patterns, resulting in more complex and diverse health service needs. Rising household incomes and growing health awareness further contributed to increasing expectations for equitable and high-quality care [52, 53]. On the supply side, government and societal investment expanded the coverage and quality of health services [50]. The number of medical institutions, bed capacity, and health personnel increased, while advances in technology, digitalisation, and infrastructure enhanced accessibility and system responsiveness [54, 55]. Collectively, these changes indicate that the health services system has become more capable of adapting to evolving population needs.

Taken together, the two systems demonstrate an interdependent and dynamically interactive relationship: demographic change shapes the scale and structure of demand, while improvements in resources and efficiency provide the capacity to respond. This bidirectional linkage highlights the strategic importance of maintaining demand–supply coordination in health governance.

Evolutionary trends and regional variations in the coupling coordination between the population and health services systems

From 2010 to 2022, the coupling coordination degree (CCD) rose steadily, moving from an initial stage of dysfunction towards higher-level coordination. This reflects not only stronger connections between demographic dynamics and service provision, but also progress in resource allocation and management. With a rising share of older adults, faster urbanisation, and intensified cross-regional mobility, demand for medical and public health services grew substantially. These shifts underscore the need for governments to anticipate health needs arising from demographic change and to ensure resources are allocated effectively.

Regionally, all areas improved but along distinct trajectories. The east sustained a relatively high level of coordination, supported by a more developed health system, stronger fiscal capacity, and substantial population inflows [56]. The economic dynamism of the eastern coastal provinces has attracted large numbers of migrants, with arrival of younger labour migrants both mitigated ageing pressures and boosted demand for high-quality services, thereby aligning demographic structure with provision capacity [57]. High population density and rapid urbanisation also fostered more intensive and efficient service delivery, accelerating the conversion of resources into services [58, 59]. By contrast, the central region initially faced a marked supply–demand gap, as ageing outpaced service expansion. This gap narrowed over time with migrant return, infrastructure improvements, and targeted policy support, reflecting a catch-up effect. In the western and parts of northeast, CCD levels remained lower, shaped by demographic imbalances. Persistent labour outflows aggravated “relative ageing,” with a shrinking younger population undermining demand-side balance and constraining the vitality of local health systems [60]. Under such conditions, even quantitative increases in resources could not fully offset demographic pressures. Overall, regional variation in CCD reflects not only disparities in supply but also the combined effects of population structure and mobility patterns. Addressing these challenges requires differentiated strategies, including improving resource efficiency in less-developed areas, fostering interprovincial cooperation to manage mobility-driven imbalances, and integrating demographic change into long-term health planning.

Spatial agglomeration and clustering patterns of coupling coordination degree

Between 2010 and 2022, the CCD between China’s population and health services systems showed significant spatial autocorrelation, with most provinces resembling their neighbours. This produced stable “high–high” and “low–low” clusters, alongside several outliers.

High–high clusters emerged in the eastern coastal and northern provinces, centred on the Yangtze River Delta (Shanghai, Jiangsu, Zhejiang), and the Beijing–Tianjin–Hebei region. Their evolution was driven not only by favourable resource endowments but also by institutional innovation and factor mobility. These provinces implemented hierarchical diagnosis and treatment, medical insurance payment reforms, and medical consortiums early on, demonstrating stronger health governance capacity [61–63]. For example, the “Medical Consortium Model” and diagnosis-related group payment piloted in Shanghai and Jiangsu were subsequently diffused to other eastern and some central provinces, which facilitated more efficient resource allocation and service delivery [61]. Meanwhile, infrastructure improvements, particularly the expansion of the high-speed rail network and the development of telemedicine, facilitated cross-regional flows of people and health services, aligning demographic change more closely with service provision [64]. Over time, these high-value clusters expanded into central provinces such as Henan and Hubei, reflecting both infrastructure-driven mobility and the diffusion of supportive policies. Initiatives such as the “Healthy Central Plains” strategy and targeted national investment in health infrastructure further strengthened regional collaboration and institutional spillovers [65]. In Henan, the development of regional medical centres, the upgrading of county-level medical centres and the expansion of smart healthcare services may have strengthened the responsiveness of the health services system to population needs [66]. In Hubei, high-level medical resources, particularly general and tertiary hospitals, are markedly concentrated in central Wuhan. Combined with policies promoting hierarchical care, medical consortia, remote consultation and improved transport accessibility, this concentration may have strengthened Wuhan’s regional service function and created favourable conditions for resource sharing and referral linkages within Hubei and neighbouring areas [67].

By contrast, western and some northeastern provinces remained in low-value clusters, forming relatively stable agglomerations at the lower end of coordination. This outcome reflects the combined effects of demographic pressures, weaker economic foundations, and mobility constraints. Provinces such as Xinjiang, Tibet, and Qinghai have experienced sustained youth outmigration and accelerated ageing, deepening structural imbalances. At the same time, fiscal limitations and geographic remoteness restricted the circulation of medical resources and populations, constraining the alleviation of supply–demand mismatches. The persistence of such patterns suggests a potential “lock-in” effect, highlighting the need for stronger regional cooperation and institutional innovation to improve coordination [63, 68].

Notably, some provinces exhibited spatial outlier patterns. Sichuan, for example, displayed a high–low (HL) configuration in 2016 and 2022, with relatively high intra-provincial coordination but lower levels in surrounding provinces. This pattern may be associated with Sichuan’s role as a medical hub in western China. Chengdu hosts a large share of tertiary hospitals and high-quality resources, supported by relatively strong public health governance, which has raised intra-provincial coordination [69]. Yet the province’s mountainous and basin-dominated terrain constrains transport links, and limited cross-regional flows of medical resources and populations have restricted external spillovers [70], resulting in a “high within, low around” spatial configuration. Conversely, northeastern provinces such as Liaoning occasionally presented a low–high (LH) outlier pattern in certain years. This may be linked to industrial restructuring, population outflows, and rapid ageing, which increased demand for long-term care while weakening fiscal capacity and service utilisation [71]. Although Shenyang and Dalian host major medical centres, the pull of higher-level facilities in the in the Beijing–Tianjin–Hebei region, together with convenient transport connections, has likely diverted part of the demand, weakening the alignment between local population needs and available services.

Driving factors and spatial spillovers for the coupling coordination degree between population and health services systems

This study, based on the spatial Durbin model (SDM), examined the key drivers and spatial effects of the CCD between the population and the health services systems. The results show that government health expenditure (GHE), health insurance coverage (HIC), digital infrastructure (DI), and transport infrastructure (TI) exert significant positive effects on CCD.

GHE, as a major channel of public investment, plays a fundamental role in fostering system coordination. Previous studies have noted that fiscal input affects health system performance and service accessibility through both both supply- and demand-side mechanisms: on the supply side, by financing health facilities, workforce training and public health systems [58]; and on the demand side, by supporting health programmes and medical assistance that alleviate the burden on vulnerable groups and improve service accessibility [72]. HIC also functions as an institutional safeguard. On the demand side, higher coverage reduces financial burdens, narrows disparities in healthcare utilisation across income groups, and improves access for older and vulnerable populations. On the supply side, payment and reimbursement arrangements guide providers to adjust services and resource allocation in line with population needs [73]. These institutional pathways provide a theoretical basis for understanding the positive association between GHE, HIC, and CCD. DI and TI also play crucial roles in strengthening coordination. Existing studies suggest that digitalisation enhances access to and efficiency of health services by optimising information flows and reducing transaction costs. On the demand side, telemedicine, internet-based hospitals, and electronic health records help residents in remote areas overcome geographical barriers, while on the supply side, big health data and smart platforms improve resource allocation and promote interregional collaboration [74]. Similarly, existing studies indicate that improved transport conditions reduce the time and travel costs of accessing healthcare, particularly in rural and remote areas, and facilitate the cross-regional movement of staff, equipment, and pharmaceuticals, thereby improving system efficiency [75, 76]. Collectively, these mechanisms provide useful insights into the positive associations between DI, TI, and CCD.

The SDM results also reveal significant spatial spillover effects. In particular, DI and GHE exhibited significant positive spillovers, suggesting that investments in information networks and public health financing can generate benefits beyond local boundaries by strengthening cross-regional information flows, connectivity, and institutional capacity. These effects can be understood through several channels, including the movement of health services resources, the diffusion of service demand driven by mobility, fiscal transfers extending governance capacity, and the integrative functions of digital and transport networks that transcend administrative borders. This suggests that fiscal transfers and regional health funds may help amplify the spillover effects of public investment, while cross-regional platforms and information-sharing mechanisms may foster digital coordination.

Finally, the SDM was applied to conduct standard deviation–based scenario simulations to quantify marginal effects. A one standard deviation increase in DI was associated with an average rise of 0.045 in CCD (5.6%), while TI increased it by 0.044 (5.4%). The effects of HIC and GHE were 0.029 (3.6%) and 0.025 (3.1%), respectively. These findings suggest that the four drivers play distinct roles in promoting system coordination, and that higher levels of coordination require integrating multidimensional policy instruments. Infrastructure and digital development improve accessibility and resource mobility, while fiscal input and health insurance provide institutional safeguards that strengthen stability and equity. Together, they constitute complementary foundations that enhance the alignment between population needs and health service provision and provide strategic support for equity, accessibility, and sustainability.

Limitation

This study has several limitations. Firstly, although the indicator system captures key dimensions of the population and health services systems, data availability constrained the breadth of measurement, particularly in relation to health management and service coverage. The indicators available for this dimension were nationally comparable and reflected important components of essential public health services, but did not fully capture broader areas such as chronic disease management, health management for older people, mental health services or service quality. Future studies should incorporate a wider range of indicators when consistent and comparable provincial-level data become available. Secondly, the entropy weight method used for weighting, while systematic, may obscure certain demographic characteristics such as age distribution and dependency ratios, which could substantially influence health system performance. Third, in calculating the CCD, equal weights were assigned to the two subsystems; although this choice is reasonable, the static assumption cannot capture potential variation in subsystem importance across regions or over time. Fourth, while lagging explanatory variables helps to mitigate reverse causality, it does not fully address potential endogeneity. Future research could explore instrumental variable methods or other causal inference techniques as data availability improves. Lastly, although the study identifies the main drivers of CCD, it does not further uncover the mechanisms through which these factors influence coordination, such as resource allocation, institutional arrangements, or behavioural incentives.

Conclusion

Based on the temporal evolution of indicators for China’s population and health services systems between 2010 and 2022, this study adopts the coupling coordination degree model to evaluate the degree of coupling coordination between the two systems at national, regional, and provincial scales. Overall, coordination progressed from a low to a medium–high level nationwide, with the east sustaining higher coordination, while the central and western regions gradually progressed from low to primary coordination. Spatial analysis confirmed significant dependence in provincial CCD, with patterns dominated by “high–high” clusters in the Yangtze River Delta and the Beijing–Tianjin–Hebei region that gradually extended into central provinces, and by persistent “low–low” clusters in west and north-east, alongside several outliers. The coordination is influenced by multiple factors and exhibits spatial spillovers. Government health expenditure, health insurance coverage, digital and transport infrastructure all had positive effects, with government health expenditure and digital development showing stronger cross-regional spillovers.

Overall, the coordinated development of the population and health services systems requires closer alignment between demographic dynamics and health service capacity. Given the regional variations identified, future policy design should adopt differentiated regional strategies. Eastern provinces could further translate their resource agglomeration advantages into wider regional spillover benefits by strengthening cross-provincial referral networks, regional medical consortia, interoperable digital health platforms, interprovincial information sharing and mutual recognition of examination results. Central provinces could consolidate their catch-up momentum by developing regional medical centres as service hubs, improving county-level medical capacity, strengthening primary care networks and referral coordination, and enhancing their capacity to absorb and further diffuse health resources across neighbouring areas. Resource planning in these provinces should also respond more closely to population return, urban agglomeration and ageing. Western provinces, as well as low-coordination areas in north-eastern China, could prioritise targeted fiscal transfers, workforce retention incentives, telemedicine, mobile health services, emergency transport networks and interprovincial counterpart assistance to address geographical barriers, population outflow, ageing pressures and limited local service capacity.

Abbreviations

AGE

Ageing rate

CCD

Coupling coordination degree

DI

Digital infrastructure

DIL

Disposable income level

ED

Economic development

GHE

Government health expenditure

HH

High-high cluster

HIC

Health insurance coverage

HL

High-low outlier

LH

Low-high outlier

LISA

Local indicators of spatial association

LL

Low-low cluster

OLS

Ordinary least squares

SDM

Spatial Durbin Model

TI

Transport infrastructure

UR

Urbanisation rate

Author contributions

LL and LZ conceived the study. LL wrote the initial draft which was updated by RH, YX. LL, WR performed the statistical analyses. All authors have read and approved the manuscript.

Funding

This study was funded by the National Natural Science Foundation of China (Grant No: 72104073; 72474162). The funders did not participate in study design, data collection, analysis, interpretation of data and manuscript writing.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.